Geographical variation in the standard physiology of brushtail possums (Trichosurus): implications for conservation translocations

Brushtail possums have a wide geographical distribution. Possums from warmer, drier habitats have more frugal energy and water use and increased heat loss capacity at high temperatures. Consideration of geographical patterns in physiological attributes of source populations will likely improve the success and welfare outcomes of translocations.


Introduction
A fundamental goal of comparative physiology is to determine how variation in physiological traits and processes differs spatially and/or temporally between and within species (Feder and Block, 1991;Bozinovic et al., 2011). Macrophysiological studies have commonly focused on broad interspecific comparisons to achieve a wide geographical and taxonomic perspective for environmental or life history influences on a particular variable (Chown et al., 2004;Bozinovic and Naya, 2015). However, this approach generally ignores potential variation within a particular taxonomic group, assuming that the trait(s) in question are fixed for each species being studied (Bozinovic et al., 2009). Variation in physiological traits between spatially separated populations of a single or closely related species has been relatively understudied (Bozinovic and Naya, 2015;McClelland et al., 2016), but recognition of lower-taxonomic-level physiological variation is gaining prominence along with attempts to identify the underlying mechanisms and functional significance when such variability is found (Bozinovic et al., 2009).
Studies of low-level taxonomic variation over a wide spatial scale not only have application to theoretical considerations of physiological function, evolutionary processes, ecological interactions and species assemblages (Chown et al., 2004;Bozinovic et al., 2009;Chown and Gaston, 2016), but they can also address more applied questions concerning impacts of human-induced habitat modification on biodiversity, and of global climate change (Bozinovic et al., 2011). These two key environmental perturbations of the Anthropocene are gaining ever more social, economic and political attention (Feder and Block, 1991;Steffen et al., 2007). We consider here whether there is sufficient spatial variation in the physiology of a widely distributed mammal to warrant consideration in conservation translocation programmes. More broadly, we consider how studies of comparative physiology can enhance the success of wildlife management and ecosystem conservation in a changing environment.
Translocation of species is a widely used conservation tool, but the success rate of translocations is very low worldwide, with only one third to one half of translocation programmes deemed successful. Translocation outcomes in Australia are particularly poor (Griffith et al., 1989;Wolf et al., 1996;Fischer and Lindenmayer, 2000). The appropriateness of the new habitat and characteristics of a translocated species affect the success of these programmes (Sarrazin and Barbault, 1996;Armstrong and Seddon, 2007), and we posit that characterizing the environmental physiology of the animals to be translocated will further improve their chances for survival. Wildlife managers commonly consider factors such as predation, competition, genetics, habitat, reproductive biology, behaviour and disease as influencing the success of conservation measures (Griffith et al., 1989;Sarrrazin and Barbault, 1996;Snyder et al., 1996;Fischer and Lindenmayer, 2000), but basic environmental physiology is generally overlooked (Tarszisz et al., 2014). This is despite the IUCN Guidelines for conservation translocations clearly indicating that the biotic and abiotic habitat needs and basic biology of a species should be known prior to planning a translocation, and specifically recognizing that an assessment of physiological suitability should be made (IUCN/SSC 2013). Physiological assessment can predict a mammal's likely capacity for survival and reproduction in a particular environment, and therefore evaluation of pertinent physiological variables should improve the success of conservation translocations beyond that achieved with current ecological, behavioural, health and genetic considerations (Tarszisz et al., 2014).
