Home range and movements of male feral cats (Felis catus) in a semiarid woodland environment in central Australia

Austral Ecology (2001) 26, 93 101 Home range and movements of male feral cats (Felis catus) in a semiarid woodland environment in central Australia G. P. EDWARDS,* N. DE PREU, B. J. SHAKESHAFT, I. V. CREALY AND R. M. PALTRIDGE Parks and Wildlife Commission of the Northern Territory, PO Box 1046, Alice Springs, Northern Territory 0871, Australia (Email: glen.edwards@nt.gov.au) Abstract There is a paucity of data on the movement patterns of feral cats in Australia. Such data can be used to refine control strategies and improve track-based methods of monitoring populations of feral cats. In this study the home ranges and movements of male feral cats were examined over 3.5 years in a semiarid woodland environment in central Australia. Two home range estimators were used in the examination: (i) minimum convex polygon (MCP); and (ii) fixed kernel. The most widely used method of estimating home range in feral cats is MCP, while the fixed kernel method can be used to identify core areas within a home range. On the basis of the MCP method, the long-term home ranges of feral cats in central Australia were much larger than those recorded elsewhere (mean, 2210.5 ha). Twenty-four hour home ranges were much smaller (mean, 249.7 ha) and feral cats periodically shifted their 24 h ranges within the bounds of their long-term home ranges. Core area analysis indicated marked heterogeneity of space use by male feral cats. Several instances where feral cats moved large distances (up to 34 km) were recorded. These long distance movements may have been caused by nutritional stress. Using data from the literature, it is shown that prey availability is a primary determinant of long-term home range size in feral cats. The relevance of the results to the design of management strategies for feral cats in central Australia is also discussed. Key words: feral cat, fixed kernel, home range, minimum convex polygon, movement patterns. INTRODUCTION *Corresponding author. Accepted for publication July 2000. The house cat (Felis catus L.) has become established in many parts of the world in a variety of situations (Konecny 1987). Cats probably arrived in Australia about 200 years ago (Jones 1989) and feral populations (i.e. those with minimal reliance on humans, sensu Dickman 1996) are now common over much of mainland Australia and Tasmania (Jones 1989). Pioneering research into the ecology of feral cats in Australia took place in the late 1970s and early 1980s (Jones 1977; Jones & Coman 1981, 1982a,b). Recently there has been concern over the impact of feral cats on native fauna (Gibson et al. 1994; Dickman 1996), which has rekindled interest in the dietary ecology of the species in Australia (Paltridge et al. 1997; Molsher et al. 1999; Risbey et al. 1999), methods of monitoring populations (Mahon et al. 1998; Edwards et al. 2000) and in the development of effective control strategies for feral populations (Edwards et al. 1997; Risbey et al. 1997; Short et al. 1997; Anonymous 1999). Despite the renewed interest in feral cats, the early study of Jones and Coman (1982b) in Victoria is the only published account of the home range and movement patterns of feral cats in Australia. From a management viewpoint, the movement patterns of feral cats are of interest from several perspectives. Information on feral cat long-term home range and movement patterns can be used to refine control programmes targeting the species in specified management areas. For example, the information could be used to gauge the spacing distance of control units (traps or poisoned baits) or to delineate the total area over which feral cats need to be controlled in order to remove resident animals and confine immigration to buffer zones on the perimeter of core conservation areas (sensu Saunders et al. 1995). Similarly, information on short-term movement patterns can be used to refine the increasingly used track-based methods of monitoring populations of feral cats and other mammalian predators (Allen et al. 1996; Mahon et al. 1998; Edwards et al. 2000). For example, Edwards et al. (2000) were able to assess the minimum number of feral cats detected during track-based population surveys on the basis of 24 h home range size. In the present paper, we present data on the longterm (>10 months) home range and movement patterns of feral cats at a semiarid site in central Australia, which is very different from the site used by Jones and Coman (1982b). The earlier study was conducted in semiarid mallee habitat where rabbits (Oryctolagus cuniculus) were abundant and were the major prey

