Management of the Terrestrial Small Mammal and Lizard Communities in the Dune System Of

Home , Ctenophorus pictus, Ctenotus regius, Desert mouse, Dusky hopping mouse, Fawn hopping mouse, Hopping mouse

Management of the terrestrial small mammal and lizard communities in the dune system of

Sturt National Park, Australia:

Historic and contemporary effects of pastoralism and fox predation

Ulrike Sabine Klöcker (Dipl. – Biol., Rheinische Friedrich-Wilhelms Universität Bonn, Germany)

Thesis submitted in fulfilment of the requirements for the degree of Doctor of Philosophy School of Biological, Earth and Environmental Sciences

The University of New South Wales, Sydney, Australia

2009

Abstract This thesis addressed three issues related to the management and conservation of small terrestrial vertebrates in the arid zone. The study site was an amalgamation of pastoral properties forming the now protected area of Sturt National Park in far-western New South Wales, Australia. Thus firstly, it assessed recovery from disturbance accrued through more than a century of Sheep grazing. Vegetation parameters, Fox, Cat and Rabbit abundance, and the small vertebrate communities were compared, with distance to watering points used as a surrogate for grazing intensity. Secondly, the impacts of small-scale but intensive combined Fox and Rabbit control on small vertebrates were investigated. Thirdly, the ecology of the rare Dusky Hopping Mouse (Notomys fuscus) was used as an exemplar to illustrate and discuss some of the complexities related to the conservation of small terrestrial vertebrates, with a particular focus on desert rodents. Thirty-five years after the removal of livestock and the closure of watering points, areas that were historically heavily disturbed are now nearly indistinguishable from nearby relatively undisturbed areas, despite uncontrolled native herbivore (kangaroo) abundance. Rainfall patterns, rather than grazing history, were responsible for the observed variation between individual sites and may overlay potential residual grazing effects. In this study reduction of Fox activity was successful, whereas the reduction of Rabbit activity was not. The treatment had no evident effect on the small vertebrate community, total abundance or species richness and the majority of individual species. There was strong indication however that the House Mouse (Mus musculus) increased in abundance as a consequence of reduced Fox activity. Cat abundance also increased following Fox reduction but low sample size prevented analysis and discrimination between treatment or rainfall related effects. New insights were gained on the ecology of the Dusky Hopping Mouse, including temporal and spatial distribution, diet, home-range, density and population structure. A key result was the identification of the seeds of the Sandhill Wattle (Acacia ligulata) as a major food resource, which significantly influenced the mice’s movements, habitat use, distribution and density. The new information allows a more accurate assessment of threats, the identification of critical habitat and refuge areas and ultimately a focus for conservation actions for the species.

I Acknowledgements First and foremost I would like to thank Dr. David Croft and Dr. Ingrid Witte. Without either of them my candidature and this project would never have eventuated. David, my supervisor, I greatly thank for his continued support in all conceptual, analytical and other varied aspects of the project. Ingrid, the Area Manager of Sturt National Park, I greatly thank for making funding for the project available and for the never-ending support, herself and her staff readily provided in the day to day running of the project. Also I would like to thank my co-supervisor Dr. Peter Banks for initial advice on study design. I am grateful for all the varied assistance the National Parks staff provided. In particular I would like to thank Johnny Illies for his genuine interest in the doings of the ‘crazy mouse-catcher’, for his diligent care for my vehicle (s) and for the manufacture of various pieces of equipment exactly to my wishes; Dan Hough for his assistance with all aspects of the control of introduced animals; Billy Thomas for ensuring I got back from the field safely and in one piece and the odd emergency car repair; Errol Nye and Lisa Montgomery without whom I would have never been able to dig all those pitfall holes, John Jackson for providing support and assistance whenever needed and Stephen Naven for various help with GIS related issues. I sincerely thank the many volunteers for their assistance and enthusiasm in the field. To Christiane and Tony I owe particular thanks for their great contribution to the radio-tracking of the hopping mice. To my friend Ingrid and her ever changing (generally increasing) entourage of two and four-legged creatures, I am indebted beyond words. Her support and the distraction, love and joy the furred and beaked faces provided were invaluable in overcoming the challenges posed by a PhD candidature and life in general. I wish to heartedly thank my friends the ‘Tiffies’ and the Bonner Clique as well as my relatives back home for letting the distance between Germany and Australia seem insignificant through the frequent email and phone communication and for understanding (well, trying to understand) my love for the life in the Australian outback. To Alexandra Ley and Anke Frank I am forever grateful for their never failing enthusiasm, constructive criticism and encouraging words. Anke and her flatmate Yvonne I also like to thank her for their hospitality during my repeated stays in Sydney. For the countless times I was welcomed in their home in Broken Hill I sincerely thank Beris and Andy Jenkins and Chris and Paul Adams. A shared problem is a problem halved and so I am grateful to have had Isabelle Wolf as a ‘Leidensgenossin’ to discuss, whinge and whine about the bureaucracy involved with being an international PhD student, ill-behaving software, problematic vehicles ect. Naomi Callen, Jaz Lawes and my sister, I thank for reading and commenting on chapter drafts and Alexandra Ley for proof-reading the final draft. It goes without saying that I lovingly thank my parents and my siblings for letting me embark on and supporting me during the endeavour that is and was my PhD candidature. I am immensely grateful to my sisters partner Markus, my aunt Heidi and her partner Lothar, my cousin Claudia and her husband Kai and many other relatives and friends for holding things together at home during times of familial hardship. This project was funded by DECC-Wildlife Division Tibooburra area and an Endeavour International Postgraduate Research Scholarship. The project was approved by the Animal Care & Ethics committee (ACEC) with the University of New South Wales (ACEC 06/14B) and the Department of Environment, Climate Change and Water (Scientific Licence No S12358).

This thesis is dedicated to my family.

Table of Contents Chapter 1: General Introduction and Study Rationale_____ - 1 - 1.1 The decline of native fauna in Australia since European settlement __ - 1 - 1.1.1 Declines in mammal and lizard diversity in the arid and semi-arid zones ______- 2 - 1.2 Key threats to small native vertebrates in the arid and semi-arid zones ______- 4 - 1.2.1 Habitat modification and disturbance through pastoral activity______- 4 - 1.2.2 Non-native fauna ______- 6 - 1.2.3 Climate change ______- 7 - 1.3 Conservation management______- 7 - 1.3.1 Legislation ______- 7 - 1.3.2 Natural heritage versus cultural heritage and visitor recreation ______- 8 - 1.3.3 Science in conservation ______- 8 - 1.3.4 Research and conservation in the arid zone ______- 8 - 1.3.5 Focus of conservation actions______- 9 - 1.4 Study rationale ______- 10 - 1.5 Thesis structure ______- 11 - Chapter 2: Study Area and General Methods ______- 12 - 2.1 Study area ______- 12 - 2.1.1 Description of topography, landforms and vegetation ______- 12 - 2.1.2 Climate ______- 13 - 2.1.3 Rainfall prior to and during study period ______- 14 - 2.2 Reasons for selecting the study area ______- 14 - 2.3 Sampling sites ______- 15 - 2.4 General methods ______- 16 - 2.4.1 Timeline ______- 16 - 2.4.2 Weather recording ______- 18 - 2.4.3 Data manipulation and analysis ______- 18 - Chapter 3: Review of the Effects of Grazing and Artificial Watering Points in the Australian Rangelands - 20 - 3.1 History of pastoralism in the rangelands ______- 20 - 3.2 Overview of impacts of livestock grazing and artificial watering points ______- 21 - 3.2.1 Direct and indirect effects of artificial watering points ______- 21 - 3.3 Impacts on the edaphic environment ______- 23 - 3.4 Impacts on vegetation ______- 24 - 3.5 Impacts on fauna______- 25 - 3.5.1 Impacts resulting from grazing activities ______- 25 - 3.5.2 Impacts related to artificial watering points ______- 26 - 3.5.3 Invertebrates ______- 27 - 3.5.4 Small mammals and Lizards______- 28 - IV

3.5.5 Non-native fauna ______- 30 - 3.6 Grazing management, management of artificial watering points and rangeland recovery ______- 31 - 3.6.1 Rangeland recovery ______- 32 - Chapter 4: Residual Effects of Grazing and Artificial Watering Points on Flora and Fauna in Sturt National Park ______- 34 - 4.1 Artificial watering points in Sturt National Park ______- 34 - 4.2 General Aim______- 34 - 4.3 Experimental design, site information and general methodology ___ - 35 - 4.3.1 General data analysis ______- 38 - 4.4 Effects on Vegetation ______- 40 - 4.4.1 Hypotheses ______- 40 - 4.4.2 Methods ______- 41 - 4.4.3 Results ______- 44 - 4.5 Effects on Fauna ______- 58 - 4.5.1 Hypotheses ______- 58 - 4.5.2 Methods ______- 58 - 4.5.3 Results ______- 66 - 4.6 Overall Discussion and Conclusion ______- 82 - 4.6.1 General ______- 82 - 4.6.2 Vegetation______- 83 - 4.6.3 Fauna ______- 83 - 4.6.4 Non-native plants and animals ______- 84 - 4.6.5 Threatened grazing-sensitive species ______- 87 - 4.6.6 Timeframe for post-pastoralism recovery ______- 88 - 4.6.7 Study limitations______- 89 - 4.6.8 Contribution of other herbivores than livestock to impacts of water- focussed grazing ______- 91 - 4.7 Conclusion ______- 92 - Chapter 5: The effect of combined Fox and Rabbit reduction on small vertebrates and Cats ______- 93 - 5.1 Introduction______- 93 - 5.1.1 The Red Fox - a brief profile ______- 93 - 5.1.2 The Fox as a conservation threat ______- 93 - 5.1.3 Fox control ______- 96 - 5.1.4 Interactions between Foxes, Rabbits and Cats ______- 97 - 5.1.5 Management ______- 101 - 5.2 Aims and Hypotheses ______- 102 - 5.3 Methods ______- 103 - 5.3.1 Study area ______- 103 - 5.3.2 Experimental Design ______- 104 - 5.3.3 Rabbit reduction ______- 105 - 5.3.4 Fox reduction ______- 108 -

5.3.5 Monitoring ______- 110 - 5.3.6 Data manipulation and analysis ______- 111 - 5.4 Results ______- 112 - 5.4.1 Effectiveness of Fox reduction______- 112 - 5.4.2 Effectiveness of Rabbit reduction______- 114 - 5.4.3 Effect of Fox reduction on Rabbits______- 115 - 5.4.4 Effect of Fox reduction on Cats ______- 116 - 5.4.5 Effect of Fox reduction on small mammals ______- 117 - 5.4.6 Effect of Fox reduction on lizards ______- 123 - 5.5 Discussion______- 132 - 5.5.1 General ______- 132 - 5.5.2 Effectiveness of control actions ______- 133 - 5.5.3 Impact of reduced Fox activity on Rabbits______- 135 - 5.5.4 Impact of reduced Fox activity on Cats ______- 135 - 5.5.5 Impact of reduced Fox activity on small mammals and lizards _____ - 136 - 5.5.6 Conservation implications ______- 137 - 5.5.7 Study limitations______- 138 - 5.6 Conclusion ______- 142 - Chapter 6: The Ecology of the Dusky Hopping Mouse (Notomys fuscus) ______- 144 - 6.1 Introduction______- 144 - 6.1.1 Australian arid zone rodents______- 144 - 6.1.2 The Dusky Hopping Mouse (Notomys fuscus) ______- 145 - 6.2 Aims ______- 147 - 6.3 Temporal and spatial distribution______- 147 - 6.3.1 Methods ______- 148 - 6.3.2 Results ______- 149 - 6.3.3 Discussion______- 153 - 6.4 Diet ______- 155 - 6.4.1 Methods ______- 156 - 6.4.2 Results ______- 156 - 6.4.3 Discussion______- 158 - 6.5 Dusky Hopping Mice and Sandhill Wattles (Acacia ligulata) ______- 159 - 6.5.1 Methods ______- 160 - 6.5.2 Results ______- 161 - 6.5.3 Discussion______- 165 - 6.6 Temporal and spatial activity ______- 166 - 6.6.1 Methods ______- 166 - 6.6.2 Results ______- 170 - 6.6.3 Discussion______- 173 - 6.7 Social organisation and behaviour ______- 175 - 6.7.1 Methods ______- 175 - 6.7.2 Results ______- 176 - 6.7.3 Discussion______- 179 - 6.8 Miscellaneous observations ______- 183 - 6.8.1 Burrow systems ______- 183 -

6.8.2 Native predators ______- 186 - 6.9 Summary ______- 186 - 6.10 Conservation and management ______- 187 - 6.10.1 Threats to the species ______- 188 - Chapter 7: Thesis Synthesis ______- 192 - 7.1 Conservation management in arid and semi-arid Australia______- 192 - 7.2 Effectiveness of livestock removal ______- 193 - 7.3 Outcome of Fox control ______- 193 - 7.4 Ecology and conservation of the Dusky Hopping Mouse ______- 194 - 7.5 Issues in the management of small vertebrate conservation______- 194 - 7.5.1 Environmental stochasticity ______- 194 - 7.5.2 Habitat connectivity ______- 196 - 7.5.3 Difficulties in controlling Foxes and Rabbits______- 197 - 7.5.4 Goats an emerging issue______- 199 - 7.6 Suggestions for future research ______- 201 - 7.6.1 Effects of livestock removal: Species specific investigation in plants and invertebrates______- 201 - 7.6.2 Diet of the Fox and Cat in Sturt National Park______- 201 - 7.6.3 Movements of Foxes and Cats in arid and semi-arid areas ______- 201 - 7.6.4 The House Mouse as a conservation threat to small vertebrates ____ - 202 - 7.6.5 Relationship between N. fuscus and A. ligulata and refuge areas ___ - 202 - 7.6.6 Mating system of N. fuscus and comparison to related species _____ - 203 - References ______- 204 - Appendices ______- 226 -

VII

List of Tables TABLE 2.1: OVERVIEW OF THE TIMING OF FIELD WORK AND THE CONTRIBUTION OF FIELD WORK PERIODS/SESSIONS TO THE THREE TOPICS COVERED WITHIN THE THESIS...... - 17 - TABLE 4.1: SITE NAMES AND ABBREVIATIONS USED IN THIS CHAPTER...... - 37 - TABLE 4.2: DESCRIPTION OF GROUND COVER CATEGORIES...... - 41 - TABLE 4.3: TEMPORAL VARIATION IN GROUND COVER...... - 45 - TABLE 4.4: TOTAL ABUNDANCE (ALL REPLICATES AND SITES) AND PERCENTAGE OF ALL LIVING SHRUBS COUNTED...... - 51 - TABLE 4.5: MONTHLY PRECIPITATION (MM) RECEIVED AT SAMPLING SITES DURING 2006...... - 55 - TABLE 4.6: INVERTEBRATE CAPTURES...... - 66 - TABLE 4.7: SIGNIFICANT DIFFERENCES IN THE INVERTEBRATE ASSEMBLAGE BETWEEN SITES...... - 69 - TABLE 4.8: COMPOSITION OF THE LIZARD ASSEMBLAGE...... - 70 - TABLE 4.9: COMPOSITION OF THE SMALL MAMMAL ASSEMBLAGE...... - 74 - TABLE 4.10: ACTIVITY OF FOXES, CATS AND RABBITS...... - 75 - TABLE 4.11: ASSOCIATIONS BETWEEN LIZARD SPECIES AS IDENTIFIED USING A PRINCIPAL COMPONENT ANALYSIS...... - 78 - TABLE 4.12: FLORISTIC HABITAT FACTORS IDENTIFIED WITH A PRINCIPAL COMPONENT ANALYSIS USED FOR SUBSEQUENT LINEAR REGRESSION ANALYSIS...... - 79 - TABLE 4.13: CORRELATION OF LIZARD COMMUNITY VARIABLES AND INDIVIDUAL SPECIES ABUNDANCES WITH HABITAT FACTORS...... - 79 - TABLE 4.14: CORRELATION OF LIZARD VARIABLES WITH TEMPERATURE, HUMIDITY AND WIND SPEED...... - 80 - TABLE 4.15: CORRELATION OF LIZARD VARIABLES WITH RAINFALL...... - 81 - TABLE 5.1: DETAILS OF STEPS UNDERTAKEN FOR RABBIT AND FOX CONTROL...... - 110 - TABLE 5.2: SMALL MAMMAL NUMBERS; TOTAL CAPTURES AND PERCENT OF TOTAL ...... - 117 - TABLE 5.3: RESULTS OF COMPARISON BETWEEN TREATMENTS (FRIEDMAN TEST) FOR SMALL MAMMAL VARIABLES. DF = 7...... - 119 - TABLE 5.4: LIZARD NUMBERS, TOTAL CAPTURES AND PERCENT OF TOTAL ...... - 124 - TABLE 5.5: RESULTS OF COMPARISON BETWEEN TREATMENTS (FRIEDMAN TEST, DF = 7) FOR LIZARD SPECIES. SPECIES WHERE THE FRIEDMAN TEST WAS SIGNIFICANT FOR ‘IMPACT’ SITES ONLY ARE MOST LIKELY TO HAVE BEEN AFFECTED BY THE REDUCTION IN FOX ACTIVITY...... - 126 - TABLE 6.1: CRITERIA USED TO SCORE NOTOMYS FUSCUS ABUNDANCE ON TRANSECTS ...... - 149 - TABLE 6.2: LIST OF FOOD ITEMS THAT ARE PART OF THE DIET OF NOTOMYS FUSCUS IN STURT NP...... - 158 - TABLE 6.3: COMPARISON OF THE MINIMUM DISTANCE TO AN ACACIA LIGULATA SHRUB BETWEEN THE ACTUAL OBSERVED LOCATIONS OF A NOTOMYS FUSCUS INDIVIDUAL AND A RANDOM SAMPLE OF LOCATIONS...... - 164 - TABLE 6.4: SUBJECT CHARACTERISTICS, SAMPLING INTENSITY, HOME (KERNEL, 95 %) AND CORE RANGES (KERNEL, 50 %) IN HA, AND MEAN AND MAXIMUM LINEAR DISTANCE MOVED BETWEEN FIXES IN METRES FOR NOTOMYS FUSCUS...... - 171 - TABLE 6.5: NUMBER OF N. FUSCUS KNOWN TO BE ALIVE AT PARTICULAR DUNE SECTIONS (1 HA) AND THE PERCENTAGE OF RECAPTURES...... - 177 - TABLE 6.6: OBSERVATIONS ON THE NUMBER OF NOTOMYS FUSCUS SHARING A BURROW SYSTEM ...... - 178 -

List of Figures FIGURE 2.1: LOCATION OF SAMPLING SITES...... - 13 - FIGURE 2.2: ANNUAL RAINFALL (MM) MEASURED AT THE FORT GREY HOMESTEAD...... - 14 - FIGURE 2.3: MAP OF STUDY AREA AND SAMPLING SITES IN THE WESTERN PART OF STURT NATIONAL PARK...... - 17 - FIGURE 4.1: COMPARISON OF MEAN COVER OF PLANT CATEGORIES BETWEEN DISTURBANCE LEVELS...... - 46 - FIGURE 4.2: COMPARISON OF MEAN PATCHINESS OF GROUND COVER CATEGORIES BETWEEN DISTURBANCE LEVELS...... - 47 - FIGURE 4.3: COMPARISON OF MEAN COVER OF PLANT TYPES BETWEEN DISTURBANCE LEVELS...... - 48 - FIGURE 4.4: COMPARISON OF MEAN PATCHINESS OF PLANT TYPES BETWEEN DISTURBANCE LEVELS...... - 49 - FIGURE 4.5: COMPARISON OF GROUND COVER SPECIES COMPOSITION BETWEEN DISTURBANCE LEVELS...... - 50 - FIGURE 4.6: COMPARISON OF MEAN DENSITY OF SHRUB CLASSES BETWEEN DISTURBANCE LEVELS...... - 52 - FIGURE 4.7: DENSITY OF BRASSICA TOURNEFORTII AT SAMPLING SITES...... - 54 - FIGURE 4.8: SIMILARITY IN GROUND COVER BETWEEN SAMPLES...... - 57 -

FIGURE 4.9: ARRANGEMENT OF PAIRS OF PITFALL TRAPS () AND TRACK PLOTS () IN A ONE HECTARE TRAPPING GRID ON A DUNE...... - 59 - FIGURE 4.10: TEMPORAL VARIATION IN THE INVERTEBRATE ASSEMBLAGE...... - 67 - FIGURE 4.11: COMPARISON OF THE INVERTEBRATE ASSEMBLAGE BETWEEN SITES...... - 68 - FIGURE 4.12: TEMPORAL VARIATION IN THE LIZARD ASSEMBLAGE...... - 71 - FIGURE 4.13: COMPARISON OF THE LIZARD ASSEMBLAGE BETWEEN DISTURBANCE LEVELS...... - 72 - FIGURE 4.14: COMPARISON OF THE SMALL MAMMAL ASSEMBLAGE BETWEEN SITES...... - 74 - FIGURE 4.15: COMPARISON OF RABBIT, FOX AND CAT ACTIVITY BETWEEN DISTURBANCE LEVELS...... - 76 - FIGURE 4.16: TEMPORAL AND SPATIAL VARIATION IN THE LIZARD ASSEMBLAGE AND HABITAT VARIABLES AS CAUSATIVE FACTORS...... - 77 - FIGURE 5.1: MAP OF STUDY AREA SHOWING LOCATION OF MONITORING SITES AND ARRANGEMENT OF TREATMENT AREAS...... - 107 - FIGURE 5.2: COMPARISON OF FOX ACTIVITY BETWEEN TREATMENTS; (A) MEAN ABUNDANCE (B) MEAN RANK PER SURVEY...... - 114 - FIGURE 5.3: RABBIT ACTIVITY OVER STUDY PERIOD; (A) MEAN ABUNDANCE (B) MEAN RANK PER SURVEY...... - 115 - FIGURE 5.4: COMPARISON OF RABBIT ACTIVITY BETWEEN TREATMENTS; (A) MEAN ABUNDANCE (B) MEAN RANK PER SURVEY...... - 116 - FIGURE 5.5: COMPARISON OF CAT ACTIVITY BETWEEN TREATMENTS; (A) MEAN ABUNDANCE (B) MEAN RANK PER SURVEY...... - 117 - FIGURE 5.6: MDS PLOT OF THE SMALL MAMMAL ASSEMBLAGE AND ITS CHANGES WITH SURVEY PERIOD AND TREATMENT...... - 118 - FIGURE 5.7: COMPARISON OF SMALL MAMMAL CAPTURES BETWEEN TREATMENTS; (A) MEAN ABUNDANCE (B) MEAN RANK PER SURVEY...... - 120 - FIGURE 5.8: COMPARISON OF SMALL MAMMAL RICHNESS BETWEEN TREATMENTS; (A) MEAN ABUNDANCE (B) MEAN RANK PER SURVEY...... - 120 - FIGURE 5.9: COMPARISON OF MUS MUSCULUS CAPTURES BETWEEN TREATMENTS; (A) MEAN ABUNDANCE (B) MEAN RANK PER SURVEY...... - 121 - FIGURE 5.10: COMPARISON OF NOTOMYS FUSCUS CAPTURES BETWEEN TREATMENTS; (A) MEAN ABUNDANCE (B) MEAN RANK PER SURVEY...... - 122 - FIGURE 5.11: COMPARISON OF PSEUDOMYS HERMANNSBURGENSIS CAPTURES BETWEEN TREATMENTS; (A) MEAN ABUNDANCE (B) MEAN RANK PER SURVEY...... - 122 - FIGURE 5.12: COMPARISON OF SMINTHOPSIS CRASSICAUDATA CAPTURES BETWEEN TREATMENTS; (A) MEAN ABUNDANCE (B) MEAN RANK PER SURVEY...... - 123 - FIGURE 5.13: MDS PLOT OF THE LIZARD ASSEMBLAGE AND ITS CHANGES WITH SURVEY PERIOD AND TREATMENT...... - 125 - FIGURE 5.14: COMPARISON OF CTENOPHORUS NUCHALIS CAPTURES BETWEEN TREATMENTS; (A) MEAN ABUNDANCE (B) MEAN RANK PER SURVEY...... - 127 - FIGURE 5.15: COMPARISON OF VARANUS GOULDII CAPTURES BETWEEN TREATMENTS; (A) MEAN ABUNDANCE (B) MEAN RANK PER SURVEY...... - 127 - FIGURE 5.16: COMPARISON OF NEPHRURUS LEVIS CAPTURES BETWEEN TREATMENTS; (A) MEAN ABUNDANCE (B) MEAN RANK PER SURVEY...... - 128 -

FIGURE 5.17: COMPARISON OF CTENOPHORUS PICTUS CAPTURES BETWEEN TREATMENTS; (A) MEAN ABUNDANCE (B) MEAN RANK PER SURVEY...... - 128 - FIGURE 5.18: COMPARISON OF CTENOTUS TAENIATUS CAPTURES BETWEEN TREATMENTS; (A) MEAN ABUNDANCE (B) MEAN RANK PER SURVEY...... - 129 - FIGURE 5.19: COMPARISON OF EREMIASCINCUS FASCIOLATUS CAPTURES BETWEEN TREATMENTS; (A) MEAN ABUNDANCE (B) MEAN RANK PER SURVEY...... - 129 - FIGURE 5.20: COMPARISON OF THE TOTAL LIZARD CAPTURES BETWEEN TREATMENTS; (A) MEAN ABUNDANCE (B) MEAN RANK PER SURVEY...... - 130 - FIGURE 5.21: COMPARISON OF LIZARD RICHNESS BETWEEN TREATMENTS; (A) MEAN ABUNDANCE (B) MEAN RANK PER SURVEY...... - 130 - FIGURE 5.22: COMPARISON OF CTENOTUS SCHOMBURGKII CAPTURES BETWEEN TREATMENTS; (A) MEAN ABUNDANCE (B) MEAN RANK PER SURVEY...... - 131 - FIGURE 5.23: COMPARISON OF RHYNCHOEDURA ORNATA CAPTURES BETWEEN TREATMENTS; (A) MEAN ABUNDANCE (B) MEAN RANK PER SURVEY...... - 132 - FIGURE 6.1: DISTRIBUTION OF THE DUSKY HOPPING MOUSE, NOTOMYS FUSCUS...... - 146 - FIGURE 6.2: ABUNDANCE OF NOTOMYS FUSCUS BETWEEN JANUARY 2006 AND APRIL 2008. THE SOLID LINE REFERS TO N. FUSCUS CAPTURES, THE DOTTED LINE TO TRACK PLOT VISITATION BY THE SPECIES...... - 150 - FIGURE 6.3: RAINFALL AND NOTOMYS FUSCUS CAPTURES OVER TIME...... - 151 - FIGURE 6.4: PERCENTAGE OF TRAPPING GRIDS WITH NOTOMYS FUSCUS EVIDENCE IN THE FORM OF CAPTURES AND FOOTPRINTS ON TRACK PLOTS...... - 152 - FIGURE 6.5: SPATIAL DISTRIBUTION AND DENSITY (ACTIVITY INDEX) OF NOTOMYS FUSCUS OVER THE STUDY AREA IN A) JANUARY 2008 AND B) IN JULY 2008...... - 153 - 2 3 FIGURE 6.6: ABUNDANCE (A), MEAN PROJECTED COVER IN M (B) AND MEAN VOLUME IN M (C) OF A. LIGULATA AT N. FUSCUS PRESENCE AND ABSENCE SITES...... - 162 - FIGURE 6.7: CORRELATION BETWEEN N. FUSCUS ACTIVITY AS A MEASURE OF DENSITY AND THE TOTAL VOLUME OF A. LIGULATA...... - 163 - FIGURE 6.8: EXEMPLARY ILLUSTRATION OF DISTRIBUTION OF THE MINIMUM DISTANCE TO THE NEXT A. LIGULATA SHRUB BETWEEN THE OBSERVED LOCATIONS OF A NOTOMYS FUSCUS INDIVIDUAL AND A RANDOM SAMPLE OF LOCATIONS. N OBSERVED = 78, N RANDOM = 1000...... - 164 - FIGURE 6.9: HOME RANGES (95 % OF FIXES) AND PARTIAL HOME RANGES (25 %, 50 % AND 75 % OF FIXES) OF A) 9 FEMALE AND B) 5 MALE NOTOMYS FUSCUS. THE VARIOUS LINES REPRESENT THE HOME RANGE ESTIMATES OF EACH INDIVIDUAL MOUSE...... - 172 - FIGURE 6.10: HOME RANGES OF THREE FEMALE AND A MALE N. FUSCUS...... - 172 -

List of Appendices APPENDIX 1: HISTORICAL MAP OF STUDY AREA...... - 226 - APPENDIX 2: LIST OF PLANT SPECIES IDENTIFIED IN THE STUDY AREA...... - 227 - APPENDIX 3: COMPARISON OF OVERALL GROUND COVER BETWEEN SAMPLING SITES...... - 229 - APPENDIX 4: COMPARISON OF INDIVIDUAL GROUND COVER VARIABLES BETWEEN SITES...... - 230 - APPENDIX 5: DISSIMILARITIES IN GROUND COVER CATEGORIES BETWEEN SAMPLING SESSIONS...... - 231 - APPENDIX 6: DISSIMILARITIES IN GROUND COVER CATEGORIES BETWEEN SITES...... - 232 - APPENDIX 7: DIFFERENCES IN INVERTEBRATE COMMUNITY VARIABLES AND INDIVIDUAL INVERTEBRATE GROUPS BETWEEN SITES...... - 232 - APPENDIX 8: SITE DIFFERENCES IN LIZARD COMMUNITY VARIABLES AND INDIVIDUAL LIZARD SPECIES...... - 233 - APPENDIX 9: DISSIMILARITIES IN INVERTEBRATE GROUPS BETWEEN SAMPLING SESSIONS...... - 234 - APPENDIX 10: DISSIMILARITIES IN INVERTEBRATE CATEGORIES BETWEEN SITES...... - 235 - APPENDIX 11: DISSIMILARITIES IN LIZARD SPECIES BETWEEN SAMPLING SESSIONS...... - 236 - APPENDIX 12: DISSIMILARITIES IN LIZARD SPECIES BETWEEN SITES...... - 237 - APPENDIX 13: RAINFALL IN THE YEARS 2005 (A), 2006 (B) AND 2007 (C) IN AUSTRALIA AND IN THE RANGE OF NOTOMYS FUSCUS (SHADY GREY AREA)...... - 238 - APPENDIX 14: KERNEL HOME RANGES (25 %, 50 %, 75 % AND 95 % OF FIXES) OF NOTOMYS FUSCUS...... - 240 - APPENDIX 15: EXEMPLARY PHOTOGRAPHS OF N. FUSCUS FOOTPRINTS AND ‘RUNWAYS’ TO ILLUSTRATE THE SCORING SYSTEM USED IN MEASURING N. FUSCUS ACTIVITY AS AN INDEX OF DENSITY...... - 244 - APPENDIX 16: EXAMPLE OF ACCUMULATION OF N. FUSCUS TRACKS INDICATING A FOOD SOURCE IN THIS CASE CATTLE BUSH (TRYCHODESMA ZYLANECUM)...... - 245 - APPENDIX 17: ACACIA LIGULATA SHRUB WITH FLOWERS, SEED PODS AND AN ACCUMULATION OF FALLEN SEEDS ON THE GROUND UNDERNEATH. THE SEEDS ARE A MAJOR COMPONENT OF THE DIET OF N. FUSCUS...... - 245 - APPENDIX 18: REMOTE CAMERA IMAGERY (INFRA-RED) OF A FEMALE N. FUSCUS AND HER YOUNG ON BURROW EXIT...... - 246 - APPENDIX 19: REMOTE CAMERA IMAGERY OF A GROUP OF EIGHT N. FUSCUS CONGREGATED AROUND THE BURROW ENTRANCE...... - 247 - APPENDIX 20: CO-HABITATION, PREDATION AND/OR COMPETITION? REMOTE CAMERA IMAGERY OF A SAND GOANNA (VARANUS GOULDII), A HOUSE MOUSE AND A N. FUSCUS USING THE SAME BURROW...... - 248 - APPENDIX 21: VARANUS GOULDII, A LIKELY NATIVE PREDATOR OF N. FUSCUS, INVESTIGATES A HOPPING MOUSE BURROW AND ALMOST COMPLETELY DISAPPEARS IN IT...... - 249 - APPENDIX 22: COMMENTS ON METHODOLOGY USED IN RESEARCH ON NOTOMYS FUSCUS...... - 250 -

List of abbreviations used in the Thesis ANOSIM Analysis of similarities ANOVA Analysis of variance AWP Artificial watering point CI Confidence interval COP Code of practice DECC Department of Environment and Climate Change Df Degrees of freedom e.g. from latin ‘exampli gratia’, meaning ‘for example’ et al. from latin ‘et allii’ meaning ‘and others’ i.e. from latin ‘id est’, meaning ‘that is’ km Kilometre m Metre MBACI Multiple-Before-After-Control-Impact; a study design for the investigation of environmental impacts mm Millimetre nMDS Non-parametric multidimensional scaling NSW New South Wales, state in the Commonwealth of Australia PCA Principal component analysis PERMANOVA Permutated multivariate analysis of variance pers. comm. Personal communication PRIMER Plymouth Routines in Multivariate Ecological Research QLD Queensland RHD Rabbit haemorrhagic disease SA South Australia SE Standard error SIMPER Species percentage contributions to similarity 1080 Sodium Monofluoroacetate, a poison SOP Standard operating procedure Sturt NP Sturt National Park the EPBC Act The Environment Protection and Biodiversity Conservation Act the NPW Act The National Park and Wildlife Act 1974 TSR Travelling stock route ver. Version, e.g. for computer software

A number of introduced and feral animals are frequently mentioned in the thesis. After being defined here they are only referred to by their common name in the thesis. These are: Rabbit the European Rabbit (Oryctolagus cuniculus) Cat the feral Cat (Felis catus); not including the domestic or stray Cat Fox the Red Fox (Vulpes vulpes) Sheep the domestic sheep Ovis aries Cattle Bos Taurus, Bos indicus and crossbreeds Goat Capra hircus Camel Camelus dromedarius Dingo Canis lupus dingo House Mouse Mus musculus but note that Mus domesticus and Mus musculus domesticus in present and past publications refer to the same species in Australia

Chapter 1: General Introduction and Study Rationale

1.1 The decline of native fauna in Australia since European settlement The continuing decline of native fauna is of great concern in Australia. The overall aim of this thesis is to further the knowledge and understanding of the ecology of small terrestrial vertebrates in the arid zone and thereby contributing to improved conservation efforts and reducing the rate of decline of the native Australian fauna.

Since European settlement in 1788, many faunal groups in Australia have experienced extinctions and declines in range or abundance. Species losses and declines are well documented for vertebrates. Approximately 13 % of all Australia’s known vertebrate species are now listed in Australia’s official national Environment Protection and Biodiversity Conservation Act (EPBC Act) as either ‘threatened’ or ‘vulnerable’ (Mackay 2006; Mackay et al. 2008). The summation of state and territory legislation plus authoritative national assessments, suggests that as many as 45 % of all Australian vertebrate species are in some form of serious decline in one or more parts of their range (Mackay et al. 2008). Mammals have been most adversely affected (Watts and Aslin 1981; Burbidge and McKenzie 1989) with at least 18 (ten marsupials and eight rodents) of the original 215 terrestrial species (8.4 %) having disappeared (numbers derived from Van Dyck and Strahan 2008). Australia has the highest rate of mammal extinctions of any continent and country in the world in recent historic times: 50 % of all the mammal species that have become extinct worldwide in the past 200 years were lost from the Australian fauna (Short and Smith 1994; MacPhee and Marx 1997). The situation initially appears to be less severe for birds, reptiles and amphibians, which together lost less than 1 % of the original faunas (Kennedy 1990). However, this is likely to be an underestimate since base-line information about taxa collected by the early colonists was less comprehensive and reliable for reptiles and amphibians than for more conspicuous faunal groups.

- 1 -

General introduction and study rationale

Animal extinctions and declines have not been spread evenly across the Australian continent. The arid-zone accounts for a disproportionate percentage of the species, especially mammals, that have become extinct or endangered since European settlement (Burbidge and McKenzie 1989; Beckmann 1990; Woinarski and Braithwaite 1990; Hoser 1991; Short and Smith 1994; Smith and Quin 1996; McKenzie et al. 2007).

1.1.1 Declines in mammal and lizard diversity in the arid and semi-arid zones

Arid lands The arid zone is very heterogeneous with variation in geology, topography, soil fertility, the distribution of plant communities and climatic influences (Stafford Smith and Morton 1990). The one constant characteristic of the arid zone is low rainfall: less than 250 mm annually for arid zones and between 250-500 mm for semi-arid zones, according to the simple definition by Leeper (1970). By this definition almost the entire interior of the Australian continent, more than 70 % of its landmass, is considered to be arid or semi-arid. The precipitation patterns in the arid and semi-arid zones of Australia are amongst the least predictable and most variable in time, space and intensity anywhere in the world (Low 1979). ‘Rangelands’ is an international term for areas where livestock are grazed on native vegetation and where rainfall is too low or too erratic for agricultural cropping or improved pastures (Newman and Condon 1969). It is often used as a synonym for the arid and semi-arid zone in Australia.

The arid zone fauna requires particular conservation attention for two reasons: firstly, the arid zone is biologically important. The extreme conditions of a desert environment necessitate highly adapted species resulting in a high level of endemism. Secondly, species in the arid zone are highly vulnerable to changes in the environment. The pay- off of high adaptation is reduced flexibility to adjust to rapid changes in their environment, thus the desert environment presents extreme conditions and many desert animals are living close to their limits of tolerance for one or more environmental variables (Louw and Seely 1982; Whitford 2002).

Mammals Despite the fact that almost two-thirds of land-dwelling mammals live in the more favourable mesic coastal areas of Australia (Beckmann 1990), more than half of the country’s endangered or extinct mammals are from the arid zone (Morton 1992). Across - 2 -

General introduction and study rationale the whole arid zone about one-third of all mammal species thought to have been present at the time of European settlement are now extinct (Burbidge and McKenzie 1989). Depending on the region and the mammalian group the percentage of species lost varies between 25 and 60 % (see review and tabulation in James et al. 1999). Not all species have proved equally vulnerable to decline or extinction. Most affected have been ground-dwelling, surface nesting species with an omnivorous or herbivorous diet (Burbidge and McKenzie 1989; Smith and Quin 1996; McKenzie et al. 2007) in the weight range of 35 g to 5.5 kg (Burbidge and McKenzie 1989; McKenzie et al. 2007), which corresponds to the range of prey size of Cats and Foxes.

The magnitude of species’ decline and extinction are similarly alarming for the western Division of NSW, where this study was carried out. Of the seventy-one native species recorded since European settlement, 38 % have disappeared and much of the remaining fauna has experienced dramatic declines with 28 species of mammals at risk. Marsupials (43 % of species lost) and rodents (65 % of species lost) fared particularly badly (Dickman 1993; Dickman et al. 1993; Dickman 1994). Some of these species have become extinct within the region and seven have become extinct nationally.

Lizards In contrast to the well documented demise of mammals, relatively little is known about the changes to the Australian herpetofauna since European settlement (Sadlier and Pressey 1994). The assessment of the past and/or present status of populations is hampered by the lack of baseline data. Data collection by early colonists and explorers was less intense and less reliable for reptiles than for more conspicuous mammals and so there are no historical records of consequence, very little long-term monitoring of populations and few estimates of population densities (Dickman et al. 1993; Sadlier and Pressey 1994). It is generally suggested that species’ declines have been less severe for reptiles than for mammals. No extinctions are known for any reptile species in Australia (EPBC Act: List of threatened species, Department of the Environment, Water, Heritage and the Arts (2008b)) and Burbidge and McKenzie (1989) reported a range decline in only 0.5 % of the extant reptile species despite the high endemism amongst Australian reptiles (Morton 1990). This resilience has been attributed to their largely insectivorous diets and reduced metabolic needs, particularly during food shortages when they can become inactive for long periods of time (Morton 1990). Recher and Lim (1990) have

- 3 -

General introduction and study rationale been less optimistic about the future of herpetofauna and have predicted similar declines to those experienced by mammals. The current threatened species list records 53 threatened reptiles (EPBC Act: List of threatened species, Department of the Environment, Water, Heritage and the Arts (2008b)), seven of which inhabit arid or semi-arid areas. A large proportion of the species of concern have been identified as having an association with habitat types containing Triodia groundcover (Sadlier and Pressey 1994).

1.2 Key threats to small native vertebrates in the arid and semi-arid zones The identification of factors threatening species and potentially causing species loss is critical for conservation but generally difficult to achieve. Often a combination of factors is involved and the impact of individual factors changes with the environmental conditions at a particular time. Johnson (2006) has undertaken an extensive review of available evidence for the mammals in Australia. He concludes that in many cases the past extinctions, as well as the decline in range or abundance of extant mammals, is related to a combination of habitat modification through pastoral activity and the influence of introduced animals, especially the predation from Foxes and Cats, and competition with or habitat degradation by the Rabbit.

1.2.1 Habitat modification and disturbance through pastoral activity Throughout most of the arid and semi-arid zones of Australia, the primary land use is rangeland pastoralism, where domestic stock have been introduced to graze native vegetation. Sheep are mostly grazed in the southern rangelands whilst Cattle are grazed in the central and northern rangelands (Newman and Condon 1969). In addition to the introduction of Sheep and Cattle as grazing animals, the provision of artificial sources of water, the introduction of foreign plants as forage, the elimination of a major predator, the Dingo and the clearing of overstorey trees have all significantly added to the alteration of grazed areas.

The provision of reliable sources of drinking water for stock, through trapping and storing surface runoff in constructed dams and/or the drilling of bores to access ground- water from the artesian basin, has arguably brought about the biggest change. This development of artificial watering points (AWP) allowed stock grazing to extend away from waterways to areas that were previously unavailable to stock except during wet

- 4 -

General introduction and study rationale conditions. The density of watering points increased many fold, allowing pastoralism, and all its associated problems, to extend to any areas that had at least some vegetation that could sustain livestock. Today, few areas in the rangelands remain that have never been impacted by livestock (James et al. 1995).

Apart from the impacts on soils and flora from the relatively intense grazing pressure around watering points, the provision of water in a landscape where surface water is naturally scarce, can change the temporal and spatial distribution of fauna (James et al. 1999). The increased localized activity and/or abundance of species other than livestock can intensify a range of direct and indirect impacts on the flora and fauna of the area, such as removal of vegetation, changes to the edaphic environment and localised predation or competition. Responses of biota to the presence of watering points vary as some species of both plants and animals increase in abundance closer to water (known as ‘increasers’), others decrease (known as ‘decreasers’) and the remainder show no detectable response (Landsberg et al. 1997).

In part AWP (in particular dams) have replaced the former natural waterholes in creek lines and drainage channels as they were often constructed in areas of natural waterholding capacity. Thus like natural waterholes, they are valuable to the survival and distribution of many species, both native and introduced. The increased number and high density of artificial and relatively reliable watering points has allowed continuous livestock grazing over broad expanses of the rangelands and thereby decreased heterogeneity in the landscape and lowered biodiversity (Fuhlendorf and Engle 2001). In an effort to combat this decreasing heterogeneity, the strategic closure of watering points, in addition to stock removal, is frequently used as a management tool by conservation managers. Even so there is still a lack of understanding of the benefits of AWP closure for conservation purposes in Australian rangeland reserves (Montague- Drake 2003; Croft et al. 2007; Box et al. 2008; Fensham and Fairfax 2008). This thesis aims to fill part of that knowledge gap by investigating whether a residual effect of watering points is still discernible after nearly four decades of livestock removal and the closure of artificial watering points.

- 5 -

General introduction and study rationale

1.2.2 Non-native fauna Australia has been subject to the accidental or intended introduction of many non-native species through human activity. Prior to European settlement such incidences are believed to have been rare, with the known example being the Dingo which is thought to have come to Australia with seafaring people from southeast Asia about 4000 years ago (Savolainen et al. 2004; Johnson 2006). The status of the Dingo is somewhat ambiguous as it is not endemic but due to its long existence in Australia it is generally considered a native rather than an introduced species (Trigger et al. 2008). Since European settlement however a multitude of animal species have been, and continue to be, brought to Australia. Most alien species failed to establish but many have done so successfully. Common biological attributes of successful invasive vertebrates include the following: they are abundant and widespread within their original range, they are tolerant of a wide range of abiotic conditions, and they have flexible requirements for habitat, a generalized diet and a high reproductive potential (Ehrlich 1989; Lockwood 1999; Forsyth et al. 2004). These newly arrived species pose considerable threats to Australian native species and ecosystems. In NSW for example non-native animal species cause concern for 40 % of threatened species (Coutts-Smith et al. 2007).

The European Red Fox, the feral Cat, the European Rabbit and the House Mouse have fared extremely well in Australia and have become widespread pests. Predation by Foxes and Cats, as well as habitat degradation and competition by Rabbits, are primary concerns in the conservation of small native vertebrates. Federal and state threatened species legislation recognise these species as threats to biodiversity and actions to mitigate their impacts are carried out throughout Australia. As part of this thesis an experiment is conducted to investigate the effects of the combined reduction of Fox and Rabbit activity on the small vertebrate community and Cats in the arid zone. The ubiquitous House Mouse, regarded as a pest due to the damage it causes in the agricultural zones when present in plague proportions, appears to present a minor threat to native biota (Watts and Aslin 1981; Coutts-Smith et al. 2007) but its impact on native small mammals may be underestimated. The potential threat emanating from the House Mouse is not directly assessed in the thesis but as a part of the small mammal community in the study area the House Mouse is part of the investigation and its potential impact on the small vertebrate community is discussed.

- 6 -

General introduction and study rationale

More recently increases in Goat and Camel populations cause additional concern in arid lands, a fact which is addressed as part of the discussion of the conservation of threatened species.

1.2.3 Climate change Climate change is a current challenge in conservation biology that the thesis does not directly address but nonetheless it aims to fill knowledge gaps to aid actions to address rapid climate change. Climate change is recognized as a major threat to the survival of species and the integrity of ecosystems worldwide (Thomas et al. 2004a; Thomas et al. 2004b). It has and will alter many of the environmental conditions that shape the distribution and abundance of species (Bereton et al. 1994). Range shifts induced by climate change are predicted to commonly span tens of kilometres by the middle of the next century (Kapelle et al. 1999). Conservation management therefore needs to aspire to preserve or restore conditions that ensure the greatest possible resilience of species to changing environmental conditions. Identifying and protecting important climate refugia, maintaining viable populations, maximizing intra-species genetic diversity, conserving large-scale migration and habitat connectivity, reducing other threatening processes as much as possible and protecting and restoring key large scale ecological processes are all measures that can help to achieve this (Mackay et al. 2008).

1.3 Conservation management

1.3.1 Legislation The central piece of environmental legislation in Australia is The Environment Protection and Biodiversity Conservation Act 1999 (the EPBC Act). It provides the legal framework to protect and manage nationally and internationally important flora, fauna, ecological communities and heritage places. Included within this framework is the conservation of Australia’s biodiversity and in particular nationally threatened species and ecological communities. Legal guidelines regarding the conservation management in National Parks also exist for each state. In NSW, the National Parks and Wildlife Act 1974 (the NPW Act) and the associated National Parks and Wildlife Regulation 2002 apply. The NPW Act requires the active management of National Parks in order to conserve biodiversity, maintain ecosystem function, protect geological and geomorphological features and natural phenomena and maintain natural landscapes.

- 7 -

General introduction and study rationale

While ensuring compatibility with the conservation of parks, it also requires the provision of opportunities for visitors’ recreation and education.

1.3.2 Natural heritage versus cultural heritage and visitor recreation According to the above-mentioned legislation, conservation managers are obligated to protect biodiversity and specifically threatened species (often a whole suite of threatened species with potentially very different conservation demands), within the restricted area of the conservation reserve and with typically a limited staff and budget. In the case of NSW National Parks, the attention of the managers, as well as the distribution of resources, need to be divided between the protection of natural heritage, cultural heritage, visitor recreation and education, and the general maintenance of the park. Science, including studies such as this thesis, plays a significant role in focussing efforts of conservation management to ensure the best outcome with limited resources.

1.3.3 Science in conservation The effective management of any ecosystem is dependent upon at least a rudimentary understanding of its major components, the key processes which underpin its function and its response to perturbation (Burgman and Lindenmayer 1998). In the conservation of individual species detailed knowledge of their ecology is the foundation to assess a species’ conservation status, identify threats and consequently compose and apply adequate conservation guidelines. The undertaking of scientific enquiry is required to establish the possible consequences of alternative choices among management options and is the basis for adaptive management (Walker 1998). Compared to the Americas or Europe the fauna in Australia remains little studied in great part due to Australia’s short history of science. This study makes a significant contribution to the knowledge of the fauna of arid areas which remain particularly understudied due to the challenges that the remoteness and desert environment pose to researchers.

1.3.4 Research and conservation in the arid zone The Australian arid zone and the organisms inhabiting it present particular challenges to research (Haythornthwaite 2007) and, as a consequence, to conservation. Desert areas are typically remote and difficult to access, as distances are long and many areas become inaccessible after rainfall. Extreme temperatures in summer provide harsh working conditions well above the human comfort zone.

- 8 -

General introduction and study rationale

In addition to these technical and logistical obstacles, variable rainfall patterns, which determine food resources, create extreme population fluctuations in many desert species and further complicate their study and conservation. Small mammals, and particularly rodents, show especially large fluctuations in their densities (Dickman et al. 1999; Letnic and Dickman 2005). Although they can provide good study opportunities on the rare occasions when they are at high population densities, their abundances are generally low and adequate sample sizes are hard to obtain. Many of these so called ‘boom and bust’ species are of conservation concern as they are extremely vulnerable to extinction at the low density phase of their population cycle due to stochastic environmental events and predation. Detailed knowledge of a species’ ecology is thus necessary to distinguish population declines that are due to environmental stochasticity from those caused by threatening factors.

Similar to rainfall, fires are a part of the ecology of many desert areas and can have significant impacts on small vertebrates (Southgate and Masters 1996; Letnic 2003a; Letnic 2003b; Letnic et al. 2004; Letnic and Dickman 2005). However in the eastern third of the arid zone and particularly in western NSW, fire does not represent a feature of the current ecological dynamic. Livestock grazing has reduced the plant response to rainfall, diminishing fuel loads across extensive areas, which prevents fires.

1.3.5 Focus of conservation actions Conservation can be addressed on several levels of ecological organisation: genes, individuals, populations, geographic variants, species communities or ecosystems. A complete management strategy ideally will operate at all levels (Falk 1990), but social and political practicalities usually result in individual species or ecosystems being the principal targets of management activities.

Most areas in Australia have experienced some form of environmental change through modern land use. Many reserves have been established on land previously used for Sheep or Cattle grazing. Conservation strategies for these reserves often involve the identification of environmental change and its causes, and the implementation of management practices to limit those changes or modify their effects (Caughley and Gunn 1996). In such cases conservation requires recovery and restoration actions in addition to other conservation actions. Recovery and restoration actions generally target

- 9 -

General introduction and study rationale the ecosystem and aim to improve the general state of the landscape in the hope of thereby indirectly increasing biodiversity and aiding in the conservation of threatened species. Often such a strategy is successful and actions used in ecosystem management are equally applicable and beneficial in individual species’ management.

1.4 Study rationale This study was conducted on a large protected area, Sturt NP (see Chapter 2 for details) where livestock removal, the strategic closure of artificial watering points, and the control of Foxes have been conducted as management actions in common with many other rangeland areas. Such management actions aim to improve the ecosystem by reducing grazing pressure and benefit a whole range of species, mainly small vertebrates, by reducing the pressure from Fox predation. However, the outcomes of such actions often remain undocumented and the benefits to relevant threatened species uncertain.

The design of species-specific conservation actions is often unachievable due to limited ecological information available on many species. The Dusky Hoping Mouse (Notomys fuscus) is an endangered species in NSW and has the highest conservation priority of any small vertebrate in Sturt NP. Yet very little is known about this rodent species, including basic aspects of its biology. Notomys fuscus is used in this thesis as an example to address some of the issues associated with small mammal conservation in the arid zone of Australia.

The thesis has three aims:

1. To investigate the outcome of more than three decades of livestock removal and strategic closure of artificial watering points (in this case bores) on a) vegetation parameters, b) arthropods, c) small mammals d) lizards and e) the populations of non-native mammals (Foxes, Cats, Rabbits and House Mice). 2. To assess the impacts of combined Fox and Rabbit control on a) small mammals b) lizards and c) the abundance of Cats. 3. To use the threatened Dusky Hopping Mouse as an example to illustrate how gains in knowledge of the ecology of a rare species can and will affect management actions.

- 10 -

General introduction and study rationale

The thesis aims to contribute to the improvement of biodiversity and threatened species management in Sturt National Park. The focus will be on management of small vertebrate communities with the Dusky Hopping Mouse used as a case study. It is hoped that extrapolating the results of this case study will aid in the conservation of other small vertebrates in the arid zone, a region that is of high conservation concern but which remains understudied.

1.5 Thesis structure In the preceding sections a general introduction and background to the thesis was given. This outlined the losses of biodiversity in Australia with a particular focus on small vertebrates in the rangelands, identified the key threatening factors affecting them and introduced management issues faced in the conservation of small vertebrates in the arid zone. This concluded with a statement of the aims and objectives of this thesis.

In Chapter 2 the study area is introduced, a timeline provided of the field work conducted and the general methods are described. The extensive topic of grazing disturbance and artificial watering points is covered over two chapters. A literature review regarding pastoralism in the rangelands and the impacts of grazing in the context of AWP is the content of Chapter 3. The impacts of livestock removal and closure of AWP on vegetation parameters and faunal characteristics are presented in Chapter 4. The experimental removal of Foxes and Rabbits and its impacts is the content of Chapter 5. Chapter 6 is concerned with the Dusky Hopping Mouse and its ecology and conservation. The thesis concludes in Chapter 7 with a synthesis of the most important findings and addresses issues that arise in the application of conservation actions.

- 11 -

Chapter 2: Study Area and General Methods

2.1 Study area

2.1.1 Description of topography, landforms and vegetation Research was conducted on the former Sheep station ‘Fort Grey’, which has been a part of Sturt NP since 1972. Sturt NP is located in the far-western corner of New South Wales, 330 km north of Broken Hill, near the small township of Tibooburra (population 80) (Figure 2.1). The park covers an area of more than 330,000 ha and comprises a variety of landforms typical of an arid and semi-arid environment, including Gibber or Mitchell grass plains, the Grey Range of ‘Jump Up’ Country, granite hills, riverine woodlands and stabilised red sand dunes. The country surrounding the park is still used for grazing Sheep and Cattle.

‘Fort Grey’ is situated in the far north-western corner of the Park, at the south-western edge of the Strzelecki dune fields on the border to South Australia and Queensland. Most of the land belonging to the former Sheep station extends west of the still extant Fort Grey Homestead (Figure 2.1). The dominant landscape feature in the area is parallel, vegetated and thus stabilized sand dunes of up to 15 m in height which alternate with interdunal areas of varying width and character. The interdunes can be rocky plains, clay pans, swamps or a combination of the three. Vegetation is varied and patchy, depending on the substrate. Most sand dunes feature a relatively dense permanent upper shrub cover, the most common species being Acacia ligulata (Sandhill Wattle), A. aneura (Mulga), Dodonea spp. (Hopbushes), Hakea leucoptera (Needlewood) and Atalaya hemiglauca (Whitewood). The lower shrub cover consists mainly of species from the genera Senna, Eremophila, Atriplex, Sida, Hibiscus, Enchylaena and Ptilotus. After sufficient rainfall ephemeral forbs and grasses are abundant. The interdunal areas are in general devoid of trees and shrubs, except for occasional small patches. The soil of the interdunes, which is mainly clay dominated, is often covered with small rocks and supports a diverse range of herbaceous vegetation, grasses, saltbushes (Atriplex, Rhagodia) and copperburrs (Sclerolaena). A particular

- 12 -

Study area and general methods feature of many interdunes is temporary canegrass (Glyceria ramigera) or lignum (Muehlenbeckia cunninghami) swamp communities with deep cracking clay soil.

Fort Grey Homestead

Tibooburra

Broken Hill Sydney

Figure 2.1: Location of sampling sites. Sampling sites (black dots) within Sturt National Park (main) and the location of Sturt National Park (shaded black) within NSW (bottom left) and Australia (bottom right).

2.1.2 Climate Consistent meteorological data for Fort Grey are not available and so data from Tibooburra, about 84 km east of Fort Grey, are presented. Tibooburra (Bureau of Meteorology Station No. 046037) receives a mean annual rainfall of 227.9 mm per year (averaged over 116.1 years) (Australian Bureau of Meteorology 2004) and is thus classified as an arid location (Leeper 1970). Rain can fall at any time of the year with a slightly higher occurrence in summer (Bell and Stanley 1991). Rainfall is highly variable and so many months pass without rain but occasional intense rainfall events can deliver a large percentage of the mean annual rainfall in a day. The mean (90 year average) daily maximum and minimum annual temperature of 27.4 and 13.8 °C,

- 13 -

Study area and general methods respectively, mask the extremeness of summer temperatures. One fifth (20.9 %) of days per year has temperatures higher than 35°C. Frost events are rare and temperatures lower than 2°C occur on only 8.7 days a year. The relative humidity in Tibooburra is low (mean relative humidity at 9 AM is 52 % and 31 % at 3 PM) (Australian Bureau of Meteorology 2004). Due to high temperatures and frequent winds, evaporation rates are high (2,800 mm per annum), exceeding monthly rainfall, particularly in summer, and even under wet conditions (Bell and Stanley 1991).

2.1.3 Rainfall prior to and during study period The Fort Grey area received well below average rainfall throughout the study period (2005 to 2008) and in most of the ten years preceding the commencement of this study (Figure 2.2). Since 1995 only two years, 1998 and 2000, had above average rainfall. Since 2001 the Fort Grey area has experienced re-current dry conditions. The rainfall during the study period between 2005 and 2008 was well below average. The first two years of the study were particularly dry with less than half of the average annual precipitation.

400

350

300

250

200

150

100

0 1995 1996 1997 1998 1999 2000 2001 2002 2003 2004 2005 2006 2007 2008

Figure 2.2: Annual rainfall (mm) measured at the Fort Grey Homestead. Precipitation in the ten years prior to the study and for the study years 2005, 2006, 2007 and 2008 in relation to the long-term (116 years) average (black line) of 228 mm. Data courtesy of the then Parks and Wildlife Service Tibooburra.

2.2 Reasons for selecting the study area Studies on small terrestrial vertebrates in the arid zone have been mostly carried out in Spinifex communities with little comparable work in portions of the arid zone with

- 14 -

Study area and general methods other vegetation types, such as western NSW. The Fort Grey area in Sturt NP was selected for this study because of its rich small mammal and reptile fauna, many of which are of conservation concern in NSW (i.e. Sandy Inland Mouse (Pseudomys hermannsburgensis) – vulnerable, Central short-tailed Mouse (Leggadina forresti) – vulnerable, Dusky Hopping Mouse (Notomys fuscus) – endangered and the Stripe-faced Dunnart (Sminthopsis macroura) – vulnerable (Threatened Species Conservation Act 1995)). Due to its remoteness the area also supported largely uncontrolled and therefore presumably large Fox, Cat and Rabbit populations which made it an ideal location for a manipulative removal experiment. A large (ca. 700 km2) and readily accessible area of very uniform habitat within the park allowed experimental and survey replication. The land is typical of much of far-western NSW, north-western South Australia and southwestern Queensland, giving results from this study a wider relevance. Furthermore, the management at Sturt NP encouraged on-park research and provided great support in the development and implementation of this project.

2.3 Sampling sites Eight sampling sites were established along the existing service tracks to the west of the Fort Grey Homestead (Figure 2.3). To investigate the residual impact of Sheep grazing and to account for this as a potential confounding factor in the investigation of Fox and Rabbit removal, the sites were chosen to represent areas of relatively high and low historical disturbance through grazing. Four sites were situated close to former bores and the other four half-way between the bores. The bores were the closest historical or extant watering point to the undisturbed sites and so the undisturbed sites were at least four kilometres from every other watering point past or present. At each site an identical setup for monitoring was installed on three parallel neighbouring dunes. The same set of sites was used for the investigation into the residual effect of AWPs and the fox/rabbit experiment.

A Travelling Stock Route (TSR) traverses the study area between site one and two. In Australia, the TSR is an authorised thoroughfare for the walking of domestic livestock such as sheep or cattle from one location to another, either to the market or to better pastures. Stock routes were important mainly in the early days of pastoralism but have become superseded since the development of the road network and the possibility of transporting large numbers of livestock by truck. The TSR is not marked in any way nor

- 15 -

Study area and general methods is it fenced. No reliable information is available on the time, frequency and intensity of its use. Grazing pressure on the TSR would have been short (livestock are ‘travelling’) but very intense early in its history, with its use declining over time and with better development of the area. Since the area became part of Sturt National Park in 1972 the TSR has definitely not been used. The edges of the TSR are more than a kilometre away from the closest study sites. It appears unlikely that the presence of the TSR would have impacted on the study results. Even so, potential impacts of the TSR have been discussed where relevant.

