MANAGEMENT OF AN INTRODUCED AND ABUNDANT

Transcription

1 MANAGEMENT OF AN INTRODUCED AND ABUNDANT ISLAND POPULATION OF KOALAS Kris Michael Carlyon A thesis in fulfillment of the requirements for the degree of Doctor of Philosophy School of Biological, Earth and Environmental Sciences Faculty of Science University of New South Wales September 2013

2 ii

3 Originality Statement I hereby declare that this submission is my own work and to the best of my knowledge it contains no materials previously published or written by another person, or substantial proportions of material which have been accepted for the award of any other degree or diploma at UNSW or any other educational institution, except where due acknowledgement is made in the thesis. Any contribution made to the research by others, with whom I have worked at UNSW or elsewhere, is explicitly acknowledged in the thesis. I also declare that the intellectual content of this thesis is the product of my own work, except to the extent that assistance from others in the project's design and conception or in style, presentation and linguistic expression is acknowledged. Signed... Date 20 September iii

4 Preface This thesis includes five stand-alone data chapters; therefore each of these chapters is selfcontained and some repetition occurs. Two of these chapters have already been published in peerreviewed scientific journals and include the work of several co-authors. My contribution to each of the chapters is stated below: Chapter 3: Whisson, D.A. and Carlyon, K. (2010) Temporal variation in reproductive characteristics of an introduced and abundant island population of koalas. Journal of Mammalogy, 91: Candidate contribution: 50% Chapter 7: Whisson, D.A., Holland, G.J. and Carlyon, K. (2012) Translocation of overabundant species: implications for translocated individuals. Journal of Wildlife Management, 76: Candidate contribution: 30% Declaration I certify that these publications were a direct result of my research towards this PhD, and that reproduction in this thesis does not breach copyright regulations. Signed... Date 19 September Supervisor s Statement I hereby certify that all co-authors of the published papers agree to Kris Carlyon submitting these papers as part of his doctoral thesis. Signed... Date 19 September iv

5 Acknowledgements First and foremost my sincere thanks go to my supervisors: Des Cooper, Cath Herbert, Desley Whisson and Kath Handasyde. Des, you leave behind a huge legacy in the many students you mentored I m so thankful to have been a small part of that. Cathy, thank you so much for the faith you showed in taking me on as a student, your help in the field and for your patience when I would shift my focus to other things. We got there in the end. Thanks Desley for your friendship, encouragement and positivity I always came away from our koala discussions feeling better about my work. Kath, thanks so much for starting me on the zoology path, you are a major reason why I do what I do I ll hold you to that drink at Boho, I can t think of a better place to celebrate than back where this whole koala thing began. To the fantastic people of Kangaroo Island thank you for making the four years I spent on that incredible island some of the most enjoyable of my life so far. There are far too many of you to mention here, but you know who you are. Thanks in particular to Janelle, Laura, Evelyn and Ange; you guys are the best and now, with this thesis finished, I can t wait to have a few more adventures beyond KI. Thanks also to Pip, Rick and Matilda for sharing your home I won t live anywhere with a better view. I owe a huge amount of gratitude to staff at the KI Conservation Programs Unit. Desley, Dave D, Trish, Phil, Dave T, Bill, Robyn, Janet: thanks for supporting this work and being so inclusive. Many thanks to Glenn Shimmin for input into project design. The various koala catch teams what a crazy, eclectic, awesome bunch of people you are. Thanks for making the work both easier and fun. Steve, I turned up thinking I knew a bit about catching koalas how wrong I was. You taught me so much. Thanks must also go to the landowners of Kangaroo Island, whose permission for me to access their land at all hours over a number of years made this project possible. As a result, I saw many beautiful parts of the island I would never have got to otherwise (and, I admit, possibly eaten a few more marron than I might have otherwise). Cheers. Thanks to my co-authors on the two published manuscripts that form a part of this work I think these chapters add significant breadth to the thesis. Thank you Greg for sharing your statistical skills at such short notice. v

6 My thanks also go to the rest of the students and project team of the Koala and Kangaroo Contraception Program it was a privilege to be part of such a dynamic and passionate group of researchers. This research was made possible by funding from a number of agencies that supported the KKCP in an ARC linkage grant, particularly Peptech Animal Health and the South Australian Department for Environment and Heritage (now DEWNR). To my current colleagues at DPIPWE, particularly Rachael Alderman, I am extremely grateful for the support, time and encouragement you have given me to complete this thesis I would not have got there otherwise. Love and thanks go to my family who support me unconditionally in everything I do. Even when disappearing to start a PhD on a remote island. Or then moving south to live on another island. At least you have good holiday options. Thank you for being there no matter what I rely on you more than you know. While I credit Kath for starting me on the professional zoology path, it is Mum and Dad who instilled a love of the wild I will be forever grateful. Shannon, thank you for sharing this with me you have watched this whole process unfold and provided much needed perspective, statistical genius and so much understanding. And a mad dog. Your love and support mean so much. vi

7 Publications and Conference Presentations Peer-reviewed publications Whisson, D.A., Holland, G.J. and Carlyon, K. (2012) Translocation of overabundant species: Implications for translocated individuals. Journal of Wildlife Management, 76: Whisson D.A. and Carlyon, K. (2010) Temporal variation in reproductive characteristics of an introduced and abundant island population of koalas. Journal of Mammalogy, 91: Cristescu, C., Cahill, V., Sherwin, W.B., Handasyde, K.A., Carlyon, K., Whisson, D., Herbert, C.A., Carlson, B.L.J., Wilton, A.N. and Cooper, D.W. (2009) Inbreeding and testicular abnormalities in a bottlenecked population of koalas. Wildlife Research, 36: Conference presentations Whisson, D.A. and Carlyon, K. (2008) Demographics of an island population of koalas. Australian Mammal Society Conference, Darwin, NT, Australia. Whisson, D.A., Eggleston, K., Carlyon, K., Dowie, D. and Molsher, R. (2008) Habitat use by koalas translocated from Kangaroo Island to south eastern South Australia. Australian Mammal Society Conference, Darwin, NT, Australia. Carlyon, K., Whisson, D. A., Herbert, C. A., Handasyde, K., Dowie, D., Molsher, R. and Cooper, D.W. (2008) Ranging behaviour and habitat use of koalas on Kangaroo Island: implications for management. Australasian Wildlife Management Society Conference, Fremantle, WA, Australia. Carlyon, K., Whisson, D.A., Herbert, C.A., Handasyde, K.A., Trigg, T. and Cooper, D.W. (2007) Slow release contraceptive implants containing the GnRH agonist deslorelin: effects on ranging behaviour of female koalas (Phascolarctos cinereus) on Kangaroo Island. Fenner Conference on the Environment incorporating the Australasian Wildlife Management Society Conference, Canberra, ACT, Australia. Cristescu, R., Tanaka, M., Herbert, C.A., Carlyon, K., Wilton, A., Whisson, D.A. Handasyde, K.A., Cahill, V. and Cooper, D.W. (2007) Genetic variation in koalas on French Island and Kangaroo Island and the likely effect of contraception protocols on its retention. Genetics Society of Australasia Annual Conference, Sydney, NSW, Australia. Carlyon, K., Herbert, C.A., Handasyde, K.A., Whisson, D.A., Trigg, T. and Cooper, D.W. (2006) Contraception for the masses: deslorelin implants as a tool for population-scale management of koalas on Kangaroo Island. Australian Mammal Society Conference, Melbourne, VIC, Australia. vii

8 Table of Contents ORIGINALITY STATEMENT... III PREFACE...IV PUBLICATIONS AND CONFERENCE PRESENTATIONS... VII TABLE OF CONTENTS... VIII LIST OF FIGURES... XII LIST OF TABLES... XV ABSTRACT GENERAL INTRODUCTION HERBIVORE OVERABUNDANCE THE KOALA CONSERVATION AND MANAGEMENT POTENTIAL POPULATION MANAGEMENT TECHNIQUES Translocation Fertility control KANGAROO ISLAND This study MATERIALS AND METHODS STUDY AREA KOALA CAPTURE AND HANDLING COLLECTION OF MORPHOMETRIC AND REPRODUCTIVE DATA, AND FITTING OF RADIO- COLLARS GNRH AGONIST CONTRACEPTIVE IMPLANT BLOOD SAMPLING GnRH challenge HORMONE ASSAYS SURGICAL STERILISATION MONITORING KOALA MOVEMENTS Radio-tracking and recapture for monitoring viii

9 Modification of collars after koala mortality DATA ANALYSIS AUTHORISATION FEMALE REPRODUCTIVE CHARACTERISTICS OF AN INTRODUCED AND ABUNDANT ISLAND POPULATION OF KOALAS INTRODUCTION Study area Kangaroo Island koalas and their management Koala capture and processing Data Analysis RESULTS Timing of births Young sex ratio Maternal mass Maternal age Maternal condition DISCUSSION CONCLUSION EFFICACY OF A SLOW-RELEASE CONTRACEPTIVE IMPLANT CONTAINING THE GNRH AGONIST DESLORELIN IN FREE-RANGING FEMALE KOALAS INTRODUCTION MATERIALS AND METHODS Study sites Koala capture and treatment Analysis RESULTS Management simulation Northeast River Fertility/presence of young Northeast River Fertility/presence of young telemetry study, Pioneer Bend GnRH challenge - radio-collared female koalas, Pioneer Bend DISCUSSION THE EFFECTS OF TWO FERTILITY CONTROL TECHNIQUES ON THE HEALTH AND CONDITION OF FREE-RANGING FEMALE KOALAS ix

10 5.1 INTRODUCTION MATERIALS AND METHODS Study animals Body weight change Muscle condition Survival RESULTS Treatment Body weight changes Muscle condition Adult survival DISCUSSION THE EFFECTS OF TWO FERTILITY CONTROL TECHNIQUES ON THE MOVEMENT PATTERNS AND RANGING BEHAVIOUR OF FEMALE KOALAS INTRODUCTION MATERIALS AND METHODS Study area Capture, collection of morphometric data and treatment Animal observations Analysis RESULTS Site fidelity and long-distance movement Movement rate Core home-range size DISCUSSION TRANSLOCATION OF OVERABUNDANT SPECIES: IMPLICATIONS FOR TRANSLOCATED INDIVIDUALS INTRODUCTION MATERIALS AND METHODS Study areas Koala density at release sites Immediate versus delayed translocation after sterilisation ( ) Translocated individuals versus residents on Kangaroo Island ( ) x

11 7.3 RESULTS Koala population density at release sites Immediate versus delayed translocation after sterilisation ( ) Translocated individuals versus residents on Kangaroo Island ( ) DISCUSSION DISCUSSION A REVIEW OF MANAGEMENT OPTIONS FOR CONTROL OF AN OVERABUNDANT POPULATION OF KOALAS ON KANGAROO ISLAND, SOUTH AUSTRALIA Treatment efficacy Treatment duration Availability and delivery of treatment Humaneness Target specificity and environmental acceptability Cost-effectiveness Reduction in population size and environmental damage CONCLUSION Management recommendations REFERENCES xi

12 List of Figures Figure 1.1: Current koala distribution 7 Figure 2.1: Kangaroo Island, South Australia, showing study sites Figure 2.2: Mean monthly ( ) minimum and maximum temperatures (ºC) and rainfall (mm) for Kingscote, Kangaroo Island Figure 2.3: Annual rainfall at Pioneer Bend, Kangaroo Island Figure 2.4: Slow-release GnRH agonist implant (Suprelorin ) Figure 3.1: Kangaroo Island showing major catchments and areas of high- and mediumquality habitat preferred by koalas Figure 3.2: Cumulative percent of male (solid line) and female (dotted line) births of koalas from November to May Figure 3.3: Percentage of independent females in each weight class with dependent young Figure 3.4: Mean body mass of adult koalas on Kangaroo Island (black) compared to the founder population on French Island Figure 3.5: Mean weight of female koalas giving birth to male young (solid line) or female young (dotted line) during the November April breeding period Figure 4.1: Kangaroo Island, South Australia, showing virtual boundaries of management simulation sites Figure 4.2: Percentage of recaptured and new female koalas caught and implanted in the Northeast River site, Kangaroo Island, each management year Figure 4.3: Fecundity (percentage of females >6.0 kg with dependent young) of newly caught and recaptured females in each management year Figure 4.4 Percentage of female koalas in each treatment group at Pioneer Bend, Kangaroo Island that produced young in each breeding season during the course of this study xii

13 Figure 4.5: Mean plasma concentrations of LH following administration of a synthetic GnRH (2 ug/kg) to control (dashed line; n=8) and suprelorin-treated (solid line; n=7) female koalas Figure 5.1: Predicted percentage change in body weight of adult female koalas from the initial mean capture weight by treatment group during the course of the study Figure 5.2: Survival of female koalas in each treatment group during the course of the study as a percentage of koalas in each treatment group Figure 6.1: Location of the study site at Pioneer Bend, Kangaroo Island, South Australia Figure 6.2: Examples of the different movement patterns exhibited by female koalas during the study in the Pioneer Bend area, Kangaroo Island Figure 6.3: Plots for individual koalas showing distance (m) of each fix from original capture point (y-axis) over the course of the movement study November 2006 July 2008 (xaxis) Figure 6.4: Percentage of female koalas in each treatment group that undertook a longdistance movement during each breeding season and then both seasons pooled Figure 6.5: Mean duration of long-distance movements (where applicable) for female koalas in each treatment group at Pioneer Bend, Kangaroo Island Figure 6.6: Frequency of long-distance movements by female koalas (treatments pooled) that began in each month Figure 6.7: GAMM-predicted mean distance from original capture of female koalas in each treatment group as a function of time Figure 6.8: Mean distance (in metres) between successive fixes over the course of the study (seasons pooled) for female koalas within each treatment group Figure 6.9: Estimated marginal means of movement rates (distance between successive fixes) for female koalas in each treatment group for each season Figure 6.10: Mean size of core home-ranges as estimated by cluster analysis for female koalas in each treatment group at Pioneer Bend, Kangaroo Island xiii

14 Figure 7.1: Kangaroo Island and the mainland koala release region in the Lower South East, South Australia Figure 7.2: The predicted distance moved by translocated male (dashed line) and female (solid line) koalas ( study) as a function of time Figure 7.3: The predicted distance moved by sterilised and translocated koalas (dashed line), and intact koalas remaining on Kangaroo Island (solid line) in , as a function of time xiv

15 List of Tables Table 2.1: Scale for converting tooth wear class (TWC) to an estimate of age (years), based on data from known-age Victorian koalas Table 3.1: Variation in female condition and maternal characteristics among years Table 4.1: Number of individual adult female koalas caught and implanted in single or multiple management years in the Northeast River site, Kangaroo Island Table 5.1: Output from generalised additive mixed model used to investigate changes to body weight through time of adult female koalas subjected to fertility control Table 5.2: Survival rates of known young born from koalas in each treatment group across different seasons Table 6.1: Output from generalised additive mixed model used to investigate changes to the distance from original capture site for female koalas subjected to fertility control Table 6.2: Summary of home-range size estimates (ha; in ascending order) for adult female koalas reported from other studies at various locations around Australia Table 7.1: Total number of koalas released and subsequent koala density in 16 mainland release sites that were regularly surveyed Table 7.2: Survival of radio-collared koalas in all treatment groups from radio-tracking studies in and Table 7.3: Results from the optimal generalized linear mixed effects model relating the distances moved by individual koalas to sex, treatment, and week Table 8.1: Successful female koala fertility control agents and their effects on individuals Table 8.2: Considerations for management and further research resulting from this study xv

16 Abstract High-density free-ranging animal populations have potential to cause ecological and economic problems often necessitating active management to reduce numbers or to limit population growth. Fertility control is a more ethically acceptable management alternative to lethal methods, and is particularly attractive when the target population is a native species and eradication is undesirable. Some high-density koala populations, particularly those occupying habitat isolates with limited dispersal opportunities, have the potential to cause extensive damage to their preferred habitat. Without intervention this can result in death of food trees, welfare impacts for the resident koala population and destruction of habitat for other species. Clearly these populations need managing. This thesis examines potential and current management options for controlling the introduced and overabundant koala population on South Australia s Kangaroo Island. I examined the impact of these techniques on individual animals, with particular focus on trialling the commercially available GnRH-agonist contraceptive implant (Suprelorin, Peptech Animal Health, Sydney) and comparing the outcomes to the current management practice of surgical sterilisation and translocation. An assessment of the reproductive characteristics of the population was also undertaken as management on Kangaroo Island has, to-date, been based on reproductive parameters derived from Victorian populations. Suprelorin implants are simple to deliver in the field, provide a contraceptive effect in female koalas for 1-2 years and have no apparent adverse impacts on health or ranging behaviour of treated individuals. Likewise, surgical sterilisation, whilst costlier and more invasive, had no adverse effects on the behaviour or health of female koalas if animals were released back at the site of capture. However, increased mortality was observed in surgically sterilised koalas in the first 12 months after translocation to mainland South Australia. 1

17 Due to the relatively short duration of contraception provided by Suprelorin and the difficulty and large cost of capturing koalas on Kangaroo Island, it is not currently considered practical for large-scale koala population management at this site. Suprelorin implants have potential for smaller-scale management programs where capture is straightforward or in situations when permanent sterilisation is not desirable. Development of remote delivery techniques would significantly increase Suprelorin s efficacy for management. 2

18 3

19 1. General Introduction 1.1 Herbivore overabundance Clearing of land, resulting in the fragmentation of natural environments, has a profound influence on the distribution and abundance of many wildlife species (Bennet, 1987). As a result, habitat fragmentation has become an issue of increasing concern for the conservation of biota in environments throughout the world, particularly in regard to the protection of threatened species (Mac Nally et al., 2000). In a somewhat paradoxical situation however, these same processes that threaten vulnerable species can also result in the local overabundance of some animal populations, largely due to a reduction in dispersal opportunities (Caughley, 1981) but also if there is an absence of predators, competitors or parasites in anthropogenically-disturbed ecosystems (Torchin et al., 2003; Ripple and Beschta, 2012). Overabundance, as discussed in this thesis, refers to a human-defined and caused lasting increase in abundance or density of a wildlife population beyond that which would occur under natural circumstances. This is in contrast to a species that may exhibit a temporary eruptive or boom population increase in response to natural changes in environmental conditions, typically followed by a phase of rapid decline ( bust ) back to normal levels following exhaustion of resources (e.g. Whitford, 1976; Dickman, 1999). Caughley (1981) identified a number of categories for declaring a wildlife population overabundant: (1) if the high density of the population threatens human life or livelihood, for example impacting negatively on agriculture (e.g. Hill, 1997; Bleier et al., 2012), human health (e.g. Nemtzov, 2002) or animal health (e.g. Vial and Donnelly, 2012); (2) if animals depress the density of favoured species (e.g. Coulson, 1998; Augustine and DeCalesta, 2003); (3) 4

20 endanger their own species (e.g. Clarke and Ng, 2006); or (4) threaten ecosystem balance (e.g. Beschta, 2005). Herbivores have the potential to directly impact on their environment and other coexisting species, particularly if they occur at high density and their movement is restricted (e.g. Ben-Shahar, 1993; Kelly, 2000). Under these circumstances, populations can overbrowse available vegetation. For example, in Africa, elephants (Loxodonta africana) have been forced into small patches of habitat through fragmentation processes. As a result, the browse-pressure from large numbers of elephants has damaged vegetation, and the structure and composition of savannah woodlands has been severely altered (Ben-Shahar, 1993, 1998; Teren and Owen- Smith, 2010). In North America similar problems of biomass depletion and environmental damage are caused by several overabundant ungulate species, particularly elk (Cervus elaphus) and white-tailed deer (Odocoileus virginianus) (McShea and Rapole, 1997). The decline of top order predators is another factor that can contribute to such processes. For example the decline of wolves (Canis lupus) and resultant increase in ungulate prey populations has had negative impacts on forest regeneration in North America (see Ripple and Beschta, 2005). Fragmented or isolated vegetation patches are assumed to be at higher risk from herbivore foraging pressure than larger patches because they place restrictions on animal movements, concentrating pressure onto a relatively small number of individual plants. Surviving on a relatively poor diet, herbivores are subject to energy constraints, and most select forage from a variety of plant species in order to meet their energy requirements (Dearing et al., 2000). Specialist foragers, such as obligate folivores, that concentrate their feeding on only a few plant species, and larger animals (which have a greater food intake) have a superior potential to impact on vegetation, particularly when at high densities (Martin, 1983). Offshore islands may be considered analogous to a patch of fragmented habitat, as for most species dispersal and movement from the island is also inhibited. 5

21 Some wildlife species have a natural ability to regulate population growth in response to changes in environmental conditions. For example, natural population regulation mechanisms in African elephants include localised reductions in fertility in response to a reduction in food availability (Moss, 2001) and dispersal from areas of locally-high density (Owen-Smith 1983). These mechanisms can be impeded by anthropogenic interference, however, with culling regimes, rotational block burning, and provision of artificial waterholes upsetting these natural regulatory processes (Dickson and Adams, 2009). Kangaroos and wallabies (Macropus sp.) can also regulate reproduction in response to environmental changes, for example, during periods of severe drought females will cease breeding (Russell, 1974). This is achieved by either extending embryonic diapause, i.e. seasonal quiescence, or by a lack of follicles developing in the ovaries, i.e. seasonal anoestrus (Tyndale-Biscoe, 2005). Other species, such as the koala (Phascolarctos cinereus), which is overabundant in parts of its range, are unable to regulate reproduction in response to adverse conditions (Handasyde et al., 1990a), potentially exacerbating the impacts of overabundance in this species. 1.2 The koala The koala is amongst the largest of Australia s arboreal marsupials, inhabiting the eucalypt forests of eastern Australia (Martin, 1983). Large-scale de-forestation has had a severe impact on its distribution, however the current range remains vast, with populations distributed from the Atherton Tablelands in Queensland to temperate forests in Victoria and into SE South Australia (Fig. 1.1; Martin and Handasyde, 1999; Martin et al., 2008). Koalas occupy a variety of habitat types, occurring in coastal woodland, wet sclerophyll forests, tropical rainforest and dryer inland areas of semi-arid woodland. The most noticeable feature of the koala s ecology is its folivorous diet, consisting almost entirely of foliage from the genera Eucalyptus and Corymbia (Martin et al., 2008). 6

22 They can be extremely selective in their choice of food trees and some populations have been shown to feed solely on a single species of eucalypt (Martin, 1985a). As a result, the number of koalas that a particular area can support is highly dependent on the presence and density of these preferred food trees (Martin and Handasyde 1999). Figure 1.1 has been removed due to Copyright restrictions. Figure 1.1: Current koala distribution. Source: Department of Sustainability, Environment, Water, Population and Communities, Female koalas are polyoestrus and, during good conditions, give birth each year during a defined breeding season, with births concentrated between November and January (Handasyde et al., 1990a; Martin and Handaysyde, 1999). Koalas have a long breeding lifespan, with females able to successfully reproduce from around two years of age until they 7

23 are up to 15 years old (Martin and Handasyde, 1999). Females give birth to a single, altricial young after a gestation of 35 days, which spends at least the first six months in the pouch (Martin et al., 2008). The dependent young then travels on its mothers back until it is weaned about a year after birth. The female enters oestrus as the young approaches independence (Handasyde et al., 1990b). After weaning, the young may remain in association with the mother for a number of months, typically making their first movements away from the natal area between the months of July and November (Martin and Handasyde, 1999; Tucker, 2008). 1.3 Conservation and management A large koala fur trade that continued until the 1930s, along with hunting, disease, fire and drought, resulted in severe declines of koala populations following European settlement (Natural Resource Management Ministerial Council, 2009). Protection measures put in place following the end of the fur trade included a number of reintroduction programs that were used to re-establish koalas in parts of their former range (Martin and Handasyde, 1999). Today, the conservation status of the species varies across its range (see Melzer et al., 2000; Natural Resource Management Ministerial Council, 2009), from secure in some regions to vulnerable or even extinct in others. Nationally, the koala is listed as vulnerable under Commonwealth legislation in Queensland, New South Wales and the Australian Capital Territory due to sustained declines of these populations and ongoing threats (Threatened Species Scientific Committee, 2012; Burke, 2012). Populations in these areas typically occur at low density and continue to decline; they are also listed as threatened under relevant State legislation (Reed and Lunney, 1990; Natural Resource Management Ministerial Council, 2009). In contrast, koalas in southeastern Australia are not listed as threatened under Commonwealth or State legislation and now occupy most of their historical range due to large-scale active re-introduction programs 8

24 from island populations (Menkhorst, 2008), with some populations occurring at very high density. In some areas, such as Kangaroo Island, the Mt Lofty Ranges and the Eyre Peninsula in South Australia, koalas now occur beyond their historic range. In areas with limited dispersal opportunities and few predators, populations can increase rapidly, potentially doubling in size every 2.8 years (Martin and Handasyde, 1999). Chlamydiosis (infection of the reproductive tract with Chlamydia bacteria, principally C. pecorum or C. pneumonia; Polkinghorne et al., 2013)is prevalent in many populations, resulting in a reduction in fecundity and depressed population growth-rate (Martin and Handasyde, 1990a; McLean, 2003), however some Chlamydia-infected populations in habitat isolates have still increased to a point where they are at very high density, e.g. Raymond Island (Victoria) (Martin and Handasyde, 1999; Greenfield, 2012). At some locations, the apparent success of re-establishment of the species is tempered by unforeseen impacts resulting from over-browsing by high-density populations (Natural Resource Management Ministerial Council, 2009). Overabundant koala populations can have severe impacts on their habitat, having been identified as the cause of defoliation and a decline in the health of eucalypts at a number of forested sites. In severe cases koalas have killed their favoured food species (Martin, 1983, 1985a; Hindell and Lee, 1990; Kelly, 2000; Department of Sustainability and Environment, 2004; Masters et al., 2004). This has primarily been in areas with a high proportion of manna gum, Eucalyptus viminalis, the koala s favoured food tree in the southern Australian states (Martin et al., 2008), however koalas can cause serious defoliation to other southern eucalypt species such as E. globulus, E. ovata, E. leucoxylon, E. baxteri, E. obliqua, E. camphora and E. radiata. Overabundance (and the resultant impacts on habitat) appears to be a feature of southern koala populations only (i.e. those occupying habitat isolates in Victoria and South Australia). Koalas were introduced to Brampton, St. Bees, Newry, Rabbit, and Magnetic Islands in Queensland (Melzer et al., 2000), but, for reasons not 9

25 understood, have not reached the density of populations introduced to southern islands (Natural Resource Management Ministerial Council, 2009). Overbrowsing can have a number of effects on both the koalas themselves and local biodiversity: animal welfare concerns result when a dramatic reduction in food resources leads to starvation and death of resident koalas; and vegetation communities in isolated patches are threatened which may impact on other aspects of the biota (Herbert et al., 2001). In such situations, management to reduce koala population densities and their impacts on habitat (including subsequent impacts on other forest-dependent species), and to address koala welfare concerns may be desirable. Management of overabundant koala populations has been carried out in Victoria since the 1920s, typically involving translocation of individuals (Martin and Handasyde 1990b; Menkhorst, 2008; Natural Resource Management Ministerial Council, 2009). Culling of koalas has been proposed for several populations but has been rejected by successive governments due to public opposition (St John, 1997; Natural Resource Management Ministerial Council, 2009). More recently, fertility control techniques such as surgical sterilisation, hormonal sterilisation (Menkhorst et al., 1998; Menkhorst, 2008; Hynes et al., 2010; Greenfield, 2012) and immunocontraception using porcine zona pellucida and recombinant possum ZP3 (Kitchener et al. 2001) have been trialled. Formal management programs are ongoing at a number of southern Australian sites, and success (i.e. koala densities reduced to below target levels through intervention) has been reported at a number of locations, e.g. Snake Island (K. Handasyde, pers. comm.). 10

26 1.4 Potential population management techniques At present, authorities aiming to control overabundant koala populations have access to just two government sanctioned methods: translocation and fertility control (Natural Resource Management Ministerial Council, 2009). Translocation and surgical sterilisation have been the predominant koala population control strategies to date, however with recipient sites for translocation becoming saturated and less invasive fertility control methods becoming available, other options must be considered. In particular, hormonal contraception shows promising results and is currently used by government managers on French Island. Translocation Translocation involves the removal of animals from an overabundant population for free-release into suitable, available habitat elsewhere, resulting in a reduction of pressure on the original population and its habitat. A number of government programs aimed at reducing koala population numbers have involved translocation to other suitable habitat (Menkhorst et al., 1998; Duka and Masters, 2005; Whisson et al., 2012). From a population-establishment perspective, such programs were very successful, and in Victoria, in particular, the koala was re-established in many areas where populations had declined after European settlement (Department of Sustainability and Environment, 2004). However many of these sites now also hold populations that are beyond the habitat s carrying capacity. Thus most suitable locations currently support populations and there simply is not enough suitable habitat left for future translocations (Menkhorst, 2008). In recent years, translocation has been carried out in conjunction with surgical sterilisation (as with the current koala management practice on Kangaroo Island), with only sterile animals being translocated (e.g. Duka and Masters, 2005); an approach that avoids the problem of overabundance developing at the new (recipient) site. The major advantage of translocation is that, if sufficient koalas are removed from the population, it provides immediate relief to preferred plant species from the impacts of 11

27 overbrowsing. The major disadvantage is that there is now a critical shortage of suitable recipient habitat to release koalas into. Further issues include the logistical complexity of the task, particularly when removing animals from an offshore island; the relatively high cost operations, and the potential for poor animal welfare outcomes and high mortality (Whisson et al., 2012). Due in particular to the lack of remaining suitable habitat, translocation is unlikely to remain a viable option for koala population management in the future (Menkhorst et al., 1998; Department of Sustainability and Environment, 2004; Menkhorst, 2008). Fertility control Surgical Sterilisation Surgical sterilisation of koalas involves vasectomy of the males and tubal ligation of the females (Duka and Masters, 2005). Another option would be to totally remove the gonads, however, at least in eutherians, this has the potential to impact on mating and ranging behaviour (e.g. Tuyttens and Macdonald, 1998; Jacob et al., 2004). The advantage of surgical sterilisation is that it is a proven technique and it is permanent, i.e. it provides fertility control for the life of the animal. The disadvantages are that it is relatively expensive compared to other fertility control methods, comparatively invasive to the animal, must be carried out by a trained veterinarian, and sterility cannot be reversed at a later date if required (Adderton Herbert, 2004). Immunocontraception This technique involves eliciting an immune response in an animal against its own eggs, sperm or hormones with vaccines containing self-antigens or their relatives, preventing successful reproduction. This has been trialled extensively in domestic animals and various ungulate species (Kirkpatrick et al., 2011). A technique using zona pellucidae (proteins which surround the egg) immunisation has caused failed ovulation in several marsupials; the tammar 12

28 wallaby (Macropus eugenii (porcine ZP); Kitchener et al., 2002) brushtail possum (possum ZP; Duckworth et al., 2007) and eastern grey kangaroos (Macropus giganteus (porcine ZP); Kitchener et al., 2009a). Preliminary research has been conducted on immunocontraception in koalas with limited success in controlling fertility (see Kitchener et al., 2009b). An advantage of immunocontraception is that it is less invasive than surgery, however there appear to be several fundamental problems that have not been addressed in research to date: vaccines may be ineffective due to the difficulty of obtaining a strong immune response to self-antigens; the use of an adjuvant to enhance the immune response can cause chronic suffering in the form of painful lesions; a substantial minority of the vaccinated animals may remain fertile, and, as this is most likely due to genetics, there would almost certainly be subsequent selection for these non-responders. There are little data which will allow estimates of the heritability of non-response to immunocontraception in any species on which this technique has been trialled (Magiafoglou et al., 2003), however Cooper and Larson (2006) argue that selection for non-responders will be rapid given the high percentage of nonresponders that typically occur within one generation. Even so, the technique could be useful in species with long generation times, as genetic changes in them, if they occur, would take decades and slow evolution of contraception resistance might therefore be anticipated (Magiafoglou et al., 2003; Cooper and Larsen, 2006). An additional issue that could result from breeding by non-responders is that the offspring from these treated animals are likely to have altered immune responses, with selection favouring a reduced immunological response(see Cooper and Larsen, 2006). In addition, to produce an effective immune response, vaccines are typically delivered as two doses, separated in time, which, if managing a freeranging population, is likely to be costly and inefficient. Further, for the control of large populations this may not be achievable in terms of logistics - it may be impossible to find and recapture the majority of individuals again for the second dose of immunocontraceptive within a suitable time frame. Techniques are improving however, for example one vaccine (using an 13

29 adjuvant) has been effective in white-tailed deer and horses for up to four years following a single dose (Fraker et al., 2002; Killian et al., 2006). Hormonal contraceptives Research in Australia into hormone-based contraceptives for marsupials has involved two types of contraceptive implants designed to provide contraception through disruption of the reproductive hormones: synthetic progestins and non-steroidal gonadotropin-releasing hormone (GnRH) agonists. Both types of implant are inserted into an animal subcutaneously and slowly release chemical to control fertility. Synthetic progestins, most commonly levonorgestrel (Norplant), act by reducing the concentration of luteinising hormone (LH) and follicle stimulating hormone (FSH), inhibiting follicular development and ovulation (Hynes et al., 2007). If ovulation does occur however, changes to the cervical mucous make it impenetrable to spermatozoa and implantation is inhibited through interference to the development of the endometrium (Brache et al., 1985). Levonorgestrel implants have been used with success in Australia to control fertility in eastern grey kangaroos (Nave et al., 2002; Wilson, 2012), tammar wallabies (Nave et al., 2000; Herbert et al., 2005) and koalas (Middleton et al., 2003; Hynes et al., 2010). The advantages of using levonorgestrel are its longevity: up to six years in macropods (e.g. Coulson et al., 2008) and over eight years in koalas (K. Handasyde, pers. comm), its low cost when compared to surgical sterilisation, its reversibility (Hynes et al., 2010), and in koalas, minimal animal welfare impacts (Hynes, 2007; Hynes et al., 2011). The disadvantages of using levonorgestrel are the need for minor surgery (under sedation or anaesthetic) to insert the implant due to its relatively large size; and the potential for side-effects, such as changes in ranging behaviour (Hynes et al., 2010) and metabolism (Sivin, 2003) and potentially pathology of the reproductive tract, which, while not recorded in marsupials, has been observed in felines treated with synthetic gestagens (Munson et al., 2002). 14

