Movement and ranging behaviour of long-nosed potoroos (Potorous tridactylus) in south-west Victoria, Australia

Introduction

The movement of an animal influences many aspects of its ecology, from individual fitness and reproductive success to population dynamics, geneflow, and species distributions (Bohonak 1999; Bowler and Benton 2005; Elith and Leathwick 2009). Therefore, a detailed understanding of animal movements is fundamental to developing and enacting effective conservation strategies. For example, prior knowledge of home range size can be used to ensure survey sites are spatially independent for occupancy modelling (MacKenzie et al. 2002), a practice often used to monitor populations of conservation concern through time (Ahumada et al. 2013), or before and after a management intervention such as fire (Pons et al. 2003) or invasive predator control (Robley et al. 2014). This same knowledge can be used to parameterise sophisticated decision support tools, such as agent-based models (DeAngelis and Mooij 2005; Hradsky et al. 2019), to model the potential benefits and consequences of different management regimes. Finally, home range estimates may be used as informative priors for density estimation when individuals are able to be uniquely identified using spatially explicit capture–recapture modelling within a Bayesian framework (Royle et al. 2013). Despite these benefits, a detailed understanding of home range size for many species is lacking.

Historically, insight into an animal’s home range was generated either through direct observation (if they were individually identifiable), from capture locations (Kitchener 1973), or by attaching very high frequency (VHF) transmitters to individuals and relocating them using a VHF receiver (Long 2001). Both approaches are often characterised by small sample sizes, logistical challenges, and analytical issues. More recently, Global Positioning System (GPS) loggers that are physically attached to an animal are increasingly being used to generate high-volume and high-quality movement data (Cagnacci et al. 2010; Hebblewhite and Haydon 2010; Tomkiewicz et al. 2010). Although weight restrictions have historically limited the application of this technology to medium- to large-sized animals (Merrill and Mech 2003; Dussault et al. 2005), technological advances reducing the size and weight of GPS loggers have expanded the suite of species that can be tracked using GPS loggers (Price-Rees et al. 2013; Stevenson et al. 2013; Zhang et al. 2019).

A wide range of analytical approaches have been developed to take advantage of ever-increasing sample sizes of animal locations. The most basic of these, the minimum convex polygon (MCP), remains commonly used despite its shortcomings (Harris et al. 1990; Börger et al. 2006), because it allows for comparison of home range areas with older studies (Nilsen et al. 2008). Kernel density estimators (KDE) represent an advance on polygon-based estimators, because they use the density of fixes to estimate the area an animal ranges in addition to how intensely different areas are used, referred to as a utilisation distribution (UD) (Worton 1989; Kranstauber et al. 2012). KDEs commonly ignore the temporal structure of a movement dataset and often require fixes to originate from a regular sampling schedule or to be temporally independent (Fieberg 2007; Kranstauber et al. 2012). More recently, the development of the dynamic Brownian bridge movement model (hereafter referred to as ‘dBBMM’) has further improved upon traditional KDEs by incorporating both temporal and spatial structures of tracking data (Horne et al. 2007; Kranstauber et al. 2012). This development allows for variable movement patterns and irregular fix schedules, better capturing movement pathways compared with traditional KDE estimators (Horne et al. 2007; Kranstauber et al. 2012).

The long-nosed potoroo (Potorous tridactylus – hereafter ‘potoroo’) is a cryptic and largely nocturnal species that utilises dense vegetation, making direct observations difficult (Bennett 1993; Claridge et al. 2007; Norton et al. 2010). Despite belonging to a faunal group that has been highly disadvantaged by the landscape changes and introduction of invasive predators that followed European invasion in Australia (Burbidge and McKenzie 1989; Woinarski et al. 2015), the behaviour and movements of this species remains poorly understood. Potoroos are considered solitary yet non-territorial (Seebeck et al. 1989; Long 2001), and occur in a patchy distribution along the eastern coast of Australia, from south-eastern Queensland through to the south-eastern coast of South Australia and Tasmania (Johnston 2008). Their small size and generally low trapping success have likely contributed to the small number of studies that describe their home range.

