Adaptive management of a remote threatened-species population on Aboriginal lands
J. L. Read
A * , R. West B , A
B
C
D
Abstract
Adoption and refinement of monitoring and management techniques is important for improving the conservation status of threatened fauna, especially in remote areas with high logistical and financial costs. In Australia, many of these remote conservation projects are conducted by Traditional Owners and Indigenous Ranger groups, with input and support from various stakeholders including government and non-government organisations, and third-party ecologists. A collaborative approach to project development and adaptive management in response to stakeholder objectives is essential for long-term project success.
In the remote context of the Anangu Pitjantjatjara Yankunytjatjara (APY) Lands, we collaboratively develop a sustainable and robust monitoring and management method for warru (black-flanked rock wallabies) that recognises the skills, interests, and capacity of all project members. In developing this method over 22 years, we also evaluate the influence of rainfall and predator-control strategies on warru populations to inform optimised management.
Practical and economic capabilities and preferences of both Traditional Owners and external scientists for monitoring and management actions were evaluated. Data from long-term cage trapping, and scat and spotlight counts of warru were compared with population trajectories, annual rainfall, and management strategies.
Population indices with lower resource requirements, such as scat and spotlight counts, provide trends consistent with more resource-intensive capture–mark–recapture studies. The warru scat index was negatively associated with the scat index for a competing herbivore (kanyala, euro). Warru growth rates were positively influenced by rainfall. Contrary to initial results from other regions, warru populations declined during periods when we undertook predator baiting. By contrast, targeted shooting of feral cats and foxes was associated with increased warru population growth rates.
These results suggest that a minimum of four groups of five scat quadrats in prime refuge areas and equivalent sampling in prime feeding zones (total 40 quadrats) is appropriate for rock-wallaby monitoring. Predator control via shooting appears preferable to baiting, which may have negative effects by removing dingoes, which prey on the competing kanyala.
Predator control via shooting is an effective way to support warru populations, and repeated scat counts are an effective monitoring approach for warru.
Abstract
Abstract in Pitjantjatjara
Nganana, ngurintjaku ngaranyi, panya yaaltji-yaaltjingkula. Anangu Tjuta alpamilalku kuka panya mawiyaringkunytja tjuta atunymara kanyintjaku ngurangka.
Waaka tjutawanungkula nyakukatingi munu numpa tjuta tjunangi munu alpamilaningi warru tjuta New Well-lawanu. Waaka nyanga paluru ngarakatingi, 1999-nguru, 2021 ku wangkara. Warruku kuna tjutala nyangangi munu nampa tjunangi munula nyangangi tili pulkawanungku palunyatjananya nyakunytjikitjangku. Nganana kulira witinintjikitjangku palya warru nyakunytjikitjangku, palu mani pulka munu waaka kunpu warru witintjaku. Nyanga paluru wirunya ngarangi munu mani tjukutjuku, ka alatjingkalta Anangu tjutaku kutjupa kutjupa tjutaku mani ma-tjunkuku.
Nganana wangkanyi panya alatji palyantjaku. Panya manta 8-pala ngaranyi. Kala alatji pukarangku warruku kuna tjuta nampa tjunkuku, manta 8-pala nyara palulanguru.
Warru Tjuta winkiringu kapingku pulkara year winkingku puintjitjangka munu ngaya inura tjuta, papa inura tjutakulula pauningi nyara palularangka. Palu patjina tjunkuntjitjangku ma-mankurparingu munu kanyala tjutangku warru tjutaku mai ngalkula wiyaningi. Patjinangkukula papa inura tjuta iluntanangi panya papa paluru tjana kanyala munu papa panya nyatji pulkanya patjara ngalkupayi. Ka ngaya inura tjuta munu papa inura nyantji pulka tjutakutju nyakukatinytjaku munu palunyatjananya kutja iluntankutjaku. Alatjikula mukuringanyi palyantjikitja.
Keywords: Anangu, dingoes, feral cats, First Nations, monitoring, population modelling, predator baiting, rock-wallaby, scat counts.
Introduction
In attempts to reverse the global biodiversity crisis, many threatened species populations are adaptively managed to improve their conservation status. Practitioners need to optimise monitoring and management techniques through feedback from previous management actions and through consideration of site-specific logistical and resourcing constraints. Where multiple stakeholders are supporting conservation projects, a collaborative approach maximises project sustainability. In Australia, Indigenous goals and objectives should be central to the design of monitoring and evaluation frameworks to ensure that conservation projects are appropriate for local contexts.