2.4 General methods

2.4.1 Timeline Field work for this thesis extended over a period of more than three years and was conducted between July 2005 and October 2008 (Table 2.1). Between January 2006 and April 2008 regular surveys (‘sessions’) including ground cover measurements, small vertebrate trapping and estimation of Fox, Rabbit and Cat activity, were carried out four times a year i.e. in January (summer), in April (autumn), in July (winter) and in October (spring). The surveys were conducted seasonally to account for variation over the course of a year and thus obtain more reliable results. The data of the sessions one to four were used to investigate the residual effects of grazing and AWP. The Fox removal experiment was conducted on the same experimental sites and so the data collected to investigate the effect of grazing and AWPs also served as part of the benchmark (i.e. pre-treatment: session one to five) for the investigation of the Fox experiment. The data collected in later sessions represented the post-treatment data. These sessions represented fixed periods in the field work schedule and all other work such as the control of Foxes and Rabbits was conducted in the time between those sessions.

Field work towards the third aim of the thesis, the investigation of the ecology of N. fuscus was disjunct. An initial survey consisting of transect walks was carried out in July 2005 to establish if and in what numbers Notomys fuscus were present in the study area. The survey did not yield any hopping mouse evidence and thus further work on this aspect of the thesis was postponed. The main part of the work on Notomys fuscus was carried out during a continuous period (January to October 2008) towards the end of the study, when N. fuscus numbers were high.

- 16 -

Study area and general methods

QLD

Site 1 SA NSW (Devis Bore)

Site 2 Tr a Fort Grey Site 3 ve Homestead lli (Collins Bore) ng s to ckr ou te

Site 4 Site 5 (Watties Bore)

Site 8

Site 6

Site 7 (Yebna Bore)

Figure 2.3: Map of study area and sampling sites in the western part of Sturt National Park.

Table 2.1: Overview of the timing of field work and the contribution of field work periods/sessions to the three topics covered within the thesis.

Year Session Month Season AWP`s Fox/Rabbit N. fuscus 2005 July/August Winter x 2006 1 January Summer xx 2April Autumn xx 3July Winter xx 4 October Spring xx 2007 5 January Summer x 6April Autumn x 7July Winter x 8 October Spring x 2008 9 January Summer xx 10 April Autumn xx May to x October Winter

- 17 -

Study area and general methods

2.4.2 Weather recording A logging weather station (WeatherMaster 2000, wireless; Environdata) was set up at the Fort Grey homestead and continuously recorded data on air temperature, relative humidity, wind speed and direction, solar radiation, evaporation rate and precipitation. In contrast to all the other parameters, rainfall varies highly even on a small spatial scale. To record site-specific precipitation data, each site was equipped with a logging rain gauge (Odyssey TM Tipping Bucket rain gauge recording system, Dataflow Systems), positioned on the top of the central dune at each site from February 2006.

2.4.3 Data manipulation and analysis All data were entered into either Microsoft Excel (Excel 2003, Microsoft ®, USA) or a Microsoft Access database (Microsoft Access 2003, Microsoft ®, USA). The data were manipulated and examined using the afore-mentioned programs or SPSS Statistics for Windows ver. 17.0 (SPSS Inc. 2008). SPSS was used for all statistical analysis except for the performance of permutated multivariate analyses of variance (PERMANOVAs) which were done using PRIMER ver. 6.1.12 (PRIMER-E Ltd, 2006) in conjunction with the PERMANOVA+ add-on ver. 1.0.2 (Anderson 2001; Anderson 2005; Anderson et al. 2008)). PRIMER was also used to calculate community indices (species diversity, evenness and richness) and to generate non-parametric multi-dimensional scaling plots (MDS plots) for graphic illustration of potential clustering of similar sites.

Variables were considered to have explanatory power if the two-tailed probability fell below 10 % (this was considered a ‘trend’) but were only truly considered significant where the two-tailed probability fell below 5 %. In the case of the PERMANOVAs the p-values given are Monte Carlo asymptotic p-values. The number of unique permutations in the analysis was generally quite small and so the Monte Carlo p-value is recommended (Anderson 2005). As a consequence of the calculation of the pseudo F-ratio in unbalanced designs (for example as used in Chapter 5), the degrees of freedom are not always whole numbers. However, this does not affect the determination of the correct p-values (Anderson et al. 2008).

Throughout this thesis the Shannon Diversity index (also called Shannon-Wiener or Shannon-Weaver Index) and the Pielou’s index for evenness and species richness (number of species) were used to describe the floral and faunal communities. The

- 18 -

Study area and general methods advantage of the Shannon Diversity index is that it takes into account the number of species and the evenness of the species. The index is increased either by having additional unique species, or by having a greater species evenness. The Pielou’s evenness index is based on the Shannon diversity index and was used for consistency. All indices were calculated using the DIVERSE functionality in PRIMER.

- 19 -

Chapter 3: Review of the Effects of Grazing and Artificial Watering Points in the Australian Rangelands

3.1 History of pastoralism in the rangelands Settlement closely followed the expeditions of the early explorers. Pastoralism in the rangelands was well established in most parts of Australia by the middle of the 19th century (James et al. 1999). Almost all of western New South Wales, the area of this study, was included in Pastoral Runs by 1880 (Green 1989). In those early days, the pastoral industry relied on natural watering points that reliably held water for months after rainfall and were large enough to be able to sustain a large number of animals which had a need to drink regularly. Those watering points were sparse and thus grazing of livestock was largely restricted to the immediate surroundings of major waterways, except in good conditions, when stock could graze away from them. The sinking of wells and bores, and the construction of dams that began in the 1870s, enabled the stock runs and settlements to extend permanently beyond the rivers. By the 1950’s large numbers of watering points had been established. Pastoralism flourished and extended to all areas that had at least some vegetation that could sustain livestock (James et al. 1999).

In the 1890’s, when only few watering points existed, stocking rates were unsustainably high, as pastoralists ran as many Sheep as possible within the limited areas that could be grazed. Stocking levels in the western Division of NSW for example were as high as 15.5 million Sheep equivalents in 1887 following a period of high rainfall in the 1880s, and remained close to these numbers for a decade before plummeting to five million in 1902 as a result of overstocking, drought and high Rabbit numbers (Green 1989). Since then stock numbers have rarely exceeded eight million and are currently low due to the effects of prolonged drought and weak markets for wool. The density of watering points, on the other hand, is so high that few areas exist that are more than 10 km from water and have thus remained relatively unaffected by grazing of domestic livestock (Landsberg and Gillieson 1996).

- 20 -

Review of the effects of grazing and artificial watering points in the Australian rangelands

3.2 Overview of impacts of livestock grazing and artificial watering points Enormous changes in the landscape, in the edaphic environment, and in the flora and fauna were brought about by pastoralism. Fleischner (1994) has summarized the ecological costs of livestock grazing as follows: 1. Alteration of species composition of communities, including decreases in density and biomass of individual species, reduction of species richness and changes in community organization 2. Disruption of ecosystem function, including interference in nutrient cycling and ecological succession 3. Alteration of ecosystem structure, including changes in vegetation structure, contributing to soil erosion and decreasing the availability of water to biotic communities The latter two costs result in a loss of landscape functional integrity, which is the intactness of natural vegetation and soil structural patterns and the processes that maintain these patterns (Ludwig et al. 2004). If the functional landscape integrity is lost, changes to species composition and communities follow (see 1).

In the following a review of the effects of livestock grazing in combination with AWP is given. AWP are often considered as a permanent source of water and the term ‘permanent watering point’ appears frequently in discussions about this topic, but obviously dams dry up or are breached and bores collapse and fail. Thus watering points should more aptly be referred to as reliable rather than permanent water sources, a terminology which has been adopted throughout this thesis. The focus in this review is on those variables that are investigated in this study, namely characteristics of vegetation, the invertebrate, small mammal and lizard assemblages and the abundance of Foxes, Cats and Rabbits. Abiotic or edaphic effects are introduced briefly as they have significant flow-on effects on the vegetation and those in turn have influences on faunal distribution and densities.

3.2.1 Direct and indirect effects of artificial watering points AWP have profoundly contributed to changes in the landscape, both directly and indirectly. These changes have affected native as well as non-native flora and fauna. Direct effects of artificial water sources on native flora and fauna include: 1) the development of wetlands that support native plants and animals, 2) the expansion of the - 21 -

Review of the effects of grazing and artificial watering points in the Australian rangelands geographic range and an increased abundance of native animals which need to drink regularly, and 3) the possible expansion of the breeding ranges of invertebrates that require water for some stage of their life cycle (James et al. 1999). These direct effects of artificial water sources are predominantly a result of earthen dams and their large amounts of standing water. To reduce evaporative water loss and cool hot artesian waters, bores typically pump into closed metal tanks that pipe water to one or more livestock troughs on demand (e.g. through float valves). In some areas in the Great Artesian Basin, bore water has been reticulated through open channels or feeds small wetlands (e.g. mound springs) but neither strategy applied to the bores on the western side of Sturt NP which followed a conventional pump (windmill), holding tank, pipe and trough system. The focus of the first part of this project is on the effects of water- focussed herbivore activity, an indirect effect, which can be more accurately studied at bores without the potential confounding influence of the direct effects. The direct effects will thus not be further considered.

The major indirect effect of AWP is as a source of drinking water for domestic stock, native and feral herbivores and predators. They become a focal point for fauna activities which results in water-focussed grazing and associated activities, and predation (James et al. 1999). Impacts of grazing animals are highest near watering points and decrease with distance from the water for two reasons: 1) the area available to graze increases with distance to the focal point resulting in a reduction in the density of stock, and 2) stock have to drink regularly so they are limited in how far they can travel from water. The spatial distribution of such impacts that extend from watering points has been termed the ‘piosphere’ (Lange 1969). Not all piospheric impacts are immediately related to grazing animals but may also be a function of focused competition and predation.

The activities of grazing animals can either have a direct impact (mainly consumption of plants or plant parts) or an indirect or environmental impact (trampling and defecation). These two groups of processes are not independent and interact considerably, as changes to the soil affect the vegetation and vice versa. Furthermore these changes to soil and vegetation (important microhabitat variables) have significant flow-on effects and may change the distribution and/or abundance of faunal species. These again may in turn influence vegetation and soil parameters.

- 22 -

Review of the effects of grazing and artificial watering points in the Australian rangelands

3.3 Impacts on the edaphic environment The causes of most edaphic changes associated with livestock activity are trampling and increased defecation. These changes include the soil surface condition, hydrological features, soil chemistry and biological activity which in turn influence the stability and potential productivity of the area (Tongway et al. 2003) and ultimately landscape function.

In contrast to the soft soles of native herbivores, Sheep and Cattle have hard hooves and exert greater foot pressure than native macropods (Bennett 1999). Trampling results in increased soil compaction in lower soil layers (Andrew and Lange 1986), removal of ground cover and decreased water infiltration levels. Compaction may also impact on the populations of soil biota resulting in further impacts on structure and infiltration. Erosion rates in the arid and semi-arid rangelands have significantly multiplied since European settlement and the introduction of livestock grazing (Noble and Tongway 1986; Thrash 1997). For example, Wasson and Galloway (1986) estimate a 50-fold increase in erosion rates for the area west of Broken Hill (far western NSW). Greater erosion has resulted in increased amounts of bare ground and pulverization of the top soil layer. The removal of cryptogamic crust cover and other plants sensitive to trampling also increases the amount of bare ground causing more erosion. The cryptogamic crust is not only important as a soil stabilizer (Eldridge 1993; Eldridge and Greene 1994b; Eldridge and Greene 1994a) but through its nitrogen-fixing cyanobacteria it contributes significantly to nitrogen fixation in desert ecosystems (Whitford 2002) and thus aids in the mineral nutrition of vascular plants (Belnap and Harper 1995).

Increased dung and urine levels lead to changes in soil nutrient levels (Tolsma et al. 1987; Turner 1998a). Dung accumulates in areas which livestock frequent such as along livestock pads, and under trees or near other forms of shelter and in the immediate surroundings of watering points (Hilder and Mottershead 1963). By picking up nutrients while feeding and depositing nutrients with their droppings livestock activity leads to a significant redistribution of nutrients in the landscape. Trampling and indirect effects of soil mobilization through plant cover removal and increased erosion also leads to a change in the depth profile of the nutrients (Tongway et al. 2003).

- 23 -

Review of the effects of grazing and artificial watering points in the Australian rangelands

Flow-on impacts of these edaphic changes are manifold and complex. For example, a change in nutrient levels will change the vegetation composition, favouring high nutrient tolerant plants and/or the plants’ nutrient content (Tolsma et al. 1987) which in turn affect its potential for reproduction or its palatability.

3.4 Impacts on vegetation Grazing, trampling and defecation by livestock can affect the vegetation biomass and cover as well as the richness of the local vegetation community. As a result of high herbivore densities and their grazing activity, plants are physically damaged and a large amount of biomass is removed. This results in an increase in the extent of bare ground, especially in dry times when soil moisture is low. Excessive trampling may contribute to this by damaging, crushing and uprooting plants, causing die-off or/and preventing recruitment. Continual selection of palatable plant species by grazing animals favours increased dominance by less palatable plant species (Moretto and Distel 1999) which affects the composition of plant communities as does the differential sensitivity to grazing between plant species (i.e. their ability to deter, escape or tolerate herbivory). Associated with the change in plant species within the communities is a change in plant attributes. Landsberg et al. (1999a) were able to identify sets of plant traits related to morphology, grazing persistence and regeneration that were typical of lightly and heavily grazed sites, respectively. Seeds may be redistributed and accumulated in livestock dung, and trampling at moderate levels may create small basins for water and seeds to catch, thereby potentially changing vegetation cover as well as species composition. Indirectly, vegetation change can result from changes to the soil pH and the redistribution of nutrients as a consequence of livestock defecation.

The amount of literature that exists on the effects of large herbivore activities on a wide array of vegetation parameters in Australia and worldwide is enormous and has been reviewed and summarized on several occasions (e.g. James et al. 1995; James et al. 1999; Landsberg et al. 1999b). Due to the complexity of grazing as a disturbance, its effects on certain parameters are often inconsistent. Climate, soil, vegetation type, water distribution, evolutionary history of grazing and perhaps most importantly herbivore type (or types, in ecosystems with a variety of herbivore species) and their density have been identified or been suggested as factors that significantly influence impacts. In addition several authors have suggested that the effects of rainfall and spatial

- 24 -

Review of the effects of grazing and artificial watering points in the Australian rangelands heterogeneity under certain (particularly arid) climates can, at particular times, be more influential than any herbivore impact confounding results (Turner 1998a; Turner 1998b; Turner 1999; Chase et al. 2000). Nevertheless, heavy grazing by herbivores appears to have some general effects under a wide variety of circumstances. These include an increase in bare ground, exotic weeds, annual species, grazing tolerant species and plants that are unpalatable due to poison, spines or an acrid flavour. In concert with these increases are a decrease in palatable perennial species, especially palatable grasses and shrubs and those plants vulnerable to trampling.

3.5 Impacts on fauna In comparison to the extensive amount of literature that exists on the effects of grazing on the vegetation, research reporting on the impacts of grazing on invertebrates and terrestrial vertebrate assemblages is relatively scarce (Woinarski 1999). Some of this is due to the fact that faunal sampling is generally time intensive and sufficiently large datasets (especially in the case of terrestrial vertebrates) require larger plots than those typically used for the assessment of pastoral impacts on flora.

3.5.1 Impacts resulting from grazing activities Animals are mainly indirectly affected by grazing, through changes in habitat structure and the availability of resources. Thus their responses are often less pronounced than those of plants, and may go undetected. The mobility of animals and their potential to reproduce prolifically in a short timeframe allows them to respond quickly to environmental factors which may be more important than or overlay any herbivore impact. Such influences and the resulting variation in the observed responses to grazing make generalizations difficult. In fact no generalizations seem possible for invertebrates. James et al. (1999) in his review of impacts of grazing and AWP on biota, draws the conclusion that the most obvious generalization is that abundance, and species composition and richness of invertebrates respond to the season of sampling and changes in the architecture and species composition of vegetation that result from grazing. In contrast, for terrestrial vertebrates of all continents some general trends in the responses to grazing regimes are evident: the abundance of terrestrial vertebrates is often higher in lightly or ungrazed areas, the species richness is similar between ungrazed and grazed areas and the assemblage composition changes in the order of 10-20 % between grazed and ungrazed areas with differences most obvious when grazing affects the structural architecture of the environment (James 2003).

- 25 -

Review of the effects of grazing and artificial watering points in the Australian rangelands

Even though generalizations can be useful for some purposes, the large number of threatened species amongst vertebrates and the concern for their conservation requires species-specific information rather than generalizations.

3.5.2 Impacts related to artificial watering points In an active pastoral landscape stock watering points can have considerable direct and indirect impacts on faunal species. Bores and dams can provide the moisture necessary for the reproduction of certain insects, such as dragon-flies or mosquitoes, allowing them to exist and persist in areas that might otherwise be devoid of them, particularly in semi-arid and arid environments. Dams also provide important habitat for a number of species, for example (migratory) waterbirds. Apart from this AWP benefit a number of species, both native and alien, by providing drinking water. While watering points may not be overly important to Fox, Cat and Rabbit populations under good environmental conditions they might support populations in drought, preventing a population crash in the vicinity of water supplies and thus preserving a source population for the post- drought recolonization of more water-distant areas.

In addition to providing drinking water AWP’s may also represent a focal point for resources and thus foraging activities of species. Predators frequent watering points due to the better foraging opportunities or hunting success that they offer (Harrington et al. 1999), a fact well known in Africa (Owen-Smith 1996) but little researched in Australia. The diet of Foxes is known to include a high percentage of carrion and this is often found near watering points especially during drought conditions. Even though feral Cats themselves generally do not rely on drinking water, many birds, which represent an important part of their diet, do. As a result of high hunting success, the populations of both of these non-native predator species may be larger and/or more stable around watering points and as a consequence increase hunting pressure on the prey population in the surrounds of AWP.

Resources for other species can accumulate around watering points (e.g. seeds in dung and dead animal stomachs) and a higher abundance of certain invertebrates may result in higher abundances of small insectivorous vertebrate species around AWP. For example, termites the preferred food item of many Australian lizards (James 1991), may

- 26 -

Review of the effects of grazing and artificial watering points in the Australian rangelands benefit from the accumulation of dung as a resource and thus be more abundant around AWP (Weir (1971) cited in James et al. 1999). During dry conditions, when supplementary feeding of livestock becomes necessary, the feed is often provided in close proximity to the stock watering source. Rabbits make use of those food sources (personal observation) and House Mice are likely to do the same, considering that lucerne and hay often contain seeds. Stock feeding sites and thus AWP, may have been the focal points of the introduction of House Mice. Undoubtedly House Mice populations were and are highest in and around homesteads and feed sheds with an ample and reliable food supply. Due to their small size they could have travelled hidden in food bales or containers as passengers to where the food was unloaded.

Whereas food and water supply around watering points may be favourable for a number of animals, the higher abundance and/or frequency of visitations of predators, such as Foxes and Cats, around watering points is not. Depending on which has the bigger effect at a given time, improved resources or increased predation risk, certain species may be more or less abundant around AWP than in water-distant areas.

The direct effects of watering points are more relevant to dams than to bores and associated troughs. With the closure of the watering points, in particular bores and troughs, the direct effects of watering points disappear within a very short timeframe. Therefore few of the above described direct effects need to be considered in the context of this study. What is of interest though is whether the previously established populations of non-native species around the watering points can persist over time even after their closure.

3.5.3 Invertebrates Invertebrates are widely regarded as powerful monitoring tools in environmental management because of their great abundance, diversity and functional importance, their sensitivity to disturbance and the relative ease with which they can be sampled. Some argue that given their overwhelming dominance, no biodiversity monitoring programme can be considered credible without the inclusion of invertebrates (Taylor and Doran 2001). In particular, ants are regarded as useful to assess disturbances in Australia (Hoffmann and Andersen 2003), but spiders (Churchill 1997), grasshoppers (Andersen et al. 2001) and moths (Kitching et al. 2000) have also been suggested as

- 27 -

Review of the effects of grazing and artificial watering points in the Australian rangelands indicator groups. However, the combination of the multitude of invertebrate species, the often difficult procedures for species identification, the large numbers of specimens and the extreme temporal variability of invertebrate assemblages complicate their investigation (Andersen et al. 2004).

Livestock grazing activities can have a direct effect on invertebrates, for example ant biomass (Majer et al. 2004), if it involves substantial disruption to the soil or other nesting sites. The major effects are more typically indirect through alteration of habitat characteristics, including vegetation characteristics, soil compactness and microhabitat temperature and humidity, which all impact on invertebrates and their food supply (Whitford et al. (1995) and Loftin et al. (2000) cited in Hoffmann and Andersen (2003); (Debano 2006)).

Some studies on some invertebrate species, or higher taxonomic levels and groups of invertebrates, have shown significant responses to grazing ((Debano 2006; Dennis et al. 2008) and reviewed by (James et al. 1999) (all invertebrates) and (Hoffmann and Andersen 2003) (ants)). Hoffman and Andersen (2003) reviewed the responses of ants to disturbance effects and found that ant species, species-groups and functional groups all show a variable response to disturbances. They suggest three confounding factors that are likely to lead to such variation: a species’ response to disturbance is 1) likely to vary with habitat (whether it lives in its optimal or suboptimal habitat) 2) unlikely to be linear and 3) a function of the time since disturbance (the immediate response may differ from the longer term response). These observations and conclusions are likely to be true not only for ants but also for other invertebrate groups. It is thus not surprising that results vary from study to study and location to location and no consistent trends on any taxonomic level are apparent. Since generalizations seem impossible to draw, the invertebrate community needs to be investigated on a case by case basis.

3.5.4 Small mammals and Lizards Soil disturbance through trampling by stock may collapse burrows, cause direct mortalities, or disturb the habitat of litter-dwelling species (Busack and Bury 1974; Ehman (1980) cited in Read 2002). Habitat preferences, habitat partitioning between species and differences in small vertebrate assemblages are usually directly related to differences in structural, thermal and edaphic variables (Brown and Lieberman 1973;

- 28 -

Review of the effects of grazing and artificial watering points in the Australian rangelands

Cogger 1974; Hallett 1982; Kotler and Brown 1988; Downey and Dickman 1993; Morris 1996; James et al. 1999; Magarey 1999). Changes to such factors are likely to go along with different grazing intensities and so it can be assumed that lizard and small mammal assemblages will undergo spatial distribution changes when grazing impacts are apparent. As a result of the habitat changes, predator pressure may increase since open habitat generally ensures higher hunting success. Also, invertebrate communities, which are the major food source for most lizards and many small mammals, may be affected by habitat change and in turn influence the vertebrate community.

In line with the generalizations for vertebrates by James (2003) and results from North America (Busack and Bury 1974; Jones 1981; Bock et al. 1990; Kutt and Woinarski 2007), Australian lizards can be less abundant and less species rich in heavily grazed areas compared to lightly or un-grazed areas (James et al. 1999; Magarey 1999; Frank 2002; Woinarski and Ash 2002; James 2003). Interestingly, James (2003) found differences between grazing regimes for abundance and species richness of diurnal lizards, but no difference for nocturnal species. Similarly Woinarski and Ash (2002) report varying responses for geckos and skinks with geckos being more abundant, but skinks less abundant on grazed sites. However, results of other studies have supported the general opinion that reptiles tend to be resilient to grazing impacts or show only limited detectable responses (Smith et al. 1996; Read 2002; Beutel et al. 2003).

As for lizards, studies of mammal responses in North America have generally found higher abundances of ground-dwelling mammals on ungrazed plots than on grazed plots (e.g.Grant et al. 1982; Hanley and Page 1982; Bock et al. 1984; Putman et al. 1989; Heske and Campbell 1991; Valone and Sauter 2005). Species richness was either not affected or was lowered by grazing (Putman et al. 1989; Woinarski and Ash 2002). Jones et al. (2003) hypothesize that variation in percent ground cover dictates the relative abundance of various rodent taxa in grass and shrublands in Arizona, USA. In more arid ecosystems with erratic rainfall patterns microhabitat variables are likely to be of lesser importance for small mammals whereas food resources (i.e. seeds, invertebrates and smaller vertebrates) are probably the greater determinant of small mammal abundance and distribution. The generally great mobility of desert small mammals (Read 1984; Dickman 1993) and their often high reproductive potential allow them to respond quickly to changes in resources. Strong responses to environmental

- 29 -

Review of the effects of grazing and artificial watering points in the Australian rangelands change and the changes in available resources may mask differences due to grazing impacts and explain the general inconclusiveness of results investigating the impacts of community parameters such as diversity, richness and abundance in the few studies including Australian arid zone small mammals (Frank 2002; Montague-Drake 2003; Dowle 2004). Given the variable and complex habitat requirements, food preferences and foraging strategies of individual species, the response of lizard and small mammal species to grazing induced changes is unlikely to be uniform. Some species will suffer from grazing impacts whereas others may directly benefit from habitat changes induced by grazing or otherwise benefit from the decline of species with which they may compete. Others still will remain unaffected by grazing impacts and populations will stay stable at different levels of disturbance. Species that prefer open habitats with little ground cover and relatively high surface temperatures (certain lizard species) might benefit from grazing whereas species preferring denser vegetation cover or those associated with shrub leaf litter are disadvantaged. Species composition tends to become more dominated by generalists at the expense of specialists under heavy grazing (Smith et al. 1996). Read and Pickering (1999) found that the ratio of open-habitat favouring agamids to skinks is typically higher in disturbed local environments than in low impact areas. Disturbed areas also typically support higher numbers of the introduced House Mouse compared with unaffected regions where native rodents are typically more abundant (Read and Pickering 1999).

3.5.5 Non-native fauna The non-native fauna considered in this section are the Rabbit, the Fox, the Cat and the House Mouse. Pastoralism has affected these species mainly through the provision of a reliable supply of water in the form of stock watering points and very little through the direct or indirect effects of the grazing activities of livestock. Access to water and benefits associated with watering points such as increased resources and/or hunting success, have arguably (depending on the species) benefited these pest species greatly by permitting and/or facilitating the establishment and persistence of their populations in otherwise potentially unsuitable areas. With or shortly after the closure of the water resource the benefit of water availability ceases to exist. Nevertheless populations of the non-native species may persist over time as a result of permanent changes in the vegetation structure due to grazing. Areas with more open vegetation and greater

- 30 -

Review of the effects of grazing and artificial watering points in the Australian rangelands amount of bare ground allow maximum movement for Cats and Foxes and provide minimum concealment for prey and may thus offer greater hunting success. Of these predators, the more mobile Fox would gain greater advantage than the more stealthy Cat.

3.6 Grazing management, management of artificial watering points and rangeland recovery Two fundamentally different types of watering points are used on pastoral properties: earthen dams and bores. The functionality of earthen dams requires close association of the tank to an area of natural water flow after rain, such as a creek, to provide enough water influx. Those areas are naturally more productive than most other parts in arid landscapes and can be very variable depending on the size and extent of the water- providing creek. Bores on the other hand do not rely on particular local characteristics since they can be erected wherever it is possible to drill deeply enough to reach the artesian water supply.

Typically grazing management policies advocate an even distribution of watering points to increase the carrying capacity of the land by accessing waterless zones and evenly spreading out the grazing pressure. Today, most Sheep-grazing focuses on areas that are within two to three kilometres of water and most Cattle-grazing areas are within six km of water (James et al. 1999). In the western division of New South Wales, where Sheep- based enterprises dominate, almost none of the land is beyond six kilometres from reliable water (Fisher et al. 2004). This figure is consistently higher in the more extensive Cattle-producing areas and in areas with a large proportion of non-pastoral land with 20 % of land more than six kilometres from water (Fisher et al. 2004). James et al. (1999) estimate that in inland Australia very few places remain that are further than 15 km from any AWP and have thus relatively little exposure to grazing pressure by livestock.

Strategies to achieve more even grazing of the landscape, as desired for pastoral activities, are likely to have detrimental effects on biodiversity. James et al. (1999) reported that about 15-38 % of species showed declines in response to grazing. These negative effects are particularly apparent in the piosphere where the landscape

- 31 -

Review of the effects of grazing and artificial watering points in the Australian rangelands experiences especially heavy grazing effects. Widespread, even grazing pressure has lead to the destruction of refugia for so called ‘decreaser’ species, a loss in water independent species, and an overall loss of species diversity. Ecological theory implies that heterogeneity in the landscape is important to preserve biodiversity. In the case of grazing, spatial and temporal variation in the disturbance regime will create situations where different species are advantaged at some time or in some place. Thus protecting some land from frequent grazing is important and is a key strategy for maintaining grazing sensitive species in pastoral lands. Approximately 10 % of the landscape should remain ungrazed or only lightly grazed (Fisher et al. 2004). To increase the level of spatial variation of grazing, the strategic closure of AWP has been proposed and implemented as a tool in the management of biodiversity conservation. Water should be supplied in a way that maintains the full suite of biodiversity and natural processes by having water-proximate and water-remote areas (James et al. 1999; Thrash and Derry 1999). This has been applied in conservation areas in the rangelands, such as National Parks, which in many cases were established on previous pastoral leaseholds, where the closure of AWP has been implemented as a tool for biodiversity conservation.

3.6.1 Rangeland recovery It has been well established that detrimental impacts result from intense grazing pressure especially around watering points. However, there is little consensus on the issue of the permanency of the disturbance, and if and to what extent changes produced by heavy grazing can be reversed. The stability of an ecosystem, defined by the resistance (the ability of the system to remain the same while external conditions change) and resilience (the ability of the system to recover after it has been disturbed) of a system, determines whether a recovery is possible and in what timeframe (Gunderson et al. 2002 in (Mayer and Rietkerk 2004). The resilience of ecosystems is likely to be lower in ‘fragile’ landscapes where water and/or nutrients limit plant growth and persistence (Dorrough et al. 2004), as is the case in the semi-arid and arid zone of Australia.

The potential for recovery from grazing disturbance and the speed of the recovery process is influenced by a multitude of factors. These include climate (change), time since de-stocking, historic grazing intensity and the resulting severity of the pastoral impact, the intensity of the remaining grazing pressure by native and non-native animals

- 32 -

Review of the effects of grazing and artificial watering points in the Australian rangelands after the removal of stock and the environmental conditions, especially the amount of precipitation received, during the recovery process. Several conceptual models to help explain and predict recovery outcomes have been generated and have recently received considerable attention in the literature (e.g. Bestelmeyer et al. 2003; Briske et al. 2003; Stringham et al. 2003; Briske et al. 2005; Vetter 2005; Bestelmeyer 2006; Briske et al. 2006; King and Hobbs 2006). Some of these models hypothesize transition breakpoints and thresholds of irreversibility which, when crossed, result in the system remaining in an alternate stable state or better dynamic regime (after Scheffer et al. 2001). These are often characterized by the dominance of particular species, trophic structures, energy flows and internal feedbacks that maintain biotic and abiotic patterns within a distinctive range. Those characteristics may be very different to the pre-disturbance system. Regimes can change over time because of ecosystem succession (the progression of community types after a disturbance from early to mature stages), changes in environmental conditions, and the continuing evolution or extinction of the species that are present.

The hope that a total recovery resembling pre-European conditions is possible in the Australian rangelands is unrealistic. A comprehensive study and subsequent series of papers investigating the degradation and recovery processes in arid grazing lands in central Australia suggest a permanent change mainly due to geomorphic stratification which was associated with resource loss and decoupling of water and nutrients (Friedel et al. 2003; Sparrow et al. 2003; Tongway et al. 2003). There are also strong cultural factors that come into play about ‘nativeness’ (Trigger et al 2008). An example of this is the Dingo within the Barrier Fence. To the pastoralists it is their worst pest (Claridge and Hunt 2008) whereas to many conservation biologists it has become a keystone species to a return to ‘nativeness’ (e.g. Glen et al. 2007).

- 33 -

Chapter 4: Residual Effects of Grazing and Artificial Watering Points on Flora and Fauna in Sturt National Park

4.1 Artificial watering points in Sturt National Park Historically the area now encompassing Sturt NP had a high density of watering points and as is typical of Sheep grazing country, 90 % of those were artificial. Of 97 AWP, 36 % were bores and 64 % were earthen tanks (varied historical information held by DECC, Parks and Wildlife Division, Tibooburra Area). Today, most bores (90 %) and many tanks (50 %) are no longer operational, due to either equipment or structural failure over time, or deliberate action undertaken by the then National Parks and Wildlife Service. Rendering all non-essential tanks and artesian bores ineffective is in accordance with the plan of Management for Sturt NP (NSW National Parks and Wildlife Service 1996). A number of selected AWP have been retained for park management and wildlife appreciation.

Similar to the other former Sheep properties that are today part of Sturt NP, ‘Fort Grey’ was run as a Sheep grazing property for more than 100 years. When it was taken over by the then National Parks and Wildlife Service in 1972 the livestock was removed. Apart from the homestead and its bore, yards and the house paddock, little of the pastoral infrastructure has been maintained. All artesian bores at ‘Fort Grey’ were either left in disrepair or were rendered dysfunctional and all but three earthen dams were altered to prevent water influx or the retention of water.

4.2 General Aim The impacts of livestock grazing have been extensively researched and documented, but relatively little is known if, and to what extent, the floral and faunal communities recover after cessation of livestock grazing in the Sheep rangelands (sensu Caughley (1987)) of Australia. Thus, the aims of this chapter are: 1) to investigate if residual effects of past Sheep grazing activity are detectable after more than 30 years of

- 34 -

Residual effects of grazing and artificial watering points on flora and fauna management for conservation of the native flora and fauna; and 2) how effective the closure of AWP and removal of livestock is as a management tool for the conservation of small terrestrial vertebrates in the sand dune ecosystem of Sturt NP.

This study cannot determine if a pre-pastoral state has been recovered (the historical state would be highly speculative in any case due to very limited historical records and observations), but it can determine if sites with different grazing history are still distinguishable. Depending on the differentiating variables and the amount of separation, deductions can then be made if and in what areas evidence of grazing persists, or whether change has taken place and what the nature of this state is. If conservation values have been enhanced by livestock removal and bore closure then the expectation is that a formerly degraded and probably dysfunctional landscape will regain function, vegetation cover and species diversity.

Variation in the degree of pastoral impact at different sites is likely to be a confounding factor in any comparative study on grazing land. Pringle and Landsberg (2004) emphasize that any experimental study conducted in the Australian rangelands needs to take into account the underlying differences due to past or present grazing activity. Therefore determining the potential residual effects of AWP on vegetation and fauna is important not only on its own but also as a prerequisite for the manipulative Fox and Rabbit removal experiment that followed this study (see Chapter 5).

4.3 Experimental design, site information and general methodology A major problem in studies evaluating the impacts of grazing and the change after cessation of grazing is the choice of points of comparison. Within any ecosystem, sites that have been least impacted by local threats are likely to capture more of the local diversity than sites that have been more heavily modified (Gibbons et al. 2002). Consequently, benchmark sites should be located in areas that have been minimally grazed by livestock, while still being representative of the ecosystem of interest. Finding suitable benchmark sites which are comparable to grazed sites (i.e. similar habitat and plant and animal communities) and are of large enough scale, is inherently difficult because few areas in Australia remain that have escaped disturbance associated with pastoral management (James et al. 1999). An alternative indirect but generally more practicable approach is to use distance from sources of water as a surrogate for

- 35 -

Residual effects of grazing and artificial watering points on flora and fauna grazing pressure and compare sites of known high disturbance; i.e. watering point proximate areas which have been relatively highly disturbed with sites that are water distant and have therefore been left relatively undisturbed. This is an approach that has been widely used (Landsberg and Gillieson 1996; Landsberg et al. 1999a; Landsberg et al. 2002; Montague-Drake 2003; Dowle 2004; Pringle and Landsberg 2004; Croft et al. 2007; Scholze 2007) and is applied here.

To eliminate as many causes for variation as possible between sites other than past grazing, bores instead of earthen tanks were chosen as the disturbed sites. Bores can be constructed anywhere irrespective of any landscape features, whereas earthen dams require the close association with an intermittently flowing creek line.

A map illustrating the location of the monitoring sites has been provided in Chapter 2 (Figure 2.3). The bores used as disturbed sites were (north to south) Devis Bore (sunk 1952; failed 1968), Collins Bore (sunk 1955, capped 1972), Watties Bore (sunk 1960, capped 1972) and Yebna Bore (sunk 1952, capped 1972). The general area had been grazed since the first lease of Fort Grey was established in the late 1800s. Earthen dams (construction dates unavailable) would have provided water for stock prior to the installation of the bores. Bores were added strategically in the most water-distant areas to increase the area that could be grazed on by stock. Thus, prior to the installation of the bores, the area around the bores would have been relatively lightly grazed for nearly 100 years. However, for the 16-20 years (depending on the bore) since the installation of the bores and before stock was removed, the grazing pressure would have been relatively intense, especially in drought years. Infrastructure remaining at the sites consists of the windmill (except Watties), one or two water holding tanks and watering troughs all of which are dysfunctional and in varying states of dilapidation.

Undisturbed sites were at a distance of at least four kilometres to any known historic or present artificial or natural watering point. At this distance from water the sites have not remained truly undisturbed as they would have received a certain level of grazing, especially during favourable conditions and thus little water-dependence of livestock. Even though this is a drawback, such short distances allow replication in a relatively small area and within the same land system, excluding other possible natural variation between sites and thus between site variability.

- 36 -

Residual effects of grazing and artificial watering points on flora and fauna

Due to the difficulty of the terrain, existing National Park tracks leading through the area needed to be used to access the sites. The tracks closely follow the old station tracks and those often run along fence lines. Fences restrict movements of grazing animals and while searching for a way to get to the opposite side of the fence livestock may spend considerable time travelling along fences. Also, fence lines facilitate the movement of stock during mustering and are thus likely to have been used on occasion to herd stock. Increased trampling and potentially increased grazing activity might have resulted. Potentially, areas adjacent to fence lines received higher disturbance through grazing animals than a random location within a paddock. All wire had been removed from the fences in the early years of the National Park’s existence; only the fence posts remain and thus have not presented a barrier to the movement of large animals for the last 30 years. Appendix 1 provides a map of the monitoring sites and the approximate course of the former fence lines. It was unavoidable that one of the 'relatively undisturbed’ sites (Site 6) was situated within approx. 100 m to a former fence line and may have experienced greater disturbance than other undisturbed sites as a result. The potential of this site as an outlier is kept in mind when discussing the results.

For the purpose of this chapter and better understanding of results and figures, the sites (see Chapter 2: Figure 2.3 for a map of their location) have been given descriptive names relevant to the chapter topic (Table 4.1).

Table 4.1: Site names and abbreviations used in this chapter

Site Name Abbreviation Site 1 Disturbed 1 D_1 (Devis Bore) Site 2 Undisturbed 1 U_1 Site 3 Disturbed 2 D_2 (Collins Bore) Site 4 Undisturbed 2 U_2 Site 5 Disturbed 3 D_3 (Watties Bore) Site 6 Undisturbed 3 U_3 Site 7 Disturbed 4 D_4 (Yebna Bore) Site 8 Undisturbed 4 U_4

- 37 -

Residual effects of grazing and artificial watering points on flora and fauna

All measurements for the assessment of residual effects of water-focussed grazing were restricted to the sandy substrate of the dunes. Pitfall traps were established on the dunes where digging was easy and drainage good in contrast to the hard clay soil or gibber of the interdunes which are also prone to flooding. For logistical reasons vegetation measurements were associated closely with the pitfall trapping grid.

4.3.1 General data analysis

Multivariate analysis Permutational multivariate analyses of variance (PERMANOVA) (Anderson 2005), were used to investigate overall differences in ground cover and the shrub assemblage and to compare fauna assemblages between disturbed and undisturbed sites. A nested design was used with ‘Dunes’ as replicates which were nested within the random factor ‘Site’ and nested within the fixed factor ‘Disturbance level’. Multi-dimensional scaling (MDS) was used to graphically explore the data and illustrate results. An analysis of similarity (ANOSIM) was used to test for significant differences between groups in cases where a PERMANOVA was not possible due to an unbalanced sampling design (e.g. ground cover species count). An analysis of species contributions to similarities (SIMPER) was used to identify the main drivers for existing grouping patterns. The whole suite of multivariate data analyses and illustrations were done using the respective functionalities in the software package PRIMER 6.1.6 (PRIMER-E Ltd. 2006). Unless otherwise noted all datasets analysed in PRIMER were square-root transformed to down weigh the most common community components (vegetation) or species (fauna). The resemblance matrices were based on the Bray-Curtis coefficient and the number of permutations used in the PERMANOVA was 9999.

Univariate analysis Nested ANOVAs (same factors and nesting of factors as for PERMANOVA, see above) were used to compare univariate data between disturbance levels. Levene’s test for homogeneity of error variances was conducted on all variables to test the null hypothesis that the error variance of the dependent variable was equal among groups. Levene’s test was chosen in preference to other tests for homogeneity, such as Bartlett’s as it is robust to non-normality in the original variable (Quinn and Keough 2002). In cases where the Levene’s test result was significant a series of transformations (square- root, log, arcsine, 1/x) were applied to normalise the data. The transformed data that

- 38 -

Residual effects of grazing and artificial watering points on flora and fauna provided the highest p-value in the Levene’s test were analysed. Where transformation did not achieve data normalization this is noted and the results of the Levene’s test included as part of the result presentation to provide the reader with means to assess the accuracy of the results as recommended by Field (2000).

- 39 -

Residual effects of grazing and artificial watering points on flora and fauna

4.4 Effects on Vegetation Plants are the primary target of many of the pressures associated with pastoralism, such as grazing, which acts most directly on the composition and structure of plant communities and only secondarily on other elements of biodiversity (Woinarski 2001). Thus measurements on plants are likely to be effective indicators of the response of biodiversity to landuse and represent the majority of core indicators proposed for monitoring rangeland condition (Whitehead et al. 2001).

4.4.1 Hypotheses Sites of heavy and light historical grazing disturbance differ in:

1. The compositional pattern of ground cover 2. The cover and patchiness of individual ground cover categories (total ground cover, cover of pasture and litter) and plant types (forbs, grasses, copperburrs) 3. The assemblage of plant species constituting the ground cover, the number of ground cover species, ground cover species richness, diversity and evenness 4. The assemblage of shrub ‘classes’ and shrub diversity, evenness and richness 5. The abundance, cover and volume of shrub categories and of individual shrub ‘classes’ 6. The ratio of palatable to unpalatable shrubs 7. The abundance of weed species

For completeness the dataset was also used to test for differences between sampling sites addressing the same hypotheses as above. The influence of rainfall was considered as an alternative causative factor for potential differences at disturbance and site level. Thus two additional hypotheses were tested:

8. Differences between disturbance level are more profound than those between individual sites 9. Historical grazing disturbance explains more of the variation in the vegetation characteristics than environmental conditions (i.e. rainfall)

- 40 -

Residual effects of grazing and artificial watering points on flora and fauna

4.4.2 Methods

Ground cover - categories Four series of ground cover measurements were collected at each site concurrent with a seasonal pitfall trapping session (Chapter 5) in January, April, July and October 2006. A point sampling method was used for the measurements. On each of the three dunes at a site, three 100-m transects (crest, slope and ecotone) were surveyed on the northern side of the dune. The consistent use of one side of the dune was important as difference in sun exposure of the two sides is likely to be reflected in the abundance and development of ground cover plants. An imaginary line connecting fixed points was followed with a wheel-point device (described in Moss and Croft 1999) and measurements taken every 50 cm on this transect until 200 sample points were collected. The ground cover at every point was classed to belong in one of ten categories (Table 4.2). For any plant, the height to the nearest five centimetres was measured with the help of markers on the wheel-spikes, and the greenness estimated to be closest to either 0 % green (dead), 25 %, 50 %, 75 % or 100 % green (100 % cell turgescence).

Table 4.2: Description of ground cover categories

Category Description Bare Bare ground Grass Perennial or ephemeral grass Forb Dicotyledonous plant without hard or woody stem Round-leafed chenopod Chenopod with cylindrical leaves and no spiky leaf bracts, primariliy Maireana spp. Flat-leafed chenopod Chenopod with flattened leaves, primarily Atriplex spp. and Rhagodia spp. Copperburr Chenopod with spiny leaf bracts primarily Sclerolaena spp. Litter Dead plant material Dung Animal dung Wood Dead wood greater than 5 cm in diameter Woody debris Dead wood less than 5 cm in diameter

Of the initial ten ground-cover variables ‘forb’, ‘grass’ and ‘copperburr’ were grouped together to form one of two ‘pasture’ categories: ‘dry pasture’ if the plant turgescence was less than 25 % of full turgescence and ‘green pasture’ for turgescence values >25 %. The cover categories were further condensed by pooling the three categories with the lowest occurrence to form the category ‘other’, thus reducing the dataset to five final categories (‘bare ground’, ‘litter’, ‘green pasture’, ‘dry pasture’ and ‘other’) which made for easier interpretation of the results.

- 41 -

Residual effects of grazing and artificial watering points on flora and fauna

The absolute count of each ground cover category was used in the analysis as it was equal for all sites and replicates but the results are presented as percentages of cover to aid comparison with other studies. The coefficient of variation (CV) was used as a measure of patchiness and included in the analysis (n = 3 at all sites). In addition to the mean abundance of the five condensed cover categories (‘bare ground’, ‘litter’, ‘green pasture’, ‘dry pasture’ and ‘other’) and their CV, the original cover categories containing a plant type (i.e. ‘forb’, ‘grass’ and ‘copperburr’, ‘flat-leafed’ and ‘round- leafed chenopod’), were compared between disturbance levels and individual sites. Separate analyses were conducted for cover and patchiness. Using trapping session as replicates in the analysis introduced too much variance in the data (see Results) and so separate analyses were run for each session where possible. The results that were consistent over sessions were investigated further.

Ground cover - species composition In addition to the seasonal monitoring of the ground cover, which was based on cover categories, two further plant surveys were conducted in which plants were identified to species. In session two (April 2006) at each trap unit all ground cover species within a square of 2 x 2 metres were identified and counted. The sampling was restricted to those five sites that had previously received rain. From this survey presence/absence data and indices of species richness, diversity and evenness were derived. A timed search (2 people, 10 minutes) was performed in session four (October 2006). A specimen of each plant species detected within the area of the trapping grid (100 x 100 m) was collected for later identification. Presence/absence data and number of species per dune were recorded. Data from both surveys (one summer and one winter survey) were used to create a plant list for the study area. The listed species of shrubs and trees were identified as part of the shrub measurements (see next section). Identification of plant species was either by reference to Cunningham et al. (1992) or, where there was uncertainty, the plant specimens were dried and sent to the John Waterhouse Herbarium at the University of NSW for identification. The plant species community data from session four were compared between disturbance levels with a PERMANOVA, whereas the data from session two were compared with an ANOSIM as the sampling design was unbalanced.

- 42 -

Residual effects of grazing and artificial watering points on flora and fauna

Shrubs Data on shrubs were only collected once, since the shrubs are slow-growing and thus neither the shrub abundance nor their size would have changed significantly over the period of the study. In October 2006 all living and dead shrubs (with the exception of copperburrs) all living and dead trees and all dead and fallen wood/shrubs were counted within an ellipse around each pitfall trapping unit (major axis 20 m, minor axis 10 m, area 157 m2). Two perpendicular measurements of the shrub width (w1, w2) and its height (h) were taken to the nearest 5 cm for shrubs (< 2 m) and to the nearest 10 cm for tall shrubs and trees (>2 m) with a measuring stick. From the measurements the projected canopy cover (w1*w2) as well as a volume estimate (V= (*w1*w2*h)*3/4) were calculated. All living shrubs/trees were identified to genus and species where possible. The density of shrubs and genera of shrubs was estimated from the total area of the sampling area (ellipse see above). Shrubs were grouped into the following broad categories ‘Tree/tall shrub’, ‘living shrub’, ‘dead shrub’, ‘dead wood’.

The shrub variables included in the PERMANOVA were density, projected cover and volume of those shrub classes with a count greater than 30, including dead shrubs and dead wood. Shrub community indices (i.e. diversity, evenness and richness) were also compared using nested ANOVAs. Counts for tall shrubs and trees were extremely low and only their density was considered.

Introduced plants In the period relevant to this chapter (January 2006 – October 2006) only one rain event occurred which provided widespread and sufficient precipitation to trigger plant growth at all sites. The sites were checked for the occurrence of introduced plant species and those species present were counted within circles of 10 metre radius at twelve locations at each site. The marker posts otherwise used for the track plots (see Fauna section 3.5) were used as the centre point of the circle. The location and distribution of sampling plots on the dunes thus corresponds to that of the track plots (i.e. four on the crest of the dune, four on mid-slope and four on the lower slope (Figure 4.9)). The mean abundance of introduced plants between disturbance levels and individual sites was compared with a nested ANOVA.

- 43 -

Residual effects of grazing and artificial watering points on flora and fauna

Rainfall as a causative factor of temporal and spatial variation in vegetative variables A number of malfunctions in the rain gauges occurred and resulted in periods of missing data for some sites. The gaps were filled by using the average of the two neighbouring sites. In the face of considerable spatial variation in rainfall patterns this was deemed to give a more accurate representation of the actual local precipitation than the average of all sites.

The timeframe of the study of one year, the rarity of rainfall events and fixed quarterly sampling sessions limited the possible rainfall variables that could be investigated. Cumulative rainfall in previous months (2, 3, 4 and 8 months prior) and lags since previous rainfall events were used as predictor variables. The lags (1, 2 and 3 months) were predetermined by the timing of significant rainfall events (>10 mm) in relation to the sampling session. Sample sizes varied depending on how many sessions could be included. Distance-based linear modelling, using the DISTLM function in PRIMER was used to analyse the relationship between rainfall and ground cover. Principal Coordinates (PCO) plots were used to explore the data and visualize results.

4.4.3 Results A total of ninety-eight native and two exotic plant species were identified as part of the ground cover and shrub survey across all sites (see Appendix 2 for a list of species). Since the surveys were restricted to the trapping grid on the sand dunes, the list omits those species found on substrates other than sand. Ninety-one of the species were dicotyledons and nine species monocotyledons and the community was dominated by annuals comprising 68 species compared to 32 perennial species.

Ground cover Plant cover was consistently low (20-30 %) in all sessions. ‘Litter’, which consists mainly of dead plant material, generally accounted for most of the ground cover, except in session three where prior rainfall produced a burst of green pasture. The ground cover composition (mean cover) and the patchiness of ground cover categories (mean CV) between sampling sessions varied significantly (PERMANOVA: F3, 64 = 36.552, p = 0.001 and F3, 64 = 4.855, p = 0.001). The ground cover composition was significantly different (p < 0.005) for all combinations of sessions, and the patchiness for all but session two and three (Table 4.3). In view of such variation, which is not

- 44 -

Residual effects of grazing and artificial watering points on flora and fauna surprising in a desert environment where rainfall greatly changes the landscape and associated plant communities, sessions were treated and analysed separately.

Table 4.3: Temporal variation in ground cover. Results of comparison of mean cover of ground cover categories and their mean patchiness between sessions (PERMANOVA).

Mean Cover Mean Patchiness Sessiontptp 1 v 2 2.093 0.012 1.827 0.019 1 v 3 5.044 0.001 2.219 0.003 1 v 4 4.026 0.001 2.105 0.004 2 v 3 5.347 0.001 1.632 0.061 2 v 4 1.996 0.030 2.062 0.004 3 v 4 7.889 0.001 2.432 0.001

Comparison at disturbance level PERMANOVA analysis revealed no significant differences between disturbed and undisturbed sites across all categories of ground cover and ground cover patchiness in any session. Likewise analysis of individual cover categories revealed no significant difference between disturbed and undisturbed sites for cover (Figure 4.1) or patchiness (Figure 4.2).

Of all plant types, the herbaceous group ‘forb’ formed the highest cover (5-15 %) in all sessions (Figure 4.3). The remaining plant types were all very sparse forming less than 0.5 % of cover. Forbs and copperburrs consistently covered more ground at undisturbed compared to disturbed sites whereas the grass cover was higher at disturbed sites. However, none of those observations differed significantly with disturbance. Likewise the patchiness of those plant types did not differ significantly consistently over sessions with disturbance (Figure 4.4).

- 45 -

Residual effects of grazing and artificial watering points on flora and fauna

Figure 4.1: Comparison of mean cover of plant categories between disturbance levels. Mean cover (%) of ground cover categories for disturbed (dark bars) and undisturbed sites (light bars) in the individual sessions. Error bars represent 95 % confidence intervals (n = 12).

- 46 -

Residual effects of grazing and artificial watering points on flora and fauna

Figure 4.2: Comparison of mean patchiness of ground cover categories between disturbance levels. Mean patchiness (coefficient of variation) of ground cover categories for disturbed (dark bars) and undisturbed sites (light bars) for individual sessions. Error bars represent 95 % confidence intervals (n = 12).

- 47 -

Residual effects of grazing and artificial watering points on flora and fauna

Figure 4.3: Comparison of mean cover of plant types between disturbance levels. Mean cover (%) of ground cover categories representing a plant type for disturbed (dark bars) and undisturbed sites (light bars) in individual sessions. Error bars represent 95 % confidence intervals (n = 12).

- 48 -

Residual effects of grazing and artificial watering points on flora and fauna

Figure 4.4: Comparison of mean patchiness of plant types between disturbance levels. Mean patchiness of ground cover categories representing a plant type for disturbed (dark bars) and undisturbed sites (light bars) in individual sessions. Error bars represent 95 % confidence intervals (n = 12). The plant type ‘Flat-leafed’ and ‘Round-leafed chenopods’ were excluded from the graphs due to their sparseness.

There was some discernible separation in the ground cover species assemblage between disturbed and undisturbed sites in the MDS plots based on a) the quadrat plant count (species abundances) and b) presence/absence data of the plant species detected during the timed search (Figure 4.5). However statistical analysis found no significant differences based on abundance data from the plant count (ANOSIM: R = 0.25, significance level 30 %) nor based on presence/absence data of the plant species detected during the plant count (ANOSIM: R = 0.25, significance level 30 %) and the timed search (PERMANOVA: F1, 16 = 1.671, p = 0.137). Likewise the number of species recorded with the species inventory was not significantly different between disturbance groups (ANOVA: F1, 6 = 0.069, p = 0.81), nor were the number of plants, number of species, species richness, diversity or evenness calculated from the ground cover plant count (ANOVA: (log-transformed) F1, 3 = 0.449, p = 0.551;

- 49 -

Residual effects of grazing and artificial watering points on flora and fauna

F1, 3 = 0.755, p = 0.449; F1, 3 = 0.957, p = 0.400; F1, 3 = 1.373, p = 0.326; F1, 3 = 0.542, p = 0.421, respectively). a) b) 3D Stress: 0.06 3D Stress: 0.14

Figure 4.5: Comparison of ground cover species composition between disturbance levels. MDS-plots based on a) ground cover species abundances (square-root transformed, Bray-Curtis similarity) and b) presence/absence data of all species detected during a timed search. Filled circles = Undisturbed, open diamonds = Disturbed.

Comparison at site level A number of significant differences in mean cover and patchiness were found for sites within the disturbance groups based on the whole suite of cover categories and in individual ground cover categories (Appendix 3 and Appendix 4). This suggests considerable heterogeneity across the landscape. However, few of the differences in ground cover were present in more than one session and so the temporal component of the spatial variation was inconsistent. Differences in individual sites were also present in the plant species composition (presence/absence data). In the disturbed group, site four was significantly different to site one and two (t4 = 2.142, p = 0.030; t3 = 1.911, p = 0.040), in the undisturbed group site differences were weaker (p > 0.1) but a number of differences between sites irrespective of disturbance group occurred. Species richness, diversity and evenness did not differ significantly between sites. The species composition was only sampled once and so temporal variation within these results could not be assessed.

- 50 -

Residual effects of grazing and artificial watering points on flora and fauna

Shrubs Of the shrubs only 40 % were living, the greatest proportion, 60 %, were dead. In addition to those a large number (523) fallen shrubs/trees (i.e. dead wood) were counted and measured. The most abundant genera of shrubs were Crotalaria, Sida/Hibiscus (grouped together since non-flowering specimens are indistinguishable), Dodonea, Enchylaena and Atriplex which together accounted for 74 % of all living shrubs (Table 4.4). The mean (± SE) living shrub density on dunes varied greatly from an average of 0.014 ± 0.008 to 0.177 ± 0.052 per m2.

Table 4.4: Total abundance (all replicates and sites) and percentage of all living shrubs counted. Palatability to livestock is indicated where applicable.

Scientific name Common name Count % of Total Crotalaria eremaea eremaea ** Loose-flowered Rattlepod 366 22.8 Sida spec. /Hibiscus krichauffianus * Sida/Hibiscus 253 15.8 Dodonea viscosa subsp. Angustifolia Narrow-leaf Hopbush 225 14.0 Enchylaena tomentosa * Ruby Saltbush 176 11.0 Atriplex stipitata Bitter Saltbush 169 10.5 Acacia ligulata Sandhill Wattle 96 6.0 Rhagodia spenescens & gaudichaudiana Cottony & Thorny Saltbush 58 3.6 Ptilotus obovatus Silvertail 55 3.4 Acacia aneura ** Mulga 51 3.2 Hakea leucoptera Needlewood 40 2.5 Senna form taxon 'sturtii' Dense Senna 20 1.2 Phyllanthus fuernrohrii * Sand Spurge 20 1.2 Eremophila longifolia ** Emubush 17 1.1 Maireana spec. Bluebushes 15 0.9 Atalaya hemiglauca Whitewood 12 0.7 Other 32 2.0 TOTAL living shrubs/trees 1605

dead shrubs 2380 dead wood 523

** highly palatable to stock, * moderately grazed, mostly in drought (Cunningham,1992)

Comparison at disturbance level There was no significant difference in the shrub assemblage between disturbance levels

(PERMANOVA: F1, 16 = 1.167, p = 0.334). However, disturbed sites had a significantly higher mean density of dead wood (ANOVA: F1, 6 = 9.473, p = 0.022) compared to undisturbed sites (Figure 4.6). The mean density of trees/tall shrubs, living shrubs and dead shrubs did not differ with disturbance level. Similarly, none of the structural characteristics of shrubs (mean cover and mean volume) or the shrub community indices (richness, evenness and diversity) were significantly different (p < 0.05) or showed a trend (p < 0.1) between disturbed and undisturbed sites. The abundance of the

- 51 -

Residual effects of grazing and artificial watering points on flora and fauna most common shrub ‘classes’ was generally similar between disturbance levels with the exception of Mulgas (Acacia aneura) which showed a trend to occur in higher abundances at undisturbed sites (ANOVA: F1, 6 = 4.405, p = 0.081; data square-root transformed). The ratio of palatable to unpalatable shrubs was also similar between groups (log transformed; ANOVA: F1, 6 = 2.506, p = 0.164).

Figure 4.6: Comparison of mean density of shrub classes between disturbance levels. Mean density of Shrubs and Trees for disturbed (dark bars) and undisturbed sites (light bars). Error bars represent 95 % confidence interval; ** indicates a significant difference, p = 0.05; n = 12.

Comparison at site level The shrub-class community was mostly homogeneous between the sites within their respective disturbed and undisturbed group. D_4 was the only anomalous site within the disturbed group. It was significantly different to D_1 and D_2 (PERMANOVA: t4 = 3.150, p = 0.020 and t4 = 2.272, p = 0.031). The interaction effect of ‘disturbance’ and ‘site’ was significant for the mean density of dead shrubs (F6, 16 = 3.971, p = 0.013, square root transformed) and living shrubs (F6, 16 = 3.909, p = 0.014), with site D_4 having less dead shrubs but more and smaller living shrubs than the other sites. The interaction effect was also significant for a number of shrub classes and thus these are

- 52 -

Residual effects of grazing and artificial watering points on flora and fauna

likely to be contributors to the anomaly of site D_4: Atriplex stipitata (F6, 16 = 2.731, p = 0.051; transformation did not improve normalization, Levene’s test: F7, 16 = 6.484, p = 0.001), Crotalaria eremea (F6, 16 = 4.441, p = 0.008; transformation did not improve normalization, Levene’s test: F7, 16 = 10.946, p << 0.001), Rhagodia sp. (F6, 16 = 14.957, p << 0.001; square root transformed) and Sida/Hibiscus spp. (F6, 16 = 3.486, p = 0.021). Even though the shrub community as a whole was homogeneous in the undisturbed group, U_3 had a significantly lower shrub ‘class’ diversity and evenness (F6, 16 = 4.533, p = 0.007 and F6, 16 = 11.058, p << 0.001) and a higher cover over of dead wood

(F6, 16 = 5.488, p = 0.003) than the other sites in the group.