Environmental conditions directly affect an animal's energy, water and thermal requirements and whether it can successfully survive and reproduce (Wikelski and Cooke, 2006). It is therefore not surprising that there are significant correlations between metabolic, thermal and hygric physiological traits and environmental factors such as temperature and rainfall at a broad inter-specific level for endotherms (e.g. Tieleman and Williams, 2000;Schleucher and Withers, 2001;Lovegrove, 2003;Rezende et al., 2004;Withers et al., 2006;White et al., 2007;van Sant et al., 2012), suggesting that these physiological characteristics are under environmental selection. However, few studies have examined such correlations at lower taxonomic levels, and even fewer describe intra-specific physiological variation, although there is some evidence that geographic variation in physiological traits is consistent with accommodating environmental variables (e.g. Tracy and Walsberg, 2000;Mueller and Diamond, 2001;Tieleman and Williams, 2002;Williams et al., 2004;Careau et al., 2007;Bozinovic et al., 2009;Cooper and Withers, 2010). These studies suggest that it is inappropriate to consider a physiological variable as a fixed species-wide trait, and that intra-specific variation may play an important role in determining the potential survival of individuals translocated to a new geographical location. Translocated individuals must have physiological traits that are appropriate for survival and subsequent reproduction in their new environment, or they must have the phenotypic flexibility to quickly acclimatize. Environmental conditions such as drought have been implicated in the decline of populations of re-introduced species (e.g. Winnard and Coulson, 2008;Facka et al., 2010), thus identifying intra-specific differences among individuals from different geographical locations most likely to physiologically cope with environmental conditions will aid in planning for more successful translocations.
The common brushtail possum (Trichosurus vulpecula) is the mostly widely distributed Australian marsupial, with a historical distribution throughout Australia (How, 1983;How and Hillcox, 2000). Despite it being one of the most successful marsupial urban adaptors, the distribution and abundance of the brushtail possum has declined post-European settlement. Declines are particularly prominent in the arid zone where the brushtail possum has disappeared from much of its former range (Kerle et al., 1992;How and Hillcox, 2000). Increasing environmental aridity, compounded by other anthropogenic impacts such as hunting, predation by introduced mammals and land clearing have been implicated in its decline (Kerle et al., 1992). As a consequence of this decline and its wide geographical distribution across the Australian continent, the brushtail possum features in many of the conservation translocation programmes that aim to re-establish functional mammal communities on the Australian mainland, particularly in the arid zone. There are a number of such programmes currently planned or in the early stages of implementation by government and non-government conservation agencies (Short and Hide, 2014;Natural Resource Management South Australia Arid Lands, 2017;Department of Biodiversity, Conservation and Attractions, 2017;Australian Wildlife Conservancy, 2018;Threatened Species Recovery Hub, 2017) and so quantifying the physiological characteristics of brushtail possums on a broad spatial scale, and assessing potential for variation consistent with local environmental conditions, is a timely contribution to these ambitious conservation initiatives.
There is considerable variation in size and colour of brushtail possums, and this has led to the recognition of various sub-species, and in some cases new species, each with distinct geographical ranges (Kerle et al., 1991;Lindenmayer et al., 2002). Body size variation is related to environmental aridity and food availability, with possums from more arid regions having a smaller body size than those from mesic areas (Yom Tov and Nix, 1986). We examine here if this observed anatomical variation is accompanied by physiological variation on a large spatial scale, from south-west Western Australia to north-east Queensland. If pertinent physiological traits are found to vary geographically within this species, then this has important implications for the success of translocation programmes.

Methods
Brushtail possums were captured in wire cage traps baited with a rolled oat, peanut butter and sardine mixture. Possums were captured and held under licence from the relevant state wildlife authorities, and experiments followed the Australian code of practice for the care and use of animals for scientific purposes, approved by the animal ethics committees of Curtin University, University of Western Australia, James Cook University, University of New England and University of Wollongong. 152.3°E). All possums were adult and no females were lactating. The total annual rainfall, mean number of rainy days, and mean minimum and maximum ambient temperatures of the capture locations are presented in Table 1, with the WA site characterized by a warm and dry climate, the two NSW sites cool and moist, and the QLD site being hot and wet (Australian Bureau of Meteorology, 2017). Brushtail possums were held in captivity for several weeks, fed on an ad libitum diet of fresh fruit and vegetables, cheese, Eucalyptus leaves, small animal muesli and rodent cubes, with ad libitum drinking water.
