Applying home-range and landscape-use data to design effective feral-cat control programs

Transcription

1 CSIRO PUBLISHING Wildlife Research, 2012, 39, Applying home-range and landscape-use data to design effective feral-cat control programs Andrew J. Bengsen A,B,C, John A. Butler A and Pip Masters A A Kangaroo Island Natural Resources Management Board, 35 Dauncey Street, Kingscote, SA 5223, Australia. B Present address: NSW Department of Primary Industries, Locked Bag 6006, Orange, NSW 2800, Australia. C Corresponding author. andrew.bengsen@uqconnect.edu.au Abstract Context. Effective feral-cat (Felis silvestris catus) management requires a sound understanding of the ways cats use their environment. Key characteristics of landscape use by cats vary widely among different regions and different conditions. Aims. The present study aimed to describe the most important characteristics of landscape use by feral cats on a large, human-populated island, and to use this information to guide the development of feral-cat management programs. Methods. We used GPS tracking collars to record the movements of 13 feral cats at two sites on Kangaroo Island, South Australia, for between 20 and 106 days. We described home-range extents by using local convex hulls, and derived management suggestions from examination of home-range and movement data. Key results. Median feral-cat home range was 5.11 km 2, and this did not differ between sexes or sites. Cats at a fragmented pastoral site tended to favour woody vegetation over open paddocks, but habitat preferences were less clear at a bushland site. Cats that preferentially used treelines at the pastoral site were almost twice as likely to be recorded close to a tree-line junction as expected. Conclusions. Control programs for feral cats on Kangaroo Island should deploy control devices at a density no less than 1.7 devices km 2. Spatial coverage should be as large as practicable or repeated frequently. Infrequent programs covering small areas can be expected only to provide short-term reductions in cat abundance. Implications. The information gained from the present study will contribute to the development of strategic sustained management plans for feral cats on Kangaroo Island. The principles from which we inferred management guidelines are applicable to other regions and species. Additional keywords: Felis silvestris catus, habitat use, Kangaroo Island, vertebrate-pest management. Received 9 June 2011, accepted 6 February 2012, published online 5 April 2012 Introduction An understanding of the ways that pest animals occupy and move through the landscape is critical to the effective and efficient implementation of pest damage-control programs. Information about the spatial ecology of pest animals can help determine where control devices should be deployed for optimal efficiency (Recio et al. 2010), how large an area should be treated (Norbury et al. 1998) and at what density control or monitoring devices should be deployed (Moseby et al. 2009). Spatial information can also be used to predict the efficacy of different control tactics or strategies (Alterio et al. 1998; Molsher et al. 2005; Guttilla and Stapp 2010), or the dispersal of zoonotic diseases (Pech and McIlroy 1990). The feral cat (Felis silvestris catus) is a widespread invasive predator (Long 2003), with substantial and diverse impacts on biodiversity and economic values across the globe (e.g. McLeod 2004; Nogales et al. 2004; Pimentel et al. 2005; Hilton and Cuthbert 2010; Medina et al. 2011). Feral cats occupy a wide range of environments, and previous investigations have shown Journal compilation CSIRO 2012 that the ways in which cats use the landscape vary widely in response to local conditions (e.g. climate, predators) and resources (e.g. prey, shelter). For example, estimated mean home-range areas of female feral cats have varied from ~0.9 km 2 in New Zealand farmland (Langham and Porter 1991) to 20.8 km 2 in Australian arid rangelands (Moseby et al. 2009). Consequently, it is difficult to generalise from previous studies when developing management programs for feral cats. Most studies of landscape use by feral cats have been descriptive, and few have specifically attempted to answer questions about the implications of landscape-use patterns for the design of cat-control programs (but see Alterio et al. 1998; Moseby et al. 2009; Guttilla and Stapp 2010). In the present study, we investigated landscape use by feral cats on a large, humanpopulated island, namely, Kangaroo Island in South Australia. Predation by feral cats is recognised as a threatening process to three resident endangered species (Isoodon obesulus obesulus, Macronectes giganteus, Sminthopsis aitkeni) and four vulnerable species (Diomedea exulans, Halobaena caerulea, Thalassarche