30 GnRH agonists are more potent analogues of the naturally produced peptide GnRH that is secreted by the hypothalamus the principle hormone governing reproductive function in males and females (Millar, 2005). GnRH stimulates the synthesis and release of FSH and LH from the pituitary, essential hormones for ovulation and spermatogenesis and the production of gonadal steroids (Conn and Crowley, 1991; Rispoli and Nett, 2005). GnRH release is normally pulsatile, however continuous administration results a down-regulation of pituitary GnRH receptors and a consequent decline in FSH and LH concentration, once again resulting in inhibition of follicular development and ovulation in females and spermatogenesis in males (Fraser, 1993; Herbert et al., 2004b). Deslorelin is a synthetic GnRH agonist that has been successfully used in a range of eutherian mammals including the domestic cat (Felis domestica), cheetah (Acinonyx jubatus), African wild dog (Lycaon pictus) and domestic dog (Canis familiaris) (Bertschinger et al., 2002; Munson et al., 2001; Trigg et al., 2001). Slow-release GnRH agonist (deslorelin) implants have also been tested on a range of marsupials, including limited tests on both captive and free-ranging koalas, with promising results (Herbert et al., 2001; Herbert, 2002; Herbert et al., 2005; Greenfield, 2012; Wilson, 2012). There are several advantages of using GnRH agonist implants: they are commercially available as a registered veterinary product (Suprelorin, Peptech Animal Health, Virbac (Australia) Pty Ltd, Sydney) which reduces costs substantially and they are administered by a simple subcutaneous injection without the need for sedation, as they are supplied loaded in a sterile trochar delivery system. There appear to be no significant undesirable side-effects following contraception (Herbert et al., 2005, Herbert et al., 2006), although ranging behaviour has been shown to increase in treated female koalas (Greenfield 2012). The disadvantage of this technique is the relatively short period of contraception (1-2 years in koalas; Greenfield, 2012); however with further research on dose rate effects, this may improve (Adderton Herbert, 2004). The comparatively brief duration of contraceptive effect may be considered advantageous in situations where it is desirable for individuals to return to the breeding 15

31 population, such as rotational captive breeding programs, or in smaller local wild koala populations where managers do not wish to risk driving the population to extinction. 1.5 Kangaroo Island Koalas were introduced to Kangaroo Island from 1923 to 1925 for conservation purposes. Eighteen adult koalas and an unknown number of young were translocated and released from French Island in Victoria (Robinson et al., 1989) and numbers rapidly increased (Masters et al., 2004). Possible explanations for this large increase in numbers are: a lack of ability to disperse from the island; the absence or lack of impact of the disease agent Chlamydia which causes variable levels off infertility in some mainland populations (McLean, 2003); a lack of predators such as the dingo (Martin and Handasyde, 1999); and the widespread availability of suitable food trees, including the Kangaroo Island endemic E. viminalis cygnetensis (Masters et al., 2004). In 1994 the South Australian Department for Environment and Heritage estimated that 50% of the canopy of most individual E. v. cygnetensis was defoliated due to koala browsing pressure (Masters et al., 2004). Urgent management was required to preserve the now threatened stands of remaining E. v. cygnetensis and in 1997 the South Australian Government implemented a control program aimed at reducing the density of the koala population on the island (Duka and Masters, 2005). Adult male and female koalas were caught and surgically sterilised, and a number of these were translocated to other suitable habitat on mainland southeastern South Australia. During this original program it became clear that the koala population was much larger and wide-spread than previously thought, and in 2002 the population was estimated at approximately 27,000 animals (Masters et al., 2004). As a result, the management program was expanded with the aim of sterilising 2000 koalas per year, with a proportion of these still translocated to the south-east of the state. Between 1997 and 2012, 10,631 koalas 16

32 were caught and sterilised, of which 3,801 koalas were translocated back to their historic range on the mainland (Molsher, 2012). These management actions are having an impact on koala numbers and the current population estimate is <14,000 animals (Molsher, 2012). This study There is no single management technique that will satisfy all goals and concerns for all population management scenarios, as each situation has its own unique circumstances and pressures. Each situation requires thorough assessment, followed by implementation of the most appropriate control method in terms of feasibility, resource availability, public and political acceptability, cost-effectiveness, humaneness, and potential for side-effects. In some situations the most appropriate strategy might be the application of a variety of management techniques, combining for an effective total result. As such, a continuing search for new and improved fertility control techniques, including an assessment of the feasibility of delivery and the efficacy of treatment, is also needed. This study investigates the potential of using GnRH agonist implants to control the breeding rate of an overabundant koala population. Preliminary research on captive koalas produced promising results with no side-effects (Herbert, 2002); however knowledge of the efficacy of using these implants to control fertility and breeding rate in a free-ranging koala population under variable conditions is limited. Greenfield (2012) conducted a trial of GnRH agonist implants in a very high density population on French Island (Victoria), and demonstrated a high level of contraceptive effect for a period of 1-2 years. There is a need however to understand the efficacy and impact of this treatment in a range of koala populations as these can be highly variable for factors such as habitat quality, population density and reproductive characteristics. This study aims to establish whether deslorelin implants are useful as an effective, humane and efficient method for large-scale koala population control on Kangaroo Island, South Australia, providing an alternative technique to the more invasive 17

33 surgical sterilisation currently used by wildlife management authorities in that state. This study also aims to discuss hormone-induced infertility in context with alternative fertility control options for management of the Kangaroo Island koala population. To achieve these aims the following studies were undertaken: 1. A summary of the reproductive characteristics of female koalas on Kangaroo Island (Chapter 3). An understanding of reproductive characteristics and how they may vary with factors like female age and environmental variables is important to any management strategy. This knowledge informs models of population growth that assist in determining management effectiveness, however decisions and planning regarding management of koalas on Kangaroo Island have historically been based, in part, on demographic parameters derived from Victorian koala populations which may be influenced by disparate environmental conditions. To provide more appropriate information to managers, this study aims to use the large dataset generated by the ongoing koala management program on Kangaroo Island to describe a range of female breeding characteristics specific to this population. 2. An investigation into the efficacy of deslorelin treatment for controlling fertility in freeranging female koalas (Chapter 4). Treatment of female koalas with the GnRH agonist deslorelin has the potential to provide a more cost-effective and less invasive option for than the available alternatives for fertility control of an overabundant koala population. This study aims to demonstrate the ability of deslorelin treatment to inhibit fertility in free-ranging female koalas, including an evaluation of the hormonal response of treated koalas subjected to a GnRH challenge. Based on limited prior studies of captive and freeranging female koalas, it is expected that deslorelin treatment of koalas on Kangaroo Island will result in high levels of efficacy. 18

34 3. An assessment of the effects of deslorelin treatment on the health and condition of freeranging female koalas (Chapter 5). Fertility control is proposed as a more humane alternative to lethal methods of wildlife population control. Using repeat captures of individuals, this study aims to determine the effect of deslorelin treatment on the body weight and muscle condition of free-ranging adult female koalas on Kangaroo Island over time, compared to control koalas and koalas surgically sterilised under current state government management practice. It is anticipated that treatment with deslorelin will have little impact on koala health and condition and hypothesised that female koalas treated with deslorelin, freed from the dual energetic constraints of undergoing oestrus and supporting young, will in fact exhibit enduring increased body weight and condition compared to control and surgically sterilised animals. 4. An investigation of the effects of deslorelin treatment and surgical sterilisation on the movement patterns and ranging behaviour of free-ranging female koalas (Chapter 6). Behavioural changes caused by the application of a fertility control technique may have important implications for population management or animal welfare. Because adult female koalas naturally show high site-fidelity, a technique that increases the probability that contracepted koalas remain in situ following treatment (i.e. little change in natural behaviour) would enhance the viability and general acceptance of this technique as an ethical management option. Using radio-collared koalas, this study aims to compare the movement patterns and ranging behaviour of adult females subjected to deslorelin treatment to those surgically sterilised under current management practice. Due to the suggested influence of oestrus on female koala spatial behaviour, it is predicted that the movement rate, distance travelled and area of core home-range of female koalas treated with deslorelin will be reduced compared to control animals, while the occurrence of long-distance movements are likely to be greater in surgically sterilised females. 19

35 5. An evaluation of the impacts of translocation on koalas (Chapter 7). Translocation is one of just two government-sanctioned techniques available for the management of overabundant koala populations, however studies of the fate of translocated individuals are few. An assessment of post-release koala densities and the survival and movement of koalas translocated from Kangaroo Island to the South Australian mainland under current management practice is presented here, along with examination of the cumulative effects of surgical sterilisation and immediate translocation. 6. A review of management options for control of the Kangaroo Island koala population (Chapter 8). Drawing on results obtained from the studies outlined above and the wider literature, this paper aims to provide a comprehensive review of available control options in the context of the Kangaroo Island koala population and the local environment. Relative efficacy, cost-effectiveness and feasibility of techniques, and considerations regarding animal welfare and behavioural changes, are discussed and a series of management recommendations are presented. Kangaroo Island was chosen as the study location for a number of reasons: The island supports a large and overabundant koala population with an identified need for management; The strong interest and support from the responsible agency for assessing the potential of new fertility control techniques to provide options for ongoing koala population management; The ongoing koala management program on Kangaroo Island offered information and a large data set resulting from the use of two traditional management techniques (surgical sterilisation and translocation). This allowed direct comparison of the feasibility and effectiveness of GnRH agonist treatment with established techniques, whilst providing new and updated information on the effects of these techniques on individual animals. 20

36 The island still had large areas of habitat supporting high densities of koalas that had not been subjected to previous population management; i.e. for relevant components of the study there is little risk that previous or ongoing management activities will influence or confound study results. A concurrent and complimentary study assessing the efficacy of GnRH agonist contraception in free-ranging koalas was undertaken in Victoria; given the spatial variability in koala reproductive characteristics (McLean et al., 2006; Whisson and Carlyon, 2010) it is essential to trial the effectiveness and feasibility of new techniques in multiple populations and in differing environmental conditions. 21

37 22

38 23

39 2. Materials and Methods This chapter provides detailed description of the study materials and methods that are common to at least two chapters. Details of experimental design and methods specific to each component of the study are given in the relevant chapters. 2.1 Study area Fieldwork for this study was conducted on Kangaroo Island (35º45 S, 137º37 E), situated approximately 12 km from the coast of South Australia, 100 km south-west of Adelaide. The island covers an area of almost 438,000 ha and approximately 47% of the native vegetation still remains (Ball and Carruthers, 1998), most of which is located towards the western end. Approximately 55% of native vegetation is conserved in National and Conservation Parks (Figure 2.1). Of all native vegetation, 35.8% (73,724 ha) is habitat considered suitable for koalas, but only 1% (750 ha) of this classed as high quality habitat (as per Masters et al., (2004), where habitat quality is based on composition of eucalypt species). Medium quality habitat makes up 17.5% (12,909 ha) of suitable koala habitat while 81.5% (60,065 ha) of koala habitat is considered low quality (Masters et al., 2004). The majority of high and medium quality habitat is confined to riparian zones. The habitat is typically characterised by linear strips of woodland vegetation dominated by roughbarked manna gum (Eucalyptus viminalis cygnetensis), South Australian blue gum (E. leucoxylon) and/or pink gum (E. fasciculosa) along the permanent watercourses, and woodland patches that may contain a mix of brown stringy-bark (E. baxteri), messmate stringy-bark (E. obliqua), cup gum (E. cosmophylla), and/or pink gum under an overstory of sugar gum (E. cladocalyx) at the upper reaches of more ephemeral creeks. Isolated or small patches of 24

40 eucalypts occur throughout the surrounding pasture, which is predominately used for grazing stock. Annual average rainfall for the island varies from mm in the east (Penneshaw) to mm in the west (Rocky River), with a peak during winter (BOM, 2008). Temperatures vary from a mean monthly minimum of 8.3ºC in August to a mean maximum of 23.7ºC in January (BOM, 2008; Figure 2.2). Elevation varies between 50 m above sea level in the east and south to 250 m in the west and north (Twidale and Bourne, 2002), and the substrate is dominated by sandy and ironstone gravelly soils (Northcote, 2002). Figure 2.1: Kangaroo Island, South Australia, showing study sites. Shaded area represents reserved land. 25

41 Temperature (C) Rainfall (mm) Mean max temp Mean min temp Mean rainfall Figure 2.2: Mean monthly ( ) minimum and maximum temperatures (ºC) and rainfall (mm) for Kingscote, Kangaroo Island (source: BOM, 2008). The island experienced severe drought in 2006, while below-average rainfall was also recorded in 2007 and 2008 (Fig. 2.3; data from Pioneer Bend has been presented here as the majority of this study was conducted in this area). Rainfall leading up to and during the peak period of the 2006/2007 koala breeding season was just 41% of the historical median (BOM, 2009a, 2009b). Mean monthly maximum temperatures were the highest on record in September and October 2006 (19.8 C, 22.3 C respectively), December 2007 (27.1 C) and March 2008 (28.3 C) (BOM 2009a, 2009b). Mean annual maximum temperatures were also above the average ( ) in all years of this study, with the 2007 mean (22.0 C) the highest on record (BOM, 2013). 26

42 Annual Rainfall (mm) Year Figure 2.3: Annual rainfall at Pioneer Bend, Kangaroo Island Black bars represent years during which this study occurred. The solid line indicates the mean annual rainfall ( ). Data for years 2000, 2001 and 2003 is from Parndana (BOM, 2009a) and other years from Pioneer Bend (BOM, 2009b). There is only 1km between these stations. 2.2 Koala capture and handling All capture, handling and processing of koalas was completed in the field, apart from the surgical sterilisation of animals for the radio-tracking study (see section 2.8). For animal welfare reasons, catching and handling of koalas was not carried out if the ambient temperature exceeded 30 ºC or during heavy rain. A standard noose and flag technique was used to capture koalas (after Martin, 1989). Depending on tree height, an experienced climber was used to gain access to the animal. Using an extendable pole, a rope noose (fitted with a stop knot so that it could not pull too tight around the animal s neck) was placed over the animal s head. Once noosed, a coloured flag was waved above the animal s head to encourage it to back down to the base of the tree. Once on the ground, koalas were immediately placed in a hessian sack for handling. Any juveniles 27

43 (either on the back or in the same tree as a female) were also caught and held with their mother. 2.3 Collection of morphometric and reproductive data, and fitting of radio-collars Study animals were selected as described in each chapter. At each capture, animals were weighed to the nearest 0.1 kg using 25 kg spring scales (Wedderburn CA150 25K). The head length of all captured koalas was measured to the nearest millimetre using 200 mm digital vernier callipers (Kincrome Australia). Head length was measured dorsally from the occipital bone at the base of the skull to the tip of the nose and the average of two measurements used. A thorough examination of each animals was conducted to determine body condition (muscle, body and coat), presence of ecto-parasites, injuries and anomalies (e.g. scoliosis, mandibular malalignment, dislocations) were all recorded. The muscle condition of individuals was assessed by palpating the muscle along the caudal edge of the scapular spine and assigning it to one of three classes based on its shape and height relative to the scapular spine. General body condition was assessed on a scale of 1-3 (with 3 representing the best condition) based on a measure of the soft tissue cover over the dorsal spinous processes of the vertebrae and the kidney area. Tooth wear was visually assessed by gently opening the mouth using a padded tooth-bar while the koala was restrained. The age class was determined using the degree of wear on both upper pre-molars and koalas were assigned to one of nine tooth wear classes (TWCs, Table 2.1) (see McLean, 2003). If the TWC differed between the right and left premolars by one age class, the higher TWC was used, while differences greater than one TWC were averaged across the two pre-molars. All tooth wear and condition assessments were conducted by the author or, for surgically sterilised animals, by trained veterinary staff. The pouch of female koalas was checked for the presence of young or lactating teats at the time of capture. Any young that were fully furred, not attached to the teat and judged to be 28

44 over 400 grams were removed from the pouch, weighed to the nearest 5 g (Salter Super Samson 1kg scales) and had their head-length recorded. Pouch-young were then returned to the pouch, while back-young were placed onto the females back as she was released. All adult koalas, and young 2 kg, were tagged in both ears (1 in the left ear, 2 in the right) with plastic sheep swivel tags (Leader Products, Melbourne) bearing a unique identification number. Tags were colour-coded to allow identification of individuals from a distance. All adult koalas, and young over 400 g, were fitted with a Passive Integrated Transponder (PIT) tag (Allflex Australia) inserted subcutaneously between the shoulder blades. Table 2.1: Scale for converting tooth wear class (TWC) to an estimate of age (years), based on data from known-age Victorian koalas (from McLean, 2003). TWC I II III IVa IVb IVc V VI VII Age > GnRH agonist contraceptive implant Commercially available implants (Suprelorin, Peptech Animal Health, Virbac (Australia) Pty Ltd, Sydney) containing 4.7 mg of the GnRH agonist deslorelin (D-Trp 6 -Pro 9 - des-gly 10 -GnRH ethylamide) were used to contracept female koalas. The implants are manufactured by a proprietary method involving extrusion of deslorelin with a matrix of lowmelting point lipids and biological surfactant (Trigg et al., 2001). Each implant was 12.5 mm long and 2.3 mm in diameter (Fig. 2.4). The in vivo release rate of these implants in koalas is unknown; however, in a real time dissolution system, Trigg et al. (2001) reported a release rate of deslorelin of > 1 µg/day for periods of > 1 year. 29

45 Implants were administered to conscious koalas and inserted subcutaneously between the shoulder-blades (on the opposite side to the PIT tag) using a pre-loaded sterile applicator specific to these implants. The implant site was then sealed with a small amount of veterinary tissue adhesive (Vetbond, 3M Animal Care Products, USA). Implants were palpated following administration to ensure their position under the skin. Placebo implants (Peptech Animal Health Pty Ltd, Virbac (Australia) Pty Ltd), identical to Suprelorin but with the active ingredient deslorelin excluded during formulation, were used as a control for the implantation process in the telemetry study. Delivery of placebo implants was the same as for Suprelorin. Figure 2.4: Slow-release GnRH agonist implant (Suprelorin ). Implant (centre) has been removed from the pre-loaded applicator (top), placed next to an Australian 5-cent coin for scale. 30

46 2.5 Blood sampling Blood samples were taken a priori from all adult koalas and young 2 kg for inclusion in a serum bank. Blood was collected from the cephalic vein after clipping a small amount of hair from the forearm and cleaning the site with 70% ethanol: 5 ml from adult koalas and 3 ml from their young. Samples were taken while koalas were restrained in a bag by a second person and a tourniquet placed around the koala s arm as high as possible above the elbow joint. Blood was collected using a 23-gauge hypodermic needle (Becton Dickinson, USA) and a 5 ml eccentric tip syringe (Codan Medical ApS, Denmark). Samples were immediately transferred to blood collection tubes containing a clotting activator (Vacuette serum separation tubes, Greiner Bio-one, Austria) and kept on ice while in the field. All samples were separated via centrifugation at 3000 rpm for 15 minutes within six hours of sampling. Serum was stored at -20ºC. GnRH challenge For the GnRH challenge, blood sampling was as described above except that a 22- gauge intravenous catheter (Terumo Corporation, USA) was inserted into the cephalic vein. The needle of the catheter set was removed, and the catheter held in place using a self-adhesive bandage (VetWrap, 3M Animal Care Products, USA). Blood samples (4 ml) were collected at the following time intervals: -15, 0, 15, 30 and 60 minutes, relative to an intravenous injection of 2 ug/kg synthetic GnRH (Fertagyl, Intervet (Aust.) Pty. Ltd., Castle Hill) in sterile 0.9% saline, injected through the catheter after collection of the second (0 min) blood sample. The catheter was filled with heparanised saline (250 I.U/ml, in sterile 0.9% NaCl) in between blood samples to keep the catheter patent, which was removed and discarded before collecting each subsequent sample. Koala were held in hessian sack in the shade throughout the full timeframe of the GnRH challenge. Samples were kept on ice while in the field and serum separated as 31

47 described above within six hours of sampling, then stored at -20ºC prior to determination of hormone concentrations by enzyme-immunoassay (EIA). 2.6 Hormone assays Serum LH concentrations were determined using an enzymeimmunoassay (EIA) procedure at the Wildlife Reproduction Centre, Taronga Western Plains Zoo, Dubbo, New South Wales. This assay employs a biotinylated ovine LH label, NIH bovine LH standards (A.F. Parlow; NIDDK National Hormone and Pituitary Program) and a monoclonal antibovine LH antibody (#518-B7, supplied by J. Roser University of California, Davis, CA) (after Graham et al., 2002). This LH assay has been used to determined serum LH concentrations in a range of exotic animals, including elephants (Graham et al., 2002), western grey kangaroos and black-flanked rock wallabies (Matson et al., 2009). Antibodies and conjugates were sourced from Coralie Munro (UC Davis USA) and all chemicals from Sigma-Aldrich (USA) and Merck (Germany). The assay was validated for use in koalas by demonstrating parallelism of serial dilutions of koala plasma to the standard curve. All samples (50μl) were assayed in duplicate according to the methods described by Graham et al. (2002). The assay sensitivity was 0.16ng/ml, and the inter- and intra- assay coefficients of variation were 2.6 ± 0.2 ng/ml inter CV = 7.59%, 0.32 ± 0.03 pg/ml inter CV = 8.41% and >0.24pg/ml intra CV = 7.7 ± 7.5%. 2.7 Surgical sterilisation The tubal ligation procedure was conducted by registered veterinarians (Kangaroo Island Veterinary Clinic) as part of the Kangaroo Island Koala Management Program. Koalas were anaesthetised with 5% isoflurane/oxygen mix administered via a face mask and open T- piece connected to an anaesthetic machine with an active vaporiser. A sterile Veress needle, 32

48 connected to a CO 2 insuflator, was placed through the abdominal wall approximately midway between the prepubic bones and the xiphoid in the midline and the abdominal pressure was increased to 12 mm Hg. A number 15 scalpel blade was then used to make a stab incision into the abdominal cavity approximately 1-2 cm cranial to each prepubic bone prominance. A 3.7 mm x 18 cm 30 rigid endoscope was then introduced into the abdomen through the incision on the left side and the light cable connected. Once the bladder was visualised and seen to be clear of the mid-abdominal wall, a stab incision was made in the midline slightly cranial to a line between the two initial incisions using a number 24 scalpel blade. A pair of Allis forceps was introduced through this incision up to the box joints to maintain insuflation of the abdomen. The left uterine horn was visualised with the endoscope and grasped with the Allis forceps just distal to the utero-tubal junction. An endoscopic monopolar cautery scissor was then introduced in to the right abdominal incision and positioned near the uterus. The Allis forceps were then used to manoeuvre the utero-tubal region away from adjacent structures and provide tension of the utero-tubal and tubal region. The utero-tubal area was grasped with the monopolar scissor and cautery applied at a setting of between 40 and 80 (Valleylab Force 300 electrocautery machine) depending on the thickness of the tissue. Current was then applied until obvious blanching of the tissue either side of the scissor was seen, and the utero-tubal area was transected in the cauterised area with the scissor. The scissor was then used to make further cuts through the almost transparent broad ligament of the uterus to achieve good separation between the uterine horn and ovarian tube. The process was repeated on the right side of the uterus and all instruments withdrawn from the abdomen. The stab incisions in the abdominal wall were closed using simple interrupted sutures of 0 Vicryl once the koala was returned to horizontal. The wounds were cleaned and sprayed with a fly repellent and a long acting amoxycillin injection was administered before the anaesthetic mask was withdrawn. 33

49 Following surgery, each koala was moved to a quiet dark room and placed into a large plastic pet-pack (Ruddocks Pty Ltd, Australia) to recover whilst lying stretched on its belly. Any young were reunited with their sterilised mother approximately half an hour post-surgery. Once fully recovered, sterilised koalas were transported back to the capture site in their petpack and released into the tree in which they were caught. 2.8 Monitoring koala movements Koalas were fitted with a standard two stage VHF transmitter sealed in epoxy resin attached to a 400 X 20 mm haultain collar fastened with two brass bolts (Sirtrack Ltd, New Zealand). Each transmitter included a 300 mm whip antenna and had a stated battery life of 53 months. Collars were fitted so that two fingers could fit comfortably under the band and the transmitter positioned under the chin. Transmitter frequencies were in the MHz range and frequencies spaced in 20 KHz steps, and emitted at a pulse rate of 40 ppm. Collars weighed 100 grams or approximately 1.3% of the body weight of an average adult Kangaroo Island female koala (7.6 kg; DEWNR unpublished data). Reflective tape in a colour combination identical to the koala s ear-tags was added to each collar to aid identification of individual koalas from a distance. Radio-tracking and recapture for monitoring Tracking was carried out with the use of a scanning hand-held telemetry receiver (R1000, Communications Specialists Inc., California) and a hand-held folding 3-element yagi antenna (Sirtrack Ltd, New Zealand). Koalas were tracked on foot from their previous location directly to a visual sighting during daylight hours, using binoculars (8 X 40, Nikon Australia Pty. Ltd., Sydney) to confirm identity, after permission was granted from landholders to track on their land. A koala s location ( fix ) was recorded at each sighting using a hand-held global positioning system (GPS) (Garmin Ltd, UK) once an accuracy of position of < ±10 m was 34

50 achieved. Positions were recorded as Australian Map Grid coordinates (MGA94). The tree species that each koala occupied was noted and the tree assigned a unique number to identify any repeat usage of individual trees. Information about tree diameter and height, dominant eucalypt species within 50 m, height and position of the koala in the tree, behaviour and condition of the koala and presence and location of young was recorded at each sighting. Tracking was undertaken by the author until June 2008 (mean 10.1 ± 0.2 days between successive radio fixes) after which it was taken over by DEWNR staff and the frequency of tracking reduced. Data on movement patterns after June 2008 was not used. All radio-collared koalas were periodically re-caught during the course of the study, at the beginning and at the conclusion of each breeding season. Koalas were caught and handled using the methods outlined in section 2.2. Information on weight, tooth-wear, health and condition and the presence of young was recorded as described in section 2.3. Collars were checked for fit and adjusted if necessary, and the animals neck examined for any rubbing caused by the collars. Microchip function was tested and presence of ear-tags noted. Ear-tags were not replaced if missing as there was generally no difficulty observing the presence of a collar. Modification of collars after koala mortality Unfortunately, two koalas (one placebo-treated and one surgically sterilised) were found dead, hanging from their collars, in December 2008 and February 2009 respectively. These koalas had been wearing their collars for a continuous period of 26 and 28 months respectively prior to these events. Following the discovery of the second koala, all remaining collared koalas were re-caught and their collars fitted with a rubber break-away section to prevent any further incidents (WEC approval 43/2007-M3). Another placebo-treated koala was found dead at the base of a tree in May 2009, cause of death unknown. As these koalas died after a change in monitoring protocol, their deaths did not affect analyses of movement 35

51 patterns, however they were naturally removed from all analyses of fecundity affected by their loss. All deaths were reported to the DEH Wildlife Ethics Committee. 2.9 Data analysis Data were analysed using the statistical software R (R Core Team, 2013) or SPSS 15.0 for Windows (SPSS Inc. 2006), except for contingency tables which were analysed in Microsoft Excel for Windows (Microsoft Corporation 2003). Results were taken as significant at p< Koala home-ranges were examined using the animal tracking and location analysis software Ranges6 v1.2 (Kenward et al., 2003). Maps were produced using ArcView 3.3 for Windows (ESRI, 2002) and Manifold System 8.0 (Manifold.net, 2010) utilising shapefiles and raster data provided by the South Australian Department for Environment and Heritage Authorisation Field procedures specific to this project involving live koalas were approved by the South Australia DEH Wildlife Ethics Committee (approvals 45/2005, 43/2007) and meet guidelines recommended by the American Society of Mammalogists (Gannon et al., 2007). Translocation studies were conducted under AEC approvals 34/2004 and 11/2007-M1. Animal capture and sample collection was conducted under Scientific Research Permit E25038, E24926 and M25454 and Licence for Teaching, Research or Experimentation Involving Animals

52 37

53 3. Female reproductive characteristics of an introduced and abundant island population of koalas * We seldom know enough about the dynamics of a wild population for either research or management needs. (Eberhardt, 1985) 3.1 Introduction The demographic parameters of a wildlife population can be influenced profoundly by temporal and spatial variation in environmental conditions (Kareiva, 1990). Because reproductive parameters are especially sensitive to environmental variation, this can have significant implications for population dynamics and persistence and growth of populations over time and in different locations. In mammals, reproductive output and thus the growth rate of a population is directly affected by or correlated with a host of intrinsic factors including age (Gilchrist et al., 2004), health (Soulsbury et al., 2007), body condition (Pitcher et al., 1998), size (Hodge et al., 2008), previous breeding experience (Broussard et al., 2008), adult sex ratio (Solberg et al., 2002), and social status (von Holst et al., 2002). Factors including rainfall (Hellgren et al., 1995), food and resource availability (Dennis and Marsh, 1997) or quality (Higginbottom, 2000), population density and inter- and intraspecific competition (Clutton-Brock et al., 1987; Watts and Holekamp, 2008), day length (Malpaux et al., 1999), disturbance (Shively et al., 2005), and predation risk (Creel et al., 2009) can directly influence reproductive timing and success in individuals and populations or exacerbate the effects of other factors. * This chapter results from a published peer-reviewed manuscript (Whisson and Carlyon, 2010) see page iv. 38

54 An understanding of the population dynamics of the koala (Phascolarctos cinereus), particularly the factors affecting reproduction in this species, is critical for effective long-term management and conservation of the species (Natural Resource Management Ministerial Council, 2009). A decline in koala populations in the early 1900s resulted in the translocation and establishment of island populations including on French Island, Victoria (Martin and Handasyde, 1999) and Kangaroo Island, South Australia (Masters et al., 2004). The Kangaroo Island population, like many others in southern Australia, is now considered overabundant, with high densities of koalas causing defoliation of their preferred food trees and thus threatening their own habitat (Masters et al., 2004). Because the Kangaroo Island population has been derived from limited genetic stock (18 koalas from French Island, Victoria; Robinson et al., 1989), it has low genetic diversity, which raises concerns about inbreeding effects and the conservation value of the population (Seymour et al., 2001; Cristescu et al., 2009;). Management decisions on Kangaroo Island have been based on population parameters derived from studies of Victorian populations (e.g. Martin, 1985b; Martin and Handasyde, 1990a; Mitchell and Martin, 1990). Spatial variation in these parameters (McLean and Handasyde 2006), however, means that they might not be reliable in predicting growth of the Kangaroo Island population. Koalas are seasonal breeders, with births generally occurring between October and May (White and Kunst, 1990; McLean and Handasyde, 2006) and a peak between December and March (Ellis et al., 2010). Female koalas begin breeding at two or three years of age and typically produce one young per year (Eberhard, 1972; Smith, 1979; Martin, 1981; White and Kunst, 1990; McLean and Handasyde, 2006). Females in good condition are able to breed in consecutive years (Eberhard, 1972; McLean and Handasyde, 2006). In Victorian populations females begin breeding at around 2.5 years of age, corresponding to a body mass of around 6 kg (Handasyde, 1986; Martin and Handasyde, 1990a; McLean and Handasyde, 2006). Fecundity generally declines after 10 years of age (Martin and Handasyde, 1990a). 39

55 Spatial and temporal variation in reproduction of southern koala populations has been documented (Martin and Handasyde, 1990a; McLean and Handasyde, 2006). The most significant factor influencing fecundity is incidence of chlamydiosis (i.e. infection of the reproductive tract with Chlamydia bacteria, principally C. pecorum or C. pneumoniae (Martin and Handasyde, 1990a; Polkinghorne et al., 2013). In Chlamydia-infected populations fecundity is between 0% and 56% compared to between 66% and 81% in Chlamydia-free populations (Martin and Handasyde, 1990a; McLean and Handasyde, 2006). It has recently been proposed that incidence of chlamydiosis may be associated with incidence of koala retrovirus (KoRV), a disease known to induce immunodeficiency in infected individuals and prevalent in many free-ranging koala populations, particularly in the north of their distribution (Denner and Young, 2013; Meers et al., 2014). KoRV has only recently been detected in the Kangaroo Island koala population, however it currently appears to be geographically isolated to the central region of the island (Meers et al., 2014). Variation in timing of breeding between sites has been attributed to differences in climate and its influence on energy allocation to reproductive activities (McLean and Handasyde, 2006). In Victoria, McLean and Handasyde (2006) observed breeding to begin earlier and extend over a longer period in island populations than at mainland sites. They suggested that under the milder weather conditions that are typical of Victorian offshore islands, koalas might require less energy for thermoregulation and could instead allocate energy to reproduction. Other environmental factors, such as the abundance, diversity, and health of preferred food trees and soil type, also have the potential to affect reproductive parameters through their effect on food resources (McLean and Handasyde, 2006); however, Martin and Handasyde (1990) suggested that these factors are more likely to affect survival of young than fecundity. Temporal variation in koala reproduction also occurs but is not well understood. McLean and Handasyde (2006) observed a longer breeding season on French Island in

