UBC Theses and Dissertations

UBC Theses Logo

UBC Theses and Dissertations

Does immigration "rescue" populations from extinction? Clinchy, Michael 1999

Your browser doesn't seem to have a PDF viewer, please download the PDF to view this item.

Item Metadata

Download

Media
831-ubc_1999-463338.pdf [ 13.37MB ]
Metadata
JSON: 831-1.0089265.json
JSON-LD: 831-1.0089265-ld.json
RDF/XML (Pretty): 831-1.0089265-rdf.xml
RDF/JSON: 831-1.0089265-rdf.json
Turtle: 831-1.0089265-turtle.txt
N-Triples: 831-1.0089265-rdf-ntriples.txt
Original Record: 831-1.0089265-source.json
Full Text
831-1.0089265-fulltext.txt
Citation
831-1.0089265.ris

Full Text

DOES IMMIGRATION "RESCUE" POPULATIONS FROM EXTINCTION? by MICHAEL CLINCHY B.Sc, University of Toronto, 1988 M.Sc, Queen's University, 1990 A THESIS SUBMITTED IN PARTIAL FULFILMENT OF THE REQUIREMENTS FOR THE DEGREE OF DOCTOR OF PHILOSOPHY in THE FACULTY OF GRADUATE STUDIES (Department of Zoology) We accept this thesis as conforming to the required standard: THE UNIVERSITY OF BRISTISH COLUMBIA July 1999 © Michael Clinchy, 1999 In presenting this thesis in partial fulfilment of the requirements for an advanced degree at the University of British Columbia, I agree that the Library shall make it freely available for reference and study. I further agree that permission for extensive copying of this thesis for scholarly purposes may be granted by the head of my department or by his or her representatives. It is understood that copying or publication of this thesis for financial gain shall not be allowed without my written permission. The University of British Columbia Vancouver, Canada Department Date DE-6 (2/88) ABSTRACT I measured the rate of immigration by female common brushtail possums (Trichosurus vulpecula) in response to the removal of resident breeding females, in a landscape with no physical barriers to dispersal. I removed 10 residents from one 36 ha study grid and 9 from another, and monitored immigration over the next two years. Only one immigrant settled in one of the two removal areas. Sixteen breeding females resident on the periphery of the removal areas expanded their ranges into the removal areas. The one immigrant was a subadult that did not give birth in the breeding season following her arrival. Parentage analysis using microsatellite DNA indicated that the immigrant had moved only one home range away from her putative mother's home range (= 200 m). All of the known daughters of resident females settled beside their mothers. Parentage analysis indicated that 39 % of adjacent pairs of resident females were putatively mother and daughter, which is close to the 42 % expected if daughters always settle beside their mothers. The sex ratio of pouch-young was significantly male-biased, as predicted by the 'local resource competition' hypothesis, if most males disperse and most females settle beside their mothers. A deterministic, stage-based model of demography indicated that the birth rate was insufficient to balance the death rate (r = - 0.1), suggesting that the site was a 'dispersal sink'. However, even with immigration, the projection was that density would decline by 84 %. Neither 'old age', starvation, predation or disease could explain 11 of the 24 deaths among resident females. Most of these females demonstrated prior symptoms of stress. Females that were captured and handled more frequently had a significantly lower probability of survival, and the estimated adult survival rate was significantly lower than that expected from the observed age distribution. I suggest apparent 'dispersal sinks' may often be of our own making. Deaths due to capture and handling are analogous to removals. Consequently, since there were evidently more than 19 'removals', I conclude that the results of the experiment likely overestimate the importance of immigration in replacing losses among breeding females. ii T A B L E OF CONTENTS Abstract ii Table of Contents iii List of Tables vi List of Figures x Preface xiii Acknowledgements xiv CHAPTER 1: Does immigration "rescue" populations from extinction? General introduction 1 References 8 CHAPTER 2: Does immigration "rescue" populations from extinction? Implications regarding movement corridors and the conservation of mammals Introduction 11 Alternative hypotheses 12 Behavioural data 13 Conclusions 17 References 18 CHAPTER 3: Methodological problems in the study of immigration and extinction Introduction 25 The correlational-survey approach 25 Identifying immigrants in open populations in the field 27 An integrated approach 30 References 33 CHAPTER 4: Does immigration "rescue" populations from extinction? Evidence from a large-scale field experiment on common brushtail possums Introduction 40 Methods Study species 46 Study site 48 Detection, capture, and handling 50 Definitions and experimental design 52 Radio telemetry and den mapping 56 Live-trapping 58 Spotlighting and darting 59 Complete enumeration 60 Adequacy of location data from radio telemetry 71 Collection of tissue samples and analysis of microsatellite DNA 73 iii Results Female immigration following removals 74 The vacuum effect and breeding dispersal 76 Potential immigrants captured in the core 77 Adverse effects of capture and handling and the measured rate of immigration 79 Locations of known mothers and daughters 82 Locations of mothers and daughters identified using microsatellite DNA 83 Direct observations of restricted movements between the grids 84 Restricted movements between the grids identified using microsatellite DNA 85 Sex ratios of pouch-young 85 Discussion 86 References 93 CHAPTER 5: Connectivity or carnivory: what's more important to the demography of common brushtail possum populations? Introduction 139 Methods 142 Age estimates from cementum annuli 142 Date of birth of pouch-young 143 Results Seasonality of births 144 Sex ratios of dependent young 145 Stages in the weaning of young 146 Pre-weaning survival of young 147 Pre-weaning loss of young as a result of capture and handling 149 Pre-weaning loss of young as a result of predator chases 154 Pre-weaning loss of young as a result of infanticide 154 Pre-weaning loss of young as a result of food shortage 155 Post-weaning survival of young 160 Age at first breeding 168 Reproductive success of first-time breeders 169 Discussion 174 References 179 CHAPTER 6: Dispersal sinks and handling effects: interpreting the role of immigration in common brushtail possum populations Introduction 212 Methods 215 Results Adult female survival 217 A model of possum demography 220 Wobbly possums 224 Other indices of poor condition 229 Weight loss associated with trapping, among apparently healthy animals 231 Deaths due to predation 232 Necropsies 234 Age at death 240 Survival and the frequency of handling 243 Population growth rates and "censored" survival estimates 246 iv Discussion 247 References 253 CHAPTER 7: Conclusions and implications for conservation General discussion 282 References 288 APPENDIX 1: Assumptions underlying the analysis of microsatellite DNA Introduction 290 Linkage disequilibrium and deviations from Hardy-Weinberg equilibrium (HWE) 290 Parentage analysis 292 Absolute likelihoods 298 References 300 APPENDIX 2: Estimating the date of birth of pouch-young Introduction 304 Derivation of methods 304 Protocols adopted 307 References 309 v LIST OF TABLES Table 2.1 Summary of results of examining populations illustrated in Figure 2.2 and asking the question "do both populations persist from time 1 to time 2?" Table 2.2 Contingency table for analysis of relationship between proximity and persistence using evidence from Table 2.1 and Figure 2.2. 21 22 Table 3.1 Distance to the nearest occupied patch at time 1, as compared between populations found to be extinct or extant at time 2, based on results from four surveys of the Bodie, California, pika metapopulation, as reported in Smith and Gilpin (1997, their Fig. 2) 37 Table 4.1 Number and date of, and level of effort associated with, full-grid trapping sessions on the West grid. 102 Table 4.2 Number and date of, and level of effort associated with, full-grid trapping sessions on the East grid. 103 Table 4.3 Level of effort associated with, and number of animals seen during spotlight transects in the core areas of the two study grids. Table 4.4 Level of effort associated with, and number of animals seen during spotlight transects on the periphery of the two study grids. Table 4.5 Population size estimates regarding all adult and subadult females on the West grid. 104 105 106 Table 4.6 Population size estimates regarding all adult and subadult females on the East grid. 107 Table 4.7 Population size estimates regarding all adult and subadult males on the West grid. 108 Table 4.8 Population size estimates regarding all adult and subadult males on the East grid. 109 Table 4.9 Allele frequencies observed at six microsatellite loci. 110 Table 4.10 Putative mother-daughter relationships identified among adult (and subadult) 111 females on the West grid. Table 4.11 Putative mother-daughter relationships identified among adult (and subadult) females on the East grid. 112 Table 4.12 Differences in mean likelihood scores, as shown in Tables 4.10 and 4.11, between adjacent and non-adjacent putative parent-offspring pairs of adult (and subadult) females. 113 Table 4.13 Reported sex ratios of pouch-young. 114 vi Table 5.1 Differences in the reproductive biology of common brushtail possums in disturbed and undisturbed populations. 185 Table 5.2 Number of resident adult females producing a primary young in either autumn or winter, compared to the number of females that did not reproduce. 186 Table 5.3 Sex ratios (F:M) of dependent young, categorized by study area and breeding season. 187 Table 5.4 Pre-weaning survival of primary young. Table 5.5 Mean interval, in days, between occasions on which females were handled, while they were carrying a pouch-young, prior to 175 days after the birth of their young, as compared between instances where the young was known to have survived to the onset of weaning (at 175 days), and where it did not. 188 189 Table 5.6 Pre-weaning loss of young in relation to ear-tagging. 190 Table 5.7 Survival of young to October-November, 1996, as compared between mothers 191 subject to repeated capture and handling (On-grid) and others that had not been previously trapped and handled (Off-grid). Table 5.8 Predator activity during the pre-weaning phase of development of primary young, from mid-March to mid-December, in 1995 and 1996, based on sightings, sounds, signs, and diagnoses of the proximate cause of death of adult possums. 192 Table 5.9 Survival of primary young to August 1, in both 1995 and 1996, on both principal study grids. 193 Table 5.10 Survival of pouch-young to the onset of weaning (175 days) in 1995 and 1996, 194 among females demonstrating any or all of four symptoms of poor condition. Table 5.11 Changes in the status of individual adult female possums examined in a 195 given season in 1995 for evidence of rump wear as compared to the same individual's status in the same season in 1996. Table 5.12 Median difference in mean weight (g) of individual adult possums measured in a given season in 1996 as compared to the same individual's mean weight in the same season in 1995. 196 Table 5.13 Weight (g) of adult females weighed with or without a young in the pouch, in October-November, 1996, as compared between On-grid and Off-grid locations. 197 Table 5.14 Number of young tagged as pouch-young in the year indicated, that were later seen as subadults, independent of their mothers. 198 Table 5.15 Mean, median, and other summary statistics, regarding the distribution of mean autumn body weights (g), among known-aged animals. 199 Table 5.16 New individuals detected during grid trapping from the beginning of spring 1995 onwards. 200 vii Table 5.17 Birth dates of pouch-young (expressed as the number of days from the beginning of autumn) among first-time breeders and females that had given birth at least once before (repeat breeders). 201 Table 5.18 Birth dates of pouch-young (expressed as the number of days from the beginning of autumn) in relation to the pre-weaning survival of primary young (see Table 5.4) on each of the two principal study grids in each of the two complete years for which data were available. 202 Table 6.1 Definitions regarding the parameters used in the model of possum demography illustrated in Figure 6.4. 259 Table 6.2 Sources of parameter values for the Paddys Land population, used in the model of possum demography illustrated in Figure 6.4. 260 Table 6.3 Sources of parameter values for the Orongorongo Valley population, used in the model of possum demography illustrated in Figure 6.4. 261 Table 6.4 Population growth rate (r) predicted using the model illustrated in Figure 6.4, as regards possums at the Paddys Land site, given three sets of assumptions regarding their breeding biology, and as compared to the predicted r at the Orongorongo Valley site. 262 Table 6.5 Total number of trap-captures per season, in each year, on the two principal study grids. 263 Table 6.6 Number of times animals were observed to have either lost weight or not lost weight between successive occasions on which they were weighed. 264 Table 6.7 Total number of times females were trapped and handled per season, in each year, on the two principal study grids. 265 Table 6.8 Potential predators of common brushtail possums recorded at the Paddys Land site over the three years of the study. 266 Table 6.9 . Relationship between the results of tests conducted to determine whether the bandaging wrapped around an individual's radio-collar was inundated with blood, and the condition of the individual's carcass, in 21 of the 22 cases where a radio signal was tracked to a dead female. 267 Table 6.10 Prevalence and intensity of parasitic infections among the 23 possums from the Paddys Land site that were necropsied, categorized by the fate of the host. 268 Table 6.11 Prevalence and intensity of ectoparasitic infections compared between adult females observed with symptoms of rump wear on at least one occasion, versus females never observed with symptoms of rump wear. 269 Table 6.12 Prevalence and intensity of Parastrongyloides trichosuri infections, categorized by the fate and condition of the host. 270 viii I Table 6.13 Mean (and S.E.) age in years, as determined from cementum annuli, among the 23 possums that were necropsied, categorized by fate (Died versus Killed) and sex. 271 Table 6.14 Observed number of individuals in each of two age classes, among 38 possums aged from cementum annuli, as compared to the expected number in each class if survival is assumed to be 81.3 % per annum. 272 Table 6.15 Seasonal survival of radio-collared adult females (1997 data excluded), under four scenarios regarding the treatment of censored data, and the effects on the value of r predicted by the model of possum demography (assuming 'Typical' parameter values for the Paddy Land site). 273 Table A 1.1 Observed (H0) end expected (HE) heterozygosity at each of the six microsatellite loci described in Table 4.9, as calculated using genotypic frequency data from both grids, both combined and separately. 301 Table A1.2 Known mother-young pairs where both mother and young were successfully typed at one or more of the microsatellite loci described in Table 4.9. 302 Table A1.3 Average exclusion probabilities based on allelic frequencies on each grid, of each of the six microsatellite loci listed in Table 4.9. 303 ix LIST OF FIGURES Figure 2.1 Conventional portrayal of relationship between proximity and persistence. 23 Figure 2.2 Environmentally induced synchrony of population dynamics signified 24 by clustering of populations. Figure 3.1 Steps in delineating a peripheral "buffer strip" for use in the 38 unambiguous identification of "true" immigrants. Figure 3.2 Example of how immigration may be overestimated if handling has 39 a significantly adverse effect on resident reproduction or survival. Figure 4.1 Schematic map of the study site showing the relative position of the two 115 principal study grids. Figure 4.2 Schematic map of the West grid. , 116 Figure 4.3 Schematic map of the East grid. 117 Figure 4.4 Location of all females captured on the West grid prior to August 1, 1995. 118 Figure 4.5 Location of all females captured, or known to be present, on the West grid, 119 between August 1, and September 15, 1995. Figure 4.6 Location of all females captured, or known to be present, on the West grid, 120 between September 15, 1995, and March 15, 1996. Figure 4.7 Location of all females captured, or known to be present, on the West grid, 121 between March 15, 1996, and August 1, 1996. Figure 4.8 Location of all females captured, or known to be present, on the West grid, 122 between August 1, 1996, and March 15, 1997. Figure 4.9 Location of all females captured, or known to be present, on the West grid, 123 between March 15, and August 1, 1997. Figure 4.10 Location of all females captured on the East grid prior to August 1, 1995. 124 Figure 4.11 Location of all females captured, or known to be present, on the East grid, 125 between August 1, and September 15, 1995. Figure 4.12 Location of all females captured, or known to be present, on the East grid, 126 between September 15, 1995, and March 15, 1996. Figure 4.13 Location of all females captured, or known to be present, on the East grid, 127 between March 15, 1996, and August 1, 1996. Figure 4.14 Location of all females captured, or known to be present, on the East grid, 128 between August 1, 1996, and March 15, 1997. x Figure 4.15 Location of all females captured, or known to be present, on the East grid, 129 between March 15, and August 1, 1997. Figure 4.16 Cumulative number of newly-identified females initially captured in the core 130 of one of the two study grids, from the start of spring 1994, until the end of the study in July 1997. Figure 4.17 Cumulative number of newly-identified females initially captured on the 131 periphery of one of the two study grids, from the start of spring 1994, until the end of the study in July 1997. Figure 4.18 Cumulative number of newly-identified males initially captured in the core 132 of one of the two study grids, from the start of spring 1994, until the end of the study in July 1997. Figure 4.19 Cumulative number of newly-identified males initially captured on the 133 periphery of one of the two study grids, from the start of spring 1994, until the end of the study in July 1997. Figure 4.20 Plot of the minimum number of females known to be alive on a grid 134 (MNA-Grid), in a given trapping session, versus the Jolly-Seber estimate of the number (N) of females present. Figure 4.21 Proportion of full-grid trapping sessions during which 135 radio-collared females were captured. Figure 4.22 Total number of dens used by a given female as compared to 136 the total number of radio fixes for that female. Figure 4.23 Total denning range length (the maximum distance between any two 137 dens, in m) calculated for a given female, as compared to the total number of radio fixes for that female. Figure 4.24 Mean (+/- S.E.) proportion of radio fixes tracked to each of the array of dens 138 used by a given female, among females located a minimum of 15 times. Figure 5.1 Deviation, in years, between age estimates, based on cementum annuli, 203 made using two teeth (PM1 and M3) from the same individual. Figure 5.2 Number of births per week during autumn and winter in each of 204 the three years of the study. Figure 5.3 Number of observations concerning the three stages in the 205 weaning of young, vs. the age of the young. Figure 5.4 Percentage of adult possums seen feeding upon being released from a trap, 206 in each season, among animals that were known to have survived to the end of the season. Figure 5.5 Percentage of adult female possums demonstrating symptoms or behaviours 207 potentially indicative of poor condition, in each season, among animals that were known to have survived to the end of the season. xi Figure 5.6 Mean (+ S.E.) body weight (g) of adult possums, categorized by season. 208 Figure 5.7 Total monthly rainfall (solid black line) recorded at the Kookabookra weather 209 station (20 km from the Paddys Land site) over the course of the study. Figure 5.8 Body weight (g) of known-age pouch-young plotted against age in days. 210 Figure 5.9 Mean autumn body weight (g) of known-age adults and subadults, 211 plotted against age in years. Figure 6.1 Number of radio-collared adult females at risk, from the beginning 274 of autumn 1995, until the end of the study in July 1997. Figure 6.2 Cumulative probability of survival of radio-collared adult females, 275 based on the Kaplan-Meier estimator, modified for staggered entry. Figure 6.3 Thirteen-week seasonal survival of radio-collared adult females, 276 based on the Kaplan-Meier estimator, modified for staggered entry. Figure 6.4 Structure of the model of possum demography . 277 Figure 6.5 Number of individuals demonstrating a given rate of proportionate weight loss, 278 measured as the percentage of body weight lost within 5 days, among: possums found moribund or dead in a trap (in grey) and apparently healthy animals (in black). Figure 6.6; Number of individuals (N = 38) aged 2 to 14 years (open bars), as determined 279 from cementum annuli, compared to the expected frequency distributions (given 38 individuals) if survival is assumed to be either 81.3 % (black line), or 89.9 % (grey line), per annum. Figure 6.7 Cumulative probability of survival of radio-collared adult females categorized 280 by median handling interval, based on the Kaplan-Meier estimator, modified for staggered entry. Figure 6.8 Change in population size with time (in years) predicted given five 281 different population growth rates (r). Figure A2.1 Head length (in mm) of pouch-young versus known minimum (grey circles) 310 and maximum (open circles) age (in days) based on trapping records. xii PREFACE Publications Chapter 2 of this thesis has already been published as: Clinchy, M. 1997. Does immigration "rescue" populations from extinction? Implications regarding movement corridors and the conservation of mammals. Oikos, 80: 618-622. xiii ACKNOWLEDGEMENTS All research is collaborative research. In initial drafts, I used "we" instead of "I" to reflect the fact that this is a collaborative effort, and I only reluctantly changed this to conform with the required use of "I" in a thesis. I had four principal collaborators whose help was indispensable: Liana Zanette, Beryl Clinchy, and Win and Len Perry. It was Liana's scholarship that put food on the table for the four years we were in Australia, and it was Liana's scholarship that allowed me to obtain a visa to work in Australia. It was mostly mom's money that paid for the research. God forbid that the scum-sucking capitalists that run this country should have to pay taxes to provide funding, far better that working people should have to bankrupt themselves! Viva la Revolution!!! Aside from material support both Liana and mom provided an inexhaustible supply of love and unfailing faith in me. Win and Len Perry's generosity was truly astounding. Not only did they let me disturb their solitude, they provided me with a place to stay, on the spot data collection, and truck repairs. More importantly, they gave me infinite cups of tea and hundreds of hours of great (and sometimes heated) conversation. I miss them very much. Special thanks as well to Bert Jenkins for saving my sanity (and ruining my waistline). I wanted to work with Charley Krebs because I greatly admired the clarity and rigour of his research. As a born cynic and skeptic the highest praise I can offer is that I admire Charley even more now. Charley's advice has been invaluable, because I knew if I could convince Charley, the idea must be right. Peter Jarman acted as my unofficial co-supervisor in Australia. Peter was immensely generous and always supportive of my research. I had hoped that the continuation of the research at Paddys Land would partly repay my debt to Peter and my biggest regret is that this did not come to pass. Hopefully my thesis will provide an impetus for some more inspired student of Peter's to restart the research there. I would also like to thank the members of my research committee: Judy Myers, Dolph Schluter, Tony Sinclair and Jamie Smith. I first met Andrea Taylor at the Australian Mammal Society meeting in September, 1995. I knew what information I wanted to obtain from genetics data when I first spoke to Andrea, but I am still occasionally struck with disbelief that it actually all worked out. Andrea not only accomplished the mammoth task of conducting all the typing, but also undertook the even greater task of patiently answering my endless questions about genetics. Dave Spratt and Pete Haycock undertook the unenviable task of poking through all the dead possums. I long ago concluded that wading through rotting carcasses all day was what was responsible for the bizarre sense of humour shared by most parasitologists. David Obendorf conducted the histopathology. I am delighted that my collaboration with Dave, Pete and David has produced results of interest to all concerned. Steve Hum and Bob Coverdale, formerly of the now defunct Armidale Regional Vet Lab, generously provided assistance, advice and facilities in which to conduct the necropsies. The N.S.W. government's decision to close the Lab was another triumph of economic rationalism. Murray Efford kindly provided me with a stack of invaluable reports regarding the Orongorongo Valley population. Paul (Dickson) Forest and Claudia Frosch were both outstanding field assistants and good friends. Paul's unceasing optimism was an excellent foil to my unending cynicism. Claudia was the polar opposite, always on the lookout to protect herself from having a single optimistic thought cross her mind. Over the years many people at UBC have provided assistance.and advice, thanks especially to: Lance Barrett-Lennard, Alistair Blanshard, Cole Burton, Andrea Byrom, Dennis Chitty, Dan Haydon, Karen Hodges, Greg Hood, Tim Karels, Alice Kenney, Andy Kliskey, Don Reid, Christoph Rohner, Karl Vernes, Irene Wingate and Deb Wilson. Stuart Green deserves special commendation for spending so many hours in the middle of the night in the middle of nowhere staring up at tranquilized possums hanging by their tails. I would also like to thank: Nihal Agar, John Duggin, Frank Dyer, Kate MacGregor, Annette McLeod and Sally Radford. Christa Clinchy, Andy Collins, Georgia de Biasi, Lindsay Fulloon, Nicole Gammie, Peter Hancock, Rick James, Robyn Martin, Natasha Noble, Josh Rapp, Lyndal Redman, Brit Rollo, Rebecca Scrivener, Michael Stanford, Piers Thomas, Steve Tremont, Jeanette van der Lee, Will Weir, Susan Wright and Gilly Young all provided valuable assistance in the field. For hospitality and friendship, thanks to: Chris Johnson, Allen Mcllwee and Richard Rowe, in Townsville; Elsie Krebs and David Green, in Canberra; and Murray Evans, for making Armidale a little more bearable. xv CHAPTER 1 DOES IMMIGRATION "RESCUE" POPULATIONS FROM EXTINCTION? The focus of my research has been on the question: does immigration "rescue" populations from extinction? Despite the fact that dispersal is the process underlying metapopulation dynamics, this question has received scant attention in the literature on metapopulations, evidently because the answer is always assumed to be "yes". On the other hand, researchers interested in cyclic species of small mammals might not only answer "no", but may also suggest that immigration can induce local population extinctions. I will begin by briefly sketching the historical origins of these different attitudes to the role of immigration in animal population dynamics. Ultimately, I think the difference in attitude derives from both: 1) an artificial separation of the ideas of distribution and abundance; and 2) a preponderance of generalizations regarding the role of dispersal in local and global extinctions, in contexts in which there is little or no direct evidence. In 1954, Andrewartha and Birch proposed a "General theory of the number of animals in natural populations", which was a challenge to the then dominant paradigm of population regulation by "density-dependent factors". Andrewartha and Birch contrasted the empirical evidence that most natural populations demonstrate large and erratic fluctuations in size, with the density-dependent paradigm's theoretical expectation of "balance" or "steady-density". Andrewartha and Birch's theory emphasized the importance of stochastic factors (catastrophes) and environmental heterogeneity as determinants of population size. Stochastic fluctuations in population size must logically include the possibility of population extinction. Andrewartha and Birch proposed a patch-occupancy model of local population dynamics to account for this possibility. Andrewartha and Birch argued that most natural populations are composed of a series of local populations inhabiting "favourable localities" (patches), among which local extinctions are commonplace, while the aggregate population remains extant, because local extinction in a given locality is eventually followed by colonization by immigrants from other localities. While Andrewartha and Birch's patch-occupancy model encompasses all of what is now referred to as metapopulation dynamics theory, their model was only one facet of a broader theory. Andrewartha and 1 Birch (1954, p. 5) stressed that "it has become customary to separate distribution [between-patch dynamics] and abundance [within-patch dynamics]", however, "the separation should never be allowed to persist in the final synthesis, for distribution and abundance are but the obverse and reverse of the same problem". Unfortunately, Andrewartha and Birch's patch-occupancy model "remained little developed in the mainstream population dynamics literature", apparently because their work "became identified as the losing side in the controversy" about the role of density-dependent factors in local population dynamics (Hanski and Gilpin 1991, p. 4). Consequently, the term "metapopulation", as used to describe "a population of populations", derives from work by Richard Levins in the late 1960's, rather than the earlier work by Andrewartha and Birch (Hanski and Gilpin 1991). Levins' patch-occupancy model consists of just three components: 1) the proportion of occupied patches; 2) the extinction rate; and 3) the colonization rate. Levins' model is often described as a two-state model because populations are either present or absent. The Levins model can be treated as analogous to a single-species version of MacArthur and Wilson's (1967) model of the equilibrium theory of island biogeography, if the probability of local population extinction in the Levins model is made to be a function of local population size (Hanski and Gilpin 1991). When the Levins model is extended to include the effects of local population size on the probability of local population extinction, a positive feedback relationship is apparent between the proportion of occupied patches and metapopulation persistence (Hanski 1991). A greater proportion of occupied patches entails a greater exchange of immigrants between occupied patches, which is assumed to increase the population size in each occupied patch, and thereby reduce the probability of both local and global extinction. Brown and Kodric-Brown (1977, p. 445) described this as the "rescue effect", whereby the "demographic and genetic contributions of immigrants" are assumed to "rescue" recipient populations from extinction. Andrewartha and Birch's (1954, p. 644) patch-occupancy model also included the idea that the probability of local population extinction was a function of local population size, an idea which they cited as originating with Darwin. In addition, Andrewartha and Birch (1954, p. 13) emphasized the importance of differences between individuals, stressing that "every individual in a population of animals is part of the environment of other individuals". According to Andrewartha and Birch, persistence within each patch is not simply a function of population size, but also depends on the sex, age, and genotype of the individuals composing the population. 2 While ignored in some quarters, Andrewartha and Birch's (1954) emphasis on the importance of dispersal between patches, in conjunction with socially-mediated interactions among different types of individuals, has been influential among researchers interested in population cycles in small mammals (Chitty 1996, p. 119). In a recent review, Krebs (1996, p. 21) concluded that "spacing behaviour in voles and lemmings produces dispersal, which is necessary for population regulation [and] breeding females appear to be the key". Researchers working on population cycles in small mammals have generally not concerned themselves with measuring the pattern of patch occupancy, but rather the number and nature of individuals moving in and out of patches, and the effects on residents within the patch (Stenseth and Lidicker 1992). In contrast to expectations from Levins-type patch-occupancy models, small mammal ecologists have reported adverse effects of immigration on the demography of recipient populations associated with acts of infanticide perpetrated by adult immigrant females against resident juveniles (Sherman 1981, Lambin and Krebs 1993, Wolff et al. 1997). The "sociobiological hypothesis", which is one of several general explanations proposed to explain population cycles (Krebs 1996, Boonstra et al. 1998), suggests that population declines are actually induced by an influx of unrelated immigrants. Wolff (1997) proposed that defence against infanticide is a key factor in the evolution of mechanisms of population regulation among mammals in general. In a recent review, Noss (1999, p. 115) credited the development of "modern conservation biology" in the mid-1970's to the "explosion of interest" in the potential application of MacArthur and Wilson's (1967) theory of island biogeography, and associated ideas regarding metapopulations, to the design of nature reserves. Noss (1999, p. 116) argued that there are nine general principles of conservation biology that have become "sufficiently well established" in the intervening 25 years, to guide conservation policy in particular cases. Two of these nine general principles are based of the idea that the exchange of immigrants helps "rescue" populations from extinction. One principle is that blocks of habitat that are closer together are better than blocks that are far apart, and the other is that blocks of habitat connected by "movement corridors" are better than isolated blocks. Unfortunately, there has been little attention given in the literature on conservation biology to the results of research on population cycles. In part, this is because applied research on cyclic species has 3 more often focussed on eradication of perceived "pests" than on the conservation of these species (Hansson 1992). Efforts have been made to use small mammals in "experimental model systems" (EMSs; Ims et al. 1993, Wolff et al. 1997) to test the effects of habitat fragmentation and the use of corridors on demographic parameters. However, these have been criticized as being so artificial as to be of "little relevance to real conservation problems and decisions" (Beier and Noss 1998, p. 1246). In Chapter 2,1 argue that the idea that immigration necessarily "rescues" populations from extinction can be challenged on the grounds that there may be adverse effects on the dynamics of recipient populations resulting from an influx of immigrants. The direct evidence in support of this contention largely comes from the research on population cycles in small mammals described above. It may be that small mammals behave differently from other species. However, small mammals and large mammals alike share a common characteristic with respect to dispersal, insofar as most dispersers are male. The fact that almost all Levins-type metapopulation models fail to distinguish between the sexes appears to be a consequence of the artificial separation of the ideas of distribution and abundance that Andrewartha and Birch (1954) cautioned against. Demographic models of abundance within a patch always account for sex and often also account for age-structure (Beissinger and Westphal 1998), and it only seems logical that this should also be the case when considering distribution between patches. In a recent review, Reed (1999) presented similar arguments with respect to birds to those I present regarding mammals. Andrewartha and Birch (1954, p. 5) noted that part of the reason for the artificial separation of distribution and abundance involves methodological problems. In Chapter 3,1 argue that this appears to be just as true today. Researchers interested in Levins-type metapopulation dynamics tend to conduct presence/absence surveys of a series of "favourable localities". Such studies rarely actually measure whether dispersal is even occurring, but instead, draw inferences regarding dispersal from the pattern of patch-occupancy. By contrast, researchers working on small mammals have considered the problems associated with the direct measurement of dispersal in great detail (Lidicker and Stenseth 1992), and have developed an array of techniques for doing so (Stenseth and Lidicker 1992). Andrewartha and Birch (1954, p. 89) recognized that it is of foremost importance to be able to distinguish between "movement associated with the ordinary activities of searching for food, a mate, and other requirements from movement which leads to finding a new place to live". One aspect concerning the direct measurement of 4 dispersal, which has not received much attention, is the degree to which rigorous measurement may itself produce inaccurate results. This potential source of error can be evaluated by comparison between direct and indirect measurements of the relative importance of dispersal. In Chapter 4,1 report the results of a large-scale field experiment conducted on common brushtail possums (Trichosurus vulpecula) in old-growth Eucalypt forest in south-eastern Australia, which was designed to address the question: does immigration "rescue" populations from extinction? I chose to study the common brushtail possum as a "model" medium-sized (1-3 kg), herbivorous, marsupial, in a "real landscape". The decline of medium-sized mammals in Australia clearly represents a "real conservation problem". Short and Smith (1994) reported that 47 % of all mammalian species extinctions in the past two centuries have occurred in Australia. Of the 245 species of mammals present in Australia at the start of European settlement in the 1780's, 16 are now extinct and a further 26 now occupy less than 20 % of then-former ranges (Short and Smith 1994). The majority of these extinctions and declines have involved medium-sized (35-5500 g), herbivorous, marsupials (Burbidge and McKenzie 1989). The common brushtail possum had the largest geographic range of any Australian marsupial at the time of European settlement, but now occupies only 37 % of its historical range (How and Kerle 1995). Numerous causes have been suggested to explain the declines and extinctions among medium-sized mammals in Australia. These include: the introduction of exotic herbivores (sheep, cattle and rabbits) and predators (foxes and cats); clearing of land for agriculture; and changed fire regimes (Burbidge and McKenzie 1989, Recher and Lim 1990, Short and Smith 1994). However, no single cause is sufficient to explain the decline of all of the affected species, and in many cases no single cause is sufficient to explain the decline of a given species across the affected range (Morton 1990, Recher and Lim 1990). Morton (1990) proposed that the natural distribution of native mammalian herbivores was particularly patchy, as these species were restricted to areas of high plant productivity on fertile soils, which in Australia are relatively scarce. According to Morton (1990), these same areas of high plant productivity acted as focal points of disturbance. Graziers naturally stocked their sheep and cattle at highest densities on these fertile patches, the invading rabbits would also congregate in these patches, and the introduced predators would then be attracted to these patches because of the rabbits. Morton (1990) argued that it was the medium-sized mammals that suffered more extinctions and declines because: 1) they were more 5 likely to be locally extirpated than smaller mammals because of their lower rates of natural increase; and 2) they were less likely than larger mammals to be "rescued" by immigration because of their poorer dispersal abilities. As part of the process of European settlement, state governments in many parts of Australia implemented a policy of retaining narrow strips of Crown land between adjacent blocks of freehold land. These narrow strips were retained as "travelling stock routes" (TSRs), to permit farmers to transport their products to market without having to cross privately held land. In most cases TSRs were not cleared and retained some natural features, while the surrounding freehold land was extensively cleared to create paddocks or cropland. TSRs appear to be ideal movement corridors for wildlife, although they were not designed for that purpose. A large number of studies have been conducted on these apparent movement corridors, primarily involving surveys for the presence or absence of native and introduced mammals. Consequently, Australian mammals feature prominently in the literature on movement corridors. In a recent review, Beier and Noss (1998) found 16 studies on mammals suggestive of a positive role for conservation corridors, of which, 6 involved Australian marsupial possums and gliders. Similarly, Laurance and Gascon (1997) advocated the retention of corridors in association with logging operations, based in part on Laurance's (1990) survey work in Australia on four species of possums, including the common brushtail possum. Comparatively little is actually known about the role of dispersal in the population dynamics of medium-sized Australian mammals (Lee and Cockburn 1985, Johnson 1989). For the most part, what is known about dispersal in these species derives from research on common brushtail possums conducted in New Zealand. At the same time as common brushtail possums have disappeared from large parts of their former range in Australia they have become a significant pest species in New Zealand, where they were first introduced in the 1830's (Cowan 1990). While there has been considerable interest in New Zealand concerning dispersal by common brushtail possums, the focus has been on the role of possums as vectors in the transmission of bovine tuberculosis to domestic livestock (Efford 1991). The importance of immigration in "rescuing" populations from extinction has been of little interest to New Zealanders as their goal since the late 1940's has been the eradication of the species (Cowan 1990). 