2 Movements and control of feral cats Wildlife Research chrysostoma, T. melanophris; Department of Environment Water Heritage and the Arts 2008). Feral cats are also vectors of livestock diseases, such as toxoplasmosis and sarcosporidiosis, which have been prominent on the island (O Donoghue and Ford 1986; O Donoghue et al. 1987). The present study aimed to estimate (1) the form of cat home ranges, (2) which habitat types should be targeted to increase the probabilities of cats encountering control tools, (3) minimum densities at which cat-control tools such as baits or traps should be deployed for optimal efficiency and (4) what spatial scales cat-control programs should be conducted over for optimal efficacy. Materials and methods Study site The study was conducted at two sites on the eastern end of Kangaroo Island, South Australia ( S, E), between March and December About 51% of the island s 4405 km2 is covered in native vegetation, predominantly low-eucalyptus spp. woodlands, with pasture and mixed crops comprising most of the remainder. The island s climate has been classified as mediterranean (Schwerdtfeger 2002). Foxes (Vulpes vulpes) and dingoes (Canis lupus dingo) are absent from the island, and the feral cat is the principal terrestrial predator. The absence of competition from foxes can have important behavioural and ecological implications for the ways that cats use the landscape (Dickman 1996; Risbey et al. 2000; Glen and Dickman 2005). Rabbits (Oryctolagus cuniculus), which are a staple prey item for cats throughout much of Australia (Molsher et al. 1999, and references therein), are also absent. Cats have been present on the island since the 1830s, and are now common across the entire island. Feral-cat control programs are currently limited to trapping or opportunistic shooting, which is usually conducted at the scale of individual properties. One study site (bush site) was situated on the island s southern coast, on limestone plains and sandhills that supported low shrubland grading to medium-height mallee (Eucalyptus spp.) with increasing distance from the coast. Vegetation on the coastal limestone plains, which covered ~16% of the study area, comprised patches of low heath within a matrix of low forbs, which provided little physical cover. Areas of woody vegetation away from the immediate coastal fringe were dominated by dense mallee regrowth up to ~3-m height with isolated patches of shrubland (22% of the study area). Other woodland areas were dominated by taller mature mallee woodland with a near-closed canopy and an open understorey (12% of the study area). The greatest structural and floristic diversity at the bush site occurred in areas of mixed shrubland with emergent eucalypts, which was most common in the centre of the study site (17% of the study area). Large areas adjacent to the bush site had been cleared for agriculture (33% of the study area). The second study site (pastoral site) was situated over three sheep- and cattle-producing properties on undulating terrain which supported substantial patches of low and mediumheight open eucalypt woodlands, linked by corridors along fencelines (Fig. 1). Woodland at the pastoral site differed from that at the bush site by generally being taller, and having a more floristically and structurally complex understorey. Grasstrees (Xanthorrhoea semiplana tateana) were prominent throughout, and the interlocking aprons of leaves formed by dense stands of this species provided shelter for cats. The density of feral cats at the pastoral site was estimated at 0.7 cats km 2 using camera-trap capture recapture methods in December 2010 (Bengsen et al. 2011). Data collection We captured 12 feral cats in cage traps baited with chicken wings or tinned fish between April and October 2010 at the bush site, and 13 during the same period at the pastoral site. We fitted 16 cats that weighed >2.6 kg with combination VHF transmitter GPS logger collars (Sirtrack, Havelock North, New Zealand) equipped with timed collar-release mechanisms. Six cats were collared at the bush site and 10 at the pastoral site. Collars weighed either 137 g or 154 g, and no cat was fitted with a collar that weighed more than 5.2% of its bodyweight (following Gannon et al. 2007). GPS units were programmed to attempt a location fix either every 5 or 2 hours (n = 11 and 5 collars, respectively). Cats that were instrumented early in the study had the longer intervals to Study sites 36 Native vegetation N Kilometres Agricultural and other land uses Fig. 1. Location of two study sites for the examination of landscape use by feral cats on Kangaroo Island, South Australia.

