Ectoparasite infestation patterns, haematology and serum biochemistry of urban-dwelling common brushtail possums

Urban environments support high concentrations of humans, domestic pets and introduced animals, creating conditions conducive to the transmission of parasites. This study compared patterns of ectoparasite infestation of the common brushtail possum Trichosurus vulpecula in urbanised Sydney (n = 161) to those from a remote woodland site (n = 18) from February 2005 – November 2006. We found diff erences in ectoparasite species prevalence between the two groups: the flea Echidnophaga myrmecobii was only found on urban possums and the tick Ixodes trichosuri was much more prevalent in the urban habitat, while the mite Atellana papilio was more prevalent on woodland possums. E. myrmecobii and I. trichosuri diff ered from other ectoparasites by showing an association with host sex and host age. Potential physiological costs of ectoparasitism to urban-dwelling possums were determined using multivariate analysis of haematology, serum biochemistry and body condition. Changes in serum iron levels were seen in the presence of both the tick Ixodes trichosuri and the flea E. myrmecobii, and E. myrmecobii was associated with elevated serum levels of the liver enzyme ALT. However, ectoparasite-related changes in haematology and serum biochemistry were not indicative of long-term pathology. In this urban possum population, the costs of ectoparasitism appear to be limited and unlikely to pose a major threat to the health of the population.

Urbanisation, the growth of cities to accommodate an increasing human population, is regarded as the most destructive anthropocentric pressure on wildlife, as it results in fragmentation or complete destruction of habitat (McKinney 2002, Garden et al. 2006. Less overt consequences of urbanisation such as changes to food webs, competition for resources and alterations in host -parasite relationships may also threaten the viability of wildlife populations (Patz et al. 2004, Spratt 2005, Bradley and Altizer 2007. In particular, urbanisation leads to a restructuring of species composition that may destabilise wildlife -pathogen relationships. Evidence of this exists in the growing number of studies in which urban areas act as a focal point for the emergence of zoonotic disease, including West Nile virus, Dengue fever (Dauphin et al. 2004, Weaver 2005, toxoplasmosis (Smith and Frenkel 1995, Meireles et al. 2004, Sukthana 2006, Lyme disease (Stanczak et al. 2004, Jobe et al. 2007) and leptospirosis (Ko et al. 1999). Despite the risks to the health of wildlife and humans, few studies have systematically investigated whether wildlife species experience a higher parasite burden as a result of living in an urban environment.
Ectoparasites are arthropods that complete at least one part of their life cycle on the surface of their host and include the ticks, lice, fl eas and mites. Th e survival and spread of ectoparasites may be favoured as a consequence of urbanisation (Spratt 2005, Bradley andAltizer 2007). Increasing population densities of wildlife that exploit anthropogenic sources of food and shelter may facilitate ectoparasite transmission, whilst the co-existence of humans, domestic pets and wildlife provides ectoparasites with a range of novel hosts (Dautel et al. 1991, Comer et al. 2001, Maetzel et al. 2005, Pearce and O'Shea 2007. Urban areas are also associated with warm microclimates due to lack of shade from vegetation, smog and effl uent from industry, retention of heat by impervious surfaces (Saaroni et al. 2000, Baker 2002) and the high number of human occupants (Torok et al. 2001). Th e warmer temperatures and reduced seasonality caused by this ' urban heat island eff ect ' may increase the survival, breeding success and activity of arthropod ectoparasites (Bradley and Altizer 2007).
As a consequence of inhabiting a disturbed environment with potentially higher rates of ectoparasite infection, wildlife inhabiting human settlements may be adversely impacted. By acquiring nutrients at the expense of the host, feeding on blood or skin tissue, ectoparasites impose a cost on the host (Price 1977). Ectoparasites may also infl ict physical damage, such as creating skin lesions, injecting salivary toxins into the wound and causing blood loss (Aeschlimann 1991); such physical damage may be costly to repair. Behavioural changes that deter ectoparasites, such as grooming or relocation of nest sites, can diminish reproductive success (Loehle 1995, Moore andWilson 2002) and immune responses to infestation are energetically demanding (Brossard et al. 1991). Ectoparasites are also vectors and intermediate hosts for a range of highly-pathogenic viral, protozoan, bacterial and rickettsial diseases (Walker et al. 1996). Owing to these detrimental impacts, ectoparasites have the potential to infl uence the growth, survival and reproduction of their host and consequently act as an important selective agent (Renaud et al. 1996).
Th e common brushtail possum Trichosurus vulpecula is one of the most abundant native mammals in urban Australia, having adapted to utilise anthropogenic resources (Hill et al. 2007). Despite success in colonising urban habitats, there are limits to the adaptability of this species. Tree clearing and predation by the European red fox Vulpes vulpes has been responsible for the decline of possums from more than 50% of their former range (Kerle 1992, Statham and Statham 1997. Th is study sought to determine whether adaptation to the urban environment enhanced ectoparasite infestation and whether such infestations were associated with changes in host physiology including body condition, haematological and serum biochemical parameters. To achieve this, we compared two populations of possums, one from metropolitan Sydney and the other inhabiting woodland remote from urban areas. Th is is the fi rst detailed investigation into multiple host -ectoparasite relationships of common brushtail possums, providing baseline data against which the health of the species may be compared as humans increasingly modify their original habitat.