94 G. P. EDWARDS ET AL. consumed by the feral cats (Jones & Coman 1982b). Rabbits have been identified as the key prey determining the behaviour and abundance of feral cats (Alterio et al. 1998) and this dietary pattern is typical of cats inhabiting rabbit-infested areas (Bayly 1978; Catling 1988; Paltridge et al. 1997; Molsher et al. 1999; Risbey et al. 1999). The present study was conducted in mixed mulga (Acacia aneura) woodland where rabbits were relatively uncommon and the feral cats relied primarily on small native prey species weighing less than 100 g (G. P. Edwards & N. de Preu, unpubl. data, 1994 1997). In the present paper, we also provide the first published account of feral cat movements over a 24 h period. METHODS Study site The study was conducted in the Northern Territory, approximately 110 km north-west of Alice Springs (Fig. 1). The study site is about 550 km 2 in area (Fig. 1) and is situated on the western part of Hamilton Downs station (23 31 S, 132 58 E) and the bordering areas of two adjacent stations: Milton Park and Narwietooma. The study site is dominated by a flat plain elevated 650 m above sealevel. The southern edge of the plain abuts a range of low hills (elevation 900 1000 m above sealevel) that make up the northern fringe of the MacDonnell Ranges. In the north-east of the study site is Mount Hay (elevation 1250 m above sealevel) and to the north-west is Redbank Hill (elevation 1000 m above sealevel). The temperature regime at the study site is typical of central Australia. Summers are hot with daily temperatures commonly greater than 40 C. January and July are, respectively, the hottest and coldest months (mean maximum temperatures, 36.1 C and 19.3 C; mean minimum temperatures, 21.2 C and 4.0 C; Bureau of Meteorology 1991). During 1984 1996 the mean annual rainfall recorded 30 km east of the study site at Hamilton Downs homestead was 256 mm with a coefficient of variation of 54%. The study site encompasses three land systems (Perry et al. 1962). The Harts land system comprises crystalline ranges with stony shallow soil supporting sparse shrubs and grasses. The Hamilton land system comprises plains with texture contrast soils, some red earths and red clay soils flanking crystalline ranges. This supports a mosaic of open grassland dominated by wiregrass (Aristida contorta) and open mixed woodland with scattered mulga, ironwood (Acacia estrophiolata), witchetty bush (Acacia kempeana), whitewood (Atalaya hemiglauca) and bloodwood (Corymbia opaca). The Bushy Park land system comprises plains with red earths flanked by the Hamilton land system. This supports open to dense stands of mulga woodland with scattered ironwood, witchetty bush and bloodwood over grassland. A series of ephemeral creeks supporting stands of river red gum (Eucalyptus camaldulensis) are scattered throughout the study site (Fig. 1). These rise in the ranges to the south and drain onto the plain. The study site has a history of cattle (Bos taurus) grazing and there are a number of permanent artificial water points (bores and dams in Fig. 1). However, Milton Park paddock, which comprises much of the study site, contained very few cattle until January 1995. Dingoes (Canis lupus dingo) and feral cats occur throughout the study area, but red foxes (Vulpes vulpes) and rabbits are rare or absent (Edwards et al. 2000). Fig. 1. Map of the study area showing (, ) water points, ( ) station roads, ( ) areas of relief and (- - - -) ephemeral creeks. Inset map of the Northern Territory shows ( ) the study site in relation to ( ) Alice Springs.

MOVEMENT PATTERNS OF FERAL CATS 95 Movement patterns and home range determinations The movements of feral cats were studied using radio telemetry over the period July 1994 to December 1997. Feral cats were captured either in bait hook cage traps (325 mm 325 mm 740 mm) baited with fresh kangaroo meat or commercial cat food, or in Victor 1 Soft Catch traps (Woodstream Corporation, Lititz, USA) baited with a scent attractant (Edwards et al. 1997). Captured animals were anaesthetized with an intramuscular injection of one part xylazine (Rompun; Bayer Australia, Sydney, Australia) to two parts ketamine hydrochloride (Ketalar; Parke Davis Australia, Sydney, Australia) at a dose rate of 0.1 ml per kg of body mass. Cats were measured, weighed and their sex determined together with the reproductive condition of females. Cats were assigned a body condition score of one (poor) to five (excellent) on the basis of visual assessment and by considering body mass in relation to body length. All feral cats were fitted with 115 g radio collars (model TX2; Biotelemetry, Adelaide, Australia), which were approximately 3.4% of mean body mass, and subsequently released at the point of capture when they had recovered from the anaesthetic. We collared only subadult (minimum mass 1.9 kg; Jones & Coman 1982b) and adult cats. Radio-collared feral cats were hand-tracked approximately once every 2 weeks using a folding three element hand-held directional antenna (Sirtrack; Havelock North, New Zealand) and a portable scanner/receiver (model RX3; Biotelemetry). When signal strength indicated that the tracked animal was close, care was taken to minimize disturbing the animal. Once located, the cat s position was recorded using a Global Positioning System (GPS) (Magellan Pro Mark 10; Magellan Systems Corporation, San Dimas, USA). In cases where the cat was disturbed, the approximate initial position of the animal was recorded. The time and habitat in which the cat was located were also recorded. GPS fixes are updated every second (Magellan Systems Corporation 1994) and can vary due to satellite geometry and Selective Availability (McElroy et al. 1998). In 95% of cases, fixes are within 100 m of the true location (McElroy et al. 1998). Norbury et al. (1998) recorded an average error of 83 m in random directions for uncorrected