56 and 2001 than was found in previous studies (McNally, 1957; Martin and Handasyde, 1990a) but were unable to explain the cause. Other reproductive parameters, including the sex ratio of offspring and mass at sexual maturity, were consistent across years (McLean and Handasyde, 2006). On Kangaroo Island the capture and examination of thousands of koalas from 1997 to 2007, as part of a management program to reduce koala densities and their impacts, provides a rare opportunity to examine reproductive characteristics of an introduced and overabundant koala population. We documented the timing of the breeding season, fecundity, and maternal characteristics, examined their temporal variation with respect to environmental conditions and population density, and compared them with those of other southern koala populations. 41

57 3.2 Materials and Methods Study area Kangaroo Island is situated approximately 12 km from the coast of South Australia, 100 km south of Adelaide (Fig. 3.1). It covers an area of approximately 438,000 ha and has approximately 47% of native vegetation still remaining (Ball and Carruthers, 1998). Approximately 55% of native vegetation is conserved in national and conservation parks. Annual average rainfall for the island varies from mm in the east to mm in the west, and temperatures are moderate, varying from a mean monthly minimum of 8.3 C in August to a mean monthly maximum of 23.7 C in January (Bureau of Meteorology; Figure 3.1: Kangaroo Island showing major catchments and areas of high- and medium-quality habitat preferred by koalas (modified from Masters et al., 2004). 42

58 Kangaroo Island koalas and their management Koalas were introduced to Kangaroo Island in the 1920s. Between 1923 and 1925, 18 koalas and an unknown number of young from French Island, Victoria (a Chlamydia-free population), were introduced into Flinders Chase National Park at the west end of the island (Robinson et al., 1989). Surveys in 2001 determined that koalas were distributed widely across the island and at high densities in habitats dominated by their preferred food trees (Masters et al., 2004). At that time the koala population was estimated at 27,053 koalas ± 2,830 SE (Masters et al., 2004). In 2006 this estimate (based on the same sites and methods used in 2001) was revised to 15,993 ± 3,173 koalas (R. Molsher, Department for Environment and Heritage, South Australia, Australia, pers. comm.). The Kangaroo Island Koala Management Program was implemented in January 1997 with the goal of reducing koala densities and their impacts on their preferred food trees. The program, which is ongoing, is based on the surgical sterilisation of koalas, with translocation of a proportion of those to mainland South Australia (Duka and Masters, 2005). Annual catching periods generally have coincided with the koala breeding season; however, the location of management, duration of the catching period, numbers caught, and sex targeted for capture have varied annually depending on management resources and objectives. Following an intensive effort to catch koalas from January 1997 to June 1998, the program entered a phase of reduced effort until June 2005, with catching periods each year of up to four months duration. The program was then expanded from July 2005 to June 2007 with catching occurring from November to May each year (R. Molsher, Department for Environment and Heritage, South Australia, Australia, pers. comm.). Most koalas have been caught from habitat dominated by preferred food trees, including rough-barked manna gum (Eucalyptus viminalis cygnetensis), South Australian blue gum (E. leucoxylon), and red gum (E. camaldulensis), in the Cygnet River Catchment and 43

59 Flinders Chase National Park (high- and medium-quality habitats Masters et al., 2004; Fig. 3.1). Annual monitoring suggests that koala population density in preferred habitats has declined from 3.1 koalas/ha ± 0.8 SE, and a maximum of 13 koalas/ha in 1996 (prior to management), to 1.0 ± 0.2 koalas/ha, with a maximum of 3 koalas/ha at a site, in 2006 and 2007 (R. Molsher, Department for Environment and Heritage, South Australia, Australia, pers. comm.). Koala capture and processing An extendable pole (to 7 m) with a hook was used to drop a loop of rope (fitted with a stop knot to ensure it could not pull too tight around the neck) over the head of a subject. Next, a flag was waved above the koala to encourage it to back down to the base of the tree. This process often involved a trained climber first climbing the tree to get closer to the koala. Any juveniles (either on the back or in the same tree as a female) also were caught. Koalas were placed immediately in a hessian sack for handling and were ear-tagged (Allflex sheep tag; Allflex Australia, Capalaba, Queensland, Australia) with a unique number. All koalas sterilised also were microchipped (Destron LifeChip; Digivet. com, Baulkham Hills, New South Wales, Australia) at the time of surgery. Koalas were held for a maximum of two nights prior to surgery and either released at the site of capture or translocated to the mainland following surgery. For each koala (including dependent young), sex, date and time, and location (recorded using a handheld global positioning system [Garmin International, Inc., Olathe, Kansas]) were recorded at the time of capture. Body mass, age class, head length, reproductive status (presence of young and their development), and condition were recorded by trained veterinary staff when koalas were sterilised. Age was determined by assessing the degree of wear on the upper premolars and assigning the koala to one of seven tooth-wear classes (after McLean, 2003; see Chapter 2). The muscle condition of individuals was assessed by palpating the 44

60 muscle along the posterior edge of the scapula and classing it as poor, average, or good based on its shape and height relative to the scapula spine (McLean, 2003). Independent koalas were those caught in the absence of a lactating female. Breeding female koalas were defined as those that were lactating. Data Analysis Catching periods in , , and were the longest and encompassed the November May period. For comparative purposes analysis was restricted to koalas caught in these years. In these years 412, 1,146, and 1,357 independent female koalas, respectively, were caught. Chi-square goodness-of-fit tests were used to compare the percentage of females in age classes >III breeding each year, percentage of females < 6 kg and 6 kg breeding each year, distribution of breeding females between condition classes, and sex ratio of offspring within a breeding season and between years. Birth dates were estimated for 1,824 dependent young with head length (H) 125 mm using a relationship between head length and age developed from Victorian koalas (McLean, 2003) for males: ln 1 H /158.7 Age and females: ln 1 H /135.8 Age The earliest estimated birth date was 19 September and the number of days between this date and the birth date of each young was calculated. These data were analysed with 1-45

61 way analysis of variance (ANOVA). Variation in mass of breeding females over the breeding period was analysed with a 2-way ANOVA, with month and sex of the offspring included as factors. Tukey s post hoc test was used to identify homogeneous subsets. Prior to analyses, data were visually inspected to ensure they approximated normality and Levene s test was used to test for homogeneity of variances. Analyses were undertaken in SPSS for Windows 17.0 (SPSS Inc., 2006). 46

62 3.3 Results Timing of births The earliest recorded birth occurred on 19 September and the latest on 30 July. The timing of births (defined here as the period from the first to last recorded birth) did not vary between years (F 2,968 = 0.38, P = 0.68). From pooled data for the three seasons of management 7% of births occurred in November, with most (76%) births occurring between December and February and a peak occurring in January (Fig. 3.2). Figure 3.2: Cumulative percent of male (solid line) and female (dotted line) births of koalas from November to May (pooled data for , , and ). Young sex ratio The ratio of male to female births varied significantly over the November May period (x 2 3 = 8.21, P = 0.001), with 65% of November and December births being male (Fig. 3.2). The overall offspring sex ratio was close to unity in (48% male; x 2 = 0.02, P = 0.89) and in (54% male; x 2 1 = 2.42, P = 0.12) but was male-biased in (55% male; x 2 1 = 4.66, P = 0.03). 47

63 Maternal mass The mean mass of an adult female (i.e. female age class III) on Kangaroo Island was 7.6 ± 0.5 kg (n=747). Between 52 and 67% of adult females were observed with young, depending on season (x 2 2 = 7.75, P = 0.021). The smallest mass of a female with young was 3.9 kg (in May ). Most (89.2%) females with young were >6 kg (Fig. 3.3). The percentage of these females breeding varied between years (Table 3.1). Figure 3.3: Percentage of independent females in each weight class with dependent young (seasons pooled). Data collected , , and Of the young recorded for females <6 kg, only 3% (3 of 96 where head length of young was measured) had a head length >50 mm (size at which a joey generally becomes furred; Eberhard 1972), compared to 18.5% of young for females that were >6 kg. The percentage of females <6 kg that were lactating also varied between years (Table 3.1), with a maximum of 19.3% with young in

64 Mass (kg) Table 3.1: Variation in female condition and maternal characteristics among years. Age class is based on tooth wear (see 3.2). Females in good condition Breeding in age class >III Females <6 kg breeding Females 6 kg breeding Year % females n % females n % females n % females n , , , , χ P < <0.001 < female male female male Sample site and sex Figure 3.4: Mean body mass of adult koalas on Kangaroo Island (black) compared to the founder population on French Island (white) (McLean, 2003). The number of females in each group is shown above the bar. Error bars represent SEs. 49

65 We found no significant interaction between month and sex of offspring on female mass (F 5, 858 = 1.98, P = 0.08); however, female mass varied significantly with month (F 5,858 = 12.62, P < ) and sex of offspring (F 1,858 = 5.26, P = 0.022). The mean mass of females with young decreased from November to March, increasing slightly in April (Fig. 3.5). This corresponded to an increase in the percentage of females in tooth-wear class III with young from 8% in November to 35% in April and a decrease in the percentage of females in toothwear class IV with young from 68% in November to 58% in April. Overall, females with female offspring were significantly heavier than those with male offspring (t 428 = -0.27, P = 0.79). Figure 3.5: Mean weight of female koalas giving birth to male young (solid line) or female young (dotted line) during the November April breeding period. Error bars represent SEs. 50

66 Maternal age Highest fecundity was associated with tooth-wear class III ( years: 53% with young, 336 of 634), and tooth-wear class IV (5.5 9 years: 66% with young, 921 of 1,396). Breeding (lactation) rates declined in older age classes with 46% (175 of 380) and 36% (26 of 72) of females with young in tooth-wear class V (9 10 years) and tooth-wear class VI (10 14 years), respectively. Only 3% (8 of 298) of independent females in tooth-wear class II ( years) produced young. This pattern was observed each year. The age of the female had no influence on timing of breeding (P=0.28) or sex of offspring (P=0.19). Maternal condition Fecundity was higher for females in good condition than those in poor/average condition (x 2 2 = 14.97, P < 0.001). Across all years, 60% (1,349 of 2,234) of females in good condition produced young compared to 40% (97 of 239) of females in poor average condition. The percentage of females in good condition varied among years (Table 3.1). 51

67 3.4 Discussion The study years encompassed a period of change in koala population densities and environmental conditions on Kangaroo Island. As a result of management activities, koala population densities in preferred habitats decreased from approximately 3 to 1 koala/ha, with maximum densities decreasing from 13 to 3 koalas/ha from 1996 to 2007 (R. Molsher, Department for Environment and Heritage, South Australia, Australia, pers. comm.). Corresponding to the decline in koala density and browsing pressure on food trees was an improvement in canopy cover of the preferred food tree species, rough-barked manna gum (E. viminalis cygnetensis; R. Molsher, Department for Environment and Heritage, South Australia, Australia, pers. comm.). Despite higher leaf biomass, food quality may have been affected by variable rainfall among years. In rainfall was much lower than that recorded in and In the May September period prior to the breeding period only 210 mm of rainfall was recorded in 2006 compared to 447 mm and 428 mm in 1997 and 2005, respectively (Bureau of Meteorology; A higher proportion of koalas were in average or poor condition in compared to previous years, suggesting that koala condition could have been impacted negatively by the effect of low rainfall on food quality. Despite the variation in population density and environmental factors, the breeding season was well defined and the timing consistent among years, with most births occurring between December and May with a peak in January. Eberhard (1972) observed a similar but slightly shorter breeding period for koalas in Flinders Chase National Park, with births occurring between late December and April and a peak in February. The breeding season on Kangaroo Island is also similar to that observed for mainland koala populations in Victoria (Framlingham and Mt. Eccles), but not the island populations of Snake Island and French Island, where births occur over a longer period and up to 40% of births occur before December 52

68 (McLean and Handasyde, 2006). Kangaroo Island has a Mediterranean climate that is more similar to that of Victorian islands than mainland locations such as Mt. Eccles. Lowest temperatures generally occur in July (Bureau of Meteorology; when mean minimum temperature is 6.5 C, compared to 7.1 C for French Island (Stony Point meteorological station) and only 4.5 C for Mt. Eccles (Hamilton meteorological station). If temperature (and its influence on food availability, maternal energy budget and offspring survival) is a major regulator of temporal breeding activity, as suggested by McLean and Handasyde (2006), the length of the breeding season on Kangaroo Island would be similar to that of Victorian islands. The lack of similarity, however, suggests that, for the Kangaroo Island population, other factors such as rainfall or the availability and spatial arrangement of food resources that also affect resources and energy expenditure for free-ranging wildlife (e.g. Nunes, 1995) may also have a strong influence on reproduction. In a study of three koala populations in Queensland, Ellis et al. (2010) suggested that rainfall patterns and their effect on food resources might explain variation in timing of births between sites. They observed the most restricted timing of births to occur at a site with the most restricted period of rainfall. Differences in rainfall patterns also might explain some of the variability in the timing of births in southern Australia. Annual rainfall is slightly lower for Kangaroo Island (627 mm) than French Island (764 mm) or Mt. Eccles (688 mm) and also occurs over a more restricted period, with 81% of annual rainfall recorded from April to October compared to only 70% of annual rainfall occurring over the same period at both island and mainland locations in Victoria. Lower rainfall during the warmer months when koalas are undergoing late-lactation and breeding on Kangaroo Island might result in a shorter period over which the food resources needed for weaning young are available and a more restricted breeding period than at Victorian island sites. 53

69 Extensive fragmentation of habitats preferred by koalas on Kangaroo Island also could result in koalas expending more energy moving between patches than koalas on Victorian islands. This potentially could reduce the amount of energy available for reproduction. Lower fecundity of the Kangaroo Island population (52 67%) compared to Chlamydia-free Victorian populations (66 81% McLean and Handasyde, 2006) also offers support for this hypothesis. Body mass of breeding females and sex ratio of the offspring varied during the breeding period. The earliest females to breed tended to be larger and produced more male offspring. McLean and Handasyde (2006) also observed a male-biased offspring sex ratio early in the breeding season and suggested that this could allow for a longer period of maternal investment in males allowing them to attain a greater size and have a greater competitive advantage. This is consistent with the Trivers Willard hypothesis that predicts that in species with male-biased sexual dimorphism females with greater access to resources, and therefore in better condition, should produce more male offspring (Trivers and Willard, 1973). Alternatively, if males require more maternal investment to achieve high reproductive fitness, mothers could invest more in males by prolonging the period of investment (McLean and Handasyde, 2006). Also, for males born early, favourable weather and the spring growth flush at weaning could promote best development after weaning and ultimately larger body mass. In contrast to our observations and those of McLean and Handasyde (2006), Ellis et al. (2010) found no evidence for a sex difference in the timing of births for three koala populations in Queensland. They suggested that because male and female growth curves during the period of dependency are identical (Tobey et al., 2006), maternal investment in sons and daughters should be identical. However, densities in the Queensland populations were considerably lower than the southern populations studied. If male-male competition increases with increasing population density (Kokko and Rankin, 2006), it may be advantageous for a female to invest more in male offspring to give them a competitive advantage. 54

70 Although we observed that heavier females were the earliest to breed, the heaviest of the breeding individuals in the early and late months of the breeding season tended to produce female rather than male offspring. Clark (1978) suggested that for species with male-biased dispersal females are more likely to produce males early in their reproductive lives to minimise the duration and reproductive cost associated with mother daughter competition. Because koalas exhibit male-biased dispersal (Dique et al., 2003; Ellis et al., 2002; Gordon et al., 1990; Mitchell, 1990b; Mitchell and Martin, 1990), it is possible that younger females (that might be less heavy) have a higher probability of producing male offspring. Our data on koala age were not sensitive enough to determine if this was the case. Despite more males being produced early in the season, sex ratio of offspring for the whole breeding period was close to unity in and , similar to that observed elsewhere (Martin and Handasyde, 1990a; McLean and Handasyde, 2006). In the offspring sex ratio was male biased (55% male); however, the reasons for this are not clear. Trivers and Willard (1973) suggested that a male-biased sex ratio could result from females being in good condition. In a study of captive koalas Tobey et al. (2006) found no relationship between female body mass and sex of progeny, however captive animals with comparatively low energy expenditure and a ready supply of high quality browse may not reflect reality for free-ranging individuals. Furthermore, Ellis et al. (2010) could not attribute a male bias in young in south-eastern Queensland to females being in better condition than those on the island of St. Bees where the sex ratio of young was close to unity. Fecundity of breeding-age individuals (body mass of 6 kg; McLean and Handasyde, 2006) was variable among years, ranging from 52% in to approximately 67% in and Although lower than previous estimates of fecundity obtained for Kangaroo Island koalas, those estimates were based on relatively small samples derived from only 1 or 2 months (Eberhard, 1972; Robinson et al., 1989). The cause of temporal variation in fecundity is not clear but is possibly due to environmental conditions, density-dependent factors and their effects on physiological condition of individuals, or social factors within the 55

71 population, or a combination of these factors. Although fecundity was lower for individuals in poor average condition, variation in condition of females was not paralleled by changes in fecundity. The highest proportion of females in poor average condition was observed in , but the breeding rate in that year was higher than in when a higher proportion were in good condition. This result is not surprising considering that females in poor average condition comprised a relatively small percentage (between 6% and 11% per year) of the female population each year. The status of Chlamydia and its potential influence on fecundity of the Kangaroo Island koala population is uncertain and warrants further investigation. The population frequently is referred to as Chlamydia-free because it was derived from koalas from the Chlamydia-free population of French Island (Masters et al., 2004). Robinson et al. (1989) found no evidence of Chlamydia in blood samples collected from 63 females captured in 1986 but identified the disease in post-mortem examination of two female koalas recovered in Flinders Chase National Park in In addition, serological testing for Chlamydia undertaken in revealed that of 201 koalas tested, 37 (18%) showed possible signs of infection, with six individuals (3%) being strongly positive (R. Molsher, Department for Environment and Heritage, South Australia, Australia, pers. comm.). None of these koalas showed clinical signs of chlamydiosis, and clinical signs were not recorded for any of the koalas in our study. If Chlamydia is present in koalas on Kangaroo Island, the relatively high fecundity of this population compared to Chlamydia-infected mainland populations (32 38% McLean and Handasyde, 2006) suggests that current impact of chlamydiosis on breeding is limited. The effect of environmental conditions on fecundity in many species can vary with effects such as age (Gaillard et al., 2000), genetic variation (Keller et al., 1994), and prior energy expenditure (Gaillard et al., 2000). Whilst young or senescing individuals show yearto-year variation in survival and fecundity (Gaillard et al., 2000), middle-aged individuals 56

72 exhibit the lowest variance in these parameters (Gaillard et al., 2000; Festa-Bianchet et al., 2003). We did not observe a change in the distribution of breeding individuals among age classes among years, suggesting that the effect of environmental conditions was consistent across age classes. However, variation did occur in breeding within females that were <6 kg. In , when fecundity was lowest, only 2% of females that were <6 kg were breeding compared to 19% and 13% breeding in and , respectively. McLean and Handasyde (2006) suggested that although females <6 kg can breed, they might be unable to carry the young to independence. This could indeed be the case on Kangaroo Island because most of the breeding females <6 kg were observed with young at early stages of development (head length <50 mm corresponding to a maximum age of about 17 weeks; after Eberhard, 1972). On Kangaroo Island the increase in fecundity between and and corresponded to a significant decline in koala population density. One explanation is that density-dependent factors might influence reproduction in koalas, however formal testing of this hypothesis would require additional years of data. Koalas maintain a nonterritorial social system that incorporates extensive home-range overlap with little interaction among individuals (Mitchell, 1990b; Ellis et al., 2009). Ellis et al., (2009) suggested that this strategy might enable them to exploit restricted and patchy resources without the cost of direct competition. At high population densities competition or interactions among individuals potentially could result in koalas expending more energy searching for food, on territorial-like behaviour, or on other social interactions than on reproductive activities. Martin (1985c) found that larger older males dominate a site as it becomes heavily defoliated as a result of koala browse pressure, and suggested that they may be excluding females and younger animals from the area. 57

73 Genetic and environmental factors could have contributed to the lower fecundity of the Kangaroo Island population relative to Chlamydia-free Victorian populations. The Kangaroo Island koala population was established from approximately 18 individuals translocated from French Island, which, in turn, was established from as few as two or three animals at the end of the nineteenth century (Martin and Handasyde, 1990a). Cristescu et al., (2009) determined that the Kangaroo Island population has lower genetic diversity than French Island koalas, with 2.4 and 3.8 alleles per locus for koalas in each population, respectively. Coupled with the low genetic diversity of the Kangaroo Island population was a high proportion of males with testicular abnormalities (cryptorchidism). This condition can result in a reduction in fertility (Nistal et al., 1980) and an increased risk of testicular malignancy in older animals (Gorlov et al., 2002; Ferlin et al., 2003; Klonisch et al., 2004). 3.5 Conclusion Spatial variation in koala reproduction highlights the importance of obtaining locationspecific information on reproductive parameters for predicting population growth and the effectiveness of management strategies. An understanding of the potential causes of temporal variation in productivity also is important. Large-scale management programs such as that implemented for koalas on Kangaroo Island provide a unique opportunity for providing this information. 58

74 59

75 4. Efficacy of a slow-release contraceptive implant containing the GnRH agonist deslorelin in free-ranging female koalas There is a growing consensus in Australia and New Zealand that fertility regulation is a desirable management tool for marsupials provided that it is shown to be effective... (Mate et al., 1998) 4.1 Introduction High-density free-ranging animal populations have the potential to cause ecological and economic problems (Caughley, 1981), often necessitating active management to reduce numbers or to limit further population growth. Both lethal and non-lethal management approaches may be used (see Chapter 1). Lethal methods may be employed as a way to immediately reduce the density of these populations, however such techniques can be both politically and publically unacceptable (e.g. Oojes, 1997; Herbert, 2007; Dandy et al., 2012). Fertility control has been proposed as a more ethically acceptable alternative to lethal methods, and is a particularly attractive option when the target population is a native species and eradication is undesirable (Tuyttens and Macdonald, 1998; Adderton Herbert, 2004; Lauber and Knuth, 2007). A number of koala (Phascolarctos cinereus) populations in south-east Australia have rapidly increased in density due to a lack of natural predators and dispersal opportunities resulting from clearing of land, urban development and previous management practices such as introductions to islands (Martin and Handasyde, 1999). Overabundant koala populations in Victoria and on South Australia s Kangaroo Island have impacted severely on their habitat, having been identified as the cause of defoliation and a decline in the health of eucalypts at a number of forested sites. In severe cases this has led to the death of favoured food tree species 60

76 (Martin, 1983, 1985a; Hindell and Lee, 1990; Kelly, 2000; DSE, 2004; Masters et al., 2004). Overbrowsing can have a number of impacts on both the koalas and local biodiversity, and animal welfare concerns arise when a dramatic reduction in food resources leads to the starvation and death of resident koalas. In addition there is concern that vegetation communities in isolated habitat patches may be threatened, impacting on other components of the biota (Martin and Handasyde, 1999; Herbert et al., 2001). Clearly, such koala populations need to be managed. Fertility control is one of two government-approved methods (the other being translocation) for management of overabundant koala populations (Natural Resource Management Ministerial Council, 2009), and is increasingly being viewed as the preferred long-term choice, because suitable recipient habitat patches for translocation are now becoming saturated (Department of Sustainability and Environment, 2004; Menkhorst, 2008; Natural Resource Management Ministerial Council, 2009). The primary aim of developing any fertility control agent is to induce infertility in the target species, with the treatment s efficacy being subsequently judged on its ability to successfully achieve this. When determining the utility of a fertility control technique for population management purposes however, its efficacy is not determined solely on its contraceptive capacity, but also the longevity of induced infertility, ease and effectiveness of treatment delivery (to an individual and to a desired proportion of the population), humaneness, target specificity, environmental acceptability, cost-effectiveness, and, ultimately, whether it results in a reduction in target population size or the damaging impacts of that population (Bomford and O Brien, 1992). A range of fertility control techniques have been trialled on both free-ranging and captive koalas. Gestagen-based contraceptive implants have been trialled on French Island, Victoria (e.g. Handasyde, et al., 1990b; Menkhorst et al., 1998; Middleton et al., 2003; Hynes 61

77 et al., 2010). The most efficacious of these, the gestagen levonorgestrel (see Hynes et al., 2010), is currently used by management authorities to control this population (K. Handasyde, pers. com.). A trial of the contraceptive effect of immunisation with zona pellucida antigens was conducted on free-ranging koalas on Snake Island, Victoria, with varying success (Kitchener et al., 2009b). Surgical sterilisation (tubal ligation) has been used in an intensive and ongoing koala management program on Kangaroo Island since 1997 (Duka and Masters, 2005; Whisson and Carlyon, 2010), and in conjunction with large-scale translocation of koalas in Victoria (Menkhorst, 2008). In all cases, only female koalas were targeted, as targeting female fertility has been shown to be the most effective and efficient approach for managing vertebrate populations (Bomford, 1990; Caughley et al., 1992; Barlow et al., 1997), particularly in polygynous species (Mitchell, 1989) such as the koala. Another promising anti-fertility agent, the GnRH agonist deslorelin, has been shown to reduce fertility in a range of mammalian species (e.g. Trigg et al., 2001; Herbert et al., 2005; Bertschinger et al., 2008; Arlt, 2010; Wilson, 2012; Eymann et al., 2013), including koalas, both captive and free-ranging (Herbert et al., 2001; Herbert, 2002; Greenfield, 2012). Deslorelin is delivered via a slow-release subcutaneous implant. In females, deslorelin treatment results in a down-regulation of pituitary GnRH receptors, causing a decline in FSH and LH concentration, resulting in inhibition of follicular development and ovulation in females (Fraser, 1993; Herbert et al., 2005). Several of the characteristics of an effective fertility control technique for koala population management are readily met by slow-release GnRH agonist implants. Deslorelin implants are commercially available (Suprelorin, Peptech Animal Health, Virbac (Australia) Pty Ltd) and simple to deliver to animals. They are relatively cost-effective because treatment can be conducted at the site of capture, because there is no need for anaesthetic or sedation (Trigg et al., 2001; Herbert and Trigg, 2005), thus there is also no need to transport animals (Greenfield, 2012). One other possible advantage is that because oestrus is likely inhibited, its 62

78 use may reduce or eliminate energetic costs associated with oestrus (see Chapter 6). Conversely, one of the disadvantages of Suprelorin is its relatively short duration of contraceptive effect, particularly for management programs aiming to control species with a long reproductive lifespan. There is, however, some evidence to suggest a dose-dependent response, i.e. higher doses, or prolonged exposure, may lead to a more sustained effect (D Occhio et al., 2000; Herbert et al., 2006; Greenfield, 2012). The most important factor, for Suprelorin to be useful as a tool for koala management, is that it must cause infertility in a high proportion of treated individuals. Koalas are seasonal breeders, with births generally occurring between October and May (White and Kunst, 1990; McLean and Handasyde, 2006; Whisson and Carlyon, 2010) and a peak between December and March (Ellis et al., 2010; Whisson and Carlyon, 2010). Female koalas begin breeding at two or three years of age and typically produce one young per year (Eberhard, 1972; Martin, 1981; Smith, 1979; White and Kunst, 1990; McLean and Handasyde, 2006). Females in good condition are able to breed in consecutive years (Eberhard, 1972; McLean and Handasyde, 2006), and in Victorian populations females begin breeding in their third season, corresponding to a body mass of around 6 kg (Handasyde, 1986; Martin and Handasyde, 1990a; McLean and Handasyde, 2006). Fecundity generally declines after ten years of age (Martin and Handasyde, 1990a). Herbert (2002) showed Suprelorin treatment can cause temporary infertility for at least one year in captive female koalas, whilst Greenfield (2012) observed that Suprelorin suppressed fertility in free-ranging females on French and Raymond Islands (Victoria) by %. Following a single treatment with a standard 4.7 mg implant, fertility in all free-ranging animals had returned by the third breeding season posttreatment (Greenfield, 2012). In any population, there may be a number of animals that are non-responders to a fertility control agent, i.e. animals that continue to breed following contraceptive treatment due to variation in individual response (Herbert, 2002). Non-responders were not observed during 63

79 trials of levonorgestrel on female koalas (Hynes et al., 2010), however non-response to deslorelin treatment has been observed in several marsupial species (Herbert, 2002; Herbert and Trigg, 2005). Non-response to a contraceptive agent may be due to genetic variation, low dose rates or the failure of the contraceptive to bind to receptors (e.g. Herbert et al., 2005; Hynes et al., 2010). Particularly given non-response may be a heritable trait (Herbert et al., 2005), trials of the efficacy of a contraceptive agent are essential before undertaking largescale management of a new species. My research aimed to establish the efficacy of Suprelorin treatment in free-ranging koalas on Kangaroo Island, involving an assessment of the contraceptive effect of treatment on female koalas at two separate sites on Kangaroo Island. Both of these sites required active management because of the high-density resident population. Animals at one site were subjected to the standard intensive management procedures (normally delivered during the State government program = management simulation) in order to assess both the efficacy of the contraceptive and the feasibility of delivering treatment to a large proportion of females in a population at a particular site. At the other site, radio-collared females, (involved in a study of the effects of Suprelorin on ranging behaviour and movement patterns, see Chapter 6), were used to assess fertility following treatment, because these koalas were being re-captured on a sufficiently regular basis to accurately assess fertility. 64

80 4.2 Materials and Methods Study sites (i) Management simulation Northeast River This component of my research was conducted in a 95 ha site comprising tall eucalypt woodland on private property along a 4 km stretch of the Northeast River, 75 km from Kingscote (35º56 25 S, 137º0 26 E) (Fig. 4.1). This site was chosen because it contained a large proportion of Eucalyptus leucoxylon, a species of high management concern that was showing considerable evidence of koala browse-damage. The site was also chosen for its high density of koalas (based on Department for Environment and Heritage (DEH) monitoring data) and its relative isolation, being surrounded by pasture or unsuitable koala habitat. Prior to this study, there was no management of koalas within this site, and no sterilised koalas had been observed migrating in from other sites. The vegetation within the site is classed overall as medium quality koala habitat (as per Masters et al., 2004) and consists of a tall E. cladocalyx/e. leucoxylon overstory. Other eucalypt species found in this site, more commonly with increasing distance from the river, are E. fasciculosa, E. cosmophylla, and E. baxteri. Large stands of dead E. viminalis dominate the alluvial soils characterised by extensive patches of bracken fern (Pteridium esculentum). It is assumed these dead trees succumbed to koala browse pressure, as this large-scale tree mortality is typical of koala damage seen in similar areas across the island. Only 12 individual E. viminalis were alive within the site, with the majority of these fitted with metal collars to exclude browsers by managers prior to this study. Close to the river, the understory was sparse (aside from the bracken fern/e. viminalis associations), with patchy distributions of Melaleuca gibbosa, Leptospermum lanigerum, 65

81 Callistemon rugulosus and Acacia retinodes. Further from the river the understory was often dense, dominated by Xanthorroea semiplana and Acacia paradoxa, and in some places consisted of a near monoculture of dense Allocasuarina meulleriana or A. verticillata. The site was surrounded by cleared pastoral land to the south and west, and E. globulus plantation at various stages of development to the north and east (Fig. 4.1a). An electric fence (6-9,000 volts), running north-south, formed the eastern boundary of the site and was erected to prevent herbivores accessing E. globulus plantations from the river. (ii) Telemetry study Pioneer Bend This component of the study was conducted on private property at Pioneer Bend (35º43 33 S, 137º15 31 E) in the Cygnet River catchment on Kangaroo Island, approximately 35 km west of Kingscote (Fig. 4.1). This area is considered high-quality habitat for koalas (see Masters et al., 2004) and supports the highest density of koalas on the island (up to 3.4/ha at the time of my study, previously up to 13/ha; DEWNR, unpublished data 2007). It is characterised by linear strips of woodland vegetation dominated by rough-barked manna gum (Eucalyptus viminalis cygnetensis), South Australian blue gum (E. leucoxylon) and pink gum (E. fasciculosa) along the permanent watercourses, and woodland patches that may contain a mix of brown stringy-bark (E. baxteri), messmate stringy-bark (E. obliqua), cup gum (E. cosmophylla), and pink gum under an overstory of sugar gum (E. cladocalyx) at the upper reaches of more ephemeral creeks. Isolated or small patches of eucalypts occurred throughout the surrounding pasture, which is predominately used for grazing stock. 66

82 Figure 4.1: Kangaroo Island, South Australia, showing virtual boundaries of management simulation sites. Grey shading represents National/Conservation Park. a: Northeast river treatment site, yellow line represents virtual boundary of study site, white line represents electric fence. b: example of habitat, Northeast River. Koala capture and treatment All koalas were caught using the methods described in Section 2.2. (i) Management simulation Northeast River Management of this koala population using Suprelorin implants was conducted using the same methods used currently by DEWNR when they sterilise koalas, which allowed me to make a direct comparison with current management practice. Koalas were captured over a period of 20 days each (four working weeks) in each of September 2005, October 2006 and September 2007, prior to the start of each breeding season. Where possible, all females sighted 67

83 were captured for Suprelorin treatment, with the aim of treating as many females as possible each year. Dependent young (on the back or in the same tree) were caught with the female. Capture duration (duration for whole capture process, i.e. from initial setting of climbing ropes or catch gear to the time the koala was on the ground and held in a hessian sack), was recorded for each individual for assessment of the efficiency of the method. Depending on the number of female koalas requiring capture, approximately 20 ha were searched per day; therefore the whole site was surveyed at least four times during each 20-day capture period to ensure that any koalas not sighted on a previous survey, or animals moving in and out of the study site, were treated. During each capture period, all female koalas identified in the site were caught where possible. In 2006 and 2007, this included any koalas previously treated (identified from the ground by their unique ear-tag colour combination). As the duration of contraception using Suprelorin was unknown in free-ranging koalas at this time, these animals received an additional 4.7 mg implant at each recapture to ensure they carried a functional implant into each breeding season. Following capture, the pouch of each female was checked for presence of pouchyoung (PY) and the sex of any young recorded. Furred young that were not attached to the teat were temporarily removed from the pouch for measurement. Each adult koala was tagged using plastic, colour-coded, uniquely numbered ear tags. Dependent young >2.0 kg were also tagged for subsequent identification (for details, see Chapter 2). All females, and dependent female young >400 g, were treated with one 4.7 mg deslorelin implant (Suprelorin, Peptech Animal Health Pty Ltd, Virbac (Australia) Pty Ltd) inserted subcutaneously between the shoulder blades. An electronic Passive Integrated Transponder (PIT) tag (Allflex Australia) was also inserted with the implant as a back-up for 68