Study sites
Th e principal urban study area was Taronga Zoological Park (33 ° 50 ′ S, 151 ° 14 ′ E) in the northern Sydney suburb of Mosman, Australia. Th e zoo is inhabited by an abundant free-ranging possum population that forage nightly on the leftover foodstuff s in the exhibits and on anthropogenic food from garbage bins. Also included in the ' urban ' fi eld site were the properties of Mosman residents who lived in proximity to the zoo ( Ͻ 1 km) and had experienced possum activity. Possums inhabiting this urban area achieved a population density of approximately 5.03 ha Ϫ1 , one of the highest recorded in Australia (Hill and Deane unpubl.). For comparison, trapping took place at a non-urban site located within Jenolan Caves Reserve Trust land, Blue Mountains, Australia (33 ° 49 ′ S, 150 ° 01 ′ E). Th e reserve was characterised by open eucalypt woodland and was inaccessible to the public. A wire fence prevented invasion by exotic species and control of feral animals was routinely performed throughout the enclosure.
Climate information was obtained from the closest Australian Bureau of Meteorology weather station to the urban study site (Observatory Hill, Sydney; station number 066062) and the woodland site (Oberon, Springbank; station number 063063). Records included monthly averages of rainfall, number of rainy days, maximum temperature and minimum temperature for each month of trapping ( Ͻ www. bom.gov.au/climate/averages Ͼ ).

Trapping
At the urban site, trapping was performed on a monthly basis between February 2005 and November 2006 (n ϭ 22 trapping surveys). At the woodland site, the remote location limited trapping to November 2005, May 2006 and October 2006 (n ϭ 3 trapping surveys). Treadle-operated cage traps (59 ϫ 22 ϫ 22 cm) were baited with apples and a rolled oats/peanut butter mixture and set at locations considered to be suitable possum ' runways ' (ground routes used to travel between trees). Th e top of each trap was covered in durable plastic sheeting to provide protection from the elements. Captured animals were transported either to the Veterinary and Quarantine Centre, Taronga Zoo or sampled on site at Jenolan Caves. All possums were permanently implanted with a passive integrated transponder to enable identifi cation of recaptured animals. Recaptured possums were included in the survey if the time since original capture exceeded three months (to avoid pseudo-replication). Ethics approval was obtained from the Macquarie University Animal Ethics Committee (no. 2004/16), the Zoological Parks Board of New South Wales Animal Care and Ethics Committee (no. 4c/08/04) and the NSW Dept of Environment and Conservation (no. S11107).

Animal handling
Possums were anaesthetised via a face-mask using 5% isofl urane in oxygen during sampling, eliminating the need for physical restraint and minimising stress for the animal. Age was estimated by observing the degree of molar tooth wear as per Winter (1980). Body measurements, weight and sex were recorded. Males were classifi ed as ' juvenile ' if their tooth wear class was 1 or 2 (corresponding to an age of Ͻ 2 years; Winter 1980) and their testis size was less than 15 mm long and less than 10 mm wide; ' adult ' males had larger testes (i.e. Ͼ 15 mm long and Ͼ 10 mm wide) and a tooth wear class of 2 or more. (In this species males can reach sexual maturity as early as 14 -15 months; Johnson et al. 2001). Females were classifi ed as ' juvenile ' if their pouch was undeveloped (indicating a lack of sexual maturity), which in all cases corresponded with a tooth wear class of 1 or 2. Females were considered ' adult ' if they were lactating or had a fully developed (non-breeding) pouch; this typically corresponded with tooth wear classes of 3 or greater (i.e. an age of 2 ϩ years), but a few individuals of tooth wear class 2 were found to be sexually mature. Th e age at which females reach sexual maturity varies between populations, but most females in most populations studied begin breeding sometime in their second year (How and Hilcox 2000). All females of breeding age were classifi ed as either ' with young ' if they carried off spring and/or were lactating, or ' without young ' if they exhibited neither of these features.

Ectoparasite collection and identifi cation
Ectoparasites were collected from six sites on the body: eyes, ears, nose, shoulders, rump and belly. To maximize consistency in sampling eff ort, a single researcher (NJH) performed all ectoparasite collections. Th e number of ticks and fl eas present at each site were counted and samples stored in sealed containers. Feeding mites were detected by parting the pelage on the rump in 5 -6 locations and dousing with paraffi n oil to allow collection with forceps. Fur-clasping mites were collected in hair samples removed with a fi ne-toothed comb. Th e small size of mites and the time-consuming methods necessary to locate them meant counts could not be performed. Ectoparasites were stored in 70% ethanol and later identifi ed under a stereo microscope to species-level on the basis of shared morphology with published descriptions of ticks (Roberts 1970), fl eas (Dunnet and Mardon 1974) and mites (Domrow 1987(Domrow , 1992.