GPS readings. On three occasions a light aircraft was used to locate feral cats whose radio collar transmitter signals could not be picked up within the study area. If radio-collared cats died during the study, the condition of the carcass was recorded in order to estimate an approximate date of death. On four occasions (August/September 1994, November 1994, March 1995, July 1995) intensive tracking sessions from two fixed stations were conducted to assess 24 h movements of radio-collared cats over two to three successive days. The tracking stations were situated 3 4 km apart atop hills in the south of the study area. Each tracking station was fitted with two 4 m, eight-element Yagi antennas (model RA-4B, gain 11.8 db; Telonics, Mesa, USA) mounted in parallel for vertical polarization. A two-port precision phase combiner (model TAC-5; Telonics) put the signals from each antenna 180 out of phase. The bearing of a transmitter signal was obtained using a fixed compass rose by finding the null point between the two loudest peaks of the combined signal. Bearings were taken on one fixed-point transmitter placed at a known location, which was used to calibrate the system and on two to five feral cats every 15 min during a tracking session. Data collection times were synchronized by radio. The width of the null point was recorded to the nearest 1 together with the strength of the signal (weak or strong). Occasionally, outside the intensive tracking sessions, the location of feral cats was determined using the station tracking system rather than by hand-tracking. Data analysis We used hand tracking radio locations to calculate the long-term home ranges of feral cats. We used the intensive station tracking data partitioned into contiguous 24 h sets to calculate the 24 h home ranges of feral cats. Successive locations over short times tend to be autocorrelated (Swihart & Slade 1985) and this can lead to large errors with some home range estimators, especially when sample sizes are small (Anderson 1982). It can also preclude analysis of home range estimates using parametric statistics (de Solla et al. 1999). Nonetheless, attempts to eliminate autocorrelation prior to analysis may be unwise (Lair 1987; de Solla et al. 1999) because it is often an intrinsic pattern arising from the behaviour of the animal under study and removing it may limit the biological significance of the result (de Solla et al. 1999). Furthermore, autocorrelation should not introduce unnecessary bias to home range estimates if the time interval between successive locations is relatively constant (de Solla et al. 1999). The 24 h data sets of the present study conformed to this requirement, precluding any need to eliminate autocorrelation. For the 24 h surveys, the positions of feral cats were calculated by triangulation and converted to Universal Transverse Mercator units in metres (UTM) for home range analysis. Data were edited using Arcview 3.0a (ESRI, Redlands, CA, USA). Locations with wide nulls (>10 ), which appeared inconsistent with the readings taken immediately before and after, were deleted. Home range size was assessed using the fixed kernel method (Worton 1995), the minimum convex polygon method (MCP; Mohr 1947) and an index based on the mean distance of the location of each feral cat from its

96 G. P. EDWARDS ET AL. home range centre (determined as the harmonic mean centre; Dixon & Chapman 1980), following Norbury et al. (1998). The MCP is the most commonly used technique for estimating the home ranges of cats. Home ranges based on the MCP method were calculated using the CALHOME package (Kie et al. 1994). Among several problems of the MCP method (Jennrich & Turner 1969; Anderson 1982; Boulanger & White 1990) is that a convex polygon is fitted to all data points, irrespective of their actual distribution (Anderson 1982). If the distribution is not convex, the fitted polygon includes unused areas that can inflate home range estimates (Anderson 1982). The MCP method provides an estimate of total home range only. The fixed kernel method compares well with the best methods available for estimating home range size (Worton 1995). It can model a data distribution of any shape (Seaman & Powell 1996) and the resultant utilization distribution density estimate can be converted into probability contours delineating regions of differential space use (Worton 1995). Fixed kernel estimates were calculated using 64 64 grids, the boundary limits of which are determined by the individual data sets. The smallest area under the utilization distribution encompassing 95% of points was used to define the total home range area and the smallest area encompassing 50% of points defined the core home range. Home ranges based on the fixed kernel method were calculated using the HOME RANGERsoftware package (Version 1.5; Hovey 1999). HOME RANGER uses least squares cross-validation to select the optimal value of the smoothing parameter or bandwidth (h) in calculating the utilization distribution. We used the HOME RANGER standardizing procedure to remove covariance between x and y coordinates in selecting h. This procedure does not remove autocorrelation and the selected h is then applied to the original data. Worton (1995) found that post-hoc adjustments to h were needed to