Warru, or black-flanked rock-wallaby (Petrogale lateralis centralis MacDonnell Ranges race), was widespread and abundant in the Anangu Pitjantjatjara Yankunytjatjara (APY) Lands of northern South Australia before and during the 1930s, when it was ‘one of the commonest mammals, with swarming populations’ (Finlayson 1961, p. 165). Warru are also important culturally, formerly being an important food source for Anangu and featuring in traditional Tjukurpa (law and culture). A series of threats, most likely precipitated by the arrival of rabbits and the elevated predator populations they supported, dramatically reduced the extent and abundance of warru to the point it became South Australia’s rarest mammal, with only two small, isolated populations persisting in South Australia into the 21st century (Read and Ward 2011). Dramatic declines were also recorded within its WA and NT ranges (Pearson 2010). The South Australian Warru Recovery Team (WRT) was formed in 2007 to recover warru populations through formalised and genuine collaboration with Indigenous owners (Anangu). The structure of the WRT and subsequent project plans includes Anangu objectives through active WRT membership and through a formalised decision-making and communication process encapsulated in the Warru Steering Committee; a representative group of Anangu Traditional Owners. Central to both the WRT and the Warru Steering Committee approach is that warru conservation plays a critical role in providing training and employment opportunities for Aṉangu and strong connections to historical and contemporary Tjukurpa, which is reflected in this study.
Management of warru has become a key component of Anangu Working on Country and Healthy Country plans since 2007 and Warru Rangers benefit from training, employment, continuing their connections to country, and opportunities for travelling and sharing their knowledge with others, and scientific stakeholders claim a range of benefits, including cross-cultural learnings (WRT 2023). It is intended also, of course, that warru benefit from the sustained and adaptive efforts of the WRT.
Control of foxes with 1080 poison baits was successfully pioneered in the Western Australian goldfields in the 1980s (Kinnear et al. 1988, 2010), to improve the conservation status of black-flanked rock-wallabies and then other rock-wallabies (DEH 2006; Kinnear et al. 2010; Pearson 2010). Subsequent research has shown that predation by feral cats may also contribute to declines of rock-wallabies (Spencer 1991; Read et al. 2019). Competition with kanyala (euro, Macropus robustus) and potentially other herbivores can also affect warru abundance (Read and Ward 2011). Intense fires, including those enabled by invasion of flammable buffel grass (Cenchrus ciliaris) that burn perennial food plants, and climate change are other key threats identified for warru (Read and Ward 2011).
Given the ongoing nature of threats to this species, systematic monitoring of warru populations is essential. Long-term population-monitoring data may identify the key drivers of change in abundance and the species’ response to adaptive management. Most macropods are typically surveyed using distance-sampling, including through spotlighting (Coulson et al. 2021). However, owing to their small size and occupancy of rugged hills, rock-wallabies are typically monitored by aerial surveys, trapping and scat monitoring (Piggott et al. 2006; Sharp et al. 2015; Phillips et al. 2022). The efficacy and cost effectiveness of these monitoring tools will vary according to locations, resourcing and the required precision of data. For example, capture–mark–recapture (CMR) trapping data may be justifiable where survivorship or population census are required, but is more expensive and risky than passive population-trend monitoring.
The New Well colony in the northern Musgrave Ranges is the largest and best-studied warru population in Australia. Routine monitoring commenced in the late 1990s, when concerns for the viability of the species intensified (Nesbitt and Wikilyiri 1994). Concerned by the rapid decline in warru populations, Anangu Traditional Owners, the ‘Warru Minyma’, in collaboration with external scientists from the WRT, initiated the Warru Project through monitoring, management and community-engagement initiatives; including the development of the warru inma (song).
In this study, we use two datasets to examine warru abundance over time; a 16-year CMR study (2005–2021), and a 22-year scat/spotlight study (1999–2021). We compare these datasets to the adaptive predator management undertaken at New Well. A key objective of this study was to develop robust practical monitoring tools to assess the value of management at remote warru colonies in an adaptive management framework. Therefore, we critically assess our monitoring methods, asking whether population indices (spotlighting and scat counts) are sufficient to provide robust, cost-ffective insights into trends in abundance and response to management.