Estimation of body condition
Body mass (M b ) was measured on electronic scales to the nearest single gram. Skeletal body length (L) was measured from the tip of the nose to the cloaca. To ensure consistency, only a single researcher (NJH) performed measurements of possum skeletal structures. A typical body condition index (BCI) estimate, such as that used for the closely related mountain brushtail possum, Trichosurus caninus (Viggers et al. 1998), is to divide body mass by length: BCI ϭ M b / L. Possums are sexually dimorphic, with males signifi cantly heavier than females (but not signifi cantly longer), so the typical BCI estimation could lead to over-estimating the body condition of males. Th erefore, we instead calculated a " scaled mass index " , SMI Green 2009, 2010), for each population sub-group (adult males, adult females, juvenile males and juvenile females), where the body mass was scaled to the mean body length for each sub-group.

Blood collection and analyses
Approximately 5 ml of blood was drawn from the lateral tail vein and distributed between two vacuette tubes; one lined with ' EDTA ' for haematological analysis and a serum clot activator tube for biochemical analysis. After clotting, the serum clot activator tube was centrifuged at 200 ϫ g for 10 min and serum collected in a microtube. All testing was conducted within 72 h of blood collection at PaLMS (Pacifi c Laboratory Medicine Services, NSW, Australia).
Haematological parameters were measured on a Beckman -Coulter counter. To limit erroneous measurements arising from the use of equipment designed for human blood cells (Whittington and Comer 1984), instrumentation was optimised for use with marsupial cells. Electronic white cell counts were verifi ed against diff erential counts from blood fi lms to confi rm accuracy of the equipment. Red cell parameters measured included haemoglobin concentration, erythrocyte count, mean corpuscular volume (MCV), mean corpuscular haemoglobin (MCH) and mean corpuscular haemoglobin concentration (MCHC). White cell parameters measured were lymphocytes, neutrophils and eosinophils. Also measured were mean platelet volume (MPV) and iron parameters including iron and iron saturation.
Serum biochemical parameters were measured on a modular analyzer, optimised for use with marsupial serum. General biochemistry measurements included sodium, potassium, creatinine, urate, total protein, albumin, total bilirubin, calcium, phosphate, magnesium, lipase and hormones, alkaline phosphatase (ALP), aspartate aminotransferase (AST), alanine transaminase (ALT) and g-glutamyl transferase (GGT). Assays for bilirubin and GGT were colorimetric, whereas those for ALP, AST and ALT were UV based protocols. Also measured were lipid chemistry parameters including cholesterol and triglyceride.

Possum population demographics and climate data
Pearson ' s χ 2 -tests were used to investigate whether the urban possum sub-groups (zoo and backyard) were similar with respect to age, gender and maternal status. Th e same test was used to assess similarity between the possum populations in urban and woodland habitats. One-way ANOVA was used to assess whether body condition (as estimated by SMI) varied between urban and woodland populations.
Th e average rainfall, number of rainy days, maximum temperature and minimum temperature at each habitat was determined for the period of trapping and compared with a one-way ANOVA.

Ectoparasite data
Prevalence, defi ned as the number of infected individuals expressed as a percentage of the study population (Bush et al. 1997), was determined for each ectoparasite species in each trapping area. Abundance, defi ned as the number of individual parasite specimens occurring on the host (Bush et al. 1997), was determined for ticks and fl eas, but not mites (since mites were scored as present/absent but not counted). Th e abundance of each ectoparasite species was not normally distributed (Shapiro -Wilk test of normality, p Ͻ 0.05) and when plotted revealed an aggregated distribution, whereby the majority of animals were not infected with an individual species of ectoparasite. It was therefore more informative to use either prevalence or intensity (the number of parasite specimens per ' infected ' individual; always a number Ͼ 0) for comparative purposes (e.g. across trapping sites).
Pearson ' s χ 2 -tests were used to identify diff erences in prevalence of ectoparasites between the trapping sites. For three ectoparasite species with prevalence greater than 25% (in any trapping site), variables infl uencing the prevalence of ectoparasite species were investigated using generalized linear mixed models (GLMMs). In this analysis the individual microchip ID of each possum was included as subject variable, to control for possible eff ects of re-trapped individuals on the analysis. Factors included in each model were trapping site (woodland, zoo or backyard), host demography (sex, age and maternal status of females), host body condition (scaled mass index, SMI), and season. Factors were centred around their respective mean values ( ' grand mean centering ' ) unless there was more variation within animals than between animals, in which case factors were centred around individual animal means ( ' group mean centering ' ).
Species richness by host individual was defi ned as the total number of parasite species occurring on the host individual (Magurran 1988). Th e species richness of