obtain unbiased estimates of the true home range size (based on the 1.0 h estimate which is the best estimate of the true home range) and advocated using Monte Carlo procedures to select an appropriate value for the adjustment. We used the HOME RANGER bootstrap routine to perform this task for each data set using 100 randomizations. Harmonic mean centres of the home range data sets were calculated using software written by D. B. Croft. The size of the error polygons associated with locational data points obtained with tracking systems like the one used here is influenced by several factors. These include the precision and accuracy of the bearings, the distance between the tracking stations and the location of the animal in relation to the tracking stations (Telonics 1982). To give some idea of the resulting error in 24 h home range determinations, we used a Monte Carlo procedure to vary the bearings from the tracking stations (null midpoints) by integer values within the range 2 to + 2. This procedure was performed 25 times for each 24 h data set. The indices of long-term and 24 h total home range size (mean distance from home range centre) were compared with t-tests using SYSTAT 5.03 (Systat, Evanston, IL, USA), and t-tests were also used to compare the mean percentage ratios of core home ranges; total home range for the fixed kernel estimations to the value of approximately 53% expected if feral cats distributed their activities uniformly within their home ranges. Percentage ratios were arcsine transformed before analysis. Correlation analyses were performed with SYSTAT. RESULTS Home range size In total, 19 feral cats were radio-collared during the study, but only four adult males provided sufficient data to determine long-term home ranges. These were monitored over periods of 15 24 months when 25 50 locations were recorded for each animal. A fifth adult male occupied a stable home range near its point of capture for approximately 4 months, then moved 17 km to a new home range, which it occupied for 8 months before dying. Ten radio-collared adult cats captured during the winter of 1994 died within 1 2 months. These cats had a mean body condition score of only 2.2/5, well below the mean for cats captured subsequently (3.7/5). Four of these animals (two males, two females) moved out of the study area shortly after being released and were later found 15.8 33.8 km from their points of capture. A subadult male cat captured in May 1996 also died. For three cats, radio signals ceased shortly after collars were fitted. During the daytime, the four hand-tracked feral cats were mainly found in areas affording good cover including creeklines, mulga woodland and rocky hills (mean, 90% of 104 locations; SD 9%). Cats were found sheltering in hollow stumps, brush piles, rock crevices and under bushes. Long-term total home range estimates were similar for the MCP (mean, 2210.5 ha) and fixed kernel methods (mean, 2206.3 ha) (Table 1). Mean long-term core home ranges were 578.6 ha (Table 1). The areas used by the four male cats were exclusive (non-overlapping) and contiguous. Post-hoc values for fixed kernel smoothing parameter (h) ranged between 0.8 and 1.0. Twenty-four hour home ranges were measured for three adult males, two of which were used in the calculation of long-term home ranges. Seventeen sets of locations over 24 h were obtained (range, 3 9 per individual). Each data set comprised 44 95 locations. The MCP estimates of 24 h total home range (mean,

MOVEMENT PATTERNS OF FERAL CATS 97 249.7 ha) were larger than fixed kernel estimates (mean, 103.1 ha) (Table 1). Mean 24 h core home ranges were 18.6 ha (Table 1). Post-hoc values for h ranged between 0.8 and 1.0. Feral cats remained on average within 2.1 km of the centres of their long-term home ranges (Table 1) and within 0.8 km of the centres of their 24 h home ranges. On this basis, long-term total home ranges were significantly larger than 24 h total home ranges (t 0.05,(2),19 = 6.5, P < 0.001). Feral cats periodically shifted their 24 h home ranges within the bounds of their long-term home range (Fig. 2). The mean distance between the centres of successive 24 h home ranges was 1.2 km (SD 1.0 km, n = 7). The core areas used by feral cats were on average 25.4% (SD 2.0%) of the long-term total home ranges and 19.9% (SD 4.4%) of 24 h total home ranges. These values were significantly lower than the Table 1. Characteristics of the long-term and 24 h home ranges of feral cats. Shown are the mean total home ranges based on the minimum convex polygon (MCP) and fixed kernel methods, the core home ranges (fixed kernel method) and the mean distance of locations from the home range centre. Standard deviations are in parentheses Home range MCP total (ha) Kernel total (ha) Kernel core (ha) Distance from centre (m) n Long-term 2210.5 ( 469.3) 2206.3 ( 1265.5) 578.6 ( 355.2) 2130.0 ( 219.2) 4 Twenty-four hour 249.7 ( 269.5) 103.1 ( 91.9) 18.6 ( 13.9) 790.3 ( 391.8) 17 Fig. 2. Map of the study area showing ( ) the long-term home range and five examples of 24 h home ranges (, 31 Aug 1 Sep 1994;, 1 2 Sep 1994;, 7 8 Sep 1994;, 23 24 Nov 1994;, 11 12 July 1995) of a 3.6 kg male feral cat. All data points are shown. Water points, roads, areas of relief and ephemeral creeks are shown in Fig. 1.