In addition to the myriad of ecological factors that can influence potoroo movements and behaviour (Claridge et al. 1992; Bennett 1993; Norton et al. 2011), variability in survey technique, analytical approach, and sample sizes among the few studies that have estimated potoroo home range size makes direct comparisons challenging. For example, one study of Tasmanian potoroos reported home ranges of 19.4 ha for males and 5.2 ha for females using trapping data and MCP’s (Kitchener 1973), whereas a study of Victorian potoroos reported much smaller home ranges (4 ha for males and 2.9 ha for females) based on similar sample sizes but using VHF tracking and a KDE estimator (Long 2001). Another potential reason for this discrepancy may be due to the considerable variation in body mass displayed by potoroos throughout its geographic range (Norton et al. 2010; Frankham et al. 2011). Individuals in south-west Victoria are among the smallest described, whereas Tasmanian potoroos weigh almost twice as much on average (Hird 1996; Long 2001; Frankham et al. 2011). Regardless, females are consistently estimated to have smaller ranges than males (Kitchener 1973; Hird 1996; Long 2001), a characteristic potoroos share with many other macropod species (Fisher and Owens 2000). To date, there have been no reported instances of GPS datalogger technology being used to generate home range estimates for potoroos, and no instances of more sophisticated home range estimators being used to estimate the home range size of this species.

Here we present the first study to use GPS dataloggers to estimate the home range size of potoroos. Our aim was to provide robust estimates of home range size, evaluate the influence of physiological factors such as body mass and sex on home range size, and describe patterns of home range overlap between and within sexes, in a population of potoroos in south-west Victoria, mainland Australia.

Materials and methods

Site description

Our study was conducted in the western Heathy Woodlands of the Great Otway National Park (38.54S, 143.47E), Victoria, Australia (Fig. 1). This locality is 130–250 m above sea level and has a mild, temperate climate: maximum daily temperatures average 26.1°C in summer and 11.6°C in winter; annual rainfall averages 538 mm with more rainfall in winter and spring (June to November) relative to the warmer months of summer and autumn (December to May) (Bureau of Meteorology 2020). The overstorey is dominated by Eucalyptus baxterii and E. obliqua, with a mid-storey dominated by Banksia marginata, Epacris impressa, Xanthorrhoea australis, and Leptospermum continentale, with Melaleuca squarosa becoming the dominant mid-storey species in drainage lines. The understorey predominantly comprises a diverse array of herbs, graminoids, and lichens.

Fig. 1.

Study area and location of all potoroo fixes collected during this study in reference to (a) Australia and (b) Victoria.

Results

We successfully captured and attached GPS collars to 39 individual potoroos (23 males and 16 females) (Figs 1 and 2, Table 1). Although traps were open continuously, no potoroos were captured during the day. One collar failed to successfully collect any GPS fixes and was discarded from this study. Collar deployments are described in the Supplementary Information (Tables S1–S3). Beyond minor hair loss around the neck of some individuals, no observable negative effects of collar attachment, such as loss of body mass, were observed. Individuals varied in mass, but a two-sample Wilcoxon rank sum test revealed there was no significant difference in mean body mass between males and females (w = 219, P = 0.16). Males weighed 805 g (95% CI: 748–860) and females weighed 765 g (95% CI: 705–815) on average.

Fig. 2.

95% home ranges of long-nosed potoroos collared in 2020 (top two panels), 2021 (middle panels) and 2022 (bottom panels), as estimated by dBBMM analysis.

Given any animal larger than 500 g was suitable for inclusion in our study, it is likely our sample included both resident adults as well as subadults yet to disperse (Table 1). Halfway through deployment, one male (‘Doody’) began to display extra-territorial movements and eventually dispersed into adjacent habitat approximately 2 km away. Therefore, only movement data before the first example of potential dispersal movement behaviour (assessed visually and characterised by marked departures from previously consistent movement patterns) were included in further analysis. The dispersal behaviour we observed in one individual male weighing 620 g (Doody) concurs with the suggestion of Hughes (1964) that males weighing less than 640 g are likely to be subadults. However, because Hughes (1964) based his study upon Tasmanian potoroos, which are larger on average relative to the potoroos found in south-eastern Australia, it is possible that potoroos in our study region could reach sexual maturity at lower body masses. Four of the 23 males and three of the 15 females included in this study were deemed to be subadults according to this criterion. An additional two-sample Wilcoxon rank sum test with subadults removed from the sample again suggested no significant difference in mean body mass between adult males and females (w = 158, P = 0.08). Individuals were tracked for an average of 14 days (range: 4–41 days). After data cleaning, 7534 fixes (range: 43–843 fixes per individual) were available for home range analysis.

Home ranges were roughly circular and the predicted 95% and 50% ranges of potoroos varied substantially among individuals (Fig. 2, Table 1). The largest 95% dBBMM range was predicted for the male ‘Snape’ (29.50 ha), and the smallest 95% dBBMM range was predicted for the female ‘Black Widow’ (3.44 ha). The average 95% dBBMM range was 13.73 ha (95% CI: 10.9–16.6) for males and 6.67 ha (95% CI: 5.49–7.85) for females. The average 95% dBBMM range across individuals was 10.94 ha (95% CI: 8.86–13.00).