Possum population
In total, across all trapping areas, there were 278 capture events of 179 individual possums. Th e majority of these were at Taronga Zoo, where 140 individuals were trapped over 219 capture events. Th e backyard group consisted of 21 individual possums, captured on 22 occasions. Th ere were 37 capture events at the woodland site, consisting of 18 individual possums. Because of the large number of recapture events, only fi rst capture data was used for comparisons of the sex structure of the populations. However, because both the age of individual possums and maternal status of females changed over the time-course of the study, all data points were used in comparisons of age and maternal status across the diff erent possum populations. A summary of the demographic data is shown in Table 1.
Th e mean ( Ϯ SEM) scaled mass index (SMI) of zoo possums was 2297 Ϯ 52 g (n ϭ 227); possums trapped in backyards had a mean SMI of 2617 Ϯ 155 g (n ϭ 22); and woodland possums had mean SMI of 2320 Ϯ 145 g (n ϭ 27). Th is index of body condition was not signifi cantly diff erent across the three trapping sites (one-way ANOVA; F 2,273 ϭ 1.711, p ϭ 0.183). ectoparasites per host individual was compared across habitats using ANOVA. In addition, the variables infl uencing species richness per host individual were examined using a GLMM with the factors trapping site, host sex, host age, maternal status and SMI.
Th e large number of health variables measured (29 in total; 11 haematological measurements, 17 serum biochemical parameters and scaled mass index), were reduced to limit statistical analysis to biologically meaningful groups of variables. We hypothesised that ectoparasite infestation was most likely to aff ect red blood cell parameters, immune function, iron levels and body condition, and that severe infestation could potentially compromise host liver function. Selection of the specifi c variables for investigation was assisted by using a Pearson ' s correlation matrix to identify variables that were correlated, although in some cases groupings were selected for their biological importance, rather than solely on correlations data (e.g. variables within both liver function parameters and immune parameters were only weakly, though significantly, correlated). Th e following groupings or single variables were selected for analysis: red blood cell parameters (variables: haemoglobin, erythrocyte count), iron status (variables: iron levels, iron saturation), immune parameters (variables: lymphocytes, neutrophils and eosinophils), liver function (variables: ALT, AST, total bilirubin) and scaled mass index.
Multivariate ANOVA (MANOVA) was used to examine the eff ect of ectoparasite prevalence on red blood cell parameters, iron status, immune parameters and liver function of urban possums only (the small study population in the woodland habitat precluded use of such statistics). Th e independent variables in this analysis were presence/absence of each of six ectoparasite species (analysed simultaneously to account for potential interactions between ectoparasites), species richness (to allow for potential interactions between species prevalence and the number of species parasitising a host to be assessed), age, a combined sex/female lactation status variable, season and number of re-captures (these variables are known to aff ect haematology and blood biochemistry of possums; Barnett et al. 1979, Presidente and Correa 1981, O ' Keefe and Wickstrom 1998. Two-way interactions between ectoparasites were only included in the model if the number of individuals hosting both parasites was greater than twelve. MANOVA was also used to examine the eff ect of the abundance of the most abundant ectoparasite (a fl ea) on red blood cell parameters and iron status.
Univariate ANOVA was used to examine the eff ect of ectoparasite prevalence on scaled mass index of urban possums. Again, the independent variables were presence/ absence of each of six ectoparasite species, species richness, age, combined sex/lactation status, season and number of re-captures. A seventh ectoparasite species (the fl ea Ctenocephalides felis ) was excluded from these analyses as it was only found on a single possum in the urban habitat.
Statistical analysis was performed using SPSS Statistics, ver. 20. For all analyses, an alpha value Յ 0.05 indicated a signifi cant diff erence. All data are reported as mean Ϯ SEM (standard error of the mean), unless otherwise noted.

Ectoparasite fauna of possums
In total, seven species of ectoparasites were identifi ed from 179 possums (161 from urban and 18 from woodland habitat). Th e ectoparasite fauna of possums included three species of tick, Ixodes trichosuri , Ixodes tasmani and Ixodes holocyclus ; two species of mite, one astigmatid, Atellana papilio and one mesostigmatid mite, Trichosurolaelaps crassipes ; and two species of fl ea, Echidnophaga myrmecobii and Ctenocephalides felis . Th e two fl ea species were unique to urban-dwelling possums, but all other ectoparasite species were found in both urban and woodland possum populations. All ticks, fl eas and mites detected in this study had been previously identifi ed from this species of possum.