98 G. P. EDWARDS ET AL. expected value of approximately 53% (long-term: t 0.05,(2),3 = 24.5, P < 0.001; 24 h: t 0.05,(2),16 = 25.6, P < 0.001). Accuracy of locational data from the tracking stations The Monte Carlo procedure disrupted the tight nature of the 24 h data sets resulting in total home range estimates that were larger than those based on the original data (MCP Monte Carlo mean, 306.4 ha, SD 303.3 ha; fixed kernel Monte Carlo mean, 187.9 ha, SD 109.0 ha). Nevertheless, the Monte Carlo simulations still grossly underestimated the long-term total home ranges. In reality the error in 24 h home range estimates was much smaller than indicated by the Monte Carlo analysis because bearings were often accurate to within 0.5. Therefore, we are confident that our estimates of 24 h home range are reliable. The fact that the 24 h home range distributions of each feral cat tended to fall within the bounds of its long-term home (Fig. 2) reinforces our position in this regard. DISCUSSION Despite considerable effort being spent catching and radio-collaring feral cats, only four individuals provided locational data of sufficiently high quality to determine home range size. Our main problem was the low density of feral cats at the study site (approximately 0.1 km 2 ; Table 2). Other problems were the mortality of collared animals, possible collar failures and/or the disappearance of animals from the study site. Only male home range was measured, although other studies indicate that females have similar or smaller home ranges (Table 2). The results of this and other studies (Table 2) indicate that most adult feral cats are sedentary and occupy home ranges for 10 months or longer. The male feral cats in the present study had the largest home ranges ever recorded in the species (Table 2), possibly as a result of low availability of prey (Corbett 1979; Hixon 1980). It is difficult to compare prey availability across studies because often only a subset of the prey was monitored (e.g. introduced rabbits: Corbett 1979; Jones & Coman 1982b; Norbury et al. 1998). Furthermore, estimates of prey abundance may not reflect actual prey availability. For example, birds that inhabit lower vegetation strata or that frequently drink at waterholes, are more prone to predation by feral cats than are those living in the upper vegetation strata, and young rabbits are more vulnerable than older rabbits (Jones 1977; Corbett 1979; Catling 1988; Paltridge et al. 1997). As it happens, the number of consumers present in a habitat often reflects a great deal of information concerning the availability of resources in that habitat (Rosenzweig 1991). For feral cats, density provides a useful surrogate for available prey as the two are directly related (Jones 1977; Corbett 1979; Jones & Coman 1982b). Pearson s correlation analysis following logarithmic transformation on the data from Table 2 showed that home range size and density in feral cats are significantly negatively correlated (males: r = 0.92, P = 0.003; females: r = 0.86, P = 0.013), which suggests that prey availability is a primary determinant of home range size in feral cats. Preyinduced changes in home range behaviour have been shown in other mammalian predators including lynx (Lynx canadensis) (Ward & Krebs 1985; Poole 1994), European polecats (Mustela putorius) (Brzezinski et al. 1992), feral ferrets (Mustelo furo) (Norbury et al. 1998), red foxes (MacDonald 1981; Marlow 1992) and coyotes (Canis latrans) (Mills & Knowlton 1991). Secondary factors influencing home ranges in feral cats Table 2. Comparison of the mean long-term total home ranges of adult feral cats across studies. The minimum convex polygon method (MCP) was used to calculate home range except where indicated Male home Female home Mean density range (ha) range (ha) (no. km 2 ) Location Source 20.6 9.30 000. 0 15.0 22.0 Scotland & Outer Hebrides, UK a Corbett (1979) 42.0 ( 25.5) 3.7 Monarch Isles, UK Corbett (1979) 615.0 ( 274.5) 170.0 ( 141.4) 0.6 b Victoria, Australia Jones & Coman (1982b) 155.0 ( 112.7) c 84.0 ( 54.1) c 1.1 d North Island, New Zealand Fitzgerald & Karl (1986) 304.4 ( 297.1) 93.3 ( 102.0) 2.2 2.5 Galápagos Islands Konecny (1987) 239.0 ( 97.0) e 154.0 ( 21.0) e. 3.5 North Island, New Zealand a Langham & Porter (1991) 189.0 ( 218.0) 249.0 ( 208.0) 0.6 1.4 South Island, New Zealand Norbury et al. (1998) 2210.5 ( 469.3) 0.1 f Central Australia Present study a Farm cats with some reliance on humans; b based on uncorrected spotlight data in Table 2 of Jones & Coman (1982b); c MCP not used, approximate home range only; d based on the minimum number known to be present (Fitzgerald & Karl 1979); e night data only; f based on spotlight data in Edwards et al. (2000) using 280 m strip width (sensu Jones & Coman 1982b). Standard deviations of home ranges are in parentheses.