Introduced plants Only two introduced plant species, Camel Melon (Citrullus lanatus) and Wild Turnip (Brassica tournefortii), were recorded in the study area. Of those only B. tournefortii was sufficiently abundant to be compared between disturbance levels.

Comparison at disturbance level Disturbed sites had a higher mean abundance (± SE) of Brassica tournefortii per m2 (Disturbed: 0.053 ± 0.010; Undisturbed: 0.031 ± 0.005) but this difference was not statistically significant between the two groups (ANOVA: F1, 6 = 0.409, p = 0.546; data transformation did not achieve normalization; Levene’s test: F7, 280 = 22.795, p << 0.001).

Comparison at site level There was a significant interaction between status and site indicating that individual sites differed in the abundance of B. tournefortii (ANOVA: F6, 16 = 4.441, p = 0.008). However, most sites had variable but similar abundances of this species (Figure 4.7). The exceptions were site D_1 with very few B. tournefortii and D_3 with many. Wild turnip is known to occur in clumps, caused by young plants sprouting from seeds that have fallen close to the mother plant. Therefore for most sites the variance was very high as indicated by the large confidence intervals.

- 53 -

Residual effects of grazing and artificial watering points on flora and fauna

D 1 U 1 D 2 U 2 D 3 U 3 D 4 U 4

Figure 4.7: Density of Brassica tournefortii at sampling sites. Grey bars: disturbed sites, white bars: undisturbed sites. Error bars represent 95 % confidence interval. Non-overlapping CI’s indicate significant differences between the respective sites. N = 36.

- 54 -

Residual effects of grazing and artificial watering points on flora and fauna

Rainfall as a driver for temporal and spatial variation Site-specific precipitation records were restricted to the timeframe of this thesis and spanned over a year. Rainfall in general was sparse over the year of the study with only two widespread and significant rainfall events (Table 4.5). Thus the exploration of rainfall lags was limited.

Table 4.5: Monthly precipitation (mm) received at sampling sites during 2006. The months that included sampling sessions are marked (*). Total values in parenthesis indicate that at least one value for the site was missing and was replaced with the average of the neighbouring sites.

Month (2006) D_1 U_1 D_2 U_2 D_3 U_3 D_4 U_4 Avg Jan* 0.0 0.0 0.0 0.0 0.0 0.0 0.0 0.0 0.0 Feb 0.0 0.0 17.8 17.8 25.2 16.5 37.7 27.6 23.8 Mar 0.0 0.0 0.0 0.0 0.0 0.0 0.0 0.0 0.0 Apr* 0.0 0.0 0.0 0.0 0.0 0.0 0.0 0.0 0.0 May 0.0 0.0 0.0 0.0 0.0 0.0 0.0 0.0 0.0 Jun 4.0 5.2 4.6 5.8 7.7 7.3 9.3 6.5 6.3 Jul* 22.6 21.2 33.6 26.5 20.8 22.0 26.0 21.0 24.2 Aug 0.0 0.0 0.0 0.0 0.0 0.0 0.0 0.0 0.0 Sep 0.0 0.0 0.0 0.0 0.0 0.0 0.0 0.0 0.0 Oct* 6.2 7.9 8.2 8.5 7.6 7.5 6.7 8.4 7.6 Nov 0.0 0.0 0.0 0.0 0.0 0.0 0.0 0.0 0.0 Dec 11.2 5.4 3.4 2.1 5.1 5.1 3.6 5.1 5.1 TOTAL 44.0 39.7 [49.8] [42.89] [53.68] 58.3 83.2 68.6 58.8

Ground cover: A SIMPER analysis was used to explore the driving variables behind the temporal and spatial variation in ground cover. The main contributor (> 30 %) to the significant differences between the sessions (Table 4.3) was in most cases ‘green pasture’ except in the comparison of session one and four where ‘litter’ contributed most to the dissimilarity (Appendix 5). In contrast to this the differences between sites within the sessions were mainly driven by the percentage of bare ground, contributing to 50 % or more of the observed dissimilarity in all sessions and ‘dry pasture’ (> 23 % contribution), except in session three where ‘green pasture’ was the second highest contributor (22 %) (Appendix 6). These results are well illustrated in Figure 4.8 a, where the ground variables have been added as vectors to a PCO plot based on the similarities of samples in ground cover. The length and direction of the vectors indicate the strength and sign, respectively, of the relationship between that variable and the PCO axes. ‘Green pasture’ and to a lesser extend ‘dry pasture’ separates the sessions along the x-axis, whereas ‘bare ground’ and ‘litter’ separate the sites along the y-axis.

- 55 -

Residual effects of grazing and artificial watering points on flora and fauna

The variables contributing most to the spatial and temporal variation in ground cover between samples (i.e. green pasture, litter, bare ground and dry pasture) are all likely to be affected by precipitation. Thus the significance of a number of rainfall variables in explaining the variation in ground cover was explored using distance-based linear modelling (DistLM in PRIMER). A rainfall lag of two months (R2 = 0.39, Pseudo-

F1, 6 = 3.872, p = 0.030, n = 8) and cumulative rainfall in the previous three months 2 (R = 0.31, Pseudo-F1, 30 = 13.276, p << 0.001, n = 32) were significant but explained relatively little of the variation in the dataset. The best results were obtained for 2 cumulative rainfall in the previous two months (R = 0.62, Pseudo-F1, 30 = 47.904, p = 0.001, n = 32) which explained 62 % of the variation in ground cover (Figure 4.8 b). The vector of cumulative rainfall in the previous two months extends along the same direction as the vector for ‘green pasture’ and ‘dry pasture’ and thus explains the temporal variation (i.e. the separation between the sessions). The variation along the y- axis, related to the amount of ‘bare ground’ and ‘litter’ could not be explained by any of the rainfall variables. Other aspects of the precipitation pattern or factors unrelated to rainfall may have caused the variation along the y-axis and thus the spatial variation.

Introduced plants: B. tournefortii were counted several months after the growth- triggering rain event as they were easiest to detect when they had dried up. The precipitation received as a result of that rain event was used as the predictor variable in the regression analysis. There was no correlation between rainfall and the average 2 = abundance of B. tournefortii (R = 0.373, R 0.139, F1, 6 = 0.970, p = 0.363).

- 56 -

Residual effects of grazing and artificial watering points on flora and fauna a) 10 1 1

2 5 Bare ground 1 2 1 4 1 2 3 3 3 4 3 0 14 1 1 3 3 4 2 3 Dry pasture 2 3 4 4 2 -5 Green pasture PCO2 (13.6% of total variation) 4 Litter 4 2 2 -10 -15 -10 -5 0 5 10 15 20 PCO1 (76.8% of total variation) b) 10 1 1

5 2 1 2 1 4 1 2 3 3 3 4 3 lag1 0 14 1 1 3 3 4 2 3 cum4 lag3 cum2 3 cum8 2 4 lag2 cum3 4 2 -5 PCO2 (13.6% of total variation) 4 4 2 2 -10 -15 -10 -5 0 5 10 15 20 PCO1 (76.8% of total variation) Figure 4.8: Similarity in ground cover between samples. PCO plot based on ground cover variables (square-root transformed; Bray Curtis similarity). Numbers (1-4) represent sampling sessions. The length and direction of the vector indicate the strength and sign, respectively, of the relationship between that variable and the PCO axes. a) Vectors are ground variables (Pearson correlation >0.55) b) Vectors are the investigated rainfall variables: rainfall lag of 1, 2 and 3 months and cumulative rainfall in previous 2, 3, 4 and 8 months. Cumulative rainfall of the previous 2 months explained the variation best and was superimposed as bubbles. Bubble-size correlates with the amount of rain received; no bubble indicates no rainfall.

- 57 -

Residual effects of grazing and artificial watering points on flora and fauna

4.5 Effects on Fauna

4.5.1 Hypotheses A series of hypotheses were tested in relation to the effects of disturbance on elements of the fauna. The disturbance levels were hypothesised to differ in: 1. The composition of the invertebrate, reptile and small mammal assemblage 2. The species diversity, evenness and richness of invertebrates, reptiles and mammals 3. The ratio of open habitat favouring agamids (Ctenophorus) and skinks (Ctenotus) as well as the ratio of House Mice to native rodents 4. The activity of Foxes, Cats, Rabbits and the abundance of House Mice Like the flora, spatial and causal effects were tested in the following hypotheses: 5. The differences between disturbance level are more pronounced than those between individual sites 6. Historical grazing disturbance explains more of the variation in the fauna dataset than environmental factors; e.g. microhabitat factors, temperature, rainfall

4.5.2 Methods

Small vertebrate sampling Ground-dwelling vertebrates were live-captured using a combination of pitfall- and Elliott trapping. On each dune seven pit-trap units were established within a dune section of approximately one hectare (Figure 4.9). A trap unit consisted of a pair of pitfall traps (stormwater-pipe, 60 cm long, 16 cm diameter) buried 10 m apart and flush to the ground. The two traps were connected by a drift fence (plastic weed mesh buried ca. 10 cm deep and standing up ca. 30 cm held by metal pegs) to increase capture rate. The connection of at least a pair of pitfall traps with a relatively long drift fence was recommended by Friend et al. (1989) and Hobbs et al. (1994) for capturing animals in the arid zone, in particular small mammals. As a bottom for the pipes shade cloth (durable plastic mesh) was used, which prevented the escape of any captured animals through the bottom of the trap and at the same time allowed drainage of traps in the event of unexpected rainfall. A tightly fitting galvanized iron lid was used to close the traps when not in use. In association with each pitfall trap unit one standard sized Elliott trap was used. They were placed within a radius of ten meters of the pitfall unit in a shady location underneath or near vegetation cover to prevent the Elliott traps from heating up in hot weather. Traps of both types were lined with a small handful of - 58 -

Residual effects of grazing and artificial watering points on flora and fauna

‘Topstuff’ (synthetic, odourless, non-allergenic ‘cotton wool’) to cushion the fall into the pit, to minimize thermal stress, to provide some shelter from predators and to maintain the general well-being of captured animals. During times of high ant activity the rim of the pitfall traps as well as the back of the Elliott traps was sprayed with a low-irritant insect spray to prevent ants from entering the traps and causing injuries or fatalities. Elliot traps were baited with a mixture of rolled oats, peanut butter and honey. Pitfall traps were left unbaited.

Slope

Top 100 m

Slope

100 m

Figure 4.9: Arrangement of pairs of pitfall traps ( ) and track plots ( ) in a one hectare trapping grid on a dune.

Seasonal trapping sessions (January, April, July and October) were conducted in 2006. During each session every site was trapped for three consecutive nights, comprising 126 pitfall and 63 Elliott trap-nights per site, which adds up to 1512 trap nights per session and a total of 6048 trap nights for the results presented in this chapter. Traps were set in the late afternoon/early evening and inspected (and closed on the last day) early in the morning each day.

For every capture, records of the date, site, dune and trap unit were taken. Each vertebrate capture was identified to species level, sexed where possible and weighed to the nearest 0.5 g on a small set of digital scales (jewellery scales, various models).

Reptiles were not marked for two reasons: a) the lack of ethical, easily applicable and reliable marking techniques available for small reptiles and b) the high rate of reptile

- 59 -

Residual effects of grazing and artificial watering points on flora and fauna captures and thus ultimately time constraints. The dataset for lizards may therefore include recaptures within and between trapping sessions. The implications of potential recaptures in the dataset should be negligible since other studies in the area have shown that recapture rates are low (Dowle 2004; Scholze 2007).

Small mammals weighing 10 g or more were microchipped for reliable long-lasting identification of individuals. The implantable ID-100 transponders (Microchips Australia) were inserted between the shoulder blades using an Implanter (Trovan Deluxe (IME)). For animals lighter than 10 g the rice-kernel sized microchip was deemed to be too large relative to body size and might have impaired movement. Instead those individuals were marked with individual ear notches by cutting out small triangles of ear cartilage with disinfected nail clippers. Microchipping and cutting of ears was performed while the animals were sedated using Isoflurane. Isoflurane is a stable, non-explosive inhalation anaesthetic that is relatively free from significant side effects. It has been recommended by Anstee and Needham (1997), as a safe field anaesthetic for small mammals as its administration is simple, induction is rapid, control of depth of anaesthesia is easy, recovery of animals is quick and complications are rare. A mouse-sized induction mask was home-made according to instructions by Anstee and Needham (1997) and with the help of and material provided by the RSPCA in Broken Hill. Female marsupials carrying young or obviously suckling young left behind in a nest, and obviously pregnant or suckling native mouse species, were not put at risk of an anaesthetic. They were identified and weighed but remained unmarked.

After measurements and marking procedures all captured individuals were released at the point of capture into the area surrounding the pitfall traps and some sort of shelter (holes in the ground or dense litter) wherever possible.

Invertebrate sampling Invertebrates were captured in the same pitfall traps as the vertebrates. All invertebrate captures were identified to a broad taxonomic category (class in most cases) and released in the near vicinity of capture.

Measures of Fox, Cat and Rabbit activity A variety of methods have been used in the past to monitor the activity and or measure abundance/density of Foxes, Cats and Rabbits respectively. For the purpose of this - 60 -

Residual effects of grazing and artificial watering points on flora and fauna project a method was needed that 1) monitored all three species simultaneously and 2) measured the activity of the three species on the monitoring sites but not attract these non-native species to the site from a wider area and thus potentially increasing their impact of predation and competition on the native small vertebrate communities. Spotlighting and the use of track plots were the two feasible methods.

Spotlighting surveys to estimate Fox, Cat and Rabbit abundance were initially conducted on four consecutive nights (starting 1 hour after sunset) along a set route of about 35 km of service track traversing all study sites. The route was driven in opposite direction every second day to account for potential temporal differences in the activity of animals from the start of the survey to the time it finished. The vehicle was driven at 15 km/h per hour whereby a spotlight (Lightforce) was slowly and continuously swept from side to side to detect any Foxes, Cats and Rabbits. Spotlighting surveys were conducted in sessions one and two but despite good visibility in an open semi-arid environment, the extremely low frequencies of sightings, even for Rabbits, led to the discontinuation of the surveys.

Whereas the low abundances of Foxes and Cats typical for rangelands made spotlighting unfeasible, track-based activity estimates detect changes most reliably in this lower density range and the index becomes less sensitive as density increases (Edwards et al. 2000). This together with an optimal sandy substrate made track plots the most suitable method and it was used throughout the project to monitor Fox, Cat and Rabbit activity.

The Fox, Cat and Rabbit activity at the monitoring sites was assessed using twelve circular (radius of 90 cm) track plots on each dune (Figure 4.9). Such a set-up promised a high likelihood to detect a Cat or Fox if it visited the site but because the plots are so close together that the detection of a single individual at more than one site cannot be ruled out abundance cannot be accurately assessed Using this methodology. However, in the case of predators at least measuring the activity of predators at the sites is more meaningful than measuring abundance as a decrease in Fox and/or Cat abundance in the area does not necessarily equate to lower predation pressure at the experimental sites.

- 61 -

Residual effects of grazing and artificial watering points on flora and fauna

Typically olfactory lures are used in conjunction with track plots set for carnivores to increase the visitation rate. Scent can carry a long way and attract Foxes and Cats from a wide area, an unwanted effect in this study, where the natural activity of the non- native species at the study sites needed to be assessed. Furthermore a scent that works for one animal may be unpleasant for another thus discouraging it from visiting the plot. Molsher (2001) even found that scent lures designed for cats decreased capture efficiency whereas visual lures increased capture efficiency; these results were statistically significant however. For these two reasons visual rather than olfactory lures were used to potentially increase the visitation rate to the plots. Visual lures can only attract those animals that are in sighting distance to the lure, thus those that unaffectedly utilize the study sites. A combination of pink or orange biodegradable flagging tape, a feather and a small metal tag on a string were used as lure. Flagging tape appears to be quite attractive to Foxes who regularly chew the tape off markers set by staff of DECCW, the Road and Traffic Authority and Country Energy working in the area them (pers. comm. Dan Hough, Kim Piddington, John Illies, George Williams, Cindy McGeorge). It has also been used in some studies to attract Cats (Edwards et al. 1997; Risbey et al. 1997; Molsher 2001) as have been tape metal tags and feathers (Molsher 2001). The disturbance to the soil resulting from setting the plots may have provided an additional olfactory lure to the animals but as it is a natural scent it is unlikely to discourage visitation by either of the species. Rabbits are not generally known to respond to lures (Moseby et al. 2004) and were thought to be attracted by the upturned soil to the plots if they were attracted at all and did not traverse the plots incidentally. Surprisingly, Rabbits were found to respond well to the visual lures, with many imprints of only the hind feet found very close to the post holding the lure where the animals must have stood up on their hind legs to investigate the lure.

On the first sampling day in each session the plots were raked and then smoothed using a custom made device. The visual lure was hooked onto the stake in the middle of the plot. The track plots were checked for foot prints of Rabbits, Cats and Foxes every morning for three consecutive days and presence/absence of prints for each species recorded. All prints found were smoothed over to ‘reset’ the plot for the next 24-h period. The locations of the plots were fixed. The metal post in the centre of the track plot remained on site at all times but the lures were only attached for the three days of sampling.

- 62 -

Residual effects of grazing and artificial watering points on flora and fauna

Data manipulation and analysis The small vertebrate survey effort was constant in all sessions and thus the number of captures (small mammals, reptiles and invertebrates) was used as an index of species abundance. For small mammals, which were individually marked, within-session recaptures were removed from the dataset. Community variables (diversity, evenness and richness) for all three faunal groups were computed from the original capture rates as was the ratio of open-habitat favouring agamids (Ctenophorus) to skinks (Ctenotus) and the ratio of introduced rodents to native rodents (Read and Pickering 1999). Using sessions as replicates in the analysis generally introduced too much variance in the small vertebrate data due to thermal preferences of some of the fauna (i.e. reptiles which are highly active in summer but less active or inactive in winter) and so separate analyses were run for each session wherever possible. The results that were consistent over sessions were investigated further. The dataset for small mammals was relatively small (71 captures, minus five recaptures). Separate analysis of sessions (i.e. small mammals) was thus not possible and the data were pooled over sessions and then analysed. The small dataset was accommodated by limiting the analysis to the small mammal community. Species-specific analyses were only conducted for the most common species, the House Mouse.

Initially six sand plots instead of twelve were used to measure the activity of Foxes, Cats and Rabbits. With these fewer plots however a number of instances a Cat/Fox was not recorded, even though the species had been visiting as evident from footprints discovered elsewhere in the trapping grid or along the access track. To increase the detection rate the number of track plots per monitoring site was doubled with satisfying results. Due to trials of methodology and sampling intensity until it yielded robust results, data regarding the non-native species are available from session three onwards. From then on the survey effort was constant in all sessions. From the presence/absence data of footprints on the individual track plots (no differentiation between individuals was attempted) two measures were derived that describe two different dimensions of the potential impact of the non-native species. These were 1) visitation intensity expressed as the total number of track plots visited by a particular species on a given site over the course of three days, 2) the frequency of visitations as the number of days (max. 3) a species was present at a sampling site.

- 63 -

Residual effects of grazing and artificial watering points on flora and fauna

Analysis of disturbance effects: PERMANOVA (nested design, based on Bray-Curtis similarities, 9999 rotations) were used to identify potential differences between disturbed and undisturbed sites in the invertebrate, small mammal and lizard assemblages. The species abundance data were square-root transformed before analysis to down weigh the abundant species. MDS plots provided an additional way of data exploration and illustration of results. The BEST and SIMPER functionalities in the PRIMER software were used to identify the variables driving the differences between disturbance levels or between individual sites. Nested ANOVA were used to compare potential differences in individual variables. Individual species abundances were only compared for species with more than 20 total captures.

Causative factors of fauna distribution: Investigating the factors underlying the distribution of the invertebrate assemblage was not essential to the project and because invertebrates were only broadly classified the interpretation of correlations would be difficult. The capture rate of small mammals was too low to validate analysis with the exception of the House Mouse which was sufficiently abundant for its relationship with rainfall to be analysed for the October session. Thus, the analysis of causative factors of distribution was focussed on the lizard assemblage where sufficient capture rates were achieved.

A principal component analysis (PCA) and subsequent regression analysis were used to investigate the causative factors determining the lizard distribution and were performed with two aims in mind. The first aim was to identify correlations of lizard community variables and abundances with microhabitat variables. The microhabitat variables entered were cover and patchiness of vegetation variables measured in the ground cover survey, as well as density, projected cover and volume of living shrubs, dead shrubs and dead wood. The latter variables were derived from the data of the shrub survey. The second aim of the analysis was to investigate if some of the observed differences reflect variations in meteorological conditions. Logistically it was impossible to trap at all sites simultaneously and even though surveys were completed as quickly as possible (weather permitting) the first and last survey day were at least 16 days apart. Variability in the weather in this timeframe is a potential factor impacting on the activity and thus trappability of fauna species (Read 1988; Read and Moseby 2001). Observed differences between sites may in actual fact be an artefact of the trapping schedule and

- 64 -

Residual effects of grazing and artificial watering points on flora and fauna the weather effect can confound true differences in abundance. A PCA was also used to explore potential associations between the most common lizard species. The identification of species associations focussed and facilitated the interpretation of the correlations found with habitat factors.

Minimum and maximum temperature, humidity and wind speed were used to describe the weather (in the following referred to as ‘weather’ variables), and rainfall was considered separately. Weather data was collected centrally at the weather station at the Fort Grey homestead whereas rainfall was recorded for each individual site. Due to equipment failure temperature, humidity and wind-speed recordings were available for the first two sessions only. A number of rainfall variables were derived from the actual rainfall records; these were two and five month rainfall lags and rainfall over the past two, four and eight months.

A Principal Component Analysis (PCA) with a subsequent Varimax rotation was performed on the variable lists for habitat and ‘weather’ prior to regression to account for possible co-linearity and reduce the number of predictor variables. Factors with an Eigenvalue > 1 were considered for further analysis. Linear stepwise multiple regressions with the PCA factors or rainfall as the independent variables were then performed for each of the dependent variables (lizard abundance, richness, diversity, evenness, abundance of the ten most common species, House Mouse abundance). Adequate sample size (at least 15 samples per predictor variable is recommended by Field (2000)) and efforts to maximize the same put restrictions on the analysis. It prevented session specific analysis (i.e. all samples across all sessions were included in analysis) and also required that variable groups (microhabitat, weather and rainfall) had to be analysed separately as the ‘weather’ and rainfall dataset included a number of missing values which would have reduced the sample. As a result of the inclusion of all sessions, the habitat analysis could only detect broad correlations and would not have detected season specific habitat correlations (e.g. vegetation cover and the shade it provides is likely to have a greater impact in summer than in winter). The separate analysis of microhabitat, ‘weather’ and rainfall variables determined if and then which variables within each group had predictive power but not whether, for example, rainfall had a bigger influence than habitat. The relation between habitat and lizards was

- 65 -

Residual effects of grazing and artificial watering points on flora and fauna examined for the total lizard abundance, species richness, diversity and evenness and for the abundances of the ten most common lizard species.

4.5.3 Results

Invertebrates A total of 6190 invertebrates were recorded and identified to belong to one of 14 morphogroups. Ants were by far the most frequently captured invertebrates, representing nearly 63 % of all captures, spider captures were close to 14 % and crickets, grasshoppers and locusts comprised 7 % of all captures (Table 4.6).

Table 4.6: Invertebrate captures. Number and percentage of captures for invertebrate groups at disturbed and undisturbed and all sites combined.

Disturbed sites Undisturbed sites Overall

Taxonomic Morpho- Number Percent Number Percent Number Percent group group captured Total captured Total captured Total Formicidae Ants 2137 65.7 1778 60.6 3915 63.2 Araneae Spiders 423 13.0 449 15.3 872 14.1 Caelifera Crickets 167 5.1 264 9.0 431 7.0 Coleoptera Beetles 167 5.1 104 3.5 271 4.4 Blattodea Cockroaches 84 2.6 103 3.5 187 3.0 Thysanura Silverfish 114 3.5 63 2.1 177 2.9 Scorpiones Scorpions 52 1.6 49 1.7 101 1.6 Lepidoptera Moths 38 1.2 41 1.4 79 1.3 Chilopoda Centepedes 26 0.8 29 1.0 55 0.9 Apoidea Bees 24 0.7 25 0.9 49 0.8 Heteroptera Bugs 10 0.3 14 0.5 24 0.4 Other Other 13 0.4 16 0.5 29 0.5

The trappable invertebrate assemblage changed significantly over the trapping sessions

(PERMANOVA: F3, 64 = 12.144, p = 0.001) with each session being distinct as revealed by a subsequent pair-wise comparison. The difference between sessions is clearly illustrated by the separation of the data points into sessions as seen in a three- dimensional MDS-plot (Figure 4.10). The invertebrate groups responsible for these differences between sessions as analysed by a SIMPER varied between session combinations, but members of the Formicidae, Araneae and Caelifera were identified most frequently as the contributing groups to the differentiation between sessions one, two and four whereas Lepidoptera were most important in distinguishing session three from all others (Appendix 9). These invertebrate groups showed the highest correlation (Pearson correlation) with the axis in the plot and were added as vectors to the plot for

- 66 -

Residual effects of grazing and artificial watering points on flora and fauna better illustration. The length and direction of the vectors indicate the strength and sign, respectively, of the relationship between the invertebrate groups and the MDS axes. In view of such variation over time, the sessions were analysed separately.

3D Stress: 0.1

3 4 Lepidoptera3 4 4 Araneae 3 2 4 1 4 3 2 4 3 1 4 3 2 3 2 Caelifera 11 3 1 4 2 2 1 1 2 Formicidae 2 1

Figure 4.10: Temporal variation in the invertebrate assemblage. MDS-diagram of the invertebrate assemblage (based on square-root transformed abundances, Bray-Curtis similarities). Numbers represent trapping sessions. Vectors symbolize the relationship of the highest correlating (Pearson) invertebrate groups with the plot axis.

Comparison at disturbance level In no session was the invertebrate assemblage distinct between the levels of disturbance

(PERMANOVA, Session 1: F1, 6 = 0.665, p = 0.710; Session 2: F1, 6 = 0.737, p = 0.610;

Session 3: F1, 6 = 1.292, p = 0.180; Session 4: F1, 6 = 0.974, p = 0.500). Accordingly no grouping with disturbance level is obvious in the MDS-plots (Figure 4.11). The total captures of invertebrates, the number of invertebrate groups and their diversity were also similar for disturbance levels in each session. The richness of invertebrate groups was lower and evenness higher at disturbed sites in one of the sessions, but such

- 67 -

Residual effects of grazing and artificial watering points on flora and fauna isolated findings do not have much weight. Similarly, the few differences that occurred for individual invertebrate groups between disturbance levels were inconsistent between sessions and are thus of little relevance. a) Session 1 b) Session 2

3D Stress: 0 3D Stress: 0

D_2 U_3 D_1 U_3 D_3 D_4U_4 U_4 D_4 U_2 D_1 U_1 U_1 D_3 U_2D_2

c) Session 3 d) Session 4 3D Stress: 0.03

3D Stress: 0.01

U_4 U_3 D_2 U_1 D_1 D_3 D_4

U_2 D_2 D_1 D_4 U_2 D_3 U_1 U_4 U_3

Figure 4.11: Comparison of the invertebrate assemblage between sites. MDS-Diagrams of invertebrate assemblage at site level for each session. The diagrams are based on similarity matrix (Bray-Curtis coefficient) of invertebrate abundances (square-root transformed). D (filled circle) = disturbed, U (open diamond) = undisturbed.

Comparison at site level A number of significant results occurred between individual sites with regard to the overall invertebrate assemblage (Table 4.7). In particular site one in both disturbance

- 68 -

Residual effects of grazing and artificial watering points on flora and fauna groups was different to other sites, usually to the respective sites three and four. These results are reflected in the distribution of sites in the MDS-plots (Figure 4.11) where especially in session two and four sites D1 and U1 are furthest apart from sites D4 and U4. The three most important invertebrate groups driving those differences in the assemblages were the Formicidae and Araneae (all sessions), Caelifera (session one), Coleoptera (session two), Lepidoptera (session three) and Thysanura (session four) as identified by the SIMPER-functionality in PRIMER (Appendix 7). The discrepancy between the sites furthest apart spatially, is also evident in the significant results for individual community variables and abundances of individual invertebrate groups when tested separately in the various sessions (Appendix 10). Especially frequent were differences between sites in spider and cricket captures. The sites that differed varied with the sessions, but generally sites one and/or two opposed sites three and/or four.

Table 4.7: Significant differences in the invertebrate assemblage between sites. Significant results of PERMANOVA analysis.

Status Site Session t p_MC Disturbed 1&2 2 2.345 0.040 Disturbed 1&4 2 2.343 0.029 Disturbed 1&3 4 2.063 0.036 Disturbed 1&4 4 2.240 0.027 Undisturbed 2&4 1 2.185 0.028 Undisturbed 1&4 2 1.853 0.042 Undisturbed 1&3 4 2.021 0.038 Undisturbed 1&4 4 1.981 0.041

Lizards Twenty-one species of lizard were identified and a total of 1159 lizard captures recorded (Table 4.8). By far the most abundant species was the skink Ctenotus taeniatus (323 captures) closely followed by the skink Ctenotus schomburgkii (208 captures) and the dragon Ctenophorus pictus (181 captures). Those three species together accounted for 61 % of all reptilian captures.

- 69 -

Residual effects of grazing and artificial watering points on flora and fauna

Table 4.8: Composition of the lizard assemblage. Number and percentage of captures for disturbed and undisturbed sites and all sites combined. Taxonomy and nomenclature after Wilson & Swan (2008)

Disturbed sites Undisturbed sites Overall

Abundance Percent Abundance Percent Abundance Percent Scientific name Total Total Total Ctenotus taeniatus 163 27.4 160 28.3 323 27.9 Ctenotus schomburgkii 109 18.4 99 17.5 208 17.9 Ctenophorus pictus 94 15.8 87 15.4 181 15.6 Lucasium damaeum 67 11.3 86 15.2 153 13.2 Lerista labialis 29 4.9 29 5.1 58 5.0 Ctenophorus fordi 31 5.2 18 3.2 49 4.2 Eremiascincus 19 3.2 21 3.7 40 3.5 fasciolatus Nephrurus levis 21 3.5 10 1.8 31 2.7 Rhynchoedura ornata 16 2.7 15 2.7 31 2.7 Ctenophorus nuchalis 12 2.0 16 2.8 28 2.4 Heteronotia binoei 10 1.7 5 0.9 15 1.3 Ctenotus regius 71.250.912 1.0 Pogona vitticeps 30.550.980.7 Varanus gouldii 20.350.970.6 Lucasium stenodactylum 20.330.550.4 Menetia greyii 30.510.240.3 Lucasium byrnei 20.300.020.2 Gehyra variegata 20.300.020.2 Lerista aericeps 10.200.010.1 Morethia adelaidensis 10.200.010.1

TOTAL 594 565 1159

Like the invertebrate community the trappable lizard community changed significantly over the trapping sessions (PERMANOVA: F3, 28 = 5.486, p << 0.001) which is well illustrated by the clear separation of sessions in the three-dimensional MDS-plot (Figure 4.12). All sessions were distinct from each other with the exception of session two and three, which were similar in their lizard assemblage. As for the invertebrate assemblage the variation between sessions in the lizard assemblage required separate analyses. The lizard species that were primarily responsible for the heterogeneity between sessions (as identified by SIMPER analysis) were added as vectors to the plot for better illustration. The length and direction of the vectors indicate the strength and sign, respectively, of the relationship between the lizard species and the MDS axes.

- 70 -

Residual effects of grazing and artificial watering points on flora and fauna

3D Stress: 0.11

Ctenophorus nuchalis Ctenophorus pictus Ctenotus taeniatus 1 11 3 2 2 1 11 2 1 2 3 Lerista labialis 44 4 2 3 3 1 4 4 22 3 3 Lucasium damaeum 2 3 4 4 4 3

Figure 4.12: Temporal variation in the lizard assemblage. MDS-diagram of the lizard assemblage (based on square-root transformed abundances, Bray- Curtis similarities). Numbers correspond to trapping sessions.

Comparison at disturbance level Graphic exploration of the lizard assemblage (Figure 4.13) did not suggest any separation with disturbance level and indeed disturbed and undisturbed sites did not differ in their overall lizard community in any session (PERMANOVA, Session 1:

F1, 6 = 0.526, p = 0.657; Session 2: F1, 6 = 0.836, p = 0.573; Session 3: F1, 6 = 0.248, p = 0.888; Session 4: F1, 6 = 0.573, p = 0.772). Likewise there was no significant effect of disturbance on the total abundance of lizards or the lizard community variables (species richness, diversity, evenness). The total lizard biomass showed a trend only in session four when disturbed sites had a higher biomass than undisturbed sites (ANOVA:

F1, 6 = 4.943, p = 0.068). Disturbance levels were also indistinguishable in the abundance of captures in the groups of scincids, geckos and agamids, captures in the genus Ctenotus and Ctenophorus, the abundance of individual lizard species and the ratio of agamids (Ctenophorus) to scincids (Ctenotus).

- 71 -

Residual effects of grazing and artificial watering points on flora and fauna a) Session 1 b) Session 2

3D Stress: 0.01 3D Stress: 0.01

D_2 D_4D_3U_2U_4U_3 U_3 U_4 U_1 U_2 U_1 D_4 D_1 D_3

D_2 D_1

c) Session 3 d) Session 4 3D Stress: 0.01 3D Stress: 0.02

U_4 U_1 D_4 D_3 U_3D_3U_2 U_1 D_4D_2 D_2 U_4 U_2 D_1 D_1 U_3

Figure 4.13: Comparison of the lizard assemblage between disturbance levels. MDS-Diagrams of lizard assemblage at site level for each session. The diagrams are based on similarity matrix (Bray-Curtis coefficient) of lizard abundances (square-root transformed). D (filled circle) = disturbed, U (open diamond) = undisturbed.

Comparison at site level Despite the clear separation of some sites, in particular sites D1 and U1 in the MDS diagrams (Figure 4.13 a – d)and the highly significant interaction effect between the factors ‘status’ and ‘site’ in each session, the pair-wise tests found few differences between the sites within each disturbance group with regard to the lizard assemblage. Only sites one and two within the undisturbed group differed in two of the sessions

- 72 -

Residual effects of grazing and artificial watering points on flora and fauna

(PERMANOVA: t4 = 2.567, p = 0.014 and t4 = 1.992, p = 0.040). Further site differences, however, may exist independently of disturbance level. A number of differences between sites existed in community variables and in the abundances of some of the lizard groups and/or species (Appendix 8). Most of those results, however, only occurred in one session and thus have little relevance. Only the uneven distribution of the agamid C. fordi and the scincid C. taeniatus between sites was temporally consistent. Ctenophorus fordi was restricted to just two neighbouring sites, occurring nowhere else but at site one in each disturbance group. For C. taeniatus the site effect was significant but which sites were responsible for the spatial variation remained unclear as sites had overlapping 95 % confidence intervals.

Small mammals The small mammal community comprised five species (Table 4.9). The House Mouse was the dominant species with 65 % of all captures. Only one individual was captured for the Central short-tailed Mouse (Leggadina forresti), and the Dusky Hopping Mouse (Notomys fuscus). Small mammal captures were generally low and also very variable over time (Table 4.9). In total seventy-one small mammal captures were recorded and five of those were recaptures that had been marked in a previous trapping session. An additional ten animals were caught which were recaptures within the same trapping sessions and were removed from the dataset. In session one just five small mammals were caught, a total of 18 in session two and three, and 38 in session four. The increase in numbers in session two was mainly due to increased captures of the Sandy Inland Mouse (Pseudomys hermannsburgensis), the higher captures in session three and four were due to increased House Mouse abundance.

Comparison at disturbance level The small mammal assemblage was not significantly different between disturbed and undisturbed sites (PERMANOVA: F1,6 = 0.527, p = 0.739). This result is supported by the MDS-diagram (Figure 4.14) which does not show any grouping according to the disturbance level. The abundance of House Mice was homogeneous between disturbance levels as was the ratio of House Mice to native mice.

- 73 -

Residual effects of grazing and artificial watering points on flora and fauna

Comparison at site level The small mammal community was similar across all sites as was the abundance of the most abundant small mammal species, the House Mouse. The captures of all other species were too low to be analysed.

Table 4.9: Composition of the small mammal assemblage. Abundance and percentage of total of small mammal captures at disturbed and undisturbed sites, and overall.

Disturbed sites Undisturbed sites Overall

Number Percent Number Percent Number Percent Species Common name captured Total captured Total captured Total Mus musculus House Mouse 27 75.0 20 57.1 47 66.2

Pseudomys Sandy Inland 7 19.4 10 28.6 17 23.9 hermannsburgensis Mouse Sminthopsis Fat-tailed Dunnart 1 2.8 4 11.4 5 7.0 crassicaudata Leggadina forresti Central short- 12.800.011.4 tailed Mouse Notomys fuscus Dusky Hopping 00.012.911.4 Mouse TOTAL 36 35 71

3D Stress: 0

D_2 U_2 D_3U_4U_3

U_1 D_4

D_1

Figure 4.14: Comparison of the small mammal assemblage between sites. MDS-Diagrams of the small mammal assemblage at site level. The diagrams are based on similarity matrix (Bray-Curtis coefficient) of small mammal captures (square-root transformed), pooled over sessions. D (filled circle) = disturbed, U (open diamond) = undisturbed.

- 74 -

Residual effects of grazing and artificial watering points on flora and fauna

Foxes, Cats and Rabbits A total of 3456 track plots were set and investigated over the study period on 280 occasions (visits to dunes). Rabbits were quite abundant and visited 10.3 % of all track plots (Table 4.10), and were present on 53.2 % of possible occasions. Foxes visited 1.9 % of all plots and were present on 5.7 % of occasions. Cats visited the track plots rarely (0.2 % of plots) with their tracks found on nine plots on 3.2 % occasions.

Table 4.10: Activity of Foxes, Cats and Rabbits. Intensity and frequency of site visitations of vertebrate pests at disturbed, undisturbed and all sites together measured as the number of visited track plots.

Intensity Frequency Species Disturbed Undisturbed Overall Disturbed Undisturbed Overall Cat5 495 49 Fox 52 14 66 35 12 47 Rabbit 164 193 357 75 74 149

Comparison at disturbance level The frequency of Fox visitation was significantly higher at disturbed sites (ANOVA: = F1, 6 8.451, p = 0.027 (transformation did not achieve normalization; Levene’s test: = F7, 88 5.023, p << 0.001)) and likewise the intensity of Fox visitation showed a strong trend to be higher at disturbed sites (ANOVA: F1, 6 = 5.682, p = 0.054 (transformation did not achieve normalization; Levene’s test: F7, 88 = 6.781, p << 0.001)) (Figure 4.15). The Rabbit visitation frequency and intensity was homogeneous between sites.

Comparison at site level The intensity of Rabbit visitations was heterogeneous between sites with sites one in each disturbance group visited less by Rabbits than other sites in the same group

(ANOVA F6, 88 = 6.021, p << 0.001). The frequency of Rabbit visitation and measures of Fox visitations were similar between all sites. Cat visitation frequency as well as the intensity differed significantly between sites (ANOVA F6, 88 = 5.583, p << 0.001 and

F6, 88 = 6.021, p << 0.001 (square-root transformed) respectively) with less frequent and intense visits at sites one in each disturbance group compared to the other sites. However, considering the generally low abundance of Cats (total number of track plots visited: n = 9) this result is not representative.

- 75 -

Residual effects of grazing and artificial watering points on flora and fauna a) b)

Figure 4.15: Comparison of Rabbit, Fox and Cat activity between disturbance levels. Mean frequency (a) and mean intensity (b) of Rabbit, Fox and Cat visitation at disturbed and undisturbed sites. Error bars represent 95 % confidence interval. ** indicates a significant difference, * a trend.

Effects of microhabitat and weather on the temporal and spatial variation of small vertebrates The previous analyses have identified considerable differences between sampling sessions and between sites in the small vertebrate assemblages and captures of some species. The following section investigates whether this temporal and spatial variation can be explained through microhabitat and environmental variables.

Lizard assemblage: The lizard assemblage was highly correlated with ground cover (i.e. habitat variables) which account for 73 % of the variation in the dataset. Two axes (x-axis mainly driven by the amount of green pasture and the y-axis mainly driven by the density of dead wood) explain 30 % and 22 % of the variation, respectively (Figure 4.16). Green pasture and the x-axis is most closely associated with the spread of samples according to session and thus represents the temporal variation, whereas dead wood density and the y-axis explain part of the spread of samples within sessions and thus the spatial variation.

- 76 -

Residual effects of grazing and artificial watering points on flora and fauna

20 1 41 1 4 4 1 12 4 4 2 1 2 2 2 4

0 2 1 3 1 3 4 3 3 3 4 3

-20 2

3 2 green pasture cover dead wood density -40 3 dbRDA2 (30.2% of fitted, 22% of total variation) total of 22% fitted, of dbRDA2 (30.2%

-60 -40 -20 0 20 40 dbRDA1 (41.5% of fitted, 30.3% of total variation)

Figure 4.16: Temporal and spatial variation in the lizard assemblage and habitat variables as causative factors. Similarity in the lizard assemblage between samples (data square-root transformed, Bray Curtis Similarity). Numbers represent sampling sessions (1-4). The length and direction of the vectors indicate the strength and sign, respectively, of the relationship between that variable and the axes. a) Vectors are habitat factors (Pearson correlation >0.50). Green pasture cover explained the greatest part of the variation and was superimposed as bubbles. Bubble-size correlates with the amount of green pasture measured.

Apart from green pasture, a great part of the temporal variation in lizard captures seemed to have been influenced by ‘weather’ factors (40.3 %). Amongst them humidity 2 showed the highest correlation (R = 0.228, F1, 14 = 4.13, p = 0.001). The cumulative rainfall in the previous four months was the highest correlated with the lizard 2 assemblage (R = 0.328, F1, 14 = 6.83, p << 0.001) of all rainfall variables.

Associations between lizard species: A PCA was performed with the ten most common lizards in the assemblages to reveal potential associations or disassociations between species which may indicate habitat separation and/or interspecific competition between the species concerned. The majority of species loaded onto a single factor

- 77 -

Residual effects of grazing and artificial watering points on flora and fauna

(Table 4.11) and all with a positive sign. A second factor was formed by the species pairs Ctenophorus pictus and Ctenotus taeniatus and a third factor by Nephrurus levis and Eremiascincus fasciolatus.

Table 4.11: Associations between lizard species as identified using a Principal component analysis. Presented is the rotated component matrix (Varimax Rotation with Kaiser Normalization; Rotation converged in 5 iterations)

Component Species 1 2 3 Rhynchoedura ornata 0.879 Lucasium damaeum 0.716 Ctenotus schomburgkii 0.706 Lerista labialis 0.697 Ctenophorus fordi 0.660 Ctenophorus nuchalis 0.547 Ctenophorus pictus 0.917 Ctenotus taeniatus 0.862 Nephrurus levis 0.809 Eremiascincus fasciolatus 0.804

Correlation with habitat-, weather and rainfall variables Lizards and Habitat: A number of significant correlations were found between habitat factors (see Table 4.12 for loading of factors) and lizard community measures and capture rates of selected species. In particular factor one, mainly related to little green pasture cover but high dry pasture and litter cover and high patchiness of dry pasture, was often part of the final regression model. However, most of the factors could explain relatively little of the overall variation in lizard community measures (R2 < 0.5) and are thus influenced by habitat variables. Only lizard diversity (correlation with factor 1 and 5) and lizard species richness (correlation with factor 1) had a correlation coefficient greater than 0.5 (Table 4.13).

- 78 -

Residual effects of grazing and artificial watering points on flora and fauna

Table 4.12: Floristic habitat factors identified with a Principal component analysis used for subsequent linear regression analysis. Presented is the rotated component matrix (Varimax Rotation with Kaiser Normalization; Rotation converged in 7 iterations)

Component Habitat variable12345 green pasture cover -0.911 dry pasture cover 0.847 dry pasture CV -0.675 green pasture CV 0.612 dead wood cover 0.937 dead shrub density -0.904 dead shrub cover 0.695 -0.586 living shrub cover 0.549 litter CV 0.814 bare ground 0.796 litter cover 0.6 -0.65 living shrub density 0.94 bare ground CV 0.818 dead wood density 0.665

Table 4.13: Correlation of lizard community variables and individual species abundances with habitat factors. Final stepwise regression model scores. R2 is an indicator of the proportion of the variance explained. The Beta coefficient refers to a negative or a positive correlation to the dependent variable.

Dependent ANOVA 2 Beta- Factor df1, df 2 F R Variable p-Value coefficient

Total abundance 1 1, 30 20.963 << 0.001 0.411 0.641 0.745 Diversity 1, 5 2, 29 23.880 << 0.001 0.622 0.259 Eveness 1 1, 30 9.891 0.004 0.248 -0.498 Richness 1 1, 30 44.024 << 0.001 0.595 0.771 0.419 Ctenophorus fordi 5, 3 2, 29 5.857 0.007 0.288 0.335 0.468 Ctenophorus nuchalis 3, 1 2, 29 9.739 0.001 0.402 0.428 Ctenophorus pictus not sign. 0.442 Ctenotus taeniatus 1, 2 2, 29 7.221 0.003 0.332 -0.371 Ctenotus schomburgkii 1 1, 30 5.165 0.030 0.147 0.383 Lucasium 0.621 1, 2 1, 30 13.682 << 0.001 0.485 damaeum 0.315 Eremiascincus 0.585 1, 3 2, 29 10.798 << 0.001 0.427 fasciolatus -0.290 Lerista labialis 1 1, 30 17.563 << 0.001 0.369 0.608 Nephrurus levis 1 1, 30 8.243 0.007 0.216 0.464 Rhynchoedura 0.376 1, 4 2, 29 5.353 0.011 0.270 ornata 0.358

Lizards and ‘weather’: All measured weather variables were co-linear and thus comprised a single factor, on which temperature related variables loaded highest

- 79 -

Residual effects of grazing and artificial watering points on flora and fauna

(positive sign) followed by humidity (negative sign). Only the variation in the total lizard abundance and species richness could be explained to more than 50 % by the ‘weather’ factor (Table 4.14 b). The considerable lack of correlations between lizard species abundances and the weather factor, and, in those cases where significant correlations were found, the low correlation coefficients, are slightly surprising at first glance. After all ectothermal lizards and their activity are heavily dependent on ambient temperatures and lizard activity varies greatly over seasons, with most activity in summer and least activity in winter. Since sampling sessions were not analysed separately, this might mean that considerable spatial variation (i.e. differences between sites) exists in the dataset and this cannot be well explained by the ‘weather’ factor.

Table 4.14: Correlation of lizard variables with temperature, humidity and wind speed. a) Results of PCA: All measured weather variables load onto one factor. b) Linear regression model 2 scores for correlation of lizard variables with the weather factor. Df 1 = 1, df 2 = 14. R is an indicator of the proportion of the variance explained. The Beta coefficient refers to a negative or a positive correlation to the dependent variable. a) Component Meteorological Variable 1 Max Temperature 0.981 Min Temperature 0.958 Heat Index mean 0.947 Humidity mean -0.816 Windspeed mean 0.545 b) Dependent ANOVA Beta- F R2 Variable p-Value coefficient

Total abundance 30.881 << 0.001 0.688 0.829 Diversity 11.163 0.005 0.444 0.666 Eveness 9.748 0.007 0.410 -0.641 Richness 23.639 << 0.001 0.628 0.792 Ctenophorus fordi not sign. Ctenophorus nuchalis not sign. Ctenophorus pictus not sign. Ctenotus taeniatus 7.013 0.019 0.334 0.578 Ctenotus schomburgkii not sign. Lucasium damaeum 29.521 << 0.001 0.678 0.824 Eremiascincus fasciolatus not sign. Lerista labialis 12.836 0.003 0.478 0.692 Nephrurus levis not sign. Rhynchoedura ornata 8.552 0.011 0.379 0.616

Lizards and rainfall: Due to the small sample size the regression results have exploratory rather than predictive value. Compared to the habitat and weather variables

- 80 -

Residual effects of grazing and artificial watering points on flora and fauna the rainfall variables generally achieved a high R2 value (Table 4.15) which indicates that they are able to explain a great part of the variation in the respective lizard variables. In particular, variation in the total lizard abundance and the abundances of the most common lizard species C. taeniatus and C. schomburgkii were correlated with rainfall variables. Unfortunately the rainfall variables entered into the model varied with each session as a result of the short study period and the rarity of rainfall events. The least variables were available for the early session (April), with both a higher number and more long-term variables for the later sessions (October). In the October session cumulative rainfall in the previous eight or four months were the best predictor variables. The long-term (relatively speaking) rainfall history of sites and the cumulative rainfall as opposed to time since the last significant rainfall event, therefore appear to be more important in explaining the variation in lizard variables than the short-term rainfall variables.

Table 4.15: Correlation of lizard variables with rainfall. Significant results of multiple stepwise regression model scores. R2 is an indicator of the proportion of the variance explained. The Beta coefficient refers to a negative or a positive correlation to the dependent variable.

Dependent ANOVA 2 Beta- Session Factor df1, df 2 F R Variable p-Value coefficient April Total abundance 2 months lag 1, 6 19.497 0.004 0.765 0.874 July Total abundance 5 months lag 1, 6 10.030 0.019 0.626 0.791 October Total abundance cumulative 8 months 1, 6 49.389 << 0.001 0.892 0.944 October Diversity cumulative 4 months 1, 14 34.702 << 0.001 0.713 0.844 April Evenness 2 months lag 1, 6 31.887 0.001 0.842 -0.917 July Evenness 5 months lag 1, 6 9.156 0.023 0.604 -0.777 October Richness cumulative 4 months 1, 14 80.712 << 0.001 0.852 0.923 April Ctenophorus fordi cumulative 2 months 1, 6 6.974 0.038 0.538 -0.733 April Ctenotus taeniatus 2 months lag 1, 6 30.702 0.001 0.837 0.915 July Ctenotus taeniatus 5 months lag, 2, 5 18.953 0.005 0.940 1.426 cumulative 4 months -0.702 October Ctenotus taeniatus cumulative 8 months 1, 6 15.452 0.008 0.720 0.849 July Ctenotus schomburgkii cumulative 2 months 1, 6 25.526 0.002 0.810 0.900 October Ctenotus schomburgkii cumulative 8 months 1, 6 40.102 0.001 0.870 0.933

House Mice and distribution determinants: No significant correlations were found between House Mouse abundance and habitat or weather variables. Cumulative rainfall in the previous eight months explained a great proportion of the variation in the 2 abundance of House Mice (linear regression: F1, 6 = 6.991, p = 0.039, R = 0.535, Beta coefficient – 0.732) but was surprisingly negatively correlated; i.e. the more rain, the less House Mice.

- 81 -

Residual effects of grazing and artificial watering points on flora and fauna

4.6 Overall Discussion and Conclusion

4.6.1 General As is typical for a desert environment ground cover, especially green ground cover, was generally scarce and highly variable over time. Reports of large variation in herbage biomass are not uncommon in arid Australian rangelands (Wilson and Mulham 1980; Robertson 1987), with 10 to 100-fold increases from drought to high rainfall conditions, even under grazing. In accordance with this, green pasture (forbs, grasses and copperburrs) was the main driver for temporal variability in the ground cover and was explained through unevenly distributed rainfall patterns. Of the plants present (living or dry) ephemeral ‘forbs’ were the most abundant in all sessions and the main contributor to the broader category ‘pasture’. Grasses were extremely sparse (< 1 % of ground cover). The dunes in the study area were quite abundantly vegetated with a variety of shrub species and a high proportion of dead woody material.

The trappable invertebrate assemblage was dominated by typical ground-dwelling arthropods, the ants and spiders. These were not only the most abundant invertebrates captured but were also identified as the main drivers for seasonal as well as site- differences in the invertebrate assemblage. Scincid lizards of the genus Ctenotus and agamids of the genus Ctenophorus were the most abundant species in the lizard assemblage. The small mammal assemblage consisted of six species of which only two, the House Mouse and the Sandy Inland Mouse, were common. Two species were represented by only one individual.

Temporal variability in the capture rates of certain invertebrate groups and small vertebrate species was high. Seasonal temperature differences as well as variable amounts of precipitation received in the months prior to sampling are obvious explanations. Temperature influences the activity times, mobility and reproduction of animals, with many species being less active in the colder conditions of winter and reproduction limited to the warmer months. In arid regions where fauna is independent of drinking water, rain affects animal species significantly but mainly indirectly through the triggering of vegetation growth. This enhances the resource base and thereby the activity and the reproduction of animals and ultimately the distribution and composition of faunal assemblages.

- 82 -

Residual effects of grazing and artificial watering points on flora and fauna

4.6.2 Vegetation There was little evidence to suggest the persistence of a grazing effect in vegetation parameters. Variation between sites and a small sample size masked significant differences in ground cover with disturbance level. A potential disturbance effect was only detectable by the consistency of mean differences in bare ground, dry pasture and litter across sessions. Precipitation patterns were identified as the underlying causative factor for session differences, but the cause of site heterogeneity was not obvious. There was a hint of a north-south axis (e.g. commonalities between U_1 and D_1, and dissimilarities to U_4 and D_4) which may suggest a residual signature of the travelling stock route (TSR) passing between U_1 and D_1 (sites 1 and 2 on Figure 2.3). Grazing pressure on the TSR would have been short but very intense early in its history.

Of the shrub variables only dead wood was significantly different between disturbance levels and had a higher mean density on disturbed sites. Dead material, in particular something as large and solid as wood, is extremely slow to decay in the dry conditions of a desert environment and it thus accumulates over time. As such dead wood is likely to represent an exacerbation of the potential trend indicated by the higher shrub abundances on disturbed sites and possibly hints at former shrub encroachment.

4.6.3 Fauna As for the vegetation, there was little evidence to suggest the persistence of a grazing effect in parameters of the fauna. The invertebrate, lizard and small mammal assemblages and other faunal variables were virtually indistinguishable between disturbance levels. However, multiple differences occurred in many faunal aspects between seasons and individual sites. The variability between sites in addition to the small sample sizes of many species in particular those of small mammals, may have masked any significant differences which could be attributed to the different disturbance levels. Most differences between sites were temporally and spatially inconsistent. It is likely they were a reflection of the natural temporal and spatial variability in the faunal communities. Noticeable was the frequent distinctiveness in faunal attributes of site one in each disturbance group. Ctenotus fordi was only caught frequently on sites D_1 and U_1 but was absent from most other sites whereas C. taeniatus, the most common skink, was remarkably less abundant on those sites. Even though site-differences in floral variables did not show the same marked separation of sites one in each

- 83 -

Residual effects of grazing and artificial watering points on flora and fauna disturbance group (most sites were distinct and differed from all others), some ground cover and shrub variables were able to explain at least some of the discrepancy in the distribution of the two lizard species. The two species apparently have opposing habitat preferences as the sign of the correlation with microhabitat factors was positive for one species but negative for the other. Ctenotus taeniatus (but not C. fordi) abundance was also highly correlated with rainfall (two and five months lag, four and eight months cumulative rainfall) and sites D_1 and U_1 had received least rainfall over the study period. Apart from the lizard assemblage, those sites also had greater House Mouse numbers than other sites. House mouse numbers were found to be negatively correlated with rainfall. Thus the likely explanation for the distinctness of the lizard and mammal assemblage of sites D_1 and U_1 is the rainfall history, which would affect microhabitat factors and resource availability and thereby influence fauna distribution and abundance indirectly.

4.6.4 Non-native plants and animals Of the two introduced plants found in the study area, the Camel Melon was present but restricted to few and isolated individuals. Camel Melon is typically found on roadsides and in cultivated or disturbed areas (Cunningham et al. 1992). Similarly the Wild Turnip is common along roadsides, around waste deposits, Sheep yards and areas of habitation. It is prevalent also on sandy soils, particularly in severely grazed or disturbed areas. After cool-season rainfall, disturbed areas are often densely infested with this species to the almost complete exclusion of other plants (Green 1989; Cunningham et al. 1992). Such dense infestations were not found in the study area but the plant was quite common with up to 0.82 plants per m2 and was very patchily distributed with some dune areas being practically devoid and others having large stands of it. Disturbed sites had a slightly higher abundance of the Wild Turnip but not significantly so and differences between individual sites were far more pronounced than differences between disturbance levels. Rainfall could not be identified as the underlying cause for the heterogeneity between sites. However, the germination of Wild Turnip may require a certain rainfall threshold which might have been met at all sites during this study so that no differences were identified.

Of concern was the dominance of the introduced House Mouse within the small mammal assemblage. During the sampling periods for this part of the study as well as in

- 84 -

Residual effects of grazing and artificial watering points on flora and fauna subsequent sampling periods (10 sampling periods) they accounted for an average of 73.7 % of rodent captures (min 29 %, max 100 %). The data could be biased due to a proportionally higher trappability of House Mice with (baited) Elliott traps compared to native small mammals (Read 1985). Nevertheless even when only pitfall captures are taken into account, House Mice represent 48 % (average of 10 sampling periods) of all rodents. In previous surveys in the study area with a nearly identical set-up and methods, the House Mouse represented on average 31 % (three sampling periods) of rodents in 2003/2004 (Dowle 2004) but 88 % (dune-habitat, one sampling period) and 56 % (interdune-habitat, one sampling period) in 2005 (DECC survey). Disturbance level had no impact on House Mouse abundance but significant differences existed on site level. In particular sites U_1 and D_1 had high abundances of House Mice. The temporal and spatial variation in House Mouse captures was negatively correlated with the rainfall received three to six months prior. This negative correlation was surprising as House Mice numbers and captures were expected to increase as a result of improved environmental conditions after rain. The high reproductive potential of House Mice should have allowed a population increase within a few months. The paucity of other small mammal species excludes interspecific competition as a causative factor and a possible explanation. It is possible that this result may have resulted from changes in the trappability with different environmental conditions. During dry times the mice needed to travel further distances to find sufficient food thus increasing the chances of being captured. When food was abundant, the mice could reduce their activity time and were thus less frequently captured. The provision of food in the form of baited Elliott traps would have acerbated the unequal capture rates. Food shortage increased the pressure for mice to find food and may have reduced their reluctance to enter a trap and thus resulted in a higher capture success of House Mice at dry sites and during dry sessions. It is a usually assumed that trappability remains constant when comparing captures between different sites, seasons or treatments. However faunal sampling techniques like trapping inherently introduce biases into the process of data collection and trapping results are not necessarily an accurate description of actual animal abundances. The result for House Mice may be an example of this. Limitations and biases introduced through faunal sampling techniques are a limitation to every study and have been acknowledged in here (see section 4.6.7).

- 85 -

Residual effects of grazing and artificial watering points on flora and fauna

Foxes and Rabbits were quite abundant in the area, as indicated by frequent and numerous visitations to track plots. Relatively few Cat footprints were recorded on the track plots but Cat activity may have been higher than this suggests. Cats are quite secretive and cautious animals and therefore difficult to monitor. Density estimates were not part of the aims of this study and thus no information can be given of how the Fox, Cat and Rabbit abundances in Sturt National Park compare to other areas. Interestingly, the frequency and intensity of Fox visitations was significantly higher, or showed a strong trend to be higher, at the disturbed sites, whereas homogeneity was suggested on the site level. Such a finding could be a remnant of historically higher and more stable Fox densities around watering points, which may persist even when the watering point has been closed. However, considering that the bores were rendered dysfunctional more than 35 years ago this seems unlikely given there was no such bias in the potential prey as measured in this study. Apart from the pastoral history several other potential explanations spring to mind including variation in the resource base, vegetation characteristics, distance to drinking water and distance to neighbouring properties and thus sources of immigration. Neither difference in the resource base nor in vegetation characteristics appears to be a likely explanation for the observation as the small mammal assemblage, the activity of Rabbits and all vegetation parameters were comparable between disturbance levels. Distance to the next reliable water source is also not a satisfactory explanation as Foxes can cover 10 km or so to the next watering point if required. Few locations in the study area would have been further than that from accessible water. The location of the monitoring sites in relation to the neighbouring properties and the dog fence may have influenced the immigration and/or emigration potential for Foxes and thereby influenced Fox activity. However, disturbed and undisturbed sites were situated in such a way that if such effects existed it should have had a similar effect on both site types. If the higher activity of Foxes at disturbed sites is not a chance result then some other explanation and unmeasured factor must be responsible for the differences. Regardless of the underlying cause of the observation, in its singularity it is probably of little relevance overall.

The presence and abundance of none of the non-native plants and animals could be correlated to the intensity of past grazing disturbance. Nevertheless the occurrence of Camel Melon and relatively high numbers of the Wild Turnip, Foxes, Cats, Rabbits and

- 86 -

Residual effects of grazing and artificial watering points on flora and fauna

House Mice collectively indicate a certain (at least historical) level of disturbance to the area.

4.6.5 Threatened grazing-sensitive species The fauna that persists in an area like Sturt NP is likely that which has been robust to the impact of livestock grazing. The presence and/or re-appearance of species believed to be threatened by habitat degradation through grazing may be an indication that an ecologically meaningful habitat improvement has been achieved since cessation of grazing.