6 While Andrewartha and Birch (1954, p. 88) emphasized the importance of dispersal in population dynamics, they also stressed the importance of differences between species in their "innate tendency toward dispersal". MacArthur and Wilson (1967, p.83) also emphasized the importance of life-history differences between "good" and "bad" colonizers, in the context of island biogeography, arguing that it is important to have a large intrinsic rate of increase to be a "good" colonizing species. Sinclair et al. (1998, their Table 1) reported per annum intrinsic rates of increase for two species of medium-sized Australian macropods (Bettongia lesueur and Lagorchestes hirsutus), and the common brushtail possum, as being, 0.932, 0.930 and 0.928, respectively. By way of comparison, Sinclair (1996, his Appendix 2) reported an intrinsic rate of increase of 2.190 for the snowshoe hare (Lepus americanus). Lee and Ward (1989) and Seebeck et al. (1989) both noted the relatively low fecundity of medium-sized, marsupial, herbivores in Australia, as compared to similar-sized species of mammals in other groups. In Chapter 5,1 describe the reproductive biology of common brushtail possums at my study site and make comparisons with the reproductive biology of the species reported at other sites. Using the results of the experiment described in Chapter 4 in conjunction with the comparisons made in Chapter 5,1 consider whether issues associated with connectivity and metapopulation dynamics, or factors such as "harvesting" by introduced predators, are more important to the demography, and likelihood of local population extinction, of common brushtail possums, and by extension, other medium-sized, marsupial, herbivores. In Chapter 6,1 present a deterministic, stage-based model of possum demography. Using the model, together with supplementary information regarding the proximate causes of adult deaths, I consider whether the rigorous methods used to accurately measure immigration, as described in Chapter 4, may confound the interpretation of the role of immigration, as suggested in Chapter 3. I argue that this is a particularly important issue to address before drawing conclusions about "dispersal sinks" (Pulliam 1988), as negative population growth rates may often be the consequence of adverse effects of repeated capture and handling. In Chapter 7,1 discuss the general implications of my results as regards the decline of medium-sized mammals in Australia, and strategies to conserve those species and populations that remain. 7 REFERENCES Andrewartha, H. G., and Birch, L. C. 1954. The distribution and abundance of animals. University of Chicago Press, Chicago. Beier. P., and Noss, R. F. 1998. Do habitat corridors provide connectivity? Conservation Biology, 12: 1241-1252. Beissinger, S. R., and Westphal, M. I. 1998. On the use of demographic models of population viability in endangered species management. Journal of Wildlife Management, 62: 821-841. Boonstra, R., Krebs, C. J., and Stenseth, N. C. 1998. Population cycles in small mammals: the problem of explaining the low phase. Ecology, 79: 1479-1488. Brown, J. H., and Kodric-Brown, A. 1977. Turnover rates in insular biogeography: effects of immigration on extinction. Ecology, 58: 445-449. Burbidge, A. A., and McKenzie, N. L. 1989. Patterns in the modern decline of Western Australia's vertebrate fauna: causes and conservation implications. Biological Conservation, 50: 143-198. Chitty, D. 1996. Do, lemmings commit suicide? Beautiful hypotheses and ugly facts. Oxford University Press, Oxford. Cowan, P. E. 1990. Brushtail possum. In The handbook of New Zealand mammals. Edited by King, C. M. Oxford University Press, Auckland, New Zealand, pp. 68-98. Efford, M. 1991. A review of possum dispersal. DSIR Land Resources Contract Report 91/73, DSIR Land Resources, Dunedin, New Zealand Hanski, I., and Gilpin, M. 1991. Metapopulation dynamics: brief history and conceptual domain. Biological Journal of the Linnean Society, 42: 3-16. Hanski, I. 1991. Single-species metapopulation dynamics: concepts, models and observations. Biological Journal of the Linnean Society, 42: 17-38. Hansson, L. 1992. Small mammal dispersal in pest management and conservation. In Animal dispersal: small mammals as a model. Edited by Stenseth, N. C , and Lidicker, W. Z., Jr., Chapman and Hall, London, pp. 181-198. 8 How, R. A., and Kerle, J. A. 1995. Common brushtail possum. In The mammals of Australia. Edited by Strahan, R., Reed Books, Sydney, Australia, pp. 273-275. Ims, R. A., Rolstad, J., and Wegge, P. 1993. Predicting space use responses to habitat fragmentation: can voles Microtus oeconomus serve as an experimental model system (EMS) for capercaillie grouse Tetrao urogallus in boreal forest? Biological Conservation, 63: 261-268. Johnson, C. N. 1989. Dispersal and philopatry in the macropodoids. In Kangaroos, wallabies and rat-kangaroos. Edited by Grigg, G., Jarman, P., and Hume, I. Surrey Beatty and Sons, Sydney, Australia, pp. 593-601. Krebs, C. J. 1996. Population cycles revisited. Journal of Mammalogy, 77: 8-24. Lambin, X., and Krebs, C. J. 1993. Influence of female relatedness on the demography of Townsend's vole populations in spring. Journal of Animal Ecology, 62: 536-550. Laurance, W. F. 1990. Comparative responses of five arboreal marsupials to tropical forest fragmentation. Journal of Mammalogy, 71: 641-653. Laurance, W. F., and Gascon, C. 1997. How to creatively fragment a landscape. Conservation Biology, 11:577-579. Lee, A. K., and Cockburn, A. 1985. Evolutionary ecology of marsupials. Cambridge University Press, Cambridge. Lee, A. K., and Ward, S. J. 1989. Life histories of macropodoid marsupials. In Kangaroos, wallabies and rat-kangaroos. Edited by Grigg, G., Jarman, P., and Hume, I. Surrey Beatty and Sons, Sydney, Australia, pp. 105-115. Lidicker, W. Z., Jr., and Stenseth, N. C. 1992. To disperse or not to disperse: who does it and why? In Animal dispersal: small mammals as a model. Edited by Stenseth, N. C , and Lidicker, W. Z., Jr., Chapman and Hall, London, pp. 21-36. MacArthur, R. H., and Wilson, E. O. 1967. The theory of island biogeography. Princeton University Press, Princeton, New Jersey. Morton, S. R. 1990. The impact of European settlement on the vertebrate animals or arid Australia. Proceedings of the Ecological Society of Australia, 16: 201-213. Noss, R. 1999. Is there a special conservation biology? Ecography, 22: 113-122. 9 Pulliam, H. R. 1988. Sources, sinks, and population regulation. American Naturalist, 132: 652-661. Recher, H. F., and Lim, L. 1990. A review of current ideas on the extinction, conservation and management of Australia's terrestrial vertebrate fauna. Proceedings of the Ecological Society of Australia, 16: 287-301. Reed, J. M. 1999. The role of behaviour in recent avian extinctions and endangerments. Conservation Biology, 13: 232-241. Seebeck, J. H., Bennett, A. F., and Scotts, D. J. 1989. Ecology of the Potoroidae: a review. In Kangaroos, wallabies and rat-kangaroos. Edited by Grigg, G., Jarman, P., and Hume, I. Surrey Beatty and Sons, Sydney, Australia, pp. 67-88. Sherman, P. W. 1981. Reproductive competition and infanticide in Belding's ground squirrels and other animals. In Natural selection and social behaviour. Editedby Alexander, R. D., and Tinkle, D. W., Chiron Press, New York, pp. 311-331. Short, J., and Smith, A. 1994. Mammal decline and recovery in Australia. Journal of Mammalogy, 75: 288-297. Sinclair, A. R. E. 1996. Mammal populations: fluctuation, regulation, life history theory and their implications for conservation. In Frontiers of population ecology. Editedby Floyd, R. B., Sheppard, A. W., and DeBarro, P. J., CSIRO Publishing, Melbourne, Australia, pp. 127-154. Sinclair, A. R. E., Pech, R. P., Dickman, C. R., Hik, D, Mahon, P., and Newsome, A. E. 1998. Predicting effects of predation on conservation of endangered prey. Conservation Biology 12: 564-575. Stenseth, N. C , and Lidicker, W. Z., Jr. 1992. Where do we stand methodologically about experimental design and methods of analysis in the study of dispersal? In Animal dispersal: small mammals as a model. Edited by Stenseth, N. C , and Lidicker, W. Z., Jr., Chapman and Hall, London, pp. 295-318. Wolff, J. O. 1997. Population regulation in mammals: an evolutionary perspective. Journal of Animal Ecology, 66: 1-13. Wolff, J. O., Schauber, E. M., and Edge, W. D. 1997. Effects of habitat loss and fragmentation on the behaviour and demography of gray-tailed voles. Conservation Biology, 11: 945-956. 10 CHAPTER 2 DOES IMMIGRATION "RESCUE" POPULATIONS FROM EXTINCTION? IMPLICATIONS REGARDING MOVEMENT CORRIDORS AND THE CONSERVATION OF MAMMALS The "rescue effect" is one of the principal rationales provided for "movement corridors for conservation" (Simberloff et al. 1992). The conventional conceptualization of the rescue effect may be paraphrased as follows: 'the closer two "island" populations are to one another, the more likely they are to exchange immigrants, which will reduce the probability of extinction of either population, thanks to the "demographic and genetic contributions of immigrants" (Brown and Kodric-Brown 1977, p. 445) to recipient populations'. Insofar as corridors facilitate the exchange of immigrants, they will aid conservation by reducing the probability of local population extinctions. The conventional conceptualization of the rescue effect masks two distinct components: 1) an empirical phenomenon involving a positive correlation between physical proximity and the persistence of populations in sequential surveys of "island" populations; and 2) the mechanism responsible for this positive correlation, which is hypothesized to be immigration. Positive correlations between proximity and population persistence are well-documented in mammals (e.g., Smith 1980, van Apeldoorn et al. 1992), amphibians (e.g., Sjogren 1991), and vascular plants (e.g., Ouborg 1993), and I do not question the existence of this correlation. However, whether immigration is the mechanism underlying this correlation can be challenged on three grounds: 1) the absence of direct empirical evidence; 2) the failure to consider alternative hypotheses: and 3) the abundance of conflicting behavioural data. Simberloff et al. (1992) commented on the absence of direct empirical evidence of a role for immigration, and this verdict still holds. Immigration is rarely measured directly (Simberloff et al. 1992), and almost never in a sufficiently rigorous fashion (Stenseth and Lidicker 1992). Despite the absence of direct empirical evidence, the idea that immigration is the mechanism responsible for the rescue effect is never challenged. The idea is accepted as dogma, as it is so "intuitively obvious" (Caughley and Gunn 1996). Unfortunately, intuitions may often be wrong (Caughley and Sinclair 1994). 11 In this Chapter, I will illustrate that alternative hypotheses can be readily formulated as regards the mechanism underlying the positive correlation between proximity and persistence. Moreover, I will argue that, for mammals at least, patterns regarding individual dispersal behaviour are well-established, and these patterns suggest that immigration may often decrease the probability of persistence of recipient populations. Alternative hypotheses The nearer any two populations are, the more likely they are to be subject to similar deterministic environmental effects, and hence, the more likely they are to fluctuate in synchrony (Harrison and Quinn 1989). It is not necessary to posit a role for immigration in inducing such synchrony. However, a correlation between proximity and synchrony in population fluctuations is as likely to lead to a negative correlation between proximity and persistence as it is to lead to a positive correlation. Yet the empirical phenomenon associated with the rescue effect involves a consistently positive correlation between proximity and persistence. I will demonstrate how a consistently positive relationship between proximity and persistence can result from: 1) environmentally induced synchrony; coupled with 2) the way. in which we ask the question. The correlation between proximity and persistence is conventionally described as follows (Fig. 2.1). Several "island" populations are surveyed for the presence/absence of a species at time 1, and then again at time 2. Persistence is defined as presence of the species in both surveys. Extinction is more likely to occur in "outlying" populations. All populations are assumed to demonstrate independent population dynamics which are modified solely by the exchange of immigrants. As an alternative, we can envisage that populations occur in clusters, wherein there is complete independence of population dynamics between clusters, but perfect synchrony within clusters, as a result of exposure to similar deterministic environmental effects. The eight independent populations illustrated in Figure 2.1 are now arranged into four independent clusters containing two synchronous populations in each cluster (Fig. 2.2). To establish whether there is a relationship between proximity and persistence we first ask: "do both populations persist from time 1 to time 2?" We can arrange the results of this query as shown in Table 2.1. Using the evidence from Table 2.1 and Figure 2.2 we can establish a simple contingency 12 table (Table 2.2), categorizing populations as to whether they persist from time 1 to time 2, and whether they are near of far from one another. Populations within clusters are categorized as near to one another, whereas populations in all other clusters are categorized as far. Sutcliffe et al. (1996) used contingency tables to analyze the relationship between proximity and population synchrony in various species of butterflies in England. Using a G-test, there is no evidence of a relationship (G = 0.904, p > 0.05). However, had we assumed there were five populations within each of four clusters, the result would reach statistical significance (G = 4.706, p < 0.05), demonstrating a significant positive relationship between proximity and persistence. Naturally, a greater number of populations within each cluster would lead to ever more statistically significant results. What the above example demonstrates is that it is not necessary to posit a role for immigration in order to explain the empirically observed positive relationship between proximity and persistence, associated with the rescue effect. While immigration may be the mechanism responsible for the rescue effect, we cannot simply assume it is, just because it is "intuitively obvious" (Caughley and Gunn 1996). Behavioural data Part of the reason why it is conventionally believed that immigration must be the mechanism underlying the rescue effect, is because in almost all current metapopulation models, immigration is invariably assumed to contribute positively to the persistence of the recipient population. There are two fundamental aspects to most models concerning the rescue effect (e.g., Hanski and Gyllenberg 1993): 1) population size, ./V, is assumed to be positively correlated with persistence; and 2) only "additive" models of demography are ever considered. I do not wish to challenge the idea that population size is positively correlated with persistence, but it must be remembered that such demographic models apply to females only. Research scientists and managers alike often seem to forget this restriction (Caughley and Gunn 1996). To emphasize this point, we can bring the scale down to the level of individual behaviour, and recognize that the persistence of a population depends on the rate of "breeding replacement". Breeding replacement is here defined as the rate, in real time, at which individual, resident, reproductively active females that die or emigrate, are 13 replaced by new reproductively active females, as a result of either immigration or local recruitment (after Watson and Moss 1970). "Additive" models of demography are those familiar from any introductory textbook in ecology: Mime = (B + I)-{D + E) Most metapopulation models assume the only alternative, involving immigration, is what might be described as an "isolation" model of demography: Nlimt = (B)-(D + E) It follows that, all else being equal, the positive rate of increase in the population will necessarily be greater with immigration, and hence, persistence will necessarily be more probable with immigration. However, it is possible to conceive of at least a second alternative model of demography, one which allows for interactions between the fundamental demographic parameters: Ntimc = (B + I)-(D + E)-(IB + ID + IE) In this case, persistence is more probable with immigration, only if the positive contribution of immigration, /, is greater than the potential negative interactions between immigration and the other fundamental demographic parameters: IB, ID, and IE. Currently, the only example in the literature involving the recognition of interactive effects comes from work by Hess (1994), who considered the detrimental effects on persistence as a result of disease transmission by immigrants. Whereas Hess (1994) suggested disease transmission as a "special case" whereby immigration decreases the probability of persistence, I argue that there is an abundance of behavioural data, at least as regards mammals, to suggest that immigration is normally irrelevant to, and may often decrease, the probability of persistence of recipient populations. The abundance of behavioural data indicating detrimental effects of immigration on recipient populations can be understood by recognizing the conflict between immigrants and residents in terms of intraspecific competition for physical and genetic resources. Intraspecific competition is a fundamental and inescapable aspect of population dynamics, and cannot be dismissed as a "special case". Competition for physical resources, such as food and shelter, has familiar consequences, most often modeled in the form of the logistic growth curve. Competition for genetic resources has markedly different consequences, and is described in detail below. 14 Several reviews of the literature (Greenwood 1980, Waser and Jones 1983, Johnson and Gaines 1990) have all concluded that in most mammalian species: 1) most dispersers are male; and 2) most females are philopatric, establishing home ranges near their mothers. As a result: 1) most immigrants are male; and 2) most populations consist of neighbourhoods of related females, with immigrant females entering a family's territory/range, rather than a lone individual's range. Returning to the idea that persistence depends on the rate of breeding replacement, it becomes obvious that, as most immigrants are male, most immigration is irrelevant to the persistence of the recipient population. Further, if males and females compete for the same physical resources, the addition of immigrant males to a population may increase competitive pressure on resident females, potentially lowering the rate of reproduction, thereby reducing the rate of breeding replacement via local recruitment, and decreasing the probability of persistence of the population. Male immigration may be modeled as: Ntime = (B)-(D + E)-(IB) Clearly, female immigration can contribute to breeding replacement, and hence increase the probability of persistence of the population. However, interactions between immigrants and residents must be acknowledged. What is important is the net rate of breeding replacement in the presence of immigrant females. As with males, immigrant females may compete with resident females for physical resources. In addition, immigrant females will compete with families of resident females for "genetic resources". The Darwinian fitness of an individual female does not depend on the absolute number of offspring she produces, but on the number relative to all other females in the population. Unlike competition for physical resources, competition for genetic resources is not necessarily a function of population size. Even if a population consists of only two unrelated females, they should both attempt to limit the other's contribution to the next generation. The effects of genetic competition have generally been overlooked in even the more sophisticated "structured" models of metapopulation dynamics (Hastings 1991), which assume logistic population growth (resulting from competition for physical resources) within each population composing the metapopulation. Genetic competition among resident females will be mild, given that, as a result of female philopatry, adjacent residents are often relatives. Levels of aggression are lower among kin in most species 15 (Waldman 1988). Intrusion by immigrant females, on the other hand, can be expected to be fiercely resisted by the family of resident females (Lambin and Krebs 1991). Infanticide can be seen as the extreme on the continuum of genetic competition, killing somebody else's offspring to create an "opening" for your own. Infanticide is well-documented in carnivores (Packer and Pusey 1984), primates (Quiatt and Reynolds 1993), rabbits (Kunkele 1992), and rodents (Cockburn 1988). Probably the best behavioural evidence of genetic competition between immigrant and resident females, leading to infanticide, comes from work by Sherman (1981) on Belding's ground squirrel. Sherman (1981) found: 1) infanticide was the largest single cause of resident juvenile mortality; 2) infanticide was never committed by resident females; 3) related resident females cooperated in defence against other individuals attempting to commit infanticide; 4) most infanticides were committed by immigrant adult females who did not eat their victims; and 5) 75 % of infanticidal adult females settled in territories adjacent to their victim's territory. Individual behaviours do not necessarily have significant demographic consequences. Lambin and Krebs (1993) documented both behavioural and demographic consequences of infanticide by immigrant female Townsend's voles. They manipulated both: 1) the proximity of relatives of resident females; and 2) levels of female immigration. Lambin and Krebs (1993) observed: 1) six direct instances of infanticide; 2) lower mortality of adult resident females if their nearest neighbour was related, suggesting the operation of kin selection; and 3) higher resident juvenile mortality in the presence of immigrant females. Given Lambin and Krebs' (1993) results, female immigration ought to be modeled as: Mime = (B + D-(D + E)-{ID) I began this section by emphasizing that in almost all current metapopulation models, immigration is invariably assumed to contribute positively to the persistence of the recipient population. By contrast, the abundance of behavioural data available, at least as regards mammals, suggests: 1) the bulk of immigration, as it involves males, is at best, irrelevant to the persistence of recipient populations; and 2) competition between immigrant females and families of resident females may often decrease the probability of persistence of the recipient population. 16 Conclusions I have argued that the general acceptance of the idea that immigration is the mechanism responsible for the rescue effect can be seen to be a function of: 1) the failure to consider alternative hypotheses; and 2) the pre-eminence of certain demographic models. The problem underlying both these criticisms is the absence of direct empirical evidence. Simple presence/absence surveys are not enough. Only rigorous identification of immigrants, experiments directly manipulating immigration, and monitoring of effects on breeding replacement and/or persistence, are sufficient to establish that immigration is in fact the mechanism responsible for the rescue effect. I have deliberately avoided any discussion of the "genetic contributions of immigrants" (Brown and Kodric-Brown 1977, p. 445) to recipient populations. Based on mathematical models and results from captive breeding, many authors (e.g., Soule 1987, Hedrick et al. 1996) have argued that a regular influx of new genetic material via immigration is absolutely crucial to the "health" of recipient populations. However, others (most notably Caughley 1994, and Caughley and Gunn 1996) have emphasized the absence of adequate evidence of inbreeding depression and like signs of genetic "ill-health" in any wildlife population. My arguments regarding the potentially negative "demographic contributions of immigrants" may or may not be ameliorated, depending upon your view of the importance of the "genetic contributions of immigrants". Simberloff and Cox (1987) questioned the values of movement corridors for conservation on the basis that there was no evidence that animals actually used corridors to move from one "island" population to another. Saunders and Hobbs (1991) presented limited evidence that animals do actually disperse through corridors. Simberloff et al. (1992) and Hess (1994) discussed negative "side effects" for conservation arising from the use of corridors. None of the authors have addressed the core issue, which concerns the nature of immigration. I have argued that, contrary to potential negative effects of immigration being a "special case", conflict between immigrants and residents is an inescapable consequence of intraspecific competition for physical and genetic resources. Not only may movement corridors be a "poor investment" (Simberloff et al. 1992), they may be something we want to avoid. 17 REFERENCES Brown, J. H., and Kodric-Brown, A. 1977. Turnover rates in insular biogeography: effects of immigration on extinction. Ecology, 58: 445-449. Caughley, G. 1994. Directions in conservation biology. Journal of Animal Ecology, 63: 215-244. Caughley, G., and Gunn, A. 1996. Conservation biology in theory and practice. Blackwell Science, Cambridge, Massachusetts. Caughley, G., and Sinclair, A. R. E. 1994. Wildlife ecology and management. Blackwell Science, Cambridge, Massachusetts. Cockburn, A. 1988. Social behaviour in fluctuating populations. Croom Helm, London. Greenwood, P. J. 1980. Mating systems, philopatry and dispersal in birds and mammals. Animal Behaviour, 28: 1140-1162. Hanski, I., and Gyllenberg, M. 1993. Two general metapopulation models and the core-satellite species hypothesis. American Naturalist, 142: 17-41. Harrison, S., and Quinn, J. F. 1989. Correlated environments and the persistence of metapopulations. Oikos, 56: 293-298. Hastings, A. 1991. Structured models of metapopulation dynamics. Biological Journal of the Linnean Society, 42: 57-71. Hedrick, P. W., Lacy, R. C , Allendorf, F. W., and Soule, M. E. 1996. Directions in conservation biology: comments on Caughley. Conservation Biology, 10: 1312-1320. Hess, G. R. 1994. Conservation corridors and contagious disease: a cautionary note. Conservation Biology, 8: 256-262. Johnson, M. L., and Gaines, M. S. 1990. Evolution of dispersal: theoretical models and empirical tests using birds and mammals. Annual Review of Ecology and Systematics, 21: 449-480. Kunkele, J. 1992. Infanticide in wild rabbits {Oryctolagus cuniculus). Journal of Mammalogy, 73: 317-320. Lambin, X., and Krebs, C. J. 1991. Can changes in female relatedness influence microtine population dynamics? Oikos, 61: 126-132. 18 Lambin, X., and Krebs, C. J. 1993. Influence of female relatedness on the demography of Townsend's vole populations in spring. Journal of Animal Ecology, 62: 536-550. Ouborg, N. J. 1993. Isolation, population size and extinction: the classical and metapopulation approaches applied to vascular plants along the Dutch Rhine-system. Oikos, 66: 298-308. Packer, C , and Pusey, A. E. 1984. Infanticide in carnivores. In Infanticide: comparative and evolutionary perspectives. Edited by Hausfater, G., and Hrdy, S. B., Aldine, New York, pp. 31-42. Quiatt, D. D., and Reynolds, V. 1993. Primate behaviour: information, knowledge, and the evolution of culture. Cambridge University Press, Cambridge. Saunders, D. A., and Hobbs, R. J. 1991. Nature conservation: the role of corridors. Surrey Beatty and Sons, Sydney, Australia. Sherman, P. W. 1981. Reproductive competition and infanticide in Belding's ground squirrels and other animals. In Natural selection and social behaviour. Edited by Alexander, R. D., and Tinkle, D. W., Chiron Press, New York, pp. 311-331. Simberloff, D., and Cox, J. 1987. Consequences and costs of conservation corridors. Conservation Biology, 1: 63-71. Simberloff, D., Farr, J. A., Cox, J., and Mehlman, D. W. 1992. Movement corridors: conservation bargains or poor investments? Conservation Biology, 6: 493-504. Sjogren, J. 1991. Extinction and isolation gradients in metapopulations: the case of the pool frog (Rana lessonae). Biological Journal of the Linnean Society, 42: 135-147. Smith, A. T. 1980. Temporal changes in insular populations of the pika (Ochotona princeps). Ecology, 61: 8-13. Soule, M. E. 1987. Viable populations for conservation. Cambridge University Press, Cambridge. Stenseth, N. C , and Lidicker, W. Z., Jr. 1992. Where do we stand methodologically about experimental design and methods of analysis in the study of dispersal? In Animal dispersal: small mammals as a model. Edited by Stenseth, N. C , and Lidicker, W. Z., Jr., Chapman and Hall, London, pp. 295-318. Sutcliffe, O. L., Thomas, C. D., and Moss, D. 1996. Spatial synchrony and asynchrony in butterfly population dynamics. Journal of Animal Ecology, 65: 85-95. 19 van Apeldoorn, R. C , Oostenbrink, W. T., van Winden, A., and van der Zee, F. F. 1992. Effects of habitat fragmentation on the bank vole, Clethrionomys glareolus, in an agricultural landscape. Oikos, 65: 265-274. Waldman, B. 1988. The ecology of kin recognition. Annual Review of Ecology and Systematics, 19: 543-571. Waser, P. ML, and Jones, T. W. 1983. Natal philopatry among solitary animals. Quarterly Review of Biology, 58: 355-390. Watson, A., and Moss, R. 1970. Dominance, spacing behaviour and aggression in relation to population limitation in vertebrates. In Animal populations in relation to their food supply. Edited by Watson, A., Blackwell Scientific, Oxford, pp. 167-219. 20 Table 2.1. Summary of results of examining populations illustrated in Figure 2.2 and asking the question "do both populations persist from time 1 to time 2?" 2 3 Populations 4 5 6 7 8 1 Yes Yes Yes Yes Yes No No 2 Yes Yes Yes Yes No No 3 Yes Yes Yes No No 4 Yes Yes No No 5 Yes No No 6 No No 7 No 21 Table 2.2. Contingency table for analysis of relationship between proximity and persistence using evidence from Table 2.1 and Figure 2.2. Distance Do both populations persist from time 1 to time 2? Yes No Near Far 3 12 12 22 Figure 2.1. Conventional portrayal of relationship between proximity and persistence. Numerals signify individual "island" populations. The letter E signifies extinction of a population. 23 Time 1 Time 2 Figure 2.2. Environmentally induced synchrony of population dynamics signified by clustering of populations. Numerals signify individual "island" populations. The letter E signifies extinction of a population 24 CHAPTER 3 METHODOLOGICAL PROBLEMS IN THE STUDY OF IMMIGRATION AND EXTINCTION The role of immigration in the dynamics of recipient populations has been the focus of attention of two groups of researchers, those interested in metapopulation dynamics, and those interested in population cycles. Within the metapopulation framework, immigration is viewed as a purely additive process (Clinchy 1997), whereby the "demographic and genetic contributions of immigrants" (Brown and Kodric-Brown 1977, p. 445) may help "rescue" recipient populations from extinction. By contrast, the potentially adverse effects of an influx of immigrants into a population has been suggested as the cause of the decline phase in cyclic populations of both birds (e.g., Watson et al. 1994) and mammals (e.g., Charnov and Finerty 1980, Hestbeck 1982, Lambin and Krebs 1991, Wolff et al. 1997). For various reasons, these two groups of researchers have adopted different methodological approaches to the study of immigration. Studies of metapopulation dynamics have tended to adopt what might be termed the "correlational-survey" approach, which draws inferences regarding immigration.from patterns of patch occupancy, while studies on population cycles have generally involved direct attempts at measuring and manipulating immigration rates. As I will discuss, there are various difficulties associated with both these approaches to the study of immigration, and there has been little attempt to integrate the two approaches. I will focus primarily on mammals, both because mammals are more likely than other vertebrate taxa to demonstrate metapopulation dynamics (Lidicker and Koenig 1996), and because studies on small mammals dominate the literature on the potential role of dispersal in producing cyclic population dynamics (Stenseth and Lidicker 1992). The correlational-survey approach The correlational-survey approach requires the measurement of three variables: 1) the distance between a number of habitat patches which are all deemed to be "suitable" for the target species; 2) the presence or absence of the target species in each suitable patch; and 3) the density of the target species in each occupied patch. Immigration is judged to be important in "rescuing" populations from extinction, if: 1) among all suitable patches, isolated patches are more likely to be unoccupied; 25 and 2) among occupied patches, isolated patches have lower density populations. The argument that immigration must be the mechanism responsible, is as follows: the nearer two occupied patches are to one another the greater will be the rate of exchange of immigrants between them, which "props up" population size in both patches, and thereby reduces the probability of stochastic local extinction (Hanski and Gilpin 1997, p. 213). Conversely, more isolated patches receive fewer immigrants, and as a result have lower densities (pattern number 2 from above), and are therefore more likely to be unoccupied (pattern number 1 from above), because the resident populations have gone extinct. While a large number of studies (e.g., Gottfried 1979, Fahrig and Merrian 1985, van Apeldoorn et al. 1992) have used the correlational-survey approach, the Bodie, California, "pika metapopulation is the best-known mammalian example" (Moilanen et al. 1998, p. 530), and it has been argued that the studies that have been done at this site (Smith 1974, Smith 1980, Smith and Gilpin 1997) "provide some of the best evidence available that [immigration] within mammalian metapopulations can ... rescue local populations from extinction" (Stacey et al. 1997, p. 283). One of the strengths of the Bodie study is that, rather than the usual single survey of presence or absence and population density, a series of surveys have been conducted on the same habitat patches, over a number of years. In each of these several surveys it has been.found that: 1) among all suitable patches, isolated patches are more likely to be unoccupied; and 2) among occupied patches, isolated patches generally have lower density populations. The existence of a series of surveys at the Bodie site permits a further, more direct test of whether immigration likely "rescues" populations from extinction. If more isolated patches are more likely to go extinct because they receive fewer immigrants, then the spatial configuration of occupied patches in one survey ought to allow us to predict which populations (i.e., those more distant from their neighbours) will go extinct, and hence, which patches will be found to be unoccupied, in the next survey. I consider this a more direct test of the "rescue effect" because extinction is a dynamic process that occurs over time, in which case measurements made over time ought to be preferable to single survey "snapshots" of a given point in time. The spatial configuration, and history of patch occupancy, in each of the four surveys conducted at the Bodie site was reported in Smith and Gilpin (1997, their Fig. 2). When the distance to the nearest occupied patch (the measure of isolation used in Smith 1974, Smith 1980, and Smith and Gilpin 1997) in one survey was compared between populations that were found to be either extinct or extant in the 26 next survey, there was no significant effect of nearest neighbour distance on the probability of extinction (Table 3.1). Consequently, while the pattern of patch occupancy within each survey of the Bodie site is consistent with that predicted by the "rescue effect" (Smith and Gilpin 1997, their Table 1), the pattern of changes in patch occupancy between surveys, is not. Simulation models incorporating a large number of variables, including population densities within each patch and estimates of "typical" dispersal distances, have been more (Moilanen et al. 1998) or less (Smith and Gilpin 1997) successful in demonstrating that the observed pattern of patch occupancy at the Bodie site can be explained by metapopulation dynamics "in the presence of some regional stochasticity" (Moilanen et al. 1998, p. 539). But, "this is not to say that the observed pattern could not have other explanations" (Moilanen et al. 1998, p. 539). As illustrated with respect to the Bodie study, difficulties can arise when interpreting the results of the correlational-survey approach because immigration is not actually measured, rather, its effects are simply inferred from the pattern of patch occupancy. Consequently, any number of alternative hypotheses (Clinchy 1997), such as localized effects of predation, or epidemic disease, could also explain the observed patterns of patch occupancy (Smith and Gilpin 1997). Identifying immigrants in open populations in the field Several very elegant experiments have been conducted in outdoor enclosures, which have measured and manipulated the rate of exchange of immigrants between habitat patches (e.g., Ims et al. 1993, Wolff et al. 1997, and Bjornstad et al. 1998). Enclosure experiments provide many advantages, the foremost, with respect to the study of immigration, being that the identity and origin of every individual is known with complete certainty, because they were put there by the experimenter. Unfortunately, some critics from among the ranks of those interested in metapopulation dynamics and associated conservation issues, have dismissed the results of enclosure experiments as being of "little relevance", because the experiments are conducted "in settings so dissimilar to [real] landscapes" (Beier and Noss 1998, p. 1246). Yet, at the same time, the level of certainty obtainable in enclosure studies seems indispensable, since "simulation models [of metapopulation dynamics] have shown that a surprisingly low number of immigrants per year ... will allow individual populations to persist" indefinitely (Stacey et al. 1997, p. 268). 