Species richness, prevalence and mean intensity of ectoparasites
Mean species richness of ectoparasites per host individual was not signifi cantly diff erent between the two urban subgroups (zoo and backyard; one-way ANOVA F 1,239 ϭ 0.246, p ϭ 0.62). Th erefore further comparisons of ectoparasite species richness per host individual were made between the urban and woodland populations. Urban possums were more likely than woodland possums to host three or more ectoparasite species (Table 2a), but the mean ectoparasite species richness per host was not signifi cantly diff erent between possums inhabiting urban and woodland habitat (F 1,276 ϭ 1.648, p ϭ 0.200). Th e GLMM analysis of factors aff ecting the species richness per host confi rmed that there was no signifi cant eff ect of habitat, as well as no eff ect attributable to host age or host scaled mass index (SMI). However, species richness per host individual was signifi cantly aff ected by host sex (GLMM F 1,145.5 ϭ 4.315, p ϭ 0.028) and season (GLMM F 1,79.6 ϭ 6.859, p ϭ 0.011). More females than males were ectoparasite-free (35%, compared with 18% of males); and one or more ectoparasite species were more common on males than females (Table 2b). In most seasons, possums were most likely to host a single ectoparasite species (45 -55% of the population aff ected; Table 2c). However, in spring the proportion of the population that was ectoparasite-free was equal to that hosting a single species (35%). Winter appeared to be the peak season for trapping possums hosting two species of ectoparasite, but possums hosting three or more parasites were most commonly trapped in spring or summer (Table 2c). A summary of the prevalence of ectoparasite species in each trapping area (i.e. zoo. backyard or woodland) is presented in Table 3. χ 2 -analysis revealed signifi cant differences in prevalence between trapping areas for the fl ea E. myrmecobii , which was not present in woodland habitat ( χ 2 ϭ 26.104, DF ϭ 2, n ϭ 278, p Ͻ 0.001); for the mite A. papilio , which was more prevalent in woodland habitat ( χ 2 ϭ 24.196, DF ϭ 2, n ϭ 278, p Ͻ 0.001); and for the tick I. trichosuri , which was most prevalent on backyard possums ( χ 2 ϭ 7.939, DF ϭ 2, n ϭ 278, p ϭ 0.019).
Th e prevalence of most ectoparasite species in most areas was low ( Ͻ 20%; Table 3). Species with prevalence greater than 25% of the population were the tick I. trichosuri (41% prevalence in backyard possums), the fl ea E. myrmecobii in urban possums (27% prevalence in backyards and 43% at the zoo) and the mite A. papilio (43% prevalence in woodland possums).
Th e mean intensity (mean number of parasites per infested individual) of each of the tick and fl ea species is presented in Table 4 (frequency histograms of intensity are available in Table 4. Mean intensity of ectoparasite species (ticks and fl eas) in the different trapping areas. Mean intensity is defi ned as the mean number of conspecifi c parasites living on an infected host.
indicates the total number of hosts trapped in each trapping area; ' n ' indicates the number of hosts infected with each parasite species in each trapping area. ' CI ' indicates the lower and upper limits of the 95% confi dence interval and ' V:M ratio ' is the variance to mean ratio. * There were 94 hosts with the presence of E. myrmecobii recorded, but only 87 with intensity of the parasite recorded (i.e. seven cases were missing).
Supplementary material Appendix 1 Fig. A1 -A4). Th e tick species all had mean intensities in the range of one to two ticks per possum host (Table 4) (Table 4), and the high mean intensity was principally due to a single possum hosting 115 individual fl eas (hence the very high variance: mean ratio for the intensity of this ectoparasite). For possums trapped at Taronga Zoo, E. myrmecobii intensity was skewed towards lower intensities, with 60 out of 84 individual hosts (71%) infested with fewer than 15 fl eas per possum, and 47 individuals (56% of infested hosts) having fewer than 10 fl eas.

Factors affecting infestation by the most prevalent ectoparasite species
Th e GLMM analysis of factors aff ecting the presence or absence of I. trichosuri indicated that the presence of this species was aff ected by host sex (GLMM F 1,116.11 ϭ 4.389, p ϭ 0.038), with males nearly twice as likely to harbour I. trichosuri than females (Table 5a). Th e presence of the fl ea E. myrmecobii was signifi cantly aff ected by trapping site (GLMM F 1,188.99 ϭ 18.283, p Ͻ 0.001) and an interaction between host sex and host age (GLMM F 1,229.91 ϭ 4.733, p ϭ 0.031). Approximately half of adult male possums hosted E. myrmecobii , compared to one fi fth of juvenile males, and around one quarter of juvenile and adult females (Table 5b). E. myrmecobii was found only on urban possums, with a higher prevalence on possums trapped at Taronga Zoo than in suburban backyards (Table 3). A trapping site eff ect was also found for the mite A. papilio (F 1,179.33 ϭ 5.477, p ϭ 0.020), with this parasite signifi cantly more prevalent in woodland habitat (43%) than in either urban habitat (Table 3). Separate analyses of females only indicated no eff ect of maternal status on prevalence of any of these three parasites.