Habitat alteration through the grazing activity of livestock is listed as a potentially threatening process for the lizard Ctenotus taeniatus listed as a vulnerable species in NSW (EPBC Act (2008b)). During the two years of trapping for this project as well as in previous studies in the same area (Dowle 2004; Scholze 2007) and a DECC survey (2006), the species was captured frequently and it often represented the most commonly captured lizard. In this study C. taeniatus was identified to be amongst the species that were the main drivers in the differences between individual sites and between sessions and thus appears sensitive to certain habitat characteristics. Even so its occurrence was not found to differ between sites with a different historical disturbance regime.

Pastoral activities are also listed as a threat to the Dusky Hopping Mouse (even though the mice have been recorded from severely grazed and degraded areas in South Australia (Moseby et al. 1999)). The rodent was listed as presumed extinct in NSW (no records since 1845) until its rediscovery in Sturt NP in 2003 (Dowle 2004). It is now listed as an endangered species under the Threatened Species Conservation Act. The survey in the present study indicates that the abundance of the Dusky Hopping Mouse fluctuates greatly in Sturt NP (see Chapter 6). It was undetectable in the beginning of this study and thus could not be considered in the investigation of potential residual impacts of pastoralism. However, the species became extremely abundant in the later stages of the study.

Another rodent, the Desert Mouse, Pseudomys desertor, is considered to be negatively affected by grazing impacts. It was found to significantly decline under grazing (Kutt and Woinarski 2007) and its decline and presumed extinction in western New South

- 87 -

Residual effects of grazing and artificial watering points on flora and fauna

Wales has been clearly linked to overgrazing (Krefft 1866; Dickman 1993). The capture of a male P. desertor during the study, put the species back on the map in NSW and extended its current known range by more than 100 km. Admittedly the capture of a single possibly transient individual and a male at that, does not provide a solid basis to presume a widespread and continued existence of the species in the area. Nevertheless, where there is one mouse there are likely to be more.

The abundance of C. taeniatus and the recent rediscovery of two rodent species, not recorded in NSW since the middle of the 19th century, are supporting evidence that effects threatening those species, amongst them grazing disturbance, have been mediated in Sturt NP since the Park came into existence and livestock was removed.

4.6.6 Timeframe for post-pastoralism recovery Many studies investigating the ‘recovery’ process that follows livestock removal and their cessation of grazing are conducted immediately after or within a few years of the reduction of grazing pressure or closure of watering points. Such studies generally fail to detect a recovery of the systems (e.g. a comprehensive study by Tongway, Friedel, Sparrow and Kinloch (Friedel et al. 2003; Tongway et al. 2003)). Much larger timescales are likely to be required to detect changes. For example, Valone et al. (2002) estimate that more than 20 years may be required to detect changes in vegetation, especially recovery of perennial grasses following the removal of grazing livestock. Friedel et al. (2003) conclude that the lack of recovery within short timeframes most likely reflects the persistence of fundamental changes to soil that have occurred as a consequence of grazing. Just as the recovery of vegetation depends on the recovery of the soil, faunal species and communities are likely to depend on the previous recovery of vegetation characteristics. Thus even longer timeframes are to be expected before the detection of recovery in the fauna.

The results of this study indicate that 35 years after the removal of livestock and the closure of watering points, historically heavily disturbed areas are nearly indistinguishable from relatively undisturbed areas across the suite of variables measured from flora and fauna. It thus appears that three decades provide a sufficient timeframe for the dissipation of grazing-induced disturbance in the wake of pastoralism

- 88 -

Residual effects of grazing and artificial watering points on flora and fauna without recourse to controlling the densities of large native herbivores, kangaroos, which may partially replace livestock.

This result contrasts somewhat with results from similar studies conducted in Sturt NP (Montague-Drake 2003; Dowle 2004) and summarised by Croft et al. (2007). They report a remaining signature on the vegetation surrounding AWP mainly in the form of a persistent reduced cover of palatable saltbushes within one kilometre of AWP in the stony downs on the eastern side of the Park, and an indication of residual invasive native shrub incursion around ground tanks compared to water-remote reference sites in the sand dunes. The choice of points (AWP types) of comparison provides an explanation for the disparity in results. In the current study only bores were considered, which were compared to relatively undisturbed sites at equal distances from any watering point historic or contemporary, whereas the above-mentioned studies compared a range of AWP (open and closed ground tanks, and closed bores). Possibly the piosphere effect was greater around ground tanks (tanks are often constructed in naturally more productive areas that ensure some run-off after rain to fill the dams) than around bores, and so it takes a longer timeframe for it to dissipate. Furthermore, Sheep were more likely to graze in the swales (interdunes) than the dunes and the dunefields were likely less productive for livestock than the chenopod shrublands (Montague- Drake’s study) and mulga woodlands (Central Australian studies) that currently are favoured for pastoralism. Scholze (2007) conducted a short unpublished study in the same sites as Dowle (see Croft et al. 2007) but in the swales rather than the dunes. Like Dowle’s dunes, those swales distant from an AWP had more vegetation cover but a lower density of shrubs (especially unpalatable invasive species). However, these vegetation differences did not translate to faunal differences (small mammals and lizards). Thus the dunefields of Sturt NP have proved more resilient to past grazing impacts than the stony downs.

4.6.7 Study limitations Friedel (1994) stresses that heterogeneity in rangelands occurs at many levels, both spatially and temporally and the scales of both in monitoring programmes will determine whether the community/landscape is perceived to be degrading or improving. In this study the survey sites were restricted to the sandy substrate of the sand dunes. This stratification eliminates a great part of the spatial heterogeneity that results from

- 89 -

Residual effects of grazing and artificial watering points on flora and fauna different soil surface types which can cause variable results or prevent the detection of any significant differences between various grazing regimes (Friedel 1994). On the other hand using such a narrow focus on one particular, even though ubiquitous, part of the landscape, significantly limits the conclusions that can be drawn from the study. A greater scope, including a broader range of land and soil types, however, was beyond the limits of this part of the thesis.

The timeframe of this study was limited to one year. A constraint imposed by the fact that the Fox/Rabbit control experiment was carried out on the same sites. The investigation of residual historical grazing impact had to be completed before any of the other treatments i.e. Fox and Rabbit control were begun as after the control of Rabbits and Foxes began, any differences between sites could have been either due to differences in the disturbance history (aim of part one of the thesis), due to the effects of Rabbit/Fox control or a combination of the two. Temporal variability is always an issue in ecological studies, regardless of the duration of a study. The year dedicated to this part of the thesis covered a range of conditions and thus sources of variability. Environmental conditions were generally very dry but the survey time also included two significant rain events.

The assumptions underlying this part of the study is that the landscape and, more precisely, the sand dunes actually were disturbed through livestock grazing and that the intensity of disturbance has been similar in all paddocks of the former station ‘Fort Grey’. Even though these are reasonable assumptions, no actual data on historical stocking rates are available to make deductions on the intensity of the livestock impact. Thus no conclusions can be drawn regarding the nature of the assumed differences between grazed and relatively ungrazed sites. The expectations were that undisturbed sites would be more heterogeneous and thus more diverse than the more heavily disturbed sites. However, if grazing pressure was low then this could have in fact led to increased heterogeneity relative to less grazed or ungrazed conditions (Earl and Jones 1996).

Faunal sampling techniques like trapping inherently introduce limitations and biases into the process of data collection and trapping results are not a pure reflection of actual animal abundances. They rather reflect the activity of animals which is at least

- 90 -

Residual effects of grazing and artificial watering points on flora and fauna influenced by the environmental conditions, food supplies and reproductive status of animals. Speed and type of movement, the intelligence level of species and their size, introduce considerable additional bias. Within a session all sites were trapped for the same number of trap nights with identical equipment and in as short a timeframe as possible. However, it was logistically impossible to monitor all sites concurrently. Thus daily variations in weather (temperature, humidity, cloud cover) and moon phase may have influenced faunal captures and confounded results. Whenever possible, these variations were considered as part of the analysis of causative factors of faunal distribution.

In parts, the study may have benefited from gathering more detailed data. Landsberg and Crowley (2004) criticise the lack of inclusion of short-lived plants (other than as part of a single functional group labelled ‘forbs’ or ‘ephemerals’) in pastoral monitoring programmes. Undeniably a detailed investigation on the basis of individual plant species may have given valuable information on differences in the plant community and for individual species. However, this is only achievable after effective rainfall which was sparse in this study. Some information was gathered at the species level after the few rainfall events that did occur, which was instructive but not definitive.

Similarly, in the fauna part, the study of invertebrates is likely to have provided considerably more insight if invertebrates had been identified to a lower taxonomic level. Ants in particular have been nominated as ideal candidates, for biological monitoring. But even if simplified sampling and sorting procedures (Andersen et al. 2002) had been applied, time-constraints and taxonomic challenges (the identification to species level requires specialist knowledge and resources such as keys, and ant taxonomy is in constant revision) prevented a more detailed investigation in this study which focused primarily on vertebrates.

4.6.8 Contribution of other herbivores than livestock to impacts of water- focussed grazing Apart from livestock other herbivores, feral and native, potentially contribute to the impacts of water-focussed grazing. In the study area, other grazing animals are the European Rabbit and two species of kangaroos. Elsewhere in the rangelands, feral populations of Goats, Camels, Donkeys and Horses and other macropods also

- 91 -

Residual effects of grazing and artificial watering points on flora and fauna significantly contribute to the grazing pressure. Rabbits are independent of drinking water and exhibit a warren-focused grazing pattern (Eldridge and Myers 2001) and thus do not contribute to any grazing gradients centred around watering points. Little direct research has been conducted to investigate if and to what extent kangaroos contribute to the water-focused grazing effect and this is assumed rather than proven. Four kangaroo species occur in western NSW, dominated by Red Kangaroos (Macropus rufus) and Euros (Macropus robustus erubescens) with fewer Eastern Grey Kangaroos (Macropus giganteus) and Western Grey Kangaroos (Macropus fuliginosus) which are at the western and eastern margins of their range, respectively. Grey Kangaroos have higher water requirements (Dawson et al. 1975; Dawson et al. 2006) than the other species and are generally more sedentary (e.g. Witte 2002) and thus the presence of watering points may be of greater importance in areas with predominantly Grey Kangaroos. In an area with the dominant species being the arid-adapted Red Kangaroo, it is less likely that kangaroos would have significantly contributed to the grazing disturbance around watering points. An intensive study in Sturt NP (Montague-Drake and Croft 2004) found that kangaroo densities were not significantly related to water proximity and did not differ between open and closed watering points. Radio-tracking of Red Kangaroos confirmed that most individuals do not exhibit water-focused grazing patterns and do not have home ranges centred on an artificial watering source. After the removal of stock and despite the closure of most of the AWP, kangaroos remained present in relatively high densities and have exhibited a continuing grazing pressure which is likely to be and have been independent of AWP.

4.7 Conclusion The signature of the piosphere on vegetation parameters around the closed bores was weak and only inferred from temporal consistency across a range of seasons. The similarity between disturbed and undisturbed sites either implies that the former have returned to a lightly grazed state or past livestock impacts were more ubiquitous and undisturbed sites are far from pristine. The latter is difficult to test as only the western deserts of Australia have never had pastoralism and even these have been invaded by feral Camels. The more optimistic view is that the signature of about a century of pastoralism does dissipate and it does not require direct management of large native herbivores to affect it.

- 92 -

Chapter 5: The effect of combined Fox and Rabbit reduction on small vertebrates and Cats

5.1 Introduction

5.1.1 The Red Fox - a brief profile The Red Fox (Vulpes vulpes) is a native species of the northern hemisphere - Eurasia, North Africa and North America. It may have been introduced to Australia as a hunting target as early as the middle of the 19th century. The wild population, however, appears to originate from releases in Victoria in the early 1870s (Rolls 1969). From its point of introduction in Victoria the Fox spread rapidly and 100 years later the Fox had reached its present distribution which includes most parts and most habitats of Australia except for the northern tropics and most offshore islands (Jarman 1986). Until recently the Fox had not successfully established in Tasmania, but evidence of Fox presence has risen since the late 1990s (Saunders et al. 2006).

Foxes are solitary hunters that stalk and ambush their prey rather than extensively chasing it (Henry 1986). They readily scavenge if the opportunity arises and they are known for their food caching behaviour for later use. Their broad, locally and seasonally variable, diet includes insects, fruits, small and medium sized mammals as well as the juveniles of large mammals (e.g. kangaroos), birds, reptiles, amphibians, carrion and human rubbish (see Newsome et al. 1997 for a review). As a generalist predator Foxes are able to survive on comparatively small home ranges which enables them to maintain high population densities (Jarman 1986) with abundance limited generally by the density and dispersion of resources.

5.1.2 The Fox as a conservation threat The threat to wildlife posed by the Fox was already recognized in the first quarter of the 20th century during which time attempts were made to conserve many declining mammal species by collecting and transferring them to Fox free off-shore islands (Wood Jones 1923-1925; Finlayson 1927; Copley 1992; Short et al. 1992). This early

- 93 -

The effect of combined Fox and Rabbit reduction on small vertebrates and Cats

recognition was then followed by a phase where many considered that the threat posed by the Fox had been overrated and it was rejected as a key threatening factor (e.g. Calaby 1960 (Numbat); Frith 1962 (Malleefowl)). Rather, the impact of grazing by domestic stock and Rabbits, as well as land clearing, were favoured as explanations for declining biodiversity, and reductions in the abundance and range of many animals. However, in the last quarter of the 20th century, the evidence of the damage Foxes do mounted and resulted in the listing of predation of wildlife by the introduced Red Fox as a key threatening process at the Commonwealth level through the Commonwealth Endangered Species Act since 1992 and in NSW under the Threatened Species Conservation Act 1995. In 1997 for example, the Fox threatened 24 mammal, 19 bird and 10 reptile species (Newsome et al. 1997).

Today a large body of evidence exists that many animal species, in particular mammals, have declined greatly in distribution and abundance through direct impacts of Foxes. Information from a great variety of areas contributes to this evidence: historical records relate the spread of the Fox through Australia to the (local) extinction and range reduction of native species. Thus current distribution ranges of a number of species are restricted to areas outside the distribution of the Fox; i.e. far north and off-shore islands. This correlative evidence is supported by diet studies that provide information on the prey preferences of Foxes, re-introduction programmes that show the dependence of success with control/eradication of the Fox in the release area, and experiments exerting Fox control and then monitoring the responses of certain species of interest that show positive results (population increases) in many cases. The evidence about the impact of Foxes on the Australian fauna has been collated and reviewed several times (see Saunders et al. 1994; Reddiex and Forsyth 2004; Reddiex et al. 2004) and those publications provide more details.

Surplus killing, a behaviour shown by many predators including the Fox, is likely to have been an important, if not the major contributor to the often rapid decline and/or (local) extinction of many medium-sized native species (Short et al. 2002). The impact has probably been exacerbated by the naïve and ineffective behaviour of prey species to these new predators. The species that have suffered most, fall into the so-called ‘critical weight range’ of 35 - 5500 g (Burbidge and McKenzie 1989), which is a reflection of the Fox’s preferred prey size. Other species, like those with an arboreal lifestyle, or with

- 94 -

The effect of combined Fox and Rabbit reduction on small vertebrates and Cats

a preference for dense mesic habitat or rock piles, species whose range extended beyond that of the Fox (the tropical north or off-shore islands), or those whose body size was outside the critical weight range and/or was associated with a high rate of increase (Short et al. 2002) were seemingly less affected.

The impact of Foxes on vertebrates below the critical weight range In many areas in the Australian arid zone the remnant mammal community consists of species far above the critical weight range (i.e. kangaroos) but the majority fall below the critical weight range (< 35 g). Whereas the impact of Foxes is relatively well known for mammals within the critical weight range (see previous section) the threats, including Fox predation, to the arid-zone small mammals and reptiles are typically poorly understood. Apart from diet studies, which have identified small vertebrates as an important part of the Fox’s diet (Marlow 1992; Read and Bowen 2001; Paltridge 2002), there is little solid evidence about the impact of Foxes on small vertebrates. Historical information that relates the decline of species with the arrival and spread of Foxes is abundant only for medium-sized but not small species. For example, attempts to determine the response of reptiles to the introduction of non-native predators havebeen hampered by the poor historical reptile collection effort and the scarcity of subfossil material.

Anecdotal or correlative evidence frequently appears in the literature. Dickman et al. (1999), for example, reported a precipitous decline in populations of the Sandy Inland Mouse (Pseudomys hermannsburgensis) and of the Spinifex Hopping Mouse (Notomys alexis) in the Simpson Desert. The impact of Foxes which had only recently colonized the area in response to a population outbreak of the Long-haired Rat (Rattus villosissimus) was suggested as a likely explanation. Cats, which had been present at the site for probably centuries, also increased as a result of the Long-haired Rat outbreak and so may have been an additional cause for the decline in the two small mammal species. Experimental studies where predators are controlled very rarely examine the effects of these operations on small vertebrates. Therefore experimental evidence from predator exclusion and/or reintroduction programmes is scarce (but see Banks 1999; Risbey et al. 2000).

- 95 -

The effect of combined Fox and Rabbit reduction on small vertebrates and Cats

Despite little steadfast evidence, the Fox is likely to pose a considerable threat to many small mammal species, especially in combination with other threatening factors just like for mammals in the critical weight range. The potential for decline, or further decline, of small vertebrate species through Fox impacts is recognized and listed as one of the key threats for most threatened small mammal and some lizard species, and actions to reduce the density of introduced predators are recommended as part of the recovery plan.

5.1.3 Fox control Pest control programmes targeting Foxes are today part of many conservation management programmes and biodiversity projects. Various methods to control Foxes are in practice and these include shooting on sight, trapping followed by shooting, hunting with hounds, den fumigation and poison baiting. Of these, only poison baiting meets the requirement of effective and efficient use of labour and containment of costs when applied over broad areas. The distribution of poison baits which contain 1080 (Sodium Monofluoroacetate) is therefore widely used in Australia.

Complete eradication of invasive predators, like Foxes over large areas, is extremely difficult and generally unrealistic. Despite the long-term and sustained efforts of government agencies and private landholders through baiting, trapping and shooting operations and the erection the Dingo Barrier Fence, even the larger predator, the Dingo (Canis lupus), has not been eradicated in the Sheep rangelands (Wallach et al. 2009). Similarly, methods for the control of the Fox, are likely doomed to fail when the goal is eradication, as baiting programmes rarely eliminate all Foxes from an area (Thomson et al. 2000). Reduction and continued suppression of Fox numbers is more feasible and may still achieve conservation goals (Sinclair et al. 1998; Baxter et al. 2008 but see Lee (2000)).

An estimated population reduction of > 65 % is required to stop maximum population growth in Foxes, a level probably achieved in most 1080 control operations (Hone 1999). However, results have been generated from a comprehensive modelling approach (individual-based, spatially explicit model that simulated the dynamics of Fox social groups, their home-range formation and their individual dispersal behaviour in artificial landscapes) to assess the effectiveness of various Fox culling techniques in the UK.

- 96 -

The effect of combined Fox and Rabbit reduction on small vertebrates and Cats

Rushton et al. (2006) found that none of the investigated methods (baiting was not one of them) were very effective at suppressing populations of Foxes, largely because of immigration from surrounding populations. The authors conclude that suppression of Fox populations in real landscapes would require a level of culling that is not feasible. Even so they predict that some levels of population suppression are possible with adequate control of inward migration. Similarly, Harding et al. (2001) found that for successful long-term Fox population suppression, efforts need to be redirected to immigrant Foxes and juveniles. Swift recolonization through dispersing individuals has not only been identified in modelling approaches but has been confirmed by observations in the field (e.g. Priddel and Wheeler 1997; Gentle et al. 2007b). Thus sustained management has to involve an initial widespread and intensive control to reduce Fox populations to very low levels and needs to be followed by maintenance control to further reduce or at least prevent the population recovery and recolonization. Reduction of the immigration of Foxes can be achieved through maximising the target area, increasing baiting frequency and duration in cases where small areas are to be treated (Armstrong 2004), as well as establishing buffer zones surrounding the protected area (Thomson et al. 2000).

5.1.4 Interactions between Foxes, Rabbits and Cats In many areas in Australia, Foxes coexist with other introduced species, including the European Rabbit (Oryctolagus cunniculus) and the Cat (Felis catus). The impacts of all three species are listed as key threatening processes under Schedule 3 of the Commonwealth Endangered Species Protection Act 1992. Foxes, Cats and Rabbtis are interconnected through a web of complex predator-prey interactions and this presents a complicating factor in the execution, evaluation and success of control programmes. Dingos are another key species in this network of interactions, but they are not considered in the context of this study as they are virtually absent in the study area, which is located on the immediate southern side of the Dingo Barrier Fence.

The Rabbit – a brief profile The Rabbit, a native species to Europe, was introduced to south-eastern Australia in 1859 (Williams et al. 1995). It spread to occupy most of the continent by 1920 (Myers et al. 1994), but occurs mostly to the south of the Tropic of Capricorn.

- 97 -

The effect of combined Fox and Rabbit reduction on small vertebrates and Cats

The Rabbit is well known for its high reproductive potential which in the absence of the population constraints present in Europe, led to the development of Rabbit plagues in Australia. The release and outbreak of the myxomatosis and the RHD (Rabbit Haemorrhagic Disease) virus in the rangelands caused significant declines in Rabbit populations (e.g. Deeker 1992; Bomford et al. 1998; Saunders et al. 1998). These diseases continue to have such an effect, though perhaps attenuated.

Rabbits have had a profound impact on the vegetation of the rangelands, where in many areas perennial pasture species have been replaced by annuals and the recruitment of palatable shrubs and trees has been suppressed through grazing by Rabbits (Lange and Graham 1983; Foran et al. 1985; Williams et al. 1995). Several authors suggest that Rabbits have played a key role in the demise of arid-zone mammals, whether directly through competition and habitat degradation or indirectly by supporting high populations of introduced predators; namely, Cats and Foxes (Morton 1990; Williams et al. 1995; Woinarski 2001).

The Cat – a brief profile The time and location of the arrival of the Cat in Australia has been discussed for many years and it remains uncertain whether they appeared in Australia before or after European settlement. Recently Abbott (2002) conducted a comprehensive search of historical sources relating to early exploration and surveying of Australia and found no evidence that the Cat was present on the Australian mainland prior to settlement by Europeans. He concludes that Cats spread from multiple coastal introductions in the period 1824-1886, with the earliest records from the Sydney region around 1820, and that they had colonised nearly the entire continent by 1890.

The Cat in Australia occupies all habitats except the wettest rainforests on mainland Australia and Tasmania and a few off-shore islands (Wilson et al. 1992). Cats are mobile, especially during food shortage (Newsome 1991) and can disperse widely. Breeding may occur at any time of the year under favourable conditions. Cats prey on a variety of animals including mammals, birds, reptiles and invertebrates but mammals comprise the major prey in most localities. In areas where Rabbits occur, especially in semi-arid and arid areas, Rabbits (mostly juvenile) are the main prey items as well as House Mice (see review Dickman 1996b). Cats rarely consume carrion (e.g. Molsher

- 98 -

The effect of combined Fox and Rabbit reduction on small vertebrates and Cats

1999) and they are able to survive without drinking water relying solely on the moisture provided in their prey (Lundie-Jenkins 1992).

There is strong evidence that feral Cats have caused the decline and extinction of native animals on islands through predation (van Rensburg and Bester 1988; Copley 1991) and that they were at least a contributor to the failure of a number of re-introduction programmes (Johnson 1991; Short et al. 1992; Gibson et al. 1994; Dickman 1996b). Feral Cats were present on the mainland before the Fox was introduced and there is some evidence of extinctions and declines due to Cats (Dickman 1996b). Conversely, Cats have been in Tasmania and on Kangaroo Island for at least a century and yet these areas have had virtually no extinctions, or none that could be attributed directly to feral Cat predation. Sound evidence that feral Cats exert a significant effect on native wildlife throughout the mainland is scarce (Jones 1989; Wilson et al. 1992; Dickman 1996b). The nature and extent of the threat posed to native wildlife by feral Cats remains poorly understood and the evidence relating to their impacts is largely inferred (Dickman 1996b).

Rabbits and introduced predators Rabbits are a serious direct agricultural and ecological pest in Australia, but their biggest impact on small vertebrates is indirect. As staple prey of Foxes and Cats (e.g. Coman 1973; Croft and Hone 1978; Catling 1988; Newsome et al. 1997; Read and Bowen 2001; Wallach et al. 2009), especially in the semi-arid and arid areas of Australia, Rabbits are likely to be important in sustaining predator populations with the potential of subsequent negative impacts on native prey (Pech et al. 1992; Saunders et al. 1994; Dickman 1996a). The rapid spread of the Fox through Australia was almost certainly facilitated by the prior spread of the Rabbit (Rolls 1969). On the other hand, a decline in Rabbits can contribute to a reduction in Cat and Fox populations (Read and Bowen 2001); for example, after the introduction of the calicivirus (King and Wheeler 1985; Holden and Mutze 2002); (Pech and Hood 1998; Holden and Mutze 2002). Unwanted and counterproductive short-term effects of the removal of Rabbits as primary prey items could be prey-switching to previously secondary and less affected prey species (Johnson et al. 1989; Newsome 1993). Prey-switching, if it does occur, may pose a serious impact on threatened native prey and potentially cause local extinctions (Sinclair et al. 1998). This process where an introduced prey species can

- 99 -

The effect of combined Fox and Rabbit reduction on small vertebrates and Cats

induce the decline or extinction of an indigenous species through the predator’s sudden increased population size has been termed ‘hyperpredation’ (Smith and Quin 1996). Even though modelling and simulation have shown that hyperpredation can occur and that it has the potential for devastating impacts on native secondary prey species (Courchamp et al. 2000) such an effect has not yet been confirmed from field data about the Australian Fox-Cat-Rabbit system (Newsome et al. 1997; Robley et al. 2004).

Not only has the prey the ability to influence predator populations but vice versa predators have an impact on the prey population through predation pressure. Modelling suggests that Cats and Foxes can maintain Rabbits at low densities but at high densities predation pressure is inconsequential (Pech et al. 1992; Blackwell et al. 2001). This was confirmed in the field at least for semi-arid areas in NSW (Newsome 1989; Pech et al. 1992). Thus the ecological cost of reducing predator numbers, such as through Fox control, can be an increase in Rabbits ((Newsome et al. 1989; Banks et al. 1998; Risbey et al. 2000) but see (Thompson and Shepherd 1995)) with subsequent negative impacts on vegetation and wildlife. Increased Rabbit abundance may accrue further ecological costs through the facilitation of increases in Cat populations (Friend 1990; Short et al. 1995; Molsher et al. 1999) and Cats possibly pose a greater threat to native wildlife than Foxes (Risbey et al. 1999). Whereas the Fox and Rabbit relationship is fairly well studied, the association between the abundance of Rabbits and feral Cats is little understood (Robley et al. 2004).

Foxes and Cats There is considerable experimental and anecdotal evidence pointing towards competitive interactions between Foxes and Cats in Australia, with Foxes being the dominant competitor and capable of limiting Cat populations (Molsher et al. 1999; Risbey et al. 1999). Even so the relationships between feral Cats and Foxes requires much further research (Robley et al. 2004). Evidence for a high potential of competition exists on many levels including distribution patterns (Smith and Quin 1996), patterns in relative abundance (Catling and Burt 1995; Read and Bowen 2001), overlap in diet (Catling 1988; Molsher et al. 1999) especially when resources are limited, overlap in habitat use (Molsher et al. 1999), changes in diet or habitat use of Cats following removal of Foxes (Molsher et al. 1999), and occasional observations of Foxes excluding Cats from food resources and Foxes killing Cats (Coman 1973; Molsher 1999).

- 100 -

The effect of combined Fox and Rabbit reduction on small vertebrates and Cats

In a system of prey (Rabbits), meso-predator (Cats) and supra-predator (Foxes) the suppression of the supra-predator may lead to an outburst in meso-predators which may in turn threaten vulnerable prey species. Modelling of this meso-predator release process in other systems has shown that this process does exist and can drive shared prey species to extinction (Courchamp et al. 1999; Fan et al. 2005). Several studies have indeed reported the meso-predator release of Cats following the control of Fox populations (Friend 1990; Short et al. 1995; Algar and Smith 1998; Risbey et al. 1999; Burrows et al. 2003), which in some cases had a negative impact on mammalian prey species (de Tores et al. 1998; Risbey et al. 2000). Cat impacts were identified to fall most heavily on species weighing <220 g (Dickman 1996b) and Smith and Quin (1996) identified the abundance of Cats to be the explanatory factor in the loss of conilurine rodents less than 35 g. An increase in Cats may therefore evoke another wave of extinctions in the small vertebrates which fall below the preferred prey range of the Fox and which have persisted thus far.

5.1.5 Management Although gaps in the knowledge of impacts by Foxes, Cats and Rabbits and their interactions are still extensive, what is known or suspected raises considerable concern about single-species control operations. Single manipulations, such as Fox or Rabbit control alone, are not necessarily sufficient for the conservation of faunal communities. Worse, such actions can even produce counterproductive results due to the indirect effects that the control of a single species can have on other non-native species by releasing these species from pressure through predation or competition (see previous sections). Recommendations are therefore for integrated control programmes targeting Foxes, Cats and Rabbits simultaneously (e.g. Newsome 1990; Smith and Quin 1996; Molsher et al. 1999). Glen and Dickman (2005) go even further and demand that managers consider whether it is appropriate to implement control programmes for individual pest species at all where integrated pest control is not practicable.

Carrying out an integrated approach remains a challenge to wildlife managers, who are frequently limited by funds, personnel and time. Whereas relatively reliable and affordable methods exist for the control of Rabbits and Foxes, existing methods are not suitable for broad scale control of Cats over most of Australia. Removal of Cats from small areas, however, is possible using a combination of exclusion fencing and

- 101 -

The effect of combined Fox and Rabbit reduction on small vertebrates and Cats

shooting, trapping and poison baiting but this is expensive. Examples of integrated removal projects or experiments are therefore rare. A few examples exist, such as ‘Arid Recovery’ (http://www.aridrecovery.org.au; Moseby et al. (2009) and ‘Project Eden’ (http://www.dec.wa.gov.au/programs/project-eden/index.html) where all three species have been suppressed inside an exclusion fence.

5.2 Aims and Hypotheses A substantial number of studies have provided valuable insights into the importance of the Fox as a conservation threat, the interaction between Foxes, Cats and Rabbits, and the effectiveness of Fox control strategies. Some authors show a high level of confidence in the knowledge that has already been gathered (Glen and Dickman 2005). Others stress how little reliable information is actually available due to the uncertainty left by shortcomings in the experimental design and methodology applied in many studies (Saunders et al. 1994; Edwards et al. 2004; Reddiex and Forsyth 2004; Reddiex et al. 2004; Robley et al. 2004).

Sturt NP, in semi-arid far north-western NSW, includes very remote parts that contained virtually uncontrolled Rabbit, Fox and Cat populations prior to the beginning of this study. With Fox and Rabbit control programmes firmly established in the other more accessible parts of the Park, the control programmes were planned to be extended to the remaining part of the Park. This provided a unique opportunity to investigate the effects of combined Rabbit and Fox control in the absence of wild dogs, on the small vertebrate community and the population of Cats.

The following hypotheses were addressed:

1) The control actions were effective in reducing the target species; i.e. Rabbits and Foxes. 2) Rabbit reduction (if effective) was less sustainable at sites where Foxes were reduced. If Rabbit reduction was ineffective, then Fox reduction led to an increase in Rabbit abundance. 3) Cat abundance increased as a consequence of Fox reduction. The effect was stronger if Rabbits were sustained.

- 102 -

The effect of combined Fox and Rabbit reduction on small vertebrates and Cats

4) The Fox and Rabbit management affected the composition of the small mammal and lizard communities, led to an increase in the total abundances and species richness of the small vertebrate groups and the abundance of individual species.

5.3 Methods

5.3.1 Study area The study was conducted in the sand dune habitat on the western side of Sturt NP, in far western NSW. A general description of the area has been given in the general methods section (Chapter 2: section 2.1). The sites used for this part of the thesis are identical to the ones investigated in regards to the residual effect of AWP (see section 2.3 and Figure 2.3).

History of the control of wild Dingos, Foxes, Cats and Rabbits in the study area Limited resources require the Park management to focus pest control efforts on those sections of the Park that are most accessible and therefore most frequently used by the public and where neighbour relations are of concern (emigration of Dingos, Foxes and Cats from the National Park onto the neighbouring properties). The area west of ‘Fort Grey’ is remote and public access is not permitted. To the north and to the west the study area is enclosed by the Dingo Barrier Fence that prevents immigration and emigration of Dingos, Foxes and Rabbits to and from neighbouring properties. Cats can climb over the fence and it presents only a partial barrier to this species. Consequently, the ‘Fort Grey’ area has been of low priority regarding the management of Dingos, Foxes, Cats and Rabbits.

Active Rabbit control had not been attempted since the inception of the National Park in the 1970s when firearms and animal welfare regulations were much less strict and Park employees could shoot non-native animals on sight wherever encountered. Outbreaks of myxomatosis have certainly had an impact on the population but significant Rabbit losses due to myxomatosis could not be recalled by long-term Park staff and local residents for the years preceding the study. Similarly the last outbreak of RHD in the western part of NSW occurred several years prior to the beginning of this study (pers. comm. Dan Hough, Pest Control Officer, DECC Tibooburra Area). No written records exist and thus reliable information is not available. Further viral outbreaks in the Rabbit

- 103 -

The effect of combined Fox and Rabbit reduction on small vertebrates and Cats

population may have occurred, but remained undetected as a result of the remoteness of the location. Apart from shooting on sight in the early days of the Park, Fox control had not been executed in the area for decades. However, baiting programmes targeting wild Dogs (using 1080 baits which are also lethal for Foxes) were carried out in the area but were locally very restricted and centred around dams (earthen tanks). Dog baiting was last conducted in the study area in 2000 (D. Hough, pers. comm.). The Dingo Barrier Fence which constitutes the northern and western boundary of the Park, together with strategic Dog baiting ensures an extremely low abundance, if not the complete absence of wild Dogs (Unpublished reports by Croft (2002; 2009)).

No wide-scale control of Cats is carried out on the Park. Cat control is opportunistic and focuses on and around areas of perceived high Cat abundance such as homesteads used for on-park accommodation, camp grounds, the major creek systems and areas near the township of Tibooburra. In the study area no Cat control has been conducted since the early years of the Park when Cats were shot on sight.

The lack of management intervention in the form of Rabbit, Fox or Cat control for decades (or at least five years prior to the study beginning when considering the Dingo baiting), and the absence of the Dingo as a top predator, suggests high and ‘naturally regulated’ relative abundances of Foxes, Cats and Rabbits. This fact made the area an excellent location for a manipulative study investigating the interactions of these exotic species and their impact on small vertebrate communities.

5.3.2 Experimental Design An unbalanced MBACI design (multiple before/after-control/impact design) (basic design by Green 1979; modifications by Underwood 1991) was used to investigate the effectiveness of 1080 baiting on Fox abundance and the impact of Fox reduction on Rabbits, Cats and small vertebrates. In order to maximise the distance between the treatment groups while staying within the boundary of the Park and a comparable land system, two Fox removal areas (‘Impact’), each spanning over two monitoring sites, were established at opposite ends of the study area. A reference area (‘Control’), also encompassing two monitoring sites, remained untreated and thus retained non-impacted Fox numbers. Two unbaited buffer sites separated the Control sites from the Impact

- 104 -

The effect of combined Fox and Rabbit reduction on small vertebrates and Cats

sites (Figure 5.1). At each site, pitfall and Elliott-trapping and monitoring of Fox, Cat and Rabbit activity using sand plots, was conducted on three parallel dunes in an identical monitoring grid (see Chapter 4, section 4.5.2). Data were collected in quarterly surveys over two and a half years, with half the surveys conducted prior to commencement of Fox reduction and the other half after Fox reduction commenced.

The reduction of Foxes was accompanied by the management of Rabbits. Rabbit destruction was carried out on all sites and thus was not replicated. Its aim was to supplement the Fox reduction by discouraging recolonization at ‘Impact’ sites and exacerbating the predation pressure of Foxes on small vertebrates at ‘Control’ sites.

Differences in the grazing history are often an undetected and confounding factor for experiments in the rangelands (Pringle and Landsberg 2004). As has been presented in Chapter 4, historically water-proximate and heavily disturbed sites in the study area were indistinguishable from water-distant and relatively undisturbed sites in the variables measured in this study and with the methodology used. Thus, grazing history can be omitted as a reason for potential observed site differences which may obscure the findings of the manipulative experiment.

5.3.3 Rabbit reduction Rabbit reduction was initially attempted via the introduction of the RHD virus into the population. RHD was released on the experimental sites and was expected to spread from there to the entire study area, favoured by concurrent high Rabbit densities. Traditionally, individual Rabbits needed to be trapped, infected with the virus via a syringe and released into the warren. A new, less labour intensive method was used here to distribute the virus (CSIRO 2006). After free feeding twice with untreated carrot shreds the carrot shreds were coated with the viral serum and laid out in trenches near Rabbit warrens on three consecutive days. Trenches were baited late in the day to prevent the quick drying out of carrots and to maximise the duration of their attraction for Rabbits. Infection occurred on contact with or ingestion of baits. Baits were laid out mid - October 2006 and thus preceded the commencement of Fox baiting (see also Table 5.1). All Rabbit warrens in a dune section of about 500 m, centred on the monitoring grids were treated with the virus. Leigh et al. (1989) estimated rabbits in central-western NSW use the habitat in a radius of 50 m around their warrens most

- 105 -

The effect of combined Fox and Rabbit reduction on small vertebrates and Cats

intensively and their activity decreased from there. Leigh et al. as well as Witte (2002) who tracked the movements of rabbits in western NSW, found 300 m to be the maximum distance rabbits foraged from their burrows. Treating all warrens within a 500 m dune section encompassing the trapping grids would thus have targeted most rabbits that utilized the area of the trapping grid. The treatment with RHD did not have the devastating effect that was expected. The rate of spread and the factors influencing it are not fully known (CSIRO 2006), but some sign of an effect of RHD was expected within 2-3 months following the treatment. However, a reinspection of the warrens after three months after the treatment showed that the number of active warren entrances had actually increased and there was no evidence of dead/sick rabbits. The activity of rabbits on the sites (measured through sand plots see Chapter 4, section 4.5.2) had slightly decreased three months after the treatment but with large variation between sites, and so this did not provide certainty that the RHD treatment had worked. The most likely reason for such a limited effect was a resistance present in the population. These results indicated that follow-up treatment was required. A repeat treatment with RHD would have been even less effective than the initial infection of the population as the level of resistance in the Rabbit population was suspected to be high. Potential alternative methods were poison baiting using 1080 or strychnine and warren destruction. Using poison baits would have posed too great a risk to the native rodents and was not considered an option. Thus the only feasible alternative method was repeated warren destruction. Whereas RHD was expected to reduce Rabbit numbers throughout the entire study area (it was released on the experimental sites only but was expected to spread throughout the rabbit population), this was not an achievable goal using warren destruction. Warren destruction is very labour intensive and can only be used to treat small areas. The best effort that could be achieved within the project restrictions regarding time, funding and personnel was the treatment of warrens within a 500 m dune section encompassing the trapping grids. The aim of the Rabbit treatment thus changed from reducing Rabbits over the entire study area (RHD treatment) to creating refuge areas from predation in the area around the trapping grids. With Rabbits, the primary prey of Foxes and Cats, absent or at low density compared to the surrounds the treatment was expected to reduce the attractiveness of the treated dune section as hunting grounds for Foxes and Cats. All Rabbit warrens in the 500-m dune section were targeted. A mixture of LPG gas and oxygen was expelled into the warrens and lit to cause an explosion. The Rabbits died of the shock waves and in many cases the warrens

- 106 -

The effect of combined Fox and Rabbit reduction on small vertebrates and Cats

collapsed. To reduce the reuse of warrens by surface-dwelling or immigrating Rabbits, any remaining open entrances were closed off and tunnels collapsed using shovels. As a result of the harsh conditions especially the extreme summer temperatures in the Australian arid zone, the proportion of rabbits that are surface-dwellers is likely to be extremely small. Hence no attempts were made to destroy surface-dwelling Rabbits. All actions to reduce Rabbits were conducted in accordance to the relevant codes of practice (COP) and standard operating procedures (SOP) for the control of Rabbits (RAB-COP, RAB-001, RAB-005 (Sharp and Glen 2005)).

Figure 5.1: Map of study area showing location of monitoring sites and arrangement of treatment areas.

QLD # # # # # # # # # # # # # # # # # # # # # # # # # # # # # # # ## # # # # # # # # # # ####### # # # SA # # # # # # # # NSW # # # # # # # # # # # # # # # # # # # Impact # # # # # # # # # # # # # # # # # # # # # # # # # # # # # # # ### # # # # # # #### # # # # # # # ######## # # # # # # # # # # # # # # # # # # # # # # # # # # # # # # # T # # # # r # # # av # # e # lli # ng ## sto ## Buffer ck r out e

##### ##### ##### Control

####### ##### ### ## ## # # # # # # # 024 6810 # # # # # # # # # # # # # # Km # # # # # # # # # # # ## # # # ####### # # # # # ######### # # # # # # # # # # # Buffer ####### # # # # ####### # # # # # # # # Monitoring Sites ####### # # # # # # # # # # # # # # # # # # # # # # # # # 1080 Bait Stations # # # # # # # # # # # # # # # # # # # # # # ##### # # # # # # ############# # # # # # # Service Tracks # # # # # # ## # # # # # # # # # # # # # Park Boundary # # # # Impact

- 107 -

The effect of combined Fox and Rabbit reduction on small vertebrates and Cats

5.3.4 Fox reduction Baits containing 1080 were used to reduce Fox numbers. The baiting operation and was carried out in cooperation with the Pest Control Officer responsible for Sturt NP. The baited areas of the two neighbouring sites within each of the two Fox removal areas were adjoining and thus resulted in a linked area of eight km 2 for each of the two removal areas. Initially an intensive baiting scheme was applied to rid the area of the resident Foxes and reduce the number of immediate immigrants from neighbouring areas. The bait stations were arranged in a grid with a cell width of 500 m centred on the monitoring grids (see Figure 5.1). An approximate spacing of 500 m is recommended in the best practice guidelines for Fox control in NSW (NSW Threat abatement plan for the predation by the Red Fox, 2001) which allows at least two bait stations within a typical Fox home range. It also allows Foxes with smaller home ranges and subordinate animals the opportunity to interact with bait stations. Closer spacing may encourage caching of baits by Foxes. The commercially manufactured ‘Foxoff’® (Animal Control Technologies, Somerton, Victoria) was the bait type that was used predominantly. Another type of pre-fabricated bait, ‘De-Fox’ (PAKS National Pty, Ltd; Mona Vale NSW) in the form of a dried meat sausage, was used in one bait-run. The latter was a relatively new product on the Australian market at the time and the Pest Control Officer sought a trial of its field applicability. The bait uptake during this stage of the control program was consistently low (average daily uptake of 0.2 baits per 350 baits offered) and suggested a low abundance of Foxes from the outset of the program. Frequently uptakes from abutting bait stations were observed an indication that the stations were unnecessarily close together. The number of stations was therefore reduced by half about five months into the baiting scheme and baiting then focussed on the perimeter and stations near the centre and around the monitoring sites. This strategy was used to prevent immigration from Foxes from other areas and their re-establishment in the core area. Coinciding with the reduction in the number of bait stations the Fox control programme needed to be incorporated into the wild Dog baiting programme that had to be conducted throughout the park. Fresh meat baits injected with the poison as are routinely applied for wild Dog control, were used from then on. Table 5.1 provides details regarding the timing of bait-runs and the usage of bait types. The inaccessibility of the study site after rains and other work commitments of the Pest Control Officer assisting with this part of the project prevented the conduct of regular bait-runs.

- 108 -

The effect of combined Fox and Rabbit reduction on small vertebrates and Cats

The baits were placed in holes about 10 cm deep and covered with soil to limit the chances of up-take by non-target species. In the study area, there are few non-target species with Goannas, Ravens and Crows the most likely. During observations, kangaroos were the non-target species that most frequently investigated the bait stations and occasionally dug up baits, but never chewed or consumed them. Due to the low risk to non-target animals the baits were left in the ground in between bait-runs and were only replaced when removed or when they started to decompose in which case the lethal dose of 1080 was no longer guaranteed. The decline of 1080 concentration in baits is known to occur as a consequence of seepage of 1080 solution, defluorination by microorganisms, decomposition by insects and leaching by rainfall and soil moisture (Korn and Livanos 1986; Kramer et al. 1987; McIlroy et al. 1988; Fleming and Parker 1991; Wong et al. 1991). Soil moisture influences the other factors (e.g. in dry soils defluorinating microorganisms are less common (Wong et al. 1992)) and is thus the most important factor involved in 1080 break-down. In the usually dry conditions of the arid zone, Fox-Off baits are likely to contain doses lethal to Foxes even after two months in the ground (Saunders 2000). Based on this information during dry conditions all baits were replaced at least every two months or when signs of decay (discolouration, insect activity) were obvious. After rainfall baits were replaced at latest after two weeks or the earliest possible time thereafter if the area was inaccessible due to flooding. All baits were collected and disposed of through deep burial on completion of the study.

All actions to reduce Foxes were conducted in accordance to the relevant codes of practice (COP) and standard operating procedures (SOP) for the control of Foxes (FOX- COP, FOX-001 (Sharp and Glen 2005)).

- 109 -

The effect of combined Fox and Rabbit reduction on small vertebrates and Cats

Table 5.1: Details of steps undertaken for Rabbit and Fox control.

Days since previous Animal targeted Action Date Bait used Fox bait run Rabbit RHD release October 2006 N/A N/A Fox 1st bait 19/04/07 N/A Fox-Off Fox Check&rebait 2/05/07 13 Fox-Off Fox Check 8/05/07 6 Fox-Off Fox Check&rebait 13/06/07 35 Fox-Off Fox Check 21/06/07 7 Fox-Off Fox Check&rebait 6/07/07 17 Fox-Off Rabbit Warren destruction July 2007 N/A N/A Fox Check&rebait 16/07/07 10 No-fox Fox Check 9/08/07 37 No-fox Fox Bait 14/09/07 36 fresh meat baits Fox Rebait 1/11/07 47 fresh meat baits Fox Rebait 18/01/07 48 fresh meat baits Rabbit Warren destruction January 2008 N/A N/A Fox Check&rebait 10/03/08 53 fresh meat baits

5.3.5 Monitoring At each site all sampling was conducted on three parallel dunes in identical set-ups. Data for small vertebrates were collected inquarterly surveys spanning from January 2006 to April 2008. Due to initial trial of methods data collection for Foxes, Cats and Rabbits began later and were surveyed in quarterly surveys between April 2006 and April 2008.

Activity of Foxes, Cats and Rabbits The aim was to measure the site utilization or activity of Foxes, Cats and Rabbits in the area of the trapping grids. Measuring the activity of Foxes at the sites is more meaningful than measuring abundance as a decrease in Fox abundance in the area does not necessarily equate to lower predation pressure at the experimental sites. Lower activity of Foxes, Cats and Rabbits at the sites however means the species spend less time at the sites which means the pressure exerted through predation (Foxes and Cats) and competition (Rabbits) at that particular site is lower. Based on this assumption the presence/absence data of footprints on sand plots provided an index of the activity of Foxes, Rabbits and Cats at the study sites and thereby an assessment of the effectiveness of the control treatments. A set of twelve track- or sand plots with visual attractants were constructed at each pitfall trap grid and then set quarterly and investigated for three consecutive days (see section 4.5.2 and Figure 4.9 for more detail).

- 110 -

The effect of combined Fox and Rabbit reduction on small vertebrates and Cats

Initially spotlighting surveys were conducted to provide an additional measure of Fox, Rabbit and Cat numbers and an estimate of relative abundance. However, despite good visibility in an open semi-arid environment, the extremely low frequencies of sightings, even for Rabbits, led to the discontinuation of the surveys.

Small vertebrates The effect of the treatments on the assemblages of small terrestrial mammals and lizards was measured by quarterly surveys using a combination of pitfall and Elliott trapping (see section 4.5.2 a Data manipulation and analysis).

5.3.6 Data manipulation and analysis The number of trap nights when monitoring small mammals and lizards was identical between sampling sites and surveys. Thus the number of captures (small mammals, reptiles and invertebrates) was used as an index of relative abundance. Recaptures of small mammals within the same survey were excluded. Lizards were not marked and individuals not identified, therefore the dataset may include recaptures. Similarly, the monitoring effort to asses Foxes, Cats and Rabbits, including track plot number and exposure time was identical between sites and surveys. The presence/absence data of footprints on the track plots was used to monitor the activity of Foxes, Cats and Rabbits. The assumption being that the species activity is correlated to the abundance of these animals, with larger numbers of footprints indicating larger Fox abundances.

Univariate statistical analyses Univariate analyses (SPSS for Windows 17.0) were performed on the community variables (diversity, richness and evenness) and on those species that were present on at least half the sites and had a total number of captures of 20 or more. If the applied Fox reduction had any effect then one would expect a breakpoint in the trend of the ‘Impact’ data at or after the beginning of the treatment application. A Friedman analysis (non- parametric test, repeated measure based on ranks) was used to test for significant changes (breakpoints) in a variable over time. Separate Friedman analyses were performed for the two treatments; i.e. ‘Impact’ and ‘Control’. Monte Carlo estimated p- values are presented due to small sample sizes. To illustrate results and establish if further analysis was needed, plots of mean ranks per survey were used. If 1) the Friedman analysis was significant for the ‘Impact’ dataset only and the plot suggested the breakpoint to lie at or after treatment began or if 2) the Friedman analysis was

- 111 -

The effect of combined Fox and Rabbit reduction on small vertebrates and Cats

significant for both datasets and the plot showed the occurrence of breakpoints to be at different points in time then a multiple range test comparison (Dunn´s test after Conover (1980)) on pairs of subsequent surveys was performed to determine when the breakpoint occurred. Additionally, plots of the mean number of captures (small vertebrates) or plot visitation (Foxes, Cats and Rabbits) and 95 % confidence intervals were used to aid in the interpretation of results.

Multivariate statistical analyses Differences in the structure of the lizard and small mammal assemblages between treatments and periods were evaluated by non-metric multi-dimensional scaling (nMDS) ordination on the square-root transformed data (abundance data), using the Bray-Curtis similarity measure. Unbalanced permutational multivariate analysis of variance (PRIMER ver. 6.1.12, PERMANOVA+ ver. 1.0.2, (Anderson 2001; Anderson 2005; Anderson et al. 2008)), based on a MBACI design was performed, which uses permutations to determine the distributions of test-statistics (Note: the degrees of freedom for F-values may be non-integers with this method). The MBACI model had four factors: treatment (Fox baiting (‘Impact’) and no baiting (‘Control’), fixed factor), site within the treatment (random factor), periods (‘Before’ / ‘After’, fixed factor) and survey within periods (random factor). For lizards ‘dunes’ were used as the repeated measure, for small mammals data were pooled over ‘dunes’ and ‘surveys’ used as the repeated measure, reducing the above model to the first three factors. The number of permutations used was 9999. The principal source of interest for impact assessment was the interaction between the treatment (‘Impact’ / ‘Control’) and the periods (‘Before’ / ‘After’).

5.4 Results

5.4.1 Effectiveness of Fox reduction The results of the Friedman analysis were highly significant for the ‘Impact’ sites (2 = 27.719, p << 0.001) but not for the ‘Control’ sites (2 = 7.401, p = 0.413). Thus there was a breakpoint at which Fox activity changed significantly at some point at the ‘Impact’ sites but not at the ‘Control’ site, a result that was expected if Fox reduction had been successful. However the graphic illustration suggested and the Dunn´s test confirmed a breakpoint between January 2007 and April 2007 (Q = 8.832, p < 0.05) (Figure 5.2 b) and thus a significant drop in Fox activity before the commencement of

- 112 -

The effect of combined Fox and Rabbit reduction on small vertebrates and Cats

baiting in May 2007 (indicated by the vertical black line in the graphs). Two factors may be responsible for this unexpected drop in Fox activity prior to the beginning of the baiting operation that occurred in both treatment groups (even though more pronounced at ‘Impact’ sites. Firstly, RHD was released into the Rabbit population in October 2006 and Rabbit activity subsequently dropped (see Figure 5.3) which in turn may have influenced the activity of Foxes. Secondly, evidence of weakened and emaciated foxes was found in a trial capture program which likely resulted in a high mortality event in the population.

As the Fox population was already at a low when poison baiting began, a decline through baiting could not be shown. However, the baiting appears to have been successful in continuously suppressing Fox populations at ‘Impact’ sites compared to ‘Control’ sites. It is quite noticeable that for the remainder of the study the activity of Foxes at the ‘Impact’ sites remained well below the Fox activity at unbaited ‘Control’ sites. The situation was reversed for most of the pre-baiting sessions. Another observation hints that Fox activity was successfully suppressed through baiting: Fox populations typically undergo increases in abundance from late spring through to autumn, coinciding with the period of independence and dispersal of juveniles (Marlow 1992; Saunders et al. 1994), and declines during winter and spring, resulting from the senescence of adults and juveniles (Marlow 1992) and the stabilisation of territories. The trajectory of the trend lines for the two treatments is parallel for most of the pre- baiting period but shows a clear disparity post-baiting. The activity data for ‘Control’ sites show distinctive peaks in October in both years, pre- and post-baiting, corresponding with the time of year when fox cubs start to leave the den. At the ‘Impact’ sites the October peak only occurred pre-baiting but instead of showing a peak in October a year later the activity of Foxes post-baiting dropped to zero. A result that is most likely due to the baiting program.

Considering all of the above it is concluded that Fox populations were significantly reduced during the study period (albeit through other causes than the applied treatment) and populations at ‘Impact’ sites were successfully suppressed (as compared to untreated ‘Control’ sites) for the remainder of the study period as a result of 1080 baiting. The subsequent analyses of the effects of a reduction in Fox activity on Rabbits, Cats and small vertebrates, are therefore warranted.

- 113 -

The effect of combined Fox and Rabbit reduction on small vertebrates and Cats

a) b)

8 4 Before After Before After 7 3.5 6 3 5 2.5 4 2

Meanrank 3 1.5 Mean activity 2 1 1 0.5 0 0 Jul Oct Jan Apr Jul Oct Jan Apr Jul- Oct- Jan- Apr- Jul- Oct- Jan- Apr- 06 06 07 07 07 07 08 08 06 06 07 07 07 07 08 08 Impact Control Control Impact

Figure 5.2: Comparison of Fox activity between treatments; (a) mean abundance (b) mean rank per survey. Graphs are based on the number of visited track plots. N= 12 for ‘Impact’ sites and n = 6 for ‘Control’ sites. Error bars in a) represent 95 % confidence intervals. Arrows in b) indicate a significant breakpoint (Dunn’s test).

5.4.2 Effectiveness of Rabbit reduction Rabbit reduction was carried out in a similar fashion on all sites and Rabbit abundance surveyed as visitation to track plots. The Friedman test was significant (2 = 20.428, p = 0.003) and the subsequent Dunn´s test identified significant variation between October 2006 versus January 2007 (Q = 6.793, p < 0.05), January 2007 versus April 2007 (Q = 3.962, p < 0.05), April 2007 versus July 2007 (Q = 5.775, p < 0.05) and July 2007 versus October 2007 (Q = 4.076, p < 0.05) (Figure 5.3 b). Thus the Rabbit abundance was in decline for about six months after the release of RHD. This may indicate at least a short-term success of the treatment which was then however followed by a significant increase in Rabbits in the subsequent six months when Rabbit numbers reached a peak abundance in October 2007 despite repeated treatments. Warren destruction using gas had no detectable effect (Figure 5.3 a).

- 114 -

The effect of combined Fox and Rabbit reduction on small vertebrates and Cats

a) b)

12 8 RHD Gas Gas 10 7 6 8 5 6 4 3

4 Mean rank 2 Mean abundance 2 1 0 0 Jul- Oct- Jan- Apr- Jul- Oct- Jan- Apr- Jul- Oct- Jan- Apr- Jul- Oct- Jan- Apr- 06 06 07 07 07 07 08 08 06 06 07 07 07 07 08 08

Figure 5.3: Rabbit activity over study period; (a) mean abundance (b) mean rank per survey. Graphs are based on the number of visited track plots. Error bars in a) represent 95 % confidence intervals. Arrows in b) indicate a significant breakpoint (Dunn’s test). N = 24.

5.4.3 Effect of Fox reduction on Rabbits The apparent ineffectivenenss of the applied Rabbit control methods (see section 5.4.2) allowed the analysis of the impact of reduced Fox activity on Rabbits. Rabbit abundance fluctuated considerably over time and between sites as indicated by the wide error bars (Figure 5.4 a). Breaks in trends were identified in both treatments (‘Control’: 2 =15.437, p = 0.020, ‘Impact’: 2 = 14.412, p = 0.037). The breaks occur at different times in the two treatments, but the trajectories of the lines for ‘Impact’ and ‘Control’ in the illustration of mean ranks are similar (Figure 5.4 b). Thus changes in Rabbit abundance were more likely due to some other effect changing conditions at sites within both treatment groups rather than a treatment effect. If there was a treatment effect, it would have been an unexpected increase in Rabbits at sites where Fox activity remained unaltered (Figure 5.4 a).

- 115 -

The effect of combined Fox and Rabbit reduction on small vertebrates and Cats

a) b)

18 8 Before After 16 Before After 7 14 6 12 5 10 4 8

Meanrank 3

Mean activity 6 2 4 1 2 0 0 Jul06 Oct Jan Apr Jul07 Oct Jan Apr Jul- Oct- Jan- Apr- Jul- Oct- Jan- Apr- 06 06 07 07 07 07 08 08 06 07 07 07 08 08 Impact Control Control Impact

Figure 5.4: Comparison of Rabbit activity between treatments; (a) mean abundance (b) mean rank per survey. Rabbit activity was measured as the number of visited track plots. N= 12 for ‘Impact’ sites and n = 6 for ‘Control’ sites. Error bars in a) represent 95 % confidence intervals.

5.4.4 Effect of Fox reduction on Cats Cat activity was low and no footrpints of Cats were recorded for most of 2007 (Figure 5.5). However, a marked increase in Cat activity was observed towards the end of the study. This seems to have occurred independently of the Fox treatment as the increase is evident in both treatment groups. At the ‘Impact’ sites the increase commenced earlier than at the ‘Control’ sites but the increase was more pronounced at the ‘Control’ sites. The generally low number of Cat visitations (a total of 26) prevented an analysis to test the significance of these observations.

- 116 -

The effect of combined Fox and Rabbit reduction on small vertebrates and Cats

a) b)

4 8 Before After Before After 3.5 7

3 6

2.5 5

2 4

Meanrank 3 1.5 Mean activity 2 1 1 0.5 0 0 Jul Oct Jan Apr Jul Oct Jan Apr Jul- Oct- Jan- Apr- Jul- Oct- Jan- Apr- 06 06 07 07 07 07 08 08 06 06 07 07 07 07 08 08 Impact Control Control Impact

Figure 5.5: Comparison of Cat activity between treatments; (a) mean abundance (b) mean rank per survey. Cat activity was measured as the number of visited track plots. N= 12 for ‘Impact’ sites and n = 6 for ‘Control’ sites. Error bars in a) represent 95 % confidence intervals.

5.4.5 Effect of Fox reduction on small mammals The small mammal community during the time period relevant to this chapter consisted of six species (Table 5.2). The House Mouse was the most abundant species, representing 68.8 % of captures and the Sandy Inland Mouse was the most abundant native species with 12.1 % of captures.

Table 5.2: Small mammal numbers; total captures and percent of total

Total Scientific name Common name captures % Mus domesticus House Mouse 347 68.8 Pseudomys hermannsburgensis Sandy Inland Mouse 61 12.1 Sminthopsis crassicaudata Fat-tailed Dunnart 50 10.4 Notomys fuscus Dusky Hopping Mouse 37 7.7 Sminthopsis macroura Stripe faced Dunnart 7 1.5 Leggadina forresti Central short-tailed Mouse 2 0.4 TOTAL 504

There was no evidence of a significant effect of the Fox reduction on the small mammal community (treatment x period: F4.54, 9.48 = 1.372, p = 0.266). This is confirmed by the MDS plot which shows no separation between pre- and post-baiting data Figure 5.6.

Significant differences did occur however between sites (F4, 32 = 2.478, p = 0.0207) and between surveys (F8, 32 = 4.85, p << 0.001).

- 117 -

The effect of combined Fox and Rabbit reduction on small vertebrates and Cats

Figure 5.6: MDS plot of the small mammal assemblage and its changes with survey period and treatment. The plot is based on Bray-Curtis similarities calculated from square-root transformed species abundances.

