Landcare Research, PO Box 69, Lincoln, New Zealand

Ecology 2002 39, Dispersal and survival of juvenile feral ferrets Mustela furo Blackwell Science Ltd in New Zealand ANDREA E. BYROM Landcare Research, PO Box 69, Lincoln, New Zealand Summary 1. Introduced feral ferrets Mustela furo are a significant pest of both conservation and economic importance in New Zealand. Ferrets prey on indigenous wildlife and they also carry bovine tuberculosis (TB), a disease which may threaten New Zealand s international beef, dairy and venison markets. 2. Very little is known about the role of dispersal and survival of juvenile ferrets in the recovery of ferret populations after control operations, and how these parameters might affect ferret impacts on native wildlife and their role in the spread of TB. 3. Fifty-two juvenile ferrets were radio-collared at emergence from their natal dens on six study sites during 2 years, 1997 98 and 1998 99. On three sites, most introduced mammalian predators (including ferrets) had been removed by kill-trapping from October to January to protect critical bird nesting areas. Three sites were left as non-removal sites. Dispersal and survival of juvenile ferrets was therefore measured at low and normal ferret densities. 4. Survival of juvenile ferrets showed a strong density-dependent response to predator removal. In year 1, survival of juvenile ferrets from emergence to the end of the dispersal period (about 4 months) was 86% (95% confidence limits 63 100%) on removal sites, whereas survival on non-removal sites was 19% (0 38%). In year 2, survival was 100% on removal sites compared with 71% (45 97%) on non-removal sites. Dispersal of juvenile ferrets was not related to population density at source. 5. A frequency distribution of dispersal distances of juvenile ferrets showed a typical negative-exponential pattern. The median dispersal distance of juvenile ferrets was 5 0 km (range 0 5 45 0 km). There was no apparent sex bias in survival, dispersal distances or timing of dispersal in either season. 6. Density-dependent survival of juvenile ferrets should be a key parameter of interest in determining frequency and seasonal timing of ferret control. Ferret control to conserve native wildlife and minimize TB spread should be carried out in late autumn after dispersal, to provide a longer lag time before juveniles reinvade an area. 7. This study fills a critical gap in the understanding of introduced mustelids in island systems such as New Zealand. It has direct implications for management of this important introduced predator, and it augments literature on dispersal and survival of mammalian carnivores. Key-words: bovine tuberculosis, density-dependence, mustelidae, predator control, secondary prey, vertebrate pest Ecology (2002) 39, Ecological Society Introduction Ferrets Mustela furo L. were introduced into New Zealand in the 1880s to control burgeoning populations of European rabbits Oryctolagus cuniculus L. They Correspondence: A.E. Byrom, Landcare Research, PO Box 69, Lincoln, New Zealand (fax + 64 3 325 2418; e-mail ByromA@landcare.cri.nz). are now considered to be a pest of both conservation and economic importance (Lavers & Clapperton 1990). While they depend to a great extent on rabbits as their primary prey, ferrets prey on indigenous wildlife as secondary prey (Pierce 1986; Alterio, Moller & Ratz 1998). After the Australian brushtail possum Trichosurus vulpecula Kerr, ferrets are regarded as the species most likely to threaten New Zealand s international beef, dairy and venison industries because they carry bovine

68 A.E. Byrom tuberculosis (TB) and transmit the disease to livestock (Ragg, Moller & Waldrup 1995; Ragg, Waldrup & Moller 1995; Lugton et al. 1997). New Zealand has the largest population of feral ferrets in the world (Blandford 1987), yet there are still gaps in our understanding of the population ecology of this important introduced predator. For example, very little research has been done on dispersal and survival of juvenile ferrets in New Zealand. Long-distance movements by ferrets are an important factor in rates of recovery of populations after control operations, and may also contribute to the spread of bovine TB (Caley & Morriss 2001). The lack of published information on dispersal behaviour of juvenile ferrets has therefore limited our understanding of the influence of immigration and emigration on population dynamics, and also of measures that can be taken to restrict dispersal for conservation or to prevent the spread of disease. In this paper, results are presented from a radio-tracking study of juvenile ferrets in New Zealand s South Island from 1997 to 1999. The South Island of New Zealand contains significant areas of semi-arid tussock grasslands and large braided riverbeds 0 5 2 0 km in width. Ferrets are particularly common in these areas, where their population dynamics are closely linked to rabbit numbers. Braided river communities support large numbers of nine species of wading birds, which breed seasonally (September January) on the open gravel of the riverbeds. Introduced mammalian