The largest 50% dBBMM range was predicted for the male ‘Snape’ (8.00 ha) and the smallest 50% dBBMM range was predicted for the female ‘Frenchie’ (0.62 ha). The average 50% dBBMM range was 2.92 ha (95% CI: 2.20–3.64) for males and 1.23 ha (95% CI: 1.02–1.44) for females. The average 50% dBBMM range across individuals was 2.25 ha (95% CI: 1.74–2.76).

There was strong evidence that both body mass and sex influenced the home range size of potoroos in our study. Models that included both body mass and sex, either additively or interactively, were more supported relative to models that only included one of these covariates (Table 2). Estimates from the top-ranked model suggested that the influence of body mass on home range size depended on sex. Body mass had no detectable influence on home range size in females, but it had a positive influence on males (Fig. 3).

Fig. 3.

Predicted relationship between home range size and body mass for male and female long-nosed potoroos. Shaded areas indicate 95% confidence intervals.

There were 24 occasions where male and female 95% ranges overlapped, 23 occasions where male ranges overlapped with other males, and six occasions where female ranges overlapped with other females (Fig. 2). There was little evidence of exclusive 95% ranges amongindividuals or between sexes (Figs 2 and 3). Where individuals of the same sex co-occurred, 95% area overlap was 38.8% (95% CI: 29.0–48.6) for males and 47.1% (95% CI: 33.2–61.0) for females (Table S4–S9). Although samples sizes were small, there was some evidence that adult males may potentially orient their 95% ranges to maximise overlap with multiple female 95% ranges. For example, one male (‘Kenickie’) had a home range area that almost completely overlapped the entirety of three different female ranges, and another (‘Hulk’) had a home range area that almost completed overlapped the entirety of two different female ranges.

There were 12 occasions where male and female 50% ranges overlapped, eight occasions where male 50% ranges overlapped with other males, and four occasions where female 50% ranges overlapped with other females. There was little evidence of consistently distinct 50% core areas between males and females. Where males and females co-occurred, their 50% range overlap was 32.9% (95% CI: 20.4–45.4) on average. In contrast, there was some evidence of distinct 50% core areas among females (Fig. 4a) but not males (Fig. 4b). Where individuals of the same sex co-occurred, 50% range overlap was 28.6% (95% CI: 14.4–42.8) on average for males and 12.7% (95% CI: 0–33.7) on average for females – although sample sizes for co-occurring females were small.

Fig. 4.

50% (solid lines) and 95% (dotted lines) home ranges of co-occurring (a) female and (b) male long-nosed potoroos.

Discussion

Our home range estimates of 13.73 ha (95% CI: 10.9–16.6) and 6.67 ha (95% CI: 5.49–7.85) for male and female potoroos, respectively, are more than twice as large as previous estimates reported in the literature for Victorian potoroos (Long 2001, Table 3). They are also the first to be produced for this species using GPS dataloggers and dynamic Brownian bridge movement models. Our results suggested that the influence of body mass on potoroo home range size depended on sex. Body mass had no detectable influence on home range size in females but a positive influence in males (Table 3, Fig. 3). Contrary to elsewhere throughout their range (Norton et al. 2010; Frankham et al. 2011) and similarly to studies nearby (Bennett 1987), we observed no substantial differences in body mass between male and female potoroos. Prior trapping surveys in the same area suggest the body mass of male and females used in our analysis were typical for potoroos in this region (Le Pla, unpubl.data).

The discrepancy between male and female potoroo ranges at larger mass, despite no significant differences in body mass between sexes, suggest that this difference is not driven by the increased energetic requirements that coincide with increasing body mass (McNab 1963; Lindstedt et al. 1986; Swihart et al. 1988). Indeed, given the largely similar range sizes we observed in females of all masses (Fig. 3), it is likely that a range of 6–7 ha is large enough to allow females and younger males to find sufficient food and shelter with minimum risk and energy expenditure in our study area (Sandell 1989; Fisher and Owens 2000). Rather, the intersexual home range overlap patterns observed in our study imply resident adult male potoroos may range across larger areas to maximise their range overlap with nearby females. Potoroos are continuous breeders with asynchronous oestrous cycles (Hughes 1964), and male reproductive success may be linked to a male’s ability to regularly assess the sexual status of nearby females (Long 2001; Frankham et al. 2012). Maximising intersexual spatial overlap may provide males with greater access to females, which in turn could provide males with detailed knowledge of female movements and potentially improve their reproductive fitness (Schwagmeyer 1988; Long 2001). For example, in a study on French Island, Victoria, male potoroos that successfully sired young also demonstrated a high degree of spatial overlap with the mothers of their offspring (Frankham et al. 2012). This behaviour may be common within our study region – we observed several large male ranges almost entirely overlapping co-occurring female ranges, and it is not uncommon for both male and female potoroos to be detected together on camera traps (Le Pla, unpubl. data).