Methods
Study site
New Well is a 590 ha granitic outcrop in the eastern Musgrave Ranges of northern South Australia (Fig. 1), surrounded by open woodland and with spinifex grasslands (Robinson et al. 2003). The area surrounding and including New Well is the traditional homelands of the Dolby, Stanley and Windlass families, Pitjantjatjara and Yankunytjatjara speakers. New Well is centrally located within the eastern Musgraves warru metapopulation (Ward et al. 2011a) and has been the largest and most intensively monitored warru colony in South Australia since the 1990s. The locations of monitoring sites and transects were all determined by Anangu and scientists to respect cultural sensitivities and preferences.
Map of the New Well outcrop showing the location of the scat quadrats, traps and spotlight transect used to monitor the warru population. Inset shows the location of the outcrop (black square) within the APY Lands (grey shading) in South Australia.
Annual rainfall readings throughout the study period were collated from the Bureau of Meteorology weather station at Pukatja, 24 km south of New Well.
CMR census data
CMR is assumed to provide a more robust measure of warru abundance than are indices that do not account for variable detection, and which do not identify and, hence, are unable to count individuals. Between 2005 and 2021, we conducted 14 CMR survey sessions for warru at New Well. Each session involved trapping over a four–six-night period, with between 11 and 26 cage traps being placed across the site in three lines, and cleared daily (Fig. 1). Captured individuals were marked with unique coloured eartag combinations and microchips, or recorded as a recapture. Pouch young were not marked. Given the repeat sampling within each session, we estimated detection probability and population size for each session (see Appendix 1). These detection-corrected estimates of population size were considered our census population size in each year, the baseline against which we compare population indices derived from scats and spotlighting. All analyses here and below were conducted in the R statistical environment (R Core Team 2023).
Population indices
Scat accumulation rate was measured in 24 1-m-radius quadrats in the core of the New Well warru colony (Fig. 1). Quadrats were not placed randomly, but were instead established in high-use areas, but not at sites where old scats were likely to be washed into the quadrats. We also avoided, on advice from Aṉangu, culturally sensitive areas. Quadrats were permanently marked by an unobtrusive central pin. Within each quadrat, warru and kanyala (euro, Osphranter robustus) scats, which have distinctly different shapes, were counted and removed twice annually from 2001 to 2021. In the analysis, scat-count data were represented as a rate of accumulation (scats per day) to account for variable survey periods. This rate variable was not normally distributed, so we used the central limit theorem to provide us with a variable, mean accumulation rate (averaged across the 24 quadrats in a survey period), that was much closer to being normally distributed.
To estimate the minimum number of scat samples that will provide a reasonable population index, we investigated the correlation between the scat index and our census population size. We subsampled (without replacement) the scat data up to a sample size, n*, representing the number of plots in a year’s survey. We then re-calculated the correlation to find the correlation coefficient , for each value of n* (from 1 to 24). We iterated this process 1000 times, resulting in 1000 resamples of for each value of n*. We then used these resamples to examine the relationship between sample size of the scat data and the correlation between scat index and census population size.
Counts of warru observed by spotlight, along with potential predators and competing herbivores, were made around a 13.4-km-perimeter track around the New Well colony on the same dates as the scat quadrats were cleared (Fig. 1). We conducted these surveys within the first 3 h after sunset; an observer with a spotlight and binoculars counted all animals seen on their (left) side of the track, whereas the driver maintained speeds of less than 20 km per hour. Again, we examined the correlation between this spotlighting index and our census population size.
Costs, precision and preferences of each monitoring method and autonomy for Indigenous Rangers
We calculated the cost of each monitoring method on a per-session basis. Scat and spotlight counts involved 1 day of field work by an ecologist plus up to three Aṉangu Rangers and costs of one vehicle. Both of these monitoring methods can also be conducted autonomously by Indigenous Rangers, with reduced costs because they live nearby and can integrate monitoring tasks with other cultural or land management activities.
By contrast, CMR required specialist expertise and substantially more person hours per survey. Because of the propensity for trapped female warru to ‘throw’ large pouch young before or on release, experienced handlers capable of suturing pouches were required at each of three traplines necessary to release all trapped animals within 3 h of dawn. This plus the logistics of installing and checking traps meant that trapping sessions typically involved three to five ecologists or animal handlers (mainly Zoos SA staff), and four to eight Anangu Rangers and their coordinators for a 4–5-day period.
The WRT hold annual on-Country meetings facilitated by translators, where cultural and scientific opinions and priorities are openly discussed by all stakeholders. Furthermore, during the 2023 AGM held at Wamitjara, all stakeholder groups were independently and then collaboratively questioned about aspects of the Warru Recovery Project that were particularly important to them (Warru Recovery Team 2023).