27 A number of steps need to be taken in order to measure immigration in open populations in "real landscapes", with anywhere near the degree of accuracy required to challenge the results from simulation studies. The central dilemma associated with measuring immigration in open populations in the field is that of how to identify who in fact is an immigrant. Immigration entails entering and settling in an occupied patch. The individuals, and their offspring, that already inhabit the patch, represent the resident population. Provided all the residents, and their offspring, have been completely enumerated, all newly-identified individuals must be immigrants. The problem here is that a proportion of the residents will almost certainly elude detection when only a single sampling technique (e.g., live-trapping) is employed (Krebs 1999). Consequently, any newly-identified individual may be an immigrant, or a resident that has previously eluded detection (Stenseth and Lidicker 1992). This problem can be substantially alleviated by employing at least two, fully independent sampling techniques (Fig. 3.1a). For example, Boonstra and Krebs (1978) used cage-trapping and pitfall-trapping to study voles and found a proportion of individuals susceptible to one, but not the other, trapping technique. Boutin (1980) used cage-trapping to monitor a population of snowshoe hares, coupled with repeated "hare-drives" involving the use of a line of volunteers acting as "beaters",, who drove the hares into nets. ,: To avoid misidentifying local recruits, as immigrants, it is also necessary to completely enumerate all of the offspring of residents (Fig. 3.1a). In species with mobile, altricial young, such as marsupials, it is possible to capture and mark the mother and young at the same time (Efford 1998). In precocial species, complete enumeration of offspring often requires that extraordinary steps be taken. O'Donoghue (1994) held pregnant snowshoe hares in portable outdoor enclosures until they gave birth, and then radio-tagged the leverets. Another approach taken by Pugh and Tamarin (1991) was to inject pregnant voles with radionuclides which would pass from mother to young across the placental barrier, the limitation being that the offspring must be trapped and marked after birth by some other means, before the radionuclides become undetectable. If the area of complete enumeration does not fully encompass the entire patch, newly-identified individuals may be neighbouring residents from just outside the area of complete enumeration that are either: 1) exploring beyond the boundaries of their normal home range (Lidicker and Stenseth 1992); or 2) are expanding their home range in response to vacancies created within the area of complete 28 enumeration (Stenseth and Lidicker 1992). Neighbouring residents from just outside the area of complete enumeration are likely to be either related to, or familiar with, residents within the area of complete enumeration, and an influx of such individuals is not expected to have adverse effects on the residents within the area of complete enumeration (Hestbeck 1982, Lambin and Krebs 1991). In terms of metapopulation dynamics, excursions (or exploratory movements) and range expansion, represent forms of "quasi-dispersal" (Lidicker and Stenseth 1992, p. 23), which are distinct from the type of "true" dispersal that is assumed when discussing the exchange of immigrants between patches (see Krebs 1992, his Fig. 7.1). According to Lidicker and Stenseth (1992, p.23), "true" dispersal involves three phases, "leaving, travelling, and arriving", and the distance traveled must be such that "the new home range [is] disjunct from the abandoned one". Whether one is interested in the role of immigrants as unrelated and unfamiliar "intruders" that induce population declines, or as "rescuers" that prevent population extinction, it is necessary to distinguish between potential immigrants and neighbouring residents from just outside the area of complete enumeration. To be able to distinguish between "true" immigrants and neighbouring residents from just outside the area of complete enumeration, the area of complete enumeration should include a "buffer strip" (or "border zone"; Stenseth and Lidicker 1992, p. 300), wherein newly-identified individuals detected in the "buffer strip" are not classified as immigrants. Once all of the residents, and their offspring, are completely enumerated within a given area (Fig. 3.1a), the area can be divided into a "core" area in which immigration will be measured, surrounded by a peripheral "buffer strip" (Fig. 3.1b). Following Lidicker and Stenseth's (1992, p. 23) definition of "true" dispersal, the "buffer strip" should be a minimum of one home range length across (Fig. 3.1b), in which case it can be said without ambiguity that newly-identified individuals that settle in the core of the area must have "traveled" across at least one home range, and their new home range is clearly "disjunct from the abandoned one". Subsequently (Fig. 3.1c): 1) if known adult residents that inhabit the "buffer strip" expand their ranges into the core of the area they can be readily distinguished from "true" immigrants (Lidicker and Stenseth 1992); similarly 2) if the offspring of residents that inhabit the "buffer strip" settle in the core area they too can be distinguished from "true" immigrants; but most importantly 3) newly-identified individuals entering and settling in the "buffer strip" are not classified as immigrants, as they are as likely to be neighbouring residents from just outside the area 29 of complete enumeration. By this means, newly-identified individuals that do enter and settle in the core area can be classified as "true" immigrants, without ambiguity (Fig. 3.1c). An integrated approach As discussed, the problem of how to identify who in fact is an immigrant, in open populations in the field, can be solved by the use of a variety of sampling techniques aimed at achieving the complete enumeration of residents, and by delineating a peripheral "buffer strip" within the area of complete enumeration. However, solving the problem of unambiguous identification may create problems when it comes to interpreting the role of immigration in the dynamics of recipient populations. To completely enumerate all of the residents in an area, and their offspring, while at the same time making home range measurements to facilitate the delineation of a peripheral "buffer strip", will in most cases require intensive handling and disturbance of the resident population. By contrast, immigrants have not been subjected to this intensive handling. Consequently, handling represents a treatment that residents receive but immigrants do not, at least not until they enter the study area. If the effect of handling is significant, then the measured rate of immigration may be incorrect. In general, animals will suffer the greatest stress in response to their initial capture, and are then expected to habituate to repeated capture and handling (Harlow et al. 1992), such that they may even become "trap-addicted" (Krebs 1999). Assuming residents have "more to lose" from abandoning their normal home range than dispersers do from not establishing within a given home range, we can predict that dispersers are more likely to be "driven out" of an area in response to the stress associated with their initial capture, than residents, who are more likely to remain and be recaptured. Consequently, the rate of immigration is expected to be underestimated under these circumstances. Alternatively, if the repeated capture and handling of residents has adverse effects on their survival or reproduction, the rate of immigration may be overestimated. For example, if local recruitment normally preempts settlement by immigrants (Fig. 3.2a), and if the intensive handling of residents induces a significant loss of local young (Fig. 3.2b), then home ranges that would have been occupied by the offspring of residents are instead occupied by immigrants (Fig. 3.2b). In the latter case, if the handling-induced loss of local young is such that the rate of local recruitment is reduced to less than that required to "compensate" for adult 30 losses (X < 1), then the recipient resident population may appear to be a dispersal "sink" (Pulliam 1988), that would otherwise go extinct if it were not continually "rescued" by immigration. Studies attempting to directly measure immigration in open populations in the field must include an evaluation of the potentially adverse effects of handling, before any interpretation is made regarding the role of immigration in the dynamics of the recipient population. Whether or not immigrants are often "driven out" of an area by the stress of their initial capture can be gauged by the likelihood of instantaneous trap-aversion. If animals are radio-collared at the time of their initial capture, or are marked in a way that allows individual identification if they are resighted, it will be possible to determine what percentage of residents are never recaptured, after initially being captured and handled. This approach can be augmented by using supplemental information regarding the characteristics of individuals that disappear after being captured on only one, or a few, occasions. Whether or not repeated capture and handling has an adverse effect on reproduction or survival can be evaluated by intentionally subjecting different groups to different handling regimes, or by determining whether reproduction or survival are correlated with the frequency of capture and handling. The fact that the accurate measurement of immigration may often confound the interpretation of the role of immigration argues for an integrated approach to the question. As discussed, interpretations of the role of immigration in producing the patterns observed using the correlational-survey approach are ambiguous because immigration is not actually measured. On the other hand, the steps necessary to measure immigration in open populations in "real landscapes" may actually lead to ambiguities when interpreting the role of immigration. Consequently, the best approach would appear to be to supplement direct measurements of immigration with a less direct, but broader, "survey" of patterns expected to arise under different scenarios regarding the role of immigration (Eberhardt and Thomas 1991). In this respect, new genetic techniques may prove quite useful in providing a less invasive means of estimating rates of immigration (Ims and Yoccoz 1997). For example, Peacock and Smith (1997) measured DNA band-sharing scores among pikas at the Bodie study site and found evidence from parentage analysis (juveniles being found in a different patch from the one inhabited by their putative parent) to suggest that there was an exchange of immigrants between patches. Unfortunately, parentage analysis is not infallible, as it is sensitive to typing errors, and there is, at present, no agreed upon means of dealing with these errors 31 (SanCristobal and Chevalet 1997, Marshall et al. 1998; see also Appendix 1). Nonetheless, when used in conjunction with direct measurements of immigration, genetic techniques, and other sources of supplemental information, ought to allow us to answer with certainty as to whether immigrants should be viewed as "intruders" that induce population declines, or as "rescuers" that prevent population extinction. 32 REFERENCES Beier. P., and Noss, R. F. 1998. Do habitat corridors provide connectivity? Conservation Biology, 12: 1241-1252. Bjornstad, O. N., Andreassen, H. P., and Ims, R. A. 1998. Effects of habitat patchiness and connectivity on the spatial ecology of the root vole Microtus oeconomus. Journal of Animal Ecology, 67: 127-140. Boonstra, R., and Krebs, C. J. 1978. Pitfall trapping of Microtus townsendii. Journal of Mammalogy, 59: 136-148. Boutin, S. 1980. Effect of spring removal experiments on the spacing behaviour of female snowshoe hares. Canadian Journal of Zoology, 58: 2167-2174. Brown, J. H., and Kodric-Brown, A. 1977. Turnover rates in insular biogeography: effects of immigration on extinction. Ecology, 58: 445-449. Charnov, E. L., and Finerty, J. P. 1980. Vole population cycles: a case for kin selection. Oecologia, 45: 1-2. Clinchy, M. 1997. Does immigration "rescue" populations from extinction? Implications regarding movement corridors and the conservation of mammals. Oikos, 80: 618-622. Eberhardt, L. L., and Thomas, J. M. 1991. Designing environmental field studies. Ecological Monographs, 61: 53-73. Efford, M. 1998. Demographic consequences of sex-biased dispersal in a population of brushtail possums. Journal of Animal Ecology, 67: 503-517. Fahrig, L., and Merriam, G. 1985. Habitat patch connectivity and population survival. Ecology, 66: 1762-1768. Gottfried, B. M. 1979. Small mammal populations in woodlot islands. American Midland Naturalist, 102: 105-112. Hanski, I., and Gilpin, M. E. 1997. Metapopulation processes. In Metapopulation biology: ecology, genetics, and evolution. Editedby Hanski, I., and Gilpin, M. E., Academic Press, San Diego, pp. 211-214. 33 Harlow, H. J., Lindzey, F. G., Van Sickle, W. D., and Gern, W. A. 1992. Stress response of cougars to nonlethal pursuit by hunters. Canadian Journal of Zoology, 70: 136-139. Hestbeck, J. B. 1982. Population regulation of cyclic mammals: the social fence hypothesis. Oikos, 39: 157-163. Ims, R. A., Rolstad, J., and Wegge, P. 1993. Predicting space use responses to habitat fragmentation: can voles Microtus oeconomus serve as an experimental model system (EMS) for capercaillie grouse Tetrao urogallus in boreal forest? Biological Conservation, 63: 261-268. Ims, R. A., and Yoccoz, N. G. 1997. Studying transfer processes in metapopulations: emigration, migration, and colonization. In Metapopulation biology: ecology, genetics, and evolution. Edited by Hanski, I., and Gilpin, M. E., Academic Press, San Diego, pp. 247-265. Krebs, C. J. 1992. The role of dispersal in cyclic rodent populations. Edited by Stenseth, N. C , and Lidicker, W. Z., Jr., Chapman and Hall, London, pp. 160-175. Krebs, C. J. 1999. Ecological methodology, 2nd ed. Benjamin/Cummings, Menlo Park, California. Lambin, X., and Krebs, C. J. 1991. Can changes in female relatedness influence microtine population dynamics? Oikos, 61: 126-132. Lidicker, W. Z., Jr., and Koenig, W. D. 1996. Responses of terrestrial vertebrates to habitat edges and corridors. In Metapopulations and wildlife conservation. Edited by McCullough, D. R., Island Press, Washington, D.C, pp. 85-109. Lidicker, W. Z., Jr., and Stenseth, N. C. 1992. To disperse or not to disperse: who does it and why? In Animal dispersal: small mammals as a model. Edited by Stenseth, N. C , and Lidicker, W. Z., Jr., Chapman and Hall, London, pp. 21-36. Marshall, T. C , Slate, J., Kruuk, L. E. B., and Pemberton, J. M. 1998. Statistical confidence for likelihood-based paternity inference in natural populations. Molecular Ecology, 7: 639-655. Moilanen, A., Smith, A.T., and Hanski, I. 1998. Long-term dynamics in a metapopulation of the American pika. American Naturalist, 152: 530-542. O'Donoghue, M. 1994. Early survival of juvenile snowshoe hares. Ecology, 75: 1582-1592. 34 Peacock, M. M., and Smith, A. T. 1997. The effect of habitat fragmentation on dispersal patterns, mating behaviour, and genetic variation in a pika (Ochotona princeps) metapopulation. Oecologia, 112: 524-533. Pugh, S. R., and Tamarin, R. H. 1991. A comparison of population characteristics and reproductive success of resident and immigrant meadow voles. Canadian Journal of Zoology, 69: 2638-2643. Pulliam, H. R. 1988. Sources, sinks, and population regulation. American Naturalist, 132: 652-661. SanCristobal, M., and Chevalet, C. 1997. Error tolerant parent identification from a finite set of individuals. Genetical Research, 70: 53-62. Smith, A. T. 1974. The distribution and dispersal of pikas: consequences of insular population structure. Ecology, 55: 1112-1119. Smith, A. T. 1980. Temporal changes in insular populations of the pika (Ochotona princeps). Ecology, 61: 8-13. Smith, A. T., and Gilpin, M. 1997. Spatially correlated dynamics in a pika metapopulation. In Metapopulation biology: ecology, genetics, and evolution. Editedby Hanski, I., and Gilpin, M. E., Academic Press, San Diego, pp. 407-428. Stacey, P. B., Johnson, V. A., and Taper, M. L. 1997. Migration within metapopulations: the impact upon local population dynamics. In Metapopulation biology: ecology, genetics, and evolution. Edited by Hanski, I., and Gilpin, M. E., Academic Press, San Diego, pp. 267-291. Stenseth, N. C , and Lidicker, W. Z., Jr. 1992. Where do we stand methodologically about experimental design and methods of analysis in the study of dispersal? In Animal dispersal: small mammals as a model. Edited by Stenseth, N. C , and Lidicker, W. Z., Jr., Chapman and Hall, London, pp. 295-318. van Apeldoorn, R. C , Oostenbrink, W. T , van Winden, A., and van der Zee, F. F. 1992. Effects of habitat fragmentation on the bank vole, Clethrionomys glareolus, in an agricultural landscape. Oikos, 65: 265-274. Watson, A., Moss, R., Parr, R., Mountford, M. D., and Rothery, P. 1994. Kin landownership, differential aggression between kin and non-kin, and population fluctuations in red grouse. Journal of Animal Ecology, 63: 39-50. 35 Wolff, J. O., Schauber, E. M., and Edge, W. D. 1997. Effects of habitat loss and fragmentation on the behaviour and demography of gray-tailed voles. Conservation Biology, 11: 945-956. 36 Table 3.1. Distance to the nearest occupied patch at time 1, as compared between populations found to be extinct or extant at time 2, based on results from four surveys of the Bodie, California, pika metapopulation, as reported in Smith and Gilpin (1997, their Fig. 2). Distances were compared between groups using Mann-Whitney U tests. Survey Fate of population at time 2 N Distance (m) to the nearest occupied patch at time 1 Median Range U z P 1972-77 Extinct 10 42 23-123 133.0 -1.15 0.2513 Extant 35 56 21-162 1977-89 Extinct 13 56 25-115 153.5 1.10 0.2720 Extant 30 46 21-94 1989-91 Extinct 9 48 27-137 107.5 -0.20 0.8451 Extant 25 55 21-137 Pooled Extinct 32 47 23-137 1408.5 -0.18 0.8545 Extant 90 52 21-162 37 Figure 3.1. Steps in delineating a peripheral "buffer strip" for use in the unambiguous identification of "true" immigrants. Step one (a), completely enumerate all residents (R), and the offspring of residents (r), using at least two, fully independent sampling techniques (x and y). Step two (b), divide the area of complete enumeration into a "core" area surrounded by a peripheral "buffer strip" which is a minimum of one home range length across. Step three (c), distinguish between movements which can (I) and cannot (NI) be unambiguously classified as "true" immigration. 38 Figure 3.2. Example of how immigration may be overestimated if handling has a significantly adverse effect on resident reproduction or survival. In the absence of handling (a), assume that the offspring (r) of adult residents (R) normally settle in adjacent home ranges (open circle) and thereby preempt (solid line) settlement by immigrants (I). If handling (b) causes deaths (barred circle) among local young, then home ranges that would have been occupied by the offspring of residents are instead occupied by immigrants. 39 CHAPTER 4 DOES IMMIGRATION "RESCUE" POPULATIONS FROM EXTINCTION? EVIDENCE FROM A LARGE-SCALE FIELD EXPERIMENT ON COMMON BRUSHTAIL POSSUMS According to metapopulation dynamics theory, landscapes are best viewed as "networks of idealized patches ... in which species occur as discrete local populations" (Hanski 1998, p. 41). Patches become unoccupied as a result of local extinctions, but are eventually reoccupied by dispersers from other occupied patches. In some metapopulation models (Hanski 1985, 1991), the exchange of immigrants between occupied patches is expected to "rescue" recipient populations from extinction, thanks to the "demographic and genetic contributions of immigrants" (Brown and Kodric-Brown 1977, p. 445). Moreover, such "models ... have shown that a surprisingly small number of immigrants per year (often between three and five adults, e.g., Stacey and Taper 1992 [Acorn Woodpeckers, Melanerpes formicivorus]) will allow individual populations to persist" indefinitely (Stacey et al. 1997, p. 268; see also Hanski 1991, Beier 1993 [cougars, Felis concolor], and Moilanen et al. 1998 [pikas, Ochotona prlnceps]). Although dispersal is the process underlying metapopulation dynamics, almost no attention is given to the nature of dispersal in most metapopulation models and applications (Clinchy 1997). In addition, studies of metapopulation dynamics rarely ever directly measure dispersal, but instead infer it from the pattern of patch occupancy (Clinchy 1997, Chapter 3). In this Chapter, I report the results of an experiment incorporating both direct and indirect measurements of dispersal, which was designed to test the role of immigration in "rescuing" populations of common brushtail possums (Trichosurus vulpecula) from extinction. The only factor considered in most metapopulation models and applications which reflects some aspect of the dispersal process itself, is the size and structure of physical barriers to dispersal. In almost all metapopulation models, the rate of dispersal between patches is assumed to be a function of the physical distance across some "uniformly unsuitable habitat" (Hanski and Gilpin 1991; Hanski 1998, p. 41). In applications of metapopulation theory to conservation biology, the issue of physical barriers to dispersal manifests itself in discussions of "connectivity", wherein "the best argument for corridors is 40 that the original landscape was interconnected ... [and] .. .corridors are simply an attempt to maintain or restore some of the natural landscape connectivity" (Noss 1987, p 162; Beier and Noss 1998). If the rate of exchange of immigrants is largely determined by physical barriers to dispersal (unsuitable habitat), it follows that the maximum rate of exchange of immigrants will be observed where there are no physical barriers to dispersal: in a uniformly suitable natural landscape. In a recent review, Lidicker and Koenig (1996) concluded that mammals are more likely than other vertebrate taxa to demonstrate metapopulation dynamics. Well-established patterns are apparent with respect to dispersal among mammals that would seem to have significant implications regarding metapopulation dynamics. Wolff (1997) argued that, in territorial species, emigration is generally inversely density-dependent, because neighbouring territory owners act as a "social fence" (Hestbeck 1982), preventing movement through each other's home range. Resident territory owners can also be expected to attempt to preempt immigration into an area, because immigrants represent additional, unrelated, intraspecific competitors, and residents will generally benefit more by having their own offspring settle in local vacancies (Clinchy 1997). "Natal dispersal", involving young, pre-reproductive individuals, is the norm among most mammals, while "breeding dispersal" by reproductively-active adults is something of a rarity (Greenwood 1980, Waser and Jones 1983, Johnson and Gaines 1980). Because the frequency of natal dispersal depends on the rate of production of young, the role of dispersal in a species' ecology will depend in part on its fecundity. The importance of this aspect of dispersal was recognized early on in the development of metapopulation theory by MacArthur and Wilson (1967, p.83), who argued that it is important to have a large intrinsic rate of increase to be a "good" colonizing species, although this seems to have been frequently ignored in subsequent metapopulation models and applications. Finally, in the vast majority of mammals, most dispersers are males, while most females are philopatric, settling on or adjacent to their mother's home range (Greenwood 1980, Waser and Jones 1983, Johnson and Gaines 1980). This aspect of dispersal seems to have been completely ignored in metapopulation theory, even though it is of immense significance, as it is the rate of female dispersal only, that determines how often new populations are established, or how often existing populations are demographically "rescued" from extinction (Clinchy 1997). 41 Several authors have commented on the difficulties associated with the accurate measurement of emigration rates and distances (Stenseth and Lidicker 1992, Koenig et al. 1996, Ims and Yoccoz 1997). Measuring immigration in open populations in the field is a task fraught with as many, if not more, methodological problems (Chapter 3). At the same time, the modeling exercises that suggest that very few immigrants are required to "rescue" populations from extinction can only be refuted if one is able to: 1) demonstrate that immigration is almost non-existent; and 2) be able to do so with, as nearly as possible, complete certainty. The foremost problem associated with the measurement of immigration is that a proportion of the existing residents will almost certainly elude detection when only a single sampling technique (e.g., live-trapping) is employed (Chapter 3). Consequently, any newly-identified individual may be an immigrant, or a resident that has previously eluded detection (Stenseth and Lidicker 1992). All attempts should be made (using multiple sampling techniques) to verify that the resident population has been completely enumerated, prior to classifying newly-identified individuals as immigrants (Chapter 3). At the same time, intensive efforts aimed at achieving complete enumeration may themselves generate errors (Chapter 3). If immigrants and residents respond differently to the stress of capture and handling, or if repeated capture and handling has adverse effects on reproduction or survival, the measured rate of immigration may be incorrect. The potentially adverse effects of capture and handling should be evaluated before drawing conclusions about the significance of immigration. Another problem associated with the measurement of immigration arises when the area of complete enumeration does not fully encompass the entire patch. In such cases a "buffer strip" (or "border zone", Stenseth and Lidicker 1992) should be delineated, wherein newly-identified individuals detected in the buffer strip are not classified as immigrants, as they may be residents from just outside the area of complete enumeration that are either: 1) exploring beyond the boundaries of their normal home range (Lidicker and Stenseth 1992); or 2) are expanding their home range in response to vacancies created within the area of complete enumeration (Stenseth and Lidicker 1992, Chapter 3), a phenomenon sometimes referred to as the "vacuum effect" (Efford 1991). The vacuum effect is most commonly expected to occur in territorial species (Carpenter 1987, Mares and Lacher 1987, Schoener 1987). "True" immigration has the potential to "rescue" a recipient population from extinction by adding new individuals to an existing population, 42 whereas the "reshuffling" of the existing population, associated with the vacuum effect, cannot be expected to "rescue" the population from extinction. Misidentifying the vacuum effect for "true" immigration will cause the rate and importance of immigration to be overestimated. Given the many methodological problems associated with measuring immigration in open populations in the field, it ought not to be surprising that there have been very few such studies conducted on mammals (e.g., Boutin 1980, Lambin and Krebs 1993; see also Stenseth and Lidicker 1992), and they have all been relatively "short-term" (extending over 2 or 3 years). The short-term nature of these studies has allowed critics to claim that the results are not representative of processes occurring over the long term (Beier and Noss 1998). The frequency of dispersal over the longer term can be indirectly gauged by the magnitude of genetic differences between populations (Ims and Yoccoz 1997). However, this approach cannot distinguish between male and female dispersal, and as mentioned above, only the latter is of relevance with respect to the rate of establishment of new populations, or the demographic "rescue" of existing populations. The likely frequency of female dispersal, over the longer term, can be indirectly gauged by the strength of female philopatry. If most females do not disperse, but instead settle on or adjacent to their mother's home range, this should be discernible from the spatial pattern of female relatedness. For example, Ishibashi et al. (1997) used estimates of relatedness derived from microsatellite DNA to compare whether adjacent pairs of female grey-sided voles (Clethrionomys rufocanus) were more closely related than would be expected from a random dispersal pattern. The strength of female philopatry may also be judged by deviations from parity in the primary sex ratio (Johnson 1988, 1989). If most daughters settle on or adjacent to their mother's home range, while most sons disperse, there will be competition for local resources between mothers and daughters, that does not exist between mothers and sons. To reduce the level of local resource competition with their philopatric daughters, mothers should reduce the number of daughters they produce, resulting in a male-biased primary sex ratio. Male-biased dispersal per se is not sufficient to produce the biased sex ratio. If both sons and daughters disperse, but males disperse farther, dispersal is said to be male-biased, but in this case mothers and daughters are not in contact with one another, and are therefore not competing with one another, so there is no reason to reduce the number of daughters produced. 43 Consequently, a significantly male-biased primary sex ratio not only suggests that sons move farther, but also that most daughters generally do not disperse at all. Ah underlying assumption of the "local resource competition" hypothesis is that sons and daughters are equally "costly" to produce (Johnson 1988), and this must be evaluated prior to drawing inferences regarding female philopatry from deviations in the primary sex ratio. The disruption of immigration between remnant populations has been identified as a key factor underlying the many declines (26 spp.) and extinctions (16 spp.) that have occurred among medium-sized (35-5500 g) mammals in Australia, in the 200 years since European settlement (Morton 1990). Correspondingly, Australian mammals feature prominently in the literature on connectivity. In a recent review, Beier and Noss (1998) found 16 studies on mammals suggestive of a positive role for conservation corridors, of which, 6 involved Australian marsupial possums and gliders (Families Burramyidae, Petauridae and Phalangeridae). Similarly, Laurance and Gascon (1997) advocated the retention of corridors in association with logging in tropical rainforests throughout the world, based in part on Laurance's (1990) survey work in Australian tropical rainforests on three species of ringtail possum (Family Petauridae) and the common brushtail possum (Family Phalangeridae). At the time of European settlement, the common brushtail possum had the largest geographic range of any Australian marsupial, encompassing approximately 94 % of the continent (How and Kerle 1995). Although considered to be "relatively adaptable to anthropogenic habitats" (Beier and Noss 1998, p. 1248) because they frequently occur in urban areas, common brushtail possums (along with most other medium-sized mammals; Burbidge and McKenzie 1989, Short and Smith 1994) have been almost entirely extirpated from the center of the continent (Finlayson 1961, Kerle et al. 1992), and presently occupy only about 37 % of their historical range in Australia (How and Kerle 1995). While both rodents and insectivorous marsupials are included among those species of Australian mammals that have suffered declines since European settlement, medium-sized, herbivorous, marsupials were more adversely affected than other groups (Burbidge and McKenzie 1989, Short and Smith 1994). I chose to study the common brushtail possum as a "model" medium-sized, herbivorous, marsupial. Common brushtail possums are similar in most respects to most of the other species of medium-sized, herbivorous, marsupials that have suffered declines. These species are all included in the suborder 44 Phalangeriformes (possums and kangaroos, Szalay 1994). Alpin and Archer (1987, p. liii) commented that "in terms of ecological diversity, [Phalangeriformes] is comprised predominantly of very generalized animals showing only minor variations in basic body plan, body size and lifestyle". Caughley and Gunn (1996, p. 61) concluded that those species that have suffered declines are all "ecologically similar enough to be considered a set". In this group, female age at first breeding ranges from 6 months to 2 years, and between 1-3 young are born each year (Lee and Cockburn 1985, Lee and Ward 1989). Calculated intrinsic rates of increase are comparable (Sinclair et al. 1998). Most of the species concerned are generally solitary, weakly territorial, have a promiscuous mating system, and demonstrate male-biased dispersal (Lee and Cockburn 1985, Croft 1989, Johnson 1989, Seebeck et al. 1989). Despite the frequency with which research on medium-sized Australian mammals appears in the literature on movement corridors and connectivity, and the prominence of place given to dispersal in Morton's (1990) explanation of the decline of medium-sized mammals in Australia, relatively little is known about the importance of dispersal in the population dynamics of these species (Lee and Cockburn 1985, Johnson 1989). Accordingly, there has been no prior attempt to experimentally quantify the role of immigration in these species. I measured the rate of immigration by female common brushtail possums in response to a spatially and temporally replicated, large-scale, field experiment involving the "pulsed" removal of resident females. The experiment was conducted in a vast (121,000 ha), uniformly suitable, natural landscape of old-growth Eucalypt forest in south-eastern Australia. I chose the seemingly counterintuitive approach of studying metapopulation dynamics in a continuous landscape under the assumption that if immigration plays little role in "rescuing" populations from extinction in a continuous landscape, it is likely to play an even smaller part where there are physical barriers to dispersal, as in fragmented or patchy landscapes. To determine if the results of my relatively "short-term" (2 year) experiment corresponded with the likely frequency of female dispersal, over the longer term, I used two methods to gauge the strength of female philopatry. I conducted parentage analysis using microsatellite DNA, and compared the frequency with which adjacent females were identified as being putative mothers and daughters, with that expected if daughters always settle on or adjacent to their mother's home range. In addition, I examined whether the sex ratio of pouch-young was significantly male-biased, as predicted by the "local resource competition" hypothesis, 45 under conditions of strong female philopatry. Contrary to inferences drawn from surveys assessing connectivity, my results suggest that immigration likely plays little part in "rescuing" common brushtail possum populations from extinction. I discuss the implications of this result in terms of the ecology of medium-sized, herbivorous, marsupials, and caution against overgeneralizations regarding the role of dispersal in mediating local and global extinctions. METHODS Study species The common brushtail possum (hereafter simply "possum"), is a solitary, nocturnal, medium-sized (1-3 kg), herbivorous, marsupial, endemic to Australia (Kerle et al. 1991, How and Kerle 1995). While having declined in Australia, possums are considered to be a significant pest species in New Zealand, where they were first introduced in 1837 (Cowan 1990). Although possums have been the subject of intensive research in New Zealand (Cowan 1990), comparatively little research has been conducted on the population ecology of possums in Australia (Green 1984, Kerle 1984). The demography of possum populations in south-eastern Australia closely resembles that observed in New Zealand (Chapters 5 and 6). Almost all adult females give birth to a single young in April or May (antipodean autumn). The young first emerges from the pouch at around 175 days of age (Sept.-Oct.), and then rides on its mother's back for another month or two. Back-young are referred to as "juveniles" by some researchers (Crawley 1973), in distinction to "subadults", which are independent (weaned) young, that have not yet become reproductive! y-active. Subadult females are readily distinguished from adult (reproductively-active) females by the condition of the pouch (Clout and Efford 1984). In subadults, the pouch is little more than a shallow indentation on the abdomen. When a female first becomes reproductively-active the pouch enlarges and invaginates, and undergoes a variety of other characteristic changes (Bolliger and Carrodus 1938, 1940). Age at first reproduction varies from 1-3 years (Chapter 5). Subadult males may be distinguished from adult males by the length of the testes (Clout and Efford 1984). Tyndale-Biscoe (1955) reported that males with testes < 18 mm in length were never observed to have 46 spermatozoa present in the epididymides, while males with larger testes were always found with spermatozoa present, independent of season. Adults in undisturbed areas may live up to 14 years (Chapter 6). While adult survival is generally thought to be about 80 % per annum, I present evidence in Chapter 6 that suggests that adult survival is adversely affected by handling, and the "true" rate of adult survival is more likely about 90 % per annum. Survival of pouch-young to the onset of weaning (at 175 days of age) may vary significantly from one year to the next, ranging from 33-90 % (Chapter 5). Less variation is apparent in the survival of young from the onset of weaning, to their first birthday (365 days of age), with estimates ranging from 43-60 % (Dunnet 1964, Efford 1998, Chapter 5). Survival of yearling females (from 1-2 years) is not significantly different from that of older females, according to results from a 15-year study in New Zealand (Efford 1998). In wooded areas, possums typically spend the day in a den in the hollow branch of a tree, and emerge at night to feed on leaves, grass, herbs, flowers and fruit (Kerle 1984). An animal's foraging range is often considerably larger than its "denning range". I define the latter as the minimum convex polygon connecting all dens used by an individual (see also Lindenmayer et al. 1997). While there is general agreement that male and female ranges overlap extensively, ^ there is some disagreement as to whether individuals are territorial with respect to others of the same sex (Green 1984). The disagreement stems in part from the different methods used to measure range size. Male ranges are always found to be larger than female ranges, regardless of the method used (Green 1984). Both male and female ranges increase and decrease in size, depending on the season. Several studies (Crawley 1973, Ward 1978, Green and Coleman 1986) have found that male ranges are largest during the autumn breeding season. Shifts in female range size do not appear to be as consistent. Crawley (1973) reported that female ranges expanded in autumn and winter, while Green and Coleman (1986) found that females had larger ranges in summer. Green and Coleman (1986) suggested that changes in female range size reflected seasonal changes in food abundance, which may be expected to show different patterns at different sites. Seasonal changes are also observed in the composition of pairs of individuals found sharing dens. During the breeding season, females in oestrous are often followed by one or more male "consorts" (Clout and Efford 1984), that may share the female's den (Caley et al. 1998). In the rare instances where den sharing is observed outside the breeding season, the animals are generally both females, that are assumed to be related (Caley et al. 1998). 47 Clout and Efford (1984) reviewed the results of studies conducted in both Australia and New Zealand and concluded that possums, like most other mammals, demonstrate male-biased dispersal (see also Efford 1991, 1998). Clout and Efford (1984) argued that movements in excess of 2 km ought to be considered long-distance dispersal, given that both male and female home ranges in continuously forested habitats are usually less than 500 m in length. Female home range lengths up to 1800 m have been recorded, but only under unusual circumstances where individuals were "commuting" from forest-interior den sites to an agricultural field, and back again, on a nightly basis (Green and Coleman 1986). Efford (1991) compiled evidence from all available sources and found 68 reported cases of long-distance dispersal, 13 of which involved females. In the majority (78 %) of cases reviewed by Efford (1991), animals dispersed from their natal range prior to their first birthday, while the remainder all dispersed prior to their second birthday. There was no evidence of long-distance dispersal by reproductively-active females (i.e., breeding dispersal), among the cases reviewed by Efford (1991). Study site The study was conducted in the eastern arm. of Paddys Land State Forest, which lies near the center of the 121,000 ha Guy Fawkes Wilderness Area (GFWA), in northeastern New South Wales (NSW), Australia. The GFWA encompasses a few small "peninsulas" of the New England Plateau (maximum elevation = 1300 m) and the surrounding eastern slopes of the Great Dividing Range (NPWS 1992). The rugged terrain has prevented substantial logging or land clearance, leaving the continuous old-growth Eucalypt forest cover largely intact (NPWS 1992). Consequently, when the GFWA was legally "identified" as wilderness by the NSW government it was judged to have met the criterion that "the area is, together with its plant and animal communities, in a state that has not been substantially modified by humans and their works" (NPWS 1992, p. 6). The GFWA is "unparalleled in south-eastern Australia" in being home to "every species of forest dependent marsupial" (in NSW; NPWS 1992, p. 35). In particular, the "high altitude open forests presently within State Forests contain the highest concentration of arboreal mammals found in NSW" (NPWS 1992, p. 20), and are "very important for arboreal mammals ... at all scales (i.e., regional, state and continental)" (NPWS 1992, p. 35). The GFWA consists of "uniformly suitable habitat" for common brushtail possums, 48 and possums are predicted to occur at moderate to high densities throughout the GFWA, based on a regional species distribution model incorporating a large number of site variables, including latitude, temperature, moisture, geology, terrain and vegetation type (NPWS 1994). I established two 36 ha study grids in the eastern arm of Paddys Land State Forest in September, 1994 (Fig. 4.1). On each grid I marked out 49 permanent trap locations at roughly 100 m intervals in a 7 x 7 matrix (Figs. 4.2 and 4.3). The grids were established at roughly 2 km distance from one another to ensure that they were independent replicates with respect to dispersal, as any movement by an individual from one grid to another could be classified as long-distance dispersal (Clout and Efford 1984). The midpoint between the two study grids lies at latitude 30°06'00" S and longitude 152°10'20" E. The Paddys Land area is considered to be part of the New England Plateau (NPWS 1992), and the elevation on the grids ranged from 940-1040 m. The bottom corner of the East grid was within 100 m of the edge of the Plateau. A number of individuals captured on the East grid had dens to the south and east of the bottom corner of the grid (Fig. 4.3) which were located in trees growing on the steep sides of the Plateau. While measured slopes on the grids never exceeded 10.0°, the mean (+/- S.E.) measured slope down the side of the Plateau was 18.1 0 (+/- 1.0), and the maximum slope I recorded was 32.0°. ,. Chapman and Binns (1995) conducted flora surveys in May, 1993, at three sites in the vicinity of the study grids, one near (within a few hundred metres) the West grid, one near the East grid, and one near the midpoint between the two grids. Chapman and Binns (1995) described the soil as a sedimentary red podsolic. The general floristic community was classified as "grassy woodland", with a very sparse shrub stratum largely composed of Acacia filicifolia and Allocasuarina littoralis (Chapman and Binns 1995). The area was lightly grazed by domestic livestock, and subject to infrequent, low intensity fires. Median annual rainfall is 837.8 mm (Range = 302.2-1247.4 mm; based on 64 years of data from the Kookabookra weather station, latitude 30°00'39" S, longitude 152°00"34" E, elevation 975 m). Seven "forest types" (Forestry Commission of NSW 1989) were identifiable in the vicinity of the study grids based on aerial survey data available from the Walcha District offices of the Forestry Commission of NSW. The seven "forest types" include: No. 70, Spotted Gum (Eucalyptus maculata); No. 72, Spotted Gum - Grey Box (E. maculata - E. moluccana); No. 82, Grey Box (E. moluccana); No. 93, Eastern Red Gums (E. amplifolia dominant); No. 122, New England Stringybark (E. caliginosa); No. 161, Roundleaved 49 Gum (E. deani); and No. 163, New England Blackbutt (E. andrewsii). Chapman and Binns (1995) classified 2 of the 3 sites they surveyed as "forest type" No. 122, New England Stringybark, and the third was classified as "forest type" No. 70, Spotted Gum. I measured the distance to, and height of, the nearest tree (defined as: > 20 cm DBH and > 2 m in height) in each of the four quadrants surrounding each permanent trap location on each grid (Figs. 4.2 and 4.3). Using program POINT QUARTER (Krebs 1999), I estimated that the mean (+/- S.E., and 95 % CL.) density of trees was 137.0 per ha (+/- 0.4, 119.0 -157.5) on the West grid, and 163.4 per ha (+/- 0.4, 141.8 -187.8) on the East grid. Median tree height was 19.8 m (Range = 9.3 - 43.8 m) on the West grid, and 20.5 m (Range = 4.0 - 43.8 m) on the East grid. Detection, capture, and handling I used two, independent sampling techniques to detect the presence of possums on the study grids: 1) live-trapping; and 2) spotlighting. Trapping involves simultaneous detection and capture. To capture animals detected during spotlighting, I followed two procedures: 1) whenever feasible, I attempted to tranquilize the animal using a C02-powered dart-gun; otherwise 2) I set one or more traps at the base of the tree in which the animal was detected. When captured for the first time, animals were weighed, tagged and sexed. Weight was recorded to the nearest 50 g using a spring scale. Aside from those individuals captured using the dart-gun, animals were generally not tranquilized prior to handling, but were instead immobilized using a variety of restraints. All animals (except pouch-young) were tagged in each ear with two types of Monel metal tags (National Band and Tag Co., Newport, Kentucky, U.S.A.). Individually numbered tags (size 1005-3) were located in a proximal and medial position. Smaller tags (size 1005-1) wrapped with coloured reflective tape were located in a distal and lateral position. Individuals were tagged in each ear with up to three coloured tags coded for sex, grid, and year of first capture. Pouch-young that were judged to have reached a sufficient size were tagged in each ear with smaller (size 1005-1), individually numbered tags, located in a proximal and medial position. The interior (non-numbered) edge of pouch-young tags were wrapped with coloured reflective tape, to ensure that surviving young were identifiable as being marked individuals, if they were later seen during spotlighting. Tags did not interfere with the development of the pinna of the ear. 50 When males were captured for the first time I measured testis length to the nearest 0.1 mm using vernier calipers. If the individual was judged to be an adult (testis length >= 18.0 mm), testis length was measured if and when it was next captured, to confirm that the animal was an adult, but was not measured on subsequent recaptures. If the individual was judged to be a subadult (testis length < 18.0 mm), testis length was measured on all subsequent occasions on which it was captured and handled, until such time as it was judged to have become an adult. The condition of a female's pouch was checked on the first occasion on which she was captured, and on all subsequent occasions on which she was captured and handled. When a female was found to be bearing a pouch-young I attempted, whenever possible, to: 1) determine the sex of the young; and 2) measure the length of its head, to the nearest 0.1 mm, using vernier calipers. Head length is the most reliable trait to use in estimating the age of pouch-young (Chapter 5). It was not always possible to examine and measure the young because: 1) the pouch-young was judged to be too small to safely handle; or 2) the mother was struggling too much. As measurements were generally attempted by a single person, in the field, on animals that were not tranquilized, a judgement had to be made in each case on the trade-off between an accurate measurement and the possibility of harming the mother or causing her to eject the pouch-young. When the pouch-young was judged to be too small to handle safely, I attempted to estimate crown-rump length (Lyne and Verhagen, 1957, their Fig. 5), which can also be used to estimate the age of pouch-young. Crown-rump length was estimated to the nearest 5.0 mm by holding a ruler adjacent to the body of the pouch-young. For slightly larger pouch-young (head length <= 10.0 mm), and in instances where the mother was struggling too much, head length was estimated to the nearest 5.0 mm by holding a ruler adjacent to the head of the pouch-young. A nursing mother may leave an older dependent young in a den, while she searches for food (Winter 1976, p. 246). Consequently, the survival of a young to the onset of weaning may be inferred even if the young itself is not seen, if the mother is found to be still nursing at the appropriate phase in the development of the young (Chapter 5). Females were classified as still nursing if: 1) they had an engorged mammary gland; and 2) they were still lactating. An engorged mammary gland was operationally defined as one that could be readily identified as a distinct oval shaped organ with minimum dimensions 51 of 20.0 x 30.0 mm, when measured with vernier calipers (Dunnet 1964, Smith et al. 1969, Crawley 1973). Lactation was determined by squeezing the female's teat. Whenever a female's pouch was examined I also noted if she was: 1) "rumpy"; or 2) "bony". "Rump wear" is a condition involving fur loss, lesions, and scab formation, generally on the lower back and base of the tail. The condition is typically observed in association with heavy infestations by the haematophagous, ectoparasitic mite, Trichosurolaelaps crassipes (Presidente 1984, Hemsley and Canfield 1993, Clark 1995, Chapter 6). To check a female's pouch it was necessary to restrain the female's hindlegs with one hand while examining the pouch with the other hand. Contrasts between females were apparent in the condition of their hindlegs. The majority of females' hindlegs felt muscular while some females' hindlegs felt like little more than skin and bone. Such contrasts have been used previously in evaluating the condition of adult possums (Humphreys et al. 1984). Females observed to be both "rumpy" and "bony" had a significantly lower probability of survival than females that were never seen to be so (Chapter 6). Aside from those selected for removal, all animals were released at their point of capture, during the day. Upon release I attempted to follow the animal as it made its way to a den or some other refuge .