Effect of ectoparasite prevalence and abundance on the health of urban-dwelling possums
A summary of the haematological and serum biochemical parameters of possums according to habitat type (urban or woodland) is presented in the Supplementary material Appendix 2 Table A1. Because of the low sample numbers for woodland possums, only the health parameters of urban possums were investigated using MANOVA or ANOVA. Host sex and lactation status interacted with host age to signifi cantly aff ect the red blood cell parameters haemoglobin and erythrocyte count (analysed together using MANOVA; Wilks λ ϭ 0.945, F 2,187 ϭ 5.407, p ϭ 0.005, partial eta-squared ϭ 0.055). Adult male possums had the Table 5. Factors signifi cantly affecting the prevalence of (a) I. trichosuri (host sex); and (b) E. myrmecobii (interaction between host sex and host age, in addition to trapping site effect shown in Table 3    highest haemoglobin and erythrocyte counts, and juvenile males had the lowest (Table 6a). Females, regardless of age or lactation status, had similar red blood cell parameter values, which were intermediate between adult and juvenile male values (Table 6a). Red blood cell parameters were unaff ected by the presence of any of the haematophagus ectoparasites. Iron status (iron levels and iron saturation) was signifi cantly aff ected by an interaction between the haematophagus ectoparasites I. trichosuri and E. myrmecobii (Wilks λ ϭ 0.965, F 2,184 ϭ 3.300, p ϭ 0.039, partial etasquared ϭ 0.035). When I. trichosuri was absent, both iron levels and iron saturation were lower in possums hosting E. myrmecobii than those without (Table 6b). When I. trichosuri was present, iron parameters were higher in hosts of E. myrmecobii compared to possums without the fl ea (Table 6b). Iron status was also aff ected by the sex and lactation status of the possums (Wilks λ ϭ 0.926, F 4,368 ϭ 3.618, p ϭ 0.007, partial eta-squared ϭ 0.038), with lactating during the study season was 14.7 ° C, signifi cantly higher than woodland habitat, 5.6 ° C (F 1,22 ϭ 54.77, p Ͻ 0.01).

Discussion
Previous studies have demonstrated that animals inhabiting human-managed habitat including game parks (Ezenwa 2003(Ezenwa , 2004 and conservation reserves (Lebarbenchon et al. 2006) are susceptible to enhanced levels of parasite exchange due to the high density of sympatric species atypical in natural environments. In addition, the diversity of hosts in an environment can also be crucial in determining the species richness of ectoparasites and may even outweigh intrinsic factors such as body size of the host (Krasnov et al. 2004). In the present study, only urban-dwelling common brushtail possums were found to host three or more species of ectoparasites per host individual (Table 2a). Th is suggests that opportunities for ectoparasite transmission may be greater in the urban environment, in part due to higher possum population densities, but also potentially due to the concentration of other hosts in the study area, including humans, domestic pets, captive zoo animals and other urban-dwelling wildlife.
Most ectoparasites hosted by brushtail possums occur at low prevalence, but three species were present on more than 40% of the population in at least one trapping area: the mite Atellana papilio , the tick Ixodes trichosuri and the fl ea Echidnophaga myrmecobii (Table 3). Th e fur-clasping mite A. papilio occurred with higher prevalence in woodland compared to urban habitat; this concurs with New Zealand studies, where A. papilio was the most prevalent ectoparasite, occurring on 81% (Stankiewicz et al. 1996) and 98% (Clark et al. 1997) of possum populations examined. Th e cooler climate may explain the dominance of the mite on possums from woodland and New Zealand habitat, compared to possums from urban Sydney. Th e denser coat and woollier tail of possums inhabiting cooler terrain (Kerle 1991) may off er a more favourable ' habitat ' for fur-clasping mites, compared to possums adapted to the urban environment and the associated warmer microclimate. Th e host ' s fur is used by A. papilio as an important site of attachment for all life stages from egg to adult (Clark 1993) implying that it may be the limiting factor causing the lower prevalence of the mite in urban-dwelling possums.
Th e fl ea Echidnophaga myrmecobii only infected possums from urban habitat (both at Taronga Zoo and in suburban backyards). It was the most prevalent ectoparasite in urban areas, and also had the highest mean intensity of all the ectoparasites in this study. Known as the common ' stickfast ' fl ea, this species possesses a broad host range encompassing native wildlife and introduced mammals such as rabbits, cats and dogs (Dunnet and Mardon 1974). Th e ability of generalist parasites to exploit a range of unrelated host species enables them to succeed in habitats that support a diversity of hosts (Ezenwa 2003, Krasnov et al. 2007. Th e transmission of E. myrmecobii may be favoured in this particular urban environment because of the overlapping host populations from the metropolitan zoo and residential areas. Th e less prevalent fl ea, Ctenocephalides felis, originates from domestic cats and has shown an ability to colonise a variety of introduced and native mammals and also humans (Dunnet and Mardon 1974). Th e absence of E. myrmecobii females having generally higher iron parameters than both males and non-lactating females (Table 6b). Diff erences in iron concentration and iron saturation due to diff erences in ectoparasite presence were of similar magnitude to diff erences between demographic groups.
Th e abundance of the most abundant ectoparasite, the fl ea E. myrmecobii , had no signifi cant eff ect on either red blood cell parameters or iron status.
Th e immune parameters lymphocyte count, neutrophil count and eosinophil count were signifi cantly aff ected only by the age of the host possums, and not by the presence of any ectoparasites. Adults had lower numbers of lymphocytes, but higher numbers of neutrophils and eosinophils, than juveniles (Table 6c).
Th e liver function parameters (ALT, AST, total bilirubin) were signifi cantly aff ected by the presence of E. myrmecobii (Wilks λ ϭ 0.949,F 3,196 F 3,196 ϭ 3.415, p ϭ 0.018, partial eta-squared ϭ 0.050). Th ese signifi cant eff ects were due to signifi cant diff erences in ALT alone, rather than the grouped parameters. Possums hosting E. myrmecobii exhibited higher mean ALT values than possums without this ectoparasite (Table 6d). By contrast, possums infested with A. papilio , I. tasmani , I. trichosuri or T. crassipes had lower mean ALT measurements than possums without these species (Table 6d). Liver function parameters were also aff ected by season (Wilks λ ϭ 0.867, F 9,477.163 ϭ 3.195, p ϭ 0.001, partial eta-squared ϭ 0.046) and species richness per host individual (Wilks λ ϭ 0.943, F 3,196 ϭ 3.964, p ϭ 0.009, partial eta-squared ϭ 0.057). Seasonal diff erences were seen in all liver function parameters, with a peak in ALT, AST and total bilirubin observed in possums trapped during autumn. However, species richness only signifi cantly aff ected ALT levels, with possums hosting one to three species of ectoparasites having a higher mean ALT measurements than those without any ectoparasites, Th e highest mean ALT was measured in possums hosting a single species of ectoparasite (Table 6d).