MOVEMENT PATTERNS OF FERAL CATS 99 include habitat patchiness (Konecny 1987) and social status (Corbett 1979). Patterns of space use by adult feral cats appear consistent with two energy-based models of animal behaviour (Konecny 1987): (i) Carpenter and MacMillen s (1976) threshold model of territoriality; and (ii) Hixon s (1980) feeding territory model for food energy maximizers. In resource-poor environments, population densities are low and feral cats occupy large exclusive home ranges (Corbett 1979; Jones & Coman 1982b; Alterio et al. 1998; present study). This ensures at least maintenance levels of energy intake (Konecny 1987). Although population densities are higher in richer environments, territoriality bestows fewer benefits and as a result individuals occupy smaller home ranges with greater overlap (Corbett 1979; Fitzgerald & Karl 1986; Konecny 1987; Langham & Porter 1991). Where prey declines suddenly or is at extremely low resource levels, home ranges may shift or be abandoned altogether (Norbury et al. 1998). On the basis of feral cat density, prey availability at the present study site in central Australia was much lower than at any other site listed in Table 2. The lack of rabbits at our site appeared to be the key factor underpinning this difference. At all of the other sites in Table 2, with the exception of the Galápagos (Konecny 1987), rabbits were abundant and were the main prey eaten by the feral cats. At our site, the feral cats ate mainly native prey including small mammals, reptiles, birds and invertebrates (G. P. Edwards & N. de Preu, unpubl. data, 1994 1997). The long-term total home ranges of feral cats in central Australia were 10 (MCP) to 20 (fixed kernel) times larger than the 24 h home ranges. Konecny (1987) similarly found that monthly home ranges of feral cats on the Galápagos averaged 31% of the long-term total home ranges. The periodic shifting of 24 h home ranges within the bounds of the larger longterm home ranges by male feral cats in central Australia may be a manifestation of territoriality, such movements providing the mechanism by which the large territory can be covered on a regular basis. On average, adult male feral cats in central Australia concentrated 50% of their space use in only 25% of their overall long-term home range areas and in only 20% of their overall 24 h home range areas, indicating marked heterogeneity in space use. Konecny (1987) reported a similar pattern of space use for feral cats occupying a patchy habitat in the Galápagos, but not for feral cats in a uniform habitat. Heterogeneous space use patterns are probably typical of all species that live in patchy environments, and have been reported in other felids including cougars (Puma concolor) (Seidensticker et al. 1973; Belden et al. 1988), leopards (Panthera pardus) (Hamilton 1976 cited in Corbett 1979) and bobcats (Felis rufus) (Wassmer et al. 1988). The results of the present study can be used to guide the management of feral cats in similar areas of central Australia. Based on long-term home ranges, control units placed at 1 2 km intervals within a grid network should be encountered by most feral cats within a specified control area, particularly if suitable attractants are used (Clapperton et al. 1994; Edwards et al. 1997). In the case where a specific problem cat needs to be removed, the placement of control units within a 2.5 km radius of a recent confirmed sign of the target animal would be an appropriate strategy to control that individual. Buffer zones approximately 4 km wide should be sufficient to protect a core area from resident cats, but widths of up to 20 km or more may be needed to absorb dispersing individuals (see also Jones & Coman 1982b; Norbury et al. 1998). ACKNOWLEDGEMENTS This study was jointly funded by the Parks and Wildlife Commission of the Northern Territory and Environment Australia (formerly the Australian Nature Conservation Agency). The study was approved by the Alice Springs Animal Ethics Committee. We thank David Croft for providing computer programs used in the calculation of home range parameters and the many volunteers who assisted with field work, particularly Allison Foster, Bev Gray, Maria McCoy, Jok Markham and Keith Drew. Comments by David Croft, Ken Johnson, Dave Lawson, Tony Bowland, Bill Freeland, Michael Bull and Allison Foster greatly improved this paper. We also thank the owners and managers for allowing us to work on Hamilton Downs Station. REFERENCES Allen L., Engeman R. & Krupa H. (1996) Evaluation of three relative abundance indices for assessing dingo populations. Wildl. Res. 23, 197 206. Alterio N., Moller H. & Ratz H. (1998) Movements and habitat use of feral house cats Felis catus, stoats Mustela erminea and ferrets Mustelo furo, in grassland surrounding yellow-eyed penguin Megadyptes antipodes breeding areas in spring. Biol. Conserv. 83, 187 94. Anderson D. J. (1982) The home range: a new non-parametric estimation technique. Ecology 63, 103 12. Anonymous (1999) Threat Abatement Plan for Predation by Feral Cats. Biodiversity Group, Environment Australia, Canberra. Bayly C. P. (1978) Observations on the food of the feral cat (Felis catus) in an arid environment. Sth Aust. Nat. 51, 22 4. Belden R. C., Frankenberger W. B. & Schwikert S. T. (1988) Panther habitat use in Southern Florida. J. Wildl. Manag. 52, 660 3. Boulanger J. G. & White G. C. (1990) A comparison of homerange estimators using Monte Carlo simulation. J. Wildl. Manag. 54, 310 5. Brzezinski M., Jedrzejewski W. & Jedrzejewska B. (1992) Winter home ranges and movements of polecats Mustela putorius in Bialowieza Primeval Forest, Poland. Acta Theriol. 37, 181 91.