3 260 Wildlife Research A. J. Bengsen et al. conserve batteries. Cats were sedated using an intramuscular injection of medetomidine hydrochloride at a rate of 150 mgkg 1, the effects of which were reversed using atipamezole hydrochloride. Nine cats that were too small to collar were photographed and fitted with brass small mammal eartags (Hauptner, Dietlikon-Zürich, Switzerland) to allow recognition on recapture. All cats were released at their site of capture. Collars were recovered after the timed-release mechanism detached as programmed, the cat was recaptured, or the cat died in the field. Data retrieved from the GPS loggers included the date, time, location, number of satellites and horizontal dilution of precision (HDOP) for each successful location fix. The HDOP provides an inverse index of precision, based on the geometric arrangement of satellites used to calculate a location. To determine a maximum acceptable HDOP threshold for the study, we placed a sample of six collars in three known locations under different levels of vegetation cover (no cover, low shrubs >80% foliage cover, medium-height shrubs >70% foliage cover) and compared the location fixes provided by the collars to the known geographic coordinates of the locations. This showed that 80% of fixes with HDOP 9 were accurate to within 20 m, and 92% were accurate to 50 m (mean error = 17.9 m 16.9 s.d.). We therefore excluded any locations with HDOP >9 from further analysis. Analysis We used an adaptive local convex hull (a-locoh) approach to describe the extent and utilisation distribution (UD) of feral cat home ranges (Getz et al. 2007). LoCoH methods produce a nonparametric UD, with a form arising directly from the data, and generally provide a more realistic characterisation of home ranges than do parametric methods, particularly when home ranges are broken by sharp disjunctions such as shorelines (Getz and Wilmers 2004). The value of a, which determines the variable scale of elements from which home-range estimates are constructed (Getz et al. 2007), was initially chosen for each cat by examining the derivative of the area of estimated home ranges plotted against different values of a to identify a range of potentially useful values. We then plotted the estimated homerange extents of each cat, using the initial a value, and applied the minimum spurious hole-covering rule to determine a final a value for each cat (Getz and Wilmers 2004). We averaged a across all cats and recalculated each cat s home-range extent and 10% isopleths by using this common value. We also calculated homerange areas by using the more commonly used 100% minimum convex polygon (MCP100) method for comparison with previous studies, although MCP results are susceptible to overestimating the actual area used by an animal and should be interpreted with caution (White and Garrott 1990). Within home ranges, we identified those areas that had the greatest intensity of use (core areas) by determining the area of each isopleth of the estimated a-locoh UD between 30% and 100% of all location fixes, at 10% intervals. We defined the core area for each cat as the area enclosed within the isopleth that followed the greatest increase in area between successive isopleths (Wray et al. 1992). We implemented the a-locoh methods using the R package adehabitat version (Calenge 2006) for R version (R Development Core Team 2010). To estimate the numbers of properties traversed by individual cats, we used a cadastral information layer in a geographical information system (GIS) to determine the number of individual properties that intersected each cat s estimated a-locoh homerange extent. All GIS operations were conducted using ArcMAP 9.3 (Environmental Systems Research Institute, Inc., Redlands, CA, USA). To test whether home-range areas varied consistently among seasons or between different classes of cat, we compared a series of nested linear models explaining a-locoh home-range area in terms of sex, landscape type (bush or pastoral) and season (winter or spring). We used likelihood ratio tests (LRTs) to identify a minimal adequate model in a stepwise manner, starting with a maximal model containing a linear combination of all fixed effects. Parameters were estimated using iterative, weighted least-squares estimation implemented with the glm function in R. Home-range estimates were square-root transformed for normality before analysis. To determine whether different classes of cats moved greater distances per day, and would therefore be more likely to encounter control or monitoring devices, we used LRT to compare nested models explaining the daily distances between location fixes in terms of season, sex and landscape type. Individual cats were specified as random effects, and parameters were estimated by restricted maximum likelihood, using the R package lme4 version (Bates and Maechler 2009). To determine whether cats showed specific habitat preferences and would be more likely to encounter control or monitoring devices in specific habitat types, we compared the numbers of fixes for each cat in different habitat types to the area of each habitat type within the home range of the cat. At the bush site, we classified each location fix as