-; (Dunnet 1956). I noted all occasions on which an animal either stopped and began eating grass, or climbed a tree and began eating leaves, as feeding during the daytime has been interpreted as a sign of poor condition (Mackintosh et al. 1995). In addition, I noted all occasions on which an animal was seen to be "weak and wobbly" upon release. Symptoms of more or less severe ataxia (loss of motor control) have been reported in association with "wobbly possum syndrome" in New Zealand (Mackintosh et al. 1995). Survivorship among females observed to be "weak and wobbly" upon release was significantly poorer than among females that were never seen to be so (Chapter 6). Definitions and experimental design I began live-trapping on the study grids in November, 1994. Residency was defined with respect to a (somewhat) arbitrarily chosen date of August 1, 1995, such that any female captured prior to August 1, 1995, was defined as being a resident female, as was the daughter of any such female (Chapter 3). While the regional species distribution model mentioned above predicted that the GFWA was "uniformly suitable habitat" for possums, the spatial resolution of the model was not such that areas as small as an individual's 52 home range could be judged to be "suitable" or not. Female immigrants would not be expected to settle on the study grids if, as a result of stupendous bad luck, the habitat on the study grids was somehow "unsuitable". To ensure that the study grids were indeed "suitable" habitat, I defined residency with respect to the latest date (August 1) by which I expected all females to have given birth, if they were going to breed in that year (1995). I could then say without ambiguity that the study grids were "suitable" habitat for at least as many breeding females as were known to be present on August 1, 1995 (Boutin 1980). Subsequent records, over three breeding seasons, indicated that the August 1 cut-off was reasonable, as the latest date on which a female was known to have given birth to a primary young (defined on p. 144) was July 11 (Chapter 5). Among those resident females captured prior to, and known to be still alive on or after August 1, 1995, 31 of 33 on the West grid, and 27 of 29 on the East grid, were known to have given birth in the preceding breeding season. On each grid, there was one subadult female with a range interior to the periphery of the trapping grid (Figs. 4.2 and 4.3), that did not breed; and one adult female with a range exterior to the periphery of the trapping grid, that was first captured prior to the beginning of the 1995 breeding season, that may have given birth, and lost her young, before being recaptured after August 1, 1995. All four females in question occupied roughly the same ranges, and were known to have given birth, in the following breeding season (1996). There were three components to the experimental design involving: 1) the complete enumeration of resident females; 2) the delineation of a peripheral "buffer strip"; and 3) the "pulsed" removal of resident females. The steps taken to ensure the complete enumeration of resident females are discussed in a subsequent section. Each study grid was divided into arbitrarily defined "core" and "peripheral" areas (Figs. 4.2 and 4.3). Resident females with denning ranges to the interior of the inner dashed lines shown in Figures 4.2 and 4.3 were defined as "core" resident females, while those with denning ranges to the exterior of the inner dashed lines were defined as "peripheral" resident females. Females first identified after August 1, 1995, were defined as "potential immigrants". Non-breeding potential immigrants that disappeared after a period of less than three months were classified as "transients" (Dunnet 1956, Crawley 1973). Any potential immigrant that was either found to be bearing a pouch-young, or was known to have been present for three or more months, was classified as a "settler". I defined the "buffer strip" as being the roughly 100 m 53 distance between the inner and outer dashed lines shown in Figures 4.2 and 4.3. Assuming I was successful in completely enumerating all resident females, any potential immigrant that settled in the core of one of the two study grids would have had to have: 1) come from outside the periphery of the trapping grid (shown by the outer dashed lines in Figs. 4.2 and 4.3); and 2) would therefore have had to have dispersed across the 100 m buffer strip to reach the core of the grid (shown by the inner dashed lines in Figs. 4.2 and 4.3). If, as I expected to be the case, the buffer strip was occupied by a more or less contiguous ring of peripheral resident females, it could also be concluded that any potential immigrant that settled in the core had dispersed across at least one home range (that of a peripheral resident female) in order to reach her destination. Lidicker and Stenseth (1992) argued that permanent movement across one home range length should be considered as a minimum criterion for any biologically meaningful definition of dispersal. Consequently, only potential immigrants that settled in the core were classified as "true immigrants". Potential immigrants that settled in the buffer strip, or exterior to it, were not classified as "true immigrants", as they were as likely to be females with home ranges just outside the initial area of complete enumeration (Chapter 3). In order for immigration to "rescue" populations from extinction, female immigrants must enter and add to an existing population of resident females (Clinchy 1997). Most metapopulation models assume that immigrants replace residents lost as a result of stochastic processes, in which case, immigrants are more likely to "rescue" populations from extinction the more rapidly they replace lost residents (Clinchy 1997). To measure the rate at which immigrants replaced lost residents, I removed all of the core resident females from both study grids on August 1, 1995. If a female's denning range straddled the dashed line defining the core of the study grid (Figs. 4.2 and 4.3), I determined the position of the arithmetic mean center of her denning range in order to judge whether she ought to be considered a core or peripheral resident female. Range expansion in response to removals (the vacuum effect) is generally not considered to be part of dispersal (Lidicker and Stenseth 1992). Consequently, if peripheral resident females expanded their ranges into the core of the study grid following the removal of core resident females, I did not consider this to be immigration (Chapter 3). The question I was interested in was not whether peripheral females replace core females, but whether immigrants replace residents. 54 I monitored the rate at which female immigrants entered and settled in the core removal areas from August 1, 1995, to August 1, 1996. In order to recreate conditions on the grids as they had existed after the August 1, 1995, removals, and thereby temporally replicate the experiment, any female that had established her denning range in the core of either grid during the 1996 breeding season (March to June, Chapter 5) was removed on August 1, 1996. I chose to remove any and all "true immigrants", as well as any originally peripheral resident female that had expanded her denning range into the core. The rate at which female immigrants entered and settled in the core removal areas was then monitored for a second year, from August 1, 1996, to August 1, 1997. Over the course of the study, several of the originally peripheral resident females either died, or expanded their denning ranges into the core and were removed, leaving gaps in the ring of peripheral females. In such cases I generally made a pointed effort to try to capture any female with a range just exterior to that of the originally peripheral resident female. I classified females captured under such circumstances as "secondary" peripheral females. Part of my aim in following this protocol was to maintain a more or less contiguous ring of peripheral females, but I also wished to maintain adequate sample sizes in order to estimate demographic parameters (Chapter 6). Following from the definitions above, if any of the secondary peripheral females, or their daughters, shifted their denning ranges into the core they would be classified as "true immigrants", as they would have dispersed across the formerly occupied range of the originally peripheral resident female. Much of the data I present is in the form of maps showing the location of individuals on the study grids (Figs. 4.4-15). The location of each female is indicated by an uppercase letter, either by itself, or followed by a number, or preceded by either a lowercase or uppercase letter. Single uppercase letters identify adult or subadult resident females. Uppercase letters followed by a number identify daughters tagged when still with their mother (who is identified by the uppercase letters). Uppercase letters preceded by the lowercase letter "d" indicate resident females that died prior to August 1, 1995. Uppercase letters preceded by the lowercase letter "r" indicate core resident females removed on August 1, 1995. Uppercase letters preceded by the uppercase letter "A" indicate females first identified after August 1, 1995, which are by my definition all potential immigrants. Letters or symbols enclosed in a rectangle indicate the location at which an individual was captured, in those cases where the animal was not radio-collared. Single 55 lowercase letters enclosed in a diamond symbol indicate where unmarked possums were seen during spotlighting. When referring to a given individual in the body of the text, the individual's identifier is preceded by an uppercase letter "W" or "E", to indicate which study grid I am discussing (e.g., "W-A" refers to resident female "A" on the West grid). Radio telemetry and den mapping All but four of the resident females known to be alive immediately prior to the August 1, 1995, removals, were radio-collared. The four exceptions included two subadult females (W-Pl and E-D, Figs. 4.4 and 4.10) and two adult females (W-B and W-U, Fig. 4.4). I did not radio-collar the subadults because I was concerned that I would lose the radios if these females dispersed off the study grids. The two adults in question were first captured prior to February, 1995, when I first began radio-collaring, and were not recaptured until after August 1, 1995. In addition to the 60 radio-collared resident females (30 per grid) known to be alive immediately prior to August 1, 1995 (Figs. 4.4 and 4.10), there were 3 radio-collared resident females on the East grid that had died prior to August 1 (E-dA, E-dB and E-dC; Fig. 4.10). „ Following the August 1, 1995, removals, I managed to radio-collar 3 of the 4 residents that were not previously radio-collared (W-B, W-Pl and E-D). In addition, I radio-collared all but three newly-identified females (i.e., potential immigrants) that were judged to have settled (i.e., found possessing a pouch-young or known to be present for three or more months) on the study grids. In one of the three cases where a newly-identified settler was not radio-collared, the female in question (W-AI, Fig. 4.7) died during capture. In the other two cases, the individuals in question (W-AA and E-AA, Figs. 4.5 and 4.11) were the very first newly-identified females captured after August 1, 1995. Both females had pouch-young, and both were captured within two weeks of August 1. By too rigidly interpreting my own definitions, I categorized both females as potential immigrants (because they were captured after August 1), and judged that I ought not risk the expense of having "dispersers" run-off with a couple of expensive radios. I subsequently changed my protocols such that any newly-identified female captured on the study grids that was found to be bearing a pouch-young was given a radio-collar. 56 Each radio-collar (total weight = 35 g; Titley Electronics, Ballina, NSW, Australia) was equipped with a mortality sensor. In addition, each radio had a whip antenna, which I wrapped with coloured reflective tape, using individually-identifiable combinations of colours. I occasionally needed to locate a female at night so that I could attempt to dart her, in order to replace her radio, or tag her pouch-young. If the female's radio had already failed (18 cases) I could still identify her from the colour combinations on the antenna. Each female's radio signal was checked on a weekly basis, to ensure that she was still alive and her radio was still operational. Radio-tracking principally involved homing in on a female's den during the day. Only daytime tracking was used in defining a female's denning range. Females were never located more than once in the same day, since daytime movements by possums are extremely rare (Mackintosh et al. 1995). The majority of radio-tracking "sessions" took place at the same time as "full-grid" trapping sessions (Tables 4.1 and 4.2), as I would generally be involved in trapping on one grid while an assistant radio-tracked on the other grid. When defining a female's denning range I excluded all cases where a female tracked during the day had been captured on the previous night. I used a female's denning range to describe her location on the study grid, in part because this ought to best reflect the core of her home range (Winter 1976). In addition, I was interested in determining whether females that entered the removal areas would use the same dens as had been used by the resident females, which would suggest that suitable dens are an important limiting resource (Caley et al. 1998). All den sites were marked with individually numbered aluminum tags. I determined the position of each den site, relative to the nearest permanent trap location, by using a sighting compass and measuring the distance to the nearest 0.1 m using a tape line. When measurements were made up or down hillsides, the slope was recorded using a clinometer. The positions of the permanent trap locations were surveyed and mapped by Resource Engineering students from the University of New England (Armidale, NSW, Australia), using theodolites, sighting compasses and tape lines. All locations shown on the grid maps (Figs. 4.2, 4.3, and 4.4-4.15) are plotted using Euclidean coordinates (i.e., corrected for elevation). 57 Live-trapping Possums were live-trapped using 30 x 30 x 60 cm, hook-operated, wire mesh cage traps (Mascot Wire Works, Sydney, Australia) set at the base of a tree and baited with apple. Animals trapped overnight were processed and released beginning at dawn the next day. In addition to live-trapping on the study grids (described below), I live-trapped along the vehicle trail running between and away from the two study grids (Fig. 4.1). In the latter case I positioned 40 traps at roughly 100 m intervals along the vehicle trail, which were baited and set for 13 nights (520 trap-nights), during October and November, 1996. My aim in trapping along the vehicle trail was to assess the reproductive success of females that had not been previously captured and handled (Chapter 5). Information was also gained regarding exploratory movements by animals normally trapped on the study grids. I followed three procedures when live-trapping on the study grids, which I define as: 1) full-grid trapping; 2) target trapping; and 3) nighttime trapping. I define a full-grid trapping session as one in which traps were set at all 49 permanent trap locations on a grid (Figs. 4.2 and 4.3). During most full-grid trapping sessions traps were set for 3 consecutive nights, however, I also include in this category both 1-night and 4-night trapping sessions. Animals were weighed and handled on the first occasion on which they were captured during a trapping session and were thereafter simply released from the trap if they were recaptured during the same trapping session. Over the course of 32 months, from the beginning of December, 1994, to the end of July, 1997,1 conducted 26 full-grid trapping sessions on the West grid (Table 4.1) and 24 full-grid trapping sessions on the East grid (Table 4.2). I conducted monthly full-grid trapping sessions on both grids during the autumn breeding seasons in each of the three years of the study (1995, 1996, 1997). During winter and spring I occasionally did not conduct a full-grid trapping session in a given month, but instead concentrated on "target trapping" (described below). I often did not trap in every month during summer because trapping success was poorer and there was less information to be gained. Instead, I concentrated on measuring den tree characteristics and other activities. Target trapping involved three procedures: 1) setting traps at a den known to be presently occupied by a radio-collared female; 2) setting traps at dens formerly occupied by peripheral resident females, with the aim of capturing secondary peripheral females; and 3) setting traps at refuge sites (which were presumed to be dens) that were seen to be used by males upon release from capture (Dunnet 1956), in areas of the grid 58 where no female had yet been caught. Being able to trap radio-collared females at their dens was very important for ensuring the complete enumeration of all resident pouch-young (Chapter 3). It was also extremely advantageous to be able to use the denning ranges of currently radio-collared females to identify areas on the grid where a female "ought" to be found, and then concentrate trapping effort in that area. Nighttime trapping was that conducted in conjunction with spotlight transects. If an unmarked possum was detected during a spotlight transect, and the dart-gun was unavailable or unusable (because it was too windy, or the animal was too high in the canopy), I set as many traps as were readily available, at the base of the tree in which the animal was seen, or had run to. Spotlighting and darting Possums have a distinctive orange eye-shine which is readily detected at night with the use of a spotlight. I used 30-watt spotlights powered by 12-volt lead-acid batteries. With the aid of binoculars I was generally able to determine the colour of the reflective ear-tags born by marked possums, at distances of up to 50 m. I usually began spotlighting about an hour after dark and typically finished before midnight. To facilitate spotlighting I marked out transect lines between each permanent trap location on each grid (Figs. 4.2 and 4.3) using nails wrapped with reflective tape. I conducted four types of spotlight transects: 1) full-grid transects; 2) core transects; 3) peripheral transects; and 4) target transects. Full-grid transects involved starting at one corner of a study grid (Figs. 4.2 and 4.3), moving up a trapline, and then across and down the next, and so forth. Core transects involved moving between core traps only (Figs. 4.2 and 4.3), while following the same procedure as for full-grid transects. Peripheral transects involved circling around the periphery of the trapping grid (shown by the outer dashed lines in Figs. 4.2 and 4.3). A target transect, like target trapping, involved an intensive search in an area of the grid where a female "ought" to be found. Possums seen on transect lines connecting two core traps (Figs. 4.2 and 4.3) were categorized as having been seen in the core (Table 4.3), while possums seen on transect lines connecting core and peripheral traps, or two peripheral traps, were categorized as having been seen on the periphery of the study grid (Table 4.4). I conducted spotlight transects on a regular basis, between late February (just prior to the beginning of the autumn breeding season) and August 1, in each of the three years of the study (1995, 1996, 1997). 59 I spent an average of 500 minutes (Range = 345-825) spotlighting each area (core vs. periphery), of each grid, in each year (Tables 4.3 and 4.4). The level of effort was comparable to similar spotlighting surveys, such as that by Laurance (1990), who spent an average of 481 minutes (Range = 175-740) spotlighting for possums at each of 20 sites in tropical rainforest in northeastern Australia. For various reasons, I conducted an additional 440 minutes worth of spotlighting in the core of the West grid in the late winter and early spring of 1995 (Table 4.3). I used a Montech Model 2 C02-powered dart-gun (Montech Pty Ltd, Melbourne, Australia), whenever possible, to attempt to capture unmarked possums detected during spotlighting, or to recapture radio-collared females that were resistant to target trapping. The dart-gun fired a modified 0.5 ml insulin syringe, which was filled with "Zoletil 100" (active ingredients include Tiletamine and Zolazepam; Virbac Pty Ltd, Sydney, Australia), at a concentration of 30 mg per 0.15 ml. Complete enumeration I enumerated a total of 200 adult and subadult possums, 105 (47 females, 58 males) on the West grid, and 95 (48 females, 47 males) on the East grid. I also enumerated a total of 154 dependent young (pouch-young or juveniles), 82 on the West grid, and 72 on the East grid. Trapping success (measured as the proportion of traps possessing possums during full-grid trapping sessions) was generally very good. Mean (+/- S.E.) trapping success was 40.2 % (+/- 2.0) on the West grid (Table 4.1), and 33.0 % (+/- 1.8) on the East grid (Table 4.2). Table 4.1 shows the number of captures during full-grid trapping, target trapping, and nighttime trapping, relative to the dates on which full-grid trapping sessions were conducted on the West grid. Table 4.2 shows the comparable data for the East grid. Tables 4.1 and 4.2 also show the number of times animals were either darted or resighted during spotlighting, relative to the dates on which full-grid trapping sessions were conducted. A resight refers to an animal detected during spotlighting, that was not captured, but whose identity could nonetheless be determined, either because it was a radio-collared female, or because it bore a unique combination of coloured reflective ear-tags. I followed five steps in evaluating whether I had achieved my goal of completely enumerating all adult and subadult possums on the study grids, which is one of the necessary preconditions that has to be met in 60 order to unambiguously identify immigrants (Chapter 3). The first step involved comparing between a total count of the population, such as that generated using the minimum-number-alive (MNA) method (Krebs 1999), and a mark-recapture estimate of population size, such as the Jolly-Seber estimate. If the population has been completely enumerated, the MNA and Jolly-Seber estimates should be nearly the same (Krebs 1999, p. 94). Deviations between the results generated by the two methods may arise, either because: 1) complete enumeration has not been achieved; or 2) some of the assumptions underpinning the population estimate have been violated. The Jolly-Seber estimate assumes that all animals marked at a given time are equally catchable, and that those not yet marked are as equally catchable as those already marked (Krebs 1999, p. 49). To evaluate the first of these two assumptions regarding equal catchability (step 2 in evaluating complete enumeration) I determined how often a radio-collared female, known to be present on a grid, was actually captured during full-grid trapping sessions on that grid. To evaluate the latter assumption regarding equal catchability (step 3 in evaluating complete enumeration) I plotted the cumulative number of newly-identified individuals captured over time, which ought to reach a more or less abrupt asymptote if animals that are not already marked are as equally catchable as those that are (Krebs 1999, p. 56). Animals that are completely wwcatchable using a given technique (e.g., live-trapping) can only be detected by using a different technique (e.g., spotlighting; Krebs 1999, p. 50). As the fourth step in evaluating the success of complete enumeration, I compared the proportion of marked and unmarked animals detected during spotlighting, with the expectation that the number of unmarked animals detected should be very nearly zero (except for transients or immigrants that have not yet been captured). As a final step I used all available information, such as the circumstances of capture, reproductive status, and likely relatedness to residents, to draw inferences regarding the potential origin of newly-identified individuals. Table 4.5 lists MNA and Jolly-Seber estimates regarding the number of females on the West grid, over the course of the study. I present three MNA tallies, concerning the number of females known to be alive based solely on full-grid trapping (MNA-Grid), the number of radio-collared females (MNA-Radio), and the number of females known to be alive based on all available data (MNA-A11). To permit comparisons with the Jolly-Seber estimate, only females that were at some point captured or resighted along or interior to the periphery of the trapping grid were included in the MNA-Radio and MNA-A11 tallies, as these females were all, at least potentially, "susceptible" to being caught during the full-grid 61 trapping sessions, and thereby stood some chance of being included in the Jolly-Seber estimate. Table 4.6 lists MNA-Grid, MNA-Radio, MNA-A11, and Jolly-Seber estimates, regarding the number of females on the East grid. Tables 4.7 and 4.8 list MNA-Grid, MNA-A11, and Jolly-Seber estimates, regarding the number of males on the two study grids. For the most part the MNA-Grid and Jolly-Seber estimates of population size were very nearly the same, regardless of sex or study grid (Tables 4.5-8). MNA-A11 was generally greater than the Jolly-Seber estimate of population size because newly-identified individuals were often first captured during target trapping (Tables 4.1 and 4.2) at peripheral den sites (Figs. 4.2 and 4.3), prior to being caught on the grid, which gave me some idea as to their (recent) point of origin. The Jolly-Seber B statistic reflects "additions" to the population (sampled by the trapping grid), composed of: 1) newly-identified animals whose origin is unknown; 2) animals from just off the grid that have been previously captured during target trapping; and 3) the surviving offspring of residents, that have begun to fend for themselves. In the latter case, the offspring of residents are included in the Jolly-Seber B statistic as newly-identified individuals, when in fact their identity is known (because they were tagged while still in the pouch). The Jolly-Seber B statistic should be compared to Figures 4.16-19, which plot the cumulative number of newly-identified individuals caught at core and peripheral locations on the grid. For purposes of comparison with the Jolly-Seber estimates of N and B, Figures 4.16-19 show what is equivalent to the cumulative number of MNA-A11, insofar as only animals caught or detected on or interior to the periphery of both grids, are included. As with MNA-A11, individuals are added to the cumulative tally shown in Figures 4.16-19 at the time at which they are detected. Because there was often some delay between the time at which individuals were first captured during target trapping, etc ... (and thereby added to the MNA-A11 tally), and when they were later caught at one of the permanent trap locations (and thereby included among the "additions" reflected in the Jolly-Seber B statistic), the timing of increases in B may not correspond perfectly with the timing of increases in MNA-A11, as shown in Figures 4.16-19. Nonetheless, the aggregate number of additions as tallied in Figures 4.16-19, and as estimated by the Jolly-Seber B statistic, were very nearly the same, regardless of sex or study grid (Tables 4.5-8). Figure 4.20 shows the relationship between the MNA-Grid and Jolly-Seber estimates of female numbers, when the two estimates are plotted against one another. There should be a 1:1 relationship if 62 attempts at complete enumeration have been successful (Krebs 1999, p. 94). If the two estimates are not equal, the Jolly-Seber estimate will always be the larger. The median deviation between the MNA-Grid and Jolly-Seber estimates was 0.6 on the West grid and 1.0 on the East grid. The largest single deviation between the estimates (5.7) occurred on the East grid in session 7 (Table 4.6), which was prior to the "deadline" (August 1, 1995) for achieving complete enumeration (Table 4.2). Moreover, the Jolly-Seber estimate of the number of females (30.7) present in session 7, was equivalent to the number of females (31) I eventually managed to capture and radio-collar prior to session 8. The number of radio-collared females shown in session 8 in Table 4.6 is 21, because one radio-collared female died and nine more were removed, in between sessions 7 and 8. The Jolly-Seber estimate of population size will exceed the estimated MNA if marked animals become trap-averse, because the marked segment of the population will then be underestimated relative to the unmarked segment. I evaluated whether marked animals became trap-averse by determining the proportion of times radio-collared females that were known to be present on a grid were actually captured during full-grid trapping sessions on that grid. Figure 4.21 shows the relevant data as regards 77 of the total of 79 females, radio-collared over the course of the study. The two females not included in Figure 4.21 are E-dA (Fig. 4.10), who was first radio-collared in session 6 and died prior to session 7 (Table 4.2), and E-AJ (Fig. 4.15), who was first radio-collared during the last full-grid trapping session (24) on the East grid (Table 4.2). Most of the 77 radio-collared females for which data were available, were captured during full-grid trapping sessions, most of the time (Fig. 4.21). However, seven females (9 %) were not recaptured during full-grid trapping sessions, during the time that they were radio-collared. Several of the seven females in question occupied denning ranges outside the periphery of the trapping grids, and rather than being trap-averse, they simply may not have been captured during full-grid trapping sessions because they were not often in the vicinity of any of the permanent trap locations. Indeed, 4 of the 7 females in question were each trapped on repeated occasions when I attempted to trap them at their dens. The remaining three females (4 % of the total of 77) were never successfully re-trapped, even when I set traps at their dens. I set traps at the dens of the three females in question (E-rA, E-AF, and W-K) on 1,5, and 18 occasions, respectively. Female W-K was unquestionably trap-averse. Fortunately, I was able to recapture W-K on three occasions using the dart-gun. Moreover, the presence of W-K would not have gone 63 undetected, even if she had avoided being trapped in the first place, because she was also successfully resighted during spotlighting. In a "closed" population (= no transients or immigrants, in this context), if unmarked animals are as equally catchable as marked animals, the cumulative number of newly-identified individuals should reach a more or less abrupt asymptote, and then level off (Krebs 1999, p. 56). In an "open" population (where some of the unmarked animals may be transients or immigrants), the level should increase more or less gradually after asymptoting. Figure 4.16 shows the cumulative number of newly-identified females first caught in core traps on both grids. On the East grid the number of newly-identified females asymptoted fairly abruptly and leveled-off, prior to the "deadline" for achieving complete enumeration (August 1, 1995). On the West grid, four females were first trapped in core traps following the August 1, 1995, "deadline". Based on supplementary information (discussed later), I was able to classify 3 of the 4 females first identified after August 1, 1995 (W-AE, W-AF and W-AK), as new "recruits" (either transients or immigrants). Supplementary information (discussed later) indicated that W-AD (one from among the 30 non-"recruits" on both grids = 3 %) did appear to be a resident female that was initially resistant to being trapped. Notably, while WrAD may have initially been somewhat trap-averse, she was readily re-trapped on seven occasions over the subsequent 11-month period, prior to her death. Figure 4.17 shows the cumulative number of newly-identified females first caught in peripheral traps on both grids. The cumulative number of newly-identified females reached an abrupt asymptote just after the August 1, 1995, "deadline" for complete enumeration. The four females (W-AA, W-AB, W-AC and E-AA) first trapped within the first two weeks of August, 1995, were all found to be bearing pouch-young. As suggested earlier, classifying the four females in question as potential immigrants because they were first captured after the (somewhat) arbitrary August 1 "deadline", can be faulted for being too rigid, and it is more biologically meaningful to classify them as residents, given that they were carrying pouch-young when they were first captured, and were therefore known to have given birth within the vicinity of the study grid during the 1995 breeding season. All of the 10 females first captured after the second week of August, 1995 (Fig. 4.17), could be classified as secondary peripheral females. From the preceding, I interpret Figure 4.17 as indicating that among the originally peripheral resident females 64 under consideration, previously unmarked females were no less catchable than those that had already been marked. Figure 4.18 shows the cumulative number of newly-identified males first caught in core traps on both grids. Comparing Figures 4.16 and 4.18, it is clear from the gentler slope prior to the asymptote, in the latter figure, that previously unmarked males were less catchable than previously unmarked females. There are at least two reasons why males may be less catchable. One reason is that males have larger home ranges (Green 1984) and spend less time in the vicinity of the trapping grid. The other reason may be that males spend more time looking for females and less time looking for food. Consistent with the suggestion that males are less catchable is the fact that the cumulative number of newly-identified males shown in Figure 4.18 did not appear to reach an asymptote until the end of winter, in 1995, six weeks after the August 1 "deadline" for complete enumeration. In addition, while the male first caught in the autumn of 1996 on the West grid (Fig. 4.18) was clearly a new "recruit" (weight - 1333 g, testis length = 10.9 mm; likely less than 1 year old, Chapter 5), the male first caught on the East grid at roughly the same time was an adult, that subsequent trapping suggested, may have been resident all along, with a range in the vicinity of that of female E-X (Fig. 4.10). The two males first caught at the end of summer, in51997 (Fig. 4.18), on the West grid, were an adult, and a large (2000 g) subadult, respectively. While I cannot rule out the possibility that these two males were residents that had eluded detection, it had been more than 17 months since a newly-identified resident male had been caught in the core of the West grid, and it seems safe to conclude that the two males in question were in fact new "recruits". The two newly-identified males first trapped in the autumn and winter of 1997 (Fig. 4.18) were both clearly new "recruits", as they were both likely less than (or were just barely) one year old (weight < 1750 g, testis length < 11.0 mm; Chapter 5). The cumulative number of newly-identified males first caught in peripheral traps reached an asymptote on both grids by the middle of spring, 1995 (Fig. 4.19), and then began gradually increasing again, from the autumn of 1996 onwards. The pattern just described is congruent with that described regarding males first trapped in core traps (Fig. 4.18). Peripheral resident males (Fig. 4.19) appear to have been less catchable than peripheral resident females (Fig. 4.17), given the gentler slope, and greater delay before asymptoting, in Figure 4.19 as compared to Figure 4.17. Also as discussed with respect to males first caught in core traps (Fig. 4.18), the gradual increase in the number of newly-identified peripheral males (Fig. 4.19) from 65 the autumn of 1996 onwards, in part reflects the addition of new male "recruits" (such as the three yearling, and two older males, described in relation to Fig. 4.18), as is to be expected, given that dispersal in possums is known to be male-biased (Clout and Efford 1984; Efford 1991, 1998). From the preceding discussion I conclude that I was successful in enumerating all but one of the females resident on both study grids, by the end of the second week of August, 1995, and that with the capture of female W-AD at the end of November (late spring), 1995,1 had achieved my goal of completely enumerating all of the resident females on both grids. As I will explain in a later section, I feel confident in classifying W-AD as a resident, based on the analysis of microsatellite DNA. My conclusion regarding the complete enumeration of resident females assumes that there are no uncatchable females. As the fourth step in evaluating the success of complete enumeration, I compared the results from live-trapping with those obtained using a second, independent, sampling methodology (Chapter 3), namely spotlighting. Table 4.3 shows the number of times both marked and unmarked possums were detected in the core areas of both grids during spotlight transects. Table 4.4 shows the comparable data as regards possums detected on the periphery of both grids. Possums were seen during spotlight transects on 302 occasions, of which 30.1 % (91) were resights (individually-identifiable, Tables 4.1 and 4.2). The last occasion on which an unmarked animal was seen in the core (Table 4.3) of the West grid, prior to the August 1, 1995, "deadline" for complete enumeration, was in early June, and the last occasion on which an unmarked animal was seen in the core of the East grid, prior to August 1, 1995, was in late February. These results are clearly congruent with those based on live-trapping (Figs. 4.16 and 4.18), that indicate that most of the animals present in the core areas of both grids had been captured and marked prior to August 1, 1995. Unmarked animals continued to be detected on the periphery of both grids (Table 4.4) up to the time of the August 1, 1995, removals. This too is in accordance with the live-trapping results (Figs. 4.17 and 4.19), wherein unmarked peripheral residents continued to be caught during the first few weeks of August, 1995. Seven of the unmarked animals detected on the periphery of the grids during spotlighting, prior to August 1, 1995, were successfully captured (Table 4.4). Six of the seven animals in question were first captured by darting, including: four peripheral resident females on the East grid (E-F, E-N, E-W and E-X; Fig. 4.10); one peripheral resident female on the West grid (W-I, Fig. 4.4); and one male. All six animals first captured by darting were eventually, and then repeatedly thereafter, caught in traps. 