84 subsequent identification in case of tag loss. All koalas were then released back into the tree from which they were caught. Individual koalas were held no longer than 40 minutes. (ii) Telemetry study Pioneer Bend Twenty-four adult female koalas were captured in October Only fertile female koalas (carrying dependent young) that were between age-class III and V inclusively, i.e years of age, (see McLean, 2003) were used. Each female was randomly assigned to one of three treatment groups (n=8 in each group): Group 1: Suprelorin-treated; Group 2: surgically sterilised; and Group 3: control (i.e., placebo implant) animals. There was no difference in initial mean body mass between groups (mean 7.7 kg, F 2,21 = 3.32, P = 0.056). All koalas were captured from very similar habitat in close proximity to each other; therefore it was assumed that all individuals had equal access to similar resources at the commencement of the study. Suprelorin-treated and control koalas were processed at the capture site. All Suprelorin-treated koalas received one 4.7 mg implant subcutaneously immediately following capture (see section 2.4), while each control female received a single placebo implant to control against any effects of the treatment procedure. The weight, head-length, tooth wear (for age class determination), and condition of each koala and any dependent young were recorded. Following assessment, each adult koala was ear-tagged with a unique colour combination, microchipped and fitted with a VHF radio-collar (see 2.8) before being released (with young) into the tree from which it was caught. Koalas were held for a maximum of 30 minutes before release. Animals in the surgical sterilisation treatment group were transported (with young) to the Kingscote Veterinary Clinic (35 km) where they were held for 1-2 days in holding pens (two females per 1.5x3x2 m pen), with unrestricted access to fresh eucalypt leaves, before sterilisation (tubal ligation, see 2.7). Each koala was ear-tagged and fitted with a radio-collar at the time of surgery. All sterilised females were released (with young) into the tree from 69

85 which they were caught approximately four hours post-surgery. Each radio-collared koala was tracked regularly (mean 10 days between radio fixes) from October 2006 (i.e., treatment) to February 2009 (see Chapter 6 for details). Each female koala and any dependent young were recaptured in May and November 2007, May and November 2008, and June Suprelorin-treated and control females were re-implanted in November 2007 to ensure they had a functioning implant during the 2007/2008 breeding season (control animals received another placebo implant at this time). Seven Suprelorin-treated (one treated koala could not be located after 2 February 2008) and all control animals were subjected to a GnRH challenge during their re-capture in May 2008 (see Chapter 2) to determine the degree of pituitary downregulation due to deslorelin treatment (after Herbert et al., 2004a), to provide an indication of contraceptive effect irrespective of environmental influences on reproductive success. The response of each koala to an intravenous injection of GnRH was assessed between 189 and 197 days following administration of their second Suprelorin or placebo implant. Serum LH concentrations were determined using an enzymeimmunoassay (EIA) procedure (see Chapter 2). One surgically sterilised koala died in January 2007, within 42 days following surgery; however this animal was too decomposed when found to determine the cause of death. This animal was excluded from analyses. Analysis (i) Management simulation Northeast River Fecundity of treated and untreated females in the management simulation site at the North-east River was determined as the proportion of recaptured and newly-caught females observed to be carrying young when caught. If a juvenile koala was observed in the same tree 70

86 as a female and deemed small enough to be from the previous breeding season, and the female had a lactating or resolving teat, the young was assumed to belong to that female. Female koalas <6.0 kg in body weight were excluded from analysis of fecundity because the majority of southern female koalas do not breed successfully until after they reach 6 kg (McLean and Handasyde, 2006; Whisson and Carlyon, 2010). Fisher s exact test was used to compare the breeding success of re-captured (Suprelorin-treated) versus newly caught adult females in each management year. A Student s t-test was used to compare capture durations (i.e. time taken to capture an individual koala, defined as from the start of setting up catch gear at the base of a tree to when the animal is restrained on the ground) with records from the DEWNR Koala Management Program ( ). (ii) Telemetry study Pioneer Bend Fisher s Exact Test was used to compare the fertility of Suprelorin-treated koalas to that of control animals. To compare the mean concentration of LH observed in control and suprelorin-treated koalas as a function of time during the GnRH challenge, a generalised additive mixed model (GAMM) was used. An additive model was used to account for a non-linear change in plasma LH concentration over time (time variable fitted with a smoothing function), and the mixed model approach used to account for repeated measures of individual koalas. A unique identifier for each koala was used to account for non-independent error structures, and a Gamma distribution was used as the residual diagnostic plot indicated a default Gaussian error distribution was not appropriate. An auto-regressive correlation structure (AR-1) was included in the model to account for a lack of independence at various time lags in the data set. Subsequent ACF plots showed inclusion of the AR-1 correlation structure clearly reduced the temporal correlation in the data. Modelling was performed using the gamm function in 71

87 package mgcv (Wood, 2006) in R v (R Development Core Team, 2012). Data at single time points were analysed using paired-sample t-tests. 72

88 4.3 Results Management simulation Northeast River Over the course of the study (three breeding seasons), 95 captures involving 56 individual female koalas (including young) were made in the Northeast River study site. In 2006 and 2007, recaptures comprised 57% and 68% of captures respectively (Fig. 4.2). Fortyfive percent of females were caught and implanted in just a single year, while 41% were captured in two of the three years. Only eight females (14%) were captured and implanted in all three years (Table 4.1). One surgically sterilised female, sterilised as part of the government management program in March 2006, was found and captured in the study site in September 2007, 5.6 km north of its original catch and release site. Figure 4.2: Percentage of recaptured and new female koalas caught and implanted in the Northeast River site, Kangaroo Island, each management year. The total number of females treated each year is shown above the bar. 73

89 Table 4.1: Number of individual adult female koalas caught and implanted in single or multiple management years in the Northeast River site, Kangaroo Island. Year caught Number of individuals 2005 only only only & & & , 2006 & A mean of 92.5% (±1.5% SE) of all females located in the site were successfully caught and treated during each capture period. Reasons for non-capture included rotten trees and bee hives that made climbing unsafe for capture. The mean duration of the capture process (i.e. length of time between setting up for a capture to the time the koala is on the ground) for koalas during this study was 50.6 (± 3.1) minutes (n=95), with 95.8% of captures at this site requiring a climber to access the koala. This is significantly greater than the mean capture duration of 22.2 (± 0.5) minutes (p < 0.001) using all capture records from the DEWNR koala management program for which duration of the capture process was recorded (n=1702; see section 2.2 for description of capture method). Fertility/presence of young Northeast River No adult female koalas that were implanted in 2005 were observed with young in 2006 (0/14), compared with 62.5% of newly-caught females (5/8) (Fisher s exact test: p = 0.002, Fig. 4.3). No recaptured adult females (0/21) were found with young in 2007 (including those 74

90 that were implanted in 2005 but not 2006) compared to 20% of newly-caught females (2/10), however the observed effect of treatment was non-significant (Fisher s exact test: p = 0.125). Figure 4.3: Fecundity (percentage of females >6.0 kg with dependent young) of newly caught and recaptured females in each management year. Total number of females caught over 6 kg is shown above the bar (note zero recaptured koalas were found with young in 2006 (n=14) and 2007 (n=21)). Fertility/presence of young telemetry study, Pioneer Bend All females (n=23) had dependent young at the time of treatment in October At the end of the study no surgically-sterilised koalas had produced young. Two Suprelorintreated females produced young following treatment: one during the 2007/2008 breeding season, and one during the 2008/2009 breeding season (Fig. 4.4). The first young did not survive it was not subsequently observed as back-young. Survival of the other young born during the 2008/2009 breeding season is unknown as collars were removed from animals in June Only two of the control females produced young during the 2006/2007 breeding season and just one of these young was subsequently observed as back-young and successfully weaned. Six control females (75%) were carrying pouch-young in May 2008 (conceived during the 2007/2008 breeding season) and three (37.5%) were carrying young in May

91 from the 2008/2009 season. A significant difference in fertility between control and Suprelorin-treated groups was observed during the 2007/2008 breeding season (Fishers Exact Test, p = 0.04), however the fertility of control females was too low in other breeding seasons to detect a difference between groups. Figure 4.4 Percentage of female koalas in each treatment group at Pioneer Bend, Kangaroo Island that produced young in each breeding season during the course of this study. Arrow indicates timing of fertility control treatment (October 2006 = tubal ligation and Suprelorin treatment; November 2007 = repeat Suprelorin treatment). The number of females in each group is shown above the bar. GnRH challenge - radio-collared female koalas, Pioneer Bend Control (n=8) and suprelorin-treated (n=7) female koalas at Pioneer Bend were subjected to an intravenous injection of GnRH between 189 and 197 days following Suprelorin or placebo administration. The mean LH response of each group to the challenge is presented in Fig The initial mean concentrations of LH in both groups prior to (-15 minutes) and during GnRH injection (0 minutes) were not significantly different (p=0.268 and p=0.201 respectively). Fifteen minutes following GnRH administration, plasma LH concentrations of 76

92 control animals increased to a mean peak of 1.33 ± 0.27 ng/ml. The mean LH concentration of Suprelorin-treated koalas followed a similar response pattern, with all animals demonstrating a peak in LH (defined as > 2 x s.d. of starting concentrations) when challenged, however the magnitude of the peak was much reduced (mean 0.64 ± 0.10 ng/ml at 15 minutes postchallenge). Overall, control koalas showed a significantly greater response in mean LH concentration compared to Suprelorin-treated animals over time (p=0.008), however there was variation in the response of Suprelorin-treated animals, suggesting that the level of pituitary desensitisation varied between individuals. Two Suprelorin-treated koalas exhibited LH peaks within the range that koalas in the control group exhibited; one of these was a koala that was observed carrying a small dependent pouch-young post-treatment in November Figure 4.5: Mean plasma concentrations of LH following administration of a synthetic GnRH (2 ug/kg) to control (dashed line; n=8) and suprelorin-treated (solid line; n=7) female koalas. GnRH was administered at time 0. Error bars represent one s.e. 77

93 4.4 Discussion This study indicated treatment with subcutaneous Suprelorin implants inhibited fertility in the majority of treated female koalas on Kangaroo Island for at least one breeding season. While the low level of breeding by control animals in two years complicates our ability to form fixed conclusions about efficacy of suprelorin treatment, no Suprelorin-treated koalas captured during the management simulation were subsequently found with young, including koalas not recaptured until the second season following treatment. Although the results presented here should be confirmed in further trials with larger sample sizes if possible, they are consistent with findings by Herbert (2002) and Greenfield (2012) who showed high efficacy of this treatment in captive female koalas in New South Wales and free-ranging females in Victoria respectively. No surgically-sterilised females produced young following surgery, demonstrating the permanent effect of this procedure and corroborating DEWNR observations that no sterilised females have subsequently been observed with young during ongoing management programs over 17 years (DEWNR, unpublished). Suppression of koala fertility following Suprelorin treatment was not 100% effective. Two Suprelorin-treated koalas in the telemetry group (25%) produced young within 12 months of treatment, although overall this equated to just 12.5% of Suprelorin-treated females observed with young each season. Although the ability to make a firm comparison is limited by the small treatment sizes available in the current study, this figure is similar to Greenfield (2012) who observed 0-11% of Suprelorin-treated females produced young in a given breeding season. The basis for the failure of Suprelorin treatment to inhibit fertility in these cases is unclear. There are several possibilities: (i.) these koalas were non-responders to the effects of deslorelin, as has been observed in some individuals in previous studies (Herbert et al., 2005; 78

94 Herbert, et al., 2006); (ii.) the implants administered to these koalas were either faulty, broken (see Lohr et al., 2009) or lost; or (iii.) there is individual variability in the responsiveness to a given dosage. D Occhio et al. (1996) suggested variation in contraceptive duration could be due to variability among implants in the duration and amount of agonist released. Although female koalas treated with Suprelorin responded to exogenous GnRH with an LH surge, their response was much reduced compared to that of control females. Two Suprelorin-treated female koalas had an LH peak within the range of that exhibited by control females, however only one of these actually produced young post-treatment. It is possible therefore that this animal was a non-responder, or that the implant was defective. If this animal was indeed a non-responder, given that non-response could be due to genetic variation and hence heritable (Herbert and Trigg, 2005; Herbert et al., 2005), it would then be reasonable to assume a proportion of female koalas on Kangaroo Island may be non-responders. Such animals could be targeted as candidates for surgical sterilisation if recaptured in subsequent years of the state government koala management programs on Kangaroo Island. The mechanism of non-response to Suprelorin treatment in koalas on Kangaroo Island needs to be investigated further before this technique can be considered as a primary management tool. The second Suprelorin-treated koala that exhibited a substantial LH surge might also have been a non-responder and we either failed to detect a young that was born and lost between capture events, or this koala failed to produce young for reasons other than treatment. Female koalas are physiologically capable of giving birth to a single young each year, and under ideal conditions most females breed every year for eight to ten years (Eberhard, 1972; Martin and Handasyde, 1999; McLean, and Handasyde, 2006). However, fecundity of breeding-age females on Kangaroo Island however was approximately 67% in 2005/2006 and 2006/2007 (Whisson and Carlyon, 2010) and just 11% in the Cygnet River area in 2007/08 (DEWNR, unpublished), suggesting that environmental factors may influence breeding success 79

95 in the Kangaroo Island population, which might also explain the absence of a young in the female described above. Rather than being non-responders, the two females that produced young may be demonstrating variability in response duration and were simply not suppressed for as long. Given the high efficacy at the management simulation site, and the efficacy demonstrated by Greenfield (2012) in Victoria, it is proposed that 4.7 mg of deslorelin is close to the threshold required to inhibit reproduction in koalas. Hence, in some females, the degree of suppression is sufficient to inhibit follicular development (or possibly just ovulation if FSH and LH are at a level that can maintain follicular development, but the pituitary is down-regulated to the extent that it cannot mount the surge of LH necessary for ovulation), while in others it is not. Conclusively determining whether this was a classical non-response (i.e. deslorelin has no inhibitive action on the pituitary cycle of these individuals, rather than a shorter duration of effect) would require more dose-response studies. The GnRH challenge results suggest pituitary desensitisation was not complete in Suprelorin-treated female koalas at the dose-rate used, even following chronic exposure to deslorelin (i.e. two 4.7 mg Suprelorin implants delivered 12 months apart). Results from several studies indicate that prolonged exposure to deslorelin may result in a more sustained contraceptive effect (D Occhio et al., 2000; Herbert et al., 2006; Greenfield, 2012), however all treated koalas in this study demonstrated some degree of a surge in LH in response to exogenous GnRH following two Suprelorin treatment events, spaced 12 months apart. Nevertheless, the significantly lower fertility exhibited by this group compared to control koalas, suggests concentrations of LH and FSH were likely reduced to levels that inhibited follicular development and/or ovulation in most animals following Suprelorin treatment. GnRH agonists act by suppressing the release of gonadotrophins rather than downregulating the ovaries (D Occhio et al., 1997). Therefore sufficient circulating FSH and LH 80

96 concentrations in some koalas may support ovarian activity in deslorelin-treated females even though the pituitary appeared unresponsive to GnRH. The administration of a higher dose at each treatment (e.g. two implants) may result in more effective down-regulation of GnRH receptors in such animals and is worthy of further research. Management implications The choice of a fertility control agent for population management depends on the management context and the aims of each individual management program. Any treatment must have an appropriate duration of effect, particularly when used for species with a long reproductive lifespan, and must be effective in inhibiting fertility in an appropriate proportion of the population to achieve the desired outcome. Most population management scenarios require long-acting fertility control agents and techniques that require a minimal number of treatments of each individual to limit the stress on individuals and to minimise the cost of ongoing management. The failure of Suprelorin treatment to inhibit fertility in 100% of female koalas, both in this study and that by Greenfield (2012), contrasts with the 100% efficacy of levonorgestrel implant treatment of free-ranging female koalas on French Island, Victoria (Hynes et al., 2010). In addition, levonorgestrel treatment provides effective contraception for a period of at least six years from a single treatment (Hynes, et al., 2010), compared to 1-2 years of contraception from a single Suprelorin treatment (Herbert, 2002; Greenfield, 2012), depending on timing of delivery. Maximum effect can be achieved by the timing of treatment. For example, in seasonally-breeding species such as the koala (McLean and Handasyde, 2006; Whisson and Carlyon, 2010), it would be most efficient to deliver treatment immediately prior to the breeding season: while this would result in an 18-month period of contraceptive-effect, it would effectively deliver inhibition of reproduction for the subsequent two breeding seasons (i.e. 2 years). 81

97 Since the delivery procedure of both contraceptive treatments is now very similar and, given there have been no direct side-effects on health or behaviour resulting from either treatment (see Herbert, 2002; Hynes et al. 2011; Greenfield 2012; Chapters 5 and 6 of this thesis), Suprelorin treatment is not as efficient as levonorgestrel if contraception of a large proportion of a population for an extended period is the primary management aim. The reproductive biology of the target species is also important. Female koalas have a reproductive lifespan of up to ten years (Martin and Handasyde, 1990a), and a proportion of animals breed until they are over 14 years old (K. Handasyde, pers. comm.); therefore every individual would require at least five treatments with Suprelorin to abolish reproductive output for the animal s lifetime. In contrast, only one or two treatments with levonorgestrel would be required to achieve the same management outcome. Duration of contraception and therefore the number of treatments needed per individual is particularly important when managing free-ranging animals. All current fertility control treatments require capture of each individual koala for administration, and the ease with which animals can be captured will dramatically affect the efficacy of the technique if multiple treatments are required. Ease of capture varies greatly between koala populations, and even between sites on Kangaroo Island due to considerable variation in the height of the eucalypts and koala population density. The management simulation component of this study showed that capture durations were significantly greater at this study site compared to the mean duration for the rest of the island, which was due almost entirely to the predominance of very tall trees at this site. Regardless of the fertility control technique used, it is clear that treatment of a substantial proportion of female koalas occupying a discrete area will only be achieved by undertaking management activities over a number of consecutive seasons. In 2007 for example, following two seasons of intensive management, only 68% of koalas caught in the 82

98 management simulation site were animals treated during the previous two years. It is likely that the remaining 32% of untreated koalas were animals that had immigrated to the study site (long-distance movements by koalas on Kangaroo Island are discussed in Chapter 6) or had been overlooked in the preceding seasons. It is generally considered that for fertility control to be effective at limiting population growth, at least 70-90% of the resident koala population must be rendered infertile in any one year, although the models on which these predictions are based assume high individual and population-level fecundity (Martin and Handasyde, 1999; McLean, 2003). Population-modelling, that includes information on immigration and emigration where possible, should be used to determine the proportion of animals that need to be treated at each site to achieve the desired result, however, clearly, a fertility control treatment that requires minimal repeat captures of individuals will enhance the ability to achieve the desired result with minimal effort and cost. Whether the 32% of new animals caught in 2007 were immigrants or whether they were overlooked during previous years has profound implications for management; if they were immigrants this would suggest that a larger area around a site targeted for management may have to be managed. Alternatively, if they were simply overlooked during previous management activities, search effort would clearly need to improve. Management situations where Suprelorin treatment may be useful include circumstances where animal capture is straightforward and where reversibility of contraceptive effect is desired. For example, some small free-ranging or captive populations may require a fertility control agent with a relatively short duration of effect to enable turnover in the breeding individuals, maximising the effective population size (Lande and Barrowclough, 1987). This may be a valid concern for koala managers due to the low genetic variability of many overabundant koala populations (e.g. Sherwin et al., 2000; Department of Sustainability and Environment, 2004; Cristescu et al., 2009; Natural Resource Management Ministerial Council, 2009; Tanaka et al., 2009). 83

99 The mode of treatment administration is another key determinant of the efficacy of a fertility control technique. Insertion of subcutaneous Suprelorin or levonorgestrel implants with a pre-loaded applicator is less invasive than surgical sterilisation, and the ability to deliver these treatments in the field reduces transport and holding costs and associated stress imposed on the animals. Implant delivery also requires minimal training for field staff. The smaller size of Suprelorin implants lends them greater potential for remote delivery, and techniques are being developed to achieve this (Herbert and Vogelnest, 2007). Remote treatment delivery, by darting or pole syringe, eliminates the stress of animal capture and reduces management time (Kirkpatrick et al., 1990) and would be particularly attractive at sites with smaller trees where koalas may be readily approached, where there is a clear line of site to the animal, and where treatment could be delivered from the ground (i.e. the most stable platform). There may be potential for implants to be delivered remotely by a climber once a tree is accessed, however this is a less stable platform and accuracy might be compromised. Although Suprelorin and surgical sterilisation (tubal ligation) both successfully inhibit fertility in female koalas (albeit with different longevity of effect), they do so in very different ways: treatment with Suprelorin results in inhibition of follicular development and ovulation, whereas the more invasive tubal ligation procedure involving cauterisation of the utero-tubal area leaves the ovaries intact and functional and sterilised females still undergo oestrus. Therefore the mechanism of any fertility control also needs to be considered by managers, as this may have impacts on animal physiology and behaviour, resulting in welfare concerns. The impact of these treatments on the welfare and behaviour of female koalas are examined and discussed in subsequent chapters. 84

100 85

101 5. The effects of two fertility control techniques on the health and condition of free-ranging female koalas overall the benefit of contraception should outweigh the risk of harm when selecting a contraceptive for a species (Munson et al., 2005) 5.1 Introduction Active management of free-ranging animal populations becomes necessary when local population growth exceeds the availability of space and resources. This is usually driven by high reproductive output combined with high survival rates and/or limited ability for animals to disperse. For some high-density koala populations in southern Australia, a reduction in koala abundance is required to reduce or limit habitat deterioration, as this may lead to poor nutritional status, or even starvation and death of koalas (Martin and Handasyde, 1999; Natural Resource Management Ministerial Council, 2009). Fertility control is frequently proposed as a more humane alternative to lethal methods of wildlife population control, particularly when the target population is a native species and eradication is undesirable (Tuyttens and Macdonald, 1998; Adderton Herbert, 2004; Lauber and Knuth, 2007; Fagerstone et al., 2010), and is currently one of only two governmentsanctioned methods (the other is translocation) for management of high-density koala populations (Natural Resource Management Ministerial Council, 2009). The general acceptance of fertility control is often due to a perception that these techniques are less invasive or more humane than the alternatives, however there is still a risk of adverse effects on individuals (Nettles, 1997; Munson et al., 2005; Gray and Cameron, 2010). The application of a fertility control method should therefore be based, in part, on knowledge that the risk of 86

102 harm to a treated individual is outweighed by the benefits of treatment at both an individual and population level (Fagerstone et al., 2002; Munson et al., 2005). Serial assessment of the health and condition of free-ranging animals is inherently difficult. As such, simple morphometric indices are often used as proxy methods for health assessment (e.g. Schulte-Hostedde et al., 2001). Body weight (e.g. Hanks, 1981; Pearson et al., 2003) or condition (e.g. Carrick and Wood, 1986; McLean and Handasyde, 2006) can be used as non-invasive indices of an animal s physical condition pre- and post-treatment. However, the scale and duration of change in a given morphometric is important and must be considered within the context of natural temporal and physiological variations. For example, the body condition or weight of free-ranging animals may fluctuate naturally in response to a range of factors, including environmental conditions (e.g. Parrott et al., 2007; Turner et al., 2012; Ratikainen and Wright, 2013), food availability (e.g. Strong and Sherry, 2000), population density (e.g. Serrano et al., 2011) and reproductive status (e.g. Reimer and Hindell, 1996; Bradford et al., 2012). Change in body weight has been used to assess the effect of fertility control techniques on the health of a range of wildlife species. Treatment with gestagen-based contraceptives has resulted in disparate trends in body weight change depending on the species treated. For example, treatment of white-tailed deer (Odocoileus virginianus) with progestins has been associated with weight loss coinciding with an extended breeding season (Turner et al., 1992) and/or reduced time spent grazing (White et al., 1994). Conversely, weight gain was reported in hamadryas baboons (Papio hamadryas) following treatment with a similar progestin-based contraceptive (Portugal and Asa, 1995) and in female Mongolian gerbils treated with the synthetic estrogen Quinestrol (Xiao-Hui and Da Zhao, 2011). Increases in the body weight of individuals treated with gestagens were also noted in marmosets (Callithrix jacchus) (Mohle et al. 1999), domestic cats (Felis domestica) (Looper et al., 2001) and Rodrigues fruit bats (Pteropus rodricensis) (Hayes et al. 1996), whilst no overall weight change was noted in 87

103 eastern grey kangaroos (Macropus giganteus) treated with the synthetic progestin levonorgestrel (Nave et al., 2002). Post-treatment survival is another straightforward measure of the effects of a given fertility control procedure, and can be influenced by the health and condition of the individual. For example, survival of immunocontracepted female rabbits was observed to be greater following fertility control treatment, likely due to their increased ability to maintain body condition during times of reduced food availability (Twigg et al., 2000). Likewise, wild horses treated with immunocontraceptives exhibited increased longevity compared to untreated horses, likely due to improved body condition associated with a lack of lactational stress (Turner and Kirkpatrick, 2002; Kirkpatrick and Turner, 2007). In contrast, Ji et al. (2000) suggested that the surgical sterilisation of female brushtail possums (Trichosurus vulpecula) may cause increased mortality of male possums resulting from a decrease in body-condition following an extended mating season. Body condition indices are ideal for assessing the impacts of treatment in koalas as, with their energy-poor diet of Eucalyptus leaves, koalas operate on a tight energetic budget and generally have an extremely low percentage of body fat (Cork and Sanson, 1990; Ellis and Carrick, 1992) and changes in condition should be obvious. Whilst adult female koalas naturally exhibit seasonal fluctuations in weight related to their limited energy intake and the demands of lactation (Krockenberger, 2003; McLean, 2003), an overall gain or maintenance of body weight and condition post-treatment should be considered positive, particularly in freeranging animals, whilst an overall decrease in the body weight of treated individuals should generally be accepted as an undesirable outcome. The risk of adverse health impacts from any fertility control treatment can be dependent on factors including the biology of the target species and the delivery method and/or dose-rate. For example, gestagen-based contraceptives are generally considered safe for most 88

104 wildlife species; however, carnivores appear especially sensitive to progestin-induced pathology following treatment with the gestagen-based contraceptive melengestrol acetate (MGA) (Munson et al., 2002). Also, the male chemosterilant, alpha-chlorhydrin, while successful in controlling fertility in rats at low doses, has toxic effects at higher dose-rates (Bowerman and Brooks, 1971; Andrews and Belknap, 1983). Clearly, the species-specific impacts of a given fertility control agent should be assessed for health impacts prior to application to ensure it is an appropriate and humane population control method. Hormone-based contraceptive implants (levonorgestrel, etonogestrel and oestradiol), that act by reducing the concentration of LH and FSH, but do not compromise ovarian activity, have been trialled extensively in wild female koalas in Victoria (Handasyde et al. 1990b, Middleton et al., 2003; Hynes et al., 2010), and levonorgestrel is now successfully used by government wildlife managers to control fertility in some populations (K. Handasyde, pers. comm.). Adverse health effects, including mortality and hyperplasia of the reproductive tract, were observed when using oestradiol at high doses in physiological trials, however at low doses, no impact was observed and most animals gained weight post-treatment (Handasyde et al., 1990b). Middleton et al. (2003) obtained similar results for health and condition, however survival rates of treated females appeared lower in this study. It is unclear whether this apparent lower survival was due to direct mortality or whether treated animals were ranging further (this latter theory supported by results in Hynes et al., 2011), as many animals were simply not resighted in subsequent years. No adverse health effects have been observed in female koalas treated with either levonorgestrel or etonogestrel, and levonorgestrel-treated koalas also gained weight following treatment (Hynes, 2007). Fertility control trials using a slow-release GnRH agonist contraceptive implant (Suprelorin ) have been conducted on captive female koalas (Herbert, 2002) and in freeranging female koalas in Victoria (Greenfield, 2012). In females, deslorelin treatment results in a down-regulation of pituitary GnRH receptors, causing a decline in FSH and LH 89

105 concentration, resulting in inhibition of follicular development and ovulation in females (Fraser, 1993; Herbert et al., 2004a). No adverse health effects of deslorelin treatment have been observed in koalas and, as for female koalas treated with gestagens, the limited information available for free-ranging koalas shows that deslorelin-treated females gain weight following contraception (Greenfield, 2012). Suprelorin implants have not negatively influenced body condition in other marsupial species, including common brushtail possums (Lohr et al., 2009), eastern grey kangaroos (Herbert et al., 2006) and tammar wallabies (Herbert et al., 2005). Surgical sterilisation of female koalas, involving endoscopic tubal ligation, has been used to control fertility in a number of populations in Victoria (Menkhorst, 2008) and South Australia (Duka and Masters, 2005). Whilst survival of surgically sterilised females has been relatively well documented (St John, 1997; Clark, 1998; Duka and Masters, 2005; Whisson et al., 2012), and initial trials resulted in no observed side-effects, the health impacts beyond the short-term (i.e., > 9 months) are relatively unknown, and this technique must be considered to be more invasive for koalas than the relatively simple delivery of a subcutaneous contraceptive implant (see Hynes et al., 2010). Whilst most studies focus on the health impacts resulting from the long-term effects of treatment, the treatment delivery method may also have impacts on the health of an individual. All of these potential impacts need to be taken into account and evaluated across the spectrum of available fertility control methods, including the acute effects of animal capture and treatment and the longer term, treatment-specific effects whilst the fertility control agent is effective. For example, current fertility control techniques available for use on koalas require capture of an individual. Some contraceptive implant techniques, such as deslorelin, may require multiple captures and re-administration to ensure continued fertility control, whereas surgical sterilisation involves a single capture and treatment procedure for permanent 90

106 infertility. One of the advantages of Suprelorin use is that a contraceptive effect occurs following a single injection without the need to transport or hold the animal over several days. In this species, the capture process, whilst causing short-term effects such as elevated stress levels (Hajduk et al., 1992), is considered to have no lasting impact on an individual s condition (see McDonald et al., 1990). Therefore, the relative impacts on koala health for surgical sterilisation versus a contraceptive implant need to take account of the (presumably) greater impacts of sterilisation initially, versus the need for repeat capture to readminister contraceptive implants throughout the animal s lifetime. The way in which a fertility control agent is administered to an individual animal may also have lasting health effects. Suprelorin contraceptive implants require a single injection without the need of adjuvants, substances that have been shown to cause adverse reactions when used during immunocontraception trials (Cooper and Larson, 2006). Occasionally animals may develop a local reaction at the injection site, e.g. cats developed minimal swelling at the implant insertion site for 3 5 days post-treatment (Munson et al., 2001), but these effects were temporary and not widespread. These potential effects need to be weighed up against the post-operative pain, inflammation and potential for post-operative infection after surgical sterilisation, which may influence short to long-term animal health. The aim of this study was to determine the effect of treatment with a GnRH agonist (Suprelorin ) contraceptive implant on the body weight and muscle condition of free-ranging adult female koalas on Kangaroo Island over time, compared to control koalas and koalas surgically sterilised under current state government management practice. 91

107 5.2 Materials and Methods Study animals Twenty-four adult female koalas were captured at Pioneer Bend as part of the annual DEWNR koala management program in October Only fertile female koalas that had dependent young, and that were between age-class III and V inclusively (3.5-9 years of age (McLean, 2003)) were included in the study. Each female was randomly assigned to one of three treatment groups (n=8 in each group): Group 1, Suprelorin-treated; Group 2, surgically sterilised and Group 3, control animals. There was no difference in initial mean body weights between koalas in each treatment group (mean 7.7 kg, F 2,21 = 3.32, P = 0.056). Suprelorin-treated and control koalas were processed at the capture site. All Suprelorin-treated koalas received one 4.7 mg implant subcutaneously immediately following capture (see section 2.4), while each control female received a single placebo implant to control against any effects of the treatment procedure. The weight, head-length, tooth wear category (for age class determination), and condition of each koala and their dependent young were recorded (see Section 2.3). Following assessment, each adult koala was uniquely eartagged, microchipped (see 2.3) and fitted with a VHF radio-collar (2.8) before being released (with young) into the tree from which it was caught. Koalas were held for a maximum of 30 minutes before release. Animals in the surgical sterilisation treatment group were transported (with young) to the Kingscote Veterinary Clinic where they were held for 1-2 days in holding pens, with unrestricted access to fresh Eucalyptus leaves, before sterilisation (tubal ligation, see Section 2.7). Each koala was ear-tagged, microchipped, fitted with a radio-collar and its weight, headlength, tooth wear class and condition recorded at the time of surgery. Any female young over 1.5 kg also underwent surgical sterilisation (as per standard DEWNR management practice). 92