The substantial difference in mean body mass between these populations of potoroos is often proposed as an explanation for why Tasmanian home range estimates are larger than Victorian estimates (Table 3; Kitchener 1973; Long 2001). Further, the Victorian and Tasmanian populations have been separated for over 2 million years (Frankham et al. 2016), potentially resulting in demographic differences influencing a range of characteristics, including home range size.

Substantial differences in sample sizes and methodological approaches may go some way to explaining the variability in potoroo home range estimates. For example, although the studies of Johnson (1988) and Hird (1996) were conducted at the same location and both used trapping data to estimate home range size, Johnson’s choice of estimator (MCP) is more vulnerable to biases compared with the estimator used by Hird (KDE) (Börger et al. 2006). At small sample sizes, such as those that characterise trapping datasets, MCPs tend to underestimate home range sizes relative to methods that estimate a UD (Girard et al. 2002; Börger et al. 2006; Fleming et al. 2019). Indeed, MCP estimates of potoroo ranges are consistently lower than those predicted using a KDE approach in both Tasmania and mainland Australia (except for Kitchener 1973; Table 3). Because all home range estimates of Tasmanian potoroos have been based upon trapping locations (Table 3), it is possible, much like we observed in our study, that Tasmanian potoroos also range across larger areas than previously predicted. If so, it is also possible that a latitudinal gradient in potoroo home range size exists and has gone undetected due to methodological and analytical differences among studies. Future studies that incorporate GPS tracking of Tasmanian potoroos will be helpful in determining if this is indeed the case.

Additionally, landscape context may also play a significant role in determining home range size and arrangement in potoroos. For example, the studies of Bennett (1987) and Long (2001) were undertaken within a relatively small habitat patch surrounded by cleared agricultural land, whereas past Tasmanian studies and our study were conducted in much larger, contiguously connected habitat patches (Kitchener 1973; Johnson 1988; Hird 1996). Potoroos often display a reluctance to use cleared agricultural landscapes, so this difference in landscape context may be partially responsible for the smaller and potentially more constrained home ranges described by Bennett (1987) and Long (2001). Fine-scale differences in availability, arrangement of suitable habitat (Bennett 1993; Norton et al. 2011), and access to key food resources (Bennett and Baxter 1989; Claridge et al. 1993; Claridge and Cork 1994) also likely play a pivotal role in determining the size and spatial arrangement of potoroo home ranges. We recommend similar research into the movement ecology of potoroo populations, particularly those at different latitudes (e.g. Barren Grounds and Booderee, NSW; Norton et al. 2010, 2011; Hall et al. 2021) and in different habitats (e.g. sub-tropical Queensland; McHugh et al. 2019, 2020) before generalising our estimates of home range size to other landscapes.

The lack of exclusive 95% inter- or intrasexual ranges in our study supports the notion of potoroos being solitary yet non-territorial (Bennett 1987; Long 2001; Norton et al. 2010). We observed several examples of larger males (e.g. >800 g) demonstrating patterns of high overlap of both 50% and 95% ranges (Fig. 4b), and larger male potoroos often had ranges that encapsulated large proportions of smaller male ranges. In contrast, although sample sizes were small, the limited overlap of 50% ‘core’ activity areas observed among females suggests female potoroos may display some degree of spatial partitioning of activity (Fig. 4a). This arrangement could be driven by resource or shelter availability (Bennett and Baxter 1989; Bennett 1993) and has been observed in potoroo populations elsewhere (Norton et al. 2010; Frankham et al. 2014). Although it is likely these females are closely related, (female potoroos have been shown to be highly philopatric in other studies; Frankham et al. 2012, 2014), the behavioural mechanisms that underpin exactly how female potoroos partition these core areas remain unknown and beyond the scope of our data. Moreover, because we were unable to attach GPS dataloggers to females with large pouch young in our study, this apparent partitioning may simply be an artifact of excluding these individuals from our sample. Attaching GPS dataloggers to more co-occurring female potoroos, collecting movement data at more frequent intervals, and incorporating additional data streams (e.g. accelerometers) could allow for the classification of fine-scale movement behaviours and provide researchers with the ability to discern movements patterns associated with different behaviours (e.g. foraging or reproductive movements) (Langrock et al. 2012; Bennison et al. 2017). This approach would allow for the detailed description of encounters among individual potoroos and the context in which these encounters occur.