66 From August 1, 1995, until the end of the study in July, 1997,1 detected possums in the core areas of the study grids during spotlight transects on 68 occasions, of which 3 occasions (4 %) involved unmarked animals (Table 4.3). I detected possums on the periphery of the grids, over the same period, on 90 occasions, of which 10 occasions (11 %) involved unmarked animals. I was able to capture the unmarked animals detected during spotlighting on 1 of the 3 occasions on which they were seen in the core (Table 4.3), and 2 of the 10 occasions on which they were seen on the periphery (Table 4.4). Although I was unsuccessful at capturing several of the unmarked animals detected during spotlighting, at the time at they were first seen, I feel confident that most of the individuals in question were eventually trapped. The locations at which unmarked animals were detected, but not captured, during spotlight transects conducted after August 1, 1995, are all indicated on the appropriate range maps (Figs. 4.4-15) by lowercase letters enclosed in a diamond symbol. The sequence of letters (a, b, c, etc..) follows the sequence of occurrences. For example, W-a (Fig. 4.6) indicates the first occasion (after August 1, 1995) on which an unmarked animal was seen on the West grid, which was in late February, 1996. Secondary peripheral female W-AG (Fig 4.7) was first caught at this same location about 2 1/2 • months later, in early May. W-b (Fig. 4.7) indicates the location at which a notably smaller unmarked animal was seen in early April, 1996. A small (1770 g, Chapter 5), adult male was first trapped at an adjacent trap in late April, who was thereafter repeatedly trapped in that vicinity. The arithmetic mean center (AMC) of the trapping range (based on N = 17 records) occupied by the male in question was 24.7 m northeast of the location of W-b, shown in Figure 4.7. Similarly, W-c (Fig. 4.8) was a notably smaller animal seen in late February, 1997, and a small (1717 g, Chapter 5), adult male was first trapped at an adjacent trap in late March, with an AMC (N = 12) 50.2 m northeast of the location of W-c, shown in Figure 4.8. W-d (Fig. 4.9) signifies a larger animal seen in late March, 1997, and a large (2525 g, Chapter 5) male was first trapped in an adjacent trap in early June, with an AMC (N = 3) 22.0 m northwest of the location of W-d, shown in Figure 4.9. Larger, unmarked animals were seen at locations W-e and W-f (Fig. 4.9) in May and July, 1997, respectively. No new animals were trapped in the vicinity of these two locations in the remaining three months of the study, although new animals continued to be trapped elsewhere on the periphery of the West grid (Figs. 4.17 and 4.19). 67 The first occasion (E-a, Fig. 4.13), after August 1, 1995, on which an unmarked animal was seen during spotlighting on the East grid, was in late May, 1996. Secondary peripheral female E-AE (Fig. 4.13) was first caught at this same location about a month later, in late June, and was re-trapped at this location over the next several months. The next occasion on which an unmarked animal was seen during spotlighting on the East grid (E-b, Fig. 4.15), involved a notably smaller animal seen in the core of the grid in mid-June, 1997. Only four days later, a small (1700 g, Chapter 5), subadult male was trapped at an adjacent trap (shown by the male symbol in Fig. 4.15). No unmarked animals had been seen during spotlight transects in the core of the East grid since February, 1995 (26 months earlier, Table 4.3), and no new animals had been trapped in the core of the East grid since mid-April, 1996 (14 months earlier, Figs. 4.16 and 4.18). Consequently, I am completely certain that the subadult male that was trapped was the same individual that was seen during spotlighting four days earlier. Larger, unmarked animals were seen at locations E-c and E-d (Fig. 4.15) in June and July, 1997, respectively. No new animals were trapped in the vicinity of these two locations in the remaining two months of the study, although new animals continued to be trapped elsewhere on the periphery of the East grid (Figs. 4.17 and 4.19). From my review of the circumstances surrounding sightings of unmarked animals, I am confident that the unmarked animals detected during spotlighting transects were, on all but the last four occasions, eventually trapped. I infer from this that the unmarked animals detected on the periphery of both grids during spotlighting in the last three months of the study (occasions W-e, W-f, E-c and E-d; Table 4.4) would also have eventually been trapped, if the study had continued. Considering the ever-increasing number of newly-identified males captured on the periphery of the grids (Fig. 4.19), as compared to the small number of newly-identified females (Fig. 4.17), I conclude that the unmarked animals detected during spotlighting in the last three months of the study were most likely males. Based on my evaluation of the evidence available from spotlighting, I conclude that there were no wncatchable animals that would not have eventually been live-trapped, given that: 1) all of the individuals that were initially captured by darting were eventually trapped; 2) the large majority of animals seen during spotlighting were in fact marked and had, therefore, already been trapped (Tables 4.3 and 4.4); and 3) the fact that for most occasions on which an unmarked animal was seen during spotlighting, I was able to identify an individual of the same relative size, trapped in the same vicinity, a short time later. 68 Determining whether there were wwcatchable animals in the population was the fourth step in evaluating the success of complete enumeration. The final step involves using all available information to draw inferences regarding the potential origin of newly-identified individuals. All of the supplemental information, which I will describe later, reinforces the following two conclusions: 1)1 was entirely successful in completely enumerating all adult and subadult possums present in the core areas of both study grids; and 2) I was successful in completely enumerating all adult and subadult possums present on the periphery of both grids over most of the duration of the study. In addition to completely enumerating all adults and subadults, all of the offspring of resident females must also be accounted for, as a necessary precondition for the unambiguous identification of immigrants (Chapter 3). Including the females with pouch-young initially caught in the first two weeks of August (W-AA, W-AB and W-AC; Fig. 4.5), and female W-AD (Fig. 4.6), there were 26 adult resident females known to be alive and "present" (= MNA-A11) on the West grid, immediately following the August 1, 1995, removals. In 21 of the 26 cases in question I was able to either tag the female's young while it was still in the pouch (18 cases), or I was certain that her young had died (3 cases). In 2 of the 3 cases where I was certain the young had died, the mother was found with no pouch-young, and was no longer nursing, 141 and 166 days, respectively, after the young had been born. Evidence presented in Chapter 5 indicates that surviving young never leave the pouch before 175 days, and are never completely weaned before 200 days of age. In the third case where I was certain the young had died, the young was found dead and rotting in the mother's pouch (the mother being apparently still healthy). There were five cases on the West grid, in 1995, where I could not account for the whereabouts of the offspring of a resident female. Females W-B and W-AD (Figs 4.4 and 4.6) were not caught during the 1995 breeding season and may or may not have given birth and produced a surviving young. Females W-I and W-0 (Fig. 4.4) did produce young, but I was unable to recapture these females prior to the time at which their young may have weaned. Female W-U was resighted with a young on her back in late October (at the position indicated in Fig. 4.6), but I was unable to catch her. I collected tissue samples from 4 of the 5 females in question (the exception being W-U, Table 4.10), and was thus able to determine if any potential immigrants captured after August 1, 1995, were in fact the untagged offspring of the females in question, based on the analysis of microsatellite DNA. 69 There were 23 adult females known to be alive (= MNA-A11) on the West grid immediately after the August 1, 1996, removals. Among the 23 females in question, one did not breed in 1996 (W-S, Fig. 4.7), and another died before its young could have weaned (W-K, Fig. 4.7). In 20 of the 21 remaining cases I was successful in tagging the young in the pouch. In the one case where the young was not tagged, the mother (W-AD, Fig. 4.7) was found with no young, and was no longer nursing, 115 days after the birth of the young. I conclude that there were no unaccounted for offspring of resident females on the West grid in 1996. Including female E-AA (Fig. 4.11), who was found to possess a pouch-young when initially caught in the first two weeks of August, there were 22 adult resident females known to be alive (= MNA-A11) on the East grid, immediately following the August 1, 1995, removals. Two of the females in question died before their young could have weaned (E-U and E-Y, Fig. 4.11). In three more cases I was certain that the young had died. In 2 of the 3 cases where I was certain the young had died, the mother was found with no pouch-young, and was no longer nursing, 135 and 143 days, respectively, after the young had been born. In the third case, the mother had ejected the pouch-young when she was trapped and it was found dead on the floor of the trap. In .4 5 of the remaining 17 cases I successfully tagged the young while it was still in the pouch. In two cases, involving females E-K and E-M (Fig. 4.11), I could not account for the whereabouts of the offspring, but I had collected tissue samples from the two mothers in question (Table 4.11), and was thus able to evaluate whether any "potential immigrants" were in fact the untagged offspring of these two females, based on the analysis of microsatellite DNA. There were 19 adult females known to be alive (= MNA-All) on the East grid immediately after the August 1, 1996, removals. Three of the females in question died before their young could have weaned (E-A, E-N and E-O; Fig. 4.14). In three more cases I was certain that the young had died because the mother was found with no pouch-young, and was no longer nursing, 102, 118, and 162 days, respectively, after the young had been born. I was successful in tagging the young in the pouch in all of the remaining 13 cases. I conclude that there were no unaccounted for offspring of resident females on the East grid in 1996. From the preceding discussion I feel confident in stating that there was only one case (involving female W-U, Fig. 4.6), in a total of 83 cases (1.2 %), where originally peripheral resident females were 70 known to have given birth (and were still alive on August 1, in either 1995, or 1996, or both), in which the offspring of a resident female could have been misidentified as being an immigrant, if it was later recaptured. In addition, I was not at a complete loss as regards the untagged offspring of female W-U, because I at least knew its (approximate) age, given that it was already riding on its mother's back (Chapter 5) when it was resighted in late October, 1995. I have not discussed the enumeration of young during the 1997 breeding season because the study ended at the end of July, 1997, well before any of the young produced in that year would have begun to emerge from the pouch (Chapter 5). I have examined the evidence for complete enumeration in great detail because I wish it to be understood that when I present the results of the experiment, what I am discussing are (as nearly as possible) total censuses of the female population. Consequently, I have (or am as nearly as possible to having) the capacity to distinguish "true immigrants", without ambiguity (Chapter 3). Adequacy of location data from radio telemetry I radio-collared 1 subadult, and 78 adult (37 West, 41 East) females, over the course of the study. Females were tracked to their dens on 3,109 occasions. Deciding whethena female has settled in the core of one of the two study grids, and should thereby be classified as a "true immigrant", requires an accurate means of assessing her location. The adequacy with which radio-tracking data can be said to describe the extent on an animal's movements can be gauged by plotting the number of radio-fixes against some measure of the animal's movements. The point at which the plot reaches an asymptote provides an indication of the minimum number of fixes required to adequately describe the animal's movements (Lindenmayer et al. 1996). Figure 4.22 plots the number of fixes recorded for a given female against the total number of dens used by that female. There was a significant correlation between the total number of fixes and total number of dens used, on both the West grid (N = 38, Spearman's r = 0.46, t = 3.15, p = 0.0033) and the East grid (N = 41, Spearman's r = 0.75, t = 6.99, p < 0.0001). I conclude that there was no minimum number of fixes, below the maximum of 79 recorded for females on both grids (Fig. 4.22), that was adequate to describe the total number of dens used by a given female. If a female uses ever more dens that are all within a circumscribed area, this will have no effect on the estimated size of her denning range, which was defined previously as being the minimum convex polygon 71 connecting all dens used by an individual. Figure 4.23 plots the number of radio-fixes recorded for a given female against the female's total denning range length (the maximum distance between any two dens used by the same female). There was no significant correlation between the total number of fixes and the total denning range length of females on the West grid (N = 38, Spearman's r = 0.25, t = 1.54, p = 0.1327), but there was one on the East grid (N = 41, Spearman's r = 0.68, t = 5.84, p < 0.0001). On the West grid, as few as 8-10 fixes were as sufficient as up to 79 fixes, in order to adequately describe a female's denning range length (Fig. 4.23). The discrepancy between the grids derives in part from two differences in the data sets. On the East grid there were a number of females (5) for which there were very few fixes (< 8; Fig. 4.23). In addition, five of the females on the East grid (E-Q, E-R, E-S, E-T and E-AB; Figs. 4.10-4.15), that appeared to use a large number of dens, and had very large denning range lengths, had dens in trees growing on the steep sides of the Plateau on which the grids were located. These den trees on the side of the Plateau were significantly (t = 3.80, df = 187 p = 0.0002) taller (N = 41, mean height in m +/- S.E. = 27.5 +/- 1.1) than den trees elsewhere on the East grid (N - 148, 23.3 +/- 0.5). The rougher terrain and taller trees meant that it generally took much longer to locate the five females in -.question, and there was likely a correspondingly greater rate of misidentification of den trees, which may have inflated the estimated number of dens and total denning range lengths occupied by these females. I conclude that the number of radio-fixes recorded was sufficient to accurately determine the location of a female's denning range in all cases on the West grid, and in most cases on the East grid. A very few fixes may be required to determine a female's general position if one den is used much more frequently than all others. The maximum number of dens used by any given female was 15, in which case, it is possible to establish the pattern of den tree preferences for all females for which there was a minimum of 15 fixes. If den trees were selected at random by a female, the proportion of times she was found in a given den would be equal across all dens. Figure 4.24 plots the mean proportion of fixes tracked to each of the array of dens used by a given female. Females on both the West grid and the East grid were located in their most preferred den about 50 % of the time. Consequently, there was a greater than 95 % probability that a female would be found in her principal den at least once, even if she was only located five times (probability = 1 - 0.5", where n is the number of fixes). Only two females (E-dA, Fig. 4.10; and E-AJ, Fig. 4.15), from among the 79 (3 %) I radio-collared, were located fewer than five times. 72 I conclude that the number of radio-fixes recorded was sufficient to have included a female's principal den in all but these two cases on the East grid. In the various range maps used to describe the position of each female (Figs. 4.4-15), I have used "bubble plots" to indicate the frequency with which a female was located at a given den, and have drawn polygons to indicate each female's denning range. Collection of tissue samples and analysis of microsatellite DNA From October, 1995, onwards, I collected tissue samples from all adult and subadult possums captured on both study grids. Tissue samples were collected from pouch-young if they were judged to be large enough to be not too adversely affected. For various reasons, tissue samples were available from a small number of animals that had died or been removed before October, 1995. Tissue samples were collected from a total of 176 individuals (97 West, 79 East). In 27 cases (17 West, 10 East), samples were simultaneously collected from mothers and their dependent (known) young. Tissue samples were generally obtained by cutting a small notch (< 5 mm on a side) in the distal part of an animal's ear. All samples were stored in 100 % ethanol (analytic reagent rate quality), and were then forwarded to Dr Andrea Taylor (Macquarie University, Sydney, Australia), for genetic analysis. Possums at the Paddys Land site were successfully typed at 6 microsatellite loci (Table 4.9), which are described in detail in Taylor and Cooper (1998). I calculated allele frequencies, linkage disequilibrium among loci, allele frequency differences between grids, F S T , Rho sx, and deviations from Hardy-Weinberg equilibrium (HWE), using program GENEPOP, Version 3.1c (Raymond and Rousset 1995). I used Fisher's combined probabilities test (Sokal and Rohlf 1995, p. 794) to obtain a global probability value over all loci, when comparing allele frequencies between grids (Paetkau et al. 1995), and when testing deviations from HWE (Piertney et al. 1998). Mean number of alleles per locus, mean proportion of loci typed, and the estimated frequency of null alleles, were all calculated using program CERVUS, Version 1.0 (Marshall 1998). Results of tests regarding linkage disequilibrium and deviations from HWE are presented in Appendix 1. I conducted parentage analysis using both program KINSHIP, Version 1.2 (Goodnight and Queller 1998), and program CERVUS, Version 1.0 (Marshall 1998). These programs differ in: 1) their sensitivity to deviations from Hardy-Weinberg equilibrium; and 2) their treatment of typing errors. Data from 73 the 27 cases in which tissue samples were simultaneously collected from both mother and young were used to compare the two programs as to the accuracy with which the most likely mother chosen by each program correctly identified the true mother (Appendix 1). I used iterated runs to derive a typing error rate for program CERVUS that produced slightly more accurate results than program KINSHIP. When I used the method of estimating typing error rates recommended by the authors (Marshall et al. 1998) of program CERVUS, the program performed much more poorly than program KINSHIP. Since there is disagreement in the literature about the most appropriate method of treating typing errors (SanCristobal and Chevalet 1997, Marshall et al. 1998; see also Appendix 1), I report the most likely mother identified by both program CERVUS, using the iterated (empirically more accurate) error rate, and program KINSHIP, when considering parentage among adult (and subadult) females, of unknown relatedness. To identify putative parent-offspring relationships among adult (and subadult) females, each adult (and subadult) female for which microsatellite typing data were available was compared against every other typed female on the same grid, and the candidate female with the highest likelihood score was judged to be the most likely "mother" of the female under consideration (SanCristobal and Chevalet 1997). Obviously, when comparing among adult females of unknown age, an individual identified as the most likely ^  "mother", may in fact be the daughter of the female under consideration. Putative mother-daughter pairs were classified as being adjacent if: 1) their ranges overlapped; or 2) no other female's range was known to have separated the ranges of the putative mother and daughter. RESULTS Female immigration following removals Figure 4.4 shows the location of all females captured on the West grid prior to August 1, 1995. On August 1, 1995,1 removed the 10 center-most females from the core of the grid (Fig. 4.5). All 10 females removed on August 1 had pouch-young, indicating that the core of the West grid represented suitable habitat for at least 10 reproductively-active females. Figure 4.10 shows the location of all females captured on the East grid prior to August 1, 1995. Three of the peripheral resident females shown in Figure 4.10 (E-dA, E-dB and E-dC) died during June 74 and July. All three were known to have given birth earlier in the year. On August 1, 1995,1 removed the nine center-most females from the core of the grid (Fig. 4.11). All nine females removed on August 1 had pouch-young, indicating that the core of the East grid represented suitable habitat for at least nine reproductively-active females. Over the course of the year following the August 1, 1995, removals, only one newly-identified female (W-AE, Figs. 4.6 and 4.7) entered and settled in the core of one of the two study grids (Figs. 4.7 and 4.13), and thereby qualified as a "true immigrant". Also over the course of the year, two originally peripheral resident females on each grid (W-J and W-Q, Fig. 4.7; E-D and E-P, Fig. 4.13), either largely or completely shifted their denning ranges into the core areas of the grids, consistent with the vacuum effect (see below). On August 1, 1996,1 removed W-J, W-Q and W-AE from the core of the West grid (Figs. 4.7 and 4.8), and E-D and E-P from the core of the East grid (Figs. 4.13 and 4.14), in order to recreate conditions on the grids similar to those existing on August 1, 1995 (Figs. 4.5 and 4.11), and thereby temporally replicate the experiment. No newly-identified females entered and settled in the core of either of the two study grids over the course of the year from August 1, 1996, until the end of the study in late July, 1997 (Figs. 4.9 and 4.15). Consequently, there was only one "true immigrant" female (W-AE, Figs. 4.6 and 4.7) that entered and settled in the core of one of the two study grids, over the course of the two years following the August 1, 1995, removal of 19 core resident females. W-AE was first caught in early February, 1996 (Fig. 4.6). She was a small (1833 g, Chapter 5), subadult, that was between 1 and 2 years (circa 22 months) old when first captured (as determined from cementum annuli, Chapter 5). W-AE was judged to have settled in the core of the grid because she was present for more than three months (see Methods), but she did not breed during the 1996 breeding season, and was still a subadult on August 1, 1996. Given her age, W-AE could not have been the untagged offspring of female W-U (Fig. 4.6), described earlier. Parentage analysis using microsatellite DNA (Table 4.10) indicated that W-AE was almost certainly (p < 0.01, Appendix 1) the daughter of secondary peripheral female W-AG (Fig. 4.7). Given that W-AE was the daughter of W-AG, she clearly met Lidicker and Stenseth's (1992, p. 24) minimum criterion for "true" dispersal, by moving across at least one home range, occupied by peripheral resident female W-W (Fig. 4.7). Moreover, since W-AE was the only "true immigrant" female, and all 75 known daughters settled adjacent to their mothers (see below), it can also be said that W-AE's movement across one home range represented the maximum female dispersal distance recorded in this study. The vacuum effect and breeding dispersal Over the course of the year following the initial August 1, 1995, removal of the 19 core resident females, four originally peripheral resident females (W-J and W-Q, Fig. 4.7; E-D and E-P, Fig. 4.13) shifted the centers of their denning ranges into the core areas of the grids, and six more peripheral residents each expanded part of their range into the core (W-C, W-P and W-S, Fig. 4.7; E-F, E-L and E-S, Fig. 4.13). In the year following the August 1, 1996, removals (involving: W-J, W-Q and W-AE, Fig. 4.7; and E-D and E-P, Fig. 4.13), one peripheral resident female further expanded her denning range into the core (W-C, Fig. 4.9), and six other females, that had not previously done so, each expanded part of their range into the core (W-E, W-Pl, W-W and W-AC, Fig. 4.9; E-B and E-W, Fig. 4.15). In 36 of 48 cases (75 %), the core den sites occupied by the originally peripheral females, had been previously occupied by one of the 19 core resident females, removed on August 1, 1995. In nine.cases, the death or removal of an originally peripheral resident-female was followed by the occupation of all or part of her range, by either an adjacent, originally peripheral resident female (W-J vs. W-K, Figs. 4.4-7; E-I vs. E-H, Figs. 4.10-15; E-N vs. E-M, Figs. 4.13-14; E-Y vs. E-W, Figs. 4.10-13), or a secondary peripheral female (W-K vs. W-AM, Figs. 4.8-9; E-K vs. E-AD, Figs. 4.11-13; E-L vs. E-AE, Figs. 4.13-15; E-0 vs. E-AJ, Figs. 4.13-15; E-P vs. E-AC, Figs. 4.10-15). In 18 of 26 cases (69 %), the dens used by the "new" female, had been previously occupied by the female that had died or been removed. As described in the introduction (see also Chapter 3), in the absence of rigorous protocols, apparent cases of breeding dispersal by reproductively-active adult females may in fact be instances of range expansion involving the usurpation of all or part of a neighbour's territory, i.e., the vacuum effect. I conclude that there is abundant evidence of the vacuum effect given the results presented in the previous two paragraphs. This does not mean that the possibility of breeding dispersal is precluded perse. By my protocols, if any of the newly-identified adult females first caught after the August 1, 1995, removals (see Figs. 4.16 and 4.17), had entered and settled in the core of either of the two study grids, she would have been classified as a "true immigrant", and this would represent unambiguous evidence 76 of breeding dispersal. Four adult females first identified after August 1, 1995 (W-AC, W-AD, W-AF and W-AK; Figs. 4.6-9 and 4.16-7), were captured in the core of the West grid, but as I will discuss, none of them settled in the core. Consequently, there is no evidence of breeding dispersal in response to the experimental removals. Clout and Efford (1984) argued that movements by possums in excess of 2 km can be unambiguously classified as dispersal. None of the 78 radio-collared adult females in this study made movements of anywhere near this magnitude. The maximum recorded denning range length (the maximum distance between any two dens used by the same female) was 550 m (Fig. 4.23), which is comparable to the figure of 500 m cited by Clout and Efford (1984) as regards normal range lengths among possums in continuously forested habitats. The frequency of long-distance dispersal is often difficult to determine because it cannot be established whether animals that have disappeared have either died or emigrated (Stenseth and Lidicker 1992, Koenig et al. 1996, Ims and Yoccoz 1997). I can say with certainty that none of the 78 radio-collared adult females in this study dispersed because none of the radios were lost, and I was able to determine the fate (alive or dead) of all of the radio-collared females, up to the end of the study (Chapter 6). I conclude .that there is no evidence of breeding dispersal, based on Clout and Efford's (1984) criterion. Potential immigrants captured in the core Newly-identified females first caught anytime after the August 1, 1995, removals, represented potential immigrants (see Methods). In addition to the one "true immigrant", W-AE (Figs. 4.6 and 4.7), there were four potential immigrants, that were either originally (W-AD, W-AF and W-AK; Fig. 4.16) or eventually (W-AC, Fig. 4.9) captured in the core of the West grid. As noted above, these four females were all adults. In addition to the evidence against breeding dispersal by adult females (Efford 1991, see above), I will present other details regarding each of these four females that reinforces my conclusion that W-AE was the only "true immigrant". None of the 10 potential immigrants caught on the East grid (Figs. 4.16 and 4.17), were ever captured in the core of that grid (Figs. 4.10-4.15 ). W-AC was initially captured during the first two weeks of August, 1995, after the August 1 "deadline" for the complete enumeration of resident females. I decided to classify W-AC as a peripheral resident (along with W-AA and W-AB, Figs. 4.5 and 4.17), given that she was found to be bearing a pouch-young, 77 and had therefore been both, present on the grid, and reproductively-active during the 1995 breeding season (see Methods). Results from one of the two programs used to evaluate parentage from microsatellite DNA (Table 4.10), suggested that W-AC was most likely the mother or daughter of the adjacent peripheral resident female, W-I (Fig. 4.5), apparently confirming W-AC's status as an originally peripheral resident female. However, the DNA data cannot be taken as being conclusive, given the disagreement between the two programs (Table 4.10). If I had rigidly stuck to the August 1 definition of who is, or is not, a resident, W-AC would be classified as a potential immigrant, and the expansion of her denning range into the core (Fig. 4.9) might then constitute "true immigration". However, W-AC was only tracked to the core den site shown in Figure 4.9 on five occasions, and then returned to her most frequently used den (as shown in Fig. 4.7), which was located in the peripheral buffer strip surrounding the core of the grid. W-AC was found dead in late June, 1997, near the latter location. The arithmetic mean center of W-AC's denning range during the 1997 breeding season was located within the peripheral buffer strip, 26.8 m to the southwest of the innermost den used by female W-AM (Fig. 4.9). I conclude that W-AC's movements (Figs. 4.5, and 4.7-9) cannot be classified as "true immigration", under either interpretation as to her status. W-AD was initially captured in late November, 1995, at a core trap location (Figs. 4.6 and 4.16). When first trapped, W-AD had an enlarged pouch, and had clearly given birth on one or more occasions (Bolliger and Carrodus 1938, 1940), but she was not nursing and did not have a pouch-young. She was fitted with a radio-collar, and was thereafter repeatedly tracked to just two den sites, both of which were located in the peripheral buffer strip (Fig. 4.7). According to my definitions (see Methods), W-AD represented a potential immigrant that had settled in the peripheral buffer strip, and was therefore not a "true immigrant". Parentage analysis using microsatellite DNA (Table 4.10) indicated that W-AD was almost certainly (p < 0.01, Appendix 1) the mother or daughter of the adjacent peripheral resident female, W-P (Fig. 4.7). Consequently, I conclude that W-AD was most likely a peripheral resident female that had eluded capture until November, 1995. As previously discussed (see Methods), I am confident that, with the capture of female W-AD, I had achieved my goal of completely enumerating all of the resident females on both grids. W-AF was initially captured in late April, 1996, at a core trap location (Figs. 4.7 and 4.16). She was re-trapped on two occasions in the first two weeks of May, and was then never seen again. 78 Similarly, W-AK was initially captured in late March, 1997, at a core trap location (Figs. 4.9 and 4.16), was re-trapped the next night, and was then never seen again. Both females were adults with enlarged pouches, that had clearly given birth on one or more occasions (Bolliger and Carrodus 1938, 1940), but neither was found to be carrying a pouch-young, either on the first or last occasion on which she was captured. According to my definitions (see Methods), both W-AF and W-AK represented potential immigrants that were transients on the grid, as they did not have pouch-young and were present less than three months. No putative mother or daughter of W-AF was found among those females on the West grid for which DNA typing data were available (Table 4.10). Results from one of the two programs used to evaluate parentage from microsatellite DNA (Table 4.10), suggested that W-AK may have been the mother or daughter of female W-K (Fig. 4.8). However, given that the number of loci compared was low (Table 4.10), that there was a "mismatch" at one of the few loci compared, and that the calculated likelihood was low, I conclude that the putative relationship is almost certainly spurious (Appendix 1). While it is possible that both W-AF and W-AK stayed on the grid following their initial capture and thereafter managed to elude subsequent detection, I conclude that this is highly improbable. Arguably, W-AF and W-AK may have eluded subsequent detection because they became trap-averse. However, the three radio-collared females (out of the 77 considered, Fig. 4.21; see Methods) that I suspect had, in fact, become trap-averse, were each only trapped on a single occasion, suggesting that trap-aversion was a response to the stress of initial capture (Humphreys et al. 1984, Harlow et al. 1992). W-AF and W-AK, on the other hand, were each trapped more than once. W-AF was trapped on 3 of the 4 nights on which traps were set, during the period she was known to be present, and W-AK was trapped on 2 of the 3 nights on which traps were set, during the period she was known to be present. Even if W-AF and W-AK had become trap-averse, both had unique sets of coloured ear-tags, and would have been recognized if resighted during spotlighting (Tables 4.1 and 4.3), which they never were. Adverse effects of capture and handling and the measured rate of immigration It is possible that the two transient females discussed in the previous section, W-AF and W-AK (Figs. 4.7, 4.9 and 4.16), did not settle on the West grid because of an adverse reaction to the stress of capture and handling (Chapter 3). At the time she was first caught, W-AF only weighed 1750 g, which is 79 exceptionally low for an adult female (Chapter 5). In addition, W-AF demonstrated rapid weight loss in between the second to last, and last occasion on which she was trapped (when she weighed only 1600 g, Chapter 5), and was "weak and wobbly", upon release, when last seen. Females seen to be "weak and wobbly" upon release had a lower probability of survival (Chapter 6), and W-AF may have disappeared because she was dead. W-AK, on the other hand, was in good condition, and of above average weight (2250 g, Chapter 5), and it seems very unlikely that she disappeared because she was dead (Chapter 6). While W-AF was already in poor condition when she was first trapped, it is possible she may have died because she was repeatedly trapped and handled (Chapter 6). Moreover, while it seems unlikely, given that she was trapped more than once, it is at least conceivable that the stress of being captured and handled may have "driven off female W-AK. If the stress of capture and handling inhibited either W-AF or W-AK from settling on the West grid, this might cause me to underestimate the rate of female immigration. However, even if W-AF and W-AK had settled on the grid, there is no guarantee that they would have settled in the core, and if they settled on the periphery, there would be no effect on the estimated rate of female immigration. Notably, W-AF (Fig. 4.7) and W-AK (Fig. 4.9) both appeared in evident gaps in the otherwise contiguous ring of peripheral resident females. Given that they were adults, and that there was abundant evidence of the vacuum effect, and no evidence of breeding dispersal, I conclude that, W-AF and W-AK were most likely residents from just outside the area sampled by the grid, in the process of making "exploratory movements" (Lidicker and Stenseth 1992) into the peripheral buffer strip. While the eventual fate of W-AF and W-AK was uncertain, one potential immigrant, W-AI (Fig. 4.7), was known to have died as a direct result of capture and handling, and this also could have caused me to underestimate the rate of female immigration. However, I am confident that W-AI would not have entered and settled in the core of the grid, and thereby become a "true immigrant", if she had lived. W-AI died during her initial capture, in late July, 1996 (Fig. 4.17). She was a very large (2575 g, Chapter 5), very old (aged 14 years, as determined from cementum annuli; Chapter 5), adult female that was carrying a large pouch-young. No putative mother or daughter of W-AI was found among those females on the West grid for which DNA typing data were available (Table 4.10). However, a small (1900 g, Chapter 5), adult male, first caught in October, 1996, at a trap site adjacent to where W-AI had been captured, was almost certainly 80 (p < 0.01, Appendix 1) the son of W-AI. Based on her age, the presence of a pouch-young, and the location of her putative son, W-AI was assuredly a resident from just outside the area sampled by the grid. In 1997,1 found the remains of one unidentifiable young of the year (< 1 year old, as determined from cementum annuli; Chapter 5), on the East grid, in late summer (indicated by the "?" enclosed in a rectangle in Fig. 4.14), and another, on the West grid, in mid-autumn (Fig. 4.9). In both cases, I was unable to find the animal's ears or its abdomen, so I could not identify the individual, or its sex. There were no newly-identified young of the year females captured on either grid following the 1996 breeding season. Consequently, if the unidentifiable remains were those of potential immigrant females, these individuals did not die because of capture and handling. Since there were at least 17 resident offspring that disappeared following weaning in the spring of 1996 (Chapter 5), and the measured rate of female immigration was so low, I conclude that the remains in question were most likely those of either, resident offspring, or male immigrants. I conclude that adverse effects of capture and handling had the potential to disrupt female immigration in only three cases, and as all three cases involved older adult females, it is unlikely that any of the three were in fact an the process of dispersing. Consequently, there is little reason for supposing that the rate of female immigration was underestimated, as a result of adverse effects of capture and handling, on potential immigrants (Chapter 3). As discussed in Chapter 3 (see Fig. 3.2), adverse effects of capture and handling on resident reproduction or survival has the potential to cause immigration to be overestimated, if local recruitment or range expansion by adults normally preempts settlement by immigrants. Evidence presented in Chapter 5 suggests that the survival of the pouch-young of resident females may have been adversely affected by repeated capture and handling, and evidence presented in Chapter 6 indicates that the survival of the resident females themselves was almost certainly adversely affected by repeated capture and handling. Consequently, there are good reasons for supposing that the rate of female immigration was, if anything, overestimated. In Chapter 6,1 report that, over the course of the study, there may have been as many as 13 deaths among the 78 radio-collared adult females, that appeared to be either directly (2) of indirectly (11) associated with capture and handling. Such "unnatural" deaths are analogous to removals. Taking 81 these 13 "unintentional" removals together with the 19 intentional removals, it could be argued that I created a total of 32 adult home range vacancies. Considering both core and peripheral areas of both grids, I captured only two newly-identified subadult females, over the course of the two years following the August 1, 1995, removals. Other subadult females were captured, but these were all identifiable because they had been tagged while still with their mothers. One of the two newly-identified subadult females was the one "true immigrant" female, W-AE. The other was E-AH (Figs. 4.14, 4.15 and 4.17). E-AH was first trapped in September, 1996 (Fig. 4.17). When first captured (Fig. 4.14), she was a very small (1100 g, Chapter 5), subadult. She was later recaptured in mid-May, 1997, at which time she was found to be bearing a pouch-young. E-AH was fitted with a radio-collar on the latter occasion, and was thereafter tracked to a total of two den sites, both of which were also used by the adjacent adult female, E-AE (Figs. 4.13-15). Parentage analysis using microsatellite DNA (Table 4.11) indicated that E-AH was most probably (p < 0.05, Appendix 1) the daughter of E-AE. I conclude that female E-AH was not a disperser, but rather a philopatric female that settled and bred within her mother's home range. Assuming that breeding dispersal does not exist (or very rarely occurs; Efford 1991, see above), "true" immigration must only involve subadult females (as a consequence of natal dispersal). Again considering both core and peripheral areas of both grids, there may have been as many as 32 adult home range vacancies created as direct or indirect consequence of the experimental protocols. At the same time, only one subadult female, W-AE (Fig. 4.7), was known to have dispersed away from her mother's (W-AG, Fig. 4.7) home range, and immigrated into one of these 32 adult home range vacancies. Consequently, I conclude that the estimated rate of one female immigrant per 19 vacancies every two years, given earlier, is almost certainly an overestimate of the importance of female immigration in replacing losses among breeding females. Locations of known mothers and daughters There were a total of eight cases where daughters, first tagged while still with their mothers, were later recaptured, by themselves. In every case the daughters were recaptured at a location adjacent to their mother's denning range (Figs. 4.4-15). In 2 of the 8 cases (involving W-Pl and W-L2, Figs. 4.7 and 4.9), 82 the daughter, in turn, was known to have given birth. Female W-Pl was fitted with a radio-collar, and she and her mother, W-P, were generally found to be together in the same den tree, while both of them bore pouch-young, both in 1996 (Fig. 4.7) and in 1997 (Fig. 4.9). Female W-L2 was recaptured at a location adjacent to her mother's (W-L) denning range (Fig. 4.9), on the last day of the study, in July, 1997, and was found to be bearing a pouch-young. Locations of mothers and daughters identified using microsatellite DNA Putative mother-daughter relationships among the adult (and subadult) females on each grid, based on microsatellite DNA, are reported in Tables 4.10 and 4.11. Putative mother-daughter pairs judged to have adjacent ranges are shown in bold letters in Tables 4.10 and 4.11 for comparison with the appropriate range maps (Figs 4.4-15). If females generally settle adjacent to their mothers, putative mother-daughter relationships involving adjacent females are most likely true mother-daughter relationships while those involving non-adjacent females are most likely "false positives". Parentage analysis may generate "false positives" when the true parent has been falsely excluded, as a result of typing errors, or when individuals are compared at only a few loci and the probability of exclusion is correspondingly low (Marshall et al. 1998). In both cases, likelihood scores associated with "false positives" will generally be lower than likelihood scores identifying true parent-offspring relationships. It follows that likelihood scores among adjacent putative mother-daughter pairs ought to be substantially greater than among non-adjacent putative mother-daughter pairs (Ishibashi et al. 1997). I conducted a series of t-tests comparing likelihood scores between adjacent and non-adjacent putative mother-daughter pairs. If a given relationship was reciprocal (e.g., A was the most likely mother/daughter of B, and B was the most likely mother/daughter of A), the likelihood score was only included once in the statistical comparisons. In addition, the lone immigrant female, W-AE, was not included, as her status (as an immigrant) was already established. Adjacent putative mother-daughter pairs had significantly higher likelihood scores than non-adjacent pairs, on both grids, and regardless of which program (KINSHIP or CERVUS) was used to generate the likelihood scores (Table 4.12). I conclude that adjacent putative mother-daughter pairs are most likely true mothers and daughters, and because the same inference was arrived at using both program KINSHIP and program CERVUS, I also 83 conclude that the result is robust with respect to assumptions regarding deviations from Hardy-Weinberg equilibrium, and the treatment of typing errors (Appendix 1). Given that the adjacent putative mother-daughter pairs identified in Tables 4.10 and 4.11 are most likely true mothers and daughters, 14 (CERVUS) or 15 (KINSHIP) of the 35 females under consideration on the West grid (40-43 %), and 11 (KINSHIP) or 13 (CERVUS) of the 33 females under consideration on the East grid (33-39 %), were philopatric. If all females are philopatric, and all females everywhere are sampled, then the mother or daughter of a given female would always (100 % of the time) be found to be one of the several females occupying the adjacent home ranges surrounding the home range of the female under consideration. In the present case, not all females were sampled. Because tissue sampling did not begin until early October, 1995, which was after the core females had been removed, what I sampled was a more or less contiguous ring of females. Figures 4.5 and 4.11 illustrate the configuration of females at about the time I began collecting tissue samples. Since what I sampled was a ring, or circular chain of females, I generally only sampled 2 females (from either side), from among the likely 4 to 6 females (depending on home range shape), that had home ranges adjacent to a given female. Consequently, even if almost all females are philopatric, I would only expect to find adjacent mothers and daughters in about one-third (sampling 2 of 6 adjacent females = 33 %) to one-half (sampling 2 of 4 adjacent females = 50 %) of the cases I examined, given the manner in which females were sampled. I conclude that the observed frequency with which females were found to be adjacent to their likely mother or daughter (33-43 %) is consistent with the expected frequency (33-50 %), given the hypothesis that almost all female possums at the Paddys Land site settle on or adjacent to their mother's home range. Direct observations of restricted movements between the grids There was no evidence of long-distance dispersal (> 2 km) between the grids, as animals tagged on one grid were never recaptured on the other grid. However, one male was known to have undertaken a long-distance "excursion" (or "exploratory movement"), within the area between the grids. The male in question was regularly caught in the same trap in the southeast sector of the West grid (as indicated by the male symbol in Fig. 4.8), from December, 1994, onwards. In October, 1996, he was caught at vehicle trail 84 trap-site Vt6 (Fig. 4.1), some 1263.0 m from his normal location. When he was next recaptured, in February, 1997, he was back in his usual trap. Only two other animals were caught at vehicle trail trap-sites in Oct.