Predator-management methods
Dried kangaroo meat baits impregnated with 1080 poison were first laid around perimeter tracks at New Well in 1996 (Geelen 1999). From August 2000 to 2007, approximately 500 baits per session were deployed every 2–3 months. Because of the concerns of baiting dingoes, which typically use roads, baits during this period were laid by hand in crevices deemed accessible to foxes and cats but not to dingoes. Because of this hand-delivery to warru dens, we refer to this method as ‘den baiting’. However, owing to the challenges of climbing hills with buckets of baits in hot weather, baiting then reverted to 34 marked bait stations around the perimeter of New Well in 2008. This ‘perimeter baiting’ continued on an approximately monthly basis around New Well until 2014.
Aerial baiting was instigated around the eastern Musgrave Ranges warru metapopulation in July 2004 and continued through to 2015. This method involved the use of a helicopter and distribution of approximately 7000 meat baits poisoned with 1080. Because of the concerns regarding baiting of pet dogs and dingoes, baits were dropped directly only on hills and no baiting was conducted along roads, or around Aṉangu communities and homelands. Baiting ceased in 2015, in response to Anangu concerns about the off-target impact of baiting on valued bushfoods (e.g. ngintaka, Varanus giganteus).
Indigenous rangers’ preferences for not using baits influenced a switch to shooting as an alternative predator-control method. Whereas feral cats and foxes were shot opportunistically at night during twice-yearly sampling periods from 2001 to 2012, hunting increased in intensity from 2013, to at least 10 nights of dedicated shooting per year from 2013 to 2021 (Fig. 2).
Efficacy of management on warru population dynamics
Given our predator-management strategies changed over time, there were strong correlations between our methods; for example, the correlation between baiting effort and intensive shooting was −0.75. These correlations, and the small number of years, preclude us from assessing the effect of each management strategy independently. We, thus, take the approach that population growth in warru is dependent on the environment (in particular, inter-annual variation in rainfall), and on the switch between baiting and shooting as a management technique (using a categorical variable indicating whether or not intensive shooting was undertaken in a given year). Because shooting and baiting are (negatively) correlated through time, we cannot separate whether a response by warru is due to the cessation of baiting, or to the increase in shooting, or both, but we can assess whether the overall shift in management is associated with improved warru growth rates.
To test the idea that warru populations respond to our changed management, we extended our mark–recapture analysis (Supplementary material) to estimate the carrying capacity, and a time-varying growth rate for the population. We make the growth rate a linear function of rainfall and management (shooting vs no shooting), allowing us to assess their effect on the population growth rate (see Supplementary material).
Our scat data are a population index, rather than a census estimate, so we cannot apply the same population growth function to these data. Instead, we created a new variable, i.e. change in scat deposition rate between years, and ran a simple multiple regression to assess whether this variable is affected by rainfall and the switch to intensive shooting.
Results
CMR census population size
Estimated census population of warru at New Well between 2005 and 2021 varied between 16 and 71 animals (Fig. 3, Supplementary Table S3).
Census population size of warru at New Well from 2005 to 2021 calculated from capture–mark–recapture analyses. Error bars show 95% credible intervals. The mean scat index (×250) for each year is shown as a light dotted line for visual comparison. Mark–recapture was not conducted in 2015, 2019, or 2020, but we are able to estimate population size in these years through constraints provided by the population growth model (Supplementary material).
Individual detection probability of warru in the CMR study varied among years (Supplementary Fig. S1), but was generally ~0.2 per individual per night. Detection probability showed a weak positive relationship with trapping effort (, 95% credible interval −0.029–0.063).
Population indices
Population indices derived from scat and spotlight counts both suggested a spike in warru occurring around 2002, followed by a steady decline through to 2006, followed by a steady recovery after 2012 (Fig. 4). For the scat index, there was a clear negative correlation between indices of warru and kanyala (Fig. 4). The estimated correlation between warru and kanyala scat data was (95% confidence interval −0.66 to −0.14); for the spotlight data, there was no clear correlation: (95% confidence interval −0.23–0.38).
Comparison of monitoring techniques
CMR-based population census data showed a pattern that was broadly consistent with the temporal pattern exhibited by the scat count index (Fig. 3), although there was a notable discrepancy between the scat index and estimated population size between 2011 and 2014.