108 All sterilised females were released (with young) into the tree from which they were caught approximately four hours post-surgery. Koalas were radio-tracked regularly until February 2009 (see Chapter 6) to assess survival of the adult females and their young, and were re-captured at regular intervals (May and November 2007 and 2008, and February 2009) to assess the health and condition of individuals. Suprelorin-treated and control females were re-implanted in November 2007 to ensure they had a functioning implant during the 2007/08 breeding season (control animals received another placebo implant at this time). Metrics used for assessing health and condition Body weight, muscle condition and survival are typical metrics used for non-invasive and efficient assessment of the health and condition of individual koalas (and other species) in the field (see McLean, 2003; Hynes, 2007; Greenfield, 2012; Wilson, 2012). These metrics were considered appropriate for assessing koalas in this study, although it is acknowledged that small samples sizes make firm conclusions challenging. Blood was initially collected from individual koalas in the field (see methodology description in section 2.5) with the aim of assessing biochemical and haematological parameters as a measure of health following treatment, however it became clear that variation within parameters varied greatly and sample sizes were too low for this to be of use. In addition, the time elapsed between sampling of blood in the field and testing at pathology laboratories in Adelaide meant a proportion of samples were not viable by the time they could be analysed. As such, blood collection for this purpose was abandoned early in the study and results are not presented here. 93

109 Body weight change Koalas were weighed at each capture to the nearest 0.1 kg using 25 kg spring scales (Wedderburn CA150 25K). Change in body weight, as a mean percentage change from mean weight at initial capture, was compared between treatment groups over time. The weight of any pouch young (estimated if not removed from the pouch) was subtracted from the weight of each female. A generalised additive mixed model (GAMM) was used to compare the trend in weight change between groups, using the mixed-model approach to account for repeated measurements of individuals. Additive models facilitate the modelling of non-linear relationships (Wood, 2006); inspection of raw data suggested that trends in koala weights through time may not have been linear. GAMMs provide further flexibility via the inclusion of random effects to account for non-independent error structures (Zuur et al., 2009). Time and treatment were included as fixed effects (time variable fitted with a smoothing function to account for the likelihood of non-linear changes to weight through time), while a unique identifier for each individual animal was included as a random effect to account for nonindependent error structures. To account for the possibility of temporal error correlation resulting from repeat measures of individual koalas, the fit of a model with and without a temporal correlation structure was assessed (Zuur et al., 2009); ACF plots showed markedly reduced temporal correlation in the data following the inclusion of the AR-1 correlation structure. Residuals were normal for each treatment group, indicating that use of the default Gaussian error distribution was appropriate. Original capture weights of all koalas were removed from analysis as all animals were deliberately selected as adult females with no significant difference in initial weights between groups. Modelling was performed using the gamm function in package mgcv (Wood, 2006) in R v (R Core Team, 2012). 94

110 Muscle condition The muscle condition of individuals was assessed at each capture by palpating the muscle along the caudal edge of the scapula (muscularis trapezius) and assigning it to one of three classes based on its shape and height relative to the scapula spine (see McLean, 2003). In good condition, the muscle was prominent (i.e. bulging) on either side of the scapula spine, whereas an animal in poor condition had concaved muscle on either side. A koala in average condition had slight bulging of the muscle, but the spine was easy to palpate. Comparisons of muscle condition between treatment groups at each capture were analysed using Fisher s Exact Test. In addition to determining muscle condition, koalas were visually assessed for any sign of infection or inflammatory reactions resulting from the treatment procedure (e.g. discharge, inflammation, necrosis, lesions, etc) or, in the case of control and Suprelorin-treated koalas, retention of the implant (implant site palpated to feel for the implant under the skin). Survival Survival of female koalas subjected to fertility control and the survival of their young to independence were compared to koalas in the control group using Fisher s Exact Test. Independence of young was assumed when the young were repeatedly observed off the mother s back, either foraging or sleeping autonomously. 95

111 5.3 Results Treatment Subcutaneous delivery of Suprelorin implants in the field presented no technical difficulties in this species, and no infection or other adverse reactions at the insertion site were observed in re-captured koalas (including animals given placebo implant). The mean duration of processing a female koala for Suprelorin treatment was 9.5 (± 1.3) minutes, including condition assessments, pouch checks, ear-tagging, microchipping, implanting and blood collection. As has been observed throughout the DEWNR management program (DEWNR, unpublished), no infection or adverse reactions were noted at the surgical site for koalas that were surgically sterilised. Body weight changes During the course of this study there was a significant non-linear change in body weight over time for animals in each treatment group (Table 5.1), with koalas in all groups gaining weight during the non-breeding season following initial capture before decreasing in weight at the conclusion of the following breeding season (Fig. 5.1). There was no overall difference in the weight change trend of Suprelorin-treated (p=0.856) and surgically-sterilised (p=0.242) koalas compared to control animals. Nevertheless, koalas subjected to fertility control appeared to better maintain body weight which was gained during the first posttreatment non-breeding season (i.e. May-August; Whisson and Carlyon, 2010) compared to control animals that show a trend of fluctuating around their original capture weight (Fig. 5.1). The data suggest that there was no negative effect of Suprelorin treatment or surgical sterilisation on the body weight of female koalas. 96

112 Table 5.1: Output from generalised additive mixed model used to investigate changes to body weight through time of adult female koalas subjected to fertility control. s( ) indicates an explanatory variable fitted as a non-parametric smoothing term; e.d.f. represents the estimated degrees of freedom for each smoothing term. Treatment Explanatory variable e.d.f F p-value Control s(time) <0.001 Suprelorin s(time) Surgery s(time) Figure 5.1: Predicted percentage change in body weight of adult female koalas from the initial mean capture weight by treatment group during the course of the study. Large dashed line represents control group (n=8); solid line represents Suprelorin-treated animals (n=8); small dashed line represents surgically sterilised animals (n=7). Horizontal line shows 0% reference. Shading indicates koala breeding season. Confidence intervals are not shown here to avoid confusion in the figure. 97

113 Muscle condition The muscle condition index recorded for each individual at each capture showed no difference between treatment groups at any capture during the study period (Fishers Exact Test, all p > 0.05). Just two animals (both Suprelorin-treated) were found to be in poor condition, where the muscle surrounding the scapula was sunken, and for both of these animals good scores were recorded at subsequent captures. Adult survival One surgically sterilised koala died in January 2007, within 42 days following surgery; however this animal was too decomposed when found to determine the cause of death. This animal had remained within 75 m of its original capture location in high-quality habitat. One control koala was found dead at the base of a tree in August 2008, and the cause of death also could not be determined. This koala had undertaken a long-distance movement of approximately 9 km over five months from December 2007 before re-settling in a new patch of habitat for another five months prior to death (koala 400 ; see Chapter 6). Two koalas (one control and one surgically-sterilised) were found dead, hanging from their collars, in December 2008 and February 2009 respectively (see 2.8). These two individuals were not included in statistical analyses of survival. Survival of female koalas subjected to Suprelorin treatment or surgical sterilisation did not differ to the survival rate of female koalas in the control group (Fisher s Exact test: p=1 and p=1 respectively) over the course of the study (Fig. 5.2). Overall, the mortality rate of female koalas across all groups was 17% during the study period, although this figure decreases to 9% if the collar deaths are not included. 98

114 Survival (%) Control Suprelorin Surgery * * 2006/ / /2009 Breeding season Figure 5.2: Survival of female koalas in each treatment group during the course of the study as a percentage of koalas in each treatment group (n=8 in each group). *one koala in each of the control and surgically-sterilised groups was found dead through causes unrelated to treatment. Survival of young The fertility of female koalas in this study is reported in Chapter 4. All females were carrying back-young when first captured during the 2006/2007 breeding season. There was no difference in the survival of these young in the control and Suprelorin-treated groups (Fisher s Exact Test, p=1), with the majority of these young surviving to independence (Table 5.2). In contrast, 57% of the dependent young of surgically-sterilised females at initial capture survived until independence, with three of seven young not sighted with their mother within three weeks of surgery (although survival did not differ statistically compared to control animals; Fisher s Exact test, p=0.282). All three dependent young (one of these young, a female, was sterilised) were too small to have been weaned during the three week period post-treatment (< 1 kg at capture). The third young was in the pouch, but was never subsequently sighted on the back. In subsequent seasons fecundity of the control group was low (see Chapter 4), and the survival of the young that were born was also poor (Table 5.2). All young lost (not including young at 99

115 original capture) were lost prior to emergence from the pouch. One of the two young produced by Suprelorin treated koalas for which the fate is known did not survive to independence. Table 5.2: Survival rates of known young born from koalas in each treatment group across different seasons. Unk: unknown. NA: not applicable, as females were surgically sterilised. Adult Treatment Season of birth Season of weaning Control (n=8) Suprelorin (n=8) Surgery (n=7) 2005/ /2007* 88% (7/8) 100% (8/8) 57% (4/7) 2006/ / % (1/2) 0% (0/1) NA (0/0) 2007/ / % (1/6) 0% (0/1) NA (0/0) 2008/ /2010 Unk (?/3) Unk (?/1) NA (0/0) * initial capture. 100

116 5.4 Discussion This study found no adverse health effects as a result of treatment with either Suprelorin or surgical sterilisation in free-ranging adult female koalas on Kangaroo Island. Survival rates of treated koalas did not differ to that of control animals, and those koalas either treated with Suprelorin or surgically-sterilised showed a mean increase and then maintenance of their body weight during the course of the study. Thus it appears Suprelorin and surgical sterilisation are safe and effective fertility control techniques for koala management purposes. There was however, some evidence to suggest that surgical sterilisation may affect the survival of dependent young post-surgery, but larger sample sizes would be required to investigate this in more depth. The trends observed in body weight of treated koalas were generally as expected. Adult female koalas exhibit natural seasonal fluctuations in body weight, typically increasing in weight by approximately 3% during the winter months and decreasing toward summer (McLean, 2003). These seasonal fluctuations are likely influenced by a range of factors, including reproductive status and activities (e.g. lactation or mate-searching), thermoregulatory requirements, and increased food intake during winter (Cork, 1986; Krockenberger, 1993; McLean, 2003). As such, female koalas treated with Suprelorin or surgically sterilised, and therefore freed from the energetic costs of lactation, were expected to increase in weight (and maintain this increase) by converting energy to growth rather than using it for maternal provision. Weight gain following fertility control treatment has also been observed in other species, such as immunocontracepted white-tailed deer (McShea et al., 1997) and captive female wild boar (Sus scrofa) treated with the GnRH vaccine GonaCon (Massei et al., 2008). However, Fraker et al. (2002) observed a reduction in kidney fat in immunocontracepted fallow deer (Dama dama) compared to control animals, suggesting a reduction in condition 101

117 possibly related to repeated oestrus events and increased mating activity, but expected that this would be compensated for by avoiding pregnancy and the absence of lactational stress. The results from this study, showing an increase in the weight of female koalas subjected to Suprelorin treatment, are consistent with results from previous studies with this species using this contraceptive technique (see Herbert, 2002; Greenfield, 2012). Weight gain peaked at 7.4% of the original capture weight, higher than the peak of 5.2% for control animals, and over double that observed in untreated Victorian koalas (McLean, 2003). There was no evidence of a decline in body weight of Suprelorin-treated koalas in either of the previous studies, and, as in this study, Greenfield (2012) observed that treated koalas appeared to maintain or increase their body-weight over time, whilst the weight of control animals appeared to fluctuate seasonally. It was expected that female koalas treated with Suprelorin might exhibit a greater increase in weight compared to surgically-sterilised koalas. This might be expected because, in addition to providing contraception, and therefore eliminating the stresses of lactation, Suprelorin treatment should inhibit oestrus altogether (as demonstrated in tammar wallabies; Herbert et al., 2005) and therefore potentially any associated ranging behaviour such as matesearching or avoidance of male harassment. Surgically-sterilised females still undergo oestrus, as the ovaries remain intact and functional, and therefore could be expected to maintain these energetically-costly behaviours. Surprisingly, there was no difference in the weight change trend of Suprelorin-treated versus surgically-sterilised female koalas in this study, which may have resulted from the fact that there was little difference in their ranging behaviour (see Chapter 6) and none underwent late-stage lactation following initial capture. Koalas treated with a synthetic gestagen contraceptive (levonorgestrel), that does not compromise ovarian activity and hence the animals still undergo oestrus, also showed a similar pattern of increasing weight over time (Hynes, 2007), supporting the idea that lactation and the cost of raising young has greater influence on seasonal weight-changes than the energetic costs of oestrus-influenced 102

118 behaviour. The relatively small treatment sizes in this study may have also contributed to the lack of difference between the two groups, and further studies involving more individuals should be conducted to verify the result. Whilst the weight of the control animals fluctuated, it did not follow the strict seasonal pattern that has been observed in previous studies (see McLean, 2003; Hynes, 2007; Greenfield, 2012). Control females gained weight during the first post-treatment non-breeding season (2007) as predicted, however they did not subsequently lose weight at the beginning of the breeding season (the period of greatest lactational burden for koalas; Krockenberger, 2003). This can be explained by the low fertility of this group during this period; just one control female carried back-young, and was therefore undergoing late lactation, at this time. What is surprising is the apparent decrease in body weight that occurred in the control group during the 2008 non-breeding season, however koalas were not captured between May and November 2008 and therefore may not have been constantly losing weight during this period as predicted by the model. Six control females (75%) were carrying pouchyoung when captured in May 2008, although early lactation should not cause such significant weight loss and only one of these young survived to the back-young (late-lactation) stage. It is not known exactly when these pouch-young were lost; if it was consistently close to the November 2008 capture it could explain the loss of weight in the control group, as these koalas would have been undergoing late lactation, however no regressing teats were noted, suggesting loss of young was not recent. Environmental influences, such as reduced rainfall or food availability could have played a role in the weight loss; indeed a high proportion of koalas on Kangaroo Island were in average or poor condition following low rainfall leading up to the breeding season in 2006/2007 compared to previous years (Whisson and Carlyon, 2010), suggesting that koala condition could have been impacted negatively by the effect of low rainfall on food quality. If this was the case here however, these pressures should have been felt equally across groups; 103

119 for example McLean (2003) observed that infertile female koalas (non-lactating) decreased in weight in a similar manner to fertile female koalas at Phillip Island, suggesting this weight change could be due to factors other than lactation. The weight loss exhibited by control animals should be put into context; none of these koalas exhibited poor muscle condition, and, as a percentage of body weight at initial capture, the loss was minor (mean -3.1% (±0.03 SE)). Natural mortality rates of free-ranging adult koalas are typically low ( % per year for koalas in southern Australia; McLean, 2003), likely contributing to the rapid population growth observed in some populations, including Kangaroo Island (Martin, 1985c; Masters et al., 2004). Mortality of adult female koalas in this study was comparable to previous studies (see summary in McLean, 2003), including the only other study examining Suprelorin use in free ranging koalas (Greenfield, 2012). The high survival of koalas in this study contrasts with a mortality rate of 37.5% in the first 12 months for surgically-sterilised koalas that were translocated from Kangaroo Island to mainland South Australia (Whisson et al., 2012; Chapter 7), suggesting movement of animals to a novel environment can have a much greater impact on survival than in situ management. Although a necropsy was not possible on the surgically-sterilised female that died within 42 days of surgery, it is possible, given the timing, that the capture and/or sterilisation procedure may have contributed to this animal s death. The cause of death of the control animal that was found dead in August 2008 was also unknown. This koala had undertaken a long-distance move five months prior to death. Whilst there is a small possibility that the energetic cost of undertaking such a move contributed to this animal s death, this is unlikely as other animals moved similar distances and survived in good condition, and the habitat where she settled was considered to be high-quality (as per Masters et al., 2004). This koala was in 104

120 good condition when caught in May 2008 (muscle score good ) and was carrying a small dependent young in the pouch. The effectiveness of any fertility control program in reducing population density will depend on the balance between survival rates of treated animals and the productivity of the remaining fertile animals (Brown, 2003). The reduced production of young resulting from fertility control may be compensated for by an increased survival rate of remaining adults; as such, the use of in situ fertility control, as opposed to lethal control, effectively delays the desired management outcome, i.e. a reduction in animal density. Fertility control treatments that result in high survival and increased condition of treated individuals, such as Suprelorin treatment or surgical sterilisation of female koalas, may have perceived enhancements to the long-term welfare of the animals (Adderton Herbert, 2004; Duka and Masters, 2005; Fagerstone et al., 2010), but the ultimate goal of a reduction in koala browse pressure is delayed until treated koalas die (unless fertility control treatment is accompanied by translocation). Additional long-term studies are required to determine whether the removal of the energetic requirements for reproduction is translated into an increase in longevity, as has been observed in long-term contracepted wild horses (Kirkpatrick and Turner, 2007). The biannual capture regime applied in this study, capturing koalas close to the beginning and then at the conclusion of each breeding season, may mean that not all births were detected if early mortality of pouch-young occurred prior to examination; therefore the survival rates of dependent young presented here may represent an overestimate of the true survival rate. Some fertility control treatments, such as bromocriptine, can inhibit lactation in various macropod species, leading to the death of dependent young (Tyndale-Biscoe and Hinds, 1984; Curlewis et al. 1986; Hinds and Tyndale-Biscoe, 1994). It is not clear whether Suprelorin treatment disrupts initiation or maintenance of lactation in koalas, however Herbert et al. (2006) found no evidence that deslorelin inhibited lactation in eastern grey kangaroos. 105

121 Low survival of pouch-young in the control group, and low fertility observed in the region more generally during the study period, suggests that environmental factors likely contributed to any pouch-young mortality. The potential for Suprelorin treatment to disrupt lactation in koalas requires further investigation, although this has little bearing on the treatment s potential as a population management tool. Whether Suprelorin provides a true contraceptive, inhibiting pregnancy, or whether it achieves a reduction in population recruitment resulting from death of neonates, the net management result is the same, however the public acceptability of these two modes of action are likely to differ. The potential lower survival of advanced dependent young of surgicallysterilised female koalas following treatment is unexplained. Although it cannot be proven that the three young that were not sighted again within three weeks post-treatment were not successfully weaned, their size suggests they were too young to become independent at this time. The survival of back-young in other fertility control trials involving wild populations has been high (e.g. Hynes, 2007; Greenfield, 2012); despite the small treatment size, the 57% survival of young carried by surgically sterilised females in this study is low in comparison, suggesting a possible issue with this technique. Further trials involving more animals and multiple years are needed to confirm this result; if we assume these survival rates are representative, a sample size of approximately 30 koalas in each group would be required to demonstrate a significant treatment effect, i.e. p 0.05 (power=0.80). As applies to Suprelorin treatment, the fact that surgical sterilisation of females may result in increased mortality of their dependent young does not change the management outcome. If advanced young are lost though, rather than neonates, this may raise animal welfare concerns. 106

122 107

123 6. The effects of two fertility control techniques on the movement patterns and ranging behaviour of female koalas The effectiveness and sustainability of wildlife management strategies hinges on understanding the movement patterns and thus the dispersion biology of individuals within a target population. (Wiggins et al., 2010) 6.1 Introduction Fertility control has been proposed as a more ethically acceptable alternative to lethal methods, and is a particularly attractive option when the target population is a native species and eradication is undesirable (Tuyttens and Macdonald, 1998; Adderton Herbert, 2004; Lauber and Knuth, 2007). Whilst the effects of a range of fertility control techniques on reproductive parameters in various wildlife species have been studied; for example surgical sterilisation (e.g. Bromley and Gese, 2001a; Jacob et al., 2006, 2008), immunocontraception (e.g. Kirkpatrick and Turner, 2008) and endocrine control (e.g. Nave et al., 2002; Middleton et al., 2003; Herbert et al., 2005; Hynes et al., 2010), only a limited number of studies have systematically examined the effects that these techniques may have on an individual s behaviour, particularly movement patterns, in free-ranging animals (e.g. Saunders, et al., 2002; Jacob et al., 2004; Ramsey, 2007; Hynes et al., 2011; Greenfield, 2012). The behaviour of wild animals naturally changes with age and sex and temporal factors such as annual migration (e.g. Mate et al., 2011). It may also be influenced by factors including: previous experience (e.g. Takechi et al., 2009), local or seasonal environmental conditions (e.g. Ey et al., 2009; Knight et al., 2009), resource availability/competition (e.g. Nunes, 1995), population density (e.g. Iossa et al., 2009) and/or predation risk (e.g. Valeix et 108

124 al., 2009). In addition, the health/condition of individuals (e.g. Alvarez and Sanchez, 2003) and physiological (e.g. Spence-Bailey et al. 2007), social (e.g. Fischhoff et al., 2007) or reproductive (e.g. Cooper et al., 2007) status can impact behaviour. Female mammals may alter their behaviour at different times during pregnancy and lactation in response to changing energy and behavioural requirements, such as an increased energy intake to offset mobility difficulties and lactational load or an amplified need to effectively avoid predators (Svare, 1981; McDonald-Madden et al., 2004). Because fertility control modifies an animal s reproductive status, it may therefore alter these requirements. As a consequence, ranging behaviour may also be influenced, since the distances and area in which an animal moves are influenced by its physiological and behavioural requirements (Burt, 1943). Reproduction, in particular lactation, is energetically costly (Gittleman and Thompson, 1988; Krockenberger and Hume, 2007). As such, a logical prediction is that infertile females, with no dependent young, may decrease their movement rate and area used because they don t need to travel as far to meet their energy requirements. This has been suggested for nonbreeding small mammals (Haythornthwaite and Dickman, 2006). Conversely (depending on treatment), infertile females may in fact move farther, possibly in search of males during the breeding season or simply because they are free from the energetic constraints of supporting young. For instance, surgically sterilised ricefield rats (Rattus argentiventer) will occupy a home-range twice the size of control rats (Jacob et al., 2004) and barren female alpine ibex (Capra ibex) tend to use larger home-ranges compared to pregnant females or those supporting highly-dependent young (Grignolio et al., 2007). Such increases in movements however, may only occur if the fertility control technique has not inhibited oestrus, particularly in polyoestrus species. The onset of oestrus has been shown to cause an increase in the home-range size of some female mammals (e.g. Mitchell, 1990b; Fisher and Lara, 1999), and, if oestrus is not inhibited in a polyoestrus species, the sterilised/contracepted female may continuously cycle throughout a breeding season (Whyte et al., 1998), potentially exacerbating the spatial and temporal scale of increased movement. 109

125 Clearly, the behavioural response of different free-ranging wildlife species to fertility control may vary considerably. Ideally, management manipulations should not impact on natural behavioural patterns, in particular to ensure that the technique s influence on population dynamics or exposure of individuals to increased risk is minimised. This was successfully achieved by Ramsey (2007), who found that female common brushtail possums (Trichosurus vulpecula) that were sterilised surgically (tubal ligation or gonadectomy) exhibited similar site fidelity and space-use patterns to untreated fertile females. Behavioural changes caused by application of a fertility control technique may have important implications for population management. For instance, if sterilised individuals of territorial species abandon their territories, there is potential for fertile immigrants to invade, thereby compromising the management effort (Jacob et al., 2004). Saunders et al. (2002) suggest that the fundamental attributes of sociality in foxes (Vulpes vulpes), such as mating systems and territorial integrity, need to remain unchanged following induced sterility for a fertility control program to be successful. Bromley and Gese (2001b) suggest a similar outcome is necessary for successful management of coyote (Canis latrans) populations. There may also be ethical concerns if the behaviour of the target species is greatly altered so that it impacts on animal welfare, health or survival; for example female African elephants (Loxodonta africana) treated with the contraceptive oestradiol remained in a continuous state of oestrus, leading to eviction from their breeding group and increased harassment from bulls (Whyte et al., 1998). Hence, there is a need for thorough investigation into the behavioural response of each target species to a specific fertility control treatment before widespread management application of the method. Some high-density koala (Phascolarctos cinereus) populations in south-eastern Australia require active management to reduce the impacts of browse-damage on their habitat (Martin and Handasyde, 1999; Natural Resource Management Ministerial Council, 2009). Socio-political pressures mean that culling has been rejected as a control technique for koalas, 110

126 and fertility control provides one option for managing these overabundant populations (Natural Resource Management Ministerial Council, 2009). The South Australian Department of Environment, Water and Natural Resources (DEWNR; formally Dept. for Environment and Heritage (DEH)) currently employs surgical sterilisation (tubal ligation) of female koalas as a technique to reduce population growth on Kangaroo Island (see Duka and Masters, 2005), and this method has been used for the management of some Victorian populations in the past (Middleton et al., 2003). Tubally-ligated females still undergo oestrus. This is considered an advantage by advocates of this technique who suggest that hormone-regulated behaviour should ideally remain unaffected (e.g. Chambers et al., 1999). However such methods have been criticised because they result in repeated oestrous cycles, with females continuing to cycle over a very long breeding season, potentially exacerbating stress in the individual (Munson et al., 2005). Nevertheless, there has been little research into the behavioural impacts of this technique on individual koalas, and any post-treatment monitoring has been relatively shortterm (i.e. 9 months) and has tended to focus mainly on post-treatment survival (St John, 1997; Clark, 1998; Duka and Masters, 2005; Whisson et al., 2012). Contraceptive implants have been proposed as a cost-effective, less invasive alternative to surgery and several have been trialed in koalas (e.g. Handasyde, 1986; Herbert, 2002; Middleton et al., 2003; Hynes, et al., 2010; Greenfield, 2012). Herbert (2002) found that treatment of captive female koalas with gonadotrophin-releasing hormone (GnRH) agonist implants stopped reproduction for at least two years by inhibiting the release of the reproductive hormones. Another contraceptive, the gestagen-based levonorgestrel implant, controls fertility in free-ranging female koalas for at least six years (Middleton et al., 2003; Hynes et al., 2010). However, Hynes et al (2011) found that all levonorgestrel-treated female koalas undertook at least one very long-range movement leading up to or during a breeding season, compared to only a small proportion of untreated control animals (19%). Rather than a direct consequence of levonorgestrel-treatment however, this behaviour was highly correlated with the absence of a back-young. Because all long distance movements occurred just prior to 111

127 or during the breeding season, and ovarian activity of levonorgestrel-treated females was not compromised by the treatment, it was suggested that these movements were associated with reproductive activity, in particular a response to oestrus and the elevation of plasma concentrations of oestrogen (Hynes et al., 2011). This phenomena has also been observed in other taxa; for example female white-tailed deer, Odocoileus virginianus, may increase their activity and expand their home during the rut (Labisky and Fritzen, 1998; D'Angelo et al., 2004), including large excursions outside their home range at the time of conception (Holzenbein and Schwede, 1989; Labisky and Fritzen, 1998; D'Angelo et al., 2004). If female koalas in oestrus move greater distances when freed from the restrictions of carrying back-young (Hynes et al., 2011) and the energetic costs associated with lactation (see Krockenberger, 1993; Logan and Sanson, 2003; Krockenberger and Hume, 2007), then females that have been surgically sterilised using tubal ligation (which still undergo oestrous) should exhibit similar behaviour. As female koalas are polyoestrus (Smith, 1979; Handasyde et al., 1990a), this behaviour should occur throughout the breeding season. In contrast, female koalas treated with a GnRH agonist contraceptive implant, which act in other species by shutting down ovarian function and inhibiting the release of reproductive hormones (Herbert et al., 2006), should be less likely to display this behaviour. Should this hypothesis prove correct, this would be one potential benefit of using GnRH agonist treatment compared to other fertility control methods. Because adult female koalas naturally show high site-fidelity (Mitchell, 1990b), a technique that increases the probability that contracepted koalas remain in situ following treatment (i.e. little change in natural behaviour) would enhance the viability and general acceptance of this as an ethical management option. Further, in a resource-limited environment, competition from contracepted animals that remain resident may reduce the opportunity for fertile females to invade a managed area. For example, Mitchell (1990b) observed that young female koalas often take up residence in close proximity to their mothers, increasing local population density. 112

128 If resident adult females are not breeding, increases in local population density may not be so rapid, an outcome that would further enhance the management result. An additional cost with increased ranging behaviour in some areas is that it may also expose koalas to an increased chance of mortality from exposure, vehicles or dog attack, or a koala may inadvertently find itself in unsuitable habitat (e.g. Dique et al., 2003). Should treatment with a GnRH agonist result in females maintaining normal behaviour patterns within their original home-range, this should limit the chance of contact with novel hazardous events or features. Such an approach may be considered a more ethically acceptable management option. Conversely, there is also the potential for GnRH agonist treatment to have adverse behavioural repercussions. Fertility control techniques that interfere with endocrine control of reproduction may alter dominance hierarchies, or lead to higher fertility of subordinate individuals (Caughley et al., 1992; Chambers et al., 1999). Only four published papers have documented the behaviour of wildlife following GnRH agonist treatment (Baker et al., 2002 and Conner et al., 2007, elk (Cervus elaphus sp.); Bertschinger et al., 2002, fennec fox (Vulpes zerda); Woodward et al., 2006, eastern grey kangaroos (Macropus giganteus)). These studies did not detect any significant behavioural side-effects associated with sexual interactions, social behaviour and daily activity patterns, but were limited to species with relatively complex social hierarchies. Since koalas are non-social, do not actively defend a territory, and there is little evidence of a female social hierarchy (Mitchell, 1990a,b), individuals generally only interact briefly during the breeding season. For this reason, side-effects of GnRH agonist treatment on social behaviour may not be obvious, particularly in free-ranging koalas. Due to the limited scope of a koala s behavioural repertoire (Martin and Handasyde, 1999), a relatively straightforward behavioural response is needed to accurately monitor an individual s behavioural response to treatment. An analysis of movement patterns and ranging behaviour of koalas following treatment would provide such a measure. 113

129 Here we investigate the effect of a commercially available GnRH agonist treatment (Suprelorin contraceptive implant; Peptech Animal Health Ltd, Virbac (Australia) Pty Ltd) on the ranging behaviour, movement patterns and site fidelity of female koalas on Kangaroo Island. While Herbert (2002) showed that implants releasing the GnRH agonist deslorelin control fertility in captive female koalas, the effect on fertility and behaviour in free-ranging koalas is unclear. Movement patterns of GnRH agonist treated females were compared to control animals, as well as females that were surgically sterilised (tubal ligation) as part of the current management regime. It was predicted that the movement rate, distance travelled and area of core home-range of adult female koalas treated with a GnRH agonist will be reduced compared to control animals, while the occurrence of any long-distance movements would be greater in those females surgically sterilised under the current management program. 114

130 6.2 Materials and Methods Study area Female koalas were captured on private property at Pioneer Bend (35º43 33 S, 137º15 31 E) in the Cygnet River catchment on Kangaroo Island, approximately 35 km west of Kingscote (Fig. 6.1). This area is considered high-quality habitat for koalas (see Masters et al., 2004), supporting the highest density of koalas on the island (currently up to 3.4/ha, previously up to 13/ha; DEWNR, unpublished data 2007), although eucalypt condition was improving at the time of the study due to intensive management (DEWNR, unpublished data 2007). The habitat is characterised by linear strips of woodland vegetation dominated by rough-barked manna gum (Eucalyptus viminalis cygnetensis), South Australian blue gum (E. leucoxylon) and pink gum (E. fasciculosa) along the permanent watercourses, and woodland patches that may contain a mix of brown stringy-bark (E. baxteri), messmate stringy-bark (E. obliqua), cup gum (E. cosmophylla), and pink gum under an overstory of sugar gum (E. cladocalyx) at the upper reaches of more ephemeral creeks. Isolated or small patches of eucalypts occur throughout the surrounding pasture, which is predominately used for grazing stock. Large areas of mature plantation pine (Pinus radiata) are situated to the north. 115

131 Figure 6.1: Location of the study site at Pioneer Bend, Kangaroo Island, South Australia. Average annual rainfall for Pioneer Bend is mm (BOM, 2009a, 2009b). Mean monthly maximum temperatures range from 13.8 C in August to 27.7 C in February (BOM, 2013). The area experienced severe drought in 2006 (nearby Parndana received its lowest annual rainfall on record: mm), while below-average rainfall was also recorded in 2007 and 2008 (see Figure 2.3, Chapter 2). Rainfall in the area leading up to and during the peak period of the 2006/2007 koala breeding season was just 41% of the historical median (BOM, 2009a, 2009b). Mean monthly maximum temperatures were the highest on record in September and October 2006 (19.8 C, 22.3 C respectively), December 2007 (27.1 C) and March 2008 (28.3 C) (BOM 2009a, 2009b). Mean annual maximum temperatures were also above average ( ) in all years of this study, with the 2007 mean (22.0 C) the highest on record (BOM, 2013). 116

132 Capture, collection of morphometric data and treatment Twenty-four adult female koalas were captured in October Only fertile female koalas (carrying dependent young) that were between age-class III and V inclusively (3.5-9 years of age; see McLean, 2003) were used to reduce potential variance in movement due to age. Each female was randomly assigned to one of three treatment groups (n=8 in each group): Group 1, Suprelorin-treated; Group 2, surgically sterilised (tubal ligation); and Group 3, control (i.e., placebo implant) animals. There was no difference in initial mean body mass between the treatment groups (mean 7.7 kg, F 2,21 = 3.32, P = 0.056). All koalas were captured from very similar habitat within a relatively small area (<1000 ha); therefore it was assumed that all individuals had equal access to similar resources at the commencement of the study. Suprelorin-treated and control koalas were processed at the capture site. All Suprelorin-treated koalas received one 4.7 mg implant subcutaneously immediately following capture (see section 2.4), while each control female received a single placebo implant to control against any effects of the treatment procedure. The weight, head-length, tooth wear (for age class determination), and condition of each koala and dependent young were recorded (see Section 2.3). Following assessment, each adult koala was uniquely ear-tagged, microchipped (see 2.3) and fitted with a VHF radio-collar (2.8) before being released (with young) into the tree from which it was caught. Koalas were held for a maximum of 30 minutes before release. Animals in the surgical sterilisation treatment group were transported (with young) to the Kingscote Veterinary Clinic where they were held for 1-2 days in holding pens, with unrestricted access to fresh eucalypt leaves, before sterilisation (tubal ligation, see Section 2.7). Each koala was ear-tagged, microchipped, fitted with a radio-collar and its weight, headlength, tooth wear class and condition recorded at the time of surgery. Any female young over 1.5 kg also underwent surgical sterilisation (as per standard DEWNR management practice, see 117