occurring in one of the following five broad structural habitat types: open paddock, low open shrubland, mixed shrubland and woodland, low woodland, and medium closed woodland. Shrubland types occurred mainly near the coast, on limestone plains (low open shrubland) or sand dunes (mixed shrubland). Low woodland was mainly mallee regrowth recovering from fire, whereas medium woodland was mostly intact mallee with a sparse understorey. At the pastoral site, only two structural habitat types were used, namely, open paddock and woodland. We estimated the area of each vegetation type in each cat s a-locoh home range by using vegetation-map overlays in ArcMAP. We used the proportion of each habitat type available to each cat to calculate the expected number of location fixes in each habitat type, given a non-preferential distribution of fixes, and tested goodness-of-fit between the observed and expected numbers of fixes in each habitat type using chi-square tests for each cat, at the Šidàk-corrected a level. Preferential use or avoidance of specific habitat types by each cat was assessed by estimating confidence intervals for the proportions of location fixes for the cat in each habitat type, and then determining whether those intervals overlapped with the proportion of habitat available to the cat (Neu et al. 1974). This method of habitat analysis suffers from using location fixes as the experimental unit, rather than individual animals; however, methods such as compositional analysis were inappropriate because of our small sample sizes at each site (Aebischer et al. 1993). Finally, we expected that if cats preferentially used woody vegetation at the pastoral site, they would be recorded close

4 Movements and control of feral cats Wildlife Research 261 to junctions in the treeline more often than expected from a random distribution of location fixes throughout the available woody vegetation, because junctions form nodes connecting branches of preferred habitat. To test this, we categorised each location fix within woody vegetation as occurring within 100 m of a treeline junction or farther than 100 m from a junction. We used a chi-square test to compare this distribution against an expected distribution calculated from the respective areas of each category. Results Useable data were recovered from 13 of the 16 collared cats, including five males and one female at the bush site and four males and three females at the pastoral site. Collars from two other cats could not be recovered and one collar returned insufficient location fixes to describe home-range use. Two cats at the bush site died during the course of the study; one was found emaciated with a long (>8.5 cm) stick in its stomach that may have inhibited its ability to hunt, and the other had consumed a tiger snake (Notechis scutatus). Four cats recovered alive at the conclusion of the study had lost an average of 14% of their initial bodyweight during the period over which they were collared. This was similar to an 11% difference in the mean weight of uncollared adult male cats caught on the island during winter and spring in 2010 (P. Masters, unpubl. data). None of the six cats recovered alive or dead showed any evidence of abrasion or other injury around the collar site. The median number of useful location fixes per cat was 249 (inter-quartile range (IQR) = 265), after removing a median of 8% of fixes with HDOP values >9.0 (IQR = 0.03). Cats were tracked for a median of 71 days (IQR = 21). The median 100% a-locoh home-range area for all 13 cats, calculated using an a-value of 17.6 km, was 5.11 km 2. The distribution of home-range areas was skewed by a large male at the bush site, with a home range 3.2 times larger than the median. The maximum home-range width (distance between farthest relocations) was 11.8 km, or 7.7 km (mean = 4.1 km 1.5 s.d.) excluding one cat with an unusually elongated home range. The largest home ranges at each site belonged to the largest male cats (Table 1) and overlapped the home ranges of all other cats tracked at the site. Home ranges of all cats showed substantial overlap (Fig. 2). Core areas for all cats resolved between the 90% and 100% isopleths, indicating that cats did not have distinct core areas of use within their home ranges at the scale we investigated. One male and one female cat showed distinct bimodal use of their home ranges, spending between 1 and 24 days (median = 6) in one side of their home range before returning to the other side. However, the proportions of location fixes recorded between the two home-range halves were sufficiently large to include the middle area within the total core home-range estimate for each cat. Most cats traversed fewer than 10 individual properties during the study (median = 6), and one cat crossed 32 properties. Model simplification using LRTs indicated that models containing fixed effects for sex, site or season provided no better explanatory power for variation in estimated homerange area than did a null model (c 2 1 < 1.67, P > 0.19). We were unable to test for an interaction between sex and site because of the small sample size and lack of female replication at the bush site. The distribution of daily distances moved by cats was