The population indices generated from scat and spotlight counts are substantially easier and cheaper to generate than are the CMR data (Table 1); however, they are indirect indices of population size. There was a clear positive correlation between both population indices and the census population size (Fig. S2). The estimated correlation between the scat index and census population size was (95% confidence interval 0.09–0.81), and between the spotlight index and census population size, it was (95% confidence interval 0.09–0.81).
| Method | Cost per session | Precision | Autonomy for Indigenous Rangers | Engagement by Indigenous Rangers | |
|---|---|---|---|---|---|
| Trapping | A$18,744 | 3 | 1 | 3 | |
| Spotlighting | A$2360 | 1 | 3 | 3 | |
| Scat quadrats | 2 | 2 | 1 |
Minimum scat-sampling effort
To provide guidance for minimum scat-sampling effort for other colonies, we conducted a simple ANOVA of the scat index, in which variance was partitioned within and among years. The overwhelming majority of variance in this index occurs across years (F20,18 = 5.23, P = 0.0004). We, thus, pooled the data within years, so that we had between 24 and 78 replicates in a given year. Given the methodology resurveys the same plots each session, this pooling does include some pseudoreplication in the dataset. We mitigate this by only resampling up to a sample size of 24 (the number of plots, and the minimum number of replicates in a year). Resampling the scat index versus census population correlation while varying the sample size of the scat index showed that the observed correlation is approached quite rapidly as sample size increases. Resampled data yielded correlations that rapidly approached the observed correlation by 15 quadrats (Fig. S3). However, the variance in resampled correlations was quite high until a sample size of >20 plots was reached.
Cost–benefit analysis
Cumulative costs of nine CMR trapping sessions from 2007 to 2016, including mobilisation of specialist animal handlers from Adelaide, were A$168,700 (Ireland et al. 2018) or A$18,744 per session. Together with mobilisation costs, from >1000 km away, cumulative costs for 20 scat and spotlight surveys from 2007 to 2016 were A$47,200 (Ireland et al. 2018), or A$2360 per session, making these surveys ~12% of the cost of a CMR session. Our experience over 20 years and feedback from current APY Warru Rangers (WRT 2023) suggest that CMR sessions (which involve connecting with Country for five nights), and vehicle-based spotlighting and shooting were preferred activities compared with climbing hills to conduct scat counts (Table 1). However, rangers have demonstrated that they can autonomously conduct both scat counts in clearly marked and readily relocated quadrats (Fig. 5) and spotlight counts, whereas trapping trips require considerable external assistance, including the administrative overhead of applications for animal ethics (Table 1).
Influence of management and rainfall on warru population dynamics
When we extended the CMR analysis to include population dynamics, we recovered an estimated carrying capacity for the population of 69.6 individuals (95% credible interval 59.1–85.4). With our best inference suggesting a positive effect of rainfall on the population growth rate in a given year (estimate 0.0034, 95% credible interval −0.00019–0.0066). The growth rate was also estimated to increase as we shifted from baiting and towards intensive shooting, although this result was more equivocal (estimate 0.71, 95% credible interval −0.67–2.14; Fig. 6).
Population growth rate of warru as a function of rainfall and management technique (shooting vs no shooting). Error bars represent 95% credible intervals for the growth rate in each year. Under the Ricker population model, growth-rate values of >0 denote an increasing population when the population is at low density.
As a contrast to this full analysis on the mark–recapture data, we also looked to see whether the change in the scat index among years showed similar patterns. A simple multiple regression on the change in scat index, with annual rainfall and shooting as covariates, showed a significant positive effect of switching from baiting to intensive shooting (estimate 0.038, t = 2.78, d.f. = 17, P = 0.013; Fig. S5), and a significant positive effect of rainfall in the calendar year prior to sampling (estimate 0.00010, t = 2.26, d.f. = 17, P = 0.037; Fig. S4).
Discussion
The adaptive management of warru at New Well that initiated a switch from poison-baiting to targeted shooting of foxes and feral cats was influenced by Warru Rangers’ preferences for hunting, and by Traditional Owner concerns about off-target baiting impacts. This switch to intensive shooting of predators, mainly cats, was associated with increased growth rates in warru population. The New Well warru population exhibited good adult survivorship but limited recruitment (Ward et al. 2011b); hence, improved survivorship of predator-vulnerable juveniles was the likely driver for the observed rebound in warru. Cat impact is often driven by relatively few individual ‘catastrophic’ cats that learn to hunt challenging prey, such as rock wallabies (Spencer 1991) and other species (Moseby et al. 2015, 2021; Hardman et al. 2016). Hence, it is likely that several of the feral cats shot with warru in their stomachs at New Well (Read et al. 2019) were responsible for suppressing recruitment in this population. The relative benefit of hunting versus baiting was positive and significant when we considered the full scat dataset, an effect that was also apparent, although less clearly so, in the CMR dataset that commenced only after the crash in warru populations at New Well.