133 Chapter 2). All sterilised females were released (with young) into the tree from which they were caught approximately four hours post-surgery. Animal observations Each radio-collared koala was located regularly (mean 10.1 ± 0.17 days between successive radio fixes) from October 2006 (i.e., initial treatment) to June 2008 during daylight hours (see section 2.8). Koalas were only tracked intermittently from July 2008 until May 2009, so these data were not used in analysis of movements. A minimum of six days was allowed between successive sightings of any individual to minimise the potential for autocorrelation of locations. It took up to three days to locate all koalas and the order of searching for koalas was randomised within the three days. The location ( fix ) of each koala was recorded (< 10m accuracy) with a hand-held global positioning system (GPS 12, Garmin Ltd, UK). Each female koala (and any dependent young) was recaptured in May and November 2007, May and November 2008, and June 2009 to assess body weight, condition and reproductive state (see Section 2.3). Suprelorin-treated and control females were re-implanted in November 2007 to ensure they had a functioning implant during the 2007/08 breeding season (control animals received another placebo implant at this time). The weight and headlength of dependent young were also recorded. One surgically sterilised koala died in January 2007, within 42 days following surgery; however this animal was too decomposed when found to determine the cause of death. This animal was excluded from analyses. 118

134 Analysis Site fidelity and long-distance movement The linear distance of each fix from the original capture point was calculated for each animal, as a measure of distance travelled and site-fidelity. Plotting these data over time enabled a visual examination of periods of spatial stability (or core activity areas) in addition to any large changes in the area used (long-range movements) by koalas (see Fig. 6.4). This was examined for each animal with a long-distance movement identified relative to each individual s normal movement patterns. Thus, a large, steep positive slope represented a relatively large movement away from the capture point and a large negative slope indicates an animal may be returning towards the original site of capture. A long-distance movement by an individual koala was considered to have taken place if (i) there was an obvious positive or negative slope between two relatively stable periods, and (ii) the distance between these stable periods was at least double that of the distance variation within a stable period. This method assumes that the original capture point was located within a current stable range of that animal; female koalas captured with large dependent young are unlikely to be undertaking large movements due to the energetic demands of late-lactation and carrying young (Hynes et al., 2011). The timing of any long-distance movements was specified simply as the starting date of any steep slope that indicated such a movement. The distance between two spatially stable periods (to establish the size of any considerable movement) was determined by calculating the straight-line distance between the geographic centres of location for all points within each period of stability (the mean easting and northing). For koalas that undertook more than one long-distance move during the course of the study, the mean distance was used. The duration of a long-distance movement was calculated simply as the time covered by a slope between two stable periods. The effect of treatment on distance and duration of long-distance movement was analysed in a one-way analysis of variance (ANOVA). The proportion of 119

135 koalas in each group that undertook at least one long-distance movement was calculated and the frequencies compared between groups using contingency tables. To compare site fidelity and dispersal by koalas in each treatment group as a function of time, a generalised additive mixed model (GAMM) was used. An additive model was used to account for the apparent non-linear change in distance from capture for many animals over time, and the mixed model used to account for repeated measures of individual koalas. A unique identifier for each koala was used to account for non-independent error structures (Zuur et al., 2009), and a gamma distribution was used as the residual diagnostic plot indicated the default Gaussian error distribution was not appropriate. Plots of auto-correlation functions (Zuur et al., 2009, 2010) revealed significant correlation at various time lags in the data set. Therefore, an auto-regressive correlation structure (AR-1) was included in the model to account for this lack of independence (Zuur et al. 2009). Subsequent autocorrelation function (ACF) plots showed markedly reduced temporal correlation in the data following the inclusion of the AR-1 correlation structure. Modelling was performed using the gamm function in package mgcv (Wood, 2006) in R v (R Development Core Team, 2012). Movement rate A mean distance between fixes was calculated for each breeding season and the nonbreeding season to investigate changes in movement rate between seasons. Data was fourthroot transformed to meet the assumption of normality and was compared between treatment groups and between seasons using multifactor between-subjects ANOVA. Home-range Home range was estimated using cluster analysis, a nearest-neighbour link-distance method that defines high-use, outlier-exclusive core areas in multinuclear ranges (Kenward et al., 2001). This method was chosen due to the patchy distribution of habitat and potential for koalas to move between multiple disjunct patches, and because it has the lowest and least 120

136 variable inclusion of unusable habitat and low correlation with range-span compared to other methods when examining location data from species with restricted habitats (e.g. Hodder et al., 2007; Knight et al., 2009). This method of excluding outliers assumes that movement inside and outside core areas involve different activities (Kenward et al., 2001) and it is a particularly useful method for identifying patchiness in habitat use (Kenward, 2001). The resulting core areas are assumed to be those areas in which each koala undertook its normal activities, an accepted definition of home-range (Burt, 1943). Cluster polygons were created using fixes within 95% of the nearest neighbour distance distribution (i.e., outliers excluded), and total range estimated as the sum of these polygons (analogous with plotting contours or ellipses to 95% of the density distribution (Kenward et al., 2003)). Cluster analysis was performed in Ranges 6 (Kenward et al., 2003). All fixes for each individual (October 2006 to June 2008) were used to generate a single range estimate. There were insufficient fixes within breeding and non-breeding seasons to determine seasonal ranges. The effect of fertility control treatment on the size of core home-range was examined using one-way ANOVA. 121

137 6.3 Results Koalas were located 1279 times during the course of the study (November 2006 to June 2008), with a mean of 55.6 ± 0.84 sightings per koala. Site fidelity and long-distance movement Several distinct movement patterns were observed in female koalas on Kangaroo Island during the course of this study (see Fig. 6.2 for examples). Some females showed strong site-fidelity and remained in close proximity (i.e. median < 50 m) to their original capture site (e.g. 500 in Fig. 6.2), while others spontaneously moved long distances (up to 14 km) and appeared to settle in new habitat (e.g. 620 ). Two moved long distances then subsequently returned to their original capture area, i.e. made a long-range excursive movement (e.g. 300 ), similar in profile to movement patterns observed in migratory species (e.g. Bunnefeld et al., 2011). Two-thirds (65.2%) of all collared females undertook at least one long-distance movement during the course of the study. Only two of 15 koalas (13.3%) that undertook a long-distance movement subsequently returned to their original capture site ( 300 and 970, see Fig. 6.3), whilst the remainder re-settled in novel locations. Four individuals ( 300, 320, 560 and 970 ) made two or more long-distance moves separated by periods of stability. There was no difference in the proportion of koalas from each treatment group that undertook at least one long-distance movement during the study (breeding seasons pooled) (Fisher s Exact Test: P = 0.856) (Fig. 6.4). 122

138 300 # # # # N # ## # ### # # # # # # # # # # # # # # # # # # # # # # # # # # # km # ### # #### # # # # # # ## ### # # ### # # # # # # Figure 6.2: Examples of the different movement patterns exhibited by female koalas during the study in the Pioneer Bend area, Kangaroo Island. Black dot indicates site of initial capture. Coloured lines indicate movement path of an individual koala from November 2006 to August 2008 (Red = koala 500, Yellow = 620, Green = 300 ). Plots show the distance (m) from original capture site over time with the y-axis scale equal for each koala. Koala 500 exhibits a stable spatial pattern with strong site-fidelity. 620 demonstrates a spontaneous movement and re-settlement away from her original capture site (i.e. dispersal ), while 300 shows a spontaneous movement and re-settlement followed by a return to her original capture area. Pink circles show site of re-settlement. Blue line on the x axis illustrates the period when the koala is carrying back-young, purple line illustrates the period when the koala is carrying pouch-young. All koalas exhibited a variation of one of these three main patterns (see figure 6.4 for plots for all individual koalas). 123

139 Control Suprelorin-treated Surgically-sterilised Figure 6.3: Plots for individual koalas showing distance (m) of each fix from original capture point (y-axis) over the course of the movement study November 2006 July 2008 (x-axis). Green shading indicates breeding season. Red line indicates 124 period koala is carrying back-young. Orange line indicates period koala is carrying pouch-young.

140 Proportion of females undertaking a long-distance movement (%) Control Suprelorin Surgery 6/ /7 4/8 4/7 5/8 4/8 4/8 30 2/ / / /2008 Pooled (All carrying dependent young at initial capture) Breeding season Figure 6.4: Percentage of female koalas in each treatment group that undertook a long-distance movement during each breeding season and then both seasons pooled. The number of females in each group is shown above the bar. Of those koalas that undertook long-distance movements (15/23), the maximum straight-line distance that a koala was located from its original capture point was 14.3 km (mean = 6.1 ± 1.0 km), and a number of females moved beyond the Cygnet River catchment. None of the females were carrying a young at the time of their movement. There was no effect of treatment on the observed maximum distance moved from capture (F 2, 12 = 0.745, P = 0.495), or on the mean length of time spent on a long-distance movement (F 2, 12 = 1.206, P = 0.33). Of the koalas that moved, the mean duration spent travelling was (± 14.1) days (range = ; Fig. 6.5). The majority of all long-distance movements (94.4%) by female koalas in this study commenced in the period leading up to or during a breeding season between September and March (Fig. 6.6). 125

141 Duration of movement between spatially stable periods (days) Control Suprelorin Surgery Treatment Figure 6.5: Mean duration of long-distance movements (where applicable) for female koalas in each treatment group at Pioneer Bend, Kangaroo Island. The number of females in each group that undertook a long-distance movement is shown above the bar. Error bars represent one s.e. Figure 6.6: Frequency of long-distance movements by female koalas (treatments pooled) that began in each month. Grey shading indicates typical breeding season of koalas on Kangaroo Island (see Whisson and Carlyon, 2010). 126

142 Compared to control koalas, there was no statistical difference in the distance moved from the original capture site as a function of time over the course of the study for both Suprelorin-treated (p=0.657) or surgically-sterilised (p=0.443) koalas (Fig. 6.7). All groups demonstrated a non-linear change in distance from original capture site over time (Table 6.1). It appeared that Suprelorin-treated animals in particular reduced their movement during the 2007 non-breeding season, and movement away from the original capture site appeared to plateau mid-way through the 2007/2008 breeding season for this group, suggesting ranging behaviour slowed at this period (Fig. 6.7b). The model predicted that, over the course of the study, koalas from all groups would move at least 3 km from their original site of capture. Table 6.1: Output from generalised additive mixed model used to investigate changes to the distance from original capture site for female koalas subjected to fertility control. s( ) indicates that an explanatory variable was fitted as a non-parametric smoothing term; e.d.f. represents the estimated degrees of freedom for each smoothing term. Treatment Explanatory variable e.d.f F p-value Control s(time) <0.001 Suprelorin s(time) <0.001 Surgery s(time)

143 Figure 6.7: GAMM-predicted mean distance from original capture of female koalas in each treatment group as a function of time (a. control; b. Suprelorin-treated; c. surgically sterilised). Dashed lines represent one s.e. 128

144 4th root distance (m) Movement rate The overall mean distance between successive fixes, or step-length (movement rate), was greater in both Suprelorin-treated and surgically sterilised koalas compared to controls (F 2, 1269 = 6.256, P = 0.002; Figure 6.8). A plot of the estimated marginal means for movement rate of each group in each season suggests this result is due to a difference in the first breeding season only (Figure 6.9). Post-hoc tests showed there was no difference in the movement rates of animals subjected to fertility control (Suprelorin-treated and surgically-sterilised) (p = 1.000). The movement rate of all groups was significantly greater during the breeding seasons compared to the non-breeding seasons (F 2, 1269 = , P < 0.001), and post-hoc tests (Tukey s test) revealed movement rates of koalas were higher within all groups during the 2007/2008 breeding season compared to the 2006/2007 breeding season (p = 0.003). There was a significant interaction between treatment and season (F 4, 1269 = 3.472, P = 0.008) control suprelorin surgery Treatment Figure 6.8: Mean distance (in metres) between successive fixes over the course of the study (seasons pooled) for female koalas within each treatment group. Number above bar indicates number of fixes of all individuals in treatment group (control: n=8; Suprelorin: n=8; Surgery: n=7). 129

145 4 distance (m) Figure 6.9: Estimated marginal means of movement rates (distance between successive fixes) for female koalas in each treatment group for each season. Although there was a difference in the mean rate of movement between groups for the entire study period, there was no difference in movement rate between groups as a function of time (p=0.373 and for Suprelorin-treated and surgically-sterilised koalas respectively, compared to control animals). The GAMM showed a non-linear change to movement rate through time for all groups, supporting the pattern shown by the estimated marginal means of movement rate by season. Core home-range size Home range estimates varied between individuals within groups, however there was no effect of treatment on the core area used by female koalas as determined by cluster analysis (F 1, 21 = 0.71, P = 0.41; Fig. 6.10). The minimum core area used by an individual koala (treatments pooled) was 0.20 ha and the maximum was ha (median = 16.7 ha). All but two individuals had multi-nuclear core ranges (i.e. more than one cluster of heavily-used locations). 130

146 Mean core home range (ha) The estimates of home-range differed dramatically depending on the method used; fixed-kernel analysis provided a mean estimate of 1325 ha (treatments pooled), compared to a mean of 65 ha given by cluster analyses. To facilitate comparison with other studies, estimates of homerange for each animal using both fixed-kernel estimation and cluster analysis are presented in Appendix Control Suprelorin Surgery Treatment Figure 6.10: Mean size of core home-ranges as estimated by cluster analysis for female koalas in each treatment group at Pioneer Bend, Kangaroo Island. The number of females in each group is shown above the bar. Error bars represent one s.e. 131

147 6.4 Discussion This study confirmed that movement patterns in female koalas vary temporally, and found little impact of Suprelorin treatment or surgical sterilisation on spatial behaviour on Kangaroo Island. Two-thirds of all study koalas, irrespective of treatment group, undertook at least one long-distance move during the study. In most cases these movements were associated with the breeding season and all occurred at a time when females were not carrying young. Although the proportion of females displaying this behaviour was comparable between groups, movement rates were higher for Suprelorin-treated and surgically-sterilised animals than control animals. Regardless of treatment, individual female koalas moved unexpectedly long-distances leading up to or during the breeding season. Some long-distance movement was expected in those koalas that had been surgically sterilised, potentially as a result of mate-seeking behaviour of oestrus females (Martin and Handasyde, 1999; Mitchell, 1990b). However, longdistance movements were not expected in the Suprelorin group, because the GnRH agonist should inhibit the release of all reproductive hormones and hence the hypothesised trigger for any long-distance movement (Hynes, 2007). Long-distance movements by Suprelorin-treated female koalas leading up to and during the breeding season were also observed during a concurrent study on French Island (Greenfield, 2012), and comparable movements were recorded using a gestagen-based contraceptive in a study on French Island during (Hynes et al., 2011). Both studies linked these movements to females not being burdened by back-young, freeing them to undertake long-distance movements, rather than resulting from the direct effect of fertility control treatment. In the current study, only one female in the control group carried a backyoung, reflecting low reproductive success observed throughout the region during

148 (11% fecundity; DEWNR, unpublished data); as such, the effect of back-young presence on long-distance movement could not be assessed. Chapter 4 discusses the low fecundity observed in placebo-treated animals in more detail. Long-distance movements by female koalas in my study occurred immediately prior to and during the breeding season, a phenomena also observed elsewhere (Hynes et al., 2011; Greenfield, 2012). This consistent pattern strongly supports the idea that such movements are associated with reproductive activity, and may be attributed to a response to oestrus such as mate-searching (Hynes et al., 2011). Whilst this behaviour is consistent with an animal that historically occurs at low densities (Strahan and Martin, 1982), it does not explain why half (4/8) of all females treated with Suprelorin, which acts by inhibiting the oestrous cycle, undertook such a movement during this study. If the stimulus for undertaking a long-range movement is related to both oestrus and/or the absence of back-young, this result suggests that a lack of back-young, and liberation from the associated energetic constraints, may be the more powerful driver of this ranging behaviour. This behaviour may also be influenced by temperature; energy requirements and foliage intake are greater in both lactating and nonlactating females in winter than summer, presumably due to demands of thermoregulation (Krockenberger, 2003). As such, with less energy required for thermoregulation, koalas (at least those in the southern parts of their range) may increase their movements as temperatures increase in spring, coinciding with the start of the breeding season. The majority of long-distance movements made by female koalas during this study appeared to be emigration, or dispersal (i.e. breeding dispersal, see Greenwood, 1980), rather than excursive type movements (e.g. Kolodzinski et al., 2010). Breeding dispersal refers to the movement of parous individuals between successive breeding sites (Greenwood, 1980), as long as the place of reproduction is not an exploratory movement or extension of home-range boundaries (Lidicker, 1975). Only two of the 15 koalas that undertook a long-distance movement subsequently returned to their original capture site during the course of the study, 133

149 whilst the remainder re-settled elsewhere. Whether this represented temporary or permanent dispersal is unknown. These observations contrast strongly with previous studies, which concluded that female koalas older than three years show high site-fidelity and remain philopatric after they first reproduce, comprising the majority demographic of a resident koala population (e.g. Eberhard, 1972; Gall, 1980; Martin, 1985c; Gordon et al., 1990; Mitchell and Martin, 1990; Dique et al., 2003), findings supported by Fowler et al. (2000) who found koala populations in south-east Queensland are structured spatially along matrilines, even over relatively small distances. However, the results of our study are similar to observations made during a concurrent study on French Island (Greenfield, 2012). This may simply reflect substantial variation in behaviour across populations or differing environmental pressures. Whilst long-distance movements by adult female koalas have been observed during other studies on Kangaroo Island, the proportion of female koalas undertaking these moves was much lower and the behaviour was not referred to as dispersal. Eberhard (1972) observed that two of eight adult female koalas that he regularly sighted in Flinders Chase (western end of Kangaroo Island) began to frequent a new set of trees during the period over which they were under study, i.e. exhibited dispersal-type behaviour. Horgan (1998) tracked 16 adult females (eight control, eight surgically sterilised) in the Cygnet River area for a period of eight months (August 1997 to March 1998). Fifteen koalas remained within stable home ranges for the duration of that study, whilst a single animal (no treatment), undertook several short excursive movements before shifting her home range to a new area of core habitat at the beginning of the 1997 breeding season. Likewise, of seven adult females ( age-class III) that were surgically-sterilised and tracked by Clark (1998) in the same area, none underwent dispersal. Carney (2010) successfully tracked twelve adult females (unspecified fertility status; at least six surgically sterilised) at Pioneer Bend for a period of 30 months (September 1997 to February 2000), observing dispersal-type behaviour in just three of these animals (25%). In contrast to the current study, these three koalas were never observed re-settling in a stable home-range (Carney, 2010). 134

150 The driver for the high level of dispersal exhibited by adult female koalas in this study is unclear. In other cases where long-distance movements by adult koalas (both male and female) have been previously reported (Gall, 1980; Martin, 1985b; Dique et al., 2003; Hynes et al., 2011; Greenfield, 2012), typically the underlying motive could not be clearly identified. For most wildlife species, basis for dispersal generally falls into three broad categories: (1) competition for mates, (2) competition for resources, and/or (3) avoidance of inbreeding (Johnson and Gaines, 1990). In koalas, dispersal is typically dominated by juvenile animals between one and three years of age (representing natal dispersal) and is usually male-biased (summaries in Dique et al., 2003; Tucker, 2008). This is consistent with dispersal theory; in a polygynous species such as the koala (Mitchell, 1989), dispersal is traditionally expected to be male-biased to increase the male s likelihood of finding new mates, whilst females are expected to be philopatric (Greenwood, 1980). Female dispersal (whether natal or breeding) in polygynous systems is typically attributed to resource competition or inbreeding avoidance rather than competition for mates, particularly in high-density populations. Resource competition was not anticipated in this study; overall browse pressure in the area was not acute (a result of local koala densities being reduced in recent years through translocation of animals under the KIKMP (DEWNR, unpublished data)), and the habitat where animals were initially caught was considered to be high quality (Masters et al., 2004). However, the effects of low rainfall (see Fig. 2.3, Chapter 2) and high temperatures during the study period may have reduced the amount and/or nutritional quality of available browse, potentially increasing competition between koalas and leading to increased ranging behaviour or dispersal, i.e. competition avoidance. Drought and heatwaves may affect a species directly or change the quality of habitat and/or the availability of resources (Adams, 2010; Albright et al., 2010). These events have been shown to affect the quality of nutrients and moisture available in the koala s diet (Cork and Braithwaite, 1996; Moore and Foley, 2000) and large-scale declines in koala density and 135

151 direct mortality have been observed during drought conditions in Queensland (Gordon et al., 1988; Clifton, 2010; Seabrook et al., 2011). Rather than documenting dispersal-type behaviour however, these studies found that the distribution of koalas during drought contracted to critical riparian habitat, whilst Gordon et al. (1988) also observed large-scale mortality which was not observed in my study. In an assessment of the impacts of future climate change on koalas in south west Queensland, where incidence of severe drought is predicted to increase, Seabrook et al. (2011) suggest the maintenance of core habitat along creek lines with permanent waterholes will become increasingly critical in the coming decades. This study was conducted in the Cygnet River catchment in habitat considered highquality and all koalas were caught in riparian vegetation bordering permanent watercourses and ephemeral creeks. As such, these animals could be expected to contract their movements to nearby vegetation in proximity to more permanent waterholes or drainage lines during the period of below-average rainfall that occurred during the study. However, no contraction in range was observed. In fact, this prolonged period of low rainfall resulted in the majority of local waterholes drying up (K. Carlyon, pers. obs.), and some long-distance movements observed may have represented searching behaviour by koalas looking for higher quality browse. Koalas on Kangaroo Island may therefore be more prone to wandering behaviour during dry periods due to the general lack of permanent standing water. Infrequent female dispersal or excursive behaviour in polygynous species may also be associated with mate-searching (e.g. Kolodzinski et al., 2010; Naulty et al., 2013), and has been proposed for female koalas (Mitchell, 1989; Lee et al., 1990; Mitchell, 1990b; Martin and Handasyde, 1999; Ellis et al., 2011; Hynes et al., 2011). Even in populations with balanced sex-ratios, some females may still need to search for prospective mates because the relative abundance of mature males to reproductively mature females may be low. If most males are preoccupied with receptive females, then females entering oestrus may be forced to engage in active mate-searching behaviours (Kolodzinski et al., 2010). Similarly, Holzenbein and 136

152 Schwede (1989) suggested that the behaviour of female white-tailed deer during rut may serve as an index of male breeding performance, i.e. if a large proportion of females exhibited increased mobility during peak-rut this may indicate an insufficient abundance of breeding males, whereas low mobility of females could indicate sufficient male presence. This hypothesis was supported by Labisky and Fritzen (1998) who documented increased mobility, attributed to mate-searching behaviour, in female white-tailed deer during the rut, particularly in low-density populations. Mitchell (1990b) observed a shift in home-range (i.e. to an area not contiguous with an individual s previously observed home-range) by three of 11 female koalas that were monitored over a period of four years, however this was only after a number of koalas (male and female) were removed from the area as part of ongoing population management. These shifts may suggest an effect of density on koala movement or may be an example of female mate-searching following a reduction in availability of potential mates. In contrast, no koalas were removed from the Pioneer Bend area under the KIKMP during the course of this study. The dispersal movements of female koalas in this study illustrate the species capacity to traverse large areas of cleared land or unfavourable habitat, suggesting that koalas are not reliant on corridors of preferred habitat to move through a landscape. As has been observed in previous studies (e.g. Hynes, 2007; Carney, 2010), koalas used isolated paddock trees, discrete habitat patches and even large tracts of unsuitable habitat such as pine plantation. This is similar to the habitat use of koalas translocated from Kangaroo Island to mainland South Australia that did not have established home-ranges (Whisson et al., 2012). The dispersal movements undertaken by female koalas in this study, and the linear and fragmented nature of suitable habitat, complicated analyses of home-ranges of the different individuals. Routine density-based analysis of each individual koala s home-range using parametric ellipses or contours based on location density distributions (e.g. bivariate normal 137

153 ellipses, Jennrich and Turner, 1969; and kernel/harmonic mean analyses, Dixon and Chapman, 1980), and subsequent comparison between treatment-groups, was deemed inaccurate due to many of the study animals shifting their core-use areas during the study. These techniques assume a continuous density distribution and as such can be heavily influenced by outliers that represent excursive behaviour, rather than core home-range. When trialed using data from this study this commonly resulted in home-range estimates that included substantial areas of habitat that was unsuitable for koalas (e.g. open paddock) and hence were not biologically realistic. Additionally, investigations of the data applying the commonly-used non-parametric minimum convex polygon method (Dalke and Sime, 1938), that uses a minimal-link peripheral polygon to estimate home-range size, also produced results that included large areas of unsuitable koala habitat. These two techniques were primarily designed for species that move freely throughout a landscape; however, while koalas are more than capable of crossing large open patches of ground, they are essentially restricted to patches of eucalypt vegetation for the majority of their normal activities. The overestimations of habitat use by these traditional home-range analyses may also be a reflection of the often linear nature of the habitat at Pioneer Bend (Knight et al., 2009). Cluster analyses, that defined core areas in multinuclear ranges, produced the most satisfactory estimates of home-range size for the majority of animals, however this likely makes comparison with other studies, that typically used location density distributions, less precise. There was no difference in the size of the core spatial area used by female koalas between groups during this study, however a review of home-range estimates for koalas from other studies (Table 6.2) shows that home-ranges of female koalas in this study were at the larger end of the scale, and are the largest observed on Kangaroo Island. This was potentially a result of the prolonged drought that occurred during the study period and the increased ranging behaviour that may have resulted from a search for quality browse as discussed above. 138

154 Table 6.2: Summary of home-range size estimates (ha; in ascending order) for adult female koalas reported from other studies at various locations around Australia. HR = home range; HM = harmonic mean; FK = fixed kernel. MedCP = median convex polygon. # Mean seasonal home range. *Mean annual home range. Location HR estimator Mean HR size (n) Study duration (months) Study French Island, VIC 90% HM 1 (18) 44 Mitchell 1990b French Island, VIC 95% FK 1 # (20) 35 Greenfield 2012 Brisbane Ranges, VIC 90% HM 2 (5) 8 Hindell and Lee 1988 St Bees Island, QLD 95% MedCP 3 (9) 38 Tucker 2008 Kangaroo Island, SA 75% FK 5 (8) 8 Horgan 1998 Kangaroo Island, SA 70% HM 4 (9) Unk. (<2) Clark 1998 Kangaroo Island, SA 90% FK 7 (10) 30 Carney 2010 St Bees Island, QLD 95% FK 8* (33) 72 Ellis et al Pilliga, NSW 95% FK 9 (14) 12 Kavanagh et al Springsure, QLD 95% Cluster 12 (9) Melzer 1995 Mutdapilly, QLD 95% FK 15 (16) 36 White 1999 Springsure, QLD 95% HM 39 (9) Melzer 1995 Kangaroo Island, SA 95% Cluster 64 (23) 20 This study Blair Athol, QLD 95% HM 101 (9) 24 Ellis et al Eden, NSW 90% HM 265 (2) varying Jurskis and Potter

155 There are energetic costs and risks associated with long-distance movements. Traversing unknown areas or crossing large areas of open habitat or roads would increase vulnerability to predation, vehicular collisions, and also have consequences for energy expenditure (Moore and Ali, 1984; Labisky and Fritzen, 1998; Dique et al., 2003). The use of any fertility control method that causes an increase in these movements must consider the animal welfare implications of this change in behaviour. There was no difference in the incidence or distance of long-distance movements between my groups of study animals, although this result may have differed if fertility of the control group had been greater during the study period; i.e. if the energetic costs associated with lactation and carrying a back-young had inhibited such movements in this group (as observed by Hynes et al., 2011). This study relied on locational information for koalas collected during daylight hours. Whilst this provides an estimate of the minimum distance moved between fixes, it does not take into account roaming activity that may occur between these times. In addition, a number of studies have shown there is a difference in activity levels and the use of tree species between day and night due to increased feeding activity during nocturnal hours and the need to find shelter from heat and humidity during the day (Melzer, 1995; Matthews et al., 2007; Tucker, 2008; Ellis et al., 2009; Melzer et al., 2011; Ryan et al., 2013). However, Tucker (2008) found a significant positive correlation between the estimate of distance moved on the basis of location fixes made day-day and day-night-day, suggesting that, whilst estimates of overnight distances moved (or straight-line distance between successive daytime fixes as used in this study) evidently do not present a true representation of the total distances moved by a koala, it does provides a proportionate estimate of activity. Further, Tucker (2008) found no correlation between the additional distance moved overnight and the distance estimated from daytime location fixes only, suggesting additional movement at night is relatively constant, irrespective of the distance moved from day to day. 140

156 Although no differences in movement patterns and ranging behaviour between treatment groups were detected during this study, it could be that differences occur at a finer spatial or temporal scale than what was investigated here. The tracking methods used in this study provide a resolution of koala spatial activity appropriate for estimation of overall homerange, dispersal distance and broad rate of movement through the landscape, however methods of tracking that provide finer-scale resolution of spatial and temporal activity patterns, such as GPS telemetry collars (see Matthews et al., 2013) or accelerometry (e.g. Ryan et al., 2013), may reveal differences between groups that went undetected here. There is clearly opportunity for further work in this area, however given neither fertility control treatment had a negative effect on health or survival of treated individuals (see Chapter 5), differences in activity occurring at a finer scale may be of less importance for managers having to choose between treatment options. Management implications The effectiveness of any fertility control program that aims to reduce the impacts of a target population through a reduction in population density depends on the proportion of individuals that are treated within a target area. Additionally, sterilised females of some species are thought to restrict reproduction on the population level if they maintain their territories and social status (e.g. Tyndale-Biscoe, 1994). Therefore, even though this study detected no direct effect of fertility control on ranging behaviour, the high level of dispersal demonstrated by female koalas in general has profound implications for management success. Management of koalas on Kangaroo Island is conducted on a catchment by catchment approach (see Masters et al., 2004). As such, any movement of koalas between catchment areas will have the effect of diluting the effectiveness of a management effort. Further, a reduction in competition from resident females may increase the opportunity for fertile females to invade a managed area, further diluting the management effect. On Kangaroo Island, further investigation of koala movement between 141

157 catchments or managed areas would be informative and could potentially be achieved through further long-term tracking (GPS) of a significant number of individuals, and a greater effort undertaken to identify individual tagged (sterilised) koalas when resighted. An examination of genetic diversity between catchments is unlikely to be helpful due to the inherently low diversity within the Kangaroo Island population as a whole (Cristescu et al., 2009). If koalas respond to treatment by changing their core areas, then this makes the identification and management of discrete areas problematic, and may necessitate a change in management scale. It is possible that under different environmental conditions (e.g. average rainfall), when the fecundity of untreated females may be higher, the indirect effects from treatment may be more obvious. Even if treatment has no effect, if female koalas are ranging further than expected following treatment, this should be factored into any management plans. 142

158 143

159 7. Translocation of overabundant species: implications for translocated individuals Despite the perceived humaneness of translocations to resolve human wildlife conflicts, very few studies have monitored the fate of translocated problem animals. (Massei et al., 2010) 7.1 Introduction Translocation is a widely used wildlife management tool employed to establish, reestablish, or augment wildlife populations (Seddon et al., 2007; Baker et al., 2011). It can also be used to rescue individuals from intentional habitat destruction (e.g. Ostro et al., 1999; Richard-Hansen et al., 2000; Edgar et al., 2005), to minimize human-wildlife conflicts (e.g. Wambwa et al., 2001; Jones and Nealson, 2003), and to reduce densities of overabundant populations (e.g. van Vuren et al. 1997; Garai and Carr, 2001). When translocation is used to reduce the density of overabundant populations, management success is generally gauged at the source of individuals. Measures of success relate to the issues that triggered translocation activities, and may include resolution of a human-wildlife conflict, improved health of remaining individuals, recovery of degraded habitat, or increased agricultural productivity (Pietsch, 1994; van Vuren et al., 1997; Jones and Nealson, 2003). However, we argue that the success of a translocation program should also consider the fate of translocated individuals. Measureable benefits at the source (for the environment or the animals themselves, as individuals and as a population) need to be weighed against any negative impacts on the health and wellbeing of translocated individuals. This is critical given that translocated individuals may experience increased mortality rates, low This chapter results from a published peer-reviewed manuscript (Whisson et al., 2012) see page iv. 144

160 breeding success, or wandering behaviour (Fischer and Lindenmayer, 2000; Letty et al., 2007; Massei et al., 2010). Translocation has frequently been used in the management of koalas (Phascolarctos cinereus) in southern Australia. Early translocations were undertaken as a conservation measure to establish koala populations on offshore islands (Lewis, 1954), and subsequently to re-establish populations on the mainland (Menkhorst, 1995). Since the mid-1980s, koala translocations have almost exclusively been undertaken to reduce the effects of high-density populations on forest habitats and lessen the risk of density-related animal welfare issues (Menkhorst 1995; Martin and Handasyde, 1999; Department of Sustainability and Environment, 2004). Studies of the fate of translocated koalas are few, although Lee et al. (1990) reported high survival of translocated individuals. Translocation and fertility control are currently the only government sanctioned methods of managing overabundant koala populations (Natural Resource Management Ministerial Council, 2009). Fertility control via surgical sterilisation may occur in combination with translocation to prevent overabundance at release sites (Department of Sustainability and Environment, 2004; Duka and Masters, 2005). The preferred sterilisation methods are tubal ligation for females, and vasectomy for males, because they are easily performed and do not induce changes in hormonal cycling and associated reproductive behaviours (Duka and Masters, 2005). Translocation may occur immediately after surgery (Duka and Masters, 2005), or koalas may be released back at their capture points for translocation at a later date (Department of Sustainability and Environment, 2004). The cumulative effects of surgical sterilisation and translocation on the health and survival of koalas have not been investigated. We use a koala sterilisation and translocation program as a case study to assess the impacts of translocation on individual animals. Individuals from a high-density island population were surgically sterilised and translocated to the mainland between 1997 and