-Nov., 1996, that were normally caught on the grids. Female W-B (Fig. 4.7) was trapped at vehicle trail trap-site Vtl6 (Fig. 4.1), and female E-AI (Fig. 4.15) was trapped at vehicle trail trap-site Vtl (Fig. 4.1). In these two cases the vehicle trail trap-sites were 138.1 and 163.2 m, respectively, from the normal positions at which these females were trapped on the grids. Restricted movements between the grids identified using microsatellite DNA Allele frequencies differed significantly between the two study grids (Table 4.9). There were significant differences as regards each of 3 of the 6 loci considered, and the combined probabilities test across all loci was highly significant. However, the magnitude of genetic subdivision between the grids was slight (F S T = 0.0062, R J I O S T = 0.0078), accounting for less than 1 % of the total genetic variation among individuals (Hard 1988). Restricted "migration" between the grids is consistent with, but not the only possible explanation for, the observed allele frequency differences (Hartl 1988). .-. Sex ratios of pouch-young Strong female philopatry is expected to select for a significantly male-biased primary sex ratio, according to the "local resource competition" (LRC) hypothesis (Johnson 1988, 1989). The LRC hypothesis assumes that there is male-biased dispersal in the species, and that males and females are equally "costly" to produce. As discussed previously (see Methods), possums are known to demonstrate male-biased dispersal, and this also appears to be the case at the Paddys Land site (comparing Figs. 4.16 vs. 4.18, and 4.17 vs. 4.19). There was no evidence that male and female pouch-young developed at different rates, or weaned at different weights at the Paddys Land site (Chapter 5), nor has this been reported in the literature (Lyne and Verhagen 1957, Kerle and Howe 1992). Kerle and Howe (1992) measured eight morphometries and concluded that adult body weight showed the most significant level of sexual dimorphism, with males being heavier than females. Crawley (1973) also reported that adult males were significantly heavier than 85 adult females, when data were pooled across seasons, but differences were more apparent in some seasons (summer and autumn) than others (winter and spring). In contrast to both Crawley (1973), and Kerle and Howe (1992), Cowan (1990) concluded that there was little evidence of sexual size dimorphism in the species, based on a review of nine separate studies, with adult female body weight being on average 98.6 % of that of adult male body weight. Similarly, I found few noteworthy differences in adult body weights between the sexes at the Paddys Land site (Chapter 5). I conclude that there is little evidence that either sex is more "costly" to produce than the other. Table 4.13 shows the total number of male and female pouch-young recorded in this study, as well as that recorded in all available published studies, that have reported on the breeding biology of common brushtail possums. In each of the 10 studies considered in Table 4.13, the reported number of male pouch-young was greater than the number of female pouch-young. The numbers reported in three studies, including this one, were significantly different from parity. While the numbers reported in the remaining seven studies were not significantly different from parity, there was no significant difference in the ratios reported across all 10 studies (Table 4.13). Considering the results from all 10 studies, taken together, a significantly male-biased sex ratio of pouch-young is clearly the norm for the species (Table 4.13). Since males and females are equally "costly" to produce, I conclude that the strongly male-biased sex ratio of pouch-young at both the Paddys Land site and across all 10 studies, can be interpreted as indicating that most females do not disperse, but instead settle on or adjacent to their mother's home range, both at the Paddys Land site, and in most other populations. DISCUSSION Only one female immigrant (W-AE, Fig. 4.7) entered and settled in the core of one of the two study grids, over the course of the two years following the removal of 19 core resident females. I conclude that female immigration was almost non-existent, and having dealt with the issue of complete enumeration at length, I feel confident in saying that my measurement of female immigration was, as nearly as possible, made without error. If the strict definitions and protocols followed in the design and interpretation of the experiment are relaxed, it would appear that there were as many as 32 adult female "removals", at the 86 same time as there was still only the one female immigrant. In this case, the estimated rate of one female immigrant per 19 vacancies every two years, is almost certainly an overestimate of the importance of female immigration in replacing losses among breeding females. Critics may claim that my experiment was too "short-term", and the near absence of female immigration was just the result of "a couple of bad years". The survival of young to the onset of weaning (175 days of age) was indeed very poor (33 %) in 1995, apparently as a result of very low rainfall during autumn and winter (Chapter 5), and the absence of immigration by female young of the year (born in 1995) in 1996 (W-AE was born in 1994) could be explained as a result of there being too few surviving young in the region as a whole, for there to be all that many dispersers. However, 1996 was a very wet year, adults measured in both years (1995 and 1996) were significantly heavier in 1996, and survival of young to the onset of weaning was significantly better in 1996 (81 %) than in 1995 (Chapter 5). Yet, there was still no immigration by female young of the year (born in 1996), in 1997. Efford (1991) concluded from a review of all available studies that most (78 %) dispersers are young of the year. Nonetheless, it is conceivable that young animals at the Paddys Land site generally delay dispersal until they are yearlings (1-2 years old), in which case there may not have been an influx of young born in 1996 until 1998, after the end of the study. Three of the four subadult males (75 %) first captured in the core of one of the two study grids, following the August 1, 1995 removals (Fig. 4.18), were young of the year (from both 1995 and 1996), which suggests that the typical age of dispersers at the Paddys Land site is similar to that at other sites. Efford (1991) found no significant difference between males and females with respect to age at dispersal, and it is unlikely, on theoretical grounds, that subadult females would disperse at a later age than subadult males, as they have "more to lose" from delaying breeding, than males do (Johnson 1986). Nonetheless, as the only female immigrant I observed (W-AE), was a yearling, I cannot rule out the possibility that there may have been an influx of yearling females (born in 1996) in 1998, after the end of the study. Female immigration may have been next to non-existent over the two years of the experiment because these were "a couple of bad years", or because female immigration is next to non-existent in most years, because most females do not disperse, but instead settle on or adjacent to their mother's home range. Three lines of evidence indicate that strong female philopatry is clearly the norm at the Paddys Land site. Firstly, in the two cases where daughters tagged as pouch-young were later known to have given birth, 87 the daughter either shared a den with her mother (W-Pl, Figs. 4.7 and 4.9), or was captured at a location adjacent to her mother's denning range (W-L2, Fig. 4.9). Secondly, parentage analysis using microsatellite DNA (Table 4.10 and 4.11) indicated that: 1) adjacent putative mothers and daughters were significantly more likely to be "true" mothers and daughters than non-adjacent putative mothers and daughters (Table 4,12); and 2) the observed frequency with which adjacent pairs of females were putatively identified as being mother and daughter (33-43 %), was close to that expected (33-50 %), if daughters almost invariably settle on or adjacent to their mother's home range. Thirdly, the significantly male-biased sex ratio of pouch-young at the Paddys Land site (Table 4.13) is consistent with the predictions of the "local resource competition" hypothesis, again suggesting that most females settle on or adjacent to their mother's home range. I conclude that strong female philopatry is the norm at the Paddys Land site, and infer that the near absence of female immigration over the two years of the experiment is representative of what would be found in most years, and was not merely the product of "a couple of bad years". Moreover, the fact that a significantly male-biased primary sex ratio appears to be the norm for the species (Table 4.13), suggests that strong female philopatry, and the resulting near absence of female immigration, will also be found to be the norm in most other populations of common brushtail possums. I have suggested, based on the results of the experiment, the location of known mothers and daughters, the location of putative mothers and daughters identified using microsatellite DNA, and the sex ratio of pouch-young, that female immigration will generally be "almost" non-existent. But I did of course observe an instance of female immigration. Perhaps the addition of a single subadult female, every second year, is somehow sufficient to "rescue" a population from extinction. Qualitatively, the one female immigrant, W-AE, did not "replace" any of the reproductively-active, core resident females, because she did not herself become reproductively-active (Clinchy 1997). Quantitatively, W-AE's probability of surviving and breeding in the following year was > 80 % (Chapters 5 and 6). In the latter case, it could be argued that the results of the experiment confirm that immigration has the potential to "rescue" recipient populations from extinction, provided that: 1) the populations are no more than one home range apart, as this was the extent of the distance traveled by W-AE; and 2) the populations are completely interconnected by "uniformly suitable habitat", as was the case at the Paddys Land site. Assuming that physical barriers attenuate dispersal, I conclude that the results of the experiment indicate that immigration will rarely ever "rescue" 88 common brushtail possum populations from extinction, in cases where the populations are separated by larger distances (more than one home range length), and/or "uniformly unsuitable habitat". I chose to conduct the experiment in a "uniformly suitable habitat" in view of the fact that metapopulation theory predicts that the maximum rate of exchange of immigrants ought to be observed where there are no physical barriers to dispersal. However, dispersal might also be impeded by social barriers. Hestbeck (1982, p. 157) proposed the "social fence" hypothesis, whereby emigration will be lower at higher densities, because of the increasing "social impermeability of neighbouring groups". Female possums at the Paddys Land site demonstrated evidence of territoriality, insofar as they generally occupied exclusive, non-overlapping denning ranges (Green 1984), and many of them expanded their ranges in response to the removal of neighbouring, core resident females (i.e., the vacuum effect; Carpenter 1987, Mares and Lacher 1987, Schoener 1987). Nonetheless, I suggest that it is unlikely that territoriality acts as a barrier to emigration in possums, although it may act as a barrier to immigration. Hestbeck (1982, p. 158) suggested that where a social fence exists, one would not expect to see individuals making "brief exploratory jaunts into the neighbouring areas". I recorded one such "exploratory jaunt" during the vehicle trail trapping session, and I interpret the appearance and , disappearance of females W-AF (Fig. 4.7) and W-AK (Fig. 4.9), and the ever-increasing cumulative number of newly-identified males caught on the periphery of the grids (Fig. 4.19), as further evidence of other such "exploratory jaunts". Moreover, the fact that several new males, and a new female (W-AE), were caught in the cores of the grids, following the August 1, 1995 removals (Figs. 4.16 and 4.18), suggests that the contiguous ring of peripheral resident females (Figs. 4.5 and 4.11) was quite permeable. Similarly, none of the females tagged as pouch-young that were later caught by themselves apparently dispersed into the core of either grid, despite the absence of core resident females. Female W-Pl (Fig. 4.9), did show some expansion of her range into the core, but this was almost two years after the initial removals, and even at that she was still generally (80 % of the time) found in the same den as her mother, and continued to be found there up to the end of the study. The significant difference in allelic frequencies between the grids (Table 4.9) suggests that travel between the grids was unusual, and I did not observe any instances of animals from one grid showing up on the other grid, but the very low level of genetic subdivision (< 1 %) indicates that there was no significant barrier to gene flow between the grids 89 (Hartl 1988). Finally, it is difficult to see how a territory owner could patrol an area of one or two hectares (being the approximate size of most denning ranges), on a sufficiently frequent basis to actually prevent travel through the territory. As I will discuss, suitable den sites appear to be an important resource, but even in this respect, each female's denning range generally contained several den sites and it is unlikely she could adequately patrol all of them, so as to prevent travel through the territory. As the "typical" female's denning range was about 1 hectare in size, it likely encompassed about 150 trees (see Methods), of which 10 % might be used as den sites, and several more might be used as refuge sites. Yet, most females were tracked to just a single den tree about half of the time (Fig. 4.24). Moreover, where females expanded their denning ranges into a previously occupied range (the vacuum effect), they almost always used the same den(s) as had been used by the previous occupant. I infer from this that while there is clearly a surplus of "shelter sites", there is evidently competition for the few "prime" den sites (see also Caley et al. 1998). There may be a barrier to immigration insofar as dispersers will not settle, but instead continue moving, if they find that the prime den site in an area is occupied by a resident animal (Chapter 3). Alternatively, if immigrants enter an area and occupy a prime den site before the offspring of a resident is able to do so, this may limit local recruitment (Clinchy 1997).,, Given that males and females presumably compete for the same prime den sites (Caley et al. 1998), the addition of more males to an area will result in increased competition for den sites. On four occasions, I successfully followed one of the newly-identified males first caught in a core trap, after August 1, 1995 (Fig. 4.18), to a den tree, and in 3 of 4 cases, it was a den that had been previously occupied by one of the core resident females removed on August 1, 1995. Since female immigration was found to be next to non-existent, the effects of (male) immigration on the demography of recipient populations will most likely be negative (e.g., Mansergh and Scotts 1989; Clinchy 1997), and I suggest that this will largely be a consequence of adverse effects on local recruitment resulting from increased competition for prime den sites. My conclusion is that female immigration is generally next to non-existent at the Paddys Land site, despite there being no apparent physical or social barriers to dispersal (emigration), and I infer that immigration is likely to play an even smaller part in "rescuing" populations from extinction, where there are physical barriers to dispersal, as in fragmented or patchy landscapes. Logically, female dispersal must play some part in the ecology of common brushtail possums, otherwise the species could 90 not have come to occupy 94 % of Australia (How and Kerle 1995). The one female immigrant, W-AE, dispersed about 200 m. If female possums spread from a point of origin at a rate of 200 m every two years, it would take roughly 40,000 years for them to distribute themselves from one side of Australia to the other. I propose this simply as a ballpark figure indicating order of magnitude. My point is that an infinitesimally slow rate of expansion is not contradicted by a wide geographic range given infinite time, and 40,000 years is not all that long in the history of most mammalian species (Simpson 1983). It may be objected that it should have taken 14,000 years for possums to colonize all of New Zealand, at a rate of expansion of 200 m every 2 years, whereas it only took about 140 years. However, the current distribution of possums in New Zealand is largely the result of "assisted dispersal" (Cowan 1990, p. 76). Possums were introduced to New Zealand beginning in the 1830's as part of a government policy to establish a fur industry. Numerous introductions from Australia were followed by hundreds of translocations of New Zealand-bred progeny, particularly between 1890-1940. Possums were even introduced to 17 offshore islands. Although government policy shifted towards eradication in the 1940's, the possum fur industry still exists and illegal translocations by trappers still occur (Cowan 1990). Given the similarities between common brushtail possums and other species of medium-sized, herbivorous, marsupials, I suggest that immigration does not often "rescue" populations of any of these species from extinction. Like common brushtail possums, many of the other medium-sized, herbivorous, marsupials that have declined had very large geographic ranges prior to European settlement, and also like common brushtail possums, I suggest it probably took tens of thousands of years to establish such broad distributions. All of these species demonstrate a pattern of life history involving low fecundity and high survivorship (Chapters 5 and 6) generally considered to be characteristic of larger mammals (Lee and Cockburn 1985, Sinclair 1996), particularly large ungulates (Gaillard et al. 1998). Whatever the reasons were for the evolution of this type of life history, it is clearly not one suited to being a "good" colonizer (MacArthur and Wilson 1967, p.83). Since life history theory would predict that medium-sized, herbivorous, marsupials will be "bad" colonizers, and the results of my study appear to confirm this, how is it that these species feature prominently in the literature on connectivity and conservation corridors (Laurance and Gascon 1997, Beier and Noss 1998), and that the disruption of dispersal should be suggested as a key factor in their 91 decline (Morton 1990)? I suggest that the "intuitively obvious" consequences of metapopulation models may often distract attention from the nature and details of dispersal in real animals. This raises the related question of just what is being measured in studies of connectivity and conservation corridors, such as the one by Laurance (1990), and those cited by Beier and Noss (1998), that have focussed on Australian marsupial possums and gliders? In the absence of female immigration, animals detected in corridors during presence/absence surveys of the type conducted by Laurance (1990), must either be: 1) individuals that are resident in the corridor (Beier and Noss 1998); or 2) males that are dispersing through the corridor. In already fragmented landscapes, the corridor will likely be one of only a few remaining pieces of remnant habitat, and the fact that it might sustain a resident population ought to be justification enough for its retention. However, there is no basis for suggesting that it is possible to somehow "creatively fragment" (Laurance and Gascon 1997) a continuous, natural landscape (such as that at the Paddys Land site), and still maintain connectivity through the retention of movement corridors, because the only individuals that are likely to move through the corridors will all be males, and these are as likely as not to have adverse effects on the populations they move into (e.g., Mansergh and Scotts 1989; Clinchy 1997). 92 REFERENCES Alpin, K. P., and Archer, M. 1987. Recent advances in marsupial systematics with a new syncretic classification. In Possums and opossums: studies in evolution. Edited by Archer, M. Surrey Beatty and Sons Limited, Sydney, Australia, pp. xv-lxxii. Beier, P. 1993. Determining minimum habitat areas and habitat corridors for cougars. Conservation Biology, 7: 94-108. Beier. P., and Noss, R. F. 1998. Do habitat corridors provide connectivity? Conservation Biology, 12: 1241-1252. Bolliger, A., and Carrodus, A. L. 1938. Changes in and around the pouch of Trichosurus vulpecula, as occurring naturally and as the result of the administration of oestrone. Journal and Proceedings of the Royal Society of New South Wales, 71: 615-622. Bolliger, A., and Carrodus, A. L. 1940. The effects of oestrogens on the pouch of the marsupial Trichosurus vulpecula. Journal and Proceedings of the Royal Society of New South Wales, 73: 218-227. Boutin, S. 1980. Effect of spring removal experiments on the spacing behaviour of female snowshoe hares. Canadian Journal of Zoology, 58: 2167-2174. Brown, J. H., and Kodric-Brown, A. 1977. Turnover rates in insular biogeography: effects of immigration on extinction. Ecology, 58: 445-449. Burbidge, A. A., and McKenzie, N. L. 1989. Patterns in the modern decline of Western Australia's vertebrate fauna: causes and conservation implications. Biological Conservation, 50: 143-198. Caley, P., Spencer, N. J., Cole, R. A., and Efford, M. G. 1998. The effect of manipulating population density on the probability of den-sharing among common brushtail possums, and the implications for transmission of bovine tuberculosis. Wildlife Research, 25: 383-392. Carpenter, F. L. 1987. The study of territoriality: complexities and future directions. American Zoologist, 27: 401-409. Caughley, G., and Gunn, A. 1996. Conservation biology in theory and practice. Blackwell Science, Cambridge, Massachusetts. 93 Caughley, G., and Kean, R. I. 1964. Sex ratios in marsupial pouch young. Nature (London), 204: 491. Chapman, B., and Binns, D. 1995. Flora survey of the Walcha/Nundle and Styx River Management Areas. Walcha/Nundle and Styx River Management Areas EIS, Supporting Document No. 2. Forestry Commission of New South Wales, Sydney, Australia. Clark, J. M. 1995. The diet of the parasitic mite Trichosurolaelaps crassipes Womersley, 1956, and its potential to act a disease vector. New Zealand Veterinary Journal, 43: 215-218. Clinchy, M. 1997. Does immigration "rescue" populations from extinction? Implications regarding movement corridors and the conservation of mammals. Oikos, 80: 618-622. Clout, M. N., and Efford, M. G. 1984. Sex differences in the dispersal and settlement of brushtail possums {Trichosurus vulpecula). Journal of Animal Ecology, 53: 737-749. Coleman, J. D., and Green, W. Q. 1984. Variations in the sex and age distributions of brush-tailed possum populations. New Zealand Journal of Zoology, 11: 313-318. Cowan, P. E. 1990. Brushtail possum. In The handbook of New Zealand mammals. Edited by King, C. M. Oxford University Press, Auckland, New Zealand, pp. 68-98. Crawley, M. C. 1973. A live-trapping study of Australian brush-tailed possums, Trichosurus vulpecula (Kerr), in the Orongorongo Valley, Wellington, New Zealand. Australian Journal of Zoology, 21: 75-90. Croft, D. B. 1989. Social organization of the Macropodoidea. In Kangaroos, wallabies and rat-kangaroos. Edited by Grigg, G., Jarman, P., and Hume, I. Surrey Beatty and Sons Limited, Sydney, Australia, pp. 505-625. Dunnet, G. M. 1956. A live-trapping study of the brush-tailed possum Trichosurus vulpecula Kerr (Marsupialia). CSIRO Wildlife Research, 1: 1-18. Dunnet, G. M. 1964. A field study of local populations of the brush-tailed possum Trichosurus vulpecula in eastern Australia. Proceedings of the Zoological Society of London, 142: 665-695. Efford, M. 1991. A review of possum dispersal. DSIR Land Resources Contract Report 91/73, DSIR Land Resources, Dunedin, New Zealand. Efford, M. 1998. Demographic consequences of sex-biased dispersal in a population of brushtail possums. Journal of Animal Ecology, 67: 503-517. 94 Finlayson, H. H. 1961. On central Australian mammals. Part IV. The distribution and status of central Australian species. Records of the South Australian Museum, 14: 141-191. Forestry Commission of N.S.W. 1989. Research Note No. 17. Forest types in N.S.W. Forestry Commission of N.S.W., Sydney, Australia. Gaillard, J.-M., Festa-Bianchet, M., and Yoccoz, N. G. 1998. Population dynamics of large herbivores: variable recruitment with constant adult survival. Trends in Ecology and Evolution, 13: 58-63. Goodnight, K. F., and Queller, D. C. 1998. KINSHIP Version 1.2. Goodnight software, Rice University, Houston, Texas, U.S.A. Available at [http://www.bioc.rice.edu/~kfg/Gsoft.html]. Green, W. Q. 1984. A review of ecological studies relevant to the management of the common brushtail possum. In Possums and gliders. Edited by Smith, A. P., and Hume, I. D. Australian Mammal Society, Sydney, Australia, pp. 483-499. Green, W. Q., and Coleman, J. D. 1986. Movement of possums (Trichosurus vulpecula) between forest and pasture in Westland, New Zealand: implications for Bovine Tuberculosis transmission. New Zealand Journal of Ecology, 9: 57-69. Greenwood, P. J. 1980. Mating systems, philopatry and dispersal in birds and mammals. Animal Behaviour, 28: 1140-1162. Hanski, I. 1985. Single-species spatial dynamics may contribute to long-term rarity and commonness. Ecology, 66: 335-343. Hanski, I. 1991. Single-species metapopulation dynamics: concepts, models and observations. Biological Journal of the Linnean Society, 42: 17-38. Hanski, I. 1998. Metapopulation dynamics. Nature (London), 396: 41-49. Hanski, I., and Gilpin, M. 1991. Metapopulation dynamics: brief history and conceptual domain. Biological Journal of the Linnean Society, 42: 3-16. Harlow, H. J., Lindzey, F. G., Van Sickle, W. D., and Gern, W. A. 1992. Stress response of cougars to nonlethal pursuit by hunters. Canadian Journal of Zoology, 70: 136-139. Hartl, D. L. 1988. A primer of population genetics, 2nd ed. Sinauer Associates, Inc., Sunderland, Massachusetts. 95 Hemsley, S., and Canfield, P. 1993. Dermatitis in brushtail possums (Trichosurus vulpecula). Proceedings of the Annual Meeting of the Australasian section of the Wildlife Disease Association, Wildlife Disease Association, Sydney, Australia, pp. 23-24. Hestbeck, J. B. 1982. Population regulation of cyclic mammals: the social fence hypothesis. Oikos, 39: 157-163. Hintze, J. 1996. NCSS Statistical Software. NCSS, Kaysville, Utah, U.S.A. Information available at [http://www.ncss.com]. Hope, R. M. 1972. Observations on the sex ratio and the position of the lactating mammary gland in the brush-tailed possum, Trichosurus vulpecula (Kerr) (Marsupialia). Australian Journal of Zoology, 20: 131-137. How, R. A., and Kerle, J. A. 1995. Common brushtail possum. In The mammals of Australia. Edited by Strahan, R. Reed Books, Sydney, Australia, pp. 273-275. Humphreys, W. F., Bradley, A. J., How, R. A., and Barnett, J. L. 1984. Indices of condition of Phalanger populations: a review. In Possums and gliders. Edited by Smith, A. P., and Hume, I. D. Australian Mammal Society, Sydney, Australia, pp. 59-77. , Ims, R. A., and Yoccoz, N. G. 1997. Studying transfer processes in metapopulations: emigration, migration, and colonization. In Metapopulation biology: ecology, genetics, and evolution. Edited by Hanski, I., and Gilpin, M. E., Academic Press, San Diego, pp. 247-265. Ishibashi, Y., Saitoh, T., Abe, S., and Yoshida, M. C. 1997. Sex-related spatial kin structure in a spring population of grey-sided voles Clethrionomys rufocanus as revealed by mitochondrial and microsatellite DNA analyses. Molecular Ecology, 6: 63-71. Johnson, C. N. 1986. Sex-biased philopatry and dispersal in mammals. Oecologia, 69: 626-627. Johnson, C. N. 1988. Dispersal and sex ratio at birth in primates. Nature (London), 332: 726-728. Johnson, C. N. 1989. Dispersal and philopatry in the Macropodoids. In Kangaroos, wallabies and rat-kangaroos. Edited by Grigg, G, Jarman, P, and Hume, I. Surrey Beatty and Sons Pty Limited, Sydney, Australia, pp. 593-601. Johnson, M. L., and Gaines, M. S. 1990. Evolution of dispersal: theoretical models and empirical tests using birds and mammals. Annual Review of Ecology and Systematics, 21: 449-480. 96 Kerle, J. A. 1984. Variation in the ecology of Trichosurus: its adaptive significance. In Possums and gliders. Edited by A. P. Smith and I. D. Hume. Australian Mammal Society, Sydney, Australia, pp. 115-127. Kerle, J. A., Foulkes, J. N., Kimber, R. G., and Papenfus, D. 1992. The decline of the brushtail possum, Trichosurus vulpecula (Kerr 1778), in arid Australia. Rangelands Journal, 14: 107-127. Kerle, J. A., and Howe, C. J. 1992. The breeding biology of a tropical possum, Trichosurus vulpecula arnhemensis (Phalangeridae : Marsupialia). Australian Journal of Zoology, 40: 663-665. Kerle, J. A., McKay, G. M., and Sharman, G. B. 1991. A systematic analysis of the brushtail possum, Trichosurus vulpecula (Kerr, 1792) (Marsupiala : Phalangeridae). Australian Journal of Zoology, 39: 313-331. Koenig, W. D., Van Vuren, D., and Hooge, P. N. 1996. Detectability, philopatry, and the distribution of dispersal distance in vertebrates. Trends in Ecology and Evolution, 11: 514-517. Krebs, C. J. 1999. Ecological methodology, 2nd ed. Benjamin/Cummings, Menlo Park, California. Lambin, X., and Krebs, C. J. 1993. Influence of female relatedness on the demography of Townsend's vole populations in spring. Journal of Animal Ecology, 62: 536-550. Laurance, W. F. 1990. Comparative responses of five arboreal marsupials to tropical forest fragmentation. Journal of Mammalogy, 71: 641-653. Laurance, W. F., and Gascon, C. 1997. How to creatively fragment a landscape. Conservation Biology, 11: 577-579. Lee, A. K., and Cockburn, A. 1985. Evolutionary ecology of marsupials. Cambridge University Press, Cambridge. Lee, A. K., and Ward, S. J. 1989. Life histories of macropodoid marsupials. In Kangaroos, wallabies and rat-kangaroos. Edited by Grigg, G., Jarman, P., and Hume, I. Surrey Beatty and Sons, Sydney, Australia, pp. 105-115. Lidicker, W. Z., Jr., and Koenig, W. D. 1996. Responses of terrestrial vertebrates to habitat edges and corridors. In Metapopulations and wildlife conservation. Edited by McCullough, D. R., Island Press, Washington, D.C, pp. 85-109. 97 Lidicker, W. Z., Jr., and Stenseth, N. C. 1992. To disperse or not to disperse: who does it and why? In Animal dispersal: small mammals as a model. Edited by Stenseth, N. C , and Lidicker, W. Z., Jr., Chapman and Hall, London, pp. 21-36. Lindenmayer, D. B., Welsh, A., Donnelly, C. F., and Meggs, R. A. 1996. Use of nest trees by the mountain brushtail possum (Trichosurus caninus) (Phalangeridae : Marsupialia). I. Number of occupied trees and frequency of tree use. Wildlife Research, 23: 343-361. Lindenmayer, D. B., Welsh, A., and Donnelly, C. F. 1997. Use of nest trees by the mountain brushtail possum (Trichosurus caninus) (Phalangeridae : Marsupialia). III. Spatial configuration and co-occupancy of nest trees. Wildlife Research, 24: 661-677. Lyne, A. G., and Verhagen, A. M. W. 1957. Growth of the marsupial Trichosurus vulpecula and a comparison with some higher mammals. Growth, 21: 167-195. MacArthur, R. H., and Wilson, E. O. 1967. The theory of island biogeography. Princeton University Press, Princeton, New Jersey. Mackintosh, C. G., Crawford, J. L., Thompson, E. G., McLeod, B. J., Gill, J. M., and O'Keefe, J. S. 1995. A newly discovered disease of the brushtail possum: Wobbly possum syndrome. New Zealand Veterinary Journal, 43: 126. Mansergh, I. M., and Scotts, D. J. 1989. Habitat continuity and social organization of the mountain pygmy-possum restored by tunnel. Journal of Wildlife Management, 53: 701-707. Mares, M. A., and Lacher, T. E., Jr. 1987. Social spacing in small mammals: patterns of individual variation. American Zoologist, 27: 293-306. Marshall, T. C. 1998. CERVUS Version 1.0. University of Edinburgh, Edinburgh, Scotland. Available at [http://helios.bto.ed.ac.uk/evolgen]. Marshall, T. C , Slate, J., Kruuk, L. E. B., and Pemberton, J. M. 1998. Statistical confidence for likelihood-based paternity inference in natural populations. Molecular Ecology, 7: 639-655. Moilanen, A., Smith, A.T., and Hanski, I. 1998. Long-term dynamics in a metapopulation of the American pika. American Naturalist, 152: 530-542. Morton, S. R. 1990. The impact of European settlement on the vertebrate animals of arid Australia: a conceptual model. Proceedings of the Ecological Society of Australia, 16: 201-213. 98 Noss, R. F. 1987. Corridors in real landscapes: a reply to Simberloff and Cox. Conservation Biology, 1: 159-164. NPWS. 1992. Guy Fawkes River Wilderness assessment report. N.S.W. National Parks and Wildlife Service, Sydney, Australia. NPWS. 1994. Fauna of northeast N.S.W. forests. Northeast forests biodiversity study. N.S.W. National Parks and Wildlife Service, Sydney, Australia. Paetkau, D., Calvert, W., Stirling, I., and Strobeck, C. 1995. Microsatellite analysis of population structure in Canadian polar bears. Molecular Ecology, 4: 347-354. Piertney, S. B., MacColl, A. D. C , Bacon, P. J., and Dallas, J. F. 1998. Local genetic structure in red grouse (Lagopus lagopus scoticus): evidence from microsatellite DNA markers. Molecular Ecology, 7: 1645-1654. Presidente, P. J. A. 1984. Parasites and diseases of brushtail possums (Trichosurus spp): occurrence and significance. In Possums and gliders. Edited by A. P. Smith and I. D. Hume. Australian Mammal Society, Sydney, pp. 171-190. Raymond, M., and Rousset, F. 1995...GENEPOP (Version 1.2): population genetics software for exact tests and ecumenicism. Journal of Heredity, 86: 248-249. Available at [http://www.cefe.cnrs-mop.fr]. SanCristobal, M., and Chevalet, C. 1997. Error tolerant parent identification from a finite set of individuals. Genetical Research, 70: 53-62. Schoener, T. W. 1987. Time budgets and territory size: some simultaneous optimization models for energy maximizers. American Zoologist, 27: 259-291. Seebeck, J. H., Bennett, A. F., and Scotts, D. J. 1989. Ecology of the Potoroidae: a review. In Kangaroos, wallabies and rat-kangaroos. Edited by Grigg, G., Jarman, P., and Hume, I. Surrey Beatty and Sons, Sydney, Australia, pp. 67-88. Short, J., and Smith, A. 1994. Mammal decline and recovery in Australia. Journal of Mammalogy, 75: 288-297. Simpson, G. G. 1983. Fossils and the history of life. Scientific American Books, Inc., New York. 99 Sinclair, A. R. E. 1996. Mammal populations: fluctuation, regulation, life history theory and their implications for conservation. In Frontiers of population ecology. Edited by Floyd, R. B., Sheppard, A. W., and DeBarro, P. J., CSIRO Publishing, Melbourne, Australia, pp. 127-154. Sinclair, A. R. E., Pech, R. P., Dickman, C. R., Hik, D, Mahon, P., and Newsome, A. E. 1998. Predicting effects of predation on conservation of endangered prey. Conservation Biology 12: 564-575. Smith, M. J., Brown, B. K., and Frith, H. J. 1969. Breeding of the brush-tailed possum, Trichosurus vulpecula (Kerr) in New South Wales. CSIRO Wildlife Research, 14: 181-193. Sokal, R. R., and Rohlf, F. J. 1995. Biometry, 3rd ed. W. H. Freeman and Co., New York. Stacey, P. B., and Taper, M. 1992. Environmental variation and the persistence of small populations. Ecological Applications, 2: 18-29. Stacey, P. B., Johnson, V. A., and Taper, M. L. 1997. Migration within metapopulations: the impact upon local population dynamics. In Metapopulation biology: ecology, genetics, and evolution. Edited by Hanski, I., and Gilpin, M. E., Academic Press, San Diego, pp. 267-291. StatSoft, Inc. 1995. STATISTICA for Windows. StatSoft, Inc., Tulsa, Oklahoma, U.S.A. Information available at [http://www.statsoftinc.com]. Stenseth, N. C , and Lidicker, W. Z., Jr. 1992. Where do we stand methodologically about experimental design and methods of analysis in the study of dispersal? In Animal dispersal: small mammals as a model. Editedby Stenseth, N. C , and Lidicker, W. Z., Jr., Chapman and Hall, London, pp. 295-318. Szalay, F. S. 1994. Evolutionary history of the marsupials and an analysis of osteological characters. Cambridge University Press, Cambridge. Taylor, A. C , and Cooper, D. W. 1998. Microsatellite markers for the Phalangerid marsupial, the common brushtail possum (Trichosurus vulpecula). Molecular Ecology, 7: 1780-1782. Tyndale-Biscoe, C. H. 1955. Observations on the breeding and ecology of the brushtailed possum, Trichosurus vulpecula Kerr (Marsupialia) in New Zealand. Australian Journal of Zoology, 3: 162-184. Ward, G. D. 1978. Habitat use and home range of radio-tagged opossums Trichosurus vulpecula (Kerr) in New Zealand lowland forest. In The ecology of arboreal folivores. Edited by Montgomery, G. G. Smithsonian Institute Press, Washington, D.C., pp. 267-287. 100 Waser, P. M., and Jones, T. W. 1983. Natal philopatry among solitary animals. Quarterly Review of Biology, 58: 355-390. Winter, J. 1976. The behaviour and social organization of the brush tailed possum Trichosurus vulpecula (Kerr). Ph.D. Thesis, University of Queensland, Brisbane, Australia. Wolff, J. O. 1997. Population regulation in mammals: an evolutionary perspective. Journal of Animal Ecology, 66: 1-13. Zar. J. H. 1996. Biostatistical analysis, 3rd ed. Prentice Hall, Upper Saddle River, New Jersey. 101 Table 4.1. Number and date of, and level of effort associated with, full-grid trapping sessions on the West grid. Also shown are the number of captures during full-grid trapping, target trapping, and nighttime trapping, as well as the number of times animals were either darted or resighted during spotlighting, relative to the dates on which full-grid trapping sessions were conducted. Full-grid trapping sessions Trap captures Spotlighting Session Date Trap-nights Full-grid Target Nighttime Darted Resighted 1 04/12/94 147 17 1 4 2 11/12/94 147 36 1 3 19/12/94 147 58 4 20/03/95 147 43 7 1 5 16/04/95 147 58 1 6 28/05/95 147 82 2 7 25/06/95 147 81 2 3 8 13/08/95 147 64 59 1 1 12 9 17/09/95 147 54 13 2 10 03/11/95 49 25 1 3 1 11 21/11/95 49 22 12 28/11/95 147 52 1 13 11/01/96 147 47 3 14 11/02/96 147 53 2 15 31/03/96 147 57 23 4 16 28/04/96 49 20 4 4 17 12/05/96 147 87 4 2 18 23/06/96 147 82 29 1 5 19 03/08/96 147 65 8 12 20 21/10/96 147 64 95 2 1 21 18/02/97 196 62 22 22 28/03/97 147 59 23 21/04/97 147 50 24 19/05/97 147 52 47 6 25 13/06/97 147 63 3 1 26 20/07/97 147 65 48 3 Total 3577 1418 372 7 8 58 102 Table 4.2. Number and date of, and level of effort associated with, full-grid trapping sessions on the East grid. Also shown are the number of captures during full-grid trapping, target trapping, and nighttime trapping, as well as the number of times animals were either darted or resighted during spotlighting, relative to the dates on which full-grid trapping sessions were conducted. Full-grid trapping sessions Trap captures Session Date Spotlighting Darted Resii 1 05/01/95 147 10 2 14/01/95 147 33 3 05/02/95 147 48 4 13/03/95 147 42 5 06/04/95 147 64 6 21/05/95 147 58 13 3 7 18/06/95 147 57 1 1 8 06/08/95 147 59 79 1 5 6 9 10/09/95 147 52 10 2 10 05/12/95 147 45 1 11 18/01/96 147 61 4 12 18/02/96 147 38 4 5 13 24/03/96 147 , 36 4 14 14/04/96 147 58 4 1 15 19/05/96 147 64 8 1 16 30/06/96 147 63 31 8 17 28/07/96 147 65 4 1 14 18 13/10/96 147 57 62 19 02/03/97 147 35 13 20 24/03/97 147 50 21 24/04/97 147 34 19 22 16/05/97 147 44 7 23 16/06/97 147 49 45 1 24 17/07/97 147 43 3 Total 3528 1165 309 2 7 33 103 Table 4.3. Level of effort associated with, and number of animals seen during spotlight transects in the core areas of the two study grids. The identity (if female) of individuals captured in conjunction with spotlighting is shown, if the animal was first captured after August 1,1995. In cases where unmarked animals were detected, but not captured, following August 1,1995, the occasion was assigned a lowercase letter, and the location at which the animal was seen was plotted on the appropriate range map (Figs. 4.4-15), as shown by a diamond symbol enclosing the relevant lowercase letter. 