skewed, with 95% of daily movements <2.9 km and the remaining 5% reaching up to 8.0 km. The average daily distance moved, estimated across all cats, was 975 m (95% CI = 718, 1233 m). There was substantial variability within and among cats (intra-class correlation coefficient = 0.17); however, LRTs indicated no statistically detectable differences between sexes, sites or seasons (c 2 1 < 3.03, P > 0.08). Four of the six cats at the bush site showed statistically detectable habitat preferences. All cats at the bush site that had areas of low and medium woodland within their estimated home Table 1. Characteristics of 13 GPS-collared feral cats at bushland and pastoral sites on Kangaroo Island Total and core home ranges were calculated using adaptive nearest local convex hulls (LoCoH) and 100% minimum convex polygons (MCP100). Median male and female characteristics are pooled across both sites. Identity codes are as follows: B = bush site, P = pastoral site, m = male, f = female Animal ID Capture weight (kg) Capture date No. of days tracked LoCoH home-range area (km 2 ) MCP100 area (km 2 ) Max. daily distance moved (m) Bm June Bf June Bm Apr Bm May Bm June Bm May Pf Sep Pf Sep Pm Oct Pm July Pf June Pm Oct Pm June Median Female Male

5 262 Wildlife Research A. J. Bengsen et al. (b) (a) km Fig. 2. Adaptive local convex hull home-range estimates for (a) six adult feral cats at a bushland site, and (b) seven cats at a pastoral site, on Kangaroo Island in Solid lines represent estimated home ranges of male cats and dashed lines represent females. ranges recorded fewer location fixes than expected in these habitat types. Most cats with clear habitat preferences also showed a preference for mixed shrub and woodland (Table 2), except one male which recorded a large proportion of fixes in an expanse of open paddock that separated the two peaks of its UD. Five of the seven cats at the pastoral site showed statistically detectable preferences for woody vegetation. The remaining two cats also showed more location fixes in woody vegetation than expected, but the differences were not statistically detectable at the a level (Table 3). On average, cats recorded 63% more location fixes in woody vegetation than was expected. Treeline junctions, including wooded area within a 100-m radius of the junction centre, comprised 19% of the woody vegetation available to cats at the pastoral site. Location fixes within woody vegetation were 1.9 times more likely to be recorded at a treeline junction than was expected from a random distribution within the entire wooded area (c 2 1 = , P < 0.001). Discussion The weight lost by four cats that were recovered alive at the end of the study suggests that the bulk or weight of collars may have inhibited cats movements and hunting abilities. This, in turn, may have influenced their spatial behaviour. However, these cats were collared over winter, when nutritional demand may be high and some food items such as reptiles are likely to be scarce. Some weight loss might therefore be expected regardless of the presence of collars. Declines in the weight of collared animals during the study were consistent with weight differences in uncollared animals between winter and spring in a different study (P. Masters, unpubl. data). Adult feral cats at bushland and pastoral sites on Kangaroo Island showed substantial variability in the ways that they used their environment, but there were no perceptible differences between sexes in either home-range extent or daily distances moved. These results concur with several previous studies, which have found that mean total home-range sizes tend to be slightly larger for males than females, but variability within sexes is greater than that between sexes (e.g. Jones and Coman 1982; Molsher et al. 2005; Moseby et al. 2009). Nonetheless, some studies have reported consistently larger home-range sizes for males (e.g. Goltz et al. 2008; Buckmaster 2011). The median home-range areas estimated in the present study were somewhat larger than average MCP100 home-range areas reported from comparable areas (4.7 km 2, Jones and Coman 1982; 0.9 km 2, Norbury et al. 1998; 2.48 km 2, Molsher et al. 2005), but much smaller than those reported from arid landscapes (22.1 km 2, Edwards et al. 2001; 26.0 km 2 MCP95, Moseby et al. 2009). Such a high degree of behavioural variability among populations is consistent with expectations for an obligate carnivore exposed to a diverse range of habitat types and accompanying variability in resources and conditions (Gompper and Gittleman 1991). Control programs aiming to reduce the density of pest animals across a site should strive to maximise the proportion of the population exposed to control devices such as traps or bait stations. Devices such as traps or automated toxin-delivery devices (e.g. Read 2010) that can remain in situ for extended Table 2. Habitat preferences for six feral cats at a bushland site on Kangaroo Island Preference or avoidance was determined by comparing distributions of GPS-location fixes within specific habitat types to distributions expected from a non-preferential distribution. Cells with missing values represent habitat types that were not within a cat s home range; null indicates no statistically detectable preference or aversion at the a level. Identity codes are as follows: B = bush site, P = pastoral site, m = male, f = female Cat c 2 d.f. P Habitat type Open paddock Low shrubland Mixed shrub and woodland Low woodland Medium woodland Bm <0.001 Null Prefer Prefer Avoid Avoid Bm <0.001 Avoid Prefer Avoid Bm <0.001 Prefer Null Avoid Bm <0.001 Avoid Prefer Bm Null Null Bf Null Null Combined <0.001