161 This provided the opportunity to examine post-release densities, survival, and movements. Results are discussed in relation to the welfare of translocated individuals and the suitability of translocation as a tool for managing overabundant species. 146

162 7.2 Materials and Methods Study areas Kangaroo Island is situated approximately 12 km off the coast of South Australia, Australia (Fig. 7.1). It covers an area of 438,000 ha, of which 47% retains native vegetation cover (Ball and Carruthers, 1998). Major land uses are sheep and cattle grazing. Koalas were introduced to Kangaroo Island for conservation purposes in the 1920s. Numbers increased rapidly such that large-scale habitat degradation is now occurring (Masters et al., 2004). High density koala populations are associated primarily with riparian habitats dominated by preferred food trees (rough-barked manna gum (Eucalyptus viminalis), South Australian blue gum (E. leucoxylon), and red gum (E. camaldulensis) (Masters et al., 2004). These habitats cover approximately 3% of the island, with less preferred forested habitat covering an additional 10%. Thirty-five release sites were located in the lower southeast region of mainland South Australia (Fig. 7.1). These sites were considered to be of high habitat quality for koalas: they were forest blocks of at least 10 ha in size and were dominated by healthy rough-barked manna gum. Koalas historically occurred throughout this region but disappeared in the mid-1900s, primarily as a result of hunting (Robinson, 1978). Major land uses in this region are grazing, cropping, and plantation forestry (pine (Pinus radiata) and Tasmanian blue gum (E. globulus)). Native forest covers approximately 7% of the region. Kangaroo Island and mainland release sites experience a Mediterranean climate with most rain falling in winter months (Bureau of Meteorology 2011a, b). Mainland sites have greater annual rainfall with a long term average of 707 mm compared to 430 mm on Kangaroo Island. Temperatures are similar with minimum temperatures ranging from 8.9 C to 20.9 C on Kangaroo Island and 8.2 C to 19.0 C on the mainland. 147

163 Figure 7.1: Kangaroo Island and the mainland koala release region in the Lower South East, South Australia. Release sites are indicated by solid circles. Koala density at release sites Koalas were translocated to the mainland from January to May (and December of the previous year on some occasions) in eight years between 1997 and 2007 (Table 1). To examine the effect of translocations on koala density on the mainland, we used data for 16 release sites that were randomly selected (from the pool of 35 release sites) for annual monitoring. These sites were surveyed in 2001 and 2002, and from 2004 to Although 148

164 koalas were released at all 16 sites throughout the study, both the number of individuals released and the sites used varied across years (Table 7.1). Table 7.1: Total number of koalas released and subsequent koala density in 16 mainland release sites that were regularly surveyed. In each year, koalas were released from January to May (and December of the previous year on some occasions), and surveys were undertaken between September and December. No surveys were undertaken in Year* Koalas released ,082 Region (16 sites) Koala density (koalas/ha) Sites where release occurred Mean SE Mean SE N *Includes animals translocated in December of the previous year. At each site, a single experienced observer systematically searched a 5-ha area and recorded the number of koalas. We assessed koala densities in two ways: 1) we determined the annual regional density by calculating the mean density across all 16 monitoring sites, irrespective of koala releases in any given year; and 2) we determined the annual release site density by calculating the mean density across sites where koalas had been released in the corresponding year. We also used data for 25 sites (11 of the 16 regular survey sites and 14 additional sites) surveyed for koalas in At these sites, an observer conducted a survey during the three months prior to release, and at one month post-release of koalas. 149

165 Immediate versus delayed translocation after sterilisation ( ) We aimed to determine the effect of a delay between surgical sterilisation and translocation of koalas on short-term (12 week) survival and movement behaviour. We defined two treatment groups: 1) delayed translocation involved koalas surgically sterilised more than one year prior to translocation to the mainland (6 F, 4 M); and 2) immediate translocation involved koalas surgically sterilised and translocated to the mainland on the same day (6 F, 9 M). Trained program personnel followed standard operating procedures of the Kangaroo Island Koala Management Program to catch, sterilize, and translocate koalas (Duka and Masters, 2005). We recorded capture date, location, ear tag number, sex, weight (kg), tooth wear (Martin and Handasyde, 1999), and muscle condition (McLean and Handasyde, 2006). We only included tooth wear class IV individuals (adults yr) with good muscle condition in our study, representative of most individuals that are translocated as part of the existing management program. Koalas in the resident group remaining on Kangaroo Island were not sterilised and were processed in the field before release on-site. We translocated koalas on 17 December 2004 and 21 January Koalas from each treatment group were represented in each translocation. We fitted radio-collars (Sirtrack Ltd, Havelock North, New Zealand) to allow 3 fingers space between the collar and the neck of the koala. We tracked koalas on foot using a Communication Specialists R-1000 receiver (Communication Specialists Inc., Orange, CA), and a 3-element Yagi antenna. We located koalas daily in the first week, bi-weekly for the next four weeks, and at weekly intervals thereafter until we removed their collars in April We recorded the coordinates of all locations. When we found a dead individual, we recorded its final location and collected its carcass for further examination. At the end of the study, we captured surviving individuals for removal of radio-collars, weighing, and assessment of muscle condition. 150

166 Translocated individuals versus residents on Kangaroo Island ( ) We compared the long-term survival (1 year post-translocation) and movements of sterilised and translocated koalas with those of intact animals on Kangaroo Island. We defined two treatment groups: 1) intact koalas on Kangaroo Island (8 F, 5 M); and 2) koalas surgically sterilised and translocated from Kangaroo Island to the mainland on the same day (8 F, 8 M). Koalas were captured, processed and radio-collared similarly to the study. They were translocated on 20 April and 2 May We located each koala daily for the first week, at weekly intervals for the next seven weeks, and then monthly until removing their collars in June All other procedures were as per the study. We took blood samples from koalas to be translocated to determine initial health status, and for use in determining the cause of any potential deaths occurring during the study. At the initial capture of each koala to be translocated, we took a 3-ml blood sample from the cephalic vein with a 21-gauge syringe into an ethylenediaminetetraacetic acid (EDTA) vacutainer. We placed tubes in an insulated container and transported them to a Healthscope Pathology laboratory in Adelaide, South Australia ( au) on the same day. The laboratory analysed 29 standard biochemical and haematological parameters. Analysis We performed separate data analyses for the two radio-tracking studies. We used generalized linear mixed models (GLMMs) to investigate relationships between the distances koalas moved from their release locations and explanatory variables (sex, treatment, and week). We employed a mixed model approach to account for repeated measurements of individuals. We included a unique identifier for each individual animal as a random effect in all models to account for non-independent error structures (Zuur et al., 2009). For both data sets, the dependent variable (distance moved) was right-skewed with a large coefficient of variation (>1). Residual diagnostic plots indicated that models incorporating a Gaussian error 151

167 distribution were not appropriate. We used a gamma error distribution (identity link function) instead, as it was more suitable for continuous, positive, and right-skewed data (Bolker, 2008). Repeat measurements through time also may result in temporal error correlation whereby measurements closer in time are more strongly correlated than those farther apart (Zuur et al., 2009). Plots of auto-correlation functions (ACF; Zuur et al., 2009, 2010) revealed significant correlation at various time lags in both data sets. Therefore, we included the autoregressive correlation structure (AR-1) in models to account for this lack of independence (Zuur et al., 2009). Subsequent ACF plots indicated that inclusion of the AR-1 correlation structure markedly reduced temporal correlation in the data. After defining the random term and temporal correlation structure, we explored the optimal fixed structure (i.e., combination of explanatory variables). We performed model selection by removing the least significant fixed term (based on model coefficients and their associated standard errors and P values) and then re-fitting the model (Zuur et al., 2009). We removed fixed terms in order of complexity beginning with the 3-way interaction (sex X treatment X week), then 2-way interactions, then main terms. The best model was identified when all terms remaining in the model were significant (Zuur et al., 2009) and residual plots confirmed model adequacy. Residual plots for the data set revealed 2 large outliers, both related to a single animal (Kangaroo Island female that displayed very large movements in weeks 50 and 54). Since these outliers represent an unusual movement event by a single animal, we repeated analysis with these data points excluded. We performed all modelling using the glmmpql function in package MASS (Venables and Ripley 2002) in R v (R Development Core Team, 2009). We compared blood parameters for koalas surviving translocation (5 M, 5 F) to koalas that did not survive translocation (3 M, 3 F). We normalized data for all blood parameters and constructed a resemblance matrix using Euclidean distances. We conducted an analysis of 152

168 similarity (ANOSIM) to determine if blood parameters varied between the 2 groups using Primer 6 (Primer-E Ltd, Ivybridge, United Kingdom). 153

169 7.3 Results Koala population density at release sites Between 1997 and 2007, staff of the Kangaroo Island Koala Management Program translocated 3,206 sterilised koalas from Kangaroo Island to the mainland. Each year, koala densities were <0.3 koalas/ha throughout the mainland region (averaged across 16 sites), and 0.4 koalas/ha at sites where koalas had been released in the corresponding year (Table 7.1). Densities remained low at release sites despite repeated releases from 2005 to 2007 (1,686 koalas released). Prior to translocations in 2006, we recorded 13 koalas in eight (32%) of the 25 survey sites, with a maximum of 0.5 koalas/ha recorded at a site. The 13 animals were all from previous translocations (between 1997 and 2005). Following these surveys, we released 579 koalas into the 25 sites (release rate of 1.0 koala/ha). In surveys conducted one month after release, we observed koalas in 16 (64%) sites, with a maximum of 0.6 koalas/ha per site. Immediate versus delayed translocation after sterilisation ( ) All translocated koalas survived the first 4 weeks post-translocation. We recorded two deaths (1 M, 1 F) in the immediate translocation group at three months post-release (Table 2). This represented 8.0% of all translocated individuals (n = 25) and 13.3% of individuals in the immediate translocation group (n = 15). We could not determine cause of death. We found no difference in survival between treatment groups (x 2,1 = 2.16, P = 0.15). Nineteen of 22 (86.4%) translocated koalas that were weighed at the end of the study had a mean weight gain of 0.7 kg ± 0.60 (SD). Three (13.6%) koalas (all in the delayed translocation group) lost kg with a maximum loss of 5% of body mass. One koala 154

170 could not be recaptured until another 3 months had elapsed so it was not included in this analysis. The optimal GLMM modelling distance moved for the data set contained sex, week, and their interaction (Table 7.3a). Treatment (immediate or delayed translocation) was not identified as an influential predictor of distances moved, either by itself or as an interaction with other terms. Predicted values showed that distances moved by male and female koalas increased with time over the 12-week duration of the study (Fig. 7.2). However, the rate of increase differed significantly between the sexes, with males moving greater distances than females over time. At four weeks post-translocation, 56.0% of koalas (41.7% of F, 69.2% of M) had moved >1 km from the release site. This figure increased to 76.0% (58.3% of F, 92.3% of M) at 12 weeks post-translocation. One male koala had moved 8.5 km from its release site after 12 weeks of monitoring (Fig. 7.2). Figure 7.2: The predicted distance moved by translocated male (dashed line) and female (solid line) koalas ( study) as a function of time. Open triangles represent actual data values for males and closed circles represent actual data values for females. 155

171 Translocated individuals versus residents on Kangaroo Island ( ) Of the 16 individuals that were sterilised and translocated to the mainland, six (37.5%; 3 M, 3 F) died during the 12-month radio-tracking period. Two (1 M, 1 F) of these deaths occurred during the first 3 months post-translocation, whereas the other four deaths occurred 3 12 months post-translocation (Table 7.2). Over the same period, no deaths of intact animals (n = 13) occurred on Kangaroo Island. Thus, mortality was greater than expected in the sterilised and translocated group (x 2,1 = 6.15, P = 0.01). All carcasses of dead koalas were found within large (approx. 10 ha) woodland patches dominated by rough-barked manna gum. The two koalas that died within 3 months of translocation were located in time for necropsy. The male had lost approximately 20.0% (2 kg) of its initial body weight and was dehydrated with signs of kidney damage. The female showed signs of acute heart failure resulting from a compromised vascular system. Table 7.2: Survival of radio-collared koalas in all treatment groups from radio-tracking studies in and % koalas surviving Year Treatment N 1 Month 3 Months 12 Months Delayed translocation N/A* Immediate translocation N/A Translocated to mainland Remaining on Kangaroo Island * N/A: koalas were not tracked 12 months after release. We found no apparent difference in distances moved by surviving and non-surviving koalas. The two koalas that died within 3 months post-release moved 216 m and 824 m from their release sites. Surviving koalas moved between 70 m and 3 km from release sites during 156

172 the same period. The four koalas that died 3 12 months post-release moved between 188 m and 9.5 km from release sites whereas surviving koalas moved between 137 m and 7.1 km over the same period. Nine of 10 (90.0%) surviving translocated koalas had a mean weight gain of 0.9 kg ± 0.5 kg after 12 months. One male koala had lost 2 kg (20.0% of initial weight). With outlying data points excluded (two large movements by a single Kangaroo Island female), the optimal movement model contained the 2-way interaction between treatment (sterilised and translocated to the mainland vs. intact on Kangaroo Island) and week (Table 7.3b). Sex was not an influential explanatory variable (including as an interaction with other explanatory variables). Predicted values from the model show that distances moved varied according to treatment group (Fig. 7.3). Distance moved increased for both treatment groups with time; however, animals that were sterilised and translocated moved at a greater rate over time than those left intact on Kangaroo Island. For translocated animals, 18.8% of koalas had moved >1 km from the release site at 4 weeks post-translocation. Just 7.7% of animals monitored on Kangaroo Island had moved >1 km from their capture point during this same time period. At 12 weeks, 31.3% of translocated animals were >1 km from their release site (7.7% for Kangaroo Island animals), and at 1 year, 80.0% of surviving translocated koalas were >1 km from release sites (30.8% for Kangaroo Island animals). With the two large outliers included in the data set, the optimal model again contained both treatment and week, but not their interaction (Table 7.3c). Biochemical and haematological parameters of koalas at first capture were within the range of reference values for healthy koalas (Canfield et al., 1989). Parameters did not vary between survivors and non-survivors (ANOSIM: Global R = 0.07, P = 0.21). 157

173 Figure 7.3: The predicted distance moved by sterilised and translocated koalas (dashed line), and intact koalas remaining on Kangaroo Island (solid line) in , as a function of time (excluding 2 outlying data points). Open triangles represent actual data values for sterilised and translocated animals and closed circles represent actual data values for intact Kangaroo Island animals. Table 7.3: Results from the optimal generalized linear mixed effects model relating the distances moved by individual koalas to sex, treatment, and week for (a) the data set; (b) the data set (excluding 2 large outliers); and (c) the dataset (including 2 large outliers). Data set Explanatory variable Coefficient SE DF t P (a) Intercept Sex a Week <0.001 Sex X week (b) Intercept <0.001 (excluding outliers) Treatment b Week <0.001 Week X treatment (c) Intercept <0.001 (including outliers) Treatment b <0.001 Week <0.001 a The reference category for the categorical sex variable was female. b The reference category for the categorical treatment variable was intact Kangaroo Island koalas. 158

174 7.4 Discussion Translocation of overabundant species typically aims to reduce population densities and impacts at the source location. Management success is usually measured at that location, whereas the fate of translocated individuals may either be assumed to be satisfactory, or not considered. This study highlights the need to consider the welfare of translocated individuals. We observed a mortality rate of 37.5% for translocated koalas that were monitored for 12 months post-translocation. No deaths were observed among monitored individuals that were not translocated from Kangaroo Island. We also observed long-distance movements of translocated koalas in the year following their translocation, despite release sites comprising what was considered to be high quality koala habitat. Translocation clearly can have negative consequences for translocated individuals. These consequences must be more routinely investigated and considered when evaluating the appropriateness of management actions. Translocation to reduce koala densities on Kangaroo Island was adopted as a management tool following the results of an experimental translocation of 20 surgically sterilised koalas in 1997 (Department of Environment and Natural Resources, unpublished data). In that study, koalas were monitored for seven months, and only one koala died at four weeks after release. Multiple translocations were then undertaken from 1998 to Despite a large number of koalas (1,082) being released into mainland sites over that period, our surveys revealed low koala densities throughout the release region, with a mean density of 0.13 koalas/ha of suitable habitat in From 2001 to 2007, koala densities remained low in the release region (<0.3 koalas/ha), despite ongoing releases of up to 603 koalas/year. Low densities (0.4 koalas/ha) were also recorded at sites where koalas had been released at densities of 1 koala/ha in the same year. Intensive post-release surveys undertaken in 2006 suggest that the decline in koala densities at release sites occurs within 4 weeks of release. Results from our radio-tracking studies indicate that the initial rapid decline in densities is because of dispersal 159

175 from release sites. However, mortality was also found to be a factor contributing to low koala densities throughout the region. In the radio-tracking study, 8.0% of all translocated koalas died within three months of release. In the radio-tracking study, this figure increased to 37.5% within 12 months of individuals being translocated. In contrast, none of the intact individuals that were radio-tracked during the same time period ( ) on Kangaroo Island died. Two conclusions can be drawn from these results. First, the sterilisation and translocation treatment resulted in elevated mortality rates. Second, mortality rates appeared to increase over time. An elevated mortality rate has been reported for translocated individuals of other species (Fischer and Lindenmayer, 2000). In some situations, 100% mortality has been observed (e.g., Augee et al., 1996). Reasons for greater mortality are not always clear and numerous factors have been proposed. These include the stress associated with capture, handling, transportation, and release of animals into an unfamiliar environment (Pietsch, 1994; Dickens et al., 2010); the health or nutritional status of individuals translocated (O Bryan and McCullough, 1985); individual characteristics such as age, mass, sex, or social status (Letty et al., 2007); quality of the release habitat (Fischer and Lindenmayer, 2000); density of the species at the release site (Fischer and Lindenmayer, 2000); predation pressure and vehicle mortality (Pietsch, 1994; Calvete and Estrada, 2004); weather conditions and season at the time of release (O Bryan and McCullough, 1985); and the presence of disease in the resident population at the release location (Kock et al., 2010). In our radio-tracking studies, we attempted to control for as many of these variables as possible while following the protocols of the management program. Koalas were of similar age and weight, and were all considered to be in good condition. Analysis of blood parameters confirmed that koalas were healthy at the time of initial capture. Release sites were chosen within the historic range of the koala on the mainland. As hunting by humans was the major cause of historical decline in koala 160

176 populations in southeast South Australia (Robinson, 1978), we assumed that, given this pressure no longer exists, large habitat patches in this region containing tree species preferred by koalas should be suitable release sites for translocated animals. Release sites were forest blocks at least 10 ha in size and comprised more than 80% rough-barked manna gum which is known to be a preferred food tree of koalas (Hindell and Lee, 1990). In fact, rough-barked manna gum was more common in release sites than in habitat on Kangaroo Island. Furthermore, koalas were released at a density of only 1 koala/ha, which is much less than densities on Kangaroo Island. This low release rate coupled with evidence of extensive overlap in home range and resource use by koalas (Ellis et al., 2009; Mitchell, 1990b) meant that competition between koalas at release sites should have been negligible. For the six koalas that died in the study, we could not identify any common physiological or behavioural characteristic that may have been a factor in their deaths. Both males and females were represented, and animals that died were of similar age and body condition to those that survived. All koalas were years of age (as determined by tooth wear), and had good body condition at the time of initial capture. Blood biochemical and haematological parameters were also similar between survivors and non-survivors, and within the range of reference values for healthy koalas (Canfield et al., 1989), although Canfield et al. (1989) bled some study animals up to 12 hours post-capture which may not provide robust normal reference values. Further assessment and consolidation of biochemical and haematological blood values from wild-caught koalas should be prioritised by koala researchers. Distances moved by non-survivors were within the range of distances moved by surviving koalas. Weather conditions may have been a factor contributing to the death of 4 animals in the study. Unseasonably cold temperatures (less than 8 C) occurred in the week immediately following the release of these animals. Only two of the six koalas released in that group survived. Although koalas can tolerate cold temperatures, they generally spend less time 161

177 active and feeding under these conditions (Martin and Handasyde, 1999). The cumulative effect of stress associated with the treatment and reduced foraging during cold conditions may have resulted in koalas becoming dehydrated. Dehydration appeared to be the cause of at least one of the deaths. Surgical sterilisation and immediate translocation may interact in an additive way to reduce the survival probability of translocated koalas. This has resulted in a policy in other management programs of rereleasing sterilised koalas at capture sites, then recapturing and translocating them at a later date (Department of Sustainability and Environment, 2004). Results from our study suggest that a time delay between sterilisation and translocation does not improve survival of translocated koalas. Although this was only a shortterm study, all surviving koalas in the immediate translocation group had gained weight after three months, suggesting that they had recovered from the effects of the sterilisation and translocation process. Following release, translocated koalas rapidly dispersed from release sites with some radio-collared animals moving up to 8 km within 12 weeks of release. Animals translocated to the mainland moved farther than those remaining on Kangaroo Island. Only one female in the group of koalas that remained on Kangaroo Island moved a similar distance (12 km) in the same period. The extent that translocated koalas moved away from release sites varied between our radio-tracking studies. In , the predicted distance moved over the 12-week study period was 2.0 km for females and 4.5 km for males. In , the predicted distance was less over the same period (less than 1.0 km). Differences in climatic variables and breeding season at the time of release may explain the differences in dispersal rates between our studies. Koalas were released in December and January in the study when temperatures were warm. In , koalas were released in April and May. Cooler temperatures may have forced koalas to expend more energy on thermoregulation than movement. Also, with breeding occurring over the warmer months (Whisson and Carlyon, 162

178 2010), koalas released in the study may have been moving over larger distances searching for mates. The reasons for long-distance movements by translocated koalas are not clear. Tree density, species composition, canopy condition, and patch size at release sites were carefully assessed to ensure their suitability for koalas. However, release sites were forest remnants in a highly cleared and fragmented landscape. Processes associated with habitat fragmentation (e.g. edge effects, inter-patch distances, surrounding land use, and road density) may influence habitat quality and carrying capacity for koalas (McAlpine et al., 2006), although habitats at the capture location were also highly fragmented. Furthermore, despite release sites being dominated by a preferred food species, local factors may have influenced the nutritional quality or palatability of the foliage (Moore and Foley 2005) and prompted koalas to search for a better food supply. An investigation of such factors may be necessary when assessing the suitability of release sites for koalas in the future. Long-distance koala movements may have been the result of wandering behaviour, as has been observed in individuals of other translocated species (Massei et al., 2010). Such wandering may be attributable to being placed in an unfamiliar location together with a disruption in social organisation. Similar ranging behaviour was also observed by Lee et al. (1990) during experimental translocation trials in Victoria. Regardless of the mechanism, observed long-distance movements raise further concerns for the welfare of translocated koalas and the selection of release sites. As forest cover in the release region was highly fragmented, dispersing koalas would have encountered large areas of cleared land or unsuitable habitat during movement events. Crossing such areas would increase vulnerability to predation, vehicular collisions, and have consequences for energy expenditure. Therefore, factors operating at larger scales (e.g. regional land use, degree of habitat fragmentation, and proximity to busy roads) must also be considered (McAlpine et al., 2006). 163

179 Management implications Translocation is increasingly being advocated as a management tool for overabundant species, particularly those for which alternative population control measures are not considered appropriate (e.g. high profile and charismatic species; Massei et al., 2010). Translocation is effective in immediately reducing population density of the target species at the source location, and managers generally assumed that translocated individuals quickly recover from the process, and may even benefit from being placed in good quality habitat where population densities are lower. Our study of koalas suggests that this is not necessarily always the case, with translocated individuals suffering higher rates of mortality than individuals in the source population, and undergoing rapid long-distance dispersal. In contrast, however, Lee et al. (1990) recorded high survival among koalas for 16 months post-translocation in Victoria. A wide range of factors may be responsible for determining the survival of translocated individuals. Although identifying some of these factors a priori (e.g. habitat quality) may be possible, our study indicates that many others will be difficult, if not impossible, to predict (e.g. unseasonable weather). Consequently, some mortality is likely even when best practice has been followed. Acceptability of all aspects of a translocation must be linked back to the goals of the translocation program, and due consideration by wildlife managers therefore must be given to what is an acceptable level of mortality for translocated individuals. Currently, no such guidelines for koalas exist and we could find no evidence of post-translocation mortality protocols for other species. Post-release monitoring of translocated individuals is clearly necessary for measuring success and for refining protocols of a translocation program. Our study suggests that a time lag may exist between translocation and the signs of negative consequences for animal health. Therefore monitoring of translocated individuals must occur over a sufficiently long time period. Results from such monitoring should be used to refine methods to improve the welfare 164

180 of individuals translocated in the future. Furthermore, any positive management outcomes observed at the source of individuals must be assessed in relation to the fate of translocated animals when determining the overall success of translocation programs. In some cases, animal welfare issues for translocated individuals may be of such concern that benefits at the source cannot be justified. Moving animals to a novel environment may have a greater impact on survival than in situ management, however a year effect cannot be ruled out here and repeat studies with greater sample sizes should be undertaken over multiple years to confirm the result. Regardless, the relative humaneness of translocation should be considered and further research conducted to determine how this can be addressed, particularly if translocation becomes necessary when the habitat in which the koalas are being managed is so severely degraded that there is a risk of starvation. 165

181 166

182 167

183 8. Discussion a review of management options for control of an overabundant population of koalas on Kangaroo Island, South Australia Active management of some high-density koala populations is necessary to protect local eucalypt habitat from excessive browse damage and to mitigate resulting welfare impacts for the resident koala population. Fertility control and translocation are the two governmentsanctioned options currently available to management authorities aiming to reduce koala densities or limit population growth (Natural Resource Management Ministerial Council, 2009). This thesis examines potential and current options for controlling the introduced and overabundant koala population on South Australia s Kangaroo Island and the impact of these management techniques on individual animals, with particular focus on the trial of a GnRHagonist contraceptive implant and its comparison with the current management practice of surgical sterilisation and translocation. Knowledge of the biology of the target species and basic population characteristics are essential to understanding the dynamics of any population and therefore the effort required to achieve a desired management outcome (Sinclair et al., 2006). Koala population parameters may vary geographically and temporally (Melzer et al., 2000; McLean and Handasyde, 2006) and, as a result, an understanding of the specific dynamics of the Kangaroo Island population, particularly the factors affecting reproduction, is critical to its management. Management decisions on Kangaroo Island have historically been based on population parameters derived from the extensive studies conducted on Victorian populations (e.g. Martin 1985c; Martin and Handasyde 1990; Mitchell and Martin 1990), however spatial variation in koala reproduction (McLean and Handasyde, 2006) highlights the importance of obtaining location-specific information on reproductive parameters for predicting population growth and the effectiveness 168

184 of management strategies. Whilst modelling the dynamics of the Kangaroo Island population in response to various control options was beyond the scope of this thesis, we presented data on reproductive characteristics that will be essential for this process (Chapter 3), and will allow managers to maximise the efficiency of control operations. We demonstrated temporal variation in fecundity and the sex-ratio of young, and that fecundity in this population is lower than that of other island koala populations. We also demonstrated that the breeding season on Kangaroo Island is similar to populations on the Victorian mainland, with the season progressing from September each year. This is particularly important information for maximising the efficiency of on-ground management activities, particularly for timing the delivery of contraceptives to maximise the duration (in terms of the number of breeding seasons) of effect gained from a fertility control treatment. The existing large-scale koala management program on Kangaroo Island provided a unique opportunity for acquire this information. This thesis addresses a range of issues that are critical in determining the viability and acceptability of slow-release contraceptive implants containing the GnRH agonist deslorelin (Suprelorin ) as a koala management tool. Bomford and O Brian (1992) suggest the following key issues should be considered when determining whether fertility control may be successful and practical for wildlife population management: Availability of the drug or technique; Effectiveness of the delivery mechanism; Humaneness; Target specificity; Environmental acceptability; Cost-effectiveness; and Reduction in population size and damage. These requirements assume the technique has proved successful in inhibiting fertility in the target species; here we add treatment efficacy, i.e. the treatment s capacity to 169

185 successfully inhibit reproduction in the target species, and two other important parameters of treatment (longevity and reversibility of contraceptive effect) to the list of critical considerations. These factors can be applied equally when assessing specific fertility control techniques or alternative control options, including translocation, and provide a useful framework for discussion of the various techniques available for management of the Kangaroo Island koala population. Here we discuss the potential of the available koala management options to satisfy these requirements, in particular the use of Suprelorin contraceptive implants, and outline future directions for research in this field. A list of key recommendations for management and future research can be found in the conclusion to this chapter. Treatment efficacy Clearly, one of the crucial elements to the usefulness of a fertility control technique for population management is whether treatment effectively inhibits fertility in the target species. In a polygynous species, targeting females has been shown to be most effective and efficient for managing vertebrate populations (Bomford, 1990; Caughley et al., 1992; Barlow et al., 1997) and females are typically targeted when managing free-ranging koala populations (translocation programs may also target males) (Department of Sustainability and Environment, 2004; Natural Resource Management Ministerial Council, 2009). This study has shown that treatment with subcutaneous 4.7 mg Suprelorin implants is successful in inhibiting fertility in the majority of free-ranging female koalas for at least one year (Chapter 4), supporting similar findings by Greenfield (2012) and Herbert (2002) that showed Suprelorin prevented the production of new offspring in % of both free-ranging and captive female koalas. This level of efficacy is comparable to Suprelorin effectiveness in females of other species, including tammar wallabies (Macropus eugenii) (Herbert et al., 2005), eastern grey kangaroos (Macropus giganteus) (Herbert et al., 2006), brushtail possums (Trichosurus vulpecula) (Eymann et al., 2013), African lions (Panthera leo) and tigers 170

186 (Panthera tigris) (Bertschinger et al., 2002; 2008). In contrast to control animals, female koalas treated with Suprelorin did not respond to exogenous GnRH with a strong LH surge (Chapter 4), suggesting, as in other species, GnRH-agonist treatment of female koalas acts by desensitising the pituitary gland to GnRH thus suppressing the release of gonadotrophins (e.g. Herbert et al., 2005). The efficacy of Suprelorin in female koalas is comparable to that achieved using several other fertility control methods. Table 8.1 summarises the various methods of fertility control that have demonstrated 70% efficacy during trials or use in current management of koalas. Surgical sterilisation (tubal ligation) results in 100% infertility of treated individuals (Chapter 4; DEWNR, unpublished data). Treatment with the synthetic gestagen levonorgestrel is also highly effective, with 100% infertility achieved over six consecutive breeding seasons following a single treatment of free-ranging females on French Island (Hynes et al., 2010), confirming the results of Middleton et al. (2003) during preliminary trials of this contraceptive on wild koalas. Like Suprelorin, efficacy of levonorgestrel treatment has been high in a range of other species, including eastern grey kangaroos (Nave et al., 2002; Coulson et al., 2008; Wilson, 2012), domestic cats (Felis domestica) (Baldwin et al., 1994) and some primates (e.g. Savage et al., 2002). In contrast, treatment of female koalas with another synthetic gestagen, etonogestrel, had no contraceptive effect, possibly due to a difference in its molecular structure, resulting in inability to bind to the koala progesterone receptor (Hynes et al., 2010). Likewise, immunocontraception treatments have not proved as successful as other techniques. In a study on Snake Island in Victoria, free-ranging female koalas immunised against porcine zona pellucida (PZP) antigens showed a 70% reduction in fertility following treatment (Kitchener et al., 2009b), however chlamydiosis, which typically reduces fecundity in infected populations (Martin and Handasyde, 1990a; McLean and Handasyde, 2006) and leads to a higher proportion of females not breeding in consecutive years (McLean, 2003), is prevalent in this 171

187 population and may have inflated the contraceptive effect recorded by Kitchener et al. (2009b). Further, in the same study, immunisation of female koalas against zona pellucida antigens using recombinant brushtail possum ZP3 did not reduce fertility (Kitchener et al., 2009b). Treatment duration The longevity of the contraceptive effect is also an important element of a treatment s efficacy, particularly for species with a long reproductive lifespan such as the koala (Martin and Handasyde, 1990a) or in situations where re-treating individuals is difficult. Both intraand inter-specific variation in the duration of the contraceptive effect of Suprelorin treatment has been observed, with longevity also potentially dose-dependent in some cases (Herbert, 2002). Although my study did not specifically examine the duration of treatment effect, most female koalas treated with a single 4.7 mg implant exhibited infertility for a period of at least one year, and no koalas caught during a simulation of typical DEWNR management practice two years post-treatment were observed with young (see Chapter 4). Greenfield (2012) recorded variation in the duration of contraception of female koalas (also given a single 4.7 mg implant) of between 338 and >776 days; fertility returned to 50% of the animals during the second breeding season following treatment, with all animals breeding by the third season post-treatment. A similar degree of variability in duration of action has been demonstrated in heifers (D Occhio et al., 1996), tammar wallabies (Herbert et al., 2005) and eastern grey kangaroos (Herbert et al., 2006; Wilson, 2012). Such variability is potentially a result of natural variation in individual response to deslorelin (Herbert, 2002), or differences in the duration and amount of deslorelin released from implants (D Occhio, 1996). Similar to Greenfield s (2012) observations of koalas, Herbert et al. (2010) showed high efficacy of Suprelorin in eastern grey kangaroos in the first season post-treatment, followed by a rise in fertility to 30% in the second year of treatment and 88% in the third year. 172