Metabolic rate (MR), as determined by rates of oxygen consumption (VO 2 ) and carbon dioxide production (VCO 2 ), and evaporative water loss (EWL), were measured using standard flow-through respirometry after Withers (2001). Measurements were made, in random order, at ambient temperatures (T a ) ranging from 6°C to 35°C; not all possums were measured at all T a for logistical and ethical reasons. Data for T. v. hypoleucus at 26°C and 30°C have been published by Cooper and Withers (2008). Possums were fasted overnight, then measured during their inactive phase (day) for~8 h, except at T a = 34°C where experiments were no longer than 6 h to avoid potential adverse effects of heat exposure and dehydration. Possums were measured at only a single T a per day, with at least three days between successive measurements. Body temperature (T b ) was measured using a Radiospares (Smithfield, New South Wales Australia) thermocouple metre (±0.1°C), with a plastic-tipped thermocouple inserted~2-3 cm into the cloaca immediately after the possum was removed from the chamber at the conclusion of each experiment. Possums were weighed to the nearest gram before and after each experiment, and the mean of the two measures used in calculations.
The respirometry system consisted of an 8000 cm 3 metabolic chamber placed in a controlled temperature cabinet or room (±~2°C). Airflow through the chamber was achieved via a variety of pumps or a compressed air line; water vapour was removed using Drierite (W.A. Hammond Drierite Co. Ltd, Xenia, OH, USA), and flow rate regulated by a mass flow controller (Aalborg GFC171 or Omega FMAA2412; Orangeburg, NY, USA and Stamford, CT, USA respectively) at 2.5-4.6 l min -1 (dependent on T a and animal mass). A subsample of excurrent air passed through a Vaisala HMP45A (Helsinki, Finland) relative humidity (RH) and T a probe, then through Drierite to carbon dioxide and oxygen analysers (Leybold-Heraeus Binos-C Cologne, Germany, Qubit S153 Kingston, Ontario, or Sable Systems CA-2A Las Vegas, NV, USA CO 2 analysers; Sable Systems Foxbox or Servomex 572, 574 or OA184 Crowborough, East Sussex, UK, O 2 analysers). A PC running a custom-written Visual Basic (V6; Microsoft, Redmond, WA, USA) programme recorded the voltage outputs of the analysers every 10-30 s. A baseline of background O 2 , CO 2 and RH was established for at least 20 min before and after each experiment.
The mass flowmeters were calibrated using a Gilian Gilibrator (Sensidyne, St Petersburg, Florida, USA), traceable to a national standard, or a bubble flowmeter, corrected to standard temperature and pressure dry (STPD). Gas analysers were calibrated using room air (20.95% O 2 ), nitrogen (O% O 2 and CO 2 ) and a precision gas mix (0.53% CO 2 , BOC Gases, Perth, Western Australia) and/or a butane flame after Withers (2001). Calibration of the RH probes, achieved by saturating air at a known temperature and then warming to T a after Cooper and Withers (2008), was routinely confirmed using 2 points, 1% RH (dried with Drierite) and 100% RH (saturated; by breathing on the probe). A mercury thermometer, traceable to a national standard, was used for temperature calibration.
Respirometry calculations were made after Withers (2001) using a custom-written Visual Basic programme, and resting VO 2 , VCO 2 and EWL were calculated for a period of at least 20 min during each experiment when values were steady and minimal, indicating the possums were at rest. Respiratory exchange ratio (RER) for each experiment was calculated as VCO 2 /VO 2 , and was used to convert MR to metabolic heat production (MHP) and metabolic water production (MWP) using oxy-calorific and hygric conversion coefficients at the measured RER (Withers et al., 2016). EWL was converted to evaporative heat loss (EHL) using 2.4 J mg -1 H 2 O (Withers et al. 2016). Wet (C wet ) and dry (C dry ) thermal conductance (J g -1 h -1°C-1 ) were calculated as C wet = MHP/(T b -T a ), and C dry = (MHP-EHL)/(T b -T a ). Relative water economy (RWE) was calculated as MWP/ EWL and the point of relative water economy (PRWE) as the T a where RWE was calculated to be 1.