A series of paired plots of 1) the mean abundance/ number of species and 2) the mean rank per survey, are used to illustrate results. The results of the analysis for treatment effects on total small mammal abundance, species number and abundance of individual small mammal species are presented in Table 5.3. Species diversity and evenness could not be computed for some surveys due to the low capture rates (frequent zero values that caused division by zero problems) and could therefore not be compared.

- 118 -

The effect of combined Fox and Rabbit reduction on small vertebrates and Cats

Table 5.3: Results of comparison between treatments (Friedman test) for small mammal variables. Df = 7.

Mean Chi p Variable Treatment abundance (Friedman) (Friedman) Control 4.9 13.274 0.079 Total Abundance Impact 8.6 26.531 << 0.001 Control 1.9 12.242 0.159 Richness Impact 1.7 17.833 0.016 Control 1.0 24.969 0.001 Mus domesticus Impact 2.4 63.042 << 0.001 Control 0.2 9.833 0.529 Notomys fuscus Impact 0.2 25.756 0.002 Pseudomys Control 0.2 7.235 0.704 hermannsburgensis Impact 0.4 19.444 0.014 Sminthopsis Control 0.3 22.566 0.005 crassicaudata Impact 0.3 42.720 << 0.001

Total abundance and species richness For both, total small mammal abundance and species richness significant Friedman test results indicate the existence of breakpoints in the ‘Impact’ treatment. For the small mammal abundance a subsequent Dunn’s test identified three breakpoints (Figure 5.7 b). For all three of these breakpoints the trendline for the ‘Control’ sites runs parallel to that of the ‘Impact’ sites. Thus even though the Friedman test was not significant for the ‘Control’ sites these breakpoints are unlikely to be related to a treatment effect. For the small mammal richness the Dunn’s test identified breakpoints between all surveys in the period following the significant reduction of Fox activity (Figure 5.8 b). At the time of the first four breakpoints the trendline for the ‘Control’ sites runs parallel to that of the ‘Impact’ sites. Thus even though the Friedman test was not significant for the ‘Control’ sites these breakpoints are unlikely to be related to a treatment effect. The trends in the two treatment groups diverge in the last survey period and so the breakpoint between January 2008 and April 2008 (Q = 5.605, p < 0.05) may indicate a treatment effect and species richness increased at ‘Impact’ sites compared to the ‘Control’ sites. However, generally parallel trajectories for both treatment groups together with the large confidence intervals that indicate high variability between samples (i.e. dunes) make an effect of reduced Fox activity unlikely.

- 119 -

The effect of combined Fox and Rabbit reduction on small vertebrates and Cats

a) b)

40 10 Before After Before After 9 35 8 30 7 25 6 5 20 4 15 Meanrank 3

Mean abundance 10 2 5 1 0 0 Jan Apr Jul Oct Jan Apr Jul Oct Jan Apr Jan- Apr- Jul-06 Oct- Jan- Apr- Jul-07 Oct- Jan- Apr- 06 06 06 06 07 07 07 07 08 08 06 06 06 07 07 07 08 08

Impact Control Control Impact

Figure 5.7: Comparison of Small mammal captures between treatments; (a) mean abundance (b) mean rank per survey. N= 12 for ‘Impact’ sites and n = 6 for ‘Control’ sites. Error bars in a) represent 95 % confidence intervals.

a) b)

7 10 Before After Before After 9 6 8 5 7 6 4 rank 5 3 4 Mean 3 Mean richness 2 2 1 1 0 0 Jan Apr Jul Oct Jan Apr Jul Oct Jan Apr Jan- Apr- Jul-06 Oct- Jan- Apr- Jul-07 Oct- Jan- Apr- 06 06 06 06 07 07 07 07 08 08 06 06 06 07 07 07 08 08

Impact Control Control Impact

Figure 5.8: Comparison of Small mammal richness between treatments; (a) mean abundance (b) mean rank per survey. N= 12 for ‘Impact’ sites and n = 6 for ‘Control’ sites. Error bars in a) represent 95 % confidence intervals.

Mus musculus The Friedman test was significant for both treatment groups. At the ‘Impact’ sites breakpoints occurred between January and April 2007, April 2007 and July 2007 and July and October 2007 (Figure 5.9 b). The first breakpoint coincided with the point in time where Fox activity significantly declined. Similar trajectories were observed in the ‘Control’ sites (even though breakpoints were not identified here) and hence changes in M. musculus abundance at ‘Impact’ sites from January2007 to October

- 120 -

The effect of combined Fox and Rabbit reduction on small vertebrates and Cats

2007 was likely unrelated to the treatment. From October 2007 onwards however a distinct disparity appeared in the trajectory of the trends in the two treatment types. The upwards trend continued at the ‘Impact’ sites, whereas the trajectory pointed significantly downwards at the ‘Control’ sites (Q = 3.906, p <<0.05). This suggests that a treatment effects became evident towards the end of the study with higher M. musculus abundance at ‘Impact’ sites compared to ‘Control’ sites (Figure 5.9). a) b)

12 10 Before After 9 Before After 10 8 7 8 6 6 5 4 4 Mean rank 3 Mean abundance 2 2 1 0 0 Jan- Apr- Jul-06 Oct- Jan- Apr- Jul-07 Oct- Jan- Apr- Jan- Apr- Jul- Oct- Jan- Apr- Jul- Oct- Jan- Apr- 06 06 06 07 07 07 08 08 06 06 06 06 07 07 07 07 08 08

Impact Control Control Impact

Figure 5.9: Comparison of Mus musculus captures between treatments; (a) mean abundance (b) mean rank per survey. N= 12 for ‘Impact’ sites and n = 6 for ‘Control’ sites. Error bars in a) represent 95 % confidence intervals.

Notomys fuscus Three breakpoints were identified for ‘Impact’ sites (Figure 5.10 b). The trajectory of the trend for the ‘Control’ sites mimics that of the ‘Impact’ sites but changes in trend are less marked and thus non-significant. Parallel lines for both treatment groups together with the large confidence intervals that indicate high variability between samples (i.e. dunes) make an effect of reduced Fox Activity unlikely (Figure 5.10 a).

- 121 -

The effect of combined Fox and Rabbit reduction on small vertebrates and Cats

a) b)

2 10 Before After 1.8 9 Before After 1.6 8 1.4 7 1.2 6 1 5 0.8 4

0.6 rank Mean 3 Mean abundance 0.4 2 0.2 1 0 0 Jan- Apr- Jul- Oct- Jan- Apr- Jul- Oct- Jan- Apr- Jan- Apr- Jul- Oct- Jan- Apr- Jul- Oct- Jan- Apr- 06 06 06 06 07 07 07 07 08 08 06 06 06 06 07 07 07 07 08 08

Impact Control Control Impact

Figure 5.10: Comparison of Notomys fuscus captures between treatments; (a) mean abundance (b) mean rank per survey. N= 12 for ‘Impact’ sites and n = 6 for ‘Control’ sites. Error bars in a) represent 95 % confidence intervals.

Pseudomys hermannsburgensis Breakpoints were only identified for the ‘Impact’ sites. A significant increase occurred between July and October 2007 and was followed by a significant decrease between October 2007 and January 2008. A simultaneous but less marked peak also occurred at the ‘Control’ sites. Thus if any effect of Fox reduction was evident, it was in the form of higher peak abundances (Figure 5.11). a) b)

2 10 Before After 1.8 Before After 9 1.6 8 1.4 7 1.2 6 1 5 0.8 4 0.6 Mean rank 3 Mean abundance 0.4 2 0.2 1 0 0 Jan- Apr- Jul- Oct- Jan- Apr- Jul- Oct- Jan- Apr- Jan- Apr- Jul- Oct- Jan- Apr- Jul- Oct- Jan- Apr- 06 06 06 06 07 07 07 07 08 08 06 06 06 06 07 07 07 07 08 08

Impact Control Control Impact

Figure 5.11: Comparison of Pseudomys hermannsburgensis captures between treatments; (a) mean abundance (b) mean rank per survey. N= 12 for ‘Impact’ sites and n = 6 for ‘Control’ sites. Error bars in a) represent 95 % confidence intervals.

- 122 -

The effect of combined Fox and Rabbit reduction on small vertebrates and Cats

Sminthopsis crassicaudata Breakpoints were identified for both treatment but trends followed a common trajectory for S. crassicaudata abundance at ‘Control’ and ‘Impact’ sites. Thus any breakpoints were likely unrelated to treatment effects (Figure 5.12). a) b)

3 10 Before After Before After 9 2.5 8

2 7 6 1.5 5 4 1 Mean rank 3 Mean abundance 2 0.5 1 0 0 Jan- Apr- Jul- Oct- Jan- Apr- Jul- Oct- Jan- Apr- Jan- Apr- Jul- Oct- Jan- Apr- Jul- Oct- Jan- Apr- 06 06 06 06 07 07 07 07 08 08 06 06 06 06 07 07 07 07 08 08

Impact Control Control Impact

Figure 5.12: Comparison of Sminthopsis crassicaudata captures between treatments; (a) mean abundance (b) mean rank per survey. N= 12 for ‘Impact’ sites and n = 6 for ‘Control’ sites. Error bars in a) represent 95 % confidence intervals.

In summary, these results suggest no effect of the reduction of Fox activity for Sminthopsis crassicaudata and Notomys fuscus or the overall abundance or richness of small mammals. But results indicate a potential positive effect for Pseudomys hermannsburgensis and Mus musculus. The effect in the two rodent species took the form of a higher abundance during population peaks in P. hermannsburgensis and a prolonged increase and higher abundance at the end of the study at ‘Impact’ sites compared to ‘Control’ sites in M. musculus. Captures of the remaining species i.e. S. macroura and L. forresti were too few for meaningful analysis.

5.4.6 Effect of Fox reduction on lizards The lizard community consisted of 20 species and was dominated by scincids of the genus Ctenotus (Table 5.4). By far the most abundant species, with nearly a quarter of all captures (52.9 %) was Ctenotus taeniatus.

- 123 -

The effect of combined Fox and Rabbit reduction on small vertebrates and Cats

Table 5.4: Lizard numbers, total captures and percent of total

Total Scientific name captures % Ctenotus taeniatus 586 52.9 Ctenotus schomburgkii 375 33.9 Lucasium damaeum 326 29.4 Ctenophorus pictus 308 27.8 Eremiascincus fasciolatus 138 12.5 Ctenophorus fordi 115 10.4 Lerista labialis 81 7.3 Nephrurus levis 72 6.5 Ctenophorus nuchalis 67 6.1 Rhynchoedura ornata 54 4.9 Lucasium stenodactylum 45 4.1 Heteronotia binoei 30 2.7 Ctenotus regius 28 2.5 Varanus gouldii 25 2.3 Pogona vitticeps 14 1.3 Menetia greyii 11 1.0 Lucasium byrnei 5 0.5 Gehyra variegata 3 0.3 Morethia adelaidensis 3 0.3 Ramphotyphlops bituberculata 1 0.1 TOTAL 2287

There was no evidence to suggest that the lizard community as a whole was affected by the reduction of Fox activity. No separation in respect to treatment and/or time period is apparent in the MDS plot (Figure 5.13). The lack of a treatment effect was confirmed by the results of the PERMANOVA (treatment x period: F1.75, 8.64 = 1.859 p = 0.094). However, temporal (between surveys) and spatial (between sites) differences were significant (F8, 32 = 7.823 p << 0.001 and F4, 32 = 3.372 p << 0.001).

- 124 -

The effect of combined Fox and Rabbit reduction on small vertebrates and Cats

Figure 5.13: MDS plot of the lizard assemblage and its changes with survey period and treatment. The plot is based on Bray-Curtis similarities calculated from square-root transformed species abundances.

As for the presentation of the previous results a series of paired plots of 1) the mean abundance and 2) the mean rank per survey, is used to illustrate results. Friedman tests were performed for the total lizard abundance, species richness and for the most common lizard species (a total of 50 or more captures) that occurred at more than half of all sites. Lucasium damaeum was excluded from the analysis due to initial difficulties in distinguishing it from the related and similar species L. stenodactylus. Lerista labialis has been omitted from analysis as trap mortality of the species was high, which may have affected results. Despite a relatively low number of captures (25) Varanus gouldii was included in the analysis as it has a particular importance in the arid environment as a native predator of small vertebrates. As for mammals, low capture rates in winter prevented the calculation of the diversity and evenness indices, thus the response of these community parameters to the reduction in Fox activity could not be analysed. The results of the Friedman test are presented in (Table 5.5).

- 125 -

The effect of combined Fox and Rabbit reduction on small vertebrates and Cats

Table 5.5: Results of comparison between treatments (Friedman test, df = 7) for lizard species. Species where the Friedman test was significant for ‘Impact’ sites only are most likely to have been affected by the reduction in Fox activity.

Chi p Variable Treatment Mean (Friedman) (Friedman) Control 8.7 67.403 << 0.001 Total Abundance Impact 9.5 34.474 << 0.001 Control 4.2 68.528 << 0.001 Richness Impact 4.6 35.622 << 0.001 Control 0.2 9.000 0.510 Ctenophorus nuchalis Impact 0.4 9.920 0.362 Control 1.4 17.527 0.028 Ctenophorus pictus Impact 1.1 32.681 << 0.001 Control 1.2 14.941 0.079 Ctenotus schomburgkii Impact 1.8 33.948 << 0.001 Control 3.0 29.628 << 0.001 Ctenotus taeniatus Impact 2.0 28.937 0.001 Control 0.6 17.542 0.028 Eremiascincus fasciolatus Impact 0.4 27.683 0.001 Control 0.3 17.197 0.034 Nephrurus levis Impact 0.2 5.567 0.816 Control 0.2 14.418 0.088 Rhynchoedura ornata Impact 0.3 26.839 << 0.001 Control 0.1 14.902 0.126 Varanus gouldii Impact 0.1 14.068 0.143

A lack of significant results in the Friedman test suggest that Ctenophorus nuchalis and Varanus gouldii were unaffected by the treatment (Figure 5.14 and Figure 5.15 respectively). Likewise an effect on N. levis was unlikely as the Friedman test suggested breakpoints for the ‘Control’ sites only (Figure 5.16). Significant results in both treatments were obtained for the abundance of C. pictus (Figure 5.17), C. taeniatus (Figure 5.18), Eremiascincus fasciolatus (Figure 5.19), the total lizard abundance (Figure 5.20) and species richness (Figure 5.21). As trend lines shared common trajectories post-treatment this suggests that none of the variables were affected by the reduction in Fox activity.

- 126 -

The effect of combined Fox and Rabbit reduction on small vertebrates and Cats

a) b)

1.4 10 Before After 9 Before After 1.2 8 1 7 6 0.8 5 0.6 4 Meanrank 0.4 3 Mean abundance 2 0.2 1 0 0 Jan- Apr- Jul- Oct- Jan- Apr- Jul- Oct- Jan- Apr- Jan Apr Jul Oct Jan Apr Jul Oct Jan Apr 06 06 06 06 07 07 07 07 08 08 06 06 06 06 07 07 07 07 08 08 Impact Control Control Impact

Figure 5.14: Comparison of Ctenophorus nuchalis captures between treatments; (a) mean abundance (b) mean rank per survey. N= 12 for ‘Impact’ sites and n = 6 for ‘Control’ sites. Error bars in a) represent 95 % confidence intervals.

a) b)

1 10 Before After 0.9 Before After 9 0.8 8 0.7 7 0.6 6

0.5 rank 5 0.4 4 Mean 0.3 3 Mean abundance Mean 0.2 2 0.1 1 0 0 Jan- Apr- Jul- Oct- Jan- Apr- Jul- Oct- Jan- Apr- Jan Apr Jul Oct Jan Apr Jul Oct Jan Apr 06 06 06 06 07 07 07 07 08 08 06 06 06 06 07 07 07 07 08 08 Impact Control Control Impact

Figure 5.15: Comparison of Varanus gouldii captures between treatments; (a) mean abundance (b) mean rank per survey. N= 12 for ‘Impact’ sites and n = 6 for ‘Control’ sites. Error bars in a) represent 95 % confidence intervals.

- 127 -

The effect of combined Fox and Rabbit reduction on small vertebrates and Cats

a) b)

1.6 10 Before After Before After 1.4 9 8 1.2 7 1 6 0.8 5 0.6 4 Meanrank 3

Mean abundance 0.4 2 0.2 1 0 0 Jan- Apr- Jul- Oct- Jan- Apr- Jul- Oct- Jan- Apr- Jan Apr Jul Oct Jan Apr Jul Oct Jan Apr 06 06 06 06 07 07 07 07 08 08 06 06 06 06 07 07 07 07 08 08

Impact Control Control Impact

Figure 5.16: Comparison of Nephrurus levis captures between treatments; (a) mean abundance (b) mean rank per survey. N= 12 for ‘Impact’ sites and n = 6 for ‘Control’ sites. Error bars in a) represent 95 % confidence intervals. a) b)

6 10 Before After 9 Before After 5 8 7 4 6 3 5 4 2 Meanrank 3 Mean abundance 2 1 1 0 0 Jan- Apr- Jul-06 Oct- Jan- Apr- Jul-07 Oct- Jan- Apr- Jan Apr Jul Oct Jan Apr Jul Oct Jan Apr 06 06 06 07 07 07 08 08 06 06 06 06 07 07 07 07 08 08 Impact Control Control Impact

Figure 5.17: Comparison of Ctenophorus pictus captures between treatments; (a) mean abundance (b) mean rank per survey. N= 12 for ‘Impact’ sites and n = 6 for ‘Control’ sites. Error bars in a) represent 95 % confidence intervals. a) b)

- 128 -

The effect of combined Fox and Rabbit reduction on small vertebrates and Cats

18 10 Before After 16 Before After 9 14 8 7 12 6 10 5 8 4 Meanrank 6 3 Mean abundance 4 2 2 1 0 0 Jan- Apr- Jul- Oct- Jan- Apr- Jul- Oct- Jan- Apr- Jan Apr Jul Oct Jan Apr Jul Oct Jan Apr 06 06 06 06 07 07 07 07 08 08 06 06 06 06 07 07 07 07 08 08

Impact Control Control Impact

Figure 5.18: Comparison of Ctenotus taeniatus captures between treatments; (a) mean abundance (b) mean rank per survey. N= 12 for ‘Impact’ sites and n = 6 for ‘Control’ sites. Error bars in a) represent 95 % confidence intervals.

a) b)

4 10 Before After Before After 9 3.5 8 3 7 2.5 6

2 rank 5 4

1.5 Mean 3

Mean abundance 1 2 0.5 1 0 0 Jan- Apr- Jul- Oct- Jan- Apr- Jul- Oct- Jan- Apr- Jan Apr Jul Oct Jan Apr Jul Oct Jan Apr 06 06 06 06 07 07 07 07 08 08 06 06 06 06 07 07 07 07 08 08 Impact Control Control Impact

Figure 5.19: Comparison of Eremiascincus fasciolatus captures between treatments; (a) mean abundance (b) mean rank per survey. N= 12 for ‘Impact’ sites and n = 6 for ‘Control’ sites. Error bars in a) represent 95 % confidence intervals.

- 129 -

The effect of combined Fox and Rabbit reduction on small vertebrates and Cats

a) b)

10 45 Before After 9 40 Before After 8 35 7 30 6 25 5 20 4 Meanrank 15 3

Mean abundance 2 10 1 5 0 0 Jan Apr Jul Oct Jan Apr Jul Oct Jan Apr Jan- Apr- Jul-06 Oct- Jan- Apr- Jul-07 Oct- Jan- Apr- 06 06 06 06 07 07 07 07 08 08 06 06 06 07 07 07 08 08

Impact Control Control Impact

Figure 5.20: Comparison of the total lizard captures between treatments; (a) mean abundance (b) mean rank per survey. N= 12 for ‘Impact’ sites and n = 6 for ‘Control’ sites. Error bars in a) represent 95 % confidence intervals.

a) b)

10 14 Before After Before After 9 12 8 10 7 6 ness 8 rank 5 6 4 Mean 3 Mean rich Mean 4 2 2 1 0 0 Jan Apr Jul Oct Jan Apr Jul Oct Jan Apr Jan- Apr- Jul-06 Oct- Jan- Apr- Jul-07 Oct- Jan- Apr- 06 06 06 06 07 07 07 07 08 08 06 06 06 07 07 07 08 08

Impact Control Control Impact

Figure 5.21: Comparison of lizard richness between treatments; (a) mean abundance (b) mean rank per survey. N= 12 for ‘Impact’ sites and n = 6 for ‘Control’ sites. Error bars in a) represent 95 % confidence intervals.

For some lizard species the results of the Friedman test identified significant breakpoints for the ‘Impact’ but not the ‘Control’ sites, suggesting a potential treatment effect. The potential effect of Fox reduction for these species is evaluated in more detail below.

- 130 -

The effect of combined Fox and Rabbit reduction on small vertebrates and Cats

Ctenotus schomburgkii Significant breakpoints between each of the surveys were identified in the ‘Impact’ sites post-treatment. However, the trajectory of the trendline of the ‘Impact’ sites closely followed that of ‘Control’ sites suggesting no treatment effect (Figure 5.22). a) b)

10 10 9 Before After 9 Before After 8 8 7 7 6 6 5 5 4 4 Meanrank 3 3 Mean abundance 2 2 1 1 0 0 Jan- Apr- Jul-06 Oct- Jan- Apr- Jul-07 Oct- Jan- Apr- Jan Apr Jul Oct Jan Apr Jul Oct Jan Apr 06 06 06 07 07 07 08 08 06 06 06 06 07 07 07 07 08 08

Impact Control Control Impact

Figure 5.22: Comparison of Ctenotus schomburgkii captures between treatments; (a) mean abundance (b) mean rank per survey. N= 12 for ‘Impact’ sites and n = 6 for ‘Control’ sites. Error bars in a) represent 95 % confidence intervals.

Rhynchoedura ornata On the ‘Impact’ sites a significant change in trend (increase) occurred between July and October 2007 and another between October 2007 and January 2008. Rhynchoedura ornata appears to have benefited from Fox reduction as it occurred in higher abundances at ‘Impact’ sites relative to ‘Control’ sites, after fox activity was reduced (Figure 5.23). However, low captures of the species and high variability between sites make it difficult to draw definite conclusions.

- 131 -

The effect of combined Fox and Rabbit reduction on small vertebrates and Cats

a) b)

1.8 10 9 Before After 1.6 Before After 8 1.4 7 1.2 6 1 5 0.8 4 Meanrank 0.6 3

Mean abundance Mean 0.4 2 0.2 1 0 0 Jan- Apr- Jul- Oct- Jan- Apr- Jul- Oct- Jan- Apr- Jan Apr Jul Oct Jan Apr Jul Oct Jan Apr 06 06 06 06 07 07 07 07 08 08 06 06 06 06 07 07 07 07 08 08 Impact Control Control Impact

Figure 5.23: Comparison of Rhynchoedura ornata captures between treatments; (a) mean abundance (b) mean rank per survey. N= 12 for ‘Impact’ sites and n = 6 for ‘Control’ sites. Error bars in a) represent 95 % confidence intervals.

In summary, reduction of Fox activity may have had a positive effect on R. ornata, however low captures of the species and high variability between sites make it difficult to draw definite conclusions. The other lizard species for which sufficient captures were obtained, appeared to have been unaffected, amongst those the threatened Ctenotus taeniatus. Varanus gouldii, a predator of small vertebrates which is potentially suppressed by the presence of Foxes was also unaffected by reduced Fox activity. Captures of eight further lizard species were too low to validate impact analysis.

5.5 Discussion

5.5.1 General In April and May 2007 a series of rainfall events occurred, resulting in a total of about 100 mm of rain. This wet period coincided with the commencement of the Fox baiting programme and may have affected the experiment in several ways: The rainfall complicated the 1080 baiting programme and potentially made it less effective. Deterioration of 1080 is much quicker in wet soil compared to dry soil (Gentle et al. 2007a) and in this study was compensated by increasing the frequency of bait replacements. Bait uptake is likely to have been lower than it would have been in dry conditions. Initially the moisture induced strong scents in the environment may have masked the smell of the bait, and decreased the bait-uptake. Later, the increase in prey abundance in response to rain may have diminished the attractiveness of the poison baits. - 132 -

The effect of combined Fox and Rabbit reduction on small vertebrates and Cats

Such an amount of rainfall greatly improved environmental conditions and triggered a noticeable population increase in many small mammal species. The trends in the mean abundance of the various species (Figure 5.12) illustrate the marked increase in small mammal captures in the second half of the study. Lizards are ectotherms and so their activity and thus capture, is temperature dependent. Hence they showed seasonal fluctuations (i.e. peaks during the warm periods (October and January) and distinct low points in winter (July)) related more to average temperature than a response to rainfall. However, the improved conditions resulting from the rainfall event and an abundance of food may have further reduced lizard activity in winter as over-winter starvation was less of a threat. This may explain the distinctly lower captures in July 2007 compared to the previous winter. The mean temperatures for July were similar in both years (11.8 ± 4.0 in 2006 and 11.6 ± 5.5 in 2007) and thus provide no explanation for the results.

The MBACI study design, which included sampling at ‘Impact’ and ‘Control’ sites as well as before and after the commencement of Fox reduction, provided the best chance to discriminate treatment effects from environmental variation, such as the impact of the rainfall event that coincided with the commencement of the baiting regime. Nevertheless the response to the rainfall could have masked any treatment effects and is considered as an alternative explanation for the results throughout the remainder of the discussion.

5.5.2 Effectiveness of control actions Foxes Causes of Fox mortality other than those inflicted by the applied treatment appear to have interfered with the 1080 baiting program, masking its effects. Around the time when the drop in Fox activity occurred a trial capture program was carried out which found evidence of weakened and emaciated foxes. The poor condition of the Foxes likely resulted in a high mortality event in the population and thus was responsible for the drop in the measured Fox activity. The mortality of the weakened Foxes could have been acerbated by the slight drop in Rabbit numbers following the release of RHD in October 2006. The drop in Rabbit activity thus just preceded the observed decrease in Fox activity.

- 133 -

The effect of combined Fox and Rabbit reduction on small vertebrates and Cats

Baiting with 1080 thus commenced at a time when Fox populations were very low, it is thus not surprising that no statistical evidence was found to affirm the effectiveness of the poison baiting. However, two factors suggest that the 1080 baiting scheme was respsonible for keeping Fox activity/abundance suppressed at the ‘Impact’ sites: Firstly, Fox activity changed from slightly higher values at ‘Impact’ sites pre-baiting to consistently lower values at ‘Impact’ sites compared to ‘Control’ sites in the post- baiting periods. And secondly, the natural seasonal increase in Fox activity in October (due to the cubs leaving the dens and becoming independent) was present in both treatments prior to Fox control attempts but was only present at ‘Control’ sites where Fox population were left unaltered. At the ‘Impact’ sites Fox activity dropped to zero, thus reaching a low rather than a high. Despite lack of statistical evidence, when considering all of the above there is considerable evidence that Fox activity was successfully reduced at ‘Impact’ sites compared to ‘Control’ sites.

Rabbits The outcome of the Rabbit control programme was somewhat ambiguous. The index of Rabbit activity based on track plot data suggests that the release of RHD resulted in a significant drop in Rabbit activity/abundance. However its effect was only of short duration. Rabbit populations recovered quickly as indicated by a significant increase in activity 6-9 months later. The effectiveness of RHD was presumably hindered by a high immunity towards the virus that was inherent in a previously exposed Rabbit population. A lack of suitable carriers of the virus (i.e. flies, mosquitoes, ticks, fleas etc.) could be an additional explanation but at least flies and mosquitoes were abundant at the time. Even so some Rabbits may have been affected by RHD Rabbit numbers remained relatively high allowing a rapid rebound of populations.

Warren destruction is generally a very effective method to control Rabbits (Martin and Eveleigh 1976), and so the small impact it had in this study was unexpected. A possible reason but unlikely explanation could be a high proportion of Rabbits that were surface- dwelling and not dependent on a warren. However, the utilization of a warren is likely to be crucial in the harsh environments of the arid zone which would limit the proportion of surface dwellers. Also recolonization, despite the lack of existing warrens to facilitate it, could be high in sand-dune habitats. Reopening warren entrances or

- 134 -

The effect of combined Fox and Rabbit reduction on small vertebrates and Cats

digging new warren systems in the sandy substrate is a relatively small effort compared to other areas where the soil is harder.

Rabbit reduction actions were performed in the same manner on all sites but there was no evidence of long-term effectiveness. Thus the intention of creating refuges from predation by reducing or removing the predators preferred prey and thus making sites less attractive as hunting ground, was not met. The lack of impact of the applied Rabbit reduction allowed the assessment of a potential effect of Fox reduction on Rabbits.

5.5.3 Impact of reduced Fox activity on Rabbits There was a general increase in rabbit activity following a population low that was evident just after baiting commenced, but this population increment was similar for ‘Impact’ and ‘Control’ sites. Fox reduction therefore had no obvious effect. Rainfall (see 5.5.1) was likely responsible for the increase of Rabbits in the study area and the strong response it evoked in the Rabbit population. Thus Rabbits may have escaped predator regulation due to a high population growth rate triggered by the rainfall and population may have exceeded the density above which impacts through predation are negligible as predicted by predator-prey-regulation models (Pech et al. 1992; Pech et al. 1995; Sinclair and Pech 1996; Sinclair et al. 1998). The observed result may therefore support the doomed surplus hypothesis (Errington 1946; and see also Banks 1999).

5.5.4 Impact of reduced Fox activity on Cats A wide body of anecdotal and experimental evidence suggests that Foxes are a superior competitor to Cats (Molsher et al. 1999; Risbey et al. 1999). Hence an increase in Cat numbers is a likely result of a reduction in Fox abundance. An increase of Cats was indeed observed towards the end of the study and after Fox reduction, for both ‘Impact’ and ‘Control’ sites. However, the sample size for Cats was too low to analyse whether the increase was statistically different between the treatment groups. Rather than the reduction in Fox activity, a rainfall-induced increase in prey populations is the more probable explanation for the rise in Cat abundance. Cat densities appear to be particularly variable over time in arid environments and presumably reflect the ‘boom or bust’ cycles of productivity that are driven by unpredictable rain (Dickman 1996b). Above-average rainfall has been the suspected cause for increased Cat abundance observed in other studies (e.g. Dickman et al. 1999; Burrows et al. 2003).

- 135 -

The effect of combined Fox and Rabbit reduction on small vertebrates and Cats

5.5.5 Impact of reduced Fox activity on small mammals and lizards The removal (or addition) of a top-predator species will almost certainly be followed by a substantial change in the prey-community with follow-on changes to the whole community, through lower level predation and competition. Nevertheless in this study most small mammal and lizard species remained unaffected by the reduction of Fox activity. The lack of a response of most species may reflect a relatively greater influence of rainfall and no significant impact of Fox predation on these species. Alternatively the duration of the study was too short to detect a population enhancement in response to rapidly lowered predation and/or competition for resources from other community members.

However, there was some evidence which suggested a positive effect on the small mammal species Pseudomys hermannsburgensis, Mus musculus and the geckonid Rhynchoedura ornata. The effect became evident as higher mean abundance during population peaks in P. hermannsburgensis and a prolonged increase and in M. musculus as a higher mean abundance at the end of the study at ‘Impact’ sites compared to ‘Control’ sites. The gecko R. ornata was more abundant at sites where Foxes were reduced compared to the ‘Control’ sites with unaltered Fox densities. These conclusions have been drawn from significant divergence in the trends of species abundances following treatment rather than actual significant differences in species abundance. High variance within and between sites of the same treatment and low captures negated the latter more definitive analysis.

Why some species of small mammals and reptiles may be affected by Fox predation whereas others appear seemingly unaffected, can have a multitude of reasons, including differences in activity times (nocturnal or diurnal habits), reproductive potential (small or large clutch size), habitat preferences (foraging in dense or on vegetation versus preferring open spaces), mode of predator avoidance (camouflage or ability for swift escape movements) and size (Foxes may prefer larger prey). Too little information is available on the biology of many of the small mammal and reptile species as to try and explain why P. hermannsburgensis, M. musculus and R. ornata may have been affected but not the other small mammal and reptile species. In addition to the characteristics of the potential prey species, species could also benefit, or be negatively affected, by indirect effects of Fox removal, i.e. changes in the community of non-native and native

- 136 -

The effect of combined Fox and Rabbit reduction on small vertebrates and Cats

predators and the resulting changes in the small vertebrate community. The species composition after removal of a top predator will depend on which other species replaces the predator and what type of predator it is (generalist or specialist). Fox control could have meso-predator release benefits to predators other than Cats. Olsson et al. (2005) for example found a considerable increase in Sand Goannas (Varanus gouldii) after Fox removal, which as a top predator in the absence of the Fox shaped the prey community. No separate surveys for V. gouldii were conducted as part of this study but the abundance of juveniles was monitored as part of the small vertebrate trapping. If a linear relationship between the abundance of adult and juvenile Sand Goannas is assumed, then a comparison of captures of juvenile V. gouldii with treatment allows similar conclusions for the entire V. gouldii population. However, with this assumption there was no indication in this study that V. gouldii was affected by Fox reduction. The populations of other native animals preying on small vertebrates, such as diurnal and nocturnal birds of prey, were not monitored and this is an acknowledged shortcoming of the study (see section 5.5.7).

5.5.6 Conservation implications Non-native predators like Foxes and Cats are relative newcomers in ecological terms in Australia and systems have probably not yet reached new stable states (Glen and Dickman 2005). Removal of introduced predators is assumed to return the remaining animal communities closer to a species composition and abundance that has been historically shaped by natural selection. In the case of Australia, information on the characteristics of communities prior to Fox, Cat and Rabbit impacts, is as good as non- existent (Banks et al. 1998; Burgman and Lindenmayer 1998; Trigger et al. 2008). The co-existence of several non-native species, that all have a direct effect on native communities and exert additional pressure on those communities through indirect effects resulting from interactions with each other, complicate the achievement of such conservation outcomes. A known draw-back of Fox reduction is the potential increase in abundance and thus impacts of Cats and Rabbits (see review in 5.1.4). However, an increase in House Mice in response to reduced Fox activity has not been previously recorded in the literature. Such an effect may be of concern due to the potential for competition of House Mice with native small mammal species or even predatory effects. House Mice are known for cannibalism and Wanless et al. (2007) report increased juvenile mortality in young seabirds due to predation by House Mice. Then

- 137 -

The effect of combined Fox and Rabbit reduction on small vertebrates and Cats

again nest sharing of House Mice and Sminthopsis crassicaudata was frequently observed by Morton (1978) and a number of studies have concluded that M. musculus had no apparent negative impact on sympatric native small vertebrates (Masters 1993; Moseby and Read 1999; Read 1999; Moseby et al. 2009). There may be no good reason to believe that any native species has become extinct or even suffered a large range reduction as a result of the spread of the House Mouse (Watts and Aslin 1981). In the study area the House Mouse persists at relatively high abundance even though the cause of its establishment (commensal relationship with people) has been removed (see Chapter 4). It is unclear that it now occupies a formerly vacant niche. Its potential impact remains understudied and the threat emanating from the House Mouse may be underestimated. Obviously the benefits of Fox removal should be measured by the response of the species known to be threatened by Fox predation. However, often those species occur at low densities and records are too few to establish population measures or have sufficient power for statistical analysis. This problem was also encountered in this study. The House Mouse was the most abundant small mammal and P. hermansburgensis was the most common native species whereas captures of other native small mammals were low overall. The degree to which the extent, if any, of a change in community and/or response of common (even non-native) species can be an indicator for the potential of recovery of threatened species is an issue that needs further investigation.

5.5.7 Study limitations Many control programmes of exotic animals have been criticized for their lack of clear measures of success or performance of control efforts, thus restricting the conclusions that can be drawn (for a detailed review see Reddiex et al. (2004)). Criticism includes the lack of benchmarks upon which to evaluate success and benefits of control measures, lack of non-treatment areas or random allocation of treatments in monitoring designs and the choice of appropriate spatial and temporal scale for both the pest animal and native species of interest (Braysher 1993; Hone 1994). Few studies look at the response of the community as a whole and most monitor the response of (usually) one species of interest (Davey et al. 2006). Furthermore, alternative hypotheses to predation, such as competition with herbivores, or the potential influence of native predators (e.g. birds of prey, goannas) are rarely considered (Robley et al. 2004). A general confounding factor for manipulative studies in the rangelands is the different grazing

- 138 -

The effect of combined Fox and Rabbit reduction on small vertebrates and Cats

history of the study sites which may result in underlying differences of the study sites which may affect results (Pringle and Landsberg 2004).

This study attempts to overcome at least some of the above-mentioned shortcomings of other similar studies. All sites were located in a similar habitat of low vegetated sand dunes. The existence of underlying site differences due to the grazing history prior to the beginning of the manipulative experiments was investigated and dismissed for the variables measured here (see Chapter 4:). The immediate pre-treatment condition was quantified, and treatment and reference sites were established without identifiable underlying biases. Simultaneous control of Foxes and Rabbits was applied with the goal of intensifying the Fox effect on the subject species. The abundance of an alternative non-native predator, the Cat, was estimated. The impact of the treatment was assessed at community and individual species levels.

Nevertheless budget and personnel limitations, the requirement to work in the large but nevertheless limited area of the Park, as well as the time constraints of a PhD study prevented an optimal (and somewhat idealistic) experimental design according to the recommendations proposed by Reddiex and Forsyth (2004). In a highly stochastic environment like the Australian arid zone, only a snapshot can be obtained and results may vary with a different run of seasons. Thus some study limitations were unavoidable.

Spacing of treatment sites and replication The restriction of the study to the area within the National Park and the land system of the sand dunes, provided challenges regarding the spacing of treatment sites. The distance between the treatment areas and sampling sites within treatment areas was maximised as much as possible but it did not exceed eight kilometres for treatment areas and four kilometres for sampling sites (Figure 5.1). Dispersal distances close to or exceeding the distance between treatment areas are known to occur (mean of 11 km (Coman et al. 1991), mean 3.5 km (Marlow 1992) and even 200 km (Saunders et al. 1994) which is surely an exception). However dispersal movements are only performed in a very limited time period in an animal’s life and so do not represent the usual. The size of home ranges or even core ranges is the relevant measure to determine whether treatment areas are independent or maybe not. The home ranges of Foxes, especially

- 139 -

The effect of combined Fox and Rabbit reduction on small vertebrates and Cats

core home ranges, are likely to be too small (see comparison of home ranges (340 -1611 ha) in Saunders et al. (1994)) to span over two treatment areas or even include two of the sampling sites. However, data on the movements on Foxes in arid and semi-arid areas is lacking and thus there is the possibility that treatment areas were not truly independent. Nonetheless, inter-treatment distances were sufficient to create a discrepancy in Fox abundance.

The independence of individual sampling sites and even dunes is assured for lizards as they have very small home ranges (e.g. Read 1999; Read 2002). Less certain is the independence of sites for small mammals. Although many arid zone small mammals (or at least a great part of their population) are fairly sedentary (e.g. Southgate and Masters 1996; Moseby and Read 1999 and this study (see Chapter 6)) mobility can be high with long-range movements of several kilometres (e.g. Predavec 1994)). The possibility of pseudoreplication (Hurlbert 1984) is acknowledged as a drawback in the design but statistical methods such as Friedman’s test were chosen to accommodate repeated and related samples. Some statistical power was lost and this is acknowledged in the interpretation of the results.

Study duration To quantify the indirect consequences of controlling Foxes and Rabbits it is often argued that long-term studies (greater than at least three years or better five years) are required (Reddiex and Forsyth 2004; Davey et al. 2006). Clearly, the timeframe of a PhD does not allow for studies of such duration. Data for this part of the projct were collected over the period of two and a half years with treaments applied half-way through the study. A significant rainfall event in May 2007, which coincided with the beginning of the 1080 control operation, may have masked potential treatment impacts. However, it also improved capture rates and sample sizes and therefore statistical power and compensated somewhat for the relatively short timeframe of the study. Small mammals in the arid zone are well known for their high reproductive potential and rapid population increases following rainfall. The study duration was sufficient to have picked up such population increases. In contrast it is less certain whether the study period was sufficient to detect changes in reptile populations. Reptiles can only reproduce during the warmer months of the year. For most of the reptile species investigated in this study very little if any information is available regarding their

- 140 -

The effect of combined Fox and Rabbit reduction on small vertebrates and Cats

reproductive biology and thus their ability of rapid population increases (Shine 1985). Some geckoes produce only a single egg, and a clutch size of two eggs is characteristic of almost all Australian geckoes (Bustard 1970) but knowledge on how many clutches are laid per year is lacking. It is thus difficult to judge if a post-treatment period of just over a year is long enough to expect detectable changes in the reptile populations. Conclusions are therefore drawn very cautiously.

Monitoring Ideally the impact of an introduced species should be monitored by assessing the damage it causes. It is often too difficult and/or impractical to monitor impacts, and instead monitoring the species’ population size is used as an indicator of associated damage with the untested assumption that higher densities equate to higher impacts (Edwards et al. 2004). Even though the temporal and spatial variation in environmental conditions ensures that the relationship will vary case-by-case (Lee 2000).

The evaluation of control programmes, as well as general monitoring programmes, is limited by the lack of reliable techniques to monitor changes in the abundance of many species. The current techniques available for Foxes and Cats (bait uptake, spotlight count, track plot activity and scat counts) as well as the use of traps for measuring small mammal and lizard abundance and diversity, are generally imprecise and/or have restrictions on their application. The main drawback is that the relationship between changes in the index and actual abundance remains generally untested, so that activity is equalled with abundance. Activity of animals however varies with environmental conditions. In the absence of better practicable options, the methods employed here are nevertheless commonly used. In studies like this one, where relative measures are used for comparison between treatments such inherent inaccuracies are less severe.

Native predators Varanus gouldii was measured using pitfall trapping but given the bias towards captures of juveniles (the size of adults of about a metre total length allowed escape from pitfalls) an (additional) alternative method such as recording goanna tracks on track plots, would have added valuable information. Monitoring of the abundances of other alternative predator like birds of prey (diurnal and nocturnal) were considered but encounters were infrequent (generally low populations) and site-specificity unclear (high mobility) and so an effective index of abundance could not be derived. For snakes - 141 -

The effect of combined Fox and Rabbit reduction on small vertebrates and Cats

accurate species identification would have presented an additional complication as it necessitates handling and close examination (e.g. to count scales) of the live and potentially deadly animals.

Setbacks The study also suffered from a series of setbacks. Part of the original project was to use GPS tracking technology to gain information on the movement of Foxes in the study area. Information from these data would have been used to evaluate whether the study design, in particular the relatively small distance between the treatment sites, was suitable and to help assess the effectiveness of the track plot method to determine Fox activity and how this relates to Fox abundance. It was also planned to time the commencement of the 1080 baiting in such a way that some battery life remained in the collars so that the GPS data could have been used to monitor the effectiveness of the baiting campaign. Unfortunately, the maker of the collars experienced problems in the manufacture of the devices leading to repeated delays in their delivery with the collars finally delivered more than a year later than anticipated. Obviously the movement data of the Foxes needed to be collected prior to the Fox baiting regime and hence Fox baiting needed to be held back until the tracking data were collected. This reduced the time data could be collected post-baiting, leading to increased uncertainty about the results that were obtained. The majority of Foxes that were trapped to be fitted with the GPS collars was in very poor condition and emaciated and found to be unfit for safely carrying the tracking collars. This led to the discontinuation of this part of the project.

5.6 Conclusion Fox populations were significantly reduced during the study period (albeit through other causes than the applied treatment) and populations at ‘Impact’ sites were successfully suppressed (as compared to untreated ‘Control’ sites) for the remainder of the study as a result of 1080 baiting. The application of common controls on Rabbit populations had no clear effect (i.e. one dependent on the index of Rabbit abundance chosen) in the sand-dune habitat of this study. The indication is that Rabbit control in such a habitat will prove challenging and may require alignment with a prolonged lack of rainfall.

An increase of Rabbits following Fox reduction did not occur, probably because Rabbit population growth rate was high following a rainfall event that coincided with the Fox treatment and Rabbit densities were such that they escaped predator regulation. Cat - 142 -

The effect of combined Fox and Rabbit reduction on small vertebrates and Cats

abundance did increase but the relative effects of treatments and rainfall could not be differentiated due to the low abundance of Cats. A response to rainfall rather than Fox reduction was the more parsimonious explanation consistent with some other results.

Most small vertebrate species remained unaffected by the reduction in Foxes, and so Fox impacts, if any existed, were considerably smaller than between site variability and seasonal fluctuations. In contrast, Fox removal may have benefitted some species, namely the rodent Pseudomys hermannsburgensis and the geckonid Rhynchoedura ornata. Most evident was the increased abundance of the House Mouse upon reduction of Fox predation. This may indicate a considerable impact on this introduced species.

The acknowledged shortcomings of the study prevent the extrapolation of results beyond Sturt NP but nevertheless the study and its results provide a small piece in the puzzle of understanding interactions between Foxes, Cats and Rabbits and the effects of Fox reduction on the small vertebrate species of the arid zone.

- 143 -

The effect of combined Fox and Rabbit reduction on small vertebrates and Cats

Chapter 6: The Ecology of the Dusky Hopping Mouse (Notomys fuscus)

6.1 Introduction

6.1.1 Australian arid zone rodents A minimum of 60 species of native rodents existed on the Australian continent and its surrounding off-shore islands at the time of European settlement in 1788. Since then at least eight species of native rodents have become extinct and 21 others have experienced a reduction in population size and geographical distribution (Dickman 1993; Lee 1995; Strahan 2002). Semi-arid and arid regions have experienced a disproportionally large loss of rodents. In western NSW for example 65 % of the region’s native rodent fauna is thought to have been lost (Dickman 1993) and all extant native rodents in western NSW are listed as threatened species.

Of the small vertebrates, desert rodents pose a particular challenge for conservation management. A common characteristic of many Australian desert rodents are large fluctuations in their population densities. Rainfall is usually considered to be the major influence on populations of these rodents (Finlayson 1939; Dickman 1993; Predavec 1994; Dickman et al. 1999). Populations increase after rainfall (with a time-lag) and the subsequent increase in food abundance (Predavec 1994; Southgate and Masters 1996; Dickman et al. 1999). With the onset of drought conditions populations can crash precipitously. For these ‘boom and bust’ species low densities, range reductions and local extinctions are (to current knowledge) a normal and natural part of their ecology. At the low density phase of their population cycle such species are extremely vulnerable to environmental and demographic stochasticity. Therefore, the preservation of a healthy source population from which the population can rebuild under favourable environmental conditions is imperative in the conservation of irruptive species. Separating the effects of natural environmental stochasticity from alien threatening factors and establishing if a population is in decline or experiences a natural population

- 144 -

The ecology of the Dusky Hopping Mouse (Notomys fuscus) low, after which an irruption will follow, is a challenge. The irruption phases seem to be shorter than the low-density phases and thus population densities of desert rodents are generally low, so that adequate sample sizes in studies are difficult to obtain. This together with other obstacles that complicate research in arid Australia (e.g. remoteness of and long distances to study areas, weather-dependent accessibility, extreme temperatures) (Haythornthwaite 2007) explains why Australian native rodents remain understudied and extensive gaps persist in knowledge about most aspects of the biology and ecology of many species.

6.1.2 The Dusky Hopping Mouse (Notomys fuscus) The diversity and abundance of the Australian Hopping Mice (genus Notomys) have been greatly reduced in the last two centuries. Of 10 known Notomys species (Van Dyck and Strahan 2008) five are extinct, two are classed as vulnerable and only three species are considered common (EPBC Act). The Dusky Hopping Mouse (Notomys fuscus) is one of the two Hopping Mouse species considered rare. Based on capture rates during an intensive survey in the 1990’s, Moseby et al. (1999) estimated there to be no more than 10,000. The conditions at the time were recurrently dry. A comparison of historical and present day distribution records reveals a dramatic decline in the species’ distribution since European settlement (Figure 6.1). Currently it occurs in fragmented and restricted populations (Moseby et al. 1999; Moseby et al. 2006) that are predominantly in South Australia and QLD. But it is also known from the NT and since 2003 from north-western NSW (Dowle 2004).

Nationally, N. fuscus is listed as ‘vulnerable’ under the Environment Protection and Biodiversity Conservation Act 1999 (Commonwealth) (EPBC Act). Prior to the commencement of the EPBC Act, it was listed as ‘vulnerable’ under Schedule 1 of the Endangered Species Protection Act 1992 (Commonwealth). On a State level, it is listed as ‘vulnerable’ also in South Australia under the National Parks and Wildlife Act 1972 but as ‘endangered’ in NSW (Threatened Species Conservation Act 1995), Queensland (Nature Conservation Act 1999) and in the Northern Territory (Territory Parks and Wildlife Act 2000).

- 145 -

The ecology of the Dusky Hopping Mouse (Notomys fuscus)

Figure 6.1: Distribution of the Dusky Hopping Mouse, Notomys fuscus. The dark shaded area illustrates the current distribution, the light shaded area the former distribution (since European settlement) and the black dots are locations of fossil records. Map taken from Van Dyck and Strahan (2008).

Notomys fuscus has been little studied and its biology remains poorly known. The older literature mainly reports captures of N. fuscus with descriptions of the animals, capture locations and habitat (Sturt 1848; Finlayson 1939; Aitken 1968; Aitken 1969) or focuses on aspects of reproduction and reproductive anatomy of captive animals (Crichton 1974; Aslin and Watts 1980; Breed 1982). Watts (1970) provides some insight into the diet of N. fuscus. In more recent years Moseby et al. (1999) reviewed the distribution, habitat and conservation status of N. fuscus and Moseby et al. (2006) investigated differences in population dynamics and movement between two populations in South Australia and Queensland. Apart from that there are few and often anecdotal observations on the ecology of N. fuscus.

The factors that caused the species to decline and limit its populations have not been identified or managed (Dickman et al. 1993; Lee 1995; Moseby et al. 1999; Moseby et al. 2006). As for most small to medium sized mammals in the Australian arid zone, land degradation, competition with and predation by feral animals are assumed to have

- 146 -

The ecology of the Dusky Hopping Mouse (Notomys fuscus) caused and/or contributed to the decline of N. fuscus and remain likely threatening factors (Moseby et al. 1999; Department of the Environment, Water, Heritage and the Arts 2008a).

6.2 Aims The objective of the study described in this chapter was to fill knowledge gaps in the behavioural ecology of N. fuscus within time constraints imposed on the duration of the study and the unknown duration of an irruption in the population. The results are synthesised with current knowledge and used to create a framework for conservation management. Due to the limited knowledge of the species, the initial aims were quite broad and exploratory and included the investigation of:

1) Temporal and spatial distribution in Sturt NP 2) Temporal and spatial activity 3) Diet 4) Social organization

Progress towards these aims was necessarily adaptive and further questions arose and more specific hypotheses were developed and investigated over the course of the study. These specific hypotheses are addressed as part of the broader aims from which they emerged. The methods and results for each aim are presented in the separate sections.

6.3 Temporal and spatial distribution The centre of the distribution of N. fuscus lies in north-eastern South Australia but extends across the state borders into Queensland and New South Wales (Van Dyck and Strahan 2008). In NSW the only references to the species, fossil or recent, are known from the region encompassed by Sturt NP. Prior to 2003, when a number of N. fuscus were captured during an Honours project (Dowle 2004), the species was ‘presumed extinct’ in the state. Previous to this rediscovery, the last evidence of N. fuscus was from 1845, as a reference in Charles Sturt´s journal (Sturt 1848). He relates his observations of indigenous Australians that collected sackfuls of ‘jerboas’ in the sand dunes around Fort Grey. ‘Jerboa’ is not a species-specific term but rather refers to a Hopping Mouse. Even though the location and habitat strongly suggests the species was N. fuscus, the collected Hopping Mice could have also been N. cervinus (Ellis 1993) or a combination of species. The capture of N. fuscus in 2003 provides the only confirmed

- 147 -

The ecology of the Dusky Hopping Mouse (Notomys fuscus) evidence prior to the beginning of this study but a number of local residents (e.g. pastoralists, Wild Dog Destruction Board employees and rural workers) came forward after the rediscovery to report their knowledge of the existence of Hopping Mice in the area and occasional but temporally very variable sightings.

6.3.1 Methods A strategic search for signs of N. fuscus presence (burrows and tracks) was performed by travelling along the approximately 150 km of service tracks through the Strzelecki dune system in the north-western section of the Park. At two kilometre intervals, 500 m sections of the dunes to the left and right of the track were examined for N. fuscus presence by walking along four parallel transects on the dune (each edge of the broad crest of dune and the mid-slope on either side of dune). Surveys were conducted three times, in July 2005, January 2008 and June 2008. Eighty-three dune sections were assessed on each of the surveys.

The presence of N. fuscus was readily determined on the basis of footprints. The sandy substrate of the dunes provided excellent conditions to view tracks, even those left by small animals. Due to their mainly bipedal and saltatorial form of locomotion Hopping mice leave unique footprints that can be easily distinguished from all other small mammal species occurring in the area. In addition N. fuscus are unique to other small mammals that occur in Sturt NP in that they tend to repeatedly follow the same trails when moving around thereby creating well defined and highly visible ‘runways’. Pseudomys hermansburgensis is the only other species that creates ‘runways’, but they are locally very restricted and very different in appearance. The width and detection of the ‘runways’ increased with the number of mice that travelled along them and thus they were more pronounced as the mices’ density increased. This fact was used to not only classify transects as presence or absence sites, but also to provide some measure of N. fuscus abundance. Notomys fuscus activity was rated on the basis of the abundance of footprints and the intensity of ‘runways’ (Table 6.1 and Appendix 15 for photographs of examples of the activity scores). The activity score was used as an index for the density of N. fuscus. Additional data based on captures of N. fuscus and records of footprints on track plots were collected in the course of the monitoring of small vertebrate assemblages conducted for the other parts of the thesis. For details regarding the methodology used

- 148 -

The ecology of the Dusky Hopping Mouse (Notomys fuscus) for pitfall- and Elliott-trapping and the usage of track plots, please refer to Chapter 3. Precipitation records for the Fort Grey Station as representative of the study site were obtained from the Bureau of Meteorology, Australia.

Table 6.1: Criteria used to score Notomys fuscus abundance on transects

Score Evidence of Notomys fuscus activity 1 One to few individual tracks, no runways 2 Many individual tracks and/or narrow runways 3 Network of well used and very obvious runways

6.3.2 Results

Temporal distribution Notomys fuscus was present in the area in 2003/2004 (Dowle 2004). Between then and mid-2005 the populations must have collapsed as the species could not be detected, at the start of this project, neither through transect walks nor through trapping. The first N. fuscus since 2003 was recorded in November 2006 during this project, when a male specimen was captured in a pitfall trap. About six months later evidence of N. fuscus in Sturt NP increased and they were well established in the area towards the end of the fieldwork period in October 2008, where they were virtually omnipresent on the study site (Figure 6.2). During visits to the study site after this date N. fuscus footprints continued to be readily found on the dunes. In April 2009 N. fuscus were still abundant and several individuals were caught (capture rate 0.029; 11 animals in 384 trap-nights) during a biodiversity monitoring programme (pers. comm. Martin Schulze, George Madani, Sandra Penman, Project: 'Trends in native fauna in the Western NRM Region'; Monitoring Evaluation and Reporting Unit, Conservation Science Section, Scientific Services Division, DECC).

- 149 -

The ecology of the Dusky Hopping Mouse (Notomys fuscus)

30 100

90 25 No. of track plot visitation 80

70 20 by

captures 60 N. fuscus 15 50

N. fuscus 10 30

20 5 No. of 10

0 0 Jul- Oct- Jan- Apr- Jul- Oct- Jan- Apr- 06 06 07 07 07 07 08 08

Figure 6.2: Abundance of Notomys fuscus between January 2006 and April 2008. The solid line refers to N. fuscus captures, the dotted line to track plot visitation by the species.

To investigate the influence of rainfall on the fluctuation of N. fuscus populations in Sturt NP the data from a previous study (Dowle 2004) were included. Early in 2003 N. fuscus were present (perhaps the tail of a population irruption following the above average rainfall in 2000) but declined over the course of that year despite reasonably good rainfall during the year and must have subsequently disappeared from the area altogether (Figure 6.3). Rainfall and environmental conditions were below average in 2004, 2005 and 2006. No records of N. fuscus were made in 2005 and during most of 2006. The first N. fuscus was caught in late 2006 despite ongoing drought conditions. In subsequent surveys in early 2007 N. fuscus were reliably recorded on track plots, long before the significant rainfall event in May 2007. Apparently rainfall did not trigger the reappearance of the mice in 2006/2007 but certainly helped in their persistence.

- 150 -

The ecology of the Dusky Hopping Mouse (Notomys fuscus)

Figure 6.3: Rainfall and Notomys fuscus captures over time. Continuous rainfall data were available for the entire timeframe. Small vertebrate surveys were conducted only during periods of fauna monitoring, as indicated by the bars below the x-Axis.

Spatial distribution The percentage of dunes at the pitfall trapping sites with evidence of N. fuscus, either in the form of animal captures or footprints on track plots, gives insight into their distribution on a fine scale and a measure of the patchiness in the species’ occurrence. The recording of footprints on track plots was the more sensitive method for detecting N. fuscus abundance, with consistently higher detection rates (Figure 6.4) and will therefore be the focus in the presentation of results.

Regular surveys conducted throughout the project, including Elliott and pitfall trapping and usage of track plots, indicated N. fuscus were present only on a very small percentage of trapping grids (< 20 % in April and July 2007) but within a few months they had spread to and colonized more than 50 % of the grids. Their site occupation

- 151 -

The ecology of the Dusky Hopping Mouse (Notomys fuscus) reached a peak in January 2008 when N. fuscus were present at all but 20 % of trapping grids.

100 Captures 90 Footprints evidence 80 70 60 N. fuscus 50 40 30 20 10 0

% of trapping grids with Ja Ap J Oc Ja Ap J Oc Ja Ap n- r- ul- t- n- r- ul- t- n- r- 06 06 06 06 07 07 07 07 08 08

Figure 6.4: Percentage of trapping grids with Notomys fuscus evidence in the form of captures and footprints on track plots.

The transect surveys provide a broader view on the distribution of N. fuscus in the western part of Sturt NP (Figure 6.5). Notomys fuscus were widely distributed and present in all parts of the study area. By January 2008, tracks of N. fuscus were found on 72 % of all surveyed dunes. This number increased further and in July 2008 evidence of N. fuscus was found on 89 % of all dune sections. Whereas the number of dune sections that had N. fuscus present increased, the mean of the activity scores at the presence sites slightly decreased (Jan 2008: 1.82, n = 60; July 2008: 1.77, n = 74). This indicates that N. fuscus colonized more habitat but on average their abundance at established presence sites decreased slightly.

- 152 -

The ecology of the Dusky Hopping Mouse (Notomys fuscus) a) b)

#S #S #S #S #S S #S #S#S #S #S

#S #S #S S S #S #S S #S #S S #S #S #S #S S #S #S #S S

#S #S#S #S S S #S #S #S S #S #S #S #S #S #S S S S #S #S #S #S #S #S #S #S #S #S #S S #S #S #S #S S S S #S #S #S #S #S S #S #S #S #S #S #S #S #S#S #S #S #S#S #S #S S#S #S #S#S #S S #S #S #S #S #S S #S #S #S #S #S #S #S S #S S S #S #S #S #S #S S #S S SSS #S #S S S #S #S #S S #S #S #S #S #S #S #S S #S #S #S #S #S #S #S #S #S #S #S #S #S #S Hopping Mouse #S#S #S#S Activ ity score Hopping Mouse #S Activity score #S S 0 S 0 #S 1 #S S #S #S 1 #S #S 2 N N #S #S 2 #S 3 #S #S 3

0 2 4 6 8 10 Kilometers 0246810Kilometers

Figure 6.5: Spatial distribution and density (activity index) of Notomys fuscus over the study area in a) January 2008 and b) in July 2008.

6.3.3 Discussion Results from this study as well as data from South Australia (Moseby et al. 2006) confirm that not surprisingly N. fuscus, like many other Australian desert rodents, undergoes extreme population fluctuations. In Sturt NP N. fuscus may be present for a certain time period, but become undetectable in the area for years. Once established, the populations in Sturt NP can persist for several years. Since their recent reappearance in 2006/2007 they have been continuously present for >2.5 years and to date the species is still widespread and abundant. Sturt NP lies on the edge of the species’ distribution but similar fluctuations in abundance and absences were also recorded for a location in the centre of its range (Moseby et al. 2006).

- 153 -

The ecology of the Dusky Hopping Mouse (Notomys fuscus)

For N. fuscus in Sturt NP, a convincing correlation between local rainfall and the presence of N. fuscus and/or their abundance could not be established due to lack of long-term data. A sudden reappearance and rapid population increase over just a few months, without a significant rainfall event which could have triggered such a response, suggest immigration of N. fuscus to the area. Notomys fuscus are an extremely mobile and fast moving species which enables them to perform long-distance migration and dispersal movements. Reports of high Hopping Mouse numbers in South Australia preceded the reappearance of N. fuscus in Sturt NP. In February 2007 shortly after the first capture of an individual in Sturt NP in 2006, sightings of large numbers of Hopping Mice were made at ‘Quinyambie’ Station in South Australia (about 90 km south of the study site in Sturt NP) (pers. comm. Renee Visser, ANU). About 6 months later evidence of N. fuscus in Sturt NP increased and by October 2008 they were virtually omnipresent. In July 2008 a rural newspaper reported N. fuscus in plague-proportions in South Australia (Peddler 2008), confirming that the mice occurred at extremely high densities elsewhere. The numbers must have been a multiple of those observed on Sturt NP, to warrant the attention of the press and to be described as a plague.