100 G. P. EDWARDS ET AL. Bureau of Meteorology (1991) Climatic averages, May, 1991. Bureau of Meteorology, Darwin. Carpenter F. L. & MacMillen R. E. (1976) Threshold model of feeding territoriality and test with a Hawaiian honeycreeper. Science 194, 639 42. Catling P. C. (1988) Similarities and contrasts in the diets of foxes, Vulpes vulpes, and cats, Felis catus, relative to fluctuating prey populations and drought. Aust. Wildl. Res. 15, 307 17. Clapperton B. K., Eason C. T., Weston R. J., Woolhouse A. D. & Morgan D. R. (1994) Development and testing of attractants for feral cats, Felis catus L. Wildl. Res. 21, 389 99. Corbett L. K. (1979) Feeding ecology and social organisation of wildcats (Felis silvestris) and domestic cats (Felis catus) in Scotland. Unpublished PhD Dissertation, University of Aberdeen, Aberdeen, Scotland. Dickman C. R. (1996) Overview of the impact of feral cats on Australian native fauna. Australian Nature Conservation Agency, Canberra. Dixon K. R. & Chapman J. A. (1980) Harmonic mean measure of animal activity areas. Ecology 61, 1040 4. Edwards G. P., de Preu N. D., Shakeshaft B. J. & Crealy I. V. (2000) An evaluation of two methods of assessing feral cat and dingo abundance in central Australia. Wildl. Res. 27, 143 9. Edwards G. P., Piddington K. C. & Paltridge R. M. (1997) Field evaluation of olfactory lures for feral cats (Felis catus L.) in central Australia. Wildl. Res. 24, 173 83. Fitzgerald B. M. & Karl B. J. (1979) Foods of feral house cats (Felis catus L.) in forest of the Orongorongo Valley, Wellington. NZ J. Ecol. 6, 107 26. Fitzgerald B. M. & Karl B. J. (1986) Home range of feral cats (Felis catus L.) in forest of the Orongorongo Valley, Wellington, New Zealand. NZ J. Ecol. 9, 71 81. Gibson D. F., Lundie-Jenkins G., Langford D. G., Cole J. R., Clarke D. E. & Johnson K. A. (1994) Predation by feral cats, Felis catus, on the rufous hare-wallaby, Lagorchestes hirsutus, in the Tanami Desert. Aust. Mammal 17, 103 7. Hamilton P. H. (1976) The movements of leopards in Tsavo National Park, Kenya, as determined by radio tracking. Unpublished MSc Thesis, University of Nairobi, Nairobi. Hixon M. A. (1980) Food production and competitor density as the determinants of feeding territory size. Am. Nat. 115, 510 30. Hovey F. (1999) The Home Ranger, Version 1.5. Electronic User s Manual. British Columbia Forest Service Research Branch, Revelstoke, Canada. Jennrich R. I. & Turner F. B. (1969) Measurement of noncircular home range. J. Theor. Biol. 22, 227 37. Jones E. (1977) Ecology of the feral cat Felis catus (L.), (Carnivora: Felidae) on Macquarie Island. Aust. Wildl. Res. 4, 249 62. Jones E. (1989) Felidae. In: Fauna of Australia. Mammalia 1B (eds D. W. Walton & B. J. Richardson) pp. 1006 11. Australian Government Publishing Service, Canberra. Jones E. & Coman B. J. (1981) Ecology of the feral cat, Felis catus (L.), in south-eastern Australia I. Diet. Aust. Wildl. Res. 8, 537 47. Jones E. & Coman B. J. (1982a) Ecology of the feral cat, Felis catus (L.), in south-eastern Australia II. Reproduction. Aust. Wildl. Res. 9, 111 19. Jones E. & Coman B. J. (1982b) Ecology of the feral cat, Felis catus (L.), in south-eastern Australia III. Home ranges and population ecology in semiarid north-west Victoria. Aust. Wildl. Res. 9, 409 20. Kie J. G., Baldwin J. A. & Evans C. J. (1994) CALHOME home range analysis program. Electronic User s Manual. U.S. Forest Service, Fresno and Albany, USA. Konecny M. J. (1987) Home range and activity patterns of feral house cats in the Galápagos Islands. Oikos 50, 17 23. Lair H. (1987) Estimating the location of the focal center in red squirrel home ranges. Ecology 68, 1092 101. Langham N. P. E. & Porter R. E. R. (1991) Feral cats (Felis catus L.) on New Zealand farmland. I. Home range. Wildl. Res. 18, 741 60. MacDonald