The poor response of the New Well warru populations to predator baiting, which had initially proven effective for this species in other regions (Kinnear et al. 2010), may be driven by two trophic cascades. First, baiting, which is most effective at controlling dingoes and foxes, can precipitate increases in populations or hunting efficacy of feral cats (through mesopredator release), leading to increased pressure on species efficiently predated by cats (Johnson 2006; Marlow et al. 2015). Unlike dingoes and large foxes, cats are small and nimble enough to enter most warru refuges.
The second relevant trophic cascade concerns a competing macropod species, kanyala. Both dingo and fox predation can limit kangaroo populations (Banks et al. 2000; Pople et al. 2000) and, hence, the removal of foxes and dingoes through baiting may have influenced the dramatic increase in kanyala numbers compared with scarce historic records in the region (Finlayson 1961). The negative relationship between the warru and kanyala scat index recorded in this study is likely to reflect a negative correlation in population sizes across years. Such a correlation is likely to be attributable to competition for food, especially because kanyala can browse important perennial food plants (e.g. Pandorea) higher in the vegetation than those in the reach of warru.
The growth rate of warru at New Well was also associated with annual rainfall in the previous calendar year. Elevated recruitment and survival in response to increased resource availability is typical of the response of many desert mammals to unpredictable wet years (Greenville et al. 2012).
The positive relationship between census counts and population indices derived from scat counts and spotlight counts indicated that these less intensive monitoring tools can provide valuable feedback on population trajectories. Although these indices clearly allow us to observe large changes in the population, there is variability in space and time in the rate at which scats are deposited and decay; hence, the value in regular 6-monthly count and removal of scats. Both the spotlight and particularly scat counts at New Well sample only a fraction of the available habitat, whereas most warru could potentially visit a trap site over the four nights of trapping. Because of this, broader spatial coverage of the scat monitoring is advisable, targeting core and peripheral areas of occupancy. In contrast, CMR trapping allows direct estimation of population size and, so (assuming the usual assumptions of mark–recapture studies), we might expect it to be the more sensitive metric. But this metric carries its own disadvantages, namely, CMR is not always feasible, especially in the long term, at all sites and is likely to be less sensitive than is scat monitoring for detection of low-density immigration or occupancy of transient sites. Hence, we concur with Bluff et al. (2011) that multiple monitoring tools are likely to detect different subsets of the population and, where possible, should be used complementarily for monitoring rock-wallabies. We also note that the hierarchical model we use here to infer population trajectories might, in principle, be extended to allow multiple observation types, allowing us to integrate information from multiple monitoring methods.
At one-eighth of the cost and manpower requirements of trapping trips, data collection for scat and or spotlight counts is less logistically and financially costly than that for trapping. Both scat and spotlight counts can be performed autonomously without the need for veterinary assistance, by as few as two people and, hence, are more sustainably included in workplans than are trapping sessions that require more planning, training and administrative overheads. For example, Warru Rangers, who live locally, autonomously monitor scat plots painted onto prominent rocks or ledges at Kalka and Kulitjara warru colonies in the APY Lands. Furthermore, these passive population indices do not place animals at any risk, nor do they require animal ethics and other permits required for trapping. For these reasons, scat counts provide a valuable tool for assessing general population trends, whereas CMR trapping is important when resources are available to measure fine-scale responses to adaptive management. Furthermore, through annual on-Country meetings and the independent decision-making process of the Warru Steering Committee, the social and cultural importance of CMR trapping events has been highlighted. These events, although resource-intensive, provide significant multi-generational learning and cross-cultural knowledge-sharing opportunities.