6 Movements and control of feral cats Wildlife Research 263 Table 3. Habitat preferences for seven feral cats at a pastoral site on Kangaroo Island Preference or avoidance was determined by comparing distributions of GPS location fixes within specific habitat types to distributions expected from a non-preferential distribution. null indicates no statistically detectable preference or aversion at the a level. Identity codes are as follows: B = bush site, P = pastoral site, m = male, f = female Cat c 2 d.f. P Habitat type Woodland Open paddock Pf <0.001 Prefer Avoid Pm <0.001 Prefer Avoid Pm <0.001 Prefer Avoid Pf <0.001 Prefer Avoid Pm <0.001 Prefer Avoid Pf Null Null Pm Null Null Combined <0.001 periods and interact with multiple cats should be deployed at densities that allow for each animal to encounter at least one device within its home range. The radius of the smallest home range in our study was 770 m, so a spatial arrangement that ensures that no device is farther than ~770 m from its nearest neighbour should ensure that most adult cats have the potential to encounter at least one device. This equates to a density of ~1.7 devices per square kilometre if devices are evenly distributed. This finding is consistent with the results of a concurrent study that used camera traps to monitor changes in population abundance at the pastoral site following a trapping program. Arrays of camera traps left in the field for 15 days at densities of 1.7 or 2.3 cameras per square kilometre both had 100% power to detect an estimated 55% population decline, whereas power decreased to 96% and 65% at densities of 1.13 and 0.57 cameras per square kilometre, respectively (Bengsen et al. 2011). However, because of the high spatial overlap of home ranges in the present study, control devices that are rendered unavailable to cats after their first use (e.g. baits) should be more closely spaced to ensure that more than one device is initially available to each cat. The preferential use of woody vegetation at the pastoral site suggests that control or monitoring devices should be located in these areas where possible, to maximise the rate at which cats encounter devices. Furthermore, the higher than expected number of location fixes in the vicinity of treeline junctions suggests that these areas should be targeted to increase efficiency. This may also hold for other species that preferentially use linear habitat features (e.g. Mahon et al. 1998; Meek and Saunders 2000); however, the efficacy of such tactics remains to be tested. Habitat preferences were less clear at the bushland site, which supported a wider range of habitat types. Nonetheless, all cats that had access to low or medium woodland tended to be recorded in these habitat types at a lower rate than expected. The poor use of these two habitat types by feral cats suggests that control devices might best be deployed in alternative habitat types where possible, such as the structurally and floristically diverse mixed shrub and woodland, which was apparently preferred by most cats at the bush site. The spatial extent of control programs aiming to reduce feralcat population densities will be determined by a range of technical, logistical, biological and social factors. However, there is likely to be a minimum-area threshold below which episodic control programs will have little discernable effect on local cat abundance because of rapid immigration. Reduction of the cat population at the pastoral site by 18 adults after the completion of the present study resulted in an estimated four new immigrants into the study area within 30 days (Bengsen et al. 2011). Control programs that do not cover large enough areas to inhibit population recovery by immigration may therefore be able to provide only short-term population reductions (see also Thomson et al. 2000; Berry et al. 2012). Most cats in the present study traversed areas with a maximum width of between 3 and 5 km, so cats within several kilometres of a control program might be available as potential immigrants. Furthermore, the high degree of overlap in home ranges at both sites suggests that the movement of cats into areas of reduced population density might not be greatly restricted by territorial boundaries. Regardless of the actual spatial extent of any control program, the observation that all cats traversed several different properties highlights the principle that control programs should be coordinated among neighbouring landholders for optimal efficacy (e.g. McLeod et al. 2010). Given that the average size of properties devoted to primary production or reserves on Kangaroo Island was only 3.39 km 2, and 80% of properties were smaller than the median home range observed in the present study, attempts to reduce cat abundance at the scale of individual