D. W. (1981) Resource dispersion and the social organization of the red fox (Vulpes vulpes). In: Proceedings of the Worldwide Furbearer Conference (eds J. A. Chapman & D. Pursely) pp. 918 49. Frostburg, Maryland. Magellan Systems Corporation (1994) User s Guide for the Magellan GPS Promark X and the Magellan GPS Promark X CP. Magellan Systems Corporation, San Dimas, USA. Mahon P. S., Banks P. B. & Dickman C. R. (1998) Population indices for feral carnivores: a critical study in sand-dune habitat, south-western Queensland. Wildl. Res. 25, 11 22. Marlow N. J. (1992) The ecology of the introduced red fox (Vulpes vulpes) in the arid zone. Unpublished PhD Dissertation, University of New South Wales, Sydney. McElroy S., Robins I., Jones G. & Kinlyside D. (1998) Exploring GPS. A GPS User s Guide. The Global Positioning System Consortium, Sydney. Mills L. S. & Knowlton F. F. (1991) Coyote space use in relation to prey abundance. Can. J. Zool. 69, 1516 21. Mohr C. O. (1947) Table of equivalent populations of North American small mammals. Am. Midl. Nat. 37, 223 49. Molsher R., Newsome A. E. & Dickman C. R. (1999) Feeding ecology of the feral cat (Felis catus) in relation to availability of prey in central-eastern New South Wales. Wildl. Res. 26, 593 607. Norbury G. L., Norbury D. C. & Heyward R. P. (1998) Behavioral responses of two predator species to sudden declines in primary prey. J. Wildl. Manag. 62, 45 58. Paltridge R., Gibson D. & Edwards G. (1997) Diet of the feral cat (Felis catus) in central Australia. Wildl. Res. 24, 67 76. Perry R. A., Mabbutt J. A., Litchfield W. H. & Quinlan T. (1962) Part II. Land systems of the Alice Springs area. In: Lands of the Alice Springs Area, Northern Territory, 1956 7. (ed. R. A. Perry) pp. 20 128. CSIRO, Melbourne. Poole K. G. (1994) Characteristics of an unharvested lynx population during a snowshoe hare decline. J. Wildl. Manag. 58, 608 18. Risbey D. A., Calver M. & Short J. (1997) Control of feral cats for nature conservation. I. Field tests of four baiting methods. Wildl. Res. 24, 319 26. Risbey D. A., Calver M. C. & Short J. C. (1999) The impact of cats and foxes on the small vertebrate fauna of Heirisson Prong, Western Australia. I. Exploring potential impact using diet analysis. Wildl. Res. 26, 621 30. Rosenzweig M. L. (1991) Habitat selection and population interactions: the search for mechanism. Am. Nat. 137, S5 28. Saunders G., Coman B., Kinnear J. & Braysher M. (1995) Managing Vertebrate Pests. Foxes. Australian Government Publishing Service, Canberra. Seaman D. E. & Powell R. A. (1996) An evaluation of the accuracy of kernel density estimators for home range analysis. Ecology 77, 2075 85. Seidensticker J. C., Hornocker M. G., Wiles M. V. & Messick J. P. (1973) Mountain lion social organization in the Idaho Primitive area. Wildl. Monogr. 35, 1 60.

MOVEMENT PATTERNS OF FERAL CATS 101 Short J., Turner B., Risbey D. A. & Carnamah R. (1997) Control of feral cats for nature conservation. II. Population reduction by poisoning. Wildl. Res. 24, 703 14. de Solla S. R., Bonduriansky R. & Brooks R. J. (1999) Eliminating autocorrelation reduces biological relevance of home range estimates. J. Anim. Ecol. 68, 221 34. Swihart R. K. & Slade N. A. (1985) Testing for independence of observations in animal movements. Ecology 66, 1176 84. Telonics (1982) Technical Information on Telonics RA-NS Precision Direction Finding Antenna Array Series. Telonics, Mesa, USA. Ward R. M. & Krebs C. J. (1985) Behavioural responses of lynx to declining snowshoe hare abundance. Can. J. Zool. 63, 2817 24. Wassmer D. A., Guenther D. D. & Layne J. N. (1988) Ecology of the bobcat in south-central Florida. Bull. Florida State Mus. Biol. Sci. 33, 159 228. Worton B. J. (1995) Using Monte Carlo simulation to evaluate kernel-based home range estimators. J. Wildl. Manag. 59, 794 800.