Both spotlight and scat counts may be influenced by factors other than warru population density. For example, when food resources were low adjacent to warru refuge sites and scat plots near the tops of hills, warru individuals were more likely to be detectable by spotlighting around the base of hills. Alternatively, recovery of dingo populations, which could push warru activity back to safer elevated sites near dens, may have taken several years after cessation of baiting. Such behaviours may explain the poor match between CMR estimates and scat/spotlighting estimates in the years between 2011 and 2015, immediately after the population growth rate appears to have recovered. This spatial variation in habitat use caused by food availability and predation risk means that optimal scat counts should be conducted in both refuge areas and habitat with more food remote from refuges. Our modelling suggests that 20 quadrats produces an index of 0.5 correlation with CMR calculations compared with the correlation coefficient asymptote of 0.54. Therefore, we suggest that a minimum of four groups of five scat quadrats (total 20 quadrats) in prime refuge areas and equivalent sampling in prime feeding zones should be monitored in each warru colony to provide robust sampling intensity and spatial coverage. Rather than using uniformly sized circular quadrats marked by a central pin that was often difficult to locate, as was used at New Well, population trends can more readily be assessed by rangers from boldly marked flat-topped boulders (Fig. 5). In cases where reliability of scat counts presents a challenge, presence/absence of fresh scats from multiple quadrats (Phillips et al. 2022) will likely also generate a meaningful index of occupancy at low densities.
On the basis of the findings of this study, scat monitoring established for the reintroduced Wamitjara warru population, also in the eastern Musgrave Ranges, uses 10 clusters of five scat quadrats around the colony (WRT 2023). Ongoing comparisons with CMR data at Wamitjara, including the baseline data when a known number of individuals was present, will be used to assess and, if necessary, refine the monitoring approach further. The negative correlations between warru and kanyala populations and warru trajectories and baiting, and the socio-cultural importance of dingoes, have resulted in predator baiting being replaced by targeted shooting and Felixer deployments to control feral cats and foxes at warru colonies in the APY Lands (WRT 2023). These informed refinements in monitoring and management methods highlight the value of collaboration and adaptive management of threatened species populations.
Declaration of funding
This research did not receive any specific funding, although CMR and scat-count field work were funded by Working on Country funding.
Acknowledgements
We acknowledge Anangu Traditional Owners and Knowledge Holders, Warru Rangers and APY Land Management for their past, and ongoing contributions and effort in the management of warru. The authors are indebted to the guidance and assistance provided by the Warru Minyma. Special thanks and consideration go to Kunmanara De-Rose for the translation of the abstract, and we extend condolences to her family. Other members of the Warru Recovery Team, including staff of APY Land Management, ZoosSA, South Australian Department of Environment and AW Landscape Board have been integral to the sustained applied conservation management and monitoring of warru at New Well. Graham, Robyn and Murray Miller and Ethan Dagg are thanked for their efforts and skill on many long nights shooting cats and foxes at New Well. BLP was supported by the WA Premier’s Fellowship from the Western Australian government.
References
Banks PB, Newsome AE, Dickman CR (2000) Predation by red foxes limits recruitment in populations of eastern grey kangaroos. Austral Ecology 25, 283-291.
| Crossref | Google Scholar |
Bluff LA, Clausen L, Hill A, Bramwell MD (2011) A decade of monitoring the remnant Victorian population of the brush-tailed rock-wallaby (Petrogale penicillata. Australian Mammalogy 33, 195-201.
| Crossref | Google Scholar |
Coulson G, Snape MA, Cripps JK (2021) How many macropods? A manager’s guide to small-scale population surveys of kangaroos and wallabies. Ecological Management & Restoration 22, 75-89.
| Crossref | Google Scholar |
Finlayson HH (1961) On central Australian mammals. IV. The distribution and status of central Australian species. Records of the South Australian Museum 14, 141-191.
| Google Scholar |
Greenville AC, Wardle GM, Dickman CR (2012) Extreme climatic events drive mammal irruptions: regression analysis of 100-year trends in desert rainfall and temperature. Ecology and Evolution 2, 2645-2658.
| Crossref | Google Scholar | PubMed |
Hardman B, Moro D, Calver M (2016) Direct evidence implicates feral cat predation as the primary cause of failure of a mammal reintroduction programme. Ecological Management & Restoration 17, 152-158.
| Crossref | Google Scholar |
Ireland L, Zabek M, Galindez-Silva C, Weir S, West R, Olds L, Backhouse B, Copley P, Read J (2018) More than just the animals: opportunities and costs of reintroducing threatened black-footed rock-wallabies to remote Indigenous land. Pacific Conservation Biology 24, 388-396.
| Crossref | Google Scholar |
Kinnear JE, Onus ML, Bromilow RN (1988) Fox control and rock-wallaby population dynamics. Australian Wildlife Research 15, 435-450.