Abiotic characteristics of each habitat
Th e mean maximum temperature at the urban site during the study season was 23.3 ° C, signifi cantly higher than woodland habitat, 17.3 ° C (F 1,22 ϭ 18.18, p Ͻ 0.01). Similarly, the mean minimum temperature at the urban site indicators were primarily aff ected by demographic variables such as sex, age and lactation status, as well as season (Table  6). Haemoglobin concentrations and erythrocyte count were higher in adult males than females; this was comparable with previous studies of this species (Barnett et al. 1979, Presidente andCorrea 1981). However, in a previous study juvenile males had also higher haemoglobin levels and erythrocyte counts than juvenile females (Presidente and Correa 1981); the converse was true in the present study. Serum iron concentrations and iron saturation were highest in lactating females, which is consistent with a previous study of the marsupial quokka, Setonix brachyurus (Kaldor and Morgan 1986). Adult possums had lower numbers of lymphocytes but higher numbers of neutrophils than juveniles, which was similar to patterns seen in the tammar wallaby, Macropus eugenii (McKenzie et al. 2002), but contrasted with the possum population studied by Presidente and Correa, which did not exhibit age diff erences in diff erential white cell counts. Th e seasonal eff ect of a peak in all liver function parameters in autumn is most likely related to increased energetic demands of mating (urban possums breed in both autumn and spring), as well as the higher incidence of females with immature pouch young (autumn being the peak time for births in the urban population; Hill and M. Deane unpubl.). Bilirubin in particular is known to increase during human pregnancy (Lentner 1984), and in tammar wallabies, total bilirubin levels are signifi cantly higher in autumn (a time of peak births) than summer (McKenzie et al. 2002).
In a few instances there was evidence of some impact of ectoparasite infestation on host haematology or serum biochemistry. Serum iron levels and iron saturation of the blood of urban possums were aff ected by an interaction between the presence of E. myrmecobii and I. trichosuri , two haematophagus parasites. When the tick alone was present, iron parameters were lower than in unparasitised possums (Table 6b); this was presumably due to the blood loss caused by the tick ' s bloodmeals. A reduced blood iron concentration has previously been reported for tick-infested cattle (O ' Kelly et al. 1971, O ' Kelly andKennedy 1981). Th e presence of E. myrmecobii was associated in a smaller decline in iron concentration and iron saturation than the presence of I. trichosuri (compared to possums that did not host either ectoparasite; Table 6b); this may have been due to increased blood production to replace blood lost from the potentially more frequent fl ea bloodmeals. In the presence of both ectoparasites, both iron parameters were higher than in the presence of the tick alone (Table 6b); this may be evidence of increased blood production in the presence of E. myrmecobii off -setting the decline in iron levels caused by I. trichosuri infestation. However, the mean changes in serum iron concentration and iron saturation seen in response to ectoparasite infestation were equivalent in magnitude to the diff erences between lactating females, non-lactating females and males. It therefore seems unlikely that infestation with either E. myrmecobii or I. trichosuri (or both) would result in changes in serum iron indicative of a clinical state of anaemia (or hyperaemia) in most possums.
In the present study, the liver function parameters analysed were alanine aminotransferase (ALT), aspartate aminotransferase (AST) and total bilirubin levels. Elevated levels of these three liver function parameters are often associated with liver damage and/or disease. Possums hosting one or two ectoparasite species had elevated ALT levels compared to possums with no parasites (Table 6d), indicating a possible and C. felis from possums inhabiting woodland habitat in this study and non-urban environments in South Australia (O'Callaghan and Moore 1986), Victoria (Presidente et al. 1982) and New Zealand (Stankiewicz et al. 1996, Clark et al. 1997 highlights the role of human settlement in mediating transmission of these fl eas. Th e success of the fl ea E. myrmecobii among urban possum populations may also refl ect the fl ea ' s adult male-biased pattern of infestation. Th e signifi cantly higher prevalence of E. myrmecobii amongst adult males compared to females corresponds with a wide range of ectoparasite -host interaction studies (Mooring et al. 1996, T ä lleklint-Eisen and Eisen 1999, Soliman et al. 2001. Sex-diff erences in behaviour, physical characteristics and immunocompetence have been proposed to explain this phenomenon. For example, males often maintain larger home ranges and engage in