1995 1996 1997 Prior to After Prior to Prior to August 1 August 1 August 1 August 1 West Effort (minutes) 505 440 415 475 Marked 24 16 25 14 Unmarked Captured 0 Male 0 0 Not captured 6 0 W-b 0 East Effort (minutes) 410 . . . 420 345 Marked 18 5 5 Unmarked Captured 0 . . . 0 0 Not captured 1 . . . 0 E-b 104 Table 4.4. Level of effort associated with, and number of animals seen during spotlight transects on the periphery of the two study grids. See caption to Table 4.3 for details. 1995 1996 1997 West Effort (min) 775 415 515 Marked 35 25 21 Unmarked Captured 1 W-AI 0 Not captured 7 W-a W-c,d,e,f East Effort (min) 825 560 345 Marked 33 24 10 Unmarked Captured 6 Male 0 Not captured 13 E-a E-c,d 1 0 5 Table 4.5. Population size estimates regarding all adult and subadult females on the West grid. MNA-Grid and Jolly-Seber estimates are based solely on full-grid trapping sessions. The number of radio-collared females "present on the grid" at the time of each trapping session is also shown. MNA-AII includes all females tallied in the MNA-Grid estimate, as well as all those females initially captured using procedures other than full-grid trapping (target trapping, etc ...), that were known to be "present on the grid", at the time of each trapping session. Females W-AH (Fig. 4.7) and W-AJ (Fig. 4.8) were never captured or resighted along or interior to the periphery of the trapping grid (shown by the outer dashed line in Fig. 4.2), and were not considered to be "present on the grid", for the purposes of these comparisons. The Jolly-Seber estimates of population size (N) and the number of new animals joining the population (B) were calculated using program JOLLY-SEBER (Krebs 1999). Session MNA Jolly-Seber Model A Grid Radio All N S.E.(N) B S-E.(B) 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 5 13 24 25 25 28 28 21 20 20 20 22 22 23 25 26 27 24 22 19 18 19 18 17 15 15 5 9 25 27 20 20 20 20 20 20 21 22 22 25 26 26 22 20 20 19 18 18 16 7 15 26 28 28 31 32 26 25 24 24 25 25 26 27 28 29 26 27 22 20 22 20 19 19 17 13.0 25.5 26.7 25.0 28.4 28.6 21.6 20.0 20.0 20.0 23.1 22.0 23.4 25.7 28.1 27.9 24.6 24.0 19.5 18.2 20.0 19.1 17.9 15.1 2.3 2.8 2.9 2.1 2.2 2.4 3.1 2.9 2.9 2.9 3.3 2.9 3.1 3.2 4.2 3.1 3.2 3.8 3.3 3.2 3.4 3.5 3.4 3.1 13.3 3.5 -1.7 3.3 0.7 3.4 -0.4 0.0 0.0 3.1 -1.1 1.4 2.3 2.4 -0.9 -0.2 0.0 0.0 0.0 2.7 -0.7 1.3 -0.2 2.8 2.8 2.0 0.7 0.7 0.9 0.9 0.0 0.0 1.5 1.5 0.8 1.4 3.2 3.1 0.6 0.0 0.0 0.0 1.2 1.1 0.7 0.6 106 Table 4.6. Population size estimates regarding all adult and subadult females on the East grid. See caption to Table 4.5 for details. Females E-AF (Fig. 4.13) and E-AG (Fig. 4.14) were never captured or resighted along or interior to the periphery of the trapping grid (shown by the outer dashed line in Fig. 4.3), and were not considered to be "present on the grid", for the purposes of these comparisons. Session MNA Jolly-Seber Model A Grid Radio All N S.E.(N) B S.E.(B) 1 5 5 2 14 14 14.2 1.7 10.5 3.3 3 21 21 23.8 3.1 1.3 3.0 4 23 3 23 24.1 1.9 4.2 1.8 5 26 5 26 27.2 2.2 2.0 2.0 6 27 20 27 30.0 3.0 2.3 2.1 7 25 24 28 30.7 4.7 5.9 2.3 8 22 21 23 24.0 3.6 -1.7 2.1 9 20 19 21 22.2 3.9 1.7 1.3 10 19 18 19 20.4 3.5 0.6 1.4 11 19 17 19 19.3 3.1 -0.3 0.7 12 19 17 19 19.0 3.0 0.0 0.0 13 19 17 19 19.0 3.0 0.0 0.0 14 19 17 19 19.0 3.0 0.0 0.0 15 19 19 19 19.0 3.0 1.3 0.7 16 20 22 22 20.6 3.2 -0.3 0.8 17 19 21 21 22.1 4.2 1.0 0.3 18 13 15 16 13.2 2.9 1.4 1.1 19 12 14 16 12.9 3.1 1.1 1.5 20 12 14 16 13.7 3.5 2.1 2.8 21 11 14 16 12.5 3.7 -1.6 2.8 22 11 15 16 11.6 2.9 1.1 0.5 23 11 15 17 11.6 3.1 24 10 14 15 107 Table 4.7. Population size estimates regarding all adult and subadult males on the West grid. See caption to Table 4.5 for details. Session MNA Grid All N Jolly-Seber Model A S.E.(N) B S.E.(B) 1 10 11 2 19 19 22.6 3.7 4.6 3.0 3 23 24 23.1 1.7 2.6 1.2 4 24 25 24.6 2.1 4.5 1.8 5 28 29 29.4 2.4 -1.1 1.4 6 27 28 27.2 2.0 3.8 1.2 7 29 31 30.7 2.7 1.8 1.5 8 28 32 29.2 2.9 -0.5 0.9 9 24 26 24.5 3.0 0.0 0.0 10 23 24 23.0 2.9 0.0 0.0 11 23 24 23.0 2.9 1.3 0.7 12 24 24 24.3 3.0 -0.3 0.7 13 24 24 24.0 2.9 0.0 0.0 14 24 24 24.4 3.0 1.3 0.7 15 24 25 24.3 3.0 1.7 1.7 16 25 26 26.0 3.4 1.2 1.7 17 27 28 27.2 3.0 1.0 0.7 18 28 30 28.5 3.1 1.0 0.7 19 27 29 27.7 3.2 5.4 1.1 20 29 32 29.9 3.4 2.7 1.9 21 28 30 30.4 4.0 1.4 1.7 22 28 29 28.5 3.5 1.2 1.6 23 29 30 31.2 4.0 -0.7 1.2 24 27 28 27.8 3.6 2.4 0.8 25 25 25 26.3 4.4 26 21 24 108 Table 4.8. Population size estimates regarding all adult and subadult males on the East grid. See caption to Table 4.5 for details. Session MNA Grid All N Jolly-Seber Model A S.E.(N) B S.E.(B) 1 3 3 2 13 13 9.3 1.4 6.9 0.0 3 16 16 16.3 1.7 1.2 1.3 4 17 17 17.5 1.7 5.3 1.6 5 22 22 22.9 2.0 -0.7 1.2 6 21 22 21.5 1.8 1.0 0.3 7 20 21 20.2 2.1 3.8 0.0 8 20 26 19.8 2.4 3.8 1.1 9 20 25 20.6 2.8 0.6 1.3 10 21 25 21.6 2.8 2.0 0.9 11 22 26 22.4 2.8 1.2 1.2 12 22 25 23.5 3.2 -0.4 0.8 13 20 23 20.0 2.8 4.0 0.5 14 24 27 24.2 2.9 0.0 0.5 15 20 23 20.1 3.0 0.0 0.0 16 19 22 19.4 3.1 1.1 0.5 17 19 23 19.3 3.1 2.0 0.7 18 20 23 20.2 3.1 1.3 1.0 19 20 25 20.8 3.3 1.8 1.0 20 : 21 23 21.4 3.1 1.2 1.0 21 20 24 20.4 3.2 -0.4 0.8 22 20 24 20.0 3.1 3.1 0.6 23 22 24 22.7 3.8 24 15 15 109 0) (/) a) 1 g J S. s § I S 4HI irj ID T T (D 5- O O O ' - o d d * OJ W CO 3- o o o ' - o d d CM in co i n o o o *~ d d d ^ o o o , - d d d CO U") CO CO eo •<-. T o o o CM £: t; eo o o o *~ d d d O C D If) CO CO O O O CO C\J CJ y O J O O O *~ d d d 0 i- W N CM ^ P ^ CN W ^  O CM O O O d d d o ^ ^ io C J o o o ' " o d d ^ 5 U J Tt O y- ' 0 ) 0 0 ' CM O O 0 ) 0 0 0 CM d d d CM o o o CO C D Ol i — co o o o CM o d d ^ N co in CO o o o CM d d d M n T » -00 o o o CM d d d O CM ^  W 0 0 ^ CM O O O OO CO CO CO h. o o o CM d d d C J i n o o o CM o" d d P o o o N d 6 d CO 1— T - O CM d d d (O CD i -t0 O O T— w d d d CM 5; i n to o o o <M d d d § i — o i — CM d d d co ^ m co m o o o CM d d d to o i - o to o o o CM d o d 0 ) 0 0 0 ' - o d d ^ f CO CD 0 ) 0 0 0 •»- d d d If) CD O CO o O •*— d d d CO o O O ' - o d d • j — CT) CO CM CO O O r-i - o o o O) CM CO O f*» T '""I *~ O O O ' - o d d in eg y o y * - o o o co -<r co h-* - o o o o o o r*. ^  o y *~ o o o o o i - o >>. o o o CM o o o CD r - CM to o o o CM d d d ^ i n co co o o o CM d d d © CM co CO O O 1CM o d ^ CO C O N W T ^  T CM O O O M o d co o CO o o ^ o o T3- — C O CO T - o o o CM CO CO in CM o o d ' -CO CM CM o d ' C M d d d ' -o o d d CM d d d ' -O T- C7> O o d d CO o i — o co o o o ' - o d d CO CM CM CM co o o o ' - o d d S 5 U J 0 ) 0 0 0 ' - o d d co co t- i n co T _ ; , - ; *~ o o o T _ m C D co co o o o • - o d d CO O p. T - o * -CO CM CO CM co o o o ' - o d d CM o o d ' " d d d ' - i ^ CM m CO N CM CO W O i 1 •>— T— O *y O O O CO1 c M d d d ' - o d d ^ i m p T -w p p *~ o o CO - CM co o o T - o O co p p o c o p p p CM d d d o d d ' - o d d co o o o CM d d d ^ O O i-CM O O O CM d d d CM m CM CD CM O O O CM d d d S j f i S g f i 110 Table 4.10. Putative mother-daughter relationships identified among adult (and subadult) females on the West grid. Letters refer to females whose ranges are plotted in Figures 4.4-9. Dashed lines signify cases where no putative "mother" was identified from among the available candidates. Numerals followed by an equal sign indicate instances where more than one putative "mother" had the same likelihood score. Letters in bold indicate instances in which the putative mother and daughter had adjacent ranges. "Daughter" "Mother" Program KINSHIP Program CERVUS (error rate = 0.00008) ID Loci compared Score ID Loci compared mismatch Score rA — — rB — — A B 5 2.812 B 5 5.887 B A 5 2.812 A 5 5.887 C — — D E 6 3.554 E 6 7.089 E D 6 3.554 D 6 7.089 F — — G — — I 0 6 2.383 O 6 4.799 J — — K — AK 3 1 1.861 L AH 4 1.711 AH 4 3.534 M — — N 0 6 4.281 O 6 S 7.950 O N 6 4.281 N 6 7.950 P Q 6 4.568 Q 6 9.061 Q P 6 4.568 P 6 9.061 R S 6 2.762 S 6 5.613 S R 6 2.762 R 6 5.613 V AH 4 0.583 AH 4 1.159 W X 6 4.414 X 6 8.661 X w 6 4.414 w 6 8.661 AB — — AC AJ 3 1.399 I 6 1 3.271 AD P 6 4.487 P 6 8.300 AE AG 5 4.169 AG 5 7.804 AF — — AG AE 5 4.169 AE 5 7.804 AH L 4 1.711 L 4 3.534 Al — — AJ AC 3 1.399 I 3 2.933 AK — K 3 1 1.861 AL — — AM 4 = 1 1.195 AJ 1 2.332 111 Table 4.11. Putative mother-daughter relationships identified among adult (and subadult) females on the East grid. Letters refer to females whose ranges are plotted in Figures 4.10-15. Dashed lines signify cases where no putative "mother" was identified from among the available candidates. Letters in bold indicate instances in which the putative mother and daughter had adjacent ranges. "Daughter" "Mother" Program KINSHIP Program CERVUS (error rate = 0.00631) ID Loci compared Score ID Loci compared mismatch Score dB D 3 1.544 D 3 1.703 dC X 3 1.599 X 3 2.282 A B 5 2.616 B 5 3.144 B A 5 2.616 A 5 3.144 D dB 3 1.544 dB 3 1.703 F Y 3 1.626 Y 3 2.323 H I — T Y 5 3 2 1.498 1.240 J . . . L 4 1 0.985 K AG 3 1.842 AG 3 1.726 L . . . F 5 1 1.991 M AG 3 2.281 AG 3 2.487 N — W 5 2.208 O D 3 1.103 dB 4 1 1.390 P Y 3 1.621 Y 3 2.287 Q — AE 4 1.384 R M t 3 1.09 AH 4 1 % 1.565 S — AB 3 1 1.224 T AB 4 1.771 AB 4 2.443 U AG 2 1.117 AG 2 1.539 W — X 5 1 2.264 X dC 3 1.599 dC 3 2.282 Y F 3 1.626 F 3 2.323 Z AG 3 1 1.357 AB T 4 1.771 T 4 2.443 AC M 4 1.814 M 4 2.195 AD . . . F 4 1 2.242 AE AH 4 2.146 AH 4 1.817 AF . . . dC 3 1 1.889 AG M 3 2.281 M 3 2.487 AH AE 4 2.146 AE 4 1.817 Al — AC 4 1 1.138 AJ . . . K 4 1 0.506 112 Table 4.12. Differences in mean likelihood scores, as shown in Tables 4.10 and 4.11, between adjacent and non-adjacent putative parent-offspring pairs of adult (and subadult) females. If a given relationship was reciprocal (e.g., A was the most likely mother or daugter of B, and vice versa), the likelihood score was only counted once in the tallies shown. In addition, W-AE was not included in the tallies shown, because her status (as an immigrant) was already established. Differences were compared using t-tests on untransformed likelihood scores. Assumptions regarding normality and homogeneity of variances were tested using program STATISTICA (StatSoft, Inc. 1995). Program Grid Location N Likelihood scores Mean S.E. t df P KINSHIP West Adjacent 8 3.535 " 0.401 2.68 9 0.0250 Non-adjacent 3 1.559 0.525 East Adjacent 6 2.006 0.167 2.69 11 0.0209 Non-adjacent 7 1.448 0.128 CERVUS West Adjacent 8 6.979 0.690 4.56 12 0.0007 Non-adjacent 6 2.770 0.527 East Adjacent 8 2.264 0.180 3.17 24 0.0041 Non-adjacent 18 1.595 0.116 113 Table 4.13. Reported sex ratios of pouch-young. Deviation of the observed sex ratio from parity was analyzed at each site, by means of an exact binomial test (Zar 1996, p. 466), using program NCSS (Hintze 1996). Deviation of the observed sex ratio from parity, across all 10 sites, was tested by means of a log-linear analysis of sex ratio by site', using program STATISTICA (StatSoft, Inc. 1995). Reference Percentage Number p Female Female Male Tyndale-Biscoe (1955) 46.1 59 69 0.426 Dunnet (1956) 35.3 6 11 0.332 Lyne and Verhagen (1957) 48.2 66 71 0.733 Caughley and Kean (1964) 47.0 294 332 0.139 Smith et al. (1969) 47.7 74 81 0.630 Hope (1972) 45.4 94 113 0.211 Coleman and Green (1984) 46.3 354 410 0.047 Kerle and Howe (1992) 45.6 41 49 0.461 Efford (1998) 43.9 293 374 0.002 This study 35.4 34 62 0.006 1 The analysis indicated a significant marginal association with sex (X2 = 22.83, p < 0.0001) and site, but no interaction (p > 0.65). 114 115 1200 1000 800 e DD | 600 u o Z 400 o o 200 "©"•ll e e 200 400 600 800 Easting (m) 1000 1200 Figure 4.2. Schematic map of the West grid. Squares signify permanent trap locations, while circles indicate locations where animals were trapped at den trees. The inner dashed line delineates the "core" of the study grid, while the outer dashed line indicates the periphery of the trapping grid. Symbols in black represent "core" trap locations, while those in grey represent "peripheral" trap locations. 116 1400 1200 1000 o © M. 800 en s o Z 600 • • o • a ' o • • • 400 Si 200 200 400 600 Easting (m) 800 1000 1200 Figure 4.3. Schematic map of the East grid. Squares signify permanent trap locations, while circles indicate locations where animals were trapped at den trees. The inner dashed line delineates the "core" of the study grid, while the outer dashed line indicates the periphery of the trapping grid. Symbols in black represent "core" trap locations, while those in grey represent "peripheral" trap locations. 117 200 -i Prior to August 1,1995 800 H 600 A 400 H 200 H # 1 6 to 20 0 H i i i i i 0 200 400 600 800 1000 1200 Easting (m) Figure 4.4. Location of all females captured on the West grid prior to August 1,1995. Polygons show the denning ranges of radio-collared females. Rectangles indicate where females were captured, but not radio-collared. Uppercase letters identify individual females (see text for details). Polygons not accompanied by uppercase letters represent the denning ranges of core resident females removed on August 1, 1995. 118 August 1,1995 to September 15,1995 Fixes per den • 1 to 5 • 6 to 10 • 11 to 15 • 16 to 20 200 400 600 Easting (m) 800 1000 1200 Figure 4.5. Location of all females captured, or known to be present, on the West grid, between August 1, and September 15,1995. Polygons show the denning ranges of radio-collared females. Rectangles indicate where females were captured, but not radio-collared. Uppercase letters identify individual females (see text for details). 119 600 A 400 A 200 A September 15,1995 to March 15,1996 Fixes per den • 1 to 5 • 6 to 10 • 11 to 15 • 16 to 20 200 400 600 Easting (m) 800 1000 1200 Figure 4.6. Location of all females captured, or known to be present, on the West grid, between September 15,1995, and March 15,1996. Polygons show the denning ranges of radio-collared females. Rectangles indicate where females were captured, but not radio-collared. Uppercase letters identify individual females (see text for details). Diamond symbols enclosing single lowercase letters indicate where unmarked possums where seen during spotlight transects. 120 200 -| = : March 15,1996 to 000 A 800 A 600 A 400 A 200 A • 11 to 15 • 16 to 20 o A 1 1 1 1 1 1 0 200 400 600 800 1000 1200 Easting (m) Figure 4.7. Location of all females captured, or known to be present, on the West grid, between March 15,1996, and August 1,1996. Polygons show the denning ranges of radio-collared females. Rectangles indicate where females were captured, but not radio-collared. Uppercase letters identify individual females (see text for details). Diamond symbols enclosing single lowercase letters indicate where unmarked possums where seen during spotlight transects. 121 •B X AG O w R O C °E • o . PI August 1,1996 to March 15,1997 A B ADO Fixes per den • 1 to 5 • 6 to 10 • 11 to 15 • 16 to 20 200 400 600 800 Easting (m) 1000 1200 Figure 4.8. Location of all females captured, or known to be present, on the West grid, between August 1,1996, and March 15,1997. See caption to Fig. 4.7 for details. Open circles indicate the locations of the most frequently used dens (based on data presented in Fig. 4.7) of females that were not radio-tracked, but were known to be alive throughout part or all of the period from August 1,1996, to March 15,1997. The relevance of the male symbol is explained in the text. 122 March 15,1997 000 A 800 A 600 A 400 A 200 A • 6 to 10 • 11 to 15 # 1 6 to 20 o A 1 1 1 1 1 1 0 200 400 600 800 1000 1200 Easting (m) Figure 4.9. Location of all females captured, or known to be present, on the West grid, between March 15, and August 1,1997. See caption to Fig. 4.7 for details. The rectangle enclosing a question mark indicates the location at which remains were found of an unidentifiable young of the year. 123 Figure 4.10. Location of all females captured on the East grid prior to August 1,1995. Polygons show the denning ranges of radio-collared females. Rectangles indicate where females were captured, but not radio-collared. Uppercase letters identify individual females (see text for details). Polygons not accompanied by uppercase letters represent the denning ranges of core resident females removed on August 1,1995. 124 1400 1200 1000 A 800 OX) c o Z 600 400 A 200 X H D U August 1,1995 to September 15,1995 K I M N O Fixes per den • 1 to 5 • 6 to 10 • 11 to 15 #16 to 20 200 400 600 800 Easting (m) 1000 1200 Figure 4.11. Location of all females captured, or known to be present, on the East grid, between August 1, and September 15,1995. Polygons show the denning ranges of radio-collared females. Rectangles indicate where females were captured, but not radio-collared. Uppercase letters identify individual females (see text for details). 125 1400 1200 1000 800 © Z 600 400 4 200 September 15,1995 to March 15,1996 ' M Fixes per den • 1 to 5 • 6 to 10 • 11 to 15 #16 to 20 200 400 600 800 Easting (m) 1000 1200 Figure 4.12. Location of all females captured, or known to be present, on the East grid, between September 15,1995, and March 15,1996. Polygons show the denning ranges of radio-collared females. Rectangles indicate where females were captured, but not radio-collared. Uppercase letters identify individual females (see text for details). 126 1400 1200 H 1000 H 800 H OX fi o Z 600 4 400 H 200 H March 15,1996 to August 1,1996 Fixes per den • 1 to 5 • 6 to 10 • 11 to 15 #16 to 20 200 400 600 800 Easting (m) 1000 1200 Figure 4.13. Location of all females captured, or known to be present, on the East grid, between March 15,1996, and August 1,1996. Polygons show the denning ranges of radio-collared females. Rectangles indicate where females were captured, but not radio-collared. Uppercase letters identify individual females (see text for details). Diamond symbols enclosing single lowercase letters indicate where unmarked possums where seen during spotlight transects. 127 1400 1200 o H August 1,1996 to March 15,1997 1000 800 c o Z 600 x ••. o . w A E . . # AD [AHf\ M N O AC 400 H AF 200 H ABO R . IQ Fixes per den • 1 to 5 • 6 to 10 • 11 to 15 #16 to 20 200 400 600 800 Easting (m) 1000 1200 Figure 4.14. Location of all females captured, or known to be present, on the East grid, between August 1,1996, and March 15,1997. See caption to Fig. 4.13 for details. The caption to Fig. 4.8 explains the signficance of the open circles, and their locations are based on data presented in Fig. 4.13. The rectangle enclosing the question mark indicates the location at which remains were found of an unidentifiable young of the year. 128 1400 1200 March 15,1997 to August 1,1997 1000 800 OX) c O Z 600 H 400 200 Fixes per den • 1 to 5 • 6 to 10 • 11 to 15 #16 to 20 200 400 600 Easting (m) 800 1000 1200 Figure 4.15. Location of all females captured, or known to be present, on the East grid, between March 15, and August 1,1997. See caption to Fig. 4.13 for details. The relevance of the male symbol is explained in the text. 129 1994 i 1995 1996 W-AF I W-AE W-AD 1997 W-AK .Jc J$> J> ^ Jc J> J> ^ .J> J> J" / / / / / / / / / A / / c$r ^ ^ ^ ^ ^ ^ eg? ^ Figure 4.16. Cumulative number of newly-identified females initially captured in the core of one of the two study grids, from the start of spring 1994, until the end of the study in July 1997. The solid black line represents the number caught on the West grid, while the grey line indicates the number caught on the East grid. Letters identify individual females (see text for details). The vertical lines signify the August 1 removal of core females. 130 30 25 20 A 15 ~ 10 1994 1995 1996 E-AE E-AA E-AD W-AG X E-AH E-AI W-AA E-AB W-AB W-AC E-AC W-AI 1997 E-AJ W-AM Jc ^ & Jc J> ^ ^ .Jc J> J> ^ #V^V / / / / / / / / Figure 4.17. Cumulative number of newly-identified females initially captured on the periphery of one of the two study grids, from the start of spring 1994, until the end of the study in July 1997. The solid black line represents the number caught on the West grid, while the grey line indicates the number caught on the East grid. Letters identify individual females (see text for details). The vertical lines signify the August 1 removal of core females. 131 Figure 4.18. Cumulative number of newly-identified males initially captured in the core of one of the two study grids, from the start of spring 1994, until the end of the study in July 1997. The solid black line represents the number caught on the West grid, while the grey line indicates the number caught on the East grid. The vertical lines signify the August 1 removal of core females. 132 Figure 4.19. Cumulative number of newly-identified males initially captured on the periphery of one of the two study grids, from the start of spring 1994, until the end of the study in July 1997. The solid black line represents the number caught on the West grid, while the grey line indicates the number caught on the East grid. The vertical lines signify the August 1 removal of core females. 133 12 15 18 21 24 27 30 Minimum number of females alive on a grid (MNA-Grid) Figure 4.20. Plot of the minimum number of females known to be alive on a grid (MNA-Grid), in a given trapping session, versus the Jolly-Seber estimate of the number (N) of females present. Symbols in black signify data from the West grid, while those in grey represent data from the East grid. The diagonal line shows the expected 1:1 relationship if attempts at complete enumeration have been successful. 134 14 A 1.0 0.9 0.8 0.7 0.6 0.5 0.4 0.3 0.2 0.1 0.0 Proportion of full-grid trapping sessions in which captured Figure 4.21. Proportion of full-grid trapping sessions during which radio-collared females were captured. Black bars signify data from the West grid, while grey bars represent data from the East grid. Letters identify individual females (see text for details). 135 9 9 9 • 9 9 9 m • 9 9 • 9 m • • m 9 9 > • ® • 9 9 9 9 W • • • 9 9 1 m 9 • W 9 9 9 99 9 W N I B H • • • 9 9 9 9 9 9 9 9 • • 0 20 40 60 80 Total number of radio fixes Figure 4.22. Total number of dens used by a given female as compared to the total number of radio fixes for that female. Symbols in black signify data regarding females from the West grid (N = 38), while those in grey indicate data regarding females from the East grid (N = 41). 136 600 500 400 300 200 100 20 40 60 Total number of radio fixes 80 Figure 4.23. Total denning range length (the maximum distance between any two dens, in m) calculated for a given female, as compared to the total number of radio fixes for that female. Symbols in black signify data regarding females from the West grid (N = 38) , while those in grey indicate data regarding females from the East grid (N = 41). 1 3 7 Figure 4.24. Mean (+/- S.E.) proportion of radio fixes tracked to each of the array of dens used by a given female, among females located a minimum of 15 times. Black bars signify data from the West grid (N = 27 females; minimum number of fixes for a given female was 16, maximum = 79, mean +/- S.E. = 55 +/- 4.08), while grey bars represent data from the East grid (N = 29 females; minimum number of fixes for a given female was 15, maximum = 79, mean +/- S.E. = 48 +/- 4.37). 138 CHAPTER 5 CONNECTIVITY OR CARNTVORY: WHAT'S MORE IMPORTANT TO THE DEMOGRAPHY OF COMMON BRUSHTAIL POSSUM POPULATIONS? Habitat fragmentation has three major consequences: 1) habitat loss; 2) loss of connectivity; and 3) "edge effects" (Wilcove et al. 1986, Andren 1994). Species will be lost as a consequence of habitat loss if the amount of remaining habitat is less than the species' minimum area requirements. Loss of connectivity will affect movements between populations, and is thus of greatest relevance to the metapopulation dynamics of a species (Hanski 1998). Edge effects may include: greater penetration of generalist predators into remnant habitat patches; decreases in food availability within patches; and changes in habitat structure. Edge effects may affect both movements between populations, and dynamics within populations (Lidicker and Koenig 1996). Loss of connectivity has been identified as a key factor in the decline (Morton 1990), and future conservation (Laurance 1990, Laurance and Gascon 1997, Beier and Noss 1998), of medium-sized, herbivorous, marsupials, in Australia, a group which has suffered more extinctions and declines than any other mammalian taxa (Short and Smith 1994). Others (Catling and Burt 1995, Sinclair et al. 1998) have emphasized the importance of introduced terrestrial carnivores (foxes, Vulpes vulpes; cats, Felis catus; and dingos, Canis familiaris), in the decline and future conservation of medium-sized Australian mammals. Morton (1990) proposed a "conceptual model" that attempted to integrate issues regarding connectivity and edge effects, as a way of explaining the precipitous decline of medium-sized Australian mammals. To address these issues, I chose to study the common brushtail possum {Trichosurus vulpecula) as a "model" medium-sized, herbivorous, marsupial, for reasons explained in Chapter 4. I conducted a large-scale field experiment on common brushtail possums in old-growth Eucalypt forest in south-eastern Australia, which was designed to address the question: does immigration "rescue" populations from extinction? The study was conducted near the center of a vast (121,000 ha), largely undisturbed, wilderness area, where, it could be argued, connectivity was at a maximum, and edge effects were at a minimum (Chapter 4). In Chapter 4,1 reported on the rate of female immigration recorded in 139 response to the "pulsed" removal of resident females. Based on the results of the experiment, in conjunction with supplementary information regarding relatedness among resident females, and the primary sex ratio of pouch-young, I concluded that female immigration is generally next to non-existent in this environment. I inferred that since my study was conducted in an environment where connectivity was at a maximum, the rate of female immigration would be even lower where there were physical barriers to dispersal, as in fragmented or disturbed landscapes. My results indicate that the degree of connectivity is unlikely to have much influence on the demography of common brushtail possum populations in fragmented or disturbed landscapes. Consequently, any differences between the demography of the population I studied, in an unfragmented and largely undisturbed environment, and other populations in fragmented and disturbed environments, can be primarily ascribed to differences in the level of edge effects. From the preceding paragraph, the edge effects of most relevance will likely be those influencing the density and behaviour of introduced terrestrial carnivores. While having declined in Australia (How and Kerle 1995, Chapter 4), common brushtail possums are considered to be a significant pest species in New Zealand, where they were introduced in the 19th century, in order to establish a fur industry (Cowan 1990). For the purposes of the fur industry, and because they are thought to have detrimental effects on forestry and agriculture, common brushtail possums have been the subject of intensive research in New Zealand (Cowan 1990), whereas there has been comparatively little research conducted on their ecology in Australia (Green 1984, Kerle 1984). Several authors (Tyndale-Biscoe 1974, Green 1984, Cowan 1990) have reported that "disturbed" and "undisturbed" common brushtail possum populations in New Zealand demonstrate a largely consistent set of demographic differences, which I have listed in Table 5.1. The most well-studied undisturbed population in New Zealand occurs in mixed native forest in the Orongorongo Valley, near Wellington, North Island. The Orongorongo Valley population has been the subject of a live-trapping study that has continued without interruption since its initiation in 1966 (Crawley 1973, Fitzgerald 1976, Fitzgerald and Karl 1979, Humphreys et al. 1984, Efford 1998). Although no food addition experiments have been conducted, correlations between inter-annual variation in reproductive success, condition, and food supply (Humphreys et al. 1984), suggest that common brushtail possums in undisturbed populations in New Zealand are primarily food limited (Green 1984, Cowan 1990). This interpretation is supported by 140 the fact that the effect of introduced terrestrial carnivores (cats, Felis catus; and dogs, Canis familiaris) on these populations appears to be very slight (Fitzgerald and Karl 1979, Efford 1998). A "disturbed" population in New Zealand is, for the most part, a euphemism for a hunted population. Human hunting can be seen as analogous to a predator addition or removal experiment. According to Cowan (1990, p. 88), the low fecundity and delayed maturation characteristic of the reproductive biology of undisturbed possum populations (Table 5.1), shifts to the pattern of high fecundity and early maturation typical of disturbed populations (Table 5.1), following the onset of hunting in an area, and then shifts back to the earlier pattern of low fecundity and delayed maturation, following the cessation of hunting. Increased reproductive output in response to hunting pressure is the expected outcome in all wildlife harvesting operations (Caughley and Sinclair 1994, p. 281). Green (1984) reviewed the results of seven studies conducted on common brushtail possum populations distributed over a wide geographic range in south-eastern Australia. The studies were conducted on farms (Smith et al. 1969, Winter 1976), in forestry plantations (Hocking 1981, cited in Humphreys et al. 1984; How 1981), or in city suburbs (Lyne and Verhagen 1957, Pilton and Sharman 1962, Dunnet 1964). The results from all seven studies indicated that the reproductive biology of common brushtail possums in south-eastern Australia most closely resembles that of disturbed populations in New Zealand (Table 5.1). Green inferred from this that the demography of undisturbed populations in New Zealand was, in a sense, "unnatural". Cowan (1990, p. 95) reiterated Green's argument, in stating that "density in Australia is controlled largely by social behaviour (e.g., residents commanding key resources, such as dens, and expelling transients and immatures; Winter 1976), whereas [undisturbed] New Zealand populations are limited more by food than by behaviour". Green's hypothesis was that the open Eucalypt forests of Australia, in which common brushtail possums evolved, are effectively a "2-dimensional habitat" (Green's term), in which it is possible to defend exclusive areas, whereas spacing behaviour becomes ineffective in the more complex, "3-dimensional", native forests of New Zealand, and population densities increase to the point at which they are food limited. In this Chapter, I report on the reproductive biology of common brushtail possums at my study site (Paddys Land), in south-eastern Australia. Contrary to the results of previous studies in south-eastern Australia, the reproductive biology of common brushtail possums at the Paddys Land site demonstrated 141 all of the patterns characteristic of undisturbed populations in New Zealand (Table 5.1). Moreover, there was a remarkably close correspondence between the values of the various measurements I recorded, and those reported for the Orongorongo Valley population. I evaluate the likely causes of inter-annual variation in reproductive success, and suggest that common brushtail possums at the Paddys Land site, like those at the Orongorongo Valley site, are primarily food limited. I propose that the principal reason for the different demography of common brushtail possums at the Paddys Land site, as compared to other sites in south-eastern Australia, is that possums at other sites are subject to a greater rate of "harvesting" by introduced terrestrial carnivores. I discuss the implications of this for the conservation of medium-sized, herbivorous, marsupials, in general. METHODS Details regarding the study species, study site and general methodology were discussed in Chapter 4. Animals that were captured and released were either trapped or darted. Handling, as regards females, is here defined as making a visual inspection of the interior of the pouch. Females that were trapped more than once during a standard, full-grid trapping session, were generally only handled on the first occasion. On subsequent occasions during the trapping session the individual's ear-tag numbers were read through the bars of the trap and the animal was released without further processing. On all occasions when females were captured by darting, they were handled, held until the tranquilizer had worn off, and then released. I recorded the date, location and circumstances, in all instances where I either saw, heard or observed signs of potential predators. I also attempted to identify the predator involved in all cases where possums were attacked in traps, or where they were apparently killed by a predator (Chapter 6). Age estimates from cementum annuli In all cases where animals were found dead, I attempted to retrieve the teeth from the animal's lower jaw. Previous studies on common brushtail possums (Pekelharing 1970, Clout 1982), using known-age samples, have shown that the number of layers in the cementum cushion between the roots of each tooth can be used to determine the animal's age, in years. As it is well-established that new cementum layers 142 are deposited on a yearly basis, these can be justifiably referred to as cementum annuli (Pekelharing 1970, Clout 1982). I retrieved teeth from a total of 40 animals. I removed the first premolar (PM1) and third molar (M3) from the right lower mandible. Each tooth was assigned a randomly-selected number, and then all of the teeth collected (80 in total) were sent to Matson's Laboratory (Milltown, Montana, U.S.A.; [http://www.matsonslab.com]) for microscopic examination of the number of cementum annuli. Two serial sections were made of each tooth in this initial batch. Matson's Laboratory was not informed as to which two teeth belonged to the same animal, allowing me to conduct a blind test of the reliability of the results. The age estimates were highly correlated (r2 = 0.81, t = 12.61, df = 40, p < 0.0001). In the majority of cases (80 %), the age estimates were either the same, or they differed by as little as one year (Fig. 5.1). In four cases (10 %) there was a discrepancy between the age estimates of three or more years (Fig. 5.1). A second batch of teeth (12 in total) was sent to Matson's Laboratory, consisting of the first premolar (PM1) and third molar (M3) removed from the left lower mandible of each of the four animals for which there was a discrepancy between age estimates of three or more years in the first batch, along with teeth from two other animals. Six serial sections were made of each of the teeth in the second batch to ensure greater accuracy (Clout 1982), given the initial discrepancies. As with the first batch of teeth, Matson's Laboratory was not informed as to which two teeth belonged to the same animal. The previous discrepancies were either eliminated or greatly reduced in the results from the second batch (Fig. 5.1). I used the following protocols to assign a single age to each of the 40 animals whose teeth were examined: 1)1 ignored estimates from the first batch of teeth examined if age estimates were also available from the second batch; and 2) where discrepancies existed within a batch, I used the higher of the two estimates, under the assumption that cementum layers are more likely to be overlooked than they are to be misidentified (Clout 1982). Date of birth of pouch-young To evaluate the seasonality of births, and survival of young, it was necessary to assign a date of birth to each pouch-young observed during the study. For very small pouch-young, age in days was estimated from 143 crown-rump length (mm), using the nomogram developed by Lyne and Verhagen (1957, their Fig. 7). For larger young, age in days was estimated from head length (mm), using Equation 1: Days = - 2.50757 + 1.99855 (Head) + 0.01216 (Head2) Details regarding the derivation of Equation 1, and the protocols adopted, are presented in Appendix 2. RESULTS Seasonality of births Of the 152 dependent young for which a date of birth was identified, 147 (96.7 %) were born in autumn, and 5 were born in winter (Fig. 5.2). This very closely resembles the situation in the Orongorongo Valley, where about 95 % of births occur in autumn (Efford 1998, his Fig. 1). Two of the 5 winter-born young at the Paddys Land site were "replacement" young (Fig. 5.2), as the mothers of these young had already produced a pouch-young in autumn, which was lost, and then "replaced" by the new, winter-born young. I use the term "primary" young, in contrast to "replacement" young, to refer to a pouch-young produced by a female that has not yet given birth within the calendar year in question. "Double breeding" refers to situations in which a female gives birth to a young in early autumn, successfully raises it to weaning, and then gives birth to a second young in the spring of the same year (Green 1984, Kerle 1984). There were no instances of "double breeding" observed at the Paddys Land site. Similarly, Crawley (1973, p. 88) reported that "double breeding" was never observed during the first few years of the Orongorongo Valley study, and this would also appear to be true of later years, as Efford (1998, p. 513) noted that the 0.8 % of births in the Orongorongo Valley that occurred as late as September (see also Cowan 1990, his Table 19) all involved "replacement" young. There was a great deal of variation among females as to when they gave birth to their primary young (Fig. 5.2), with births being separated by up to 115 days (on the East grid in 1995). In contrast to the variation between females, there was very little variation in median birth dates (Fig. 5.2) across years (Kruskal-Wallis H = 1.20, N = 150, df = 2, p = 0.5495) or between study grids (Mann-Whitney U = 2642.0, z = 0.60, p = 0.5516). The median birth date of primary young at the Paddys Land site was around the third week of April, the exact date differing slightly depending on year and study grid (Fig. 5.2). Efford (1998, 144 p. 504; see also his Fig. 1) reported that that the annual median birth date in the Orongorongo Valley varied between April 18 and May 14. Almost all adult females at the Paddys Land site produced a single pouch-young each year (Table 5.1). Among radio-collared, resident, adult females, there were 144 cases (96.0 %) where the female gave birth in either autumn or winter, and only 6 cases where she did not do so (Table 5.2). Four of the six non-breeding females died during the breeding season: one was killed by a predator, another died from a massive Ophidascaris robertsi infection (Chapter 6), the third was observed to be both "rumpy" and "bony", and the fourth was observed to be "weak and wobbly". As documented in Chapter 6, survival among females observed to be both "rumpy" and "bony", and females observed to be "weak and wobbly", was significantly poorer than among females that never demonstrated these symptoms of poor condition. Two females survived throughout the breeding season but apparently did not give birth. One of these two surviving females had been attacked by a predator in mid-February and was severely injured. The final "non-breeding" female may actually have given birth and lost her pouch-young before she was first captured in mid-June, as she was recorded at this time as having an engorged mammary gland, although she was not lactating. A different female was known to have given birth and lost her young prior to the end of June, indicating that the "true" autumn birth rate may be slightly underestimated if judged solely on the presence or absence of a pouch-young in late June. Efford (1998, p. 510) reported that 90.4 +/- 6.0 % (mean +/- S.D.) of adult females trapped in the latter half of June at the Orongorongo Valley site were found to possess pouch-young. Sex ratios of dependent young The sex of dependent young was identified whenever possible. As with measurements of head length it was not always possible to check the sex of the young because: 1) the pouch-young was judged to be too small to safely handle; or 2) the mother was struggling too much. The sex of young cannot be identified reliably prior to 14-21 days of age (Tyndale-Biscoe 1955). In most cases estimates of head length were easier to accomplish than examinations of sex because the head of the pouch-young generally lies adjacent to the opening of the pouch while the hind limbs are nestled in the folds of the pouch. 145 Sporadic examinations were sufficient to identify the sex of 94.0 % of young in 1996 and 1997 (Table 5.3). However, the unanticipated, poor survival of pouch-young in 1995 (described later) meant only a handful of young were examined for sex prior to their disappearance (Table 5.3). Fewer than half of the young seen during the vehicle trail trapping session in Oct.