properties may be futile unless cats are removed frequently to counter immigration. There is no potential for cats to be eradicated from Kangaroo Island in the foreseeable future because existing cat-control methods (trapping, poison baiting and hunting) are unlikely to be able to remove cats across this large island faster than they can be replaced. Furthermore, there is high potential for humanmediated recolonisation, despite strong local laws that require sterilisation of most domestic cats (Kangaroo Island Council 2010). Consequently, programs aiming to reduce the impacts of feral cats will need to be sustained indefinitely. Management programs aiming to protect spatially restricted resources, such as little-penguin (Eudyptula minor) colonies, may need to operate intensively and near-continuously throughout the period when the resource is vulnerable (e.g. before, and during, penguin breeding season). Programs aiming to reduce the impacts of cats at larger scales, such as to protect livestock from disease or widely distributed native species from predation, may need to be coordinated across much larger areas than currently occurs to have any lasting impact on cat densities. This will be difficult to achieve using trapping or shooting alone, because these methods are very labour intensive (e.g. Short et al. 2002), but may be possible with the future development of novel control methods (e.g. Read 2010), if they can increase efficiency. Acknowledgements This project was funded by the State Government of South Australia and the Invasive Animals Cooperative Research Centre. Some of the GPS collars were loaned by the Arthur Rylah Institute for Environmental Research. Ethics approval was granted by the South Australian Department for Environment

7 264 Wildlife Research A. J. Bengsen et al. and Heritage Wildlife Ethics Committee (Project 52/2009). Comments from Tarnya Cox and three anonymous reviewers improved earlier drafts. References Aebischer, N. J., Robertson, P. A., and Kenward, R. E. (1993). Compositional analysis of habitat use from animal radio-tracking data. Ecology 74(5), doi: / Alterio, N., Moller, H., and Ratz, H. (1998). Movements and habitat use of feral house cats Felis catus, stoats Mstela erminea and ferrets Mustela furo, in grassland surrounding yellow-eyed penguin Megadyptes antipodes breeding areas in spring. Biological Conservation 83(2), doi: /s (97) Bates, D., and Maechler, M. (2009). lme4: linear mixed-effects models using S4 classes. R package version ) Available at [Verified April 2009.] Bengsen, A. J., Butler, J., and Masters, P. (2011). Estimating and indexing feral cat population abundances using camera traps. Wildlife Research 38(8), doi: /wr11134 Berry, O., Algar, D., Angus, J., Hamilton, N., Hilmer, S., and Sutherland, D. (2012). Genetic tagging reveals a significant impact of poison baiting on an invasive species. The Journal of Wildlife Management. [in press] doi: /jwmg.295 Buckmaster, A. J. (2011). Ecology of the feral cat (Felis catus) in the tall forests of far east Gippsland. Ph.D. Thesis, University of Sydney, Sydney. Calenge, C. (2006). The package adehabitat for the R software: a tool for the analysis of space and habitat use by animals. Ecological Modelling 197 (3 4), doi: /j.ecolmodel Department of Environment, Water, Heritage and the Arts. (2008) Threat abatement plan for predation by feral cats. (DEWHA: Canberra.) Dickman, C. R. (1996). Impact of exotic generalist predators on the native fauna of Australia. Wildlife Biology 2(3), Edwards, G., De Preu, N., Shakeshaft, B., Crealy, I., and Paltridge, R. (2001). Home range and movements of male feral cats (Felis catus) in a semiarid woodland environment in central Australia. Austral Ecology 26(1), Gannon, W. L., and Sikes, R. S. Animal Care and Use Committee of the American Society of Mammalogists. (2007). Guidelines of the American Society of Mammalogists for the use of wild mammals in research. Journal of Mammalogy 88(3), doi: /06-mamm-f- 185R1.1 Getz, W., and Wilmers, C. (2004). A local nearest-neighbor convex-hull construction of home ranges and utilization distributions. Ecography 27 (4), doi: /j x Getz, W., Fortmann-Roe, S., Cross, P., Lyons, A., Ryan, S., and Wilmers, C. (2007). LoCoH: nonparameteric kernel methods for constructing home ranges and utilization distributions. PLoS ONE 2(2), doi: /journal. pone Glen, A. S., and Dickman, C. R. (2005). Complex interactions among mammalian carnivores in Australia, and their implications for wildlife management. Biological Reviews of the Cambridge Philosophical Society 80(03), doi: /s Goltz, D. M., Hess, S. C., Brinck, K. W., Banko, P. C., and Danner, R. M. (2008). Home range and movements of feral cats on Mauna Kea, Hawai i. Pacific Conservation Biology 14(3), Gompper, M. E., and Gittleman, J. L. (1991). Home range scaling: intraspecific and comparative trends. Oecologia 87(3), doi: /bf Guttilla, D. A., and Stapp, P. (2010). Effects of sterilization on movements of feral cats at a wildland-urban interface. Journal of Mammalogy 91(2), doi: /09-mamm-a Hilton, G. M., and Cuthbert, R. J. (2010). The catastrophic impact of invasive mammalian predators on birds of the UK overseas territories: a review and synthesis. The Ibis 