The observed irruption of N. fuscus appears to have been unusual as not only were extremely high densities observed but N. fuscus also appeared in areas where it had never been recorded previously. Waudby and How (2008) reported a range extension by ca. 70 km to the south east of its previously known distribution in South Australia. In NSW the range of the species extended eastwards, as found in a follow up to this study (Ulrike Klöcker, unpublished data). A specimen was recorded near ‘Whittabrinna’ (ca. 70 km east of closest previously known location) and indubitable Hopping Mouse tracks were found on isolated sandy rises surrounded by gibber plains, near the township of Tibooburra (75 km from closest previously known location) and close to Mt Wood Station (110 km from closest previous known location).

If rainfall was sought as an explanation for the population irruption then the population increase was probably triggered by widespread rainfall over the range of N. fuscus in June 2005. Even though the rainfall totals over the whole year were average with 100- 200 mm, winter rainfall (25-50 mm, locally possible greater amounts) in June 2005 presumably stimulated favourable environmental conditions for a population increase of small vertebrates including N. fuscus. Rainfall in 2006 was below average but

- 154 -

The ecology of the Dusky Hopping Mouse (Notomys fuscus) significant rainfall in 2007 probably allowed N. fuscus to further increase in abundance and to expand their range (see Appendix 13 for maps of rainfall totals in the years 2005- 2007 and the rainfall event in June 2005).

Morton (1990) first coined the concept of refugia , which has since been widely used to explain the population dynamics of small and medium-sized mammals in the desert regions of Australia (Dickman et al. 1995). He hypothesized refugia as resource-rich, highly productive patches scattered through the otherwise resource-poor landscape of semi-arid and arid Australia. Such patches of presumably high and regular food production act as drought refugia and species are able to persist in these critical habitats during dry times and re-colonise other areas in favourable conditions. Letnic et al. (2009) extended this concept to include not only resources as characteristics of a refuge area but also low levels of predation (e.g. due to the presence and abundance of the dingo). The area of Sturt NP does not appear to possess all characteristics required for such a drought refuge for N. fuscus (at least not during the drought period of 2004-2006 as N. fuscus disappeared from the area), even though it apparently provides suitable habitat to support a widespread and high population of N. fuscus over a period of several years.

6.4 Diet Many Australian desert rodents have typically been classified as granivorous. More recently, many of those same species have been described as opportunistic omnivores as their diet includes food items other than seeds when they are available (Murray et al. 1999). The knowledge of the diet of N. fuscus is scarce but the little information that is available suggests that N. fuscus is also an opportunistic omnivore. Watts (1970) quantitatively assessed the percentage of the food types eaten by looking at their occurrence in mouse droppings. He found the diet to consist of 74 % seeds, 23 % green plant material and 3 % insects. The sample was very small (n = 10) and it did not account for temporal variation in the diet as all scats were collected simultaneously. Moseby et al. (1999) observed field-captured N. fuscus to eat piedish beetles (Helea sp.), the flowers and stem of Cynanchum floribundum, leaves of Nitraria billardieri, flowers of Senecio cunninghamii and tips of Salsola kali leaves. Also the consumption of small trapped lizards was frequently observed.

- 155 -

The ecology of the Dusky Hopping Mouse (Notomys fuscus)

Clearly more information is needed on the actual food items and species of plants and/or invertebrates that are consumed. The identification of major food resources might help to identify critical drought refugia (Morton 1992) which are crucial in the conservation of a species with largely fluctuating populations.

6.4.1 Methods The diet was assessed by the types of plant, invertebrate, vertebrate or fungus species consumed rather than their quantity. An extended period of time spent in the field provided plentiful opportunities for casual observations on the foods eaten by N. fuscus. The ‘runways’ of N. fuscus along the dunes connected frequently visited sites and following them led to discoveries of shrubs and other plants or patches that were apparently visited in the search for food (see Appendix 16 for an image of a group of plants that are frequently visited by N. fuscus as is evident by the number of footprints). Windy conditions followed by calm weather presented excellent conditions to follow the tracks that individual mice had left the previous night. Imprints of all four feet instead of two indicated locations where the mice stopped and potentially gathered food. Mouse droppings further indicated locations of frequent visitation or an extended length of stay. Areas with tracks or droppings were investigated more closely for bite marks on the vegetation, evidence of digging activity and the remains of the consumed foods such as seed husks, inedible parts of invertebrate prey and other discarded food parts. Direct observations of diet choice and feeding activity were made when various food items were provided while mice were temporarily held captive for fitting of radio collars, and during radio-tracking.

6.4.2 Results In accordance with what is already known about their diet N. fuscus in Sturt NP fed on a variety of food items, including seeds, leaves, berries, roots and invertebrates (Table 6.2). Seeds were the main part of the diet. The seeds of Acacia ligulata, which were superabundant when produced, were highly sought after but seeds of other plants were included in the diet whenever available. Of these, the seed capsules of the Sandspurge (Phyllanthus fuernrohrii), seeds of the Cattle Bush (Trychodesma zylanecum) and the Mulga Trefoil (Tephrosia sphaeropspora) appeared to be very desirable. All these species produce only small numbers of seeds which ripen consecutively. Nevertheless, the mice visited them regularly during their foraging activities to check if new seeds had ripened and/or fallen to the ground thereby including the location of the plants as

- 156 -

The ecology of the Dusky Hopping Mouse (Notomys fuscus)

‘points of interest’ in their network of runways. The other plants that were usually included in the runway network are the Ruby Saltbush (Enchylaena tomentosa) and various spiney saltbushes (Rhagodia spp.). The berries of the Ruby Saltbush were highly sought after and the shrubs were re-visited often to check for berries that have fallen to the ground. The flesh of the berries was eaten but the core was discarded. The frequency of N. fuscus tracks in the vicinity of Rhagodia shrubs indicated that something associated with the shrub is likely to present a valuable resource but which part could not be determined. Berries were always plentiful underneath the shrubs and were apparently not eaten and no bite marks on leaves and branches were ever detected. Rather than a food resource, the spiny saltbushes may represent shelter.

Masses of caterpillars (several hundred in one trap-night) of small unidentified moth species developed after winter rainfall. The caterpillars mainly fed on the low ephemeral vegetation (most often observed feeding on Fleshy Groundsel (Senecio gregorii)) and thus presented an easily accessible, abundant and protein-rich food source that was exploited by N. fuscus. Like the caterpillars, the moth pupae were also eaten. The caterpillars pupate several centimetres below the sand surface but appear to be easily located by N. fuscus. The mice dug the pupae up, neatly opened the cocoon, ate the developing moth and left the cocoon-shell behind. The consumption of other invertebrate material was only observed once when a radio-tracked mouse grabbed a lacewing off a plant and ate it. Temporarily captive held N. fuscus readily consumed meal worms and so beetle larvae may be part of their diet in the wild as well.

Geckos and skinks were killed and partly eaten when they were trapped together with N. fuscus but the capture and consumption of lizards was never observed outside the confined space of a pitfall trap.

Potential food items that were abundant in the environment, at least seasonally, but were never observed to be consumed include wolf spiders, adult moths, seeds of Cassia spp., Large Pigweed (Portulaca intraterranea), Storksbill/Crowfoot (Erodium sp.), Fleshy Groundsel (Senecio gregorii), Native Camomile (Gnephosis eriocarpa), Sturt’s Pigface (Aizoodon quadrifidum), Clasping Twinleaf (Zygophyllum howittii), Sand Twinleaf (Zygophyllum ammophilum) and Native Parsnip (Trachymene glaucifolia).

- 157 -

The ecology of the Dusky Hopping Mouse (Notomys fuscus)

Table 6.2: List of food items that are part of the diet of Notomys fuscus in Sturt NP

Frequency Species Occurrence Parts eaten Availability of observation always seeds but variable in abundance constantly Sandhill Wattle common (Acacia ligulata) fulicle/ariel periodically, short-term frequently only during seed fall

Ruby Saltbush always common berries constantly (Enchylaena tomentosa) but variable in abundance

Cattle bush sparse seeds periodically after rain frequently (Trichodesma zylanecum)

Moths ubiquitous caterpillar & pupae periodically after rain frequently

New Zealand Spinach ubiquitous leaves periodically after rain frequently (Tetragonia tetragonoides)

Rhagodia spp. common ????? always frequently

Sandspurge seed & capsules sparse periodically after rain frequently (Phyllanthus fuernrohrii) while still green

Buckbush leaves except usually present common occasionally (Salsola kali) spiky tips not in drought

Mulga trefoil sparse seeds occasionally (Tephrosia sphaerospora)

Unknown copperburr sparse leaves occasionally Mulga common seeds once (Acacia aneura) Common Pigface ubiquitous roots periodically after rain once (Portulaca orelacea) Narrow-leafed Hopbush always common seeds once (Dodonea viscosa) but variable in abundance Poachegged Daisy ubiquitous seeds periodically after rain once (Myriocephalus stuartii) Lacewing once rarely Geckos, skinks in confines of pitfall traps

6.4.3 Discussion The observations made during this study confirm earlier quantitative estimates of the diet of N. fuscus and its classification as an opportunistic omnivore. It is primarily granivorous and obtains a great part of its energy requirements through the consumption of seeds which are rich in protein and fat. The availability of food resources is dependent on rainfall and so the seed-dominated diet is complemented with invertebrates, in particular the caterpillars and pupae of moths, and various parts of plants such as berries, leaves (of at least partly succulent species) and occasionally

- 158 -

The ecology of the Dusky Hopping Mouse (Notomys fuscus) roots. Seeds probably become more important and dominant in its diet during dry conditions, as seeds prevail long into dry conditions unlike other food resources.

Whereas the observations collated during this study did not provide quantitative information on the food types consumed, they provided previously unrecorded information on the actual food items and the plant and invertebrate species that are important to N. fuscus. A particularly important resource for N. fuscus in Sturt NP appeared to be the seeds of A. ligulata. The relationship between this shrub and N. fuscus was therefore investigated in more detail (see next section).

6.5 Dusky Hopping Mice and Sandhill Wattles (Acacia ligulata) Early in the project it was noticed that N. fuscus footprints were reliably found in association with A. ligulata and, in some locations, a network of ‘runways’ connected individual shrubs. This together with an accumulation of seed husks and mouse droppings under A. ligulata shrubs suggested that the mice frequently visit A. ligulata shrubs and spend a lot of time around and under the shrubs eating the seeds. This was confirmed later in the project by the data collected from radio-tracked mice. Each of those mice spent a great amount of its foraging time under A. ligulata shrubs.

Acacia ligulata is a shrub or tree widespread in central and southern arid Australia, occurring in all mainland states. It usually grows on sandy soils, commonly on the tops and slopes of red sand dunes. It is particularly common in South Australia (Moore 2005) but is also abundant in the dune system of the study area in the western part of Sturt NP. It produces seeds which are arranged longitudinally in pods. The individual seed is 4-6 mm long and 3-3.5 mm wide. Acacia ligulata is a prolifically seeding plant and large shrubs can produce tens of thousands of seeds. The seeds fall below or not far from the shrubs and, as they are too heavy to be easily moved by wind, considerable numbers of seeds accumulate underneath the plants (see Appendix 17 for an example). As such a high number of seeds are produced they remain part of the seed bank well into dry times. Over the three years of this study A. ligulata seeds were always abundant. Acacia ligulata seeds thus represent a highly clumped, easily locatable and (at least over timeframes of several years) reliable resource.

- 159 -

The ecology of the Dusky Hopping Mouse (Notomys fuscus)

Traditionally indigenous Australians collected the seeds of A. ligulata and ground them to a paste which was then cooked and eaten (Cunningham et al. 1992). A number of animals including ants, Emus, a multitude of other granivorous birds as well as the Sandy Inland Mouse, Pseudomys hermannsburgensis, have been observed using A. ligulata seeds. However, none of these animals appears to have the same reliance on A. ligulata as the N. fuscus.

The importance of A. ligulata for N. fuscus was investigated in more detail by testing the following hypotheses:

1) Acacia ligulata is a determinant of the distribution of N. fuscus and the density at a site 2) Within a site A. ligulata is a determinant of the movements and habitat use of N. fuscus

6.5.1 Methods To address the first hypothesis, the height, two perpendicular width measurements (at the height of greatest perimeter, to the nearest 5 cm) and a seed density estimate were recorded for every A. ligulata that occurred within the same 500 m dune section in which the activity index of N. fuscus was recorded (see section 6.3.3). Living or dead shrubs were considered as long as some seeds could be found on the ground underneath. The seed density was visually assessed within randomly located plots of 50 x 50 cm underneath each shrub and the density rated very broadly on a four point scale as ‘no seeds’ (0), 1 -50 seeds (1), 50-500 seeds (2) or >500 seeds (3). The volume of a shrub was multiplied by the seed density category to obtain a measurement of the quality of a shrub as a resource. Acacia ligulata projected cover (multiplication of width measurements), volume (multiplication of projected cover and height) and adjusted volume (combined seed density) were used in the analysis.

Whereas the N. fuscus surveys were conducted on two occasions (technically on three occasions but no evidence of N. fuscus was found during the first survey), A. ligulata dimensions and seed density were only measured once, in July 2008. While seed density can change rapidly over time, shrub dimensions however would not have changed

- 160 -

The ecology of the Dusky Hopping Mouse (Notomys fuscus) greatly within six months as these shrubs are slow growing. Therefore the one sample of plant dimensions was used to investigate the correlation of A. ligulata variables with the N. fuscus activity scores from either of the surveys. An overall activity score (sum of both surveys) was used as an additional variable to include the aspect of site persistence over time and was also correlated with A. ligulata dimension data. T-tests were conducted to identify if presence/absence sites showed differences in the A. ligulata variables. ANOVA was used for comparison of A. ligulata variables between sites with different activity scores. Subsequent stepwise regression analysis were performed on those A. ligulata variables that differed significantly to determine to what extent the A. ligulata resource can be used to predict the occurrence and density of N. fuscus.

To address the second hypothesis and establish whether A. ligulata determines the space use of N. fuscus the distance between the observed locations of the individual in its environment (location-fixes determined through radio-tracking; see section 6.6.1 for methodological details of radio-tracking) and the nearest A. ligulata shrub was calculated and compared to the distances calculated for a random sample of locations (n = 1000). The random locations were generated in a rectangle that included all observed mouse locations and so its corners and thus side length were defined by the minimum and maximum x and y coordinates of each mouse subject’s locations. The random points were generated and distances calculated using functionalities included in the Animal Movement Extension V2 to ArcView 3.2 (Hooge et al. 1999). The distances were assigned to distance classes and differences between the observed and the random samples tested using Chi- Square in a Fisher’s exact test.

6.5.2 Results

Acacia ligulata and Dusky Hopping Mouse distribution In January sites where N. fuscus were present had a greater abundance, cover and volume of A. ligulata than those sites where no mouse activity was recorded

(t41 = 2.826, p = 0.007; t41 = -2.855, p = 0.007; t41 = -2.964, p = 0.005 respectively, n presence = 30, n absence = 13) (Figure 6.6). In contrast no difference in any of the A. ligulata variables between presence and absence sites was found in July. The number of presence sites became too small (n = 2) when combining the January and July surveys and thus could not be compared.

- 161 -

The ecology of the Dusky Hopping Mouse (Notomys fuscus) a) b) c)

Figure 6.6: Abundance (a), mean projected cover in m2 (b) and mean volume in m3 (c) of A. ligulata at N. fuscus presence and absence sites. Data presented are from January 2008. Error bars represent 95 % confidence intervals.

Acacia ligulata also influenced the density of N. fuscus at a site. The results were most pronounced in January when sites with different activity scores differed significantly in all three A. ligulata variables (abundance: F3, 39 = 3.397, p = 0.027, cover: F3, 39 = 3.953, p = 0.015 and volume: F3, 39 = 4.297, p = 0.010, n = 42). Volume was the best predictor of N. fuscus density but despite yielding significant results (F1, 41 = 9.994, p = 0.003), the regression analysis could explain less than a quarter (19.6 %) of the variation in the activity of N. fuscus between sites. In July N. fuscus density was only correlated to

A. ligulata volume (F3, 39 = 2.838, p = 0.050), but showed a strong trend for A. ligulata cover (F3, 39 = 2.726, p = 0.057). Similarly for inferred site persistence, the combined score for both surveys, the volume of A. ligulata differed significantly (F6, 39 = 2.838, p = 0.050) and cover showed a trend between sites (F6, 39 = 2.022, p = 0.088). Volume of A. ligulata could explain 27 % in the variation of the N. fuscus density. Notomys fuscus density over time was higher at sites with a greater volume of A. ligulata (Figure 6.7).

- 162 -

The ecology of the Dusky Hopping Mouse (Notomys fuscus)

Figure 6.7: Correlation between N. fuscus activity as a measure of density and the total volume of A. ligulata. The N. fuscus activity scores were summed over the two surveys (Jan 08 and July08) to include the aspect of site persistence.

Acacia ligulata and space use of Dusky Hopping Mice The results of the comparison of the distance between known N. fuscus locations and A. ligulata and random locations and A. ligulata showed highly significant differences for all subjects of either sex (Table 6.3). Sandhill wattles have a significant influence on the space use of N. fuscus, with the mice concentrating their movements around the shrubs. For example, the distribution of the observed sample of the mouse ‘Bertha‘ and the random locations across 10-m distance classes, shows that instead of being more or less equally distributed across all distance classes the subject’s locations were skewed towards the shorter distances in the random sample (Figure 6.8). The random sample is also skewed to the left which indicates that the shrub itself is clumped and not uniformly dispersed.

- 163 -

The ecology of the Dusky Hopping Mouse (Notomys fuscus)

Table 6.3: Comparison of the minimum distance to an Acacia ligulata shrub between the actual observed locations of a Notomys fuscus individual and a random sample of locations. Results of chi-square test (Fisher’s exact test). Random locations n=1000.

N actual Mouse Sex df 2 p locations Bertha f 76 9 47.73 << 0.001 Daisy1 f 134 9 64.88 << 0.001 Daisy2 f 78 9 56.14 << 0.001 Maja f 38 4 63.88 << 0.001 Miffy f 86 9 138.37 << 0.001 Minnie f 59 9 18.88 0.014 Susie f 72 9 37.44 << 0.001 Bert m 65 8 108.91 << 0.001 Dennis m 36 9 42.41 << 0.001 Donald m 63 9 63.97 << 0.001 Ernie m 86 9 109.39 << 0.001

Figure 6.8: Exemplary illustration of distribution of the minimum distance to the next A. ligulata shrub between the observed locations of a Notomys fuscus individual and a random sample of locations. N observed = 78, n random = 1000.

- 164 -

The ecology of the Dusky Hopping Mouse (Notomys fuscus)

6.5.3 Discussion Acacia ligulata was a very important resource for N. fuscus and influenced space use as well as distribution and density of the species. The analysis yielded significant results even though A. ligulata is an extremely variable species and the shrubs are likely to vary in their quality as a food resource depending on age, condition and site quality. It was observed that the seeds of some shrubs were never eaten even though they were abundant whereas the seeds of neighbouring A. ligulata shrubs were favoured, although they were scarcer. N. fuscus might be able to assess the seed quality, which was not measured in this study.

The correlation between N. fuscus presence/absence and density of A. ligulata was much more pronounced in January compared to July. One explanation for such differences over time could be that the importance of A. ligulata is a function of environmental conditions and other available resources. There are also density- dependent effects. At the beginning of an irruption of N. fuscus, many sites that have high A. ligulata cover are likely to show no N. fuscus activity because they have not yet been colonized. Similarly, when populations decline there are bound to be sites where the resources are close to being exhausted and the mice have moved on to other locations. In January N. fuscus was already abundant but still increasing, whereas it may have reached its population peak or started to decline slightly by July.

In the approach taken here it is assumed that the number of tracks and the distinctiveness of ‘runways’ on a dune are correlated with N. fuscus density rather than activity. It might however be a combination of the two. With accumulating experience with the progression of the study it was observed that most tracks and the most distinct runways appeared at the time when the young emerged from their burrows and explored the dunes. The number of N. fuscus that used the tracks increased but as juveniles were generally more active than adult individuals so did N. fuscus activity. Thus A. ligulata may be more important for the distribution of reproducing females and their offspring than for the distribution of the species in general.

Moseby et al. (1999) aimed to identify and characterise typical N. fuscus habitat, by correlating habitat characteristics (these included landform pattern and type, surface soil texture and strew cover, complete floristic inventory, plant life form and

- 165 -

The ecology of the Dusky Hopping Mouse (Notomys fuscus) cover/abundance information for individual species) with the presence of N. fuscus. The analysis confirmed the common belief that the occurrence of N. fuscus is tied to sandy substrates with a preference for sand ridges. Other than that, their results did not seem to infer a particular correlation with particular habitat features or plant species and the authors concluded that N. fuscus is present in a wide variety of sand-dune habitats, with the exception of those dominated by Triodia sp. They also concluded that perennial vegetation may be important in the maintenance of stable N. fuscus populations and that it is unlikely that a specific food item determines their distribution. With the attention turned to A. ligulata, however, their results (see Table 3 in Moseby et al. 1999) support the conclusion drawn here, that A. ligulata is a major resource. Only three of the eight site groups based on presence and cover attributes of all plant species, featured A. ligulata as a characteristic species and at two of those site groups N. fuscus was recorded. Notomys fuscus was otherwise only recorded in one other site group. This independent study thus confirms a relationship between N. fuscus and A. ligulata not only for Sturt NP but over a greater part of their distribution. However, in the Cobbler Sandhill area (the location of the remaining site group where N. fuscus was recorded) in the southern part of the distribution, A. ligulata does not occur or is at very low densities based on the information provided by Moseby et al.. (1999). There, the Dillon Bush, Nitraria billardieri is likely to present an important resource for the mice. The discovery of the main resource for a threatened species is of great value for its conservation and allows the refinement of threat abatement and recovery actions.

6.6 Temporal and spatial activity Like most other Australian small mammals N. fuscus are nocturnal. Opportunistic radio- tracking carried out in South Australia (Moseby et al. 2006) provided some insight into movements, but the number of fixes per individual in that study was quite low and did not allow the calculation of a home range. Thus information on temporal and spatial behaviour is lacking and more intensive radio-tracking is required.

6.6.1 Methods The issue of auto-correlation of observations in radio-tracking studies and the consequences for the calculation of home ranges has been a much discussed topic for decades. The concern about autocorrelated location data is that the autocorrelation causes negatively biased estimates of home range size (e.g. Cresswell and Smith 1992; Swihart and Slade 1995). Others however argue that the effect of autocorrelation on

- 166 -

The ecology of the Dusky Hopping Mouse (Notomys fuscus) home range bias is relatively unimportant compared to obtaining a representative sample (i.e. maximise number of radio-tagged individuals, number of fixes per individual and observation time) (Otis and White 1999; Fieberg 2007) and that autocorrelated data may reveal better behavioural information than would independent observations as animals typically move in a non-random fashion and its experience leads it to favour particular foraging areas, travel routes and dens (Lair 1987; Rooney et al. 1998; de Solla et al. 1999). A number of studies have compared the performance of home range estimators and the influence of sample size per individual, numbers of individuals and the effects of independent and autocorrelated data sets but results were not conclusive and thus these studies have provided limited practical guidance to investigators who must develop well-designed radio-tracking field studies. To be on the safe side it is advised that autocorrelation should be avoided or at least minimized and non-parametric range estimates, such as the kernel density estimator, should be used to calculate home ranges. In this study the number of tracked individuals, the time-period they were tracked and the number of fixes were all maximized as far as possible within the constraints of the project (project deadlines, limited manpower and battery life of transmitters). To maximize the number of fixes, the time interval between fixes has to be as short as possible without producing heavily autocorrelated data. The independence of observations is commonly justified based on the knowledge of animal movements and sufficient time between observations so that the animal can relocate to any other point within its home range (Lair 1987; McNay et al. 1994; Swihart and Slade 1997; Kernohan et al. 2001). Some information of movement rates and distances of N. fuscus was available from a previous radio-tracking study (Moseby et al. 2006) which suggested that N. fuscus is a highly mobile animal that can easily cover several hundred meters in an hour. A trial period conducted at the beginning of this study found even higher movement rates with the animals covering several hundred meters in the matter of minutes. Based on this information the time interval between fixes was set to ten minutes. This is quite a short time interval but it fulfilled the commonly used criterion for independent data points as every tracked mouse would have easily been able to reach any point within its home-range within a ten minute time interval. The time interval between fixes was kept constant and kernel density estimation (non-parametric method and relatively unaffected by autocorrelation) was used to investigate home range size as recommended by de Solla et al. (1999). Thus the calculated home range

- 167 -

The ecology of the Dusky Hopping Mouse (Notomys fuscus) estimates are likely to be quite reliable. Sub-sampling to eliminate autocorrelaction prior to home-range analysis was therefore not undertaken but all fixes included in the calculation of home ranges.

Adult (> 27 g) N. fuscus were fitted with custom-made radio-transmitters (2.5 g, brass loop antenna, dental acrylic encasing, 80-100 m range, SIRTRACK Ltd – Wildlife Tracking Solutions, New Zealand) and their movements followed over five consecutive nights, whenever possible. With the aid of the radio-signal the mice were followed continuously and every ten minutes their location was recorded with a GPS receiver (Garmin® 72). A handheld spotlight (Lightforce) with a red filter was used to determine the exact location of the mouse and allow observation of the activity at the time. If the mouse could not be sighted then the approximate position based on the direction and strength of the radio signal was logged. Estimates from signal strength were made once sufficient experience had been gained and inaccuracies were likely to be small given short range of transmitters. Care was taken to keep a distance of at least 10 m to the mice at all times. Most mice were indifferent to the presence of observers and the light allowed continuous observations. Some mice were more timid than others and in those cases the distance to the animal was increased and if the mice showed obvious negative reactions to the light then the use of the spotlight was limited to the brief period needed to determine its location at every time interval but remained switched off the remainder of the time. Apart from the subjects’ GPS position, notes were taken of their location on the dune at the given time, whether they were moving or stationary, if they could be sighted and what activity they were engaged in at the time. The mice were not found to be active continuously but rather to have several periods of activity during a night interrupted by resting periods usually in their burrow. Following an initial exploratory phase, radio-tracking was therefore restricted to the first activity period of the mice.

As many pitfall traps on as many sites as possible were opened at any given time to capture suitable individuals (heavier than 27 g). Wherever possible, data on at least a female and a male mouse per dune section were collected to get insight into potential differences in the use of habitat and resources of the sexes within the same area. However, a very high percentage of the male mice that were collared and released could not be subsequently located despite extensive searching in a radius of 700 m on the dune of capture and release as well as on neighbouring dunes. Safety concerns for the

- 168 -

The ecology of the Dusky Hopping Mouse (Notomys fuscus) researchers in the remote and difficult terrain prevented work at greater distances than that from known roads and tracks, especially at night. At least ten radio-tracking collars were lost that way until it was decided to limit the tracking of male mice to those individuals that had been recaptured at least once at a particular site and thus obviously showed some degree of site fidelity.

After capture in the early mornings, N. fuscus were taken to the field camp for the fitting of radio-collars. As suggested by Anstee et al. (1997) Isoflurane, an inhalation anaesthetic with a wide overdose safety margin, was used to calm the animals for safe fitting of collars. The mice were kept in a terrarium under observation to ensure the correct fit of the collars before their release in the evening. Upon release at the site where they had been captured, the mice were followed back to their burrows. The mice were not tracked on the first night but rather tracking commenced on the following evening which gave the mice time to recuperate from the stress of being captured and handled and allowed them to adjust to carrying the radio-collars. Upon completion of data collection the radio-collared N. fuscus were re-trapped at their burrows or on the trapping grids for the removal of collars.

Home ranges (95 % of fixes) and core ranges (50 % of fixes) were calculated using Kernel Home Range estimation with LSCV as a smoothing factor. This method has been found to be very reliable for the calculation of home ranges (Seaman and Powell 1996; Seaman et al. 1999). The recommended number of fixes (at least 30, better 50 (Seaman et al. 1999)) was achieved for all subjects. Only fixes obtained while the mice were active and moving around outside their burrows were used in the home range calculations. The observation time started at the moment of burrow exit and the first location was recorded after ten minutes had passed. The observation time ended upon return to the burrow and the burrow location was used as the last datapoint. No locational information was recorded for the time they spent in their burrow. The same method was applied if an individual routinely used more than one burrow. The mean as well as the maximum linear distance moved per time interval, was computed to give an indication of their mobility.

The calculation of home ranges and successive distances as well as the plotting of the home range contours were performed in ArcView® 3.2 using the free software

- 169 -

The ecology of the Dusky Hopping Mouse (Notomys fuscus) extension ‘Animal Movement Analyst Version 2’ (USGS, Alaska Biological Science Centre). The illustration of the home ranges generated in ArcView® were imported into Google Earth and overlaid onto the satellite imagery for better illustration of habitat use.

6.6.2 Results Generally N. fuscus had a regular activity pattern, with emergence from the burrows at dusk within an hour after sunset, a period of activity (foraging, social contacts) of two to three hours, and then a return to their burrows. A similar period of activity was repeated in the hours before dawn, N. fuscus returned to their burrow at the latest at sunrise. Individuals could and did deviate from this general pattern under the presumed influences of the weather, their food requirements, the quality of their habitat, their reproductive status or possibly other undetermined factors.

Sufficient radio fixes for home range calculation were obtained for fourteen N. fuscus, five males and nine females (Table 6.4). Images of their home ranges superimposed onto the Google Earth satellite imagery are provided in Appendix 14. The movements of the mice were very much restricted to the sand ridges. Individuals rarely left the sandy substrate and visited the adjoining parts of the interdunes (swales). Even less often were forays further into the interdune or crossings to a neighbouring dune. At least in the short term, N. fuscus inhabit and utilize well defined sections of the dunes. Male N. fuscus had a larger mean (± S.E.) home range (ha) (males: 3.63 ± 1.27, n = 5; females: 1.16 ± 0.35, n = 9) and a larger core range (males: 0.44 ± 0.16; females: 0.16 ± 0.05) than females. These differences however were not significant due to the large variability in home ranges in male mice. Had it been possible to track all males and not only the sedentary ones, home range size for males would have certainly been larger and the differences between the sexes significant. Home range size did not increase linearly with the percentage of fixes that are included (Figure 6.9). The slope of the curve resembled more an exponential increase, especially for males. Some male mice had small home ranges similar to those of females with few long-distance movements. Other males travelled much larger distances outside their core area, so that a small percentage of additional fixes increased the 95 % home range estimate greatly. Not surprisingly therefore the mean linear distance travelled in 10 min (i.e. between fixes) as well as the maximum linear distance covered in 10 min were significantly greater in

- 170 -

The ecology of the Dusky Hopping Mouse (Notomys fuscus)

males than in females (mean distance: t11 = 2.229, p = 0.048; max distance: t11 = 2.627, p = 0.024).

The mice were very predictable in their movements and often visited the same locations in the same sequence every night. This was particularly true for the first hour or so after burrow emergence and more pronounced in the females. A great part of each mouse’s time was spent under A. ligulata shrubs feeding on their seeds.

Males and females had overlapping home ranges. At one site with a particularly high N. fuscus density, the opportunity arose to track three females in short succession. The home ranges of these three females had very little overlap but all three female home ranges had considerable overlap with the male’s home range (Figure 6.10). Information collated regarding the species’ social system (see section 6.7) revealed that all female mice were carrying young or suckling small young at the time of tracking.

Table 6.4: Subject characteristics, sampling intensity, home (Kernel, 95 %) and core ranges (Kernel, 50 %) in ha, and mean and maximum linear distance moved between fixes in metres for Notomys fuscus. Mean Maximum linear linear distance distance Age&repro- Nights No. Core Home moved in moved in Mouse ID Sex weight ductive state tracked fixes Range Range 10 min 10 min Noname M 35.0 adult 5630.342.34* * Daisy0 F 40.0 adult 6 140 0.03 0.32 13.26 80.30 Donald M 38.0 adult 5 68 0.10 0.54 21.39 98.95 Ernie M 37.1 adult 5 117 0.19 2.42 37.88 137.58 Minnie F 36.0 adult-pregnant 5 81 0.40 2.79 37.51 126.57 Bert M 30.0 adult 4 73 0.58 5.07 46.47 140.55 Daisy2 F 52.0 adult-lactating 5 80 0.05 0.46 23.39 125.00 Miffy F 33.3 adult-never young 5 90 0.31 1.84 22.45 112.42 Dennis M 36.0 adult 2 41 1.00 7.79 24.12 137.79 Bertha F 45.0 adult-pregnant 5 76 0.08 0.59 17.53 92.77 Bianca F 41.0 adult-never young 4 52 0.03 0.37 19.34 56.67 Susie F 41.0 adult-lactating 5 110 0.16 0.96 19.89 66.42 Maja F 42.0 adult-pregnant 3 38 0.04 0.27 15.71 68.30 Wilma F 43.0 adult-lactating 2 22 0.34 2.81 20.13 83.98

- 171 -

The ecology of the Dusky Hopping Mouse (Notomys fuscus) a) b)

9 9 8 8 7 7 6 6 5 5 4 4 3 3

Home [ha] size range 2

Home [ha] size range 2 1 1 0 0 25% 50% 75% 95% 25% 50% 75% 95% fixes fixes fixes fixes fixes fixes fixes fixes

Figure 6.9: Home ranges (95 % of fixes) and partial home ranges (25 %, 50 % and 75 % of fixes) of a) 9 female and b) 5 male Notomys fuscus. The various lines represent the home range estimates of each individual mouse.

Figure 6.10: Home ranges of three female and a male N. fuscus. The home ranges of the three female mice (yellow, pink and green) do not overlap but the home range of a male mouse (white outline) overlaps with the home ranges of all three females. The house symbol indicates the locations of the respective burrows.

- 172 -

The ecology of the Dusky Hopping Mouse (Notomys fuscus)

At least for the respective tracking period the mice used the same burrow to spend the day in. But most N. fuscus, males as well as females, had one or several other burrows that they either visited briefly during the night or spent considerable time in. In some cases these other burrows were inhabited by other N. fuscus but more often the respective radio-tracked individual appeared to be the only occupant.

6.6.3 Discussion The radio-tracking data collected during this study confirm the results from Moseby et al. (2006) that N. fuscus almost exclusively use the top and slopes of sand dunes during their nightly movements. Occasionally interdunal areas were visited but the mice remained relatively close to the dune and never stayed long before they returned to the dunes. The crossing of an interdune to move onto the neighbouring dune was only observed once, in a male mouse.

Notomys fuscus showed great consistency in their nightly movements, visiting the same sites each night, usually in the same succession. The sites visited defined a small but consistent core range. Occasional excursions to other parts of the dune resulted in a considerable extension of the range, resulting in a much larger home range. When discussing the core range of animals it is often stated that these are likely to be centred around and defined by food resources, but the resource itself is rarely known. For N. fuscus this resource could be identified. The core ranges, in particular those of female mice, included a cluster of A. ligulata shrubs, whose seeds represented a patchily distributed key resource (see section 6.5). Even though the sample size of mice tracked in short succession in one dune section was small (three females and one male), it appears that at least female mice have exclusive core ranges and possibly even home ranges.

The movements outside the core range were made in search of more widely distributed additional food resources such as copperburrs or other plants, presumably consumed to fulfil their water requirements. The search for mating opportunities, new sites suitable for burrow excavation or the need to identify new food resources are likely to be further motivators for the wider ranging movement. The visitation of burrows, other than their own, was possibly made in search for potential mating partners. The unoccupied burrows that were frequented could be back-up burrows in case the primary burrow was

- 173 -

The ecology of the Dusky Hopping Mouse (Notomys fuscus) entered by predators, flood water or had collapsed; or they could have been burrows in the construction phase. For many N. fuscus the location of the burrow was either not included in the 95 % kernel of its movements or at the edge of the home range, as the calculation of the home range was based on fixes recorded during periods of activity only. With the exception of an occasionally observed short period of grooming upon burrow exit, the mice usually exited and entered burrows swiftly and spent little time in the immediate vicinity of their burrows. Quite often the longest distance moved within a time interval was the movement to and from the burrow. This is quite different to results for another species which inhabits a permanent burrow, the Western Pebble Mound Mouse, where the mound was usually located in the core range (Anstee et al. 1997).

The home range size calculated from the data collected for this study is likely to be an underestimate of the actual size. Generally, home range is likely to increase with an increasing number of days that animals are tracked but logistics and equipment characteristics (such as battery life and costs) put limitations on the practical tracking period. Five nights of tracking as attempted during this study are apparently an adequate time period for N. fuscus and yielded enough data to provide reliable estimates of home range size. In the face of the consistency and predictability of the movements of female mice even fewer nights would have resulted in little deviation in home range size. The limitation of tracking to the first activity period of the night however could have lead to an under-representation of home range. After spending the entire day resting in the burrow the mice are likely to be hungry and thus the search for resources predominantly dictate their movements and the sites they visit. Any further activity periods later in the night, after food requirements had been met, could have other motivators and potentially may have resulted in a larger and possibly differently shaped home range.

Male mice had a greater mean home range area than the female mice, even though this difference was not significant. However, the tracking during this study was limited to recaptured males (proven site fidelity) which reduced the loss of radio-collars - a significant problem early in the study when any male that was captured was radio- collared. The recaptured males that were tracked represented the more sedentary part of the population and thus the home range estimate for males is likely to be an underestimate. In contrast to the recaptured males another part of the population appeared to be much more mobile, had larger, possibly shifting home ranges and/or

- 174 -

The ecology of the Dusky Hopping Mouse (Notomys fuscus) performed long-distance (dispersal/migratory) movements. The relatively low number of re-captures of males compared to females (see 6.7.2) is another indication for this. Similarly, Moseby et al. (2006) found a higher proportion of their male sample to be more transient. They collared six male mice at the site that is most similar to this study site, and two of these males could not be found after release (possibly because they moved out of range) and three of them moved maximum linear distances greater than 700 m (at which distance they would have probably not been found in this study). Possibly younger males make up the transient part of the population whereas the more mature males are more sedentary and can remain at a site for several months.

Trapping and recapture has provided some information on desert rodent movements in the past (e.g. Dickman et al. 1995), revealing the occurrence of long-range movements of several kilometres. Information in the literature on the home ranges of Australian arid-zone rodents is limited or data are unpublished (see tabulation of existing data in (Letnic 2002), Table 4). Comparisons of home and core ranges between species are of little value, as different methods are usually used to estimate range areas and because home ranges are believed to change or become non-existent under certain conditions, such as drought followed by rainfall. Suffice it to say that the home ranges of N. fuscus observed here fall well within the range of those previously observed in Australian desert rodents.

6.7 Social organisation and behaviour Hopping Mice are considered a social rodent group for which communal living is common. Notomys alexis is the most widespread and abundant Hopping Mouse species and a common domestic pet in Australia. Consequently its social organisation and behaviour is relatively well known and researched even though most observations have been made on captive animals and comparatively little is known about its behaviour in the natural environment. In comparison, apart from some observations on group size, next to nothing is known about the rarer species, including N. fuscus.

6.7.1 Methods A combination of methods provided insight into the social organisation and behaviour of N. fuscus. Trapping results and microchipping of individuals (methodology described in section 4.5.2) provided data on the density of N. fuscus in a certain dune section, the ratio of females to males and information on site fidelity. In addition to the monitoring

- 175 -

The ecology of the Dusky Hopping Mouse (Notomys fuscus) of the permanent study sites, traps were established on other dunes where N. fuscus activity was found to be high. At those sites, traps were dug in strategically at locations which promised a high capture success; i.e. close to ‘runways’ or near obvious feeding sites. Some traps were also added in strategic places on the existing trapping grids for example to determine which individual mice lived in particular burrows. The trapping effort differed between sites but the number of traps used at each site over time remained constant. The traps were opened opportunistically whenever time allowed within a period of seven weeks. All captured N. fuscus were sexed, weighed and microchipped.

Information gathered while radio-tracking was used to ascertain burrow occupancy. Since radio-tracked mice were as much as possible observed directly, their behaviour including social contacts was witnessed. Tracking of several individuals on one dune section within the timeframe of a couple of weeks (a short enough period that it is plausible that the mouse first tracked is still alive when the last mouse is tracked) made it possible to at least get a glimpse into potential territoriality and resource partitioning.

Motion-triggered still cameras (FaunaFocus FF 120, Faunatech, Australia) were installed at the active burrow entrances of all radio-tracked mice and others, and allowed the monitoring of the mice entering and leaving the burrow. Captured N. fuscus were not only marked with a microchip but also the upper coat was shaved off in small patches on their body revealing the darker undercoat. The latter marks distinguished between individual mice in the camera shots and so a minimum number of burrow occupants was determined.

6.7.2 Results

Density, sex ratio and site fidelity Over a seven week period eight dune sections, approximately one hectare in size, were intensively surveyed and as many N. fuscus trapped as possible. Logistics prevented the conduct of a regular trapping programme and thus capture-recapture analysis could not be used to estimate the abundance of the mice. The results therefore represent the minimum number of N. fuscus present over the survey period (Table 6.5).

- 176 -

The ecology of the Dusky Hopping Mouse (Notomys fuscus)

Table 6.5: Number of N. fuscus known to be alive at particular dune sections (1 ha) and the percentage of recaptures. The observation period was seven weeks. Recaptures recorded on the day following the initial capture of individual N. fuscus were excluded.

% females % males Site Total Females Males Immatures recaptured recaptured Donalds 16 5 6 5 80 33 Watties1 13 3 10 0 33 30 Watties3 18 8 10 0 63 30 Yetna1 4 1 3 0 100 0 Yetna2 8 3 5 0 67 20 Yetna3 9 1 4 4 100 50 Yetna4 13 2 6 5 50 0 Yetna5 6 3 3 0 67 0

Up to 18 adult N. fuscus were recorded on a single dune section. The ratio of females to males was at least 1:1 but in most cases males outnumbered the females with a maximum ratio of 1:4. More males than females frequented a particular dune section, but the percentage of adult mice that were recaptured (excluding recaptures on the day after initial capture) at least once during the survey, was much lower for males (20 %, CI: 4.5) than for females (70 %, CI: 5.5).

From the data on recaptured mice the minimum number of days they apparently inhabit a particular dune section was calculated to obtain some information on site fidelity. Females were found at a particular site for an average of 20 days (CI: 2.26); the maximum number of days recorded was 44. Males showed lower site fidelity with an average of 14 days (CI: 2.86) spent on a particular dune section. The longest time a male was found to visit a site was 43 days and thus was similar to the maximum of females. These results were obviously limited by the time of investigation and the actual period of site fidelity is expected to be much longer.

Social grouping Of the radio-tracked mice whose burrows were monitored with cameras, the males (n = 7) lived either alone or with up to two other mice in the same burrow. Unfortunately the sex of the co-inhabiting mice could not be determined based on the photo material. Conversely, all females (n = 6) lived either solitary or with their young if they had a litter at the time (Table 6.6). Even though this situation appeared to be common, larger social groups occurred. Remote camera footage was recorded of a group of eight mice (five obviously juveniles, their mother a second adult mouse and a - 177 -

The ecology of the Dusky Hopping Mouse (Notomys fuscus) further individual of which only the tail tip was seen on the photo) exiting the same burrow and aggregating near the entrance (Appendix 19).

Table 6.6: Observations on the number of Notomys fuscus sharing a burrow system

Mouse Sex Weight Age&repro- Burrow occupancy (g) ductive state Minnie F 36 adult/preg solitary, later with young Miffy F 33 adult/never young solitary Bertha F 45 adult/never young solitary, later with young Susie F 41 adult/lactating solitary, with young Maja F 42 adult solitary, later with young Wilma F 43 adult/lactating solitary, with young Noname M 35 adult with >=2 adults Donald M 38 adult solitary solitary but occasional visitation Ernie M 37 adult by subadult individual Bert M 30 adult with >=2 ad Elmo M 33 adult with >=2 ad Dennis M 36 adult solitary Mickey M 36 adult with one other male

Breeding Litter sizes of four to five were recorded. At a weight of around 15 g the young come to the surface for the first time. The first excursion from the burrow was a communal event. The mother left the burrow first and from above encouraged her young to exit the burrow, presumably using vocalisations, but she also repeatedly entered and exited the burrow coaxing the young to follow her. The family group congregated just outside the burrow entrance before moving off. A set of images showing an occurrence like this is provided in Appendix 18 and Appendix 19. Presumably, the young follow their mother at least in the first days after burrow emergence to get to know their environment in order to be able to orientate themselves in their mother’s home range and learn the location of resources. The frequent capture of mother and young together in one pitfall trap supports this. On the other hand communal foraging or young following their mother was never observed during radio-tracking and observation with spotlights. Thus presumably the period when the young follow their mother may be very short. Irregular time intervals separate the exit of mother and young after this first orientation phase which was undertaken together. Offspring remained in their mother’s burrow or at least

- 178 -

The ecology of the Dusky Hopping Mouse (Notomys fuscus) in her home range until they reached a weight of up to 30 g and thus maturity. In at least three cases, female mice dug a new burrow before delivering a new litter.

Aggression Female N. fuscus were very aggressive, particularly towards male mice. Where males got trapped in a pitfall together with a female then in most cases it resulted in the death of the male or at least to bites to body and head and potential internal injuries as a result of kicks with the feet. Even a subadult female, not yet reproductive, quite severely attacked a subadult male that she was trapped with. In cases where two females got trapped together, injuries were either not visible or consisted of slight bite marks on the ears. Males trapped or temporarily held together in a tank showed no overt aggressive behaviour towards each other.

Upon release after capture in pitfalls a hole was sought, with a preference for a N. fuscus burrow. Generally the mice readily entered the holes. In some instances however attempts for release into a burrow were unsuccessful and the mice would not enter it or would back out again if they had. The reason may not only be that it was not their own burrow but that they expected and attempted to avoid aggressive behaviour directed towards them, especially if the burrow occupant was a female.

6.7.3 Discussion Up to 18 adult N. fuscus were captured on an approximately 1-ha site in the course of seven weeks. Even though the mice were abundant in Sturt NP these densities were far below those observed in other areas, where ‘plagues’ of N. fuscus have been reported (Peddler 2008). Such conspicuously high numbers would have been noticeable to neighbouring residents and are unlikely to have occurred in Sturt NP in recent history. The species was thought to be extinct before its rediscovery in 2003. However, while settled near Fort Grey (part of the study area), the explorer Charles Sturt describes in his journal (Sturt 1848) how indigenous people captured not less than 150 to 200 jerboas. Even though no species is mentioned or described, this record most likely refers to N. fuscus as they were caught in the surrounding sand hills. To enable the capture of such numbers of Hopping Mice the densities must have been much higher than those experienced in Sturt NP recently, even the relatively high abundances observed during the recent irruptions (2002-2004 and 2007 to today). During the irruption observed throughout this study, the population of N. fuscus in Sturt NP alone, most definitely

- 179 -

The ecology of the Dusky Hopping Mouse (Notomys fuscus) exceeded the estimates of Moseby et al. (1999) for the minimum number of N. fuscus alive (2,000) and most likely the estimates of 10,000 for the proposed maximum population size of N. fuscus.

In a given area more individual males were captured than females. This could be a reflection of an actual skew of the sex ratio towards males, as is the case in many rodent and other small mammal populations. However, a range of other factors could have influenced the results and may explain higher male captures. Recapture rates of males were much lower than those for females and many of the males were only ever captured once. Maybe mortality in males is higher than that in females. Lower recapture rates can also be explained by the higher mobility and larger home range of males which dictates that a particular area is frequented less by a male than by a female. As the successful radio-tracking of some males over a period of five nights and long-term photo data of some males suggests, a part of the male population has set home ranges at least temporarily. However, another part of the male population appears to be transient. Many of the male radio-tracked mice moved out of range, and disappeared shortly after release or during tracking, despite extensive searching. Young males are likely to be more tr ansient a s their movement would include dispersing, whereas more mature males may settle in more permanent home ranges.

Radio-tracking of three females on one dune section (5.2 ha) within the time frame of 12 days, found that the home range of each female was centred around a cluster of A. ligulata shrubs and only slightly overlapped, if at all, with the home range of the other females. The fact that all females were either pregnant or had just given birth to a litter, may have represented unusual circumstances and potentially increased the solitary behaviour of females. The home range of the only male tracked on that dune section encompassed the home range of all three females and extended beyond these.

The findings from Sturt NP indicate that N. fuscus in the wild may be less social than suggested by observations in captivity where the mice readily form big social groups (Watts and Aslin 1981) and inferences based on the social organisation of the related and common species N. alexis in which mixed groups, including multimale and multifemale groups appear common even in the wild (Edwards et al. pers. comm. and Howe pers. comm. in Happold 1976). In N. fuscus, males were found to frequently

- 180 -

The ecology of the Dusky Hopping Mouse (Notomys fuscus) share burrows with a small number of other individuals, but females generally lived solitary or with their offspring. With a common litter size of four to five cubs, female- offspring groups fit well with the group size of five which is often found in the literature (Breed and Ford 2007; Owens et al. 2008).

The aggressiveness of females towards other individuals but particularly against males that has been observed here in wild N. fuscus is consistnet with the behaviour of captive N. alexis. For this species Happold (1976) and Breed and Ford (2007) describe that males housed together presented mild antagonistic behaviour initially but it subsided quickly and they shared a burrow. Females showed much stronger aggressive behaviour towards each other and to other males (Happold 1976), even though communal nesting was eventually also achieved. Strangers, introduced to an established group were initially repelled by group members, particularly by the females but were eventually accepted into the group. Male N alexis also demonstrate some degree of parental care (Happold 1976), and, like the females, retrieve cubs that wander from the communal nest (Breed and Ford 2007).

In the wild population of N. fuscus studied here, the aggressiveness of female mice against conspecifics in general but especially towards males makes the formation of groups consisting of adult females and males unlikely. As aggression between females was less pronounced than the antagonistic behaviour of females towards males, a likely scenario that may lead to increased group sizes around females is the persistence of previous off-spring (more likely females than males) in the maternal burrow throughout the rearing of the subsequent litter(s). This might explain the few occasions when larger group sizes, including several adult mice, were observed during this study. High resource availability may reduce aggression and allow larger numbers of individuals to coexist in an area which may result in larger communal groups of mice sharing a burrow system. This could also be an explanation for the large groups that form in captivity. Presumably food is always abundant and so aggression is low and group size can be high.

Mating system Hopping Mice as a group represent somewhat a paradox in regards to their mating system. Communal nesting and large social groups with little aggression between males

- 181 -

The ecology of the Dusky Hopping Mouse (Notomys fuscus) are common in Hopping Mouse species and such a social organisation suggests a polygamous mating system and so promiscuity of both sexes. Promiscuity favours the evolution of sperm competition and it is therefore generally correlated with relatively larger testes size in males (Kenagy and Trombulak (1986) in Wolff and Sherman 2007). Hopping Mice however have an unusual reproductive anatomy in which males have extremely small testes (all species see Breed and Taylor 2000) and low sperm production and storage (definitely in N. alexis (e.g. Breed 1990; Breed 1997; Bauer and Breed 2009); unstudied in other species). Such features generally suggest low levels of inter-male sperm competition perhaps due to a monogamous mating system or the occurrence of single male breeding units and/or low copulatory frequency (references in Breed 1990). Based on the pronounced female aggressive behaviour it has been suggested that females may prevent others from mating with her after she has been mated once (Breed 1986). Such behaviour by the female would reduce the potential for effective sperm competition and hence explain the evolution of small testes and low sperm numbers. However, testing this hypothesis Breed (1990; 1991) found that at least under laboratory conditions female N. alexis occasionally copulated with several males but as Breed and Adams (1992) determined in a subsequent study, multiple paternity never occurred. The time of ovulation in N. alexis appears to be quite short (< 133 min in Breed 1991) and for multiple paternity both inseminations must occur within this timeframe, otherwise all the eggs would have been fertilized before the second insemination takes place.

No research has yet been undertaken to investigate the mating system of N. fuscus. From the observations made and insights gained during this study of N. fuscus in the wild, speculation on the possible mating system can be made. Females inhabit small, at least temporarily permanent, resource-focussed home ranges which appear to overlap little with those of other females. Particularly productive resource patches may lead to a relatively high density of females in a fairly small area but otherwise females are widely dispersed. Males have larger home ranges or show transient behaviour which maximizes access to female mice. Together with the apparent lack of aggression between males, which rules out the defence of females or territories, these observations make polygamy or polygyny the most likely mating system for N. fuscus. Since females were very aggressive, particularly towards males, Breed’s (1986) hypothesis that female aggressive behaviour may prohibit a second male to copulate with her may be true for

- 182 -

The ecology of the Dusky Hopping Mouse (Notomys fuscus)

N. fuscus. Such behaviour may have prevented the evolution of sperm competition, leading to small testes size and a low sperm count. Instead of sperm competition males probably compete with each other through their locomotor activity. The male that is most persistent and fastest is most likely to find receptive females. Such competitive searching for females, which would support the evolution of smaller more agile males, could explain the sexual dimorphism in the species where males are about 10 % lighter than the females. It could also have lead to the evolution of different strategies in males. Radio-tracking data from this study suggested that at least a part of the male population was largely transient whereas the remainder was at least temporarily sedentary. Thus there may be a mixed mating strategy for males. Transient males are constantly on the move attempting to maximise access to females by covering an area as large as possible. Other males have a stable home range and thus access to a limited number of females but they have the opportunity to monitor the reproductive status of the resident females and use this knowledge to increase the likelihood of being the first and successful mate.

The sample size upon which the determination of the mating system is based is small and the conclusions drawn here are mainly speculative and clearly more research is needed to confirm them. Also, mating patterns like social grouping can be flexible and vary with population density, availability and patchiness of resources, and synchrony of breeding (Wolff and Sherman 2007). All of these depend on the environmental conditions, which in the arid zone are determined primarily by rainfall.

6.8 Miscellaneous observations

6.8.1 Burrow systems The burrow systems of N. fuscus are dug on the stable crests and slopes of dunes and sand hummocks. Their structure has been described previously (e.g. Watts and Aslin 1981; Van Dyck and Strahan 2008) and consists of a subterranean horizontal tunnel system with one or more nesting chambers. Access to the surface is through vertical shafts about two to three centimetres in diameter that are manoeuvred with amazing speed in either direction by bracing against the sides with the back and feet. The vertical shafts are dug from below back-filling existing tunnels in the process. The shafts terminate either at or just below the surface. Those unfinished tunnels can be rapidly opened from inside the burrow or from the surface, should the use of the other exits be impassable due to an entering predator or collapse.

- 183 -

The ecology of the Dusky Hopping Mouse (Notomys fuscus)

Burrows have at least two visible exits to the surface but can have many more. A maximum of six entrances have previously been recorded in the literature (Owens et al. 2008). In this study a maximum of 17 entrances was counted for a burrow system that extended over an area of at least 12 m2 on the surface. Usually only one, sometimes two of the entrances were actually used to enter and exit the burrow. New entrances were added over time probably when the previously used entrance shaft became too wide from frequent usage to be comfortably moved through. The number of entrances thus increases with the time a particular burrow is inhabited but may also be correlated with the number of animals inhabiting the burrow. The depth of the burrow is very variable and depends on the habitat, the location of the burrow on the dune and thus the depth/height of sandy soil. Vertical shafts of various depths, usually more than one metre but also more than two metres deep, have been measured in this study and similar information has been published for other locations (Owens et al. 2008). The extent of the burrow system can be inferred from the distance between the entrances and, where still recognisable, the distance to the initial tunnel on burrow construction. The observations made here fell within the lengths of between two and five metres as recorded elsewhere (Owens et al. 2008).

The location of burrows on the dunes does not appear to be associated with any particular features, vegetation or otherwise. Entrances can be in open areas as well as sheltered under dead wood or under shrubs. The reason for this is the way a burrow is dug. The first tunnel is excavated in a place where the sand is naturally piled up in some way; for example, at a vegetation hummock, or in places where erosion has created small sand cliffs by blowing away loose sand and exposing harder more compacted sand. As the tunnel system is extended the initial tunnel is backfilled and its location becomes indiscernible once the excavated surplus sand has been eroded down. Since the vertical entrance shafts are dug from below they end up anywhere on the surface.

Finely chewed vegetation as lining of the nesting chamber has been recorded (Owens et al. 2008). The photo material collected with remote cameras captured some images of mice, males and females, carrying single dry Poached-egg Daisy stems (Myreocephalus stuartii) into their burrows. These were usually individual incidences, except in one case where a mouse brought back one stem each morning for several consecutive days.

- 184 -

The ecology of the Dusky Hopping Mouse (Notomys fuscus)

All these observations were made in the winter months. The material carried into the burrows was sparse and appeared too little to provide lining and insulation of a nesting chamber. However, the cameras were set up at established burrows and so it is possible that the greater part of nesting material was transported into the burrow shortly after its creation and the individual stems that were carried in were added over time to improve the existing lining.

Notomys fuscus frequently utilize inactive as well as active rabbit warrens. They enter the warrens through the same holes as Rabbits but usually add their own typical Hopping Mouse entrance hole in the form of a vertical entrance shaft. Whether N. fuscus use separate parts of the warrens, create their own tunnel system extending from the existing warren system or truly co-inhabit with Rabbits (i.e. use the same tunnels and maybe even share a nesting chamber) remains a subject for further investigation. Individuals visiting a burrow other than their own sniffed the opening extensively before entering cautiously or deciding not to enter.

A number of other species were observed entering or emerging from N. fuscus burrows, most frequently House Mice. Also observed were juvenile Sand Goannas (Varanus gouldii), Sandy Inland Mice (Pseudomys hermannsburgensis), Stripe-faced and Fat- tailed Dunnarts (Sminthopsis macroura and S. crassicaudata), a Desert Mouse (Pseudomys desertor), Sand Swimmers (Eremiascincus fasciolatus), and the geckonids Nephrurus levis, Diplodactylus damaeus and Diplodactylus stenodactylus. Appendix 20 shows images captured with remote cameras of V. gouldii, M. musculus and N. fuscus emerging from the same N. fuscus burrow. All three species (probably the same individuals) were repeatedly recorded entering or emerging from this burrow over several days.

Notomys fuscus were never seen to forage together under one shrub. On the contrary, the food resource was strongly defended against conspecifics as well as against other species such as the Sandy Inland Mouse. However, House Mice appeared to be dominant over N. fuscus and could displace them from a resource.

- 185 -

The ecology of the Dusky Hopping Mouse (Notomys fuscus)

6.8.2 Native predators Likely native predators are snakes, goannas and owls. Snakes and goannas are able to enter (snakes, goannas) and dig up (goannas) burrows. Goannas were frequently seen investigating/entering/emerging from N. fuscus burrows (Appendix 20 and Appendix 21). Juvenile goannas even seemed to temporarily inhabit the burrows. If goannas can reach the nesting chambers, sleeping mice would be an easy prey. Photo evidence from cameras set up at burrows revealed that in some cases burrows were abandoned after visitation by a sand goanna. This could be either because the burrow inhabitants had fallen prey to the goanna or because they moved on to another burrow since the current one had been invaded by a potential predator and was no longer safe.

6.9 Summary Sturt NP appears to be a facultative location for the N. fuscus. At times the mice are seemingly absent from the area but can be present in high densities. In contrast to this Moseby et al. (2006) reported some locations in South Australia where N. fuscus was continuously present over an eight year period. At the time this thesis was written N. fuscus have been present in Sturt NP for three years. When they were first detected, no rainfall event > 50 mm had occurred in the area for at least two years. The reappearance of the mice was likely due to migration from South Australia where widespread rain over much of the species’ range triggered a population explosion which allowed the mice to extend their range from South Australia into Sturt NP. Significant rainfall after N. fuscus was established in the area allowed the continued persistence of N. fuscus.

Observations on the diet of N. fuscus have confirmed findings of earlier studies that the species is heavily reliant on seeds but includes a wide variety of items (leaves, berries, roots, invertebrates) in their diet. Like many other Australian desert rodents it is thus an opportunistic omnivore. In Sturt NP and at the time of the study Acacia ligulata, a prolific producer of relatively large seeds, was the most important food resource. The distribution of the shrub influenced the distribution, density and movements of N. fuscus.

Densities of up to 18 mice per ha were recorded. Generally males outnumbered females by about threefold. Notomys fuscus was not found to be as social as was expected from

- 186 -

The ecology of the Dusky Hopping Mouse (Notomys fuscus) comments about communal living in the literature and comparison with its apparently more social relative the Spinifex Hopping Mouse (N. alexis). Female mice lived either solitarily or with their offspring, whereas male mice shared burrows with other, probably unrelated mice. Since the typical family group consisted of the mother and her four to five young, the most frequently observed group size was around five. However in one case a larger group of at least seven individuals, consisting of several adults and juveniles was recorded.