predators, namely ferrets, feral cats Felis catus L., hedgehogs Erinaceus europaeus Barrett- Hamilton and stoats M. erminea L., prey on eggs and fledglings during the breeding season (Rebergen et al. 1998) and therefore pose a significant threat to the long-term survival of these birds. This effect is exacerbated by declines in rabbit numbers (Sinclair et al. 1998) through the hyperpredation effect (Courchamp, Langlais & Sugihara 2000). The most common methods used to control ferrets are trapping or poisoning in late winter or early spring. An important issue in the management of ferrets is how control (i.e. a reduction in density through removal of individuals) affects the survival of individuals remaining in the population, and the dispersal frequency of juveniles. For example, it is often assumed that dispersal frequency may increase with increasing density, and that mortality of dispersing individuals is high (Gaines & McClenaghan 1980). Empirical studies of a number of species have demonstrated density-dependent survival in response to harvest or predation (Staines 1978; White & Bartmann 1998; Boyce, Sinclair & White 1999). However, much of the theory of dispersal patterns in small mammals stems from studies of dispersal in herbivores (Byrom & Krebs 1999). Dispersal patterns of juvenile carnivores are less commonly examined (Waser 1996; Warrick, Scrivner & O Farrell 1999; Thomson et al. 2000), possibly because of their propensity to move long distances relative to their body size (Lavers & Clapperton 1990). In addition, the factors governing settlement of young animals (high or low conspecific density, good or poor habitat) are still relatively unclear (Barlow 1993). Whether population density affects settlement and survival of juvenile ferrets is still open to question. In this paper juvenile dispersal is defined as the movement of an individual away from its natal area upon reaching independence, to a new area where it may settle and attempt to reproduce upon reaching maturity (Howard 1960; Greenwood 1980). Four hypotheses were tested: (i) that dispersal in ferrets is malebiased, as predicted for most mammals by Greenwood (1980); (ii) that dispersal has a survival cost for juvenile ferrets, as is commonly assumed for mammals (Gaines & McClenaghan 1980; Waser & Jones 1983; Waser, Creel & Lucas 1994); (iii) that the proportion of juvenile ferrets dispersing is related to density at source; and (iv) that survival is inversely related to density at source (Gaines & McClenaghan 1980). A link between juvenile survival and population density has been observed previously in stoats (Powell & King 1997) and ferrets (Caley & Morriss 2001). Understanding dispersal, survival and settlement patterns of juvenile feral ferrets will contribute to developing better management strategies for this pest in New Zealand. It will also contribute to a better understanding of management of predators internationally, either as overabundant pests (e.g. foxes Vulpes vulpes in Australia; Newsome 1995) or as threatened species (e.g. blackfooted ferrets Mustela nigripes in the USA; Biggins et al. 1999). Methods STUDY SITES The dispersal patterns of juvenile ferrets were examined on six study sites on four braided rivers in the Mackenzie Basin, central South Island, New Zealand: the Tekapo, Pukaki, Ohau and Ahuriri rivers (Fig. 1). Study sites were located at least 1 5 km apart and five sites were > 300 ha (Table 1). Ecological communities in the grassland braided riverbed habitats of New Zealand s South Island comprise a mix of native and introduced species. Native vegetation on the beds of these rivers historically was sparse, but has increased since the introduction of exotic grasses and shrubs (e.g. brown top Agrostis capillaris L., sweet briar Rosa rubiginosa L. and willows Salix L. spp.). Riverbeds are bounded on both sides by modified tussock grasslands that are now grazed by sheep. The pasture riverbed boundaries are fenced to prevent stock from wandering onto the riverbed. Nine species of native birds use the braided rivers as breeding habitat, including the endangered black stilt Himantopus novaezealandiae Gould which is particularly at risk, with fewer than 100 individuals left in the wild (Brown & Keedwell 1998). Native species of reptile and invertebrate are found in the riverbed grassland ecosystems, and are potential

69 Dispersal of juvenile ferrets in New Zealand Fig. 1. Map of study sites on braided rivers in the Mackenzie Basin, New Zealand. R = predator removal areas. Table 1. Size, treatment, number of ferrets collared and number of ferrets that died on six study sites on braided rivers in the Mackenzie Basin, New Zealand. Numbers in parentheses are the number of juvenile ferrets collared on each site. R = predator removal; C = control; N = not used in that year Treatment No. of mortalities Site Size (ha) Year 1 Year 2 Year 1 Year 2 1 Tekapo 400 C (7) C (8) 4 3 2 Tekapo 330 C (2) C (2) 2 2 3 Tekapo Pukaki Delta 470 R (4) R (6) 0 0 4 Ohau 740 R (6) C (8) 1 1 5 Ahuriri 60 R (4) N 1 6 Tekapo 380 C (4) C (4) 1 1 secondary prey for predators. Access to the study sites is provided by a two-lane paved highway running approximately north south through the Mackenzie Basin (Fig. 1), and by single-lane four-wheel-drive tracks along both sides of all braided riverbeds. Rabbit numbers in the Mackenzie Basin were reduced by poisoning and shooting in the early 1990s from very high densities (spotlight counts of > 100 km 1 ) to low densities (indices < 10 km 1 ), which were maintained until 1997. In August 1997, just prior to the beginning of this study, rabbit haemorrhagic disease (RHD) was introduced illegally by landowners to kill rabbits on pasture land immediately adjacent to all study sites (Flux 1998). This resulted in, on average, a