| Crossref | Google Scholar |
Kinnear JE, Krebs CJ, Pentland C, Orell P, Holme C, Karvinen R (2010) Predator-baiting experiments for the conservation of rock-wallabies in Western Australia: a 25-year review with recent advances. Wildlife Research 37, 57-67.
| Crossref | Google Scholar |
Marlow NJ, Thomas ND, Williams AAE, Macmahon B, Lawson J, Hitchen Y, Angus J, Berry O (2015) Cats (Felis catus) are more abundant and are the dominant predator of woylies (Bettongia penicillata) after sustained fox (Vulpes vulpes) control. Australian Journal of Zoology 63, 18-27.
| Crossref | Google Scholar |
Moseby KE, Peacock DE, Read JL (2015) Catastrophic cat predation: a call for predator profiling in wildlife protection programs. Biological Conservation 191, 331-340.
| Crossref | Google Scholar |
Moseby KE, McGregor H, Read JL (2021) The lethal 23%: predator demography influences predation risk for threatened prey. Animal Conservation 24, 217-229.
| Crossref | Google Scholar |
Pearson DJ (2010) Recovery Plan for five species of rock-wallabies: Black-flanked Rock-wallaby (Petrogale lateralis), Rothschild Rock-wallaby (Petrogale rothschildi), Short-eared Rock-wallaby (Petrogale brachyotis), Monjon (Petrogale burbidgei) and Nabarlek (Petrogale concinna). Department of Environment and Conservation, Perth, WA, Australia.
Phillips S, Howard K, Krause J (2022) Informing the space between zero and one: a standardised, scat-based approach to monitoring changes in habitat use by brush-tailed rock-wallabies (Petrogale penicillata. Australian Mammalogy 44, 161-168.
| Crossref | Google Scholar |
Piggott MP, Banks SC, Stone N, Banffy C, Taylor AC (2006) Estimating population size of endangered Brush-tailed Rock-wallaby (Petrogale penicillata) colonies using faecal DNA. Molecular Ecology 15, 81-91.
| Crossref | Google Scholar | PubMed |
Plummer M (2023) rjags: Bayesian Graphical Models using MCMC. R package version 4-14. Available at https://CRAN.R-project.org/package=rjags
Pople AR, Grigg GC, Cairns SC, Beard LA, Alexander P (2000) Trends in the numbers of red kangaroos and emus on either side of the south Australian dingo fence: evidence for predator regulation? Wildlife Research 27, 269-276.
| Crossref | Google Scholar |
R Core Team (2023) ‘R: a language and environment for statistical computing.’ (R Foundation for Statistical Computing: Vienna, Austria) Available at https://www.R-project.org/
Read JL, Ward MJ (2011) Bringing back warru: initiation and implementation of the South Australian Warru Recovery Plan. Australian Mammalogy 33, 214-220.
| Crossref | Google Scholar |
Read JL, Dagg E, Moseby KE (2019) Prey selectivity by feral cats at central Australian rock-wallaby colonies. Australian Mammalogy 41, 132-141.
| Crossref | Google Scholar |
Sharp A, Norton M, Havelberg C, Cliff W, Marks A (2015) Population recovery of the yellow-footed rock-wallaby following fox control in New South Wales and South Australia. Wildlife Research 41, 560-570.
| Crossref | Google Scholar |
Spencer PBS (1991) Evidence of predation by a feral cat, Felis catus (Carnivora: Felidae) on an isolated rock-wallaby colony in tropical Queensland. Australian Mammalogy 14, 143-144.
| Crossref | Google Scholar |
Ward MJ, Urban R, Read JL, Dent A, Partridge T, Clarke A, van Weenen J (2011a) Status of warru (Petrogale lateralis MacDonnell Ranges race) in the Anangu Pitjantjatjara Yankunytjatjara Lands of South Australia. 1. Distribution and decline. Australian Mammalogy 33, 135-141.
| Crossref | Google Scholar |
Ward MJ, Ruykys L, van Weenen J, de Little S, Dent A, Clarke A, Partridge T (2011b) Status of Warru (Petrogale lateralis MacDonnell Ranges race) in the Anangu Pitjantjatjara Yankunytjatjara Lands of South Australia. 2. Population dynamics. Australian Mammalogy 33, 142-151.
| Crossref | Google Scholar |
WRT (2023) Warru recovery team summary report 2007–2022 recovery of Petrogale lateralis centralis in South Australia. Available at https://cdn.environment.sa.gov.au/landscape/docs/aw/Warru-Summary-Report-2007-2021.PDF