-Nov., 1996 (Fig. 4.1), were sexed (Table 5.3), because my objective was simply to check for the presence or absence of young. My aim in trapping along the vehicle trail was to assess the reproductive success of females that had not been previously captured and handled, and I wanted to minimize the level of stress experienced by these animals, in case I decided to trap along the vehicle trail trapline at some later date. A log-linear analysis of the sex ratio of primary young, by study grid (West, East) and year (1995, 1996, 1997), revealed a significantly male-biased sex ratio, independent of study grid or year (Table 5.3; see also Table 5.1). Hope (1972) reported differential mortality between the sexes during pouch life, with the proportion of males decreasing significantly with increasing age. The sex of many of the dependent young at the Paddys Land site was not identified until late in pouch life, or when they were already back-young. If the phenomenon Hope (1972) described applied at the Paddys Land site, then the magnitude of the male bias in the sex ratio reported in Table 5.3 is potentially an underestimate. Efford (1998, p. 510) reported a significantly male-biased sex ratio of pouch-young in late June at the Orongorongo Valley site. Stages in the weaning of young Young possums can potentially breed when they are a year old (Green 1984). Whether the young survives to this age is in part a function of the mother's survival, at least until the young is weaned. Unfortunately, differentiating between pre-weaning and post-weaning survival is not straightforward. Young in the pouch can be classified as pre-weaning (Kerle and Howe 1992). Young that have permanently left the pouch and begun riding on their mother's back may or may not be weaned. Kerle and Howe (1992) suggested that the initiation of weaning begins no earlier than 175 days of age, about 25 days after the young has begun riding on its mother's back. It is not clear from the literature what the minimum age is at which weaning can be said to have been completed, and the young could survive on its own. Kerle and Howe (1992) stated that the young is weaned between 175-240 days of age, whereas Smith et al. 146 (1969) suggested that young are weaned completely between 220-290 days of age. If a back-young is present and the mother is still nursing, then it is likely the young is still largely dependent on its mother. Another complication is that nursing mothers may leave their older dependent young in the den while they forage (Winter 1976, p. 246). In addition, Winter (1976, p. 250) reported that young began to be seen at night independent of their mothers, while still sharing their mother's den, once the young was 200 days of age. Consequently, a stage will be reached when mothers may be seen to be nursing, in the absence of their young being present. Only when the young is seen after the mother is known to be no longer nursing can the young be truly classified as having survived to the post-weaning phase of development. Figure 5.3 illustrates all relevant observations made at the Paddys Land site concerning the timing of the three stages in the weaning of young: 1) together with the mother and still nursing; 2) separate from the mother but still nursing; and 3) separate from the mother and no longer nursing. The earliest a dependent young was first seen out of the pouch and riding on its mother's back was at 172 days of age. The earliest a nursing mother was seen without her young was when the young was 183 days old. The latest a mother was seen to be still nursing was when the young was 287 days old. The earliest a surviving young was known to have been completely weaned was at 202 days of age, however, the next earliest, among the other 14 young known to have survived to independence, was at 271 days. Pre-weaning survival of young Pre-weaning survival depends in part on the mother's survival, which could be checked on a weekly basis simply by tuning into her radio frequency. To evaluate the fate of the young required that the mother be captured and examined. Given the irregularities in the frequency of captures the most sensible way to deal with data concerning the pre-weaning survival of the young itself was to judge whether or not the young had survived to some threshold age. Smith et al.'s (1969) suggestion that young are completely weaned (stage 3 in the process of weaning) between 220 and 290 days of age is clearly an overestimate as regards at least some of the young at the Paddys Land site (Fig. 5.3). At the same time, Kerle and Howe's (1992) suggestion that the process of weaning begins at 175 days, 25 days after the young leaves the pouch and begins riding on its mother's back, may be overly conservative, as young at the Paddys Land site were only just emerging from the 147 pouch at 175 days (Fig. 5.3). Given these alternatives, I decided to err on the side of caution and set the minimum threshold for pre-weaning survival at 175 days. Females observed to be without a young and no longer nursing prior to 175 days after the birth of their young were judged to have lost their young prior to weaning. Results regarding pre-weaning survival in the two complete years for which data were available are presented in Table 5.4. The longer a young survived the more likely it was to be seen alive. Once seen, the young's survival could then be "back-dated". A young seen on day 200 obviously survived to day 175. Since young generally survived longer in 1996, more of them were seen alive. Consequently, there were fewer young for which it was unknown (the "?" column in Table 5.4) whether they survived to 175 days. The survival of such "back-dated" young is a function of both the likelihood of pre-weaning survival as well as survival during and after weaning. Pre-weaning survival, in and of itself, can only be evaluated by limiting consideration to those young known to have been lost prior to 175 days of age (the "No" column in Table 5.4). Pre-weaning survival was significantly poorer in 1995 than in 1996 (Table 5.4, see also Table 5.1). There was an almost significant three-way interaction in the relevant log-linear analysis, with the difference.between years being much more dramatic on the West grid than the East grid (Table 5.4). However, as just discussed, this test is conservative in only using data from the "No" column in Table 5.4. If the proportion of failed young in the unknown sample (the "?" column in Table 5.4) was the same as in the known sample ("No" vs. "Yes" columns in Table 5.4) the effect of year would be more pronounced and the difference between grids less so. In a later section of the present Chapter, I report on evidence indicating that survival of pouch-young to the onset of weaning is significantly poorer among first-time breeders. Considering only females that were known or presumed to have given birth at least once before, I estimate that survival to the onset of weaning was 33.3 % in 1995, and 81.3 % in 1996. Efford (1998, p. 510) reported that the estimated survival of pouch-young from June to September at the Orongorongo Valley site was 82.4 +/- 17.6 % (mean +/- S.D.), very close to the estimated rate of survival at the Paddys Land site in 1996. Efford (1998, pp. 504 and 513) cited several pieces of evidence which suggest that the sample of females used to generate the estimated survival of pouch-young at the Orongorongo Valley site likely included only those females that had given birth at least once before. The estimated 33.3 % survival of pouch-young to the onset of weaning at the 148 Paddys Land site in 1995 is below the lower 95 % confidence limit for survival at the Orongorongo Valley site (= 47.9 % = 82.4 - [1.96 x 17.6]). Efford (1998, p. 513) noted that survival to the onset of weaning at the Orongorongo Valley site was estimated to be only 52 % during the early years of the study (1966-1975). Efford suggested that the poorer survival of pouch-young during the early years of the Orongorongo Valley study may have been attributable to the greater frequency with which mothers with pouch-young were captured and handled in the earlier years of the study as compared to later years (1980-1994). A difference in the frequency of capture and handling is one of at least four proximate causes that may explain the significant difference between 1995 and 1996 as regards the pre-weaning loss of pouch-young at the Paddys Land site. Other potential causes include: 1) a difference in the frequency with which mothers were chased by predators; 2) a difference in the frequency of infanticide; and 3) a difference in food availability. Pre-weaning loss of young as a result of capture and handling Two pouch-young were lost as a direct result of the mothe&being captured, one case involved trapping and the other involved darting. Neither of these two pouch-young are included in the tallies presented in Table 5.4. In the case where the young was lost as a direct result of trapping, the 90-day-old young was found dead, lying on the floor of the trap beside its mother. The mother had been trapped previously with her pouch-young, was apparently unharmed, and showed no signs of having been harassed by a predator while in the trap. In only one other instance was a young ejected from the pouch when the mother was trapped. In the latter instance, the 149-day-old young was, however, evidently unharmed, as its mother was trapped the very next day with the young securely back in the pouch. Moreover, this particular young was known to have survived to independence. These two cases of young being ejected in response to trapping were clearly exceptional. Females were trapped with pouch-young less than 175 days of age on 624 occasions, and in 622 cases the mother did not eject the young. There were a total of three instances where the young had not been ejected in the trap, but after having been processed, the young was not securely back in the mother's pouch upon her release. In two instances the young was out of the pouch but was seen to be securely attached to the mother's back as the mother was followed back to her den. In the third 149 instance the young did become separated from its mother, but she evidently returned and retrieved the young within the space of an hour. There were nine occasions when females with pouch-young less than 175 days of age were darted. On two of these occasions the young fell out of the pouch. In both instances the young was put back in the pouch. In one case the pouch-young was alive and well and securely in the mother's pouch when she was released the next morning. In the other case the young was found lying dead, beside its mother, even though the mother was evidently unharmed. It is likely the tranquilizer had the effect of relaxing the pouch muscles. In the instance where the young had fallen out but survived, the pouch had been loosely sealed with surgical tape (which was removed before releasing the mother), after the young was put back in the pouch. This precaution was not taken in the case where the young was later found dead. Although there were only two direct observations of young being lost in response to trapping or darting, there may have been less obvious effects of handling animals such that the poorer survival of pouch-young in 1995 was, nonetheless, the consequence of experimenter-induced loss of young. I undertook three analyses to evaluate the importance of experimenter-induced loss of young, I tested: 1) whether the fate of young was associated with the frequency with which the mother was handled (Table 5.5); 2) whether losses were induced by ear-tagging (Table 5.6); and 3) whether the survival of young was greater among a group of females that had not been trapped previously, as compared to females that had been trapped and handled repeatedly (Table 5.7). For each of the 63 pouch-young shown in Table 5.4, whose fate was known for certain (excluding those in the "?" column), I identified each occasion on which the mother was handled while the young was in the pouch, up to the time the young was 175 days of age. I then calculated the interval, in days, between each successive occasion on which the mother was handled. There were 9 young whose mother's were only handled once during the period in question, 4 of which survived to the onset of weaning, while the remaining 5 did not. For each of the remaining 54 young whose mother's were handled on more than one occasion, I calculated the mean, and median, interval (in days) between occasions on which the mother was handled. I calculated both the mean and median handling interval for each female because I had no a priori way of knowing what the shape of the distribution of handling intervals would be. If a distribution is symmetrical there will be little difference between the mean and the median value (Sokal and Rohlf 1995, 150 p. 46). To evaluate the shape of the distribution of handling intervals, I compared the mean and median handling intervals for each female using Wilcoxon's signed-ranks test. As there was no significant difference between mean and median handling intervals (T = 44.0, z = 0.53, p = 0.5936), I conclude that the mean handling interval is representative of the "typical" frequency with which the mother was handled. Mean handling intervals were compared between young that died prior to, and those that survived to, the onset of weaning (175 days), in both years, on both grids, using a three-way ANOVA (Table 5.5). The mean interval between handling occasions was significantly shorter among young that died, than those that survived. The mothers of young that died were generally handled about once a month, while the mothers of surviving young were generally handled about once every second month (Table 5.5), suggesting that mothers that were handled more frequently were more likely to lose their young. The interval between occasions on which a mother was handled while the young was in the pouch, prior to 175 days after the birth of her young, will be a function of the number of times she was handled, and the total time over which the young survived. If a pouch-young died only a few days after its birth, its mother would likely have been handled on fewer occasions while the young was alive, than the mother of a young that survived to 175 days of age, which may confound the interpretation of just what the mean interval between handling occasions represents. However, as regards the 54 pouch-young under consideration in Table 5.5, there was no significant difference in the number of times the mother was handled (Mann-Whitney U = 317.5, z = 0.73, p = 0.4644) between young that died, and those that did not, suggesting that the interval between handling occasions is not confounded by the number of times the mother was handled. Young that survived a shorter period of time had mothers that were handled as often as young that survived a longer period of time, indicating that the interval between handling occasions was what was different between the two groups. I interpret the results of Table 5.5 as suggesting that handling may have had an adverse effect on the survival of pouch-young. Tagging the young was clearly the most invasive procedure I undertook while the young was still in the pouch. Thirty-five of the 63 known-fate pouch-young shown in Table 5.4 were known to have survived after being ear-tagged. Yet it was quite often the case that a young would be ear-tagged on one occasion, and the next time its mother was seen, the young had been lost. However, among the 28 pouch-young shown in Table 5.4 that were known to have died prior to 175 days of age, 151 there was no significant association between when the young was lost and whether it had been ear-tagged (Table 5.6). About as many young died before being ear-tagged (13) as died after being ear-tagged (15). The most rigorous means of testing the importance of experimenter-induced loss of young is to compare the survival of young between females that have not been trapped previously as against females that have been captured and handled repeatedly. In October-November, 1996,1 attempted to capture every radio-collared female on the two principal study grids (which I will refer to as "On-grid" females), to determine whether their young had survived to 175 days of age. At the same time I attempted to capture as many previously untrapped adult females as possible along the vehicle trail trapline, illustrated in Figure 4.1. The 40 vehicle trail traps were baited and set for 13 nights (520 trap-nights) between October 12 and November 15. There were a total of 98 captures (18.8 % trap success) over the 13 nights. I caught 49 previously untrapped adult and subadult possums, of which 16 were adult females (which I will refer to as "Off-grid" females). Table 5.7 summarizes the fate of young of both On-grid and Off-grid mothers. The tallies regarding the On-grid mothers include the 34 pouch-young shown in Table 5.4, whose fate was known for certain in 1996 (excluding those in the "?" column), as well as the single young known to have been lost in response to darting, given that I was interested in evaluating the adverse effects of capture and handling (the single instance when a young was lost after its mother ejected it, in response to trapping, occurred in 1995). Among On-grid females, 22 were seen with dependent young, 6 were not seen with young but were clearly still nursing more than 175 days after the birth of their young, 9 were known to have lost their young, and 2 did not have young and showed no signs of nursing (Table 5.7). Among Off-grid females, 8 were seen with dependent young, 6 were not seen with young but were clearly still nursing, and 2 did not have young and showed no signs of nursing (Table 5.7). I assume that both Off-grid females that did not have young and were no longer nursing, had in fact given birth earlier in the year (p > 0.95, Table 5.2). Both females had large pouches, indicating that they had clearly given birth on at least one or more occasions (Bolliger and Carrodus 1938, 1940), neither showed signs of being in poor condition, and both were within the range of normal body weights (2000 g or more) for adult females at this time of year (Fig. 5.6) 152 The two On-grid females trapped in the October-November period that did not have dependent young and were no longer nursing (Table 5.7), were trapped 189, and 197 days, respectively, after the birth of their young. The two Off-grid females that did not have young with them and were not nursing were trapped 183, and 207 days, respectively, after the median birth date of young on the trapping grids (April 20, Fig. 5.2). If these two Off-grid females had young that were born on March 18, the earliest known birth date in 1996, the young would have been 216, and 240 days old, respectively, when their mothers were first trapped. The earliest a surviving young was known to have been completely weaned was at 202 days of age, and the next earliest, among the other 14 young known to have survived to independence, was at 271 days (Fig. 5.3). It is therefore almost certain that both of the On-grid females and at least one of the two Off-grid females had lost their young prior to the completion of weaning. Assuming that females seen without young and showing no signs of nursing had lost their young, the proportion of young surviving to the onset of weaning that were born to On-grid mothers (70.3 %), was not significantly from that among Off-grid mothers (87.5 %, Table 5.7). Only if both of the two Off-grid females, had in reality, already weaned their young, would the difference be significant. The results from Table 5.5 regarding differences in the frequency of handling among On-grid females suggest that handling may have had an adverse effect on the survival of pouch-young, while the results from Table 5.6 regarding the consequences of ear-tagging suggest that this is not the case. Assuming that the frequency of handling influences survival, but ear-tagging does not, this is not a sufficient explanation of the difference in the survival of pouch-young between 1995 and 1996 because there was no significant difference in the frequency with which females were captured and handled, between these two years (Chapter 6). However, as will be demonstrated in a later section, there was evidence of food shortage in 1995 as compared to 1996, and there may have been an interaction such that the same level of handling nonetheless resulted in poorer survival in 1995, because mothers were already in poorer condition. Unfortunately, because I did not conduct Off-grid trapping in 1995, as I did in 1996 (Table 5.7), I cannot rule out the possibility that the decreased survival of pouch-young in 1995 was the result of an interaction between food shortage and the frequency of capture and handling. 153 Pre-weaning loss of young as a result of predator chases Marsupial mothers have the distinct advantage of being able to eject their young from the pouch when being chased by a predator, thereby sacrificing the young to the predator in order to save themselves. If the greater pre-weaning loss of pouch-young in 1995 was a response to predators I would expect to have evidence of greater rates of predator activity in 1995 than in 1996. Table 5.8 summarizes the available evidence of predator sightings, sounds, and signs, as well as the number of adult possums apparently killed or scavenged by a given predator, during the pre-weaning phase of development of young in 1995 and 1996. Since I spent comparable amounts of time in the field during this period in both 1995 and 1996 (Tables 4.1 and 4.2), differences in the number of incidental observations can be attributed to differences in the level of predator activity rather than differences in the amount of sampling. While there was a trend towards greater predator activity in 1995 (Table 5.8), notably as regards dingos, the numbers are too small to provide an adequate test of the relationship to the pre-weaning loss of pouch-young. Nonetheless, it seems unlikely that the occasional pursuit by a predator could be responsible for the dramatic difference in survival between 1995 and 1996. When possums were released from traps, every attempt was made to try to follow them back to their dens. Excluding instances when the animal was not followed or when it disappeared from view, there were 502 cases where females with pouch-young were followed until they came to rest. Often this would involve running after the animal at (the experimenter's) top speed. In no case did the mothers respond to this pursuit by ejecting their young. Overall it seems females were extraordinarily reticent to give up their young. The most salient example of this was a female trapped and released 110 days after the birth of her young, who was observed to still have her young in the pouch, even though it was dead and rotting. Crawley (1973, p. 82) also reported instances of young being found dead in the pouch at the Orongorongo Valley site. Pre-weaning loss of young as a result of infanticide While it cannot be said that the activity of predators differed dramatically between 1995 and 1996, the density of female conspecifics certainly did. As documented in Chapter 4, the center-most, resident, adult females were removed from both principal study grids on August 1, 1995. A total of 10 females were removed from the West grid, while 9 were removed from the East grid, a reduction in the density of adult 154 females by about one-third (Tables 4.5 and 4.6; Figs. 4.4 vs. 4.5, and 4.10 vs. 4.11). No new adult females settled in the vacancies created, over the course of the next two years. Consequently, the density of adult females in 1996 was about one-third less than the density in July, 1995 (Tables 4.5 and 4.6; Figs. 4.4 vs. 4.7, and 4.10 vs. 4.13). Wolff (1997) argued that infanticide committed by adult females against one another is a principal driving force in the evolution of social systems in many mammalian lineages. If aggressive encounters between female possums were responsible for the loss of young (direct infanticide as opposed to resource-depletion), I would expect better survival of young in 1996, as was observed, given the lower density of females. Wolff (1997) argued that marsupials are unlikely to demonstrate infanticide since their otherwise vulnerable altricial young are protected in the mother's pouch. Wolff (1997) was apparently not aware that marsupials demonstrate behaviours such as mothers leaving their dependent young in the den while they forage, which he suggests is the phase of development when infanticide occurs in prosimian primates. In the present context I am addressing pre-weaning survival to 175 days, when most young are still in the pouch (Fig. 5.3). Hence, Wolff (1997) would predict that infanticide is not a significant threat to survival at this stage of development. Nonetheless, it is at least possible that possums can cause each other to eject their young. If infanticidal aggression between females was the principal cause of the loss of pouch-young I would expect more pre-weaning loss of young before August 1, 1995, when the reduction in female density was undertaken, than after. Moreover, I would expect poorer survival prior to August 1 in 1995 than in 1996. In both 1995 and 1996 the "typical" pouch-young was about 100 days old as of August 1, given that the median birth date of primary young in both years was around April 21 (Fig. 5.2). As is shown in Table 5.9, at most, a single pouch-young was lost prior to August 1 on a particular trapping grid in a given year. Obviously, the majority of cases of pre-weaning loss of pouch-young in 1995 occurred after the August 1 reduction in adult female densities, and not before (compare Tables 5.4 and 5.9). Similarly, there was no difference between 1995 and 1996 as to survival prior to August 1 (Table 5.9). Pre-weaning loss of young as a result of food shortage Since the majority of cases of pre-weaning loss of pouch-young occurred after August 1 (Table 5.9), this implies that most young were lost sometime after they were about 100 days old (given a median birth 155 date of April 21, Fig. 5.2). The weight of pouch-young increases very gradually from 0 -100 days of age and then increases very dramatically from 100 days onwards (Fig. 5.8; see also Lyne and Verhagen 1957, their Fig. 8). The evidence that the timing of the loss of young (sometime after August 1, Table 5.9) corresponds to the period of growth during which the young is putting the greatest energetic demands on the mother (after 100 days of age, Fig. 5.8) suggests that the survival of young to the onset of weaning at the Paddys Land site is most likely governed by the availability of food. Similarly, Humphreys et al. (1984, p. 64) noted that the average median date on which pouch-young were lost at the Orongorongo Valley site was August 19. Moreover, Humphreys et al. (1984, their Table 2) reported a significantly positive relationship between the survival of pouch-young and adult female body weight in winter, which in turn showed a significantly positive relationship to the size of the annual crop of Hinau fruit. To evaluate the possibility that the loss of pouch-young at the Paddys Land site was a function of food shortage I examined evidence regarding four potential indices of condition, as well as changes in adult body weights. The four indices of condition were, whether or not an individual: 1) was seen to stop and feed on grass or leaves, following their release from a trap during the day; 2) was observed to be "weak and wobbly" upon release; 3) felt "bony"; or 4) was noticeably "rumpy". Survival was significantly poorer among females observed to be "weak and wobbly" upon release, and among those observed to be both "rumpy" and "bony", as compared to females that never demonstrated these symptoms of poor condition (Chapter 6). Naturally, if the mother dies, so too will the pouch-young. In the present context I am interested in factors that might influence the survival of pouch-young possessed by surviving females. Table 5.10 shows the proportion of young that survived to the onset of weaning in 1995 and 1996 whose mothers demonstrated any or all of the four symptoms of poor condition under consideration. The observed proportions reported in Table 5.10 were compared to those expected in each of 1995 and 1996 (33.3 and 81.3 % survival respectively) based on the data presented in Table 5.4. Sample sizes were too small to conduct statistical tests except in the case of "rumpy" females. Mothers that were noticeably "rumpy" were significantly less likely (x 2 = 7.96, df = 3, p = 0.0468) to have a pouch-young that survived to the onset of weaning, as compared to other females. Figure 5.4 shows the percentage of surviving individuals of both sexes seen feeding on grass or leaves, following their release from a trap, in each season, over the course of the study. The percentage 156 of surviving adult females that were either observed to be "weak and wobbly" upon release, felt "bony", or were noticeably "rumpy", in each season, over the course of the study, is shown in Figure 5.5. In all cases the proportion of individuals showing symptoms of poor condition was greater during the 1996 breeding season than during the 1995 breeding season. This is the opposite pattern to that expected if the four symptoms of poor condition under consideration were causally related to the poorer survival of pouch-young in 1995. The most direct way of examining contrasts between the 1995 and 1996 breeding seasons is to compare changes in the status of the same individual. Whether or not an individual demonstrated any of the four symptoms of poor condition in a given season in 1995 was compared to whether or not it demonstrated symptoms in the same season in 1996. The proportion of individuals demonstrating a change in symptoms was evaluated using a repeated measures "exact" binomial test (Sokal and Rohlf 1995, p. 784). Pooling data across grids, there was a strong trend towards being more likely to be diagnosed as being "rumpy", in the autumn of 1996 as compared to the autumn of 1995, although the trend is not significant if the p-value is interpreted conservatively (Table 5.11). Nonetheless, the direction of the trend is the opposite to that expected if the likelihood of being diagnosed as being "rumpy" were causally related to the poorer survival of pouch-young in 1995. There were no other significant effects or noteworthy trends regarding any of the other symptoms of poor condition under consideration. While the survival of a pouch-young within a year appeared to be related to whether or not its mother was noticeably "rumpy" (Table 5.10), the difference in the proportion of young surviving between years (1995 vs. 1996) did not appear to be associated with inter-annual differences in the proportion (Fig. 5.5) and frequency (Table 5.11) with which females were diagnosed as being "rumpy". This might suggest that the significant difference between 1995 and 1996 in the proportion of young surviving to the onset of weaning (Table 5.4) was not a function of food shortage, if food shortage were the only factor influencing the severity of "rump wear". However, the cause of "rump wear" is probably multifactorial and contributing factors include behavioural stress, immunosuppression, and underlying disease (Hemsley and Canfield 1993, Viggers and Spratt 1995), all of which may be exacerbated by food shortage. That food shortage by itself is not sufficient to explain the underlying cause of "rump wear" is demonstrated by the fact that the condition has been reported in captive animals (Presidente 1984, Clark 1995). Nonetheless, as 157 food shortage can be expected to exacerbate the underlying causes of "rump wear", the argument that food shortage was responsible for the poorer survival of pouch-young in 1995 is not supported by the evidence regarding "rump wear", nor by the evidence regarding the other three indices of condition under consideration. Figure 5.6 illustrates the pattern of seasonal changes in mean body weights of adult females and males, over the course of the study. Body weights of adult females shown in Figure 5.6 are the aggregate of the mother's weight and the weight of her pouch-young. No attempt was made to remove the young from the pouch in order to get a separate measurement for mother and young. During the pre-weaning phase of development the aggregate weight of mother and young is wholly dependent on the mother's ability to find food, as the young has not yet begun to forage on its own. In essence the young at this time represents an enlargement of female reproductive tissue. The mother's weight taken alone may not reflect difficulties in finding food as accurately as the aggregate weight if the mother "chooses" to invest the bulk of her energy into the growth of her young at the expense of her own condition. In 1996 the mean weight of females, being for the most part the aggregate weight of mothers and young, as most females had surviving young (Table 5.4), increased steadily from autumn to spring on both grids (Fig. 5.6). By contrast, aggregate weights declined from autumn to spring on both grids in the 1995 breeding season. As the majority of females in 1995 had lost their young by the middle of spring (Table 5.4), weights taken at this time primarily reflect the mother's weight by itself. Given that the weight of the pouch-young in autumn is negligible (the oldest young at the end of autumn would be 90 days of age, corresponding to a weight of 60 g, according to Lyne and Verhagen 1957; see also Fig. 5.8), the decrease in weight between autumn and spring suggests that the mother's own weight was declining during this period. Interpreting changes in adult male body weights is more straightforward. Mean male body weights on both grids declined between autumn and winter in both 1995 and 1996 (Fig. 5.6). In 1995 mean weights on both grids continued to decline between winter and spring, whereas mean weights increased from winter to spring on both grids in 1996. Changes in mean body weight could reflect either changes in the weight of the same individuals measured over time or a change in the composition of the population being measured. For example, there may have been an influx of fat males in the spring of 1996. To avoid potentially confounding factors 158 associated with a change in the composition of the population, I restricted the analysis to an examination of the change in weight of the same individuals measured over time. Table 5.12 summarizes the median difference in the mean weight of individual adult possums of both sexes (female weights representing the aggregate weight of mother and young) measured in a given season in 1996 as compared to the same individual's mean weight in the same season in 1995. There were no clear differences between the two years regarding comparative weights in autumn or winter. However, an individual measured in the spring of 1996 was generally about 200 g heavier than it had been in the spring of 1995, regardless of sex. The difference in spring weights was statistically significant for males on both grids. The difference was statistically significant for females on the West grid, and all six females, for which data were available, on the East grid, were also heavier. The significant difference in spring weights between 1995 and 1996 may have been a consequence of the one-third reduction in the density of adult females associated with the removal of the center-most, resident, adult females undertaken on August 1, 1995 (Chapter 4). In this case, the reduction in intraspecific competitors is of importance as regards resource-depletion, rather than direct behavioural interactions such as would be involved in instances of infanticide. If the one-third reduction in the density of adult females were the only factor associated with the greater survival of pouch-young in 1996,1 would predict poorer survival of pouch-young, and lower body weights among Off-grid females, as compared to On-grid females in 1996 (post-removal), under the assumption that Off-grid densities were similar to pre-removal On-grid densities in 1995. I have already discussed the fact that the proportion of young surviving to the onset of weaning was, if anything, slightly greater among Off-grid females (Table 5.7). Similarly, there were no significant differences in spring (Oct.-Nov.) body weights between On-grid and Off-grid females (Table 5.13). While the one-third reduction in the density of adult females does not appear to be a sufficient explanation for the differences in the survival of pouch-young, and adult body weights, between 1995 and 1996, another, non-exclusive, explanation would be a change in environmental conditions between the two years. Total monthly rainfall in autumn and winter was dramatically different between the two years (Fig. 5.7). Total rainfall in August, 1995, fell below the 5 th percentile, based on 68 years of records, while total rainfall in May, 1996, was above the 97.5 th percentile (data are from the Kookabookra 159 weather station, 30°00'39"S 152°00'34"E, elevation 975 m; approximately 20 km from the Paddys Land site). I did not attempt to measure food availability directly at the Paddys Land site because possums are "opportunistic herbivores" (Kerle 1984, Cowan 1990), so I was unable to identify a specific food item that might act as an index of overall food availability, such as was done at the Orongorongo Valley site, as regards the size of the annual crop of Hinau fruit (Humphreys et al. 1984). Humphreys et al. (1984) were able to use the availability of Hinau fruit as an index of food availability because an extensive dietary analysis had been conducted earlier on during the Orongorongo Valley study (Fitzgerald 1976). On the other hand, the fact that possums are "opportunistic herbivores" suggests that they will almost certainly benefit from a general increase in the amount of vegetation, resulting from increased rainfall. In summary, three lines of evidence suggest that the poorer survival of pouch-young in 1995 was a function of food shortage: 1) the fact that most young were lost during the growth phase when they were putting the greatest energetic demands on the mother; 2) the fact that both males and females had significantly lower body weights in 1995; and 3) the fact that rainfall in the autumn and winter of 1995 was exceptionally low. Since none of the four indices of condition indicated that animals were in poorer shape in 1995, this clearly weakens the argument regarding food shortage, but it must be noted that it is not clear to what degree these indices of condition reflect nutritional state, as opposed to immunocompetence, stress level, or other components of general health. Post-weaning survival of young In total for the 1995 and 1996 breeding seasons combined, there were 14 primary young tagged in the pouch that were later seen as subadults, independent of their mothers (Table 5.14). One male subadult born in 1995 on the East grid was seen in January of 1996 and never again. The remaining 13 subadults were seen sometime after the beginning (mid-March) of the breeding season in the year following the year in which they were born. More than twice as many young survived to the beginning of the breeding season in the year following their birth in 1996, than in 1995 (Table 5.14). The difference between the two breeding seasons could have been a function of differences in pre-weaning (to 175 days) survival (Table 5.4) or survival after the onset of weaning. Extrapolating the proportion of young known to have survived to the onset of weaning 160 (Yes/[Yes + No], Table 5.4) to the subset of young whose fate was unknown (the "?" column in Table 5.4), the total number of primary young surviving to the onset of weaning (Yes + [[Yes/[Yes + No]] x [?]]), in 1995, was most likely 13 . Similarly, the total number in 1996 was most likely 29. Survival after the onset of weaning to the beginning on the next breeding season was evaluated by calculating the proportion of young that survived to the beginning of the breeding season in the year following their birth (Table 5.14) from among the total number that likely survived to the onset of weaning (Table 5.4). Estimated survival from the onset of weaning to the beginning of the next breeding season was 30.8 % (4/13) in 1995, and 31.0 % (9/29) in 1996. Consequently, the difference between years in the number of young surviving to the beginning of the breeding season in the year following their birth would appear to be largely a function of pre-weaning survival (Table 5.4). Since young were not radio-collared, it is not possible to conclusively differentiate between survival and disappearance due to emigration. Previous studies suggest that subadults begin to emigrate at 9-10 months of age (Clout and Efford 1984, Efford 1991). Differences in survival from the onset of weaning to the beginning of the next breeding season in 1995 and 1996 may have been masked if there was a corresponding difference in emigration rates. I have no direct way of evaluating this possibility, but-indirect evidence (described below) suggests there was no difference in emigration rates between the two years in question. Confounding survival with disappearance due to emigration causes survival to be underestimated. Pooling across years, survival from the onset of weaning to the beginning of the next breeding season is estimated to be 31.0 %. It is important to remember that this is a minimum estimate. Emigration rates in possums are often observed to be male-biased, although significant differences between the sexes are demonstrable only when large samples are generated by pooling across years or between sites (Clout and Efford 1984, Efford 1991). If my estimate of survival from the onset of weaning was confounded with disappearance due to emigration I would expect more females than males among the young surviving to the next breeding season, given male-biased emigration. I did in fact find that more females (7) than males (6) survived to the next breeding season, however, the proportion of survivors that were female did not differ significantly (%2 = 1.93, df = 1, p = 0.1643) from the number expected (4.6) given a primary sex ratio of 35.4 % female for the Paddys Land site (Table 4.13). 161 Another way of evaluating the frequency of emigration is to quantify the level of immigration. It is the level of immigration by 1-year-olds that is of most relevance with respect to post-weaning survival. To identify instances of immigration by 1-year-olds I needed to establish: 1) when it is appropriate to classify newly-identified individuals as potential immigrants (Chapters 3 and 4); 2) where they were detected relative to their potential point of departure (Chapters 3 and 4); and 3) what age they were. As discussed in Chapter 4, there was only one instance (on the West grid in 1995) in a total of 83 cases (1.2 %) where the offspring of a resident female could have been misidentified as being an immigrant (because I were unable to tag it while it was still in its mother's pouch), if it was later recaptured. As I will show, this potential source of error is only of relevance to the estimated rate of immigration by 1-year-old males. Where newly-identified individuals are first captured is of relevance because individuals caught in traps on the periphery of the trapping grids may be residents from just off the grid undertaking a long distance "excursion", rather than "true" dispersal (Lidicker and Stenseth 1992, Chapters 3 and 4). The closer to the interior of a trapping grid a newly-identified individual is caught, the more likely it is that this individual is actually in the process of emigrating from its original home range. I used five types of evidence to evaluate the likely age of newly-identified individuals first captured any time after the beginning of spring, 1995 (Chapter 4): 1) whether they were sexually active (adult vs. subadult); 2) how much they weighed; 3) the condition of their pouch (if female); 4) the length of their testes (if male); and 5) their estimated age from cementum annuli (if available). Sexual maturity, by itself, is not a sufficiently accurate indicator of age. One female young tagged while still in the pouch was known to have given birth in the autumn of the year after the year in which she was born (Fig. 5.8). On the other hand, based on my sample of 40 animals aged from cementum annuli, the earliest at which all individuals are certain to be adults is at 4 years of age (Fig. 5.9). Sexual immaturity is at least a starting point in helping me sort out the age of potential immigrants, as I could be sure that subadults were all less than 4 years old. Efford (1994, p. 6) noted that this was also true of the Orongorongo Valley population. Gilmore (1969) and Crawley (1973) both used body weight to aid in estimating the age of young possums. Adult body weights range from 1-3 kg depending upon location (Cowan 1990, Kerle et al. 1991, How and Kerle 1995), and growth rates of young vary accordingly (Efford 1991, p. 29; Appendix 2). 162 Consequently, it was necessary to determine the observed variation in adult body weight, and rate of growth of young at the Paddys Land site, prior to using body weight as an indication of age. Gilmore (1969) classified as immature any animal weighing less than 2000 g. Crawley (1973, p. 78) reported that, among known-age young at the Orongorongo Valley site, 1-year-olds never weighed more than 1450 g, while 2-year-olds never weighed less than 1700 g. Figure 5.8 illustrates the pattern of increase in body weight with age among known-age young tagged in the pouch at the Paddys Land site. I have not attempted to fit a regression line to these data because they include repeated measures from the same individual, as well as data from different individuals. The overall pattern of growth illustrated in Figure 5.8 would appear to be similar to that observed among captive animals by Lyne and Verhagen (1957, their Fig. 8): 1) weight increases slowly up to 90-110 days of age; 2) t