Radio-tracking revealed variable home ranges between 0.27 ha and 7.79 ha, with larger ranges for males compared to females, but this difference was not significant. Male and female home ranges overlapped considerably whereas female home ranges overlapped only slightly or not at all. Movements were nearly exclusively tied to the sandy substrates of the top and slopes of the sand dunes. Most of the mice became active just after sunset and on most days followed a predictable pattern in their movement through their home range, visiting the same locations every night.

The sex ratio was skewed towards males. There appears to be a mixed mating strategy amongst these males. Low recapture rates of males, suggested a large transient part of the population, but there were also more sedentary males in larger home ranges which overlapped with several (generally) smaller, home ranges of females. The lack of aggression between males suggests a social system of either polygamy or non-defence polygyny with competitive searching of mates. Small testes size and smaller body size of males supports the latter.

6.10 Conservation and management A detailed knowledge of a species’ ecology provides a solid basis for its conservation. For desert rodents with large population fluctuations, long-term monitoring of abundance is necessary to learn the scale of natural population fluctuations and to better inform the determination of conservation status. Understanding the social structure, including parameters such as adult sex ratio, number of individuals in the breeding pool, the mating system and ecological factors such as density dependence are important in determining the viability of populations. The distribution of many species is clearly linked to particular features of the environment. Identifying where species occur and the factors affecting their patterns of distribution and abundance allow the assessment of the

- 187 -

The ecology of the Dusky Hopping Mouse (Notomys fuscus) impacts of particular land use actions or other suspected threatening factors on a species and what mitigation actions may be effective.

This study has made a significant contribution to the knowledge of the ecology of N. fuscus. With this newly gained knowledge factors that may threaten the species can be re-evaluated and emerging management issues identified.

6.10.1 Threats to the species

Pastoralism The land use within the range of N. fuscus is predominantly Cattle grazing. Notomys fuscus occurs in high densities in often severely degraded habitats (Moseby et al. 1999) suggesting that land degradation resulting from Cattle grazing is not a severe threat to the species. If A. ligulata seeds are indeed a major resource for the mice not only in Sturt NP but over much of its range, then it may provide an explanation for the persistence of N. fuscus in otherwise degraded habitat. The foliage of A. ligulata to livestock is reported as not highly palatable to stock in Western Australia (Cattle country) (Mitchell and Wilcox 1994) and was found to be toxic to Cattle in Central Australia (Chippendale and Jephcott, 1963 in (Cunningham et al. 1992). However, in western New South Wales where grazing properties run mainly Sheep, it is eaten in some areas and not touched in others (Cunningham et al. 1992). Even though adult A. ligulata may not be palatable to livestock their seedlings are likely to be and thus livestock could have a significant impact on shrub recruitment. Acacia ligulata is considered by some to have weed potential due to its ability for rapid regeneration after disturbances (Cunningham et al. 1992) but it does not appear to pose a serious problem which requires control. Thus there is little risk that the species will be removed as a ‘woody weed’ to allow more palatable species for Cattle.

Potential damage through trampling or soil compaction is likely to be of little significance, except in areas of high livestock activity such as around watering points or along travelling stock routes. The burrows of N. fuscus are dug deeply enough so that only the entrances to the burrows would collapse but not the rest of the burrow. Notomys fuscus are avid diggers and can easily dig another entrance shaft to the burrow or open one of the shafts that end just under the surface.

- 188 -

The ecology of the Dusky Hopping Mouse (Notomys fuscus)

Predation by Cats and Foxes Notomys fuscus has been listed as a species susceptible to high risk from Cat predation due to its body weight, habitat use, behaviour, mobility and fecundity (Dickman 1996b). Indeed Cat predation was the cause of death for at least one of the radio-collared mice, as the collar was found within a metre from the mouse’s burrow entrance with clear Cat foot prints right next to it. Further evidence of Cat predation includes a Hopping Mouse (which was not identified to species but was most probably N. fuscus due to the lack of other Notomys species in the area) in the stomach contents of a Cat captured in the study area (G. Madani, personal communication). Similarly, Notomys alexis represent part of the diet of Foxes in the Gibson Desert in Western Australia (Burrows et al. 2003). In the Simpson Desert N. alexis declined precipitously in 1992 (Dickman et al. 1999). The authors discuss the possibility that predation by Foxes and Cats was responsible for the sudden decline and the subsequent continued low population densities despite heavy rainfall, which unexpectedly failed to trigger a population increase. Foxes had previously been unrecorded in the area but arrived and remained after peak population numbers in the Long-haired Rat (Rattus villosissimus) following a heavy rainfall event a year prior. Cats had been resident for probably the past century but also increased in numbers following the Long-haired Rat irruption. Thus predation by Cats and Foxes, which occur throughout the range of N. fuscus, is likely to be a key threatening factor for this species as for many others. Recently Letnic et al. (2009) found that N. fuscus abundance was negatively correlated with Fox activity and thus provided correlative evidence that at least Fox predation is indeed a threat to the species.

Rabbits Rabbits are present throughout the range of N. fuscus and were abundant at sites where N. fuscus occurred in Sturt NP (this study) and in South Australia (Moseby et al. 1999; Moseby et al. 2006). Many native faunal species make use of the shelter other burrowing species provide with their burrows (Moseby et al. 2006; Read et al. 2008). In Sturt NP N. fuscus were often found commensally in Rabbit warrens. The energetic costs of burrowing for Hopping Mice are great and estimated to be the equivalent of the energy expended during the terrestrial locomotion expected to occur in several weeks (White et al. 2006). By either co-inhabiting with Rabbits or by incorporating part of the warren into their own tunnel system, N. fuscus greatly reduce the amount of excavation necessary and obtain a direct energetic benefit. Thus Rabbits or rather their warrens

- 189 -

The ecology of the Dusky Hopping Mouse (Notomys fuscus) might actually be beneficial for N. fuscus by providing readily available shelter and facilitating the excavation of burrows.

However, Rabbits may pose a threat to N. fuscus by reducing seed production and recruitment of A. ligulata (and/or other shrubs) that N. fuscus may be dependent on. Rabbits are known to feed on the foliage A. ligulata (Cunningham et al. 1992) and could also harm the plant by ringbarking and limiting recruitment by eating the seedlings. The damage to A. ligulata is likely to be low during good conditions when other food in the form of ephemerals is available but could be quite severe in drought or when food resources for Rabbits get constrained due to intraspecific competition at high rabbit densities and they incorporate the foliage and bark of shrubs into their diet.

Goats Goats are becoming an increasing problem in the rangelands of Australia. As browsers Goats are able to do considerable damage to A. ligulata, a major food resource for N. fuscus , and are likely to severely restrict recruitment of the shrub (Auld 1995).

Camels A. ligulata was found to be regularly browsed by Camels (Phillips et al. 2001). Thus like Goats, increasing Camel populations pose a potential threat to N. fuscus through the consumption of or damage of A. ligulata, a major food resource for N. fuscus.

House Mice House Mice are abundant over the range of N. fuscus and may present an underestimated threat to the species. Until the N. fuscus reappeared in the study area, House Mice were the most abundant small mammal species. Photo evidence from cameras positioned at burrow entrances found House Mice to cohabit with N. fuscus quite frequently. In several cases the arrival of a House Mouse coincided or was shortly followed by the disappearance of N. fuscus from that particular burrow. Whether this was coincidental or evidence of displacement requires further investigation.

House Mice are infamous for their omnivory and thus the potential for food competition exists. During the extensive night-time observations undertaken while radio-tracking N. fuscus, House Mice were never seen to eat A. ligulata seeds. However competition is

- 190 -

The ecology of the Dusky Hopping Mouse (Notomys fuscus) likely to be limited to dry periods and food limitation and the environmental conditions were good at the time when observations were made.

- 191 -

Chapter 7: Thesis Synthesis

7.1 Conservation management in arid and semi-arid Australia The successfully management of an area for conservation requires an understanding of the ecosystem dynamics (Sinclair and Byrom 2006) and a detailed knowledge of the ecology of the species. A number of authors have synthesised the understanding of ecological functioning in arid Australia, developed a framework for conservation and identified key research questions (Stafford Smith and Morton 1990; James et al. 1995; Soulé et al. 2004). Two key themes are recurrent: habitat change caused by livestock grazing and the threat posed by feral animals. This study addressed these key themes by investigating the outcome of two conservation strategies, the removal of livestock and the control of the Fox. Both aim to improve the ecosystem and thereby aid biodiversity conservation but their effectiveness in general and the effect on species and/or faunal groups of conservation concern is rarely assessed.

Conservation efforts are often hampered by the large knowledge gaps that remain in the ecology of many arid zone species, which hinder the assessment of their population sizes and conservation status and the development and application of appropriate conservation actions. This study has taken advantage of an irruption of N. fuscus to increase the knowledge on the ecology of this little known species, which is listed as ‘vulnerable’ nationally and as ‘endangered’ in NSW and is of particular conservation concern for the management of Sturt NP. Notomys fuscus is exemplary for a great number of species inhabiting the arid zone of Australia. Large fluctuations in its population and a range confined to remote locations in the Outback pose challenging conditions for researchers that have restricted the amount of research that has been done on the species.

- 192 -

Thesis Synthesis

7.2 Effectiveness of livestock removal The results of this study have shown that more than 35 years after stock removal no substantial differences persist in either floral or faun al variables between sites of historically high and low disturbance from pastoralism. Rain fall patterns and the habitat changes they evoke rather than grazing history explained a large part of the observed variation between sites and may overlay potential residual grazing effects. This indicates that the landscape has recovered at least to the point where it is indistinguishable from areas less affected by livestock grazing. The rediscovery of two rodent species, the Dusky Hopping Mouse (Notomys fuscus) and the Desert Mouse (Pseudomys desertor), both of which were listed as ‘presumed extinct’ in the area, as well as the high abundance of the threatened scincid Ctenotus taeniatus provide additional evidence that the landscape and the fauna it once supported is approaching its original functionality. This outcome was achieved without direct intervention to control large native herbivores which continue to exert grazing pressure. The results of indirect intervention to close AWP are ambiguous in respect to both control of native herbivores and small vertebrate communities (Croft et al. 2007).

7.3 Outcome of Fox control A natural mortality event predated the experimental removal of Foxes and significantly reduced Fox activity resulting in extremely low levels of Fox activity at the beginning of the small scale, intensive Fox baiting regime that was implemented in this study. A further reduction of Fox activity in response to the applied treatment could not be shown. However, it was evident that the treatment was successful in maintaining low Fox activity at ‘Impact’ sites compared to ‘Control’ sites, for the remainder of the study. Rabbit control appeared to have only limited effect. A short-term reduction only of Rabbit activity was achieved following RHD release and the destruction of Rabbit warrens had no apparent effect. Thus a key prey species remained to encourage Fox activity and recolonization and facilitate the competitive release of Cats following Fox removal. Cat activity did increase following the beginning of Fox control but small sample size prevented a distinction of a response to improved environmental conditions after rainfall and a response to reduced Fox activity.

The applied treatments and the effective reduction of Fox activity did not have a significant effect on the small vertebrate communities, as expressed in measures of

- 193 -

Thesis Synthesis composition, number of species or total abundance. The outcome varied for the more abundant individual species. The results indicated increased abundances following Fox control for the native Sandy Inland Mouse (Pseudomys hermannsburgensis) and the geckonid Rhynchoedura ornata. The strongest response and population increase occurred in the introduced House Mouse (Mus musculus), which had a high capture rate and thus a sample sufficient for a confident result. Thus a serious unwanted secondary response of Fox removal is the potential competition between native small vertebrates and the House Mouse. However, the total abundance and species richness of small mammals, the abundance of lizards and the abundances of most individual species were unaffected. If effects existed, they were either not detected or they were confounded and overshadowed by the effects of a significant rainfall event which coincided with the reduction in Fox activity.

7.4 Ecology and conservation of the Dusky Hopping Mouse The knowledge of the ecology and behaviour of the threatened Dusky Hopping Mouse was significantly expanded. Their temporal and spatial distribution in Sturt NP, population densities and composition, temporal and spatial activities as well as the diet were described and compared to other studies in their core geographic range. A particularly important result was the identification of their major dietary item, seeds of the Sandhill Wattle (Acacia ligulata), which proved to determine the spatial activities of individual mice as well as the species’ spatial distribution and density. This newly gained knowledge allows a re-evaluation of threats to the species and a better estimate of the effectiveness of management actions suggested in any recovery plans.

7.5 Issues in the management of small vertebrate conservation A presentation of specific management recommendations would be presumptuous considering the complexities of biodiversity conservation and the scope and limitations of this study. Instead attention is drawn to some conservation issues that are of particular relevance to small vertebrates in the arid zone. These are discussed in the context of the results of this study.

7.5.1 Environmental stochasticity b Long-term monitoring Rainfall is the principal driving force in the arid zone. Australian desert regions in particular are characterized by the extreme temporal variability of rainfall and the

- 194 -

Thesis Synthesis resources it generates. The unpredictability of rainfall and the resulting stochasticity in environmental conditions have led to the evolution of strategies to avoid and overcome the challenging conditions of low resources. Such strategies include dormancy (e.g. frogs), endurance in a certain stage of the life cycle (e.g. seeds in plants, eggs or cocoons in invertebrates) and migratory movements (birds, mammals). In species where a great percentage of the population dies during times of low resources, the potential for rapid population growth on improved environmental conditions (small mammals) manifests. Consequently populations undergo perceived (e.g. frogs and plants) or actual (small vertebrates) fluctuation in their densities. Drought can result in large areas that become unsuitable for occupation or activity for many organisms and thus many species in arid areas are typically absent or quasi-absent (dormant, periods of aestivation) or exist in low abundances.

The strong influence of rainfall and rainfall-induced vegetation changes on the temporal and spatial distribution of many species may mask the impact of disturbances or experimental treatments. Environmental variation in this study was found to be more important in explaining temporal and spatial patterns in animal communities than any potential residual effects of livestock grazing. Similarly the strong response to rainfall of many species during the period of the Fox control experiment posed challenges in the data analysis.

This variability with environmental conditions illustrates the importance of long-term monitoring that covers a range of conditions. It is essential to gain information on the full inventory of species in arid environments and to be able to separate the effects of environmental stochasticity from harmful impacts when assessing the conservation status of species and the effectiveness of conservation actions. Developing and maintaining comprehensive long-term monitoring programmes and/or encouraging research (which usually includes a monitoring component) are fundamental for successful conservation as they provide essential baseline data. Adequate monitoring requires considerable resources (money, time, staff) but can save money in the long run by improving experimental design and increasing the conclusiveness of ecological experiments.

- 195 -

Thesis Synthesis

7.5.2 Habitat connectivity b Off-reserve conservation and interstate co-operation In the face of rapid climate change the preservation or even enhancement of a specie`s potential for adaptation and for speciation is crucial (Hulme 2005) and conservation strategies need to accommodate shifts and expansion of geographic ranges so that evolutionary processes can operate (Soulé 1980; Moritz et al. 2000). The habitat loss, fragmentation and degradation now present in Australia raise significant impediments and barriers to species that may need to disperse and find new habitats or refugia and thus hinder adaptive evolutionary processes and speciation. A recent assessment of the Australian National Reserve System has found that it is neither large enough, nor is it designed to meet the challenges to biodiversity conservation (Mackay et al. 2008), even before considering the impact of rapid climate change (Mackay 2007). This latest assessment confirms existing concerns about the appropriateness of the Australian reserve system (e.g. McKenzie et al. 1989; Pressey and Nicholls 1989; Woinarski et al. 1992; Soulé et al. 2004). Reserved or protected land in the rangelands is likely to be most effective for conservation if it is surrounded by carefully managed pastoral leases (Mott and Bridgewater 1992). Both points clearly emphasize the importance of off- reserve conservation.

Land acquired for conservation in the form of National Parks is routinely destocked to allow recovery from grazing disturbance. Most of the land surrounding National Parks and other reserves types consists of private land, primarily used for Cattle and/or Sheep production in the rangelands. Such properties are run to maximise economic benefit usually with little regard for and understanding of biodiversity conservation. Therefore without a monetary incentive graziers are unlikely to change their stock management to accommodate conservation efforts. To aid conservation on private land in NSW, financial compensation and technical advice is provided to landholders if they enter a conservation agreement (legally enforceable, provides permanent protection) or a voluntary non-binding agreement under the National Parks and Wildlife Amendment Act. The success of such agreements relies on the goodwill of landowners and therefore increasing awareness of conservation issues, species at risk, and possible remedial processes, are important.

- 196 -

Thesis Synthesis

Sturt NP borders on two other states, Queensland in the north and South Australia in the west. Thus in the case of Sturt NP conservation matters are further complicated because not only is the surrounding land is private land managed for sheep and cattle grazing but about half of it is also located in another state. Thus for the management of biodiversity and individual species, collaboration with the conservation bodies and other landholders in adjoining states could be beneficial. For Sturt NP in particular, the co-operation with adjoining states would be advantageous as the outskirts of the Strzlecki dune system extend from South Australia and Queensland into NSW and cover about a third of the area of the Park. Hence Sturt NP and its habitats and floral and faunal communities are more similar to areas in the neighbouring states than to other areas within NSW and the same. Currently constitutional arrangements between the Australian Commonwealth and State Governments have seen the principal responsibility for the management of rare and threatened species residing with the states. Each state, with its respective political and social constraints has its own approach to developing priorities for undertaking conservation actions. A concerted conservation effort thus relies on the commitment and dedication of the individual manager(s) but is discouraged by the discrepancies in the conservation systems between the states.

For so called boom-and-bust species, like N. fuscus, protected (drought-) refugia have been suggested in a concept that envisions areas (presumably of high resources) in which species can persist during times of difficult environmental conditions. There is some indication that for N. fuscus this concept may apply (Moseby et al. 1999). Albeit relatively small, the current range of the N. fuscus spans across at least three states. It appears that N. fuscus exists in the form of a continuous population during good times but becomes fragmented into numerous subpopulations that persist in certain patches but become locally extinct in others during less favourable conditions. Notomys fuscus is one of the species whose conservation might be greatly improved through interstate and off-reserve conservation.

7.5.3 Difficulties in controlling Foxes and Rabbits b The Dingo as a conservation management tool An important factor which allowed the continued suppression of Fox activity during this study was the restriction of recolonization to the wider area in the form of the Dingo Barrier Fence. The Fence borders the National Park on the northern and western

- 197 -

Thesis Synthesis boundary. Even though the Barrier Fence has been erected and is maintained with the aim to keep wild dogs from the sheep grazing areas, it represents an equal barrier to the movements of other animals, including the Fox. The Fence would thus have prevented immigration and re-colonization of Foxes from the north and the west and therefore allowed the concentration of baiting efforts to a relatively small target area. Intensive Rabbit reduction attempts with a variety of methods on the other hand were only effective in the very short-term if at all and Rabbit populations remained unchanged. Even at the small scale that these control efforts were carried out; they were time- consuming and required considerable resources. Despite positive outcomes in supressing Fox activity, a similar operation at a larger scale would be unrealistic.

Recently an increasing amount of research has arisen investigating the benefit of the Dingo in nature conservation by suppressing the densities of the smaller Cats and Foxes (Johnson 2006; Glen et al. 2007; Johnson et al. 2007; Claridge and Hunt 2008). Studies investigating the actual benefit of the Dingo to particular species have been scarce. An exception in relation to small mammals is a recent study which found an increased abundance of N. fuscus with Dingo activity (Letnic et al. 2009). Thus permitting a certain population density of Dingos could be an effective conservation action that would require no major investment or continuous effort into the control of smaller predators.

In areas where Cattle are grazed, Dingoes are generally allowed to exist at reasonably low densities and are only occasionally controlled. The situation is very different in the Sheep grazing country, south and east of the Dingo Barrier Fence. Here Dingos are continuously controlled to protect Sheep from predation and are virtually eradicated. A re-introduction of the Dingo would face the vehement opposition of the Sheep graziers. Sturt NP consists of former Sheep grazing country and is surrounded by grazing properties, mainly used for Sheep production and thus utilizing Dingos as a trophic regulator is an unlikely option in the control of Foxes, Rabbits and Cats in the park’s current configuration. A proposal to excise the Park by a realignment of the Dingo Barrier Fence along its southern and eastern boundaries was rejected in the 1980s but continues to be raised by the Pastoralists Association of West Darling (pers. comm. Ingrid Witte). Since the success of single species and/or integrated traditional control methods is doubtful for areas like Sturt NP the hope of constant suppression of Foxes,

- 198 -

Thesis Synthesis

Cats and Rabbits probably lies in new to be developed technologies of species-specific viral developments or sterilization methodology.

7.5.4 Goats an emerging issue b Need for additional funds The Goat is a serious ecological problem in many areas of the world (e.g. the Mediterranean, the Galapagos Islands). The potential threat posed by Goats in Australia has long been recognized but until recently in Australia densities have been relatively low and their ecological impact of comparatively little concern. A sharp rise in the world trade in Goat meat and fast growing prices over the past decade (Dubeuf et al. 2004) has changed this. Following a rise in prices for Goat meat the status of Goats in Australia has changed recently from a pest animal, whose control was in the interest of graziers due to its competition for food with Sheep and Cattle, to a valuable rangeland resource as part of a mix of livestock. In fact, Australia is the most important (60 %) Goat meat exporting country in the world (Dubeuf et al. 2004). Problems arise from the unstable and fluctuating prices paid for Goat meat (Forsyth et al. 2009) which lead to only opportunistic exploitation of the Goat resource. When prices for Goat meat are low and mustering and transport of Goats are not profitable, they remain uncontrolled on the properties in the hope that prices increase. Existing property fencing on Sheep and Cattle grazing properties is unsuitable to contain Goats and would require expensive reinforcement (Murphy 2006) to effectively hold these animals. Considering the considerable cost of fencing required for the large size of grazing properties in the rangelands this investment is uneconomical for graziers whose primary income remains in the Sheep and/or Cattle industry. Uncontrolled populations and the lack of appropriate fencing lead to the unrestricted distribution of Goats in the rangelands.

Like many other areas, Sturt NP has seen a considerable increase in Goat numbers in recent years. Considering the higher densities and the damage to vegetation Goats are capable of as well as the flow-on effects to other components of the ecosystem (for example N. fuscus which appears to be reliant on the seeds of a particular shrub species that could suffer from high Goat densities) they now require continuous control effort whereas previously only sporadic action had been necessary. Increased personnel and additional resources need to be made available to tackle this newly arising issue so that

- 199 -

Thesis Synthesis funds are not alienated from existing feral animal management actions, including Fox, Cat and Rabbit control but can be carried out in addition to those.

- 200 -

Thesis Synthesis

7.6 Suggestions for future research

7.6.1 Effects of livestock removal: Species specific investigation in plants and invertebrates This study included an investigation of residual grazing effects on invertebrates and vegetation, but the power to assess recovery was limited due to the usage of broad categorical variables rather than taxonomic data. The study has established that no broad-scale effects persist and that habitat characteristics (e.g. plant cover) have recovered. However the plant species shaping them can be different ones. Hence, a study which investigates the persistence of fine scale effects (species level) in the invertebrate assemblage and vegetation would complete the picture of persistence of historical grazing impacts.

7.6.2 Diet of the Fox and Cat in Sturt National Park The majority of individual species remained unaffected by the reduction of Fox activity. Study limitations (short duration, low replication and small sample sizes for many species) provide one possible explanation for this, as do competitive processes and the possibility that the species may not be part of the Fox’s diet in the area. Diet studies of the Fox (and the Cat) in Australia are still relatively scarce when applied to the desert regions. Hence a study of the diet of the Fox (and the Cat) in the area as representative of a sand-dune habitat would be important in the assessment of the threat posed by those introduced predators in Sturt NP and regions with similar habitat. The identification of dietary components should be accompanied by an assessment of the resource base to obtain information on dietary preferences relative to their occurrence.

7.6.3 Movements of Foxes and Cats in arid and semi-arid areas Most studies in Australia investigating the movement patterns of Foxes and Cats have focussed on urban areas and in the case of cats on farm or domestic Cats, rather than the feral cats). Limited information is available on the movements of these important predators in the arid and semi-arid zones, a knowledge gap that is in need to be addressed. Advances in GPS technology now have made automatic tracking devices available that are suitable for tracking medium-sized animals such as the Fox and Cat in remote areas like the Australian semi-arid and arid areas. This new technology makes such studies now more easily achievable than using the traditional tracking technology. A better understanding of the ecology and movement patterns of these predators will be

- 201 -

Thesis Synthesis important in developing and correctly interpreting monitoring techniques and experimental designs (such as the one used in this study) and ultimately assist in optimising control methods.

7.6.4 The House Mouse as a conservation threat to small vertebrates Throughout this study the House Mouse (M. musculus) was the most abundant small mammal species. Similarly other studies have reported the prominence of House Mice in small vertebrate communities (e.g. Moseby et al. 2009). Furthermore, the results of the Fox removal experiment conducted as part of this study indicated an increase in the abundance of House Mice. Observations on N. fuscus found that House Mice frequently co-inhabit with N. fuscus and possibly displace this native species from its burrows. Such observations together with the broad diet of House Mice imply a potential for competition with native species, especially when resources are limited during regular droughts in arid ecosystems. Thus House Mice may pose an unrecognised threat to native small vertebrates. Comparative dietary studies, behavioural experiments to investigate the competitiveness of House Mice compared to native mice (e.g. Do House Mice displace native species from their burrows or from resource patches?) and experimental removal of House Mice would provide ways to address the issue.

7.6.5 Relationship between N. fuscus and A. ligulata and refuge areas A key result of this study has been strong evidence for a dependency of N. fuscus on A. ligulata as a food resource. Similar studies in other locations and from a wider range of different environmental conditions are needed to confirm and possibly more accurately define this relationship. For example A. ligulata might prove most important for reproducing females and their young. Long-term monitoring of N. fuscus density, reproduction, and persistence of individuals in relation to A. ligulata cover and/or seed abundance and a comparison of home range sizes between sites with different density and cover of A. ligulata may present suitable avenues for confirmation of the relationship.

As an extension of such a study, food supplementation experiments could be used to determine if N. fuscus are resource limited. If this proves to be the case then A. ligulata may be a crucial factor in the characterization of potential (drought-) refugia for the species. With this knowledge potential refuge areas for the species could be identified

- 202 -

Thesis Synthesis and protected for conservation e.g. construction of small scale enclosures around A. ligulata rich areas.

7.6.6 Mating system of N. fuscus and comparison to related species Interesting findings promise to result from further investigation into the social and mating systems of N. fuscus. In contrast to the quite detailed knowledge about rodents elsewhere, in particular the Americas, relatively little is known of the Australian species. Hopping Mice show some unusual features in their reproductive anatomy, which suggest mating systems that disagree with their communal social system. A comparative study between N. fuscus and the better studied N. alexis and/or N. mitchelli could be a worthwhile extension to understand commonalities in the social organisation of the Australian Hopping Mice. Hopping Mouse burrows can be found with relative ease and their occupants caught readily by surrounding the burrow with a fencelike structure and strategic Elliott and pitfall traps (see Appendix 22 for comments on potentially useful methodology for future work on N. fuscus). Paternity tests for litters could be used to establish multiple mating in females. Data collection should take advantage of population highs when adequate samples of subjects are readily available.

- 203 -

References

Abbott, I. (2002). Origin and spread of the Cat, Felis catus, on mainland Australia, with a discussion of the magnitude of its early impact on native fauna. Wildlife Research 29: 51-74. Aitken, P. F. (1968). Observations on Notomys fuscus (Wood Jones) (Muridae- Pseudomyinae). South Australian Naturalist 43: 37-45. Aitken, P. F. (1969). The mammals of the Flinders Ranges. In: The natural history of the Flinders Ranges. (D. W. P. Corbett, Ed. Adelaide: Libraries Board of South Australia, 1969: 255-356. Algar, D. and R. Smith (1998). Approaching Eden. Landscape 13: 28-34. Andersen, A. N., A. Fisher, B. D. Hoffmann, J. L. Read and R. Richards (2004). Use of terrestrial invertebrates for biodiversity monitoring in Australian rangelands, with particular reference to ants. Australian Journal of Ecology 29 (1): 87-92. Andersen, A. N., B. D. Hoffmann, W. J. Muller and A. D. Griffiths (2002). Using ants as bioindicators in land management: Simplifying assessment of ant community responses. Journal of Applied Ecology 39 (1): 8-17. Andersen, A. N., J. A. Ludwig, L. M. Lowe and D. C. F. Rentz (2001). Grasshopper biodiversity and bioindicators in Australian tropical savannas: Responses to disturbance in Kakadu National Park. Austral Ecology 26 (3): 213-222. Anderson, M. J. (2001). A new method for non-parametric multivariate analysis of variance. Australian Ecology 26: 32-46. Anderson, M. J. (2005). PERMANOVA: A FORTRAN computer program for permutational multivariate analysis of variance, Department of Statistics, University of Auckland: New Zealand. Anderson, M. J., R. N. Gorley and K. R. Clarke (2008). PERMANOVA+ for PRIMER: Guide to software and statistical methods, PRIMER-E Ltd. Andrew, H. and R. T. Lange (1986). Development of a new piosphere in chenopod shrubland grazed by sheep- changes to the soil surface. Australian Journal of Ecology 11: 395-409. Anstee, S. D. and D. J. Needham (1997). The use of a new field anaesthesia technique to allow the correct fitting of radio collars on the Western Pebble Mound Mouse, Pseudomys chapmani. Australian Mammalogy 20 (1): 99-101. Anstee, S. D., J. D. Roberts and E. O. O' Shea (1997). Social structure and patterns of movement of the Western Pebble-mound Mouse, Pseudomys chapmani, at Marandoo, Western Australia. Wildlife Research 24: 295-305. Armstrong, R. (2004). Baiting operations: Western Shield review - February 2003. Conservation Science Western Australia 5: 31-50. Aslin, H. J. and C. H. S. Watts (1980). Breeding of a captive colony of Notomys fuscus (Rodentia- Muridae). Australian Wildlife Research 7 (3): 379-384. Auld, T. (1995). The impact of herbivores on regeneration in four trees from arid Australia. The Rangeland Journal 17 (2): 213-227.

- 204 -

Banks, B., C. R. Dickman and A. E. Newsome (1998). Ecological costs of feral predator control: foxes and rabbits. Journal of Wildlife Management 62: 766- 772. Banks, P. B. (1999). Predation by introduced foxes on native bush rats in Australia: Do foxes take the doomed surplus? Journal of Applied Ecology 36: 1063-1071. Bauer, M. and W. G. Breed (2009). Testis mass of the Spinifex Hopping Mouse and its impact on fertility potential. Journal of Zoology 274 (4): 349-356. Baxter, P. W. J., J. L. Sabo, C. Wilcox, M. A. McCarthy and H. P. Possingham (2008). Cost-effective suppression and eradication of invasive predators. Conservation Biology 22 (1): 89-98. Beckmann, R. (1990). Demise of the desert mammals. Ecos 63: 13-17. Bell, F. C. and R. J. Stanley (1991). Climate of the north-eastern corner of New South Wales. In: Lands of the North-west corner of NSW. (P. L. Milthorpe, Ed. Sydney, Australia: New South Wales Conservation Service, 1991: 160-184. Belnap, J. and K. T. Harper (1995). Influence of cryptobiotic crusts on elemental content of tissue of two desert seed plants. Arid Soil Research & Rehabilitation 9: 107-115. Bennett (1999). Foot areas, ground reaction forces and pressures beneath the feet of kangaroos, wallabies and rat-kangaroos. Journal of Zoology 247: 365-369. Bereton, R., S. Bennett and I. Mansergh (1994). Enhanced greenhouse climate change and its potential effect on selected fauna of south-eastern Australia: A trend analysis. Biological Conservation 72 (3): 339-354. Bestelmeyer, B. T. (2006). Threshold concepts and their use in rangeland management and restoration: The good, the bad and the insidious. Restoration Ecology 14 (3): 325-329. Bestelmeyer, B. T., J. R. Brown, K. M. Havstad, R. Alexander, G. Chavez and J. E. Herrick (2003). Development and use of state-and-transition models for rangelands. Journal of Range Management 56 (2): 114-126. Beutel, T. S., G. S. Baxter and R. J. S. Beeton (2003). Geckos and grazers: A perspective on reptiles in the assessment of rangeland biodiversity. www.savanna.ntu.edu.au/downloads/. Blackwell, G. L., M. A. Potter and E. O. Minot (2001). Rodent and predator population dynamics in an eruptive system. Ecological Modelling 142 (3): 227- 245. Bock, C., H. Smith and J. Bock (1990). The effect of livestock grazing upon abundance of the lizard, Sceloporus scalaris, in southeastern Arizona. Journal of Herpetology 24 (4): 445-446. Bock, C. E., J. H. Bock, W. R. Kenney and V. M. Hawthorne (1984). Responses of birds, rodents and vegetation to livestock exclosure in a Semi Desert Grassland Site. Journal of Rangeland Management 37 (3): 239-242. Bomford, M., H. Neave and L. Conibear (1998). Lessons from Rabbit calicivirus disease. 11th Australian Vertebrate Pest Conference, Western Australia, Promaco conventions. Box, J. B., A. Duquid, R. E. Read, R. G. Kimber, A. Knapton, J. Davis and A. E. Bowland (2008). Central Australian waterbodies: The importance of permanence in a desert landscape. Journal of Arid Environments 72: 1395-1413.

- 205 -

Braysher, M. (1993). Managing vertebrate pests: Principles and strategies. Canberra, Australian Government Publishing Service. Breed, B. and F. Ford (2007). Native mice and rats. CSIRO Publishing. Breed, W. G. (1982). Morphological variation in testis and accessory sex organs of Australian rodents in the genera Pseudomys and Notomys. Journal of Reproduction & Fertility 66: 607-613. Breed, W. G. (1986). Comparative morphology and evolution in the male reproductive tract in the Australian hydromyine rodents (Muridae). Journal of Zoology, London 209: 607-629. Breed, W. G. (1990). Copulatory behaviour and coagulum formation in the female reproductive tract of the Australian hopping mouse, Notomys fuscus. Journal of Reproduction and Fertility 88: 17-24. Breed, W. G. (1991). Mating behaviour and insemination in the Spinifex Hopping Mouse (Notomys alexis). Journal of Reproduction and Fertility 93: 187-194. Breed, W. G. (1997). Interspecific variation of testis size and epididymal sperm numbers in Australasian rodents with special reference to the genus Notomys. Australian Journal of Zoology 45: 651-669. Breed, W. G. and M. Adams (1992). Breeding systems of Spinifex Hopping Mice (Notomys alexis) and Plains Rats (Pseudomys australis): A test for multiple paternity within the laboratory. Australian Journal of Zoology 40: 13-20. Breed, W. G. and J. Taylor (2000). Body mass, testis mass and sperm size in murine rodents. Journal of Mammalogy 81 (3): 758-768. Briske, D. D., S. D. Fuhlendorf and F. E. Smeins (2003). Vegetation dynamics on rangelands: A critique of the current paradigms. Journal of Applied Ecology 40 (4): 601-614. Briske, D. D., S. D. Fuhlendorf and F. E. Smeins (2005). State-and-transition models, thresholds, and rangeland health: A synthesis of ecological concepts and perspectives. Rangeland Ecology and Management 58 (1): 1-10. Briske, D. D., S. D. Fuhlendorf and F. E. Smeins (2006). A unified framework for assessment and application of ecological thresholds. Rangeland Ecology and Management 59: 225-236. Brown, J. H. and G. A. Lieberman (1973). Resource utilization and coexistence of seed-eating desert rodents in sand dune habitats. Ecology 54 (4): 788-797. Burbidge, A. A. and N. L. McKenzie (1989). Patterns in the modern decline of Western Australia's vertebrate fauna: Causes and conservation implications. Biological Conservation 50: 143-198. Burgman, M. A. and D. B. Lindenmayer (1998). Conservation Biology for the Australian Environment. Surrey Beatty & Sons PTY Limited. Burrows, N. D., D. Algar, A. D. Robinson, J. Sinagra, B. Ward and G. Liddelow (2003). Controlling introduced predators in the Gibson Desert of Western Australia. Journal of Arid Environments 55: 691-713. Busack, S. D. and R. B. Bury (1974). Some effects of off-road vehicles and sheep grazing on lizard populations in the Mojave Desert California USA. Biological Conservation 6 (3): 179-183. Bustard, H. R. (1970). Australian lizards. Sydney, Collins.

- 206 -

Calaby, J. H. (1960). Observations on the banded ant-eater Myrmecobius f. fasciatus Waterhouse (Marsupialia), with particular reference to its food habits. Proceedings of the Zoological Society (London) 135: 183-207. Catling, P. C. (1988). Similarities and contrasts in the diets of Red Foxes (Vulpes vulpes) and Cats (Felis catus) relative to fluctuating prey populations and drought. Australian Wildlife Research 15: 307-317. Catling, P. C. and R. J. Burt (1995). Why are Red Foxes absent from some eucalypt forests in eastern New South Wales? Wildlife Research 22: 535-546. Caughley, G. (1987). Introduction to the sheep rangelands. In: Kangaroos their ecology and management in the sheep rangelands of Australia. (G. Caughley, N. Shepherd and J. Short, Eds.). Cambridge: Cambridge University Press, 1987: 1- 13. Caughley, G. and A. Gunn (1996). Conservation biology in theory and practice. Cambridge, Massachusetts, Blackwell Science. Chase, J. M., M. Leibold, A. L. Downing and J. B. Shurin (2000). The effects of productivity, herbivory, and plant species turnover in grassland food webs. Ecology 81 (9): 2485-2497. Churchill, T. B. (1997). Spiders as ecological indicators: An overview from Australia. Memoirs of the Museum of Victoria 56 (2): 331-337. Claridge, A. W. and R. Hunt (2008). Evaluating the role of the Dingo as a trophic regulator: Additional practical suggestions. Ecological Management & Restoration 9: 116-119. Cogger, H. G. (1974). Thermal relations of the Mallee Dragon Amphiblorus fordi Lacertilia Agamidae. Australian Journal of Zoology 22 (3): 319-339. Coman, B. J. (1973). The diet of Red Foxes Vulpes vulpes (L.) in Victoria. Australian Journal of Zoology 21: 391-401. Coman, B. J., J. Robinson and C. Beaumont (1991). Home range, dispersal and density of Red Foxes (Vulpes vulpes L.) in Central Victoria. Wildl. Res. 18: 215- 223. Conover, W. (1980). Practical nonparametric statistics. New York, Wiley. Copley, P. B. (1991). Feral and domestic Cats in South Australia. In: The impacts of cats on native wildlife. (C. Potter, Ed. Canberra: Australian National Parks and Wildlife Service, 1991. Copley, P. B. (1992). Translocations of native vertebrates in South Australia: A review. In: Reintroduction biology of Australian and New Zealand fauna. (M. Serena, Ed. Chipping Norton, NSW, Australia: Surrey Beatty & Sons, 1992: 35-42. Courchamp, F., M. Langlais and G. Sugihara (1999). Cats protecting birds: modelling mesopredator release effect. Journal of Animal Ecology 68: 282-292. Courchamp, F., M. Langlais and G. Sugihara (2000). Rabbits killing birds: modelling the hyperpredation process. Journal of Animal Ecology 69: 154-164. Coutts-Smith, A. J., A. J. Mahon, P. S. Letnic and P. O. Downey (2007). The threat posed by pest animals to biodiversity in New South Wales. Invasive Animals Cooperative Research Centre, Canberra. Cresswell, W. J. and G. C. Smith (1992). The effects of temporally autocorrelated data on methods of home range analysis. In: Wildlife telemetry: Remote monitoring and tracking of animals. (I. G. Priede and S. M. Swift, Eds.). Chichester, England: Ellis Horwood Limited, 1992: 272-284. - 207 -

Crichton, E. G. (1974). Aspects of reproduction in the genus Notomys, Muridae. Australian Journal of Zoology 22 (4): 439-447. Croft, D. B. (2002). Estimation of the abundance of wild-living dogs in Sturt National Park. University of New South Wales, Sydney. Croft, D. B. (2009). Re-estimation of the abundance of wild-living dogs in Sturt National Park. University of New South Wales, Sydney. Croft, D. B., R. Montague-Drake and M. Dowle (2007). Biodiversity and water point closure: Is the grazing piosphere a persistent effect? In: Animals of Arid Australia: out on their own? (C. R. Dickman, D. Lunney and S. Burgin, Eds.). Mosman, NSW, Australia: Royal Zoological Society of New South Wales, 2007: 143-171. Croft, J. D. and L. J. Hone (1978). The stomach content of foxes Vulpes vulpes (L.) collected in New South Wales. Australian Wildlife Research 5: 85-92. CSIRO (2006). Rabbit Haemorrhagic disease virus suspension. Sustainable Ecosystems. Crace, ACT, Australia, Commonwealth Scientific and Industrial Research Organisation. Cunningham, G. M., W. E. Mulham, P. L. Milthorpe and J. H. Leigh (1992). Plants of western New South Wales. Melbourne, Incata Press. Davey, C., A. R. E. Sinclair, R. P. Pech, A. D. Arthur, C. J. Krebs, A. E. Newsome, D. Hik, R. Molsher and K. Allcock (2006). Do exotic vertebrates structure the biota of Australia? An experimental test in New South Wales. Ecosystems 9 (6): 992-1008. Dawson, T. J., M. J. S. Denny, E. M. Russell and B. A. Ellis (1975). Water usage and diet preferences of free ranging kangaroos, sheep and feral goats in the Australian arid zone during summer. Journal of the Zoological Society of London 177: 1-23. Dawson, T. J., K. J. McTavish, A. J. Munn and J. Holloway (2006). Water use and the thermoregulatory behaviour of kangaroos in arid regions: Insights into the colonisation of arid rangelands in Australia by the Eastern Grey Kangaroo (Macropus giganteus). Journal of Comparative Physiological Biology 176: 45- 53. de Solla, S. R., R. Bonduriansky and R. J. Brooks (1999). Eliminating autocorrelation reduces biological relevance of home range estimates. Journal of Animal Ecology 68: 221-234. de Tores, P., K. Himbeck, M. Dillon, J. Cocking, B. MacArthur and S. Rosier (1998). Large scale Fox control in the northern jarrah forest of southwest Western Australia. 11th Australasian Vertebrate Pest Conference, Agriculture Western Australia, Bunbury. Debano, S. J. (2006). Effects of livestock grazing on aboveground insect communities in semi-arid grasslands of southeastern Arizona. Biodiversity and Conservation 15: 2547-2564. Deeker, W. (1992). New Rabbit biological control strategies for the 90s in Australia. CSIRO, Canberra. Dennis, P., J. Skartveit, D. I. McCracken, R. J. Pakeman, K. Beaton, A. Kunaver and D. M. Evans (2008). The effects of livestock grazing on foliar arthropods associated with bird diet in upland grasslands of Scotland. Journal of Applied Ecology 45 (1): 279-287.

- 208 -

Department of the Environment Water Heritage and the Arts (2008a). "Approved Conservation Advice for Notomys fuscus (Dusky Hopping-Mouse)." 2009, from http://www.environment.gov.au/biodiversity/threatened/species/pubs/125- conservation-advice.pdf. Department of the Environment Water Heritage and the Arts (2008b). "EPBC Act; List of threatened fauna." from http://www.environment.gov.au/cgi- bin/sprat/public/publicthreatenedlist.pl?wanted=fauna. Dickman, C. R. (1993). The biology and management of the native rodents of the arid zone in NSW. NSW National Parks and Wildlife Service, Sydney: Report No 12. Dickman, C. R. (1994). Native mammals of western New South Wales: Past neglect, future rehabilitation? In: Future of the Fauna of western New South Wales. (D. Lunney, S. Hand, P. Reed and D. Butcher, Eds.). Mosman: The Royal Zoological Society of New South Wales, 1994: 81-92. Dickman, C. R. (1996a). Impact of exotic generalist predators on the native fauna of Australia. Wildlife Biology 2: 185-195. Dickman, C. R. (1996b). Overview of the impact of feral Cats on Australian native fauna. Canberra, Australian Nature Conservation Agency. Dickman, C. R., P. S. Mahon, P. Masters and D. F. Gibson (1999). Long-term dynamics of rodent populations in arid Australia: The influence of rainfall. Wildlife Research 26 (4): 389-403. Dickman, C. R., M. Predavec and F. J. Downey (1995). Long-range movements of small mammals in arid Australia: Implications for land management. Journal of Arid Environments 31 (4): 441-445. Dickman, C. R., R. L. Pressey, L. Lim and H. E. Parnaby (1993). Mammals of particular conservation concern in the Western Division of New South Wales. Biological Conservation 65: 219-248. Dorrough, J., A. Yen, V. Turner, S. G. Clark, J. Crosthwaite and J. R. Hirth (2004). Livestock grazing management and biodiversity conservation in Australian temperate grassy landscapes. Australian Journal of Agricultural Research 55: 279-295. Dowle, M. (2004). Effects of artificial watering point on vegetation and faunal biodiversity in the sand dunes of Sturt National Park. Honours Thesis. School of Biological, Earth and Environmental Sciences. University of New South Wales, Sydney. Downey, F. J. and C. R. Dickman (1993). Macro- and micro-habitat relationships among lizards of sandridge desert in central Australia. In: Herpetology in Australia: A diverse discipline. (D. Lunney and D. Ayers, Eds.). Sydney: Royal Zoological Society of New South Wales, 1993: 133-138. Dubeuf, J.-P., P. Moran-Fehr and R. Rubino (2004). Situation, changes and future of the goat industry around the world. Small Ruminant Research 51 (2): 165-173. Earl, J. M. and C. E. Jones (1996). Is cell grazing the answer? Rangeland Journal 18: 327-350. Edwards, G. P., A. R. Pople, K. Saalfeld and P. Caley (2004). Introduced mammals in Australian rangelands: Future threats and the role of monitoring programmes in management strategies. Austral Ecology 29: 40-50.

- 209 -

Ehrlich, P. R. (1989). Attributes of invaders and the invading processes: Vertebrates. In: Biological Invasions: a global perspective. (J. A. Drake, Ed.: John Wiley & Sons Ltd, 1989: 315-328. Eldridge, D. J. (1993). Cryptogam cover and soil surface condition: Effects on hydrology on a semi-arid woodland soil. Arid Soil Research & Rehabilitation 9: 203-217. Eldridge, D. J. and R. S. B. Greene (1994a). Assessment of sediment yield from a semi-arid red earth with varying cover of cryptogams. Journal of Arid Environments 26: 221-232. Eldridge, D. J. and R. S. B. Greene (1994b). Microbiotic soil crusts: A review of their roles in soil and ecological processes in the rangelands of Australia. Australian Journal of Soil Research 32: 389-415. Eldridge, D. J. and C. A. Myers (2001). The impact of warrens of the European Rabbit (Oryctolagus cuniculus L.) on soil and ecological processes in a semi- arid Australian woodland. Journal of Arid Environments 47 (3): 325-337. Ellis, M. (1993). Extension of the known range of the Fawn Hopping-Mouse Notomys cervinus in New South Wales. Australian Zoologist 29 (1-2): 77-78. Errington, P. L. (1946). Predation and vertebrate populations. Quarterly Review of Biology 21: 144-177. Falk, D. A. (1990). Endangered forest resources in the U.S.: Integrated strategies for conservation of rare species and genetic diversity. Forestry and ecological management 35: 91-107. Fan, M., Y. Kuang and Z. Feng (2005). Cats protecting birds revisited. Bulletin of Mathematical Biology (67): 1081-1106. Fensham, R. J. and R. J. Fairfax (2008). Water-remoteness for grazing relief in Australian arid-lands. Biological Conservation 141: 1447-1460. Fieberg, J. (2007). Utilization distribution estimation using weighted kernel density estimators. Journal of Wildlife Management 71 (5): 1169-1675. Field, A. (2000). Discovering statistics using SPSS for Windows. London, Sage Publications. Finlayson, H. H. (1927). Observations on the South Australian members of the subgenus 'Wallabia'. Transactions of the Royal Society of South Australia 51: 363-377. Finlayson, H. H. (1939). On mammals from the Lake Eyre Basin. Part 4. Transactions of the Royal Society of South Australia 63: 88-117. Fisher, A., L. Hunt, C. James, J. Landsberg, D. Phelps, A. Smyth and I. Watson (2004). Review of total grazing pressure management issues and priorities for biodiversity conservation in rangelands: A resource to aid NRM planning. Alice Springs. Fleischner, T. L. (1994). Ecological costs of livestock grazing in western North America. Conservation Biology 8 (3): 629-644. Fleming, P. J. S. and R. W. Parker (1991). Temporal decline of 1080 within meat baits used for control of wild dogs in New South Wales. Wildlife Research 18: 729-740. Foran, B. D., W. A. Low and B. W. Strong (1985). The response of Rabbit populations and vegetation to Rabbit control on a calcareous shrubby grassland in central Australia. Australian Wildlife Research 12 (2): 237-247. - 210 -

Forsyth, D. M., R. P. Duncan, M. Bomford and G. Moore (2004). Climatic suitability, life-histroy traits, introduction effort, and the establishment and spread on introduced mammals in Australia. Conservation Biology 18 (2): 557- 569. Forsyth, D. M., J. Parkes, P., A. P. Woolnough, G. Pickles, M. Collins and I. Gordon (2009). Environmental and economic factors determine the number of feral goats commercially harvested in Western Australia. Journal of Applied Ecology 46 (1): 101-109. Frank, A. (2002). Comparison of reptile and small mammal assemblages between grazed and ungrazed plots in the Australian Arid Zone. Staatsexamensarbeit. Zoologisches Institut und Museum Alexander König. Rheinische Friedrich- Wilhelms Universität, Bonn. Friedel, M. H. (1994). How spatial and temporal scale affect the perception of change in rangelands. Rangeland Journal 16 (1): 16-25. Friedel, M. H., A. D. Sparrow, J. E. Kinloch and D. J. Tongway (2003). Degradation and recovery processes in arid grazing lands of central Australia. Part 2: Vegetation. Journal of Arid Environments 55: 327-348. Friend, G. R., G. T. Smith, D. S. Mitchell and C. R. Dickman (1989). Influence of pitfall and drift fence design on capture rates of small vertebrates in semi-arid habitats of Western Australia. Australian Wildlife Research 16: 241-260. Friend, J. A. (1990). The numbat Myrmecobius fasciatus (Myrmecobiidae): History of decline and potential for recovery. Frith, H. J. (1962). Conservation of the Mallee fowl, Leipoa ocellata Gould (Megapodiidae). Wildlife Research 7: 33-49. Fuhlendorf, S. D. and D. M. Engle (2001). Restoring heterogeneity on rangelands: Ecosystem management based on evolutionary grazing patterns. Biological Science 51 (8): 625-632. Gentle, M. N., G. R. Saunders and C. R. Dickman (2007a). Persistence of sodium monofluoroacetate (1080) in Fox baits and implications for Fox management in south-eatstern Australia. Wildlife Research 34: 325-333. Gentle, M. N., G. R. Saunders and C. R. Dickman (2007b). Poisoning for production: how effective is Fox baiting in south-eastern Australia? Mammal Review 37 (3): 177-190. Gibbons, P., S. V. Griggs and J. M. Shields (2002). Are economic instruments the saviour for biodiversity on private land? Pacific Conservation Biology 7: 223- 228. Gibson, D. F., G. Lundie-Jenkins, D. G. Langford, J. R. Cole, J. E. Clarke and K. A. Johnson (1994). Predation by feral Cats, Felis catus, on the rufous hare wallaby, Lagorchestes hirsutus, in the Tanami Desert. Australian Mammalogy 17: 103-107. Glen, A. S. and C. R. Dickman (2005). Complex interactions among mammalian carnivores in Australia, and their implications for wildlife management. Biological Reviews 80: 387-401. Glen, A. S., C. R. Dickman, M. E. Soule and B. G. Mackey (2007). Evaluating the role of the Dingo as a trophic regulator in Australian ecosystems. Austral Ecology 32: 492-501.

- 211 -

Grant, W. E., E. C. Birney, N. R. French and D. M. Swift (1982). Structure and productivity of grassland small mammal communities related to grazing induced changes in vegetative cover. Journal of Mammalogy 63 (2): 248-260. Green, D. R. (1989). Rangeland restoration projects in western New South Wales. 11: 110-116. Green, R. H. (1979). Sampling design and statistical methods for environmental biologists. Chichester, Wiley. Hallett, J. G. (1982). Habitat selection and the community matrix of a desert small- mammal fauna. Ecology 63: 1400-1410. Hanley, T. A. and J. L. Page (1982). Differential effects of livestock use on habitat structure and rodent populations in Great Basin Communities California and Nevada USA. California Fish and Game 68 (3): 160-173. Happold, M. (1976). Social behavior of the conilurine rodents Muridae of Australia. Zeitschrift fur Tierpsychologie 40 (2): 113-182. Harding, E. K., D. F. Doak and J. D. Albertson (2001). Evaluating the effectiveness of predator control: The non-native Red Fox as a case study. Conservation Biology 15 (4): 1114-1122. Harrington, R., N. Owen-Smith, P. C. Viljoen, H. C. Biggs, D. R. Mason and P. J. Funston (1999). Establishing the causes of the Roan antelope decline in the Kruger National Park, South Africa. Biological Conservation 90. Haythornthwaite, A. S. (2007). Postgraduate research in the Simpson Desert: The pitfalls of a PhD. In: Animals of arid Australia: Out on their own? (C. R. Dickman, D. Lunney and S. Burgin, Eds.). Mosman, NSW, Australia: Royal Zoological Society of New South Wales, 2007: 76-81. Henry, J. D. (1986). Red Fox: the catlike canine. Washington DC., Smithonian Institute Press. Heske, E. J. and M. Campbell (1991). Effects of an 11-year livestock exclosure on rodent and ant numbers in the Chihuahuan Desert Southeastern Arizona USA. Southwestern Naturalist 36 (1): 89-93. Hilder, E. J. and B. E. Mottershead (1963). The redistribution of plant nutrients through free-grazing sheep. Australian Journal of Science 26: 88-89. Hobbs, T. J., S. R. Morton, P. Masters and K. R. Jones (1994). Influence of pit-trap design on sampling of reptiles in arid spinifex grasslands. Wildlife Research 21: 483-490. Hoffmann, B. D. and A. N. Andersen (2003). Responses of ants to disturbance in Australia, with particular reference to functional groups. Austral Ecology 28 (4): 444-464. Holden, C. and G. Mutze (2002). Impact of Rabbit haemorrhagic disease on introduced predators in the Flinders Ranges, South Australia. Wildlife Research 29: 615-626. Hone, J. (1994). Analysis of vertebrate pest control. Cambridge, Cambridge University Press. Hone, J. (1999). On rate of increase (r): patterns of variation in Australian mammals and implication for wildlife management. Journal of Applied Ecology 36: 709- 718. Hooge, P. N., W. Eichenlaub and E. Solomon (1999). The animal movement program, USGS, Alaska Biological Science Center. - 212 -

Hoser, R. T. (1991). Endangered Animals of Australia. Sydney, Australia, Pierson & Company. Hulme, P. E. (2005). Adapting to climate change: Is there scope for ecological management in the face of a global threat? Journal of Applied Ecology 42: 784- 794. Hurlbert, S. H. (1984). Pseudoreplication and the design of ecological field experiments. Ecological Monographs 54 (2): 187-211. James, C. D. (1991). Temporal variation in diets and trophic partitioning by coexisting lizards Ctenotus, Scincidae, in Central Australia. Oecologia 85 (4): 553-561. James, C. D. (2003). Response of vertebrates to fenceline contrasts in grazing intensity in semi-arid woodlands of eastern Australia. Australian Journal of Ecology 28 (2): 137-151. James, C. D., J. Landsberg and S. R. Morton (1995). Ecological functioning in arid Australia and research to assist conservation of biodiversity. Pacific Conservation Biology 2: 126-142. James, C. D., J. Landsberg and S. R. Morton (1999). Provision of watering points in the Australian arid zone: A review of effects on biota. Journal of Arid Environments 41: 87-121. Jarman, P. J. (1986). The Red Fox- an exotic large predator. In: The ecology of exotic animals and plants: Some Australian case histories. (K. R. L., Ed. Brisbane, Australia: John Wiley and Sons, 1986: 45-61. Johnson, C. (2006). Australia's mammal extinctions: A 50 000 year history. New York, Cambridge University Press. Johnson, C. N., J. L. Isaac and D. O. Fisher (2007). Rarity of a top predator triggers continent-wide collapse of mammal prey: dingoes and marsupial in Australia. Proceedings of the Royal Society of Britain 274: 341-346. Johnson, K. (1991). Feral Cats: the Northern Territory perspective. In: The impact of cats on native wildlife. (C. Potter, Ed. Canberra: Australian National Parks and Wildlife Service, 1991. Johnson, K. A., A. A. Burbidge and N. L. McKenzie (1989). Australian Macropodoidea: Status, causes of decline and future research and management. In: Kangaroos, Wallabies and Rat-Kangaroos. (G. Grigg, P. Jarman and I. Hume, Eds.). New South Wales, Australia: Surrey Beatty & Sons Pty Limited, 1989: 641-657. Jones, E. (1989). Felidae. In: Fauna of Australia, Volume 1B Mammalia. (D. W. Walton and B. J. Richardson, Eds.). Canberra: Australian Government Publishing Service, 1989: 1006-1011. Jones, K. B. (1981). Effects of grazing on lizard abundance and diversity in Western Arizona USA. Southwestern Naturalist 26 (2): 107-116. Jones, Z. F., C. E. Bock and J. H. Bock (2003). Rodent communities in a grazed and ungrazed Arizona grassland, and a model of habitat relationships among rodents in southwestern grass/shrublands. American Midland Naturalist 149 (2): 384- 394. Kapelle, M., M. M. I. Van Vuuren and P. Baas (1999). Effects of climate change on biodiversity: A review and identification of key research issues. Biodiversity and Conservation 8: 1383-1397. Kennedy, M. (1990). Australia's endangered species. Sydney, Simon & Schuster.

- 213 -

Kernohan, B. J., R. A. Gitzen and J. J. Millspaugh (2001). Analysis of animal space use and movements. In: Radio-tracking animal populations. (J. J. Millspaugh and J. M. Marzluff, Eds.). San Diego, California, USA: Academic, 2001: 125- 166. King, D. R. and S. H. Wheeler (1985). The European Rabbit in south-western Australia - I. Study sites and population dynamics. Australian Wildlife Research 12: 183-196. King, E. G. and R. J. Hobbs (2006). Identifying linkages among conceptual models of ecosystem degradation and restoration: Towards an integrative framework. Restoration Ecology 14 (3): 369-378. Kitching, R. L., A. G. Orr, L. Thalib, H. Mitchell, M. S. Hopkins and A. W. Graham (2000). Moth assemblages as indicators of environmental quality in remnants of upland Australian rain forest. Journal of Applied Ecology 37 (2): 284-297. Korn, T. J. and G. Livanos (1986). The effect of dosing technique on the 1080 content of meat baits. Australian Wildlife Research 13: 455-459. Kotler, B. P. and J. S. Brown (1988). Environmental heterogeneity and the coexistence of desert rodents. Annual Review of Ecology and Systematics 19: 281-307. Kramer, H. L., P. W. Merrell and B. J. Burren (1987). Liquid chronomatographic determination of sodium monofluoroacetate (compound 1080) in the control of dingoes. 1. Meat preparation techniques. Australian Wildlife Research 14: 65- 68. Krefft, G. (1866). On the vertebrated animals of the lower Murray and Darling, their habits, economy, and geographical distribution. Trans. Phil. Soc. N.S.W. 1862- 1865: 1-33. Kutt, A. S. and J. C. Z. Woinarski (2007). The effects of grazing and fire on vegetation and the vertebrate assemblage in a tropical savanna woodland in north-eastern Australia. Journal of Tropical Ecology 23: 95-106. Lair, H. (1987). Estimating the location of the focal center in red squirrel home ranges. Ecology 68: 1092-1101. Landsberg, J. and G. Crowley (2004). Monitoring rangeland biodiversity: Plants as indicators. Austral Ecology 29 (1): 59-77. Landsberg, J. and D. Gillieson (1996). Looking beyond the piospheres to locate biodiversity reference areas in Australia's rangelands. Rangelands in a sustainable Biosphere- Proceedings of the Fifth International Rangeland Congress, Salt Lake City, Utah, Denver, CO: Society for Range Management. Landsberg, J., C. James and S. Morton (1997). Assessing the effects of grazing on biodiversity in Australia's rangelands. Australian Biologist 10: 153-162. Landsberg, J., C. D. James, J. Maconochie, A. O. Nicholls, J. Stol and R. Tynan (2002). Scale-related effects of grazing on native plant communities in an arid rangeland region of South Australia. Journal of Applied Ecology 39 (3): 427- 444. Landsberg, J., S. Lavorel and J. Stol (1999a). Grazing response groups among understorey plants in arid rangelands. Journal of Vegetation Science 10 (5): 683- 696.