We used generalized linear mixed effect models (GLMM) to examine T a and geographic location effects, while accounting for repeated measurements of individuals as a random factor, using the lmer function in library lme4 (Bates et al. 2014) and lmerTest (Kuznetsova et al. 2014), in RStudio (RStudio Team, 2015, with Satterthwaite's approximations for calculation of degrees of freedom. Individual differences between possums were examined with a likelihood ratio test of the random effect. We examined the physiological response to T a for each species, with T a as a polynomial fixed factor and individual as a random factor. To compare between locations, MR, EWL and C wet were expressed as mass-independent values using the marsupial scaling exponents of 0.74, 0.68, 0.57, respectively (Withers et al., 2006). We then compared body mass and physiological responses to T a among locations, with T a as a polynomial function, location as a fixed factor, and individual as a random factor. Pair-wise location comparisons were made with the most arid habitat sub-species (T. v. hypoleucus from WA) as the reference category. Finally, we compared mass-independent standard physiological variables (measured at basal MR; T a~2 6°C) among possums from each location using a one-way ANOVA with Student-Newman-Keuls post-hoc tests, in StatistiXL (v2.1, Nedlands, Western Australia). Values are presented as mean ± standard error (SE), with N = number of individuals and n = number of measurements.

Discussion
We have quantified significant variation in the physiological traits of spatially separated populations of brushtail possums, marsupials with a wide geographical distribution throughout the Australian continent. This variation was consistent with environmental patterns of inter-specific variation for marsupials (Withers et al., 2006) and other endotherms (e.g. Tieleman and Williams, 2000;Schleucher and Withers, 2001;Lovegrove, 2003;Rezende et al., 2004;White et al., 2007;van Sant et al., 2012), and with anatomical differences identified previously for brushtail possums (Yom Tov and Nix, 1986). We discuss here potential environmental drivers of this geographic variability and then assess the likely impacts of these patterns on management and conservation actions, particularly conservation translocations, for this species and for mammals in general.
Possums from all locations had typical endothermic responses to T a . Possums were thermoneutral at 26°C, MR increased at lower T a while T b , EWL and C remained relatively constant, and all variables increased at higher T a . This pattern was consistent with previous observations for T. v. vulpecula from NSW (Dawson, 1969;Dawson and Hulbert, 1970), although our measures of BMR and EWL were 16-24% lower, and our standard T b was 1.3°C lower (despite almost identical body masses). These differences are presumably due to possums in the earlier studies being restrained and having a thermocouple inserted in the cloaca throughout measurement.
Various macro-physiological studies have reported environmental influences on standard physiological variables that highlight the important role of these factors on the physiological phenotype of mammals. Smaller body mass is well described for many other mammalian species from hot compared to colder habitats (i.e. Bergmann's rule; reviewed by Meiri and Dayan, 2003). There are a number of environmental correlates with standard physiological variables for marsupials (Withers et al. 2006) that are broadly consistent with patterns for mammals (e.g. Lovegrove, 2003;Rezende et al., 2004;Van Sant et al., 2012). Inter-specific environmental correlates for standard marsupial T b , BMR and EWL indicated that species from arid environments with high rainfall variability have lower values than species from mesic environments with more reliable rainfall.

6
have higher BMR and EWL, and lower PRWE, than species from arid, hot and low productivity habitats (Mueller and Diamond, 2001;Williams et al., 2004;Careau et al., 2007;Cooper and Withers, 2010). Broadly, intra-specific physiological differences between populations reflect these intra-generic, and wider inter-specific, patterns for rodents, bats, marsupials and monotremes (Augee, 1978;Tracy and Walsberg, 2000;Geiser and Ferguson, 2001;Bozinovic et al., 2009;Dunbar and Brigham, 2010;Stawski and Geiser, 2011), but there are exceptions. For example, there were no differences in torpor T b for northern and southern little red bats (Lasiurus borealis), but there were for big brown bats (Eptesicus fuscus; Dunbar and Brigham, 2010), while woodrats (Neotoma spp.) from different environments had similar MR and T b but different C wet and body mass (Brown and Lee, 1969). There is currently no clear explanation as to why the established ecological drivers of physiology observed at higher levels of taxonomy are reflected in geographically separated populations of many, but not all, species. Presumably there are a suite of genetic, evolutionary, life history and environmental factors that determine populational differences a over broad geographic scales (Dunbar and Brigham, 2010). A much more comprehensive dataset of broadscale intra-specific population studies is required to address this question, but it does highlight the requirement for speciesspecific studies to examine potential geographic effects, when the aim is to inform conservation and management actions.