152(3), doi: /j x x Jones, E., and Coman, B. (1982). Ecology of the feral cat, Felis catus (L.), in south-eastern Australia. III. Home ranges and population ecology in semiarid north-west Victoria. Australian Wildlife Research 9(3), doi: /wr Kangaroo Island Council. (2010). Animal Management Plan Available at Animal_Management_Plan_v7_Final.pdf [Verified 23 November 2011]. Langham, N. P. E., and Porter, R. E. R. (1991). Feral cats (Felis catus L.) on New Zealand farmland. I. Home range. Wildlife Research 18(6), doi: /wr Long, J. L. (2003). Introduced Mammals of the World: Their History, Distribution, and Influence. (CSIRO Publishing: Melbourne.) Mahon, P. S., Banks, P. B., and Dickman, C. R. (1998). Population indices for wild carnivores: a critical study in sand-dune habitat, south-western Queensland. Wildlife Research 25(1), doi: /wr97007 McLeod, R. (2004) Counting the Cost: Impact of Invasive Animals in Australia (Cooperative Research Centre for Pest Animal Control: Canberra.) McLeod, L. J., Saunders, G. R., McLeod, S. R., Dawson, M., and van de Ven, R. (2010). The potential for participatory landscapemanagement to reduce the impact of the red fox (Vulpes vulpes) on lamb production. Wildlife Research 37(8), doi: /wr10082 Medina, F. M., Bonnaud, E., Vidal, E., Tershy, B. R., Zavaleta, E. S., Josh Donlan, C., Keitt, B. S., Corre, M., Horwath, S. V., and Nogales, M. (2011). A global review of the impacts of invasive cats on island endangered vertebrates. Global Change Biology 17, doi: /j x Meek, P., and Saunders, G. (2000). Home range and movement of foxes (Vulpes vulpes) in coastal New South Wales Australian Wildlife Research 27(6), doi: /wr98030 Molsher, R., Newsome, A., and Dickman, C. (1999). Feeding ecology and population dynamics of the feral cat (Felis catus) in relation to the availability of prey in central-eastern New South Wales. Wildlife Research 26(5), doi: /wr98058 Molsher, R., Dickman, C., Newsome, A., and Muller, W. (2005). Home ranges of feral cats (Felis catus) in central-western New South Wales Australian Wildlife Research 32, doi: / WR04093 Moseby, K., Stott, J., and Crisp, H. (2009). Movement patterns of feral predators in an arid environment implications for control through poison baiting. Wildlife Research 36(5), doi: / WR08098 Neu, C. W., Byers, C. R., and Peek, J. M. (1974). A technique for analysis of utilization-availability data. The Journal of Wildlife Management 38(3), doi: / Nogales, M., Martín, A., Tershy, B., Donlan, C., Veitch, D., Puerta, N., Wood, B., and Alonso, J. (2004). A review of feral cat eradication on islands. Conservation Biology 18(2), doi: /j x Norbury, G., Norbury, D., and Heyward, R. (1998). Space use and denning behaviour of wild ferrets (Mustela furo) and cats (Felis catus). New Zealand Journal of Ecology 22(2), O Donoghue, P., and Ford, G. (1986). The prevalence and intensity of Sarcocystis spp infections in sheep. Australian Veterinary Journal 63(9), doi: /j tb08065.x O Donoghue, P., Riley, M., and Clarke, J. (1987). Serological survey for Toxoplasma infections in sheep. Australian Veterinary Journal 64(2), Pech, R. P., and McIlroy, J. C. (1990). A model of the velocity of advance of foot and mouth disease in feral pigs. Journal of Applied Ecology 27(2), doi: / Pimentel, D., Zuniga, R., and Morrison, D. (2005). Update on the environmental and economic costs associated with alien-invasive species in the United States. Ecological Economics 52(3), doi: /j.ecolecon

8 Movements and control of feral cats Wildlife Research 265 R Development Core Team. (2010). R: A Language and Environment for StatisticalComputing. (R Foundationfor StatisticalComputing: Vienna.) Available at [Verified July 2010]. Read, J. (2010). Can fastidiousness kill the cat? The potential for target specific poisoning of feral cats through oral grooming. Ecological Management & Restoration 11(3), doi: /j x Recio, M., Mathieu, R., Maloney, R., and Seddon, P. (2010). First results of feral cats (Felis catus) monitored with GPS collars in New Zealand. New Zealand Journal of Ecology 34(3), Risbey, D. A., Calver, M. C., Short, J., Bradley, J. S., and Wright, I. W. (2000). The impact of cats and foxes on the small vertebrate fauna of Heirisson Prong, Western Australia. II. A field experiment. Wildlife Research 27(3), doi: /wr98092 Schwerdtfeger, P. (2002) Climate. In Natural History of Kangaroo Island. (Eds M. Davies, C. R. Twidale and M. J. Taylor.) pp (Royal Society of South Australia: Adelaide.) Short, J., Turner, B., and Risbey, D. (2002). Control of feral cats for nature conservation. III. Trapping. Wildlife Research 29(5), doi: /wr02015 Thomson, P., Marlow, N., Rose, K., and Kok, N. (2000). The effectiveness of a large-scale baiting campaign and an evaluation of a buffer zone strategy for fox control. Wildlife Research 27(5), doi: / WR99036 White, G. C., and Garrott, R. A. (1990). Analysis of Wildlife Radio-tracking Data. (Academic Press: London.) Wray, S., Cresswell, W., White, P., and Harris, S. (1992). What, if anything, is a core area? An analysis of the problems of describing internal range configurations. In Wildlife Telemetry: Remote Monitoring and Tracking of Animals. (Eds I. G. Priede and S. M. Swift.) pp (Ellis Horwood: Chichester, UK.)