- 214 -

Landsberg, J., T. G. O'Connor and D. Freudenberger (1999b). The impacts of livestock grazing on biodiversity in natural ecosystems. In: 5th International Symposium on the Nutrition of Herbivores. (H. G. Jung and G. C. Fahey, Eds.). Savoy, Il.: American Society of Animal Science, 1999b. Lange, R. T. (1969). The piosphere: sheep track and dung patterns. Journal of Range Management 22: 396-400. Lange, R. T. and C. R. Graham (1983). Rabbits and the failure of regeneration in Australian arid zone Acacia. Australian journal of Ecology 8: 377-381. Lee, A. (2000). Measuring predator control effectiveness: Reducing numbers may not reduce predator impact. 19th Vertebrate Pest Conference, Davis, University of California. Lee, A. K. (1995). The action plan for Australian rodents. Department of Environment and Heritage. Leeper, G. W. (1970). The Australian Environment, 4th edition. Melbourne, CSIRO and Melbourne University Press. Leigh, J. H., D. H. Wood, M. D. Holgate, A. Slee and M. G. Stanger (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. Letnic, M. (2002). Long distance movements and the use of fire mosaics by small mammals in the Simpson Desert, Central Australia. Australian Mammalogy 23: 125-134. Letnic, M. (2003a). The effects of experimental patch burning and rainfall on small mammals in the Simpson Desert, Queensland. Wildlife Research 30 (6): 547- 563. Letnic, M. (2003b). The effects of experimental patch burning and rainfall on small mammals in the Simpson Desert, Queensland. Wildlife Research 30: 547-563. Letnic, M., M. D. Crowther and F. Koch (2009). Does a top-predator provide an endangered rodent with refuge from an invasive mesopredator? Animal Conservation 12 (4): 302-312. Letnic, M. and C. R. Dickman (2005). The response of small mammals to patches regenerating after fire and rainfall in the Simpson Desert, central Australia. Austral Ecology 30: 24-39. Letnic, M., C. R. Dickman, M. Tischler, B. Tamayo and C. L. Beh (2004). The response of small mammals and lizards to fire and rainfall in arid Australia. Journal of Arid Environments 59: 85-114. Lockwood, J. L. (1999). Using taxonomy to predict success among introduced avifauna: Relative importance of transport and establishment. Conservation Biology 13: 560-567. Louw, G. N. and M. K. Seely (1982). Ecology of Desert Organisms. New York, Longman Group Limited. Low, B. S. (1979). The predictability of rain and the foraging patterns of the Red Kangaroo (Megaleia rufa) in central Australia. Journal of Arid Environments 2: 61-76. Ludwig, J. A., D. J. Tongway, G. N. Bastin and C. D. James (2004). Monitoring ecological indicators of rangeland functional integrity and their relation to biodiversity at local to regional scales. Austral Ecology 29 (1): 108-120.

- 215 -

Lundie-Jenkins, G. (1992). Feral Cats in the arid zone, an integrated plan for research and management of the feral Cat, Felis catus, in Australia's arid zone. Conservation Commission for the Northern Territory. Mackay, B. G. (2006). The state of biodiversity in Australia. Biodiversity Summit 2006, The Green Institute and Lawyers of Forests, Australia. Mackay, B. G. (2007). Climate change, connectivity and biodiversity conservation. In: Protected areas: Buffering nature against climate change. A symposium on building and managing the terrestrial protected area system to best enable Australia's biodiversity to adapt to climate change. Canberra: WWF/IUCN/WCPA, 2007. Mackay, B. G., J. E. M. Watson, G. Hope and S. Gilmore (2008). Climate change, biodiversity conservation, and the role of protected areas: An Australian perspective. Biodiversity 9 (3&4): 11-18. MacPhee, R. D. E. and P. A. Marx (1997). The 40.000 - year plague: humans, hyperdisease, and first-contact extinctions. In: Natural Change and Human Impact in Madagascar. (S. Goodman and B. Patterson, Eds.). Washington DC.: Smithonian Institution Press, 1997: 169-217. Magarey, E. R. (1999). The effect of sheep grazing on lizard assemblages in the New South Wales arid zone. BSc. Hons. School of Biological Science. University of New South Wales, Sydney. Majer, J. D., S. O. Shattuck, A. N. Andersen and A. J. Beattie (2004). Australian ant research: Fabulous fauna, functional groups, pharmaceuticals, and the fatherhood. Australian Journal of Entomology 43: 235-247. Marlow, N. J. (1992). The ecology of the introduced Red Fox (Vulpes vulpes) in the arid zone. PhD Thesis. School of Biological Science. University of New South Wales, Sydney. Martin, J. T. and J. N. Eveleigh (1976). Observations on the effectiveness of warren destruction as a method of Rabbit control in a semi-arid environment. The Australian Rangeland Journal 1 (3): 232-238. Masters, P. (1993). The effects of fire-driven succession and rainfall on small mammals in spinifex grassland at Uluru National Park, Northern Territory. Wildlife Research 20: 803-813. Mayer, A. L. and M. Rietkerk (2004). The dynamic regime concept for ecosystem management and restoration. Bioscience 54 (11): 1013-1020. McIlroy, J. C., E. J. Gifford and S. M. Carpenter (1988). The effect of rainfall and blowfly larvae on the toxicity of 1080 treated meat baits used in poinsoning campaigns against wild dogs. Australian Wildlife Research 15: 473-483. McKenzie, N. L., L. Belbin, C. R. Margules and G. J. Keighery (1989). Selecting representative reserve systems in remote areas: A case study in the Nullarbor region, Australia. Biological Conservation 50: 239-261. McKenzie, N. L., A. A. Burbidge, A. Baynes, R. N. Brereton, C. R. Dickman, G. Gordon, L. A. Gibson, P. W. Menkhorst, A. C. Robinson, M. R. Williams and J. C. Z. Woinarski (2007). Analysis of factors implicated in the recent decline of Australia's mammal fauna. Journal of Biogeography 34 (4): 597-611. McNay, R. S., J. A. Morgan and F. L. Bunnell (1994). Characterizing independence of observations in movements of Columbian Black-tailed deer. Journal of Wildlife Management 58: 422-429.

- 216 -

Mitchell, A. A. and D. G. Wilcox (1994). Arid shrubland plants of Western Australia. Western Australia: Perth, University of Western Australia Press in association with the Department of Agriculture. Molsher, R. (1999). The ecology of feral Cats, Felis catus, in open forest in New South Wales: Interactions with food resources and foxes. PhD Thesis. School of Biological Sciences. University of Sydney, Sydney. Molsher, R., A. Newsome and C. R. Dickman (1999). Feeding ecology and population dynamics of the feral Cat (Felis catus) in relation to the availability of prey in central-eastern New South Wales. Wildlife Research 26: 593-607. Montague-Drake, R. (2003). Strategic management of artificial watering points for biodiversity conservation. PhD thesis. School of Biological Sciences. University of New South Wales, Sydney. Montague-Drake, R. and D. B. Croft (2004). Do kangaroos exhibit water-focused grazing patterns in arid New South Wales? A case study in Sturt National Park. Australian Mammalogy 26 (1): 87-100. Moore, P. (2005). A guide to plants of inland Australia. Sydney, New Holland Publisher (Australia) Pty Ltd. Moretto, A. S. and R. A. Distel (1999). Effects of selective defoliation on the competitive interaction between palatable and unpalatable grasses native to a temperate semi-arid grassland of Argentina. Journal of Arid Environments 42: 167-175. Moritz, C., J. L. Patton, C. J. Schneider and T. B. Smith (2000). Diversification of rainforest faunas: An integrated molecular approach. Annual Review of Ecological Systems 31: 533-563. Morris, D. (1996). Coexistence of specialist and gengeralist rodents via habitat selection. Ecology 77 (8): 2352-2364. Morton, S. R. (1978). Torpor and nest-sharing in free-living Sminthopsis crassicaudata (Marsupialia) and Mus musculus (Rodentia). Journal Of Mammalogy 59 (3): 569-575. Morton, S. R. (1990). The impact of European settlement on the vertebrate animals of arid Australia: A conceptual model. Proceedings of The Ecological Society of Australia 16: 201-213. Morton, S. R. (1992). Sustainable land management in arid Australia. Search 23: 6-68. Moseby, K. E., B. M. Hill and J. L. Read (2009). Arid Recovery- A comparison of reptile and small mammal populations inside and outside a large Rabbit, Cat and Fox-proof exclosure in arid South Australia. Austral Ecology 34 (2): 156-169. Moseby, K. E., H. Owens, R. Brandle, J. K. Bice and J. Gates (2006). Variation in population dynamics and movement patterns between two geographically isolated populations of the Dusky Hopping Mouse (Notomys fuscus). Wildlife Research 33: 223-232. Moseby, K. E. and J. L. Read (1999). Population dynamics and movement patterns of the Bolam's Mouse, Pseudomys bolami, at Roxby Downs, South Australia. Australian Mammalogy 20: 353-368. Moseby, K. K., R. Brandle and M. Adams (1999). Distribution, habitat and conservation status of the rare Dusky Hopping-Mouse, Notomys fuscus (Rodentia: Muridae). Wildlife Research 26: 479-494.

- 217 -

Moss, G. L. and D. B. Croft (1999). Body condition of the Red Kangaroo (Macropus rufus) in arid Australia: The effect of environmental condition, sex and reproduction. Australian Journal Of Ecology 24 (2): 97-109. Mott, J. J. and P. B. Bridgewater (1992). Biodiversity conservation and ecologically sustainable development. Search 23: 284-287. Murphy, S. (2006). "Goats meat exports growing steadily." 2009, from http://www.abc.net.au/landline/content/2006/s1729765.htm. Murray, B. R., C. R. Dickman, C. H. S. Watts and S. R. Morton (1999). The dietary ecology of Australian desert rodents. Wildlife Research 26 (4): 421-437. Myers, K., I. Parer, D. Wood and B. Cooke (1994). The Rabbit in Australia. In: The European Rabbit. (H. V. Thompson and C. M. King, Eds.). Oxford: Oxford University Press, 1994: 108-157. Newman, J. C. and R. W. Condon (1969). Landuse and present condition. In: Arid lands of Australia. (R. O. Slayter and R. A. Perry, Eds.). Canberra: Australian University Press, 1969: 105-132. Newsome, A. E. (1989). Large vertebrate pests. In: Mediterranean landscapes in Australia. (J. C. Noble and R. A. Bradstock, Eds.). Melbourne: CSIRO, 1989: 406-415. Newsome, A. E. (1990). The control of vertebrate pests by vertebrate predators. Trends in Ecology & Evolution 5 (6): 187-191. Newsome, A. E. (1991). Feral Cats: An overview. In: The impacts of cats on native wildlife. (C. Potter, Ed. Canberra: Report to the Australian Nature Conservation Agency, 1991. Newsome, A. E. (1993). Biological interactions. In: Australian rabbit control conference, Adelaide, April 1993. (B. D. Cooke, Ed. Adelaide: Anti-rabbit Research Foundation orf Australia, 1993: 23-25. Newsome, A. E., I. Parer and P. C. Catling (1989). Prolonged prey suppression by carnivores - predator removal experiments. Oecologia 78: 458-467. Newsome, A. E., R. P. Pech, R. Smyth, C. R. Dickman and P. Banks (1997). Potential impacts on Australian native fauna of releasing Rabbit calicivirus disease. Environment Australia, Canberra. Noble, J. C. and D. J. Tongway (1986). Pastoral settlement in arid and semi-arid rangelands. In: Australian Soils: The human impact. (J. S. Russell and R. F. Isbell, Eds.). Brisbane: University of Queensland Press, 1986: 217-242. NSW National Parks and Wildlife Service (1996). Sturt National Park: Plan of Management. Department of the Environment and Conservation, NSW. NSW National Parks and Wildlife Service (2001). Threat Abatement Plan for predation by the Red Fox (Vulpes vulpes). NSW National Parks and Wildlife Service, Hurstville. Olsson, M., E. Wapstra, G. Swan, E. Snaith, R. Clarke and T. Madsen (2005). Effects of long-term Fox baiting on species composition and abundance in an Australian lizard community. Austral Ecology 30 (8): 907-913. Otis, D., L. and G. C. White (1999). Autocorrelation of location estimates and the analysis of radiotracking data. Journal of Wildlife Management 63 (3): 1039- 1044. Owen-Smith, R. N. (1996). Ecological guidelines for waterpoints in extensive protected areas. South African Journal of Wildlife Research 26 (107-112). - 218 -

Owens, H. M., K. M. Moseby and R. Brandle (2008). Dusky Hopping-mouse, Notomys fuscus. In: The mammals of Australia. (S. Van Dyck and R. Strahan, Eds.). Sydney: Reed New Holland Publishers, 2008. Paltridge, R. (2002). The diets of cats, foxes and dingoes in relation to prey availability in the Tanami Desert, Northern Territory. Wildlife Research 29: 389-403. Pech, R. P. and G. M. Hood (1998). Foxes, rabbits, alternative prey and Rabbit calicivirus disease: Consequences of a new biological control agent for an outbreaking species in Australia. Journal of Applied Ecology 35: 434-453. Pech, R. P., A. R. E. Sinclair and A. E. Newsome (1995). Predation models for primary and secondary prey species. Wildlife Research 22: 55-64. Pech, R. P., A. R. E. Sinclair, A. E. Newsome and P. C. Catling (1992). Limits to predator regulation of rabbits in Australia: Evidence from predator-removal experiments. Oecologia 89: 102-112. Peddler, R. (2008). Dusky Hopping Mouse plague. Across the Outback. Produced for Outback South Australia. Number 38. Government South Australia: 4. Phillips, A., J. Heucke, B. Dörges and G. O'Reilly (2001). Co-grazing cattle and camels. Rural industries and research and development corporation. Predavec, M. (1994). Population dynamics and environmental changes during natural irruptions of Australian desert rodents. Wildlife Research 21 (5): 569-582. Pressey, R. L. and A. O. Nicholls (1989). Application of a numerical algorithm to the selection of reserves in semi-arid New South Wales. Biological Conservation 50: 263-278. Priddel, D. and R. Wheeler (1997). Efficacy of Fox control in reducing the mortality of released captive-reared Malleefowl, Leiopoa ocellata. Wildlife Research 24: 469-482. Pringle, H. J. R. and J. Landsberg (2004). Predicting the distribution of livestock grazing pressure in rangelands. Austral Ecology 29 (1): 31-39. Putman, R. J., P. J. Edwards, J. C. E. Mann, R. C. How and S. D. Hill (1989). Vegetational and faunal changes in an area of heavily grazed woodland following relief of grazing. Biological Conservation 47 (1): 13-32. Quinn, G. P. and M. J. Keough (2002). Experimental design and data analysis for biologists. Cambridge, Cambridge University Press. Read, D. G. (1984). Movements and home ranges of three sympatric dasyurids, Sminthopsis crassicaudata, Planigale gilesi and P. tenuirostris (Marsupialia), in semiarid western New South Wales. Australian Wildlife Research 11 (2): 223- 234. Read, D. G. (1985). Notes on capture techniques for small mammals of the arid zone. Australian Zoologist 21 (6-7): 545-550. Read, D. G. (1988). Weather and trap response of the dasyurid marsupials Sminthopsis crassicaudata, Planigale gilesi and P. tenuirostris. Australian Wildlife Research 15 (2): 139-148. Read, J. and Z. Bowen (2001). Population dynamics, diet and aspects of the biology of feral cats and foxes in arid South Australia. Wildlife Research 28: 195-203. Read, J. L. (1999). Longevity, reproductive effort and movements of three sympatric Australian arid-zone geckos. Australian Journal of Zoology 47: 307-316.

- 219 -

Read, J. L. (2002). Experimental trial of Australian arid zone reptiles as early warning indicators of overgrazing by cattle. Austral Ecology 27 (1): 55-66. Read, J. L., J. Carter, K. E. Moseby and A. Greenville (2008). Ecological roles of Rabbit, bettong and bilby warrens in arid Australia. Journal of Arid Environments 72: 2124-2130. Read, J. L. and K. E. Moseby (2001). Factors affecting pitfall capture rates of small ground vertebrates in arid South Australia. I. The influence of weather and moon phase on capture rates of reptiles. Wildlife Research 28 (1): 53-60. Read, J. L. and R. Pickering (1999). Ecological and toxicological effects of exposure to an acidic, radioactive tailings storage. Environmental monitoring and assessment 54: 69-85. Recher, H. F. and L. Lim (1990). A review of current ideas of the extinction, conservation and management of Australia's terrestrial vertebrate fauna. Proceedings of the Ecological Society of Australia 16: 287-301. Reddiex, B. and D. M. Forsyth (2004). Review of existing Red Fox, feral Cat, feral Rabbit, feral pig and feral goat control in Australia. 2. Information gaps. Arthur Rylah Institute for Environmental Research, Department of Sustainability and Environment, Melbourne. Reddiex, B., D. M. Forsyth, E. McDonalld-Madden, L. D. Einoder, P. A. Griffioen, R. R. Chick and A. J. Robley (2004). Review of existing Red Fox, wild dog, feral Cat, feral Rabbit, feral pig and feral goat control in Australia. 1. Audit. Arthur Rylan Institute for Environmental Research, Department of Sustainability and Environment, Melbourne. Risbey, D. A., M. C. Calver and J. Short (1999). The impact of Cats and foxes on small vertebrate fauna of Heirisson Prong, Western Australia. 1. Exploring potential impact using diet analysis. Wildlife Research 26: 621-630. Risbey, D. A., M. C. Calver, J. Short, J. S. Bradley and I. W. Wright (2000). The impact of Cats and foxes on the small vertebrate fauna of Heirisson Prong, Western Australia. 2. A field experiment. Wildlife Research 27: 223-235. Robertson, G. (1987). Effect of drought and high summer rainfall on biomass and composition of grazed pastures in western New South Wales. Australian Rangelands Journal 9 (2): 79-85. Robley, A. J., B. Reddiex, T. Arthur, R. Pech and D. Forsyth (2004). Interactions between feral Cats, Foxes, native carnivores, and Rabbits in Australia. Arthur Rylan Institute for Environmental Research, Department of Sustainability and Environment, Melbourne. Rolls, E. C. (1969). And they all ran wild. Sydney, Australia, Angus and Robertson. Rooney, S. M., A. Wolfe and T. J. Hayden (1998). Autocorrelated data in telemetry studies: time to indepndence and the problem of behavioural effects. Mammal Review 28 (2): 89-98. Rushton, S. P., M. D. F. Shirley, D. W. MacDonald and J. C. Reynolds (2006). Effects of culling Fox populations at the landscape scale: A spatially explicit population modelling approach. Journal of Wildlife Management 70 (4): 1102- 1110. Sadlier, R. A. and R. L. Pressey (1994). Reptiles and amphibians of particular conservation concern in the Western Division of New South Wales: A preliminary review. Biological Conservation 69: 41-54.

- 220 -

Saunders, G. (2000). Degradation of sodium monofluoroacetate (1080) in buried fox baits. Wildlife Research 27: 129-135. Saunders, G., D. Choquenot, J. McIlroy and R. Packwood (1998). Initial effects of Rabbit haemorrhagic disease on free-living Rabbit (Oryctolagus cuniculus) populations in central-western New South Wales. Wildlife Research 26: 69-74. Saunders, G., B. Corman, J. E. Kinnear and M. Braysher (1994). Managing vertebrate pests: Foxes. Canberra, Australian Government Publishing Service. Saunders, G., C. Lane, S. Harris and C. R. Dickman (2006). Foxes in Tasmania: A report on the incursion of an invasive species. Invasive Animals Cooperative Research Centre. Savolainen, P., T. Leitner, A. N. Wilton, E. Matisoo-Smith and J. Lundeberg (2004). A detailed picture of the origin of the Australian Dingo, obtained from the study of mitochondrial DNA. Proceedings of the Naitonal Academy of Sciences in the United States of America. Scheffer, M., S. Carpenter, J. A. Foley, C. Folke and B. Walker (2001). Catastrophic shifts in ecosystems. Nature 413: 591-596. Scholze, S. (2007). The residual effect of water-focussed grazing on abundance and diversity: The small mammals, reptiles and selected invertebrates in the dunefields of Sturt National Park. Diploma Thesis. Mathematisch- Naturwissenschaftliche Fakultät. University of Potsdam, Potsdam. Seaman, D. E., J. J. Millspaugh, B. J. Kernohan, G. C. Brundige, K. J. Raedecke and R. A. Gitzen (1999). Effects of sample size on Kernel Home Range estimates. Journal of Wildlife Management 63 (2): 739-747. Seaman, D. E. and R. A. Powell (1996). An evaluation of the accuracy of Kernel density estimators for home range analysis. Ecology 77 (7): 2075-2085. Sharp, T. and S. Glen (2005). Humane Pest Animal Control: Codes of practice and standard operating procedures. New South Wales Department of Primary Industries. Shine, R. (1985). The reproductive biology of australian reptiles: a search for general patterns. In: Biology of australasion forgs and reptiles. (G. Grigg, R. Shine and H. Ehmann, Eds.): Royal Zoological Society of New South Wales, 1985: 297- 303. Short, J., S. D. Bradshaw, J. Giles, R. I. T. Prince and G. R. Wilson (1992). Reintroduction of macropods (Marsupialia: Macropodoidea) in Australia - a review. Biological Conservation 62 (3): 189-204. Short, J., J. E. Kinnear and A. Robley (2002). Surplus killing by introduced predators in Australia - evidence for ineffective anti-predator adaptations in native prey species. Biological Conservation 103: 283-301. Short, J. and A. Smith (1994). Mammal decline and recovery in Australia. Journal of Mammalogy 75 (2): 288-297. Short, J., B. Turner, S. Parker and J. Twiss (1995). Reintroduction of endangered mammals to mainland Shark Bay: A progress report. Surrey Beatty & Sons, Chipping Norton. Sinclair, A. R. E. and A. E. Byrom (2006). Understanding ecosystem dynamics for conservation of biota. Journal of Animal Ecology 75: 64-79. Sinclair, A. R. E. and R. P. Pech (1996). Density dependence, stochasticity, compensation and predator regulation. Oikos 75: 164-173.

- 221 -

Sinclair, A. R. E., R. P. Pech, C. R. Dickman, D. Hik, P. Mahon and A. E. Newsome (1998). Predicting effects of predation on conservation of endangered prey. Conservation Biology 12 (3): 564-575. Smith, A. P. and D. G. Quin (1996). Patterns and causes of extinction and decline in Australian conilurine rodents. Biological Conservation 77: 243-267. Smith, G. T., G. W. Arnold, S. Sarre, M. Abensperg-Traun and D. E. Steven (1996). The effect of habitat fragmentation and livestock grazing on animal communities in remnants of gimlet (Eucalyptus salubris) woodland in the Western Australian wheatbelt. II. Lizards. Journal of Applied Ecology 33 (6): 1302-1310. Soulé, M. E. (1980). Thresholds for survival: Criteria for maintenance of fitness and evolutionary potential. In: Conservation Biology: An Evolutionary-Ecological Perspective. (M. E. Soulé and B. M. Wilcox, Eds.). Sunderland: Sinauer Associates, 1980. Soulé, M. E., B. G. Mackey, H. F. Recher, J. E. Williams, J. C. Z. Woinarski, D. Driscoll, W. C. Dennison and M. E. Jones (2004). The role of connectivity in Australian conservation. Pacific Conservation Biology 10: 266-279. Southgate, R. and P. Masters (1996). Fluctuations of rodent populations in response to rainfall and fire in a Central Australian hummock grassland dominated by Plectrachne schinzii. Wildlife Research 23: 289-303. Sparrow, A. D., M. H. Friedel and D. J. Tongway (2003). Degradation and recovery processes in arid grazing lands of central Australia Part 3: Implications at landscape scale. Journal of Arid Environments 55: 349-360. Stafford Smith, D. M. and S. R. Morton (1990). A framework for the ecology of arid Australia. Journal of Arid Environments 18: 255-278. Strahan, R., Ed. (2002). The Mammals of Australia. Sydney, Australian Museum-Reed New Holland. Stringham, T. K., W. C. Krueger and P. L. Shaver (2003). State and transition modelling: An ecological process approach. Journal of Range Management 56 (2): 106-113. Sturt, C. (1848). Narrative of an expedition into central Australia, performed under the authority of her majesty´s government, during the years 1844, 5 and 6. Together with a notice of the province of South Australia in 1847. London: T. and W. Boone, 1848. Swihart, R. K. and N. A. Slade (1995). Influence of sampling interval on estimates of home-range size. Journal of Wildlife Management 49: 1019-1025. Swihart, R. K. and N. A. Slade (1997). On testing for independence of animal movements. Journal of Agricultural, Biological and Environmental Statistics 2: 48-63. Taylor, R. J. and N. Doran (2001). Use of terrestrial invertebrates as indicators of the ecological sustainability of forest management under the Montreal Process. SO - Journal of Insect Conservation. 5(4). December, 2001. 221-231. Thomas, C. D., A. Cameron, R. E. Green, M. Bakkenes, L. J. Beaumont, Y. C. Collingham, B. F. N. Erasmus, M. F. de Siqueira, A. Grainger, L. Hannah, L. Hughes, B. Huntley, A. S. van Jaarsveld, G. F. Midgley, L. Miles, M. A. Ortega-Huerta, A. Townsend Peterson, O. L. Phillips and S. E. Williams (2004a). Extinction risk from climate change. Nature 427: 145-148.

- 222 -

Thomas, C. D., S. E. Williams, A. Cameron, R. E. Green, M. Bakkenes, L. J. Beaumont, Y. C. Collingham, B. F. N. Erasmus, M. F. de Siqueira, A. Grainger, L. Hannah, L. Hughes, B. Huntley, A. S. van Jaarsveld, G. F. Midgley, L. Miles, M. A. Ortega-Huerta, A. T. Peterson and O. L. Phillips (2004b). Biodiversity conservation: Uncertainty in predictions of extinction risk/Effects of changes in climate and land use/Climate change and extinction risk (reply). Nature 430. Thompson, P. C. and R. Shepherd (1995). Return to Eden. Landscope 10: 21-25. Thomson, P. C., N. J. Marlow, K. Rose and N. E. Kok (2000). The effectiveness of a large-scale baiting campaign and an evaluation of a buffer zone strategy for Fox control. Wildlife Research 27: 465-472. Thrash, I. (1997). Infiltration rate of soil around drinking troughs in the Kruger National Park, South Africa. Journal of Arid Environments 35 (4): 617-625. Thrash, I. and J. F. Derry (1999). The nature and modelling of piospheres: A review. Koedoe 42 (2): 73-94. Tolsma, D. J., W. H. O. Ernst and R. A. Verwey (1987). Nutrients in soil and vegetation around two artificial waterpoints in eastern Botswana. Journal of Applied Ecology 24: 991-1000. Tongway, D. J., A. D. Sparrow and M. H. Friedel (2003). Degradation and recovery processes in arid grazing lands of central Australia. Part 1: Soil and land resources. Journal of Arid Environments (55): 301-326. Trigger, D., J. Mulcock, A. Gaynor and Y. Toussaint (2008). Ecological restoration, cultural preferences and the negotiation of 'nativeness' in Australia. Geoforum 39: 1273-1283. Turner, M. D. (1998a). Long-term effects of daily grazing orbits on nutrient availability in Sahelian West Africa: 1. Gradients in the chemical composition of rangeland soils and vegetation. Journal of Biogeography 25: 669-682. Turner, M. D. (1998b). Long-term effects of daily grazing orbits on nutrient availability in Sahelian West Africa: 2. Effects of a phosphorous gradient on spatial patterns of annual grassland production. Journal of Biogeography 25: 683-694. Turner, M. D. (1999). Spatial and temporal scaling of grazing impact on the species composition and productivity of Sahelian annual grasslands. Journal of Arid Environments 41: 277-297. Underwood, A. J. (1991). Beyond BACI: experimental designs for detecting human environmental impacts on temporal variations in natural populations. Australian Journal of Marine and Freshwater Research 42: 569-587. Valone, T. J., M. Meyer, J. H. Brown and R. M. Chew (2002). Timescale of perennial grass recovery in desertified arid grasslands following livestock removal. Conservation Biology 16 (4): 995-1002. Valone, T. J. and P. Sauter (2005). Effects of long-term cattle exclosure on vegetation and rodents at a desertified arid grassland site. Journal of Arid Environments 61 (1): 161-170. Van Dyck, S. and R. Strahan, Eds. (2008). The mammals of Australia. Sydney, Reed New Holland.

- 223 -

van Rensburg, P. J. and M. N. Bester (1988). The effect of Cat Felis catus predation on three breeding Procellariidae species on Marion Island. South African Journal of Applied Ecology 24: 36-73. Vetter, S. (2005). Rangelands at equilibrium and non-equilibrium: recent developments in the debate. Journal of Arid Environments 62 (2): 321-341. Walker, B. (1998). The art and science of wildlife management. Wildlife Research 25: 1-9. Wallach, A. D., B. R. Murray and A. J. O`Neill (2009). Can threatened species survive where the top predator is absent? Biological Conservation 142: 43-52. Wanless, R. M., A. Angel, R. J. Cuthbert, G. M. Hilton and P. G. Ryan (2007). Can predation by invasive mice drive seabird extinctions? Biology Letters 2 (3): 241- 244. Wasson, R. J. and R. W. Galloway (1986). Sediment yield in the Barrier Range before and after European settlement. Rangeland Journal 8: 79-90. Watts, C. H. S. (1970). The foods eaten by some Australian rodents, Muridae. South Australian Naturalist 44: 71-74. Watts, C. H. S. and H. J. Aslin (1981). The rodents of Australia. Sydney Angus & Robertson. Waudby, H. P. and T. How (2008). An additional record of the Dusky Hopping Mouse, Notomys fuscus, in South Australia. Australian Mammalogy 30 (1): 47- 50. White, C. R., P. G. D. Matthews and R. S. Seymour (2006). Balancing the competing requirements of saltatorial and fossorial specialisation: Burrowing costs in the Spinifex Hopping Mouse, Notomys alexis. Whitehead, P., J. C. Z. Woinarski, A. Fisher, R. Fensham and K. Beggs (2001). Developing an analytical framework for monitoring biodiversity in Australia´s Rangelands. Tropical Savannas Cooperative Research Centre (CRC) for the National Land and Water Resources Audit, Darwin. Whitford, W. G. (2002). The Ecology of Desert Ecosystems. New York, Academic Press. Williams, K., I. Parer, B. Coman, J. Burley and M. Braysher (1995). Managing vertebrate pests: Rabbits. Canberra, Australian Government Publishing Service. Wilson, A. D. and W. Mulham (1980). Vegetation changes and animal productivity under sheep and goat grazing on an arid Belah (Casuarina cristata) - Rosewood (Heterodendrum oleifolium) Woodland in western New South Wales. The Australian Rangelands Journal 2 (2): 183-188. Wilson, G. A., N. Dexter, P. O'Brien and M. Bomford (1992). Pest animals in Australia. A survey of introduced wild animals. Canberra, Bureau of Rural Resources, Department of Primary Industries and Energy. Wilson, S. and G. Swan (2008). A complete guide to reptiles of Australia. Sydney, New Holland Publishers (Australia) Pty Ltd. Witte, I. (2002). Spatio-temporal Interactions of mammalian herbivores in the arid zone. PhD thesis. School of Biological Science. University of New South Wales, Australia, Sydney.

- 224 -

Woinarski, J. C. Z. (1999). Prognosis and framework for the conservation of biodiversity in rangelands: Building on the north Australian experience. International Rangeland Congress, Aitkenvale, Australia. Woinarski, J. C. Z. (2001). A manual for biodiversity monitoring. Tropical Savannas Cooperative Research Centre (CRC) for the National Land and Water Resources Audit. Woinarski, J. C. Z. and A. J. Ash (2002). Responses of vertebrates to pastoralism, military land use and landscape position in an Australian tropical savanna. Austral Ecology 27 (3): 311-323. Woinarski, J. C. Z. and R. W. Braithwaite (1990). Conservation foci for Australian birds and mammals. Search 21 (65-68). Woinarski, J. C. Z., P. J. Whitehead, D. M. Bowman and J. Russell-Smith (1992). Conservation of mobile species in a variable environment: the problem of reserve design in the Northern Territory, Australia. Global Ecological Biogeographical Letters 2: 108-118. Wolff, J. and P. W. Sherman (2007). Rodent societies: An ecological and evolutionary perspective. University of Chicago Press. Wong, D. H., W. E. Kirkpatrick, D. R. King and J. E. Kinnear (1992). Defluorination of sodium monofluoroacetate (1080) by microorganisms isolated from Western Australian soils. Soil Biology and Biochemistry 24: 833-838. Wong, D. H., W. E. Kirkpatrick, J. E. Kinnear and D. R. King (1991). Defluorination of sodium monofluoroacetate (1080) by microorganisms found in bait materials. Wildlife Research 18: 539-545. Wood Jones, F. (1923-1925). The Mammals of South Australia. Adelaide, A. B. James.

- 225 -

Appendices Appendix 1: Historical map of study area. Location of study sites (black triangles) in relation to station fences (--.--.) as at 1971. Map not to scale.

- 226 -

Appendix 2: List of plant species identified in the study area. Note: Sampling was restricted to the sandy substrate of the dunes.

Family Scientific name Common name AIZOACEAE Aizoon quadifidum Sturt’s Pigface AIZOACEAE Mollugo cerviana Wire-stemmed Chickweed AIZOACEAE Tetragonia tetragonioides New Zealand Spinach AIZOACEAE Trianthema triquetra Small Hogweed AMARANTHACEAE Ptilotus latifolius White Foxtail/Mulla Mulla AMARANTHACEAE Ptilotus obovatus Crimson Foxtail AMARANTHACEAE Ptilotus polystachyus Longtails AMARANTHACEAE Amaranthus grandiflorus Large-flowered Amaranth APIACEAE Trachymene glaucifolia Wild Parsnip ASTERACEAE Angianthus pusillus Dwarf Cup Flower ASTERACEAE Calotis cymbacantha Showy Burr Daisy ASTERACEAE Calotis hispidula Bogan Flea ASTERACEAE Gnephosis arachnoidea -- ASTERACEAE Gnephosis eriocarpa Native Chamomile ASTERACEAE Myriocephalus stuartii Poach egg Daisy ASTERACEAE Podolepis capillaris Invisible Plant ASTERACEAE Rhodanthe floribunda Common White Sunray ASTERACEAE Rhodanthe microglossa Clustered Sunray ASTERACEAE Rhodanthe moschata Musk Sunray ASTERACEAE Senecio gregorii Fleshy Groundsel BORAGINACEAE Heliotropium paniculatum Bushy Heliotrope BORAGINACEAE Omphalolappula concava Burr-Stickseed BORAGINACEAE Trichodesma zeylanicum Cattle Bush BRASSICACEAE Blennodia pterosperma Wild Stock BRASSICACEAE Brassica tournefortii Wild Turnip BRASSICACEAE Harmsiodoxa blennodiodes/brevipes Hairy pod Cress/Short Cress BRASSICACEAE Lepidium oocytrichum Green Peppercress BRASSICACEAE Lepidium phlebopetalum Veined Peppercress CAESALPINIACEAE Senna filifolia -- CAESALPINIACEAE Senna pleurocarpa Smooth Senna CAESALPINIACEAE Senna sturtii Dense CHENOPODIACEAE Atriplex eardleyae Small Saltbush CHENOPODIACEAE Atriplex guinii ? -- CHENOPODIACEAE Atriplex holocarpa Pop Saltbush CHENOPODIACEAE Atriplex stipitata Bitter Saltbush CHENOPODIACEAE Chenopodium truncatum Black Crumbweed CHENOPODIACEAE Dissocarpus paradoxus Cannon-Ball CHENOPODIACEAE Enchylaena tomentosa Ruby Saltbush CHENOPODIACEAE Osteospermum acropterum Waterweed CHENOPODIACEAE Salsola kali Buck Bush CHENOPODIACEAE Sclerolaena decurrens Green Copperburr CHENOPODIACEAE Sclerolaena diacantha Grey Copperburr CHENOPODIACEAE Sclerolaena patenticuspis Spear-fruit Copperburr CONVOLVULACEAE Convolvulus erubescens Australian Bindweed CONVULVULACEAE Evolvulus alisnoides -- CONVULVULACEAE Ipomoea polymorpha Silky Cow Vine CUCURBIACEAE Citrullus lanatus Camel Melon EUPHORBIACEAE Chamaesyce drummondii Caustic Weed EUPHORBIACEAE Euphorbia drummondii Caustic Weed EUPHORBIACEAE Euphorbia eremophila Desert Spurge

- 227 -

EUPHORBIACEAE Euphorbia planiticola Plains Spurge EUPHORBIACEAE Phyllanthus fuernrohrii Sand Spurge EUPHORBIACEAE Phyllanthus sp. Sandhill Spurge EUPHORBIACEAE Sauropus trachyspermus Dwarf Spurge, Sandhill Spurge FABACEAE Crotalaria eremaea Loose-flowered Rattlepod FABACEAE Cullen australasicum Tall Verbine FABACEAE Glycine canescens Silky Glycine FABACEAE Indigofera colutea Rusty Indigo FABACEAE Indigofera psammophila Indigo FABACEAE Isotropis wheeleri Wheeler’s Lambpoison FABACEAE Swainsonia phacoides Lilac Darling Pea FABACEAE Tephrosia sphaerospora Mulga Trefoil GERANIACEAE Erodium crinitum Short Storksbill GOODENIACEAE Goodenia cycloptera Serrated Goodenia GOODENIACEAE Goodenia fasicularis Silky Goodenia GOODENIACEAE Scaevola parvibarbata Bushy/Common Fanflower MALVACEAE Abutilon otocarpum Desert Chinese Lantern MALVACEAE Hibiscus kirchauffianus Velvet-leafed Hibiscus MALVACEAE Sida ammophila ? Sand Sida MALVACEAE Sida corrugata ? Corrugated Sida MALVACEAE Sida cunninghamii Ridge Sida MALVACEAE Sida filiformis Fine Sida MALVACEAE Sida petrophila ? Rock sida ? MIMOSACEAE Acacia aneura Mulga MIMOSACEAE Acacia ligulata Sandhill Wattle MIMOSACEAE Acacia murrayana Sandplain Wattle MIMOSACEAE Acacia victoria Prickly Acacia MYOPORACEAE Eremophila longifolia Emubush NYCTAGINACEAE Boerhavia dominii Western Tar-Vine PLANTAGINACEAE Plantago drummondii/turrifera Sago-weed POACEAE Aristida contorta Bunched Kerosene Grass POACEAE Dactyloctenium radulans Button Grass POACEAE Enneapogon intermedius Tall Bottlewashers POACEAE Eragrostis ariopoda Wooly Butt POACEAE Eragrostis dielsii ? Mulka POACEAE Eriachne aristidea Broad-leaf Wanderrie Grass POACEAE Paractaenum novae-hollandiae Reflexed Panic POACEAE Tragus australianus Smallburr Grass POACEAE Triraphis mollis Purple Plume Grass PORTULACEAE Portulaca oleracea Common Pigweed PROTEACEAE Hakea leucoptera Needlewood SAPINDACEAE Alectryon oleifolius Rosewood SAPINDACEAE Atalaya hemiglauca Whitewood SAPINDACEAE Dodonea viscosa Broad-leaf Hopbush SOLANACEAE Solanum spec. -- THYMELACEAE Pimela simplex Desert Rice-Flower ZYGOPHYLLACEAE Tribulus micrococcus Yellow Vine, Spineles Caltrop ZYGOPHYLLACEAE Tribulus terrestris Cat-Head ZYGOPHYLLACEAE Zygophyllum howittii Clasping Twin-leaf ZYGOPHYLLACEAE Zygophyllum spec. Small-fruit Twinleaf Nomenclature is generally consistent with Cunningham (1992). ? after the scientific name indicate a likely but uncertain identification.

- 228 -

Appendix 3: Comparis on of overall ground cover between sampling sites. Significant differences (< 0.05) between individua l sites in mean cover and mean patchiness of ground cover variables (P ERMANOVA). df 1 =6, df 2 = 16; D = Disturbed and U = Undisturbed

Variable Status Site Session t p_MC Cover D 1 v 3 4 3.847 0.020 Cover D 1 v 4 2 3.005 0.010 Cover D 1 v 4 3 2.089 0.050 Cover D 2 v 3 2 1.928 0.050 Cover D 2 v 3 4 4.087 0.040 Cover D 2 v 4 1 3.623 0.020 Cover D 2 v 4 2 3.342 0.030 Cover D 3 v 4 13.280 0.010 Cover D 3 v 4 32.089 0.010 Cover U 1 v 3 24.160 0.020 Cover U 1 v 3 4 3.373 0.020 Cover2 U 1 v 4 3.241 0.020 Cover4 U 1 v 4 4 .112 0.030 Cover4 U 2 v 3 3.426 0.010 Cover4 U 2 v 4 5.232 0.010 Cover4 U 3 v 4 3.933 0.040 Patchiness D 1 v 2 1 2.059 0.050 Patchiness D 1 v 3 2 2.514 0.040 Patchiness D 1 v 4 1 3.574 0.010 Patchiness D 1 v 4 2 2.321 0.040 Patchiness D 1 v 4 4 2.586 0.020 Patchiness D 2 v 4 1 2.617 0.030 Patchiness D 3 v 4 1 2.327 0.040

- 229 -

Appendix 4: Comparison of individual ground cover variables between sites. Ground cover variables for which the interaction effect of disturbance and site was significant (nested ANOVA, n = 3). Sites with non-overlapping confidence intervals (95 %) were significantly different from each other. Wherever necessary data were transformed to achieve normalization. For variables where this was not possible the results of the Levene’s test are provided. Nested

ANOVA: df 1 = 6, df 2 = 16. Levene’s test: df 1 = 7, df 2 = 16. D = Disturbed and U = Undisturbed.

Levene´s test

Variable Ses- Trans- sion Fpformation t p Sites Cover Dry Pasture 1 6.228 0.002 none D1 D1, D2 Forb 1 5.912 0.002 none D1most others Copperburr 1 3.199 0.029 none 3.966 0.011 U3>D1, U1, D2 Bare ground 2 4.806 0.006 none D4D1, U1, D2 Forb 2 7.180 0.001 none D3>D1, D2 Grass 2 26.216 0.000 none 4.829 0.004 D4>all others Coperburr 2 3.059 0.034 none 3.333 0.022 D4>D1, U1, D2, U2 Litter 3 8.197 0.000 none U3, D4, U4>D1, D2, U2, D3 Bare ground 4 6.474 0.001 none 3.001 0.033 D1>U3, D4, U4 Dry Pasture 4 10.458 0.000 none 3.607 0.016 U4>D1, U1, D2, U2 Green Pasture 4 4.364 0.008 none U4>U1 and D4>D1, D2, U1 Litter 4 3.886 0.014 sqrt U3>D2, U2 Forb 4 11.463 0.000 none 4.014 0.010 U1, U2U2, D3 Grass 4 10.037 0.000 none 5.841 0.002 D4>all others Patchiness Dry Pasture 1 3.793 0.015 lg10 U1>D2, D4, U3 Forb 1 4.429 0.008 lg10 U1>D2, D4, U3, U4 Grass 1 4.760 0.006 none 12.976 << 0.001 D4>most others Copperburr 1 3.203 0.029 none 3.943 0.011 D1

- 230 -

Appendix 5: Dissimilarities in ground cover categories between sampling sessions. Results of SIMPER analysis of square-root transformed data. Only the three highest contributing categories are presented.

Session 1 v 2 Avg. Avg. Cover Cover Ground cover category 1 2 Avg.Diss Diss/SD Contrib% Green pasture 0.18 1.78 4.76 1.48 39.83 Dry pasture 2.57 1.88 2.36 1.83 19.73 Litter 3.93 4.25 1.40 1.20 11.73

Session 1 v 3 Avg. Avg. Cover Cover Ground cover category 1 3 Avg.Diss Diss/SD Contrib% Green pasture 0.18 3.77 10.53 10.37 45.57 Dry pasture 2.57 0.56 5.87 3.39 25.41 Litter 3.93 2.86 3.20 1.91 13.84

Session 2 v 3 Avg. Avg. Cover Cover Ground cover category 2 3 Avg.Diss Diss/SD Contrib% Green pasture 1.78 3.77 5.81 1.49 34.28 Litter 4.25 2.86 3.99 2.32 23.55 Dry pasture 1.88 0.56 3.79 3.41 22.32

Session 1 v 4 Avg. Avg. Cover Cover Ground cover category 1 4 Avg.Diss Diss/SD Contrib% Litter 3.93 4.73 2.39 1.61 27.09 Dry pasture 2.57 2.49 1.76 1.30 19.94 Bare ground 8.62 8.23 1.33 1.50 15.11

Session 2 v 4 Avg. Avg. Cover Cover Ground cover category 2 4 Avg.Diss Diss/SD Contrib% Green pasture 1.78 0.51 4.25 1.61 37.91 Dry pasture 1.88 2.49 1.98 1.47 17.68 Litter 4.25 4.73 1.74 1.52 15.49

Session 3 v 4 Avg. Avg. Cover Cover Ground cover category 3 4 Avg.Diss Diss/SD Contrib% Green pasture 3.77 0.51 9.42 8.08 39.21 Dry pasture 0.56 2.49 5.55 3.57 23.09 Litter 2.86 4.73 5.38 3.03 22.40

- 231 -

Appendix 6: Dissimilarities in ground cover categories between sites. Results of SIMPER analysis based on square-root transformed data. Only the three highest contributing categories are presented.

Session 1 Ground cover category Avg. cover Av.Sim Sim/SD Contrib% Bare ground 8.62 49.75 25.25 53.69 Litter 3.93 21.61 11.11 23.32 Dry pasture 2.57 13.08 4.09 14.12

Session 2 Ground cover category Avg. cover Av.Sim Sim/SD Contrib% Bare ground 8.40 45.86 12.79 51.25 Litter 4.25 22.55 12.48 25.20 Dry pasture 1.88 9.45 9.91 10.57

Session 3 Ground cover category Avg. cover Av.Sim Sim/SD Contrib% Bare ground 8.67 50.08 26.74 53.16 Green pasture 3.77 21.43 18.25 22.74 Litter 2.86 15.02 7.74 15.95

Session 4 Ground cover category Avg. cover Av.Sim Sim/SD Contrib% Bare ground 8.23 45.76 16.90 49.33 Litter 4.73 25.43 14.62 27.42 Dry pasture 2.49 12.27 5.90 13.22

Appendix 7: Differences in invertebrate community variables and individual invertebrate groups between sites. Significant results of nested ANOVA (df 1= 6, df 2 = 16; n = 3). For those variables where data normalization could not be achieved through transformation, the results of the Levene’s test are provided.

Levene`s test Sites with non- Ses- transfor- Variable Fpoverlapping CI df F p sion mation (95%) No. Invert.groups 1 4.021 0.012 D1>D2; U1>U2 2 8.136 << 0.001 D4>D1,D2; U4>U1 Total No. Captures 1 2.760 0.049 none lg10 2 5.355 0.003 D4>D1,D3; U4>U1 7, 16 4.948 0.004 Diversity 4 4.275 0.009 U3>U1 7, 16 3.127 0.028

Formicidae 2 3.516 0.021 D4>U2,U3 Araneae 1 12.684 << 0.001 D4, U4> all others 2 3.269 0.027 D4>D2 sqrt 4 20.621 << 0.001 D1, U1< all others sqrt Caelifera 2 2.983 0.038 U4>D1 7, 16 6.58 0.001 3 3.250 0.028 D4> all others 7, 16 4.514 0.006 4 22.961 D1,D2U3,U4 sqrt Coleoptera 3 11.111 << 0.001 U3> all others 7, 16 11.43 0 Blattodea 4 5.303 0.003 D1U1, U2 7, 16 10 0 4 4.545 0.007 D3> all others - 232 -

Appendix 8: Site differences in lizard community variables and individual lizard species. Significant results of nested ANOVA (df 1= 6, df 2 = 16, n = 3). For those variables where data normalization could not be achieved through transformation, the results of the Levene’s test are provided.

Levene`s test

Ses- Sites with non- transfor- Variable Fp df F p sion overlapping CI (95%) mation No Species 1 3.115 0.032 U1>D1, U2 Total abundance 2 2.796 0.047 U1< U4 Richness 1 3.043 0.035 D1U1 Diversity 4 3.405 0.023 D3>U1 Ct. taeniatus 1 3.306 0.026 none sqrt 2 2.872 0.043 none 7, 16 5.356 0.003 4 5.285 0.004 D1, D2 < U2, D4 sqrt C. pictus 1 3.512 0.021 D1, U1< D4 7, 16 7.050 0.001 2 5.589 0.003 D1, U1 < D2, U3, D4 3.131 0.028 C. fordi 1 6.964 0.001 D1,U1 > all others 7, 16 11.488 << 0.001 2 5.714 0.002 D1 > all others 7, 16 6.637 0.001 3 3.000 0.037 D1 > all others 7, 16 4.000 0.010 R. ornata 1 3.016 0.036 U1 > D1, D2, U2, U3 7, 16 3.879 0.012 Skinks 2 2.815 0.046 U1, D1< U4 3 2.792 0.047 none 4 2.993 0.037 D1 < D4 sqrt 7, 16 4.182 0.008 Ctenotus genus 2 3.049 0.035 U1, D1 < U4 7, 16 3.128 0.028 4 4.556 0.007 U1, D1 < D4 sqrt 7, 16 3.871 0.012

- 233 -

Appendix 9: Dissimilarities in invertebrate groups between sampling sessions. Results of SIMPER analysis based on number of captures (square-root transformed). Only the three highest contributing categories are presented.

Simper results for differences between sessions Session 1 v 2 Avg. Avg. Abund Abund Invertebrate category 1 2 Avg.Diss Diss/SD Contrib% Formicidae 14.93 9.09 10.16 1.23 28.32 Araneae 5.85 3.00 4.86 1.51 13.56 Caelifera 4.42 2.60 3.72 1.38 10.38

Session 1 v 3 Avg. Avg. Abund Abund Invertebrate category 1 3 Avg.Diss Diss/SD Contrib% Formicidae 14.93 7.81 10.79 1.30 24.51 Caelifera 4.42 0.39 6.51 3.07 14.79 Lepidoptera 0.00 2.93 5.03 2.49 11.42

Session 2 v 3 Avg. Avg. Abund Abund Invertebrate category 2 3 Avg.Diss Diss/SD Contrib% Lepidoptera 0.13 2.93 5.74 2.32 14.28 Formicidae 9.09 7.81 4.39 1.47 10.91 Caelifera 2.60 0.39 4.29 1.70 10.67

Session 1 v 4 Avg. Avg. Abund Abund Invertebrate category 1 4 Avg.Diss Diss/SD Contrib% Formicidae 14.93 7.37 10.85 1.26 33.14 Araneae 5.85 6.24 4.36 1.29 13.32 Caelifera 4.42 4.14 3.60 1.47 11.00

Session 2 v 4 Avg. Avg. Abund Abund Invertebrate category 2 4 Avg.Diss Diss/SD Contrib% Araneae 3.00 6.24 6.40 1.81 17.14 Formicidae 9.09 7.37 5.24 1.14 14.04 Caelifera 2.60 4.14 4.38 1.34 11.72

Session 3 v 4 Avg. Avg. Abund Abund Invertebrate category 3 4 Avg.Diss Diss/SD Contrib% Caelifera 0.39 4.14 6.35 1.80 14.30 Araneae 3.79 6.24 5.72 2.44 12.87 Blattodea 0.73 3.67 5.07 2.01 11.40

- 234 -

Appendix 10: Dissimilarities in invertebrate categories between sites. Results of SIMPER analysis based on invertebrate abundances (square-root transformed). Only the three highest contributing categories are presented.

Session 1

Invertebrate category Avg. Abund Av.Sim Sim/SD Contrib% Formicideae 14.93 28.44 4.43 39.72 Araneae 5.85 12.41 3.93 17.34 Caelifera 4.42 9.51 4.37 13.29

Session 2

Invertebrate category Avg. Abund Av.Sim Sim/SD Contrib% Formicidae 9.09 28.72 4.81 42.64 Coleoptera 3.80 10.37 3.48 15.39 Araneae 3.00 8.79 3.42 13.05

Session 3

Invertebrate category Avg. Abund Av.Sim Sim/SD Contrib% Formicidae 7.81 29.14 8.50 39.15 Araneae 3.79 14.12 5.97 18.97 Lepidoptera 2.93 9.67 3.73 12.99

Session 4

Invertebrate category Avg. Abund Av.Sim Sim/SD Contrib% Formicidae 7.37 18.94 3.75 26.95 Araneae 6.24 13.95 2.30 19.84 Thysanura 3.10 8.03 2.25 11.42

- 235 -

Appendix 11: Dissimilarities in lizard species between sampling sessions. Results of SIMPER analysis based on species abundances (square-root transformed). Only the three highest contributing categories are presented.

Session 1 v 2 Avg. Avg. Abund Abund Lizard Species 1 2 Avg. Diss Diss/SD Contrib% Ctenotus taeniatus 4.28 2.47 5.90 1.38 13.30 Lerista labialis 2.06 0.25 5.12 3.10 11.55 Lucasium damaeum 2.86 1.20 4.72 2.35 10.66

Session 1 v 3 Avg. Avg. Abund Abund Lizard Species 1 2 Avg. Diss Diss/SD Contrib% Ctenotus taeniatus 4.28 2.19 7.32 1.46 12.74 Lucasium damaeum 2.86 0.73 6.99 3.00 12.18 Lerista labialis 2.06 0.00 6.75 5.80 11.75

Session 2 v 3 Avg. Avg. Abund Abund Lizard Species 1 2 Avg. Diss Diss/SD Contrib% Ctenophorus pictus 2.21 1.80 6.81 1.21 15.41 Ctenotus taeniatus 2.47 2.19 6.76 1.16 15.28 Eremiascincus fasciolatus 1.20 0.00 6.17 2.05 13.95

Session 1 v 4 Avg. Avg. Abund Abund Lizard Species 1 2 Avg. Diss Diss/SD Contrib% Ctenophorus pictus 2.90 1.12 5.24 1.80 13.82 Ctenotus taeniatus 4.28 2.55 4.95 1.42 13.07 Ctenotus schomburgkii 3.13 2.27 3.34 1.34 8.82

Session 2 v 4 Avg. Avg. Abund Abund Lizard Species 1 2 Avg. Diss Diss/SD Contrib% Lucasium damaeum 1.20 2.77 5.57 1.65 12.35 Ctenophorus pictus 2.21 1.12 4.98 1.71 11.04 Ctenotus taeniatus 2.47 2.55 4.61 1.28 10.22

Session 3 v 4 Avg. Avg. Abund Abund Lizard Species 1 2 Avg. Diss Diss/SD Contrib% Lucasium damaeum 0.73 2.77 8.64 2.05 14.84 Lerista labialis 0.00 1.52 6.35 2.73 10.89 Eremiascincus fasciolatus 0.00 1.38 5.85 3.28 10.04

- 236 -

Appendix 12: Dissimilarities in lizard species between sites. Results of SIMPER analysis based on abundances (square-root transformed). Only the three highest contributing categories are presented.

Session 1

Lizard Species Avg. Abund Avg. Sim Sim/SD Contrib% Ctenotus taeniatus 4.28 15.84 3.31 22.81 Lucasium damaeum 2.86 11.24 6.39 16.19 Ctenotus schomburgkii 3.13 10.44 4.20 15.03

Session 2

Lizard Species Avg. Abund Avg. Sim Sim/SD Contrib% Ctenotus schomburgkii 2.33 16.78 7.04 26.79 Ctenotus taeniatus 2.47 13.71 1.50 21.89 Ctenophorus pictus 2.21 12.16 1.45 19.42

Session 3

Lizard Species Avg. Abund Avg. Sim Sim/SD Contrib% Ctenotus taeniatus 2.19 20.57 1.44 34.93 Ctenotus schomburgkii 1.69 18.56 4.78 31.51 Ctenophorus pictus 1.80 14.04 1.03 23.83

Session 4

Lizard Species Avg. Abund Avg. Sim Sim/SD Contrib% Lucasium damaeum 2.77 14.11 4.62 22.06 Ctenotus taeniatus 2.55 10.92 2.48 17.08 Ctenotus schomburgkii 2.27 9.61 2.94 15.03

- 237 -

Appendix 13: Rainfall in the years 2005 (a), 2006 (b) and 2007 (c) in Australia and in the range of Notomys fuscus (shady grey area). The widespread rain of 100-200 mm in 2005 over the range of the hopping mice presumably triggered a population increase. After a dry year of 2006 significant rainfall early in 2007 allowed N. fuscus population to persist and further increase leading to high densities and range expansions later in 2007 and 2008. a) 2005

b) Rainfall event June 2005

- 238 -

c) 2006

d) 2007

- 239 -

Appendix 14: Kernel home ranges (25 %, 50 %, 75 % and 95 % of fixes) of Notomys fuscus. Home ranges in red, pink, yellow or green belong to female mice; home ranges in blue, purple or white to male mice. The house symbol indicates the location of the burrow of the respective mouse. The colour of the house symbol matches the home range colour of the respective mouse. Acacia ligulata shrubs, an important resource for N. fuscus have been added to the image where appropriate, in the form of tree symbols.

a) Bertha (female): green, Wilma (female): pink, Maja (female): yellow and Bert (male): white.

- 240 -

b) Susie (female)

c) Minnie (female): pink, Miffy (female): red and Ernie (male):blue

- 241 -

d) Dennis (male): purple; unknown male: blue

e) Bianca (female)

- 242 -

f) Donald (male): blue, Daisy0 (female): pink, Daisy2 (female): yellow. Note: Unfortunately the image quality for this location is much lower than that for the other sites.

- 243 -

Appendix 15: Exemplary photographs of N. fuscus footprints and ‘runways’ to illustrate the scoring system used in measuring N. fuscus activity as an index of density. a) Score 1 b) Score 2

c) Score 3

- 244 -

Appendix 16: Example of accumulation of N. fuscus tracks indicating a food source in this case Cattle Bush (Trychodesma zylanecum).

Appendix 17: Acacia ligulata shrub with flowers, seed pods and an accumulation of fallen seeds on the ground underneath. The seeds are a major comp onent of the diet of N. fuscus.

- 245 -

Appendix 18: Remote camera imagery (infra-red) of a female N. fuscus and her young on burrow exit. The female encourages her young to emerge from the burrow and the group hops off together.

- 246 -

Appendix 19: Remote camera imagery of a group of eight N. fuscus congregated around the burrow entrance. This was the largest group recorded. It consisted of at least five juveniles (smaller individials to left and background) and two adults (to right), one of them likely the mother. A further N. fuscus is in the process of entereing the burrow.

- 247 -

Appendix 20: Co-habitation, predation and/or competition? Remote camera imagery of a Sand Goanna (Varanus gouldii), a House Mouse and a N. fuscus using the same burrow. All three images were taken within a period of 15 hours.

- 248 -

Appendix 21: Varanus gouldii, a likely native predator of N. fuscus, investigates a hopping mouse burrow and almost completely disappears in it.

- 249 -

Appendix 22: Comments on methodology used in research on Notomys fuscus

Pitfall traps were found to be a lot more successful in capturing N. fuscus than Elliott traps. A depth of at least 60 cm is recommended for pitfall traps. Even at this depth in a number of cases adult N. fuscus were able to escape from the traps.

N. fuscus is very fond of Sunflower seeds and they can be used as bait in Elliott traps. The effectiveness of pitfall traps can also be enhanced by distributing some Sunflower seeds around the rim of the trap. When specifically targeting Hopping Mice, pitfall trapping success can be enhanced further by covering the opening with a round, light piece of cloth (eg. domestic wipes) that is held in place and weighted down by small amount of sand/dirt around the edges. Sunflower seeds which are placed in the centre of the cloth entice the Hopping Mice to move onto the cloth-covered opening, the cover does not hold their weight and the mouse falls through into the pit beneath.

In winter, N. fuscus are prone to accidental death when captured in pitfall traps, juveniles especially are at risk. Accidental deaths have never been a problem in surveys during the warmer months. So, presumably N. fuscus expend a lot of energy in the attempt to escape from the pitfall traps and are then left with too little energy to retain the necessary body temperature. To minimize the rate of accidental death of N. fuscus during winter surveys, if possible food should be made available in the pitfall traps (e.g. Sungflower seeds) in addition to providing cushioning material in the bottom of the pit (cotton wool in this study).

An effective method to capture particular mice, for example to remove a radio-collar, or to determine social grouping, is to encircle the burrow with all its entrances with a barrier (e.g. pitfall fencing material) and set up pitfall traps and Elliott traps on the inner side of the fence. Sometimes the mice avoid capture by opening and using previously hidden entrances outside the ‘coral’ in which case the fence needs to be adjusted. This method is successful only for Hopping Mouse burrows that are not associated with rabbit warrens. Water dishes are readily visited by Hopping Mice and can be used to attract mice to certain places for example camera locations.

- 250 -