territorial behaviour more than females, facilitating heightened interaction with parasites (Zuk and McKean 1996). In support of this explanation, GPStracking of twelve urban-dwelling possums revealed that males occupied signifi cantly larger home ranges (1.21 -2.41 ha) than females (0.27 -0.61 ha) (Hill and Deane unpubl.). Males typically inhabit home-ranges overlapping those of several females, and the polygamous mating system of brushtail possums means that adult males associate with multiple females, potentially exposing them to more ectoparasites. Sexual dimorphism often manifests in larger-bodied males, providing ectoparasites with a larger surface area and diversity of refugia (Moore and Wilson 2002); in the present study males hosting E. myrmecobii were signifi cantly larger than their fl ea-free counterparts. Lastly, the inhibitory eff ect of testosterone on the immune system can lower the resistance of male hosts and make them more susceptible to parasite infestation (Folstad and Karter 1992). Th e lower prevalence of E. myrmecobii amongst juvenile male possums (compared to adult males) can also be explained by the above factors: juvenile males typically remain within the (smaller) maternal home range until maturity (Isaac 2005), are smaller and have lower testosterone levels than those of adult male possums, and are not engaging in mating activities that could promote the spread of ectoparasites.
Th e tick Ixodes trichosuri , sometimes known as the ' possum tick ' , similarly occurred with signifi cantly higher prevalence in urban habitat, specifi cally in possums trapped in backyards. Like E. myrmecobii , I. trichosuri displayed a male-biased pattern of infestation, with sex diff erences in home-range size, body size and immune function again likely to explain the higher prevalence of this tick amongst male possums. While this tick species was originally isolated and described from a brushtail possum host (Roberts 1970), it has since been identifi ed parasitising on other native mammal hosts, including bush rats Rattus fuscipes (Spratt and Haycock 1988), koalas Phascolarctos cinereus (Vilcins et al. 2009) and eastern bettongs Bettongia gaimardi (Portas et al. 2014). It has also been putatively identifi ed, in the larval stage only, on domestic dogs and cats (Baxter et al. 2009). Like E. myrmecobii , I. trichosuri appears to fl ourish in an environment where multiple possible hosts are present. Th e ability of the larval stage to infest domestic dogs and cats may explain the higher prevalence of I. trichosuri in urban backyards, where domestic pets may serve as a parasite ' reservoir ' .
Th e secondary focus of the present study was an examination of the haematological and biochemical health indicators of common brushtail possums. In urban possums, these eff ect on liver metabolism of multiple species of ectoparasite. However, the exact eff ect of specifi c ectoparasites is diffi cult to determine, since possums hosting E. myrmecobii exhibited higher ALT levels than possums without this ectoparasite (Table 6d), but possums infested with A. papilio , I. tasmani or I. trichosuri had lower ALT levels than possums without these species (Table 6d). Th ere was no evidence of elevated liver function parameters in association with any of the ixodid tick species, despite tick infestation being known to disrupt host liver metabolism of the host (Kaufman 1989), most likely as a direct eff ect of a toxin secreted by the parasite (O ' Kelly et al. 1971). In humans, liver function may also be impacted by the pathogens spread by ectoparasites, such as the tickborne bacteria Babesia spp. (Kjemtrup and Conrad 2000) and Borrelia spp., which cause Lyme disease (Horowitz et al. 1996). It appears that in this urban population of possums, ectoparasite infestation does not cause large changes in liver function parameters, since all ectoparasite eff ects were smaller in magnitude than the diff erences observed across seasons.
In conclusion, this study provides insight into multiple host-ectoparasite relationships, ectoparasite infestation patterns and the haematology and serum biochemistry of common brushtail possums, with a particular focus on an urban population successfully adapted to human settlement. Urban conditions favoured a higher prevalence of the fl ea E. myrmecobii and the tick I. trichosuri on urban-dwelling possums (compared to woodland possums). Both ectoparasites were associated with host sex and host age in a manner that facilitated transmission. Although the presence of several ectoparasite species was found to impact on health parameters of urban possums, such as iron status and liver function, there was no evidence of long-term pathology. Th ese fi ndings demonstrate that changes in host -ectoparasite dynamics resulting from adaptation to the urban environment do not pose a major threat to the health of possums in urban Sydney.