The primary limitations of our study were the variable fix success (i.e. variation in the proportion of attempted fixes that each GPS datalogger was able to successfully record as a location but were not removed through data screening), our inability to attach GPS dataloggers to females with pouch young, and the short duration of deployments. Fix success may be influenced by many factors (e.g. topography, fix interval, habitat characteristics; Cain et al. 2005), and fix success varied widely among individual GPS dataloggers in our study. For example, one GPS datalogger failed to record any fixes whatsoever, whereas another produced an overall average fix success of 88.7%. Similarly, fix success also varied temporally for individual dataloggers throughout their deployments. For example, one GPS datalogger’s daily fix success ranged between 8% and 80%. Despite this variability in fix success, because dBBMMs explicitly incorporate both spatial and temporal features of an individual’s path when estimating the UD, this approach can robustly estimate the UD even with variable fix success (Kranstauber et al. 2012). Additionally, although capture rates of male and female potoroos in our study region are generally even (Le Pla, unpubl. data), the inability to attach GPS dataloggers to female potoroos with large pouch young resulted in more males than females being included in our study (Table 1). This reduced the opportunity to describe occasions where female potoroos co-occurred and limited our ability to assess the degree to which co-occurring female 95% and 50% ranges overlapped. Finally, the relatively short duration of deployments (4–41 days) precluded any investigation into how potoroo movements may change through time. Given the potoroo’s propensity for exploiting hypogeal fungi (Bennett and Baxter 1989; Claridge and Cork 1994), it is possible potoroos may alter the size and spatial arrangement of their ranges in response to drivers of this seasonally available food resource.

Inappropriate fire regimes are a key threatening process for potoroos (Woinarski et al. 2014) because they often tend to inhabit environments that are burnt, through both prescribed burning and wildfire. For example, the recent ‘Black Summer’ wildfires (which took place during summer 2019–2020) in Australia burnt a substantial proportion of the long-nosed potoroo distribution (Legge et al. 2022; Ward et al. 2022), and a recent meta-analysis identified potoroos as one of nine mammal species consistently negatively affected by fire (Pocknee et al. 2023). Further, there is mounting evidence that potoroos may be particularly vulnerable to predation by invasive predators immediately after fire (Smith 2013; Robley et al. 2016; Hradsky 2020; Le Pla, unpubl. data). Estimates of potoroo home range size can directly inform the management of this functionally important (Claridge et al. 1992), yet nationally vulnerable, species after fire. For example, robust pre-fire home range size estimates could be used to assess the suitability of unburnt patches post-fire. Indeed, the small scale of burnt patches (<2 ha) relative to assumed potoroo home ranges (∼6 ha) was proposed as a primary reason for the lack of negative impact of planned burns on potoroos in sub-tropical Queensland (McHugh et al. 2020). Identifying where to prioritise recovery efforts (i.e. post-fire predator management), understanding the carrying capacity of the landscape as it recovers, and informing the ideal size of – and distance between – unburnt habitat patches to leave remaining within a prescribed burn scar to maintain viable populations of potoroos in the landscape post-fire are all potential ways land managers could use home range data to improve the management of potoroos in fire-prone landscapes.

Our results demonstrate the value of integrating GPS loggers and robust home range estimators to describe the movement ecology of a small, vulnerable native Australian marsupial. We highlight the potential that home ranges of potoroos may be larger than previously expected in mainland Australia and provide further evidence that males may arrange their ranges to maximise reproductive opportunities. Given their highly variable body size and the diverse array of habitats inhabited by this species, there would be value in conducting similar research into the movement ecology of potoroos elsewhere throughout their distribution. Doing so woulddetermine if the patterns of movement and range overlap observed in our study are consistent or vary according to different habitat types, body sizes, and landscape contexts. Improving our understanding of how range size and overlap vary within potoroo populations can inform targeted management actions before and after fire – a key threatening process for this threatened marsupial.

Supplementary material

Supplementary material is available online.

Data availability

The data that support this study will be shared upon reasonable request to the corresponding author.

Conflicts of interest

The authors declare no conflicts of interest.

Declaration of funding

Funding to purchase the GPS dataloggers used in this study was generously provided by the Hermon Slade Foundation.