Our physiological data for brushtail possums are generally consistent with the broad macro-physiological observations and lower-taxonomic level environmental patterns for other mammals. Body mass variation with T a is well documented for brushtail possums (Yom Tov and Nix, 1986), consistent with our results. Differences in thermal, metabolic and hygric traits of brushtail possums were also generally consistent with other studies, with possums from warmer/drier habitats having more frugal energy and water use and increased capacity for EHL at T a above thermoneutrality. The PRWE, an index of water economy and hence a measure of adaptation to aridity (MacMillen, 1990;MacMillen and Hinds, 1983;Hinds and MacMillen, 1986) provided further evidence that the physiology of T. v. hypoleucus from WA is more favourable for drier habitats compared to possums from other environments.
The variation we identify here for brushtail possums throughout their geographic range adds to the growing evidence that basic physiological traits are not necessarily fixed species-specific characteristics, and that local environmental conditions can be significant physiological drivers. Unfortunately, there is little evidence that translocation studies consider this variation in the planning phase. Baker and Gemmell (1999) found that possums translocated from cool-temperature Armidale (NSW) to sub-tropical Brisbane (QLD) had greater immune and hormonal responses, and mortality compared to locally captured possums. Our data, showing significant geographic variation in possum physiology, suggest that climate may have contributed to these results; possums from Armidale likely had a less favourable physiology for the Brisbane climate and this impacted on their health and survival.
Despite the potential for acclimatization to account for geographic patterns observed here for brushtail possums, data of Cooper and Withers (2010), which approximated a common garden design for another widespread genus of marsupial (Dasyurus), suggests a strong genetic basis to their physiology, reflecting environmental conditions. Even if there are significant genetic differences between brushtail possums populations, with sufficient time, re-introduced populations could respond to a new environment by adaptation (evolution, micro-evolution or epigenetic control of gene expression; Hetem et al., 2014). However, re-introduced populations are by necessity small and may not have the capacity to survive because of the reduced fitness of less-suited individuals, and the variation in the founder population might not provide sufficient adaptive scope.

Recommendations
Given our observed differences in brushtail possum physiology, together with the findings of Baker and Gemmell (1999) concerning translocation responses of brushtail possums from cool-temperature to sub-tropical environments,  we recommend that possums involved in translocation programmes should be sourced from areas with the most similar environmental conditions to the proposed release site, not necessarily possums from the closest geographical location. As it is the arid zone where brushtail possum re-introductions are most desirable (Kerle et al., 1992), then the sub-species T. v. hypoleucus would be most physiologically appropriate for these translocations, having physiological traits most favourable for the low productivity, low and variable water availability and extreme T a of arid environments.

Implications for conservation translocations
Similar intra-specific geographic variability in physiological traits for an array of other species (e.g. Tracey and Walsberg, 2000;Mueller and Diamond, 2001;Williams et al., 2004;Bozinovic et al., 2009;Dunbar and Brigham, 2010;Stawski and Geiser, 2011) suggest that our recommendation regarding the physiological suitability for translocation of brushtail possums has general significance to translocation, reintroduction and management programmes for other mammals. Tarszisz et al.'s (2014) review of 120 translocation programmes found that only 11 programmes (9%) considered aspects of 'traditional' physiology in any way. Even for these, the vast majority involved only clinical health checks or post-release monitoring. Traditional physiological variables are not usually considered as part of initial pre-translocation planning (Tarszisz et al., 2014). Our study of brushtail possums is the first to our knowledge that has assessed basic physiological variation of geographically separated populations of a mammal as a consideration in terms of their suitability for translocation. Our data provide a physiological explanation for previous observations of responses by brushtail possums to translocation (Baker and Gemmel, 1999) and support the suggestion of Cooke et al. (2013) that individual variation in physiology and environmental tolerance can potentially impact on long-term translocation success. We conclude from our study that widespread mammals can have considerable geographic variation in basic physiological variables, and that knowledge of location-dependence of thermal, metabolic and hygric physiology is a potentially useful tool in selection of which populations would be most suitable to source individuals for conservation translocations.