188 Published research shows treatment with a single 70 mg levonorgestrel implant induces infertility in female koalas for over six years (Hynes et al., 2010), and ongoing studies now show this effect continues for over nine years (K. Handasyde, pers. comm.). In koalas, the variability in response to this treatment is low and the duration of effect appears consistent between individuals (Middleton et al., 2003; Hynes et al., 2010). Where there have been failures of this treatment in other species, it has been suggested that dose-rates were too low and that higher doses would be likely to extend the effective life-span of the implants (Coulson et al., 2008). Immunisation of female koalas against PZP antigens resulted in 70% infertility for a duration of at least 33 months following a primary and subsequent booster vaccination (Kitchener et al., 2009b). Surgical sterilisation of female koalas (tubal ligation) has been used in a number of government management programs in Victoria and South Australia to manage high-density populations, usually in conjunction with translocation (Duka and Masters, 2005; Menkhorst, 2008; Whisson and Carlyon, 2010). Unlike hormonal or vaccination methods of fertility control, tubal ligation results in permanent irreversible infertility for the lifetime of the sterilised individual (Table 8.1). This may be acceptable if the sole management aim is to reduce population density, but may not be desirable for the management of some koala populations due to the potential loss of genetic diversity in areas that are already deficient, e.g. Kangaroo Island (Seymour et al., 2001; Cristescu et al., 2009). Reversibility of the contraceptive effect may therefore be an important consideration when choosing a fertility control option. In large populations, where only a proportion of females can be treated in any one year, random treatment of koalas with a temporary contraceptive will help to preserve genetic diversity, as the majority of females will breed at some stage in their life (Tanaka et al., 2009). Translocation and permanent infertility techniques such as surgical sterilisation have the most detrimental effects on genetic diversity and are analogous to culling in terms of their impact on effective population size (Tanaka et al., 2009). 173

189 Timing of treatment can be critical to its success and its longevity, particularly for treatments with a limited duration of effect. For example, in seasonally-breeding species such as the koala (McLean and Handasyde, 2006; Whisson and Carlyon, 2010), it is most efficient to time treatment delivery for immediately prior to the breeding season, which results in an 18- month period of contraception, effectively inhibiting reproduction for the subsequent two breeding seasons, i.e. achieving fertility control equivalent to a period of two years. Female koalas have a reproductive lifespan of up to ten years (Martin and Handasyde, 1990a). Therefore, an individual would require at least five treatments with Suprelorin to abolish all reproductive output for the animal s lifetime. In contrast, only two treatments with levonorgestrel or just a single surgical procedure would be needed to achieve the same outcome (Table 8.1), although surgery must be considered to be substantially more invasive than the delivery of subcutaneous implants. Justification for choosing a particular method needs to balance the impact of the delivery procedure with the cost-effectiveness and potential welfare impacts of the capture process and handling, transport and holding time. Suprelorin-induced infertility has been shown to be reversible in a range of species, including koalas, following cessation of deslorelin release and recovery of the pituitary-ovarian axis (e.g. Trigg, et al., 2001; Baker et al., 2004; D Occhio et al., 2002; Herbert, 2002; Greenfield, 2012). Likewise, removal of levonorgestrel implants from female koalas reverses the contraceptive effect by the following breeding season (Hynes et al., 2010), although, given the duration of effect (>8 years) reversal would normally require recapture of an individual rather than relying on attrition of the implant. Restoration of fertility in female koalas following immunisation against PZP antigens has not been demonstrated, however this treatment is reversible in other species such as feral horses (Equus caballus) (Kirkpatrick et al., 1990) and morphology of ovaries and oocytes in treated koalas appears normal (Kitchener et al., 2009b), suggesting fertility may return in this species following this treatment. However, it has been shown, in horses, that there is a decreasing likelihood of reversal as they become old 174

190 and have had many year successive treatments (Kirkpatrick et al., 2011). While there are no long term data for koalas, this potential reduction in reversibility needs to be considered in models driving management planning. Availability and delivery of treatment For a fertility control technique to be practical at a population-scale the treatment needs to be readily available to wildlife managers and, ideally, the mode of delivery should be simple to humanely deliver to an adequate proportion of the target population. Suprelorin implants are commercially available as a registered veterinary product in Australia and hence readily available on a veterinarian s prescription. This treatment is easily delivered to koalas with minimal training, using a sterile pre-loaded hypodermic implanter to insert the implant subcutaneously between the shoulder-blades. No anaesthetic or sedative is required and the implant site is sealed using veterinary adhesive. These implants proved extremely easy to deliver to koalas in the field at the site of capture, thus the individual could be released immediately following treatment, dramatically reducing handling time and removing the need for transporting animals. These factors also increase the efficiency of the management program because a substantially larger number of animals can be treated in a day. Levonorgestrel implants are a prescription medication registered for human use (Norplant, Leiras Pharmaceuticals, Turku, Finland) and must be inserted subcutaneously via a small incision between the shoulder-blades that requires sutures to close, thus increasing the duration and cost of treatment and the expertise required for delivery. For large-scale levonorgestrel use, koalas are typically transported to a central processing site where implantation occurs under general anaesthetic (K. Carlyon, pers. obs.) or sedation (Hynes, 2007). Another gestagen implant of similar dimensions (Implanon TM, Merck Sharp and Dohme Pty Ltd, NSW) is regularly administered to women by a simple subcutaneous injection 175

191 without sedation (Levine et al., 2008), suggesting there is scope for future delivery of levonorgestrel via an implanter. The only immunocontraception trial conducted on free-ranging koalas involved immunisation of female koalas against the zona pellucida antigens PZP and recombinant brushtail possum ZP3 (Kitchener et al., 2009b). PZP is not readily available in Australia, but some laboratories produce it in South Africa and the United States, specifically for use in wildlife. It is not currently registered as a veterinary product, thus is not currently available for broad-scale koala management. The mode of treatment used by Kitchener et al. (2009b) also involved two subcutaneous injections in the groin of either PZP or ZP3 emulsified with complete Freund s adjuvant, followed by two duplicate injections intramuscularly into the thigh (i.e. four injections in total), all under sedation. A single booster immunisation emulsified in incomplete Freund s adjuvant was administered to each animal 3 5 months after the initial immunisation. The need for sedation, subsequent recovery and then to find, and recapture animals to deliver a second booster treatment decreases the efficiency of treatment enormously thereby increasing the cost. Additionally, in a free-ranging population, a proportion of animals given the primary vaccination are unlikely to be found again for delivery of the booster, rendering the initial treatment pointless. Recent developments in the delivery of contraceptive vaccines have involved intramuscular injection of PZP/adjuvant emulsions simultaneously with a number of slow-release polymer pellets also containing PZP and adjuvant. These pellets replace the need for a second booster injection of the vaccine (and hence a second capture), releasing PZP and adjuvant at ~1, 3 and 12 months post-treatment (depending on the make-up of the pellet) and have been trialled on ungulates with promising results (Turner et al., 2007; Turner et al., 2008; Rutberg et al., 2013). This technique has not been tested in koalas, and no trials have been commenced, so the use of this approach is unlikely in the foreseeable future. A single-dose GnRH vaccine (GonaCon TM ) is registered in the US for several wildlife species, however whilst trials suggest the treatment has no long- 176

192 term effects on behaviour and physiology (e.g. wild boar; Massei et al., 2012), the treatment is yet to be trialled on marsupials and delivery currently relies on injection. Investigations into alternative delivery modes for contraceptive vaccines to wildlife on a large scale include the use of species-specific viral vectors (e.g. Hinds et al., 2003; Reubel et al., 2005) or bacterial ghosts (e.g. Walcher et al., 2008), however there are two fundamental issues with viral vectors and/or other naturally disseminating systems that need to be addressed before this could be considered a viable delivery technique for koalas: (1) the technical feasibility of such an approach has not been demonstrated in free-ranging wild populations, and (2) there is a lack of control over dissemination, i.e. even if technical feasibility were demonstrated, it may not be desirable for the management of overabundant native species, where control over which populations are treated and the time-frame of treatment is typically required (see McLeod et al., 2007; Cross et al., 2011). Surgical sterilisation does not require drug delivery, however this specialised technique must be performed under general anaesthetic by a trained veterinarian (see Chapter 2). This method also inevitably requires transportation of animals to a veterinary facility following capture, holding of animals awaiting surgery and again for recovery prior to release. On Kangaroo Island, transport distances from capture sites to the veterinary facility (and viceversa for release) can be over 80 km and animals are typically held for 1-2 days in purposebuilt holding cages with an adequate supply of suitable browse, which is a costly and less efficient system, and must be assumed to cause some level of distress. One major advantage of delivering fertility control agents at the point of capture is that this must be less distressing for the animals than the approach described above. Despite this, because fertility control techniques currently available for koalas still require capture of animals for treatment there will still unavoidably be short-term stress (Hajduk et al., 1992). The standard method for koala capture is described in Chapter 2 (after Martin, 1989) and 177

193 requires operators who are skilled in tree-climbing and koala handling techniques. The efficiency and ease of locating a koala and then capturing it varies greatly between locations, primarily in relation to the tree height and complexity of canopy cover. This was highlighted during the management simulation conducted during this study (Chapter 4), where the mean duration of the capture process (~50 minutes) was over twice the average for the rest of Kangaroo Island. This was entirely due to the greater height of trees in this area (typically >20m), which generally required multi-stage climbing before a koala could be noosed, followed by a greater time spent guiding a koala down the tree. Ease of capture should be a key parameter for evaluating the suitability of a fertility control technique for a particular population. Duration of contraceptive effect and therefore the number of treatments needed per individual per unit time is particularly important when managing free-ranging animals, and the ease with which animals can be captured will have a large influence on the value of the technique if multiple treatments are required. Clearly, a technique that requires just a single capture of an individual to prevent reproduction for its entire reproductive lifespan, is most efficient in this regard and avoids duplication of capture stress. However, a need for multiple captures of an individual to achieve the same result may not be such a concern if capture of animals is straightforward and rapid, from both a costeffectiveness and animal welfare perspective. On French Island, Victoria, for example, the relatively low height of the canopy means that a large proportion of koala captures can be completed without climbing the tree, thus the capture process is typically completed in less than ten minutes (Greenfield, 2012). The likelihood of recapturing an individual for multiple treatments will naturally be reduced if koalas fail to remain within a targeted management area following treatment. This may not be an issue if the management aim is simply to inhibit fertility in a pre-determined proportion of the population, because in this situation it matters little whether the same individuals are treated repeatedly, as long as the fertility of the desired percentage of the 178

194 population is inhibited at any one time. However approaches such as immunocontraception, which typically require a booster treatment in the weeks or months following primary inoculation, will be inefficient if treated individuals move beyond a management area and cannot be located for capture a second time. One consistent indirect effect of fertility control on free-ranging female koalas across several studies is the greater occurrence of long-distance dispersal-type movements away from the initial site of capture (Hynes et al., 2011; Greenfield, 2012; my study Chapter 6). During our trials of Suprelorin treatment, a high proportion of female koalas on Kangaroo Island, including Suprelorin-treated, surgically-sterilised and untreated control individuals, undertook long-distance movements (see Chapter 6). Not only would such ranging behaviour substantially reduce the chance of recapturing these individuals in following years, it effectively dilutes the proportion of treated individuals within the landscape. Likewise, movement of untreated animals into the management area will decrease the effectiveness of the control effort (Merrill et al., 2006). Similar movements were also observed in female koalas treated with Suprelorin (Greenfield, 2012) and levonorgestrel (Hynes et al., 2011) on French Island. This effect has also been observed in Suprelorin-treated female eastern grey kangaroos (Wilson, 2012). Increased ranging behaviour and a subsequent reduction in the ability of managers to find and re-treat individuals in following seasons will negatively impact on the success and efficiency of landscape-scale management of large free-ranging koala populations. As a result, fertility control techniques with a longer duration of treatment effect that negate the need for repeat capture of individuals are likely to be the most useful approach. Increased ranging behaviour will result in an increased proportion of the population that will require fertility control to achieve the desired effect, and population models should include a factor for different rates of migration of contracepted animals out of targeted management areas. 179

195 Remote delivery of a fertility control agent, removing the requirement to capture and handle each animal, would greatly enhance the efficiency of a fertility control program involving free-ranging animals. The smaller size of Suprelorin implants (12.5 x 2.3 mm) compared to levonorgestrel (44 x 2.4 mm) provides greater potential for the development of intramuscular remote delivery techniques for this treatment, for example by dart or polesyringe. Trials of darts to facilitate remote delivery of Suprelorin have been conducted (Herbert and Vogelnest, 2007), however the technology is still at the prototype stage. Remote delivery of the larger levonorgestrel is not considered feasible in its current form (Herbert et al., 2010). Remote delivery techniques for other fertility control agents have been developed. For example, a synthetic gestagen, norgestomet, has been successfully delivered to white-tailed deer in the form of a biobullet, in which the hormone is encased in a biodegradable coating which dissolves upon entering the animal (De Nicola et al., 1997). Successful remote delivery of both PZP and GnRH vaccines have been achieved in the field for a range of species (e.g. Kirkpatrick et al., 1990; Curtis et al., 2002) and booster vaccines are often delivered remotely (e.g. Fayrer-Hosken et al., 2000; Rutberg et al., 2013). One study (Roelle and Ransom, 2009), however, demonstrated a higher incidence of reactions at the injection site in horses treated remotely than those treated by hand-injection, possibly due to surface bacteria and debris being pushed into the injection site (Kirkpatrick et al., 2011). The ability to remotely deliver a fertility control agent to a koala would increase the effectiveness for management; however there are some practicalities that would need to be overcome before such delivery could be used. Firstly, koalas have an extremely low percentage of body fat (Cork and Sanson, 1990; Ellis and Carrick, 1992) and comparatively minimal muscle mass (Grand and Barboza, 2001). With few sizeable muscle masses to safely absorb a remotely-delivered hypodermic needle large enough to deliver a solid implant, the risk of bone or organ damage when darting a koala could be considerable. Further, the ability 180

196 of both gestagen and GnRH-agonist contraceptives implants to effectively inhibit fertility in koalas if inserted intramuscularly, rather than subcutaneously, has not yet been tested. The arboreal nature of koalas also presents unique challenges to the accurate firing/delivery of a projectile. One essential element to successfully darting free-ranging wildlife is obtaining a clear path for the projectile, particularly when a high degree of accuracy is required (Roberts et al., 2010). Lynch and Martin (2003) were successful in remotely delivering intramuscular sedative to koalas at heights of 3-15 m using small 0.25 ml darts with a 12 x 1.1 mm needle, although these darts and needles were smaller than those that would be required to deliver current forms of contraceptive implants. On Kangaroo Island, koalas are often at heights greater than 20 m and in complex canopies; darting is unlikely to be optimal under such conditions. This technique is likely to have greater feasibility in areas where tree height is lower and a clear line of sight to the target can be found, e.g. French Island. Use of a pole-syringe to remotely administer contraceptive treatments to koalas from the ground would remove much of the risk associated with darting in relation to impact trauma and accuracy of delivery. Once again, however, complex canopies may complicate this technique and climbing would still be required if the koala was at a height of greater than 8 m (length of a standard catch-pole). One potential issue that arises from remote delivery of fertility control treatments is the problem of recognising previously-treated individuals (Herbert et al., 2010). This is important in regards to efficiency of the management effort and also when considering negative consequences that may result from higher doses for individuals inadvertently treated multiple times within a single season. Repeat treatment with Suprelorin does not appear to have any negative side-effects in koalas (this study; Greenfield, 2012), and Herbert et al. (2006) noted similar findings with eastern grey kangaroos. The development of darts that can simultaneously inject a drug and mark an animal with coloured dye (e.g. Naugle et al., 2002) 181

197 may help resolve this issue, however remote marking is unlikely to last for longer than a single management season. Humaneness The general acceptance of fertility control as a management tool is often due to a perception that these techniques are less invasive or more humane than alternatives, however there is still some risk of adverse impacts on the health of individuals (Nettles, 1997; Munson et al., 2005; Gray and Cameron, 2010) or toxic effects (e.g. Cummins and Wodzicki, 1980). As such, a decision to use a particular fertility control method should be based on knowledge that the risk of harm to a treated individual is minimised and offset by the benefits of treatment at both an individual and population level (Fagerstone et al., 2002; Munson et al., 2005). The relative impacts of the mode of treatment delivery and multiple captures or holding duration on the welfare of individual koalas have been discussed above. In addition, as with any procedure involving physiological manipulation, fertility control techniques have the potential to cause ongoing issues for treated individuals beyond the initial treatment phase, including adverse chronic health problems (see Nettles, 1997) or changes in behaviour and social dynamics (e.g. Whyte et al., 1998). Alternatively, treatment may instead result in improved body-condition of an individual (e.g. Turner et al., 2002) or increased survival (e.g. Twigg et al., 2000), which may result from the elimination of reproductive and lactational stressors (e.g. Kirkpatrick and Turner, 2007). We found no adverse health effects of Suprelorin treatment on female koalas; in fact, Suprelorin-treated individuals maintained higher body weight compared to control animals (Chapter 5), supporting similar findings by Greenfield (2012). This was also true for surgically-sterilised females. This trend was also demonstrated in levonorgestrel-treated koalas that gained and maintained body weight following treatment, compared to control 182

198 animals that had no overall change in their weight-change profile because it fluctuated in response to annual lactational demands (Hynes, 2007). No cases of infection have been reported at the implant or surgery sites of Suprelorin-treated (Chapter 4; Greenfield, 2012), levonorgestrel-treated (Hynes, 2007) or surgically-sterilised (DEWNR, unpublished data) female koalas. In other species, incidences of decreased male body-condition following surgical sterilisation of females in a population (Ji et al., 2000) and increased male harassment of sterilised females (Whyte et al., 1998) have been observed, however these potential sideeffects of this technique have not been investigated in koalas. High survival of treated individuals and their dependent young is desirable for justification of the use of a fertility control method from a welfare perspective. There is no evidence to suggest treatment of female koalas with Suprelorin (Chapter 4), levonorgestrel (Hynes, 2007) or surgical sterilisation (Chapter 4) negatively affects survival rates of treated individuals or their dependent young. The increased body weight of koalas subjected to fertility control suggests that treatment and subsequent release of females from the cost of reproduction could in fact increase the survival and longevity of individuals, as has been seen in other species, (e.g. Twigg et al., 2000; Kirkpatrick and Turner, 2007), however longer studies would need to be undertaken to assess this. The high survival of koalas managed in situ contrasts with a mortality rate of 37.5% observed in the first 12 months for surgically-sterilised koalas that were translocated from Kangaroo Island to mainland South Australia (Whisson et al., 2012). This suggests moving animals to a novel environment may have a greater impact on survival than in situ management, however a year effect cannot be ruled out here and repeat studies should be undertaken over multiple years to confirm the result. Regardless, the relative humaneness of translocation should be considered and further research conducted to determine how this can be addressed, particularly if translocation becomes necessary when the habitat in which the koalas are being managed is so severely degraded that there is a risk of starvation. 183

199 In some situations, the effects of fertility control may not be immediate enough to achieve the management objective of reducing population density and browse-pressure on food trees. In this situation, removal/translocation of animals out of the population may be the only option. Because culling is not an approved management option for koalas (Natural Resource Management Ministerial Council, 2009), translocation to suitable habitat within areas of the koala s former range must be used to achieve this. To prevent overpopulation at translocation sites, animals translocated from Kangaroo Island are surgically sterilised prior to release. However, mortality rates within the first year post-translocation were higher than for animals sterilised and left in situ. While a negative effect on survival has not been observed when using Suprelorin, levonorgestrel or surgical sterilisation in the absence of translocation, increased ranging behaviour by female koalas following treatment appears to be consistent between treatments and the potential welfare impacts of this must be considered by managers. One possible issue with increased ranging behaviour in some areas is that it may expose koalas to an increased risk of mortality from vehicles or dog attack, or because koalas may inadvertently find venture into unsuitable habitat (e.g. Moore and Ali, 1984; Dique et al., 2003). Such issues may be more important when managing koala populations in urban or semi-urban settings and are not considered a major concern on Kangaroo Island. Target specificity and environmental acceptability Ensuring a fertility control treatment affects only the target species is essential to any control operation, as few drugs are species-specific and some may be toxic to non-target species (Bomford and O Brian, 1992). Suprelorin, levonorgestrel and PZP immunisation treatments are not species-specific, having been shown to be an effective contraceptive in a range of species. Target specificity is not an issue for management of koala populations, 184

200 however, as direct capture of individuals is required for delivery of treatment with all current fertility control options. The environmental acceptability of a fertility control method must also be considered before large-scale deployment in the landscape. Residues of a fertility control agent in carcasses may enter the food chain, with potential risks to non-target species (Bomford and O Brian, 1992). Suprelorin has high environmental acceptability as the active ingredient, deslorelin, is not active by the oral route (Padula, 2005) and therefore there is no chance of secondary exposure for other animals through predation or scavenging of a treated animal. Likewise, there is no evidence of conventional PZP passing through the food chain (Kirkpatrick et al., 1990; Kirkpatrick and Rutberg, 2001). In contrast, steroid-based compounds (e.g. levonorgestrel) have a high resistance to biodegradation (Besse and Garrick, 2009) and are active via the oral route (e.g. Gaspard et al., 1984), however it is unlikely that secondary exposure to ingested levonorgestrel implants would result in any more than a minor transient effect to a predator/scavenger as the implant passed through their system. Cost-effectiveness The economic cost of delivering fertility control to a sufficient proportion of the population to achieve the desired management outcome is an essential consideration for managers. Determination of the relative cost-effectiveness of a particular fertility control option in varied environments requires detailed analyses of the costs and benefits involved, however consideration of several key factors at the individual animal level are critical in this process (see Herbert, 2012). The costs involved in administering the fertility control agent (including the cost of locating animals and their capture, the cost of the fertility control agent itself, and any animal transport and/or holding costs) provide an estimate of treatment per individual. The net period of induced infertility of a treatment and the reproductive lifespan of females within the target population then allows determination of the number of successive 185

201 treatments needed to inhibit fertility in an individual for its lifespan if ongoing control is desired. As all current fertility control options currently require capture of koalas prior to treatment, the costs associated with locating and capturing animals at a particular site are identical no matter which treatment is used. These costs will vary greatly between locations however, due to variation in tree height and habitat complexity and subsequent ease of locating and capturing animals. The ongoing costs of monitoring both the koala population being managed (e.g. population trends, condition and survival of treated animals, fecundity) and the condition and recovery of habitat must also be considered. These costs are essential to determine the success of any management activity, but will vary depending on a range of factors. These include the size and density of the koala population, the area targeted for management (e.g. extent, habitat type, access), and the fertility control method (and its impacts on animal behaviour). Translocation will significantly increase the cost of monitoring should, as is recommended, monitoring of koala welfare, density and movement and their impact on vegetation be undertaken at the release site. Reduction in population size and environmental damage Control of a wildlife population, be it to reduce population density or limit population growth, is not in itself an objective; rather it is simply a management action (Sinclair et al., 2006). The success of any control operation must therefore be measured by a reduction of the damage caused by the wildlife population, or an increase in the density or condition of the value the management action aims to protect. 186

202 Management of high-density koala populations in south-east Australia is conducted with the objective of reducing browse damage, preventing forest death and subsequent suffering and death of individual koalas (Department of Sustainability and Environment, 2004; Natural Resource Management Ministerial Council, 2009). Depending on the scale of the problem, koala population density may need to be reduced immediately or managers may be satisfied with simply limiting the population to its current level to avoid issues in the future. In isolation, fertility control only reduces recruitment of new young animals into the population (and potentially limits immigration if treated residents remain in situ) and will have no immediate impact on population density. Removal of animals from the population provides an immediate reduction in population size and therefore browse-pressure, however, as culling is not an approved management option (Natural Resource Management Ministerial Council, 2009), animals must be translocated to suitable habitat within areas of their former range. To prevent overpopulation at translocation sites, these animals are surgically sterilised prior to release. 187

203 Table 8.1: Successful female koala fertility control agents and their effects on individuals. LEVONORGESTREL OESTRADIOL SUPRELORIN PZP VACCINE TUBAL LIGATION TREATMENT DETAILS Commercially available? Yes No Yes No NA Active ingredient Synthetic progestin Synthetic gestagen GnRH agonist (deslorelin) Porcine zonae pellucidae NA Implant dimensions 44 x 2.4 mm Custom. Trial implants were k 10 x 4.5 mm and 5 x 4.5 mm 12.5 x 2.3 mm NA NA Standard dose 70 mg NA 4.7 mg NA NA Trial dose 70 mg a,,k Dose not stated. 4.7 mg b,c,f ml complete Freund s adjuvant. Booster: NA Primary: 100 µg PZP in 0.5 ml citrate buffer Unspecified PZP in incomplete Freund s adjuvant. d Potential for remote delivery Not in present form Not in present form Yes Yes No Administration SC and IM injection under sedation + Small incision and SC Small incision and SC Endoscopic cauterisation SC injection booster injection weeks/months after injection under sedation injection under sedation under general anaesthetic primary inoculation Koala holding time > 3 hours > 3 hours <10 minutes > 3 hours 1-2 days PHYSIOLOGICAL EFFECTS Follicular development No effect #,n No effect Inhibited #,o Minimal or no effect d No effect FSH secretion Minimal or no effect #,n Minimal or no effect Reduced #,o No effect No effect LH secretion Surge inhibited #,n Surge inhibited Reduced b No effect No effect Ovulation Inhibited #,n Inhibited Inhibited #,o Inhibited #,r No effect Oestrous and mating No effect* No effect* Suppressed #,o No effect No effect* Ovarian progesterone Suppressed a Suppressed Suppressed c No effect No effect Lactation No effect #,q No effect No effect #,p No effect No effect EFFICACY Contraceptive success 100% a 92-95% k % b,c,f 70% d 100% Duration of contraceptive effect >8 years a,e >4 years k 1-2 years b,c,f 2-3 years d Permanent Min. number of captures required to maintain infertility of an individual for its x 2 1 lifetime Reversibility of contraceptive effect Yes a Assumed k Yes c Assumed d No Non-responders No a No k Yes #,s Yes #,j NA Weight trend following treatment Increase a Increase k Increase b,c Unknown Increase b Behavioural effects Minimal g Minimal k Minimal b,c Minimal d Minimal b Observed negative side-effects None detected None detected k None detected None detected None detected Potential negative side-effects Increased ranging behaviour g At high doses: mortality, cystic endometrial hyperplasia,increased vascularity and oedema of the endometrium,glandular proliferation of the oviducts l Increased ranging behaviour c Swelling at injection site, lesions, selection against immune response i,j *Oestrous cycle is not suppressed and koalas likely cycle multiple times. # Inferred from effects on other species. (a) Hynes et al., 2010; (b) this study; (c) Greenfield, 2012; (d) Kitchener et al., 2009b; (e) K. Handasyde, pers. comm. (f) Herbert, 2002; (g) Hynes et al., 2011; (h) Whyte et al., 1998; (i) Leenaars et al., 1998; (j) Cooper and Larson, 2006); (k) Middleton et al. 2003; (l) Handasyde et al. 1990; (m) Ji et al. 2000; (n) Hynes et al. 2007; (o) Herbert et al. 2004; (p) Herbert et al. 2006; (q) Nave et al. 2002; (r) Kitchener et al. 2002; (s) Herbert et al Increased male harassment h, decrease in male body condition m 188

204 8.1 Conclusion Active management of the introduced koala population on South Australia s Kangaroo Island has been in progress since 1997 in an attempt to reduce overbrowsing and eucalypt death caused by this high-density population. The current landscape-scale government management program involves surgical sterilisation and translocation as a means to reduce population density and limit population growth. The success of the program, in terms of the management objectives, is evident in an approximately 50% reduction in the estimated island-wide population size between 2000 and 2010 and the recovery of overbrowsed habitat. Although successful, surgical sterilisation is costlier, less efficient to deliver and relatively invasive compared to other available fertility control methods. Further, mortality rates of koalas, within the first year post-translocation, were higher than for animals sterilised and left in situ on Kangaroo Island (at least during the study period; see Chapter 7), thus the relative humaneness of the former approach must be questioned and further research conducted to determine if the result is consistent in other years and how this might be addressed. Despite this, translocation of koalas from Kangaroo Island to the South Australian mainland may be a necessary ongoing component of the management program in the medium term in order to immediately reduce koala density in vulnerable habitat. In addition, it is clear that ongoing management of this population will be required to capitalise on previous management efforts and to ensure that koala population density does not return to previous high, unsustainable levels. Investigations into alternative population control techniques are therefore necessary to guarantee that future management is conducted as efficiently as possible whilst minimising welfare impacts for the animals. One alternative management tool trialled, subcutaneous contraceptive implants (Suprelorin), inhibited fertility in a high proportion of breeding age female koalas for a period of 1-2 years, and there was no evidence of any negative impacts on the body weight or survival of treated individuals. Consistent with observations during other fertility control 189

205 studies, a high proportion of Suprelorin-treated female koalas undertook long-distance movements, however this behaviour was unrelated to treatment as it was also observed in control (non-treated) and surgically sterilised koalas. The high efficacy and lack of observed side-effects of Suprelorin suggest it has potential for use for the management of some koala populations, however the relatively short duration of contraception provided to female koalas by Suprelorin suggests this treatment is less suitable than alternatives for managing this species on Kangaroo Island. Very tall trees in many parts of the island and the high proportion of female koalas that make long-distance movements means that recapture of treated individuals for re-treatment would be relatively difficult. Thus efficient management of koalas on this island ideally requires a longer-acting form of fertility control. Levonorgestrel implants provide contraception for the majority of a female koala s breeding lifespan however, unlike surgical sterilisation, can be delivered on-site without the need for transport or extended holding of animals. Levonorgestrel is currently used by managers on French Island, Victoria, and offers a viable, safe alternative option for koala management on Kangaroo Island. The ease of administering Suprelorin, at the site of capture, without the need for sedation or anaesthesia, means Suprelorin is the least invasive of all current fertility control options, a fact that cannot be ignored when considering management options for a much-loved high-profile species. Even though Suprelorin is not currently considered the most efficient treatment option for management of the Kangaroo Island koala population, it has potential for use elsewhere. It would be ideal for small captive or semi-captive populations or in freeranging populations where maintenance of the maximum possible genetic variation is highly desirable. In the future, the development of a remote delivery technique for Suprelorin implants would greatly enhance the efficiency of delivery and increase the logistic viability of this treatment for use on Kangaroo Island. There are a number of issues that still need to be overcome before remote delivery will be feasible, however research into these techniques should continue. 190

206 Management recommendations This thesis provides a broad assessment of a commercially-available GnRH agonist implant, Suprelorin, as an alternative management tool for controlling koala population growth on Kangaroo Island. The assessment includes investigation of the treatment s efficacy for controlling the fertility of free-ranging female koalas, and its impact on a treated individual s health and behaviour. Where possible, Suprelorin treatment has been compared to the current management practice of surgical sterilisation. A summary of female koala breeding characteristics on Kangaroo Island, and an appraisal of the suitability of translocation as a tool for managing overabundant species, are also included for context and completeness. An investigation into the breeding demographics of female koalas on the island (Chapter 3) provides information critical to managers wanting to accurately model growth of this population and the effectiveness of population management strategies. Results from experimental trials (see Chapters 4-7) contribute new information regarding the use of hormonal implants for managing free-ranging koala populations, whilst Chapter 8 frames these results in the context of available management alternatives and discusses the relative suitability of alternatives in relation to a range of key criteria. Some clear advice for managers is the result, however there is further research required to confirm treatment effects. Considerations for management and future investigation are presented in Table

207 Table 8.2: Considerations for management and further research resulting from this study. Study results Management consideration Future research Chapter Spatial variation in breeding seasons exists between southern populations of koalas. Models of population growth and for testing management strategies should incorporate location-specific reproduction parameters. 3 Suprelorin treatment successfully inhibits fertility in the majority of treated female koalas for at least one breeding season. Treat females with deslorelin immediately prior to breeding season to maximise duration of contraceptive effect. Suprelorin treatment is not an efficient and cost-effective tool compared to surgical sterilisation or levonorgestrel treatment if contraception of a large proportion of the population for an extended period is the primary management aim. Investigation into dose-response effects and the mechanisms of non-response. Development and trials of remote-delivery methods for Suprelorin treatment. 4 Treatment with Suprelorin may be a useful management tool in situations when animal capture is straightforward, and where reversibility of the contraceptive effect is desired. The development to remotely deliver a Suprelorin implant to an individual koala would greatly enhance its efficiency as a management tool and provide a clear advantage over alternatives No adverse health effects were evident as a result of Suprelorin treatment of adult female koalas, although two pouch young conceived by Suprelorin-treated individuals did not survive. Suprelorin can be considered a safe and humane method of fertility control of free-ranging adult female koalas. Determine whether fertility control treatments, i.e. removal of the energetic requirements for reproduction, translates into increased longevity of treated individuals. Investigate the potential for Suprelorin treatment to disrupt lactation in koalas. 5 Surgical sterilisation may impact negatively on the survival of dependent young post-surgery. The impact of a fertility control treatment (including both the capture and delivery process and subsequent treatment effects) on the welfare of dependent young must be considered when developing any management strategy. Further investigation into the impact of surgical sterilisation on the survival of advanced young carried by sterilised female koalas should be conducted with larger sample sizes and over multiple seasons to rule out potential year effects

208 High level of breeding dispersal exhibited by female koalas on Kangaroo Island, regardless of treatment. The majority of longdistance movements by adult female koalas occur leading up to and during the breeding season. The level of breeding dispersal exhibited by female koalas on Kangaroo Island should be factored into management decisions, particularly decisions relating to identification of target areas and management scale. High levels of breeding dispersal is likely to dilute the effectiveness of a management effort if treated koalas emigrate from a targeted area and are replaced by fertile animals. Conversely, high levels of movement will help disperse treated individuals throughout the landscape. Further investigation into the extent of movement of female koalas between management units (i.e. catchments). 6 Translocating koalas to a novel environment may have a greater impact on their survival than in situ management. The relative humaneness of translocation should be factored in to management decisions. Post-release monitoring of translocated individuals should be a key component of any management program involving translocation. Repeat studies using larger sample sizes should be undertaken over multiple years to confirm the result and rule out a potential year effect

209 194