70 A.E. Byrom 79% reduction in densities in the grasslands (Parkes, Norbury & Heyward 1999). Rabbit densities in the riverbeds were already very low prior to the release of RHD and generally were unaffected by the epidemics. However, because predators such as ferrets used both riverbeds and adjacent pastureland, the New Zealand Department of Conservation (DoC) began a predator removal programme of lethal trapping on sites 3 and 4 and immediately downstream of site 5 from October 1997 to mid-january 1998, to mitigate the potential impacts of predators on indigenous prey as rabbit densities declined. Predator removal areas had been identified by DoC as critical nesting habitat for the endangered black stilt, so it was not possible to establish a random experimental design for predator removal. Three sites (sites 1, 2 and 6) were left as replicated non-removals with normal post-rhd populations of rabbits and predators (Table 1). In the second year, lethal trapping of predators was confined to site 3 (with trapping completed by mid-january 1999). For logistic reasons, juvenile ferrets were not collared at site 5 in the second year, so four sites (sites 1, 2, 4 and 6) were left as non-removals with normal populations of predators and rabbits in the second year (Table 1). POPULATION CENSUSES Populations of ferrets on the braided riverbeds were estimated in early November, early December, early February and late March in both years. On each site, Grieves Cage Traps (30 30 80 cm; Grieves Wrought Iron and Wirework, Christchurch, New Zealand) were laid at 300-m intervals along transects on both sides of the riverbed. All sites were trapped to obtain comparable indices of ferret abundance despite concurrent predator removal on sites 3, 4 and 5. Traps were baited with fresh rabbit meat. Traps were opened on day 1, checked each morning, rebaited with fresh meat when necessary, then closed on day 5. All ferrets caught were tagged with individually numbered Monel metal tags (National Band and Tag Co., Newport, Kentucky, USA) in both ears, sexed, sexual condition recorded, and weighed to the nearest 5 g with a Pesola spring scale. For each site, the number of ferrets captured per 100 trap nights (TN) was calculated. Indices of rabbit abundance were taken monthly throughout spring and summer (September February) and bimonthly during autumn and winter (March August), by making spotlight counts on two consecutive nights along the same transects as the predator transects. Rabbits were counted by driving along the transect on a motorcycle at a speed of 10 20 km h 1, with a spotlight strapped to the observer s helmet. The total number of rabbits observed was divided by the number of kilometres driven to give an estimate of the number of rabbits per spotlight kilometre (Fletcher, Moller & Clapperton 1999). The same observer was used for all spotlight counts of rabbits. Rabbit spotlight counts are reasonably precise and reliable (Caley & Morely 2001). RADIO-COLLARING AND TRACKING Adult female ferrets were radio-collared in spring and early summer with 25-g mortality-sensing transmitters (Sirtrack Ltd, Havelock North, New Zealand), to measure survival during the breeding season and to determine locations of natal dens of breeding females. Female ferrets were trapped before, during and between the November and December population censuses. All adult females caught early in the season were radiocollared, but some collars were removed during the season if females failed to breed. Adult females were radio-tracked on foot once or twice per week during spring and early summer with a hand-held Yagi antenna (Telonics Corporation, Mesa, Arizona, USA). Individual fixes were taken using a Trimble GeoExplorer II global positioning system (GPS) from Trimble Navigation Ltd., Sunnyvale, California, USA. At that time differential correction of GPS data was necessary, and GPS data were differentially corrected using base stations in Christchurch (210 km away) or Dunedin (190 km away). Presence of young in the den of an adult female was determined by several methods: (i) listening for the sound of squeaking juveniles in the den; (ii) the female remaining in the same location for several days; (iii) a stethoscope placed just inside the den entrance to listen for juveniles; or (iv) use of a burrowscope (a modified endoscope with a strong light source at the end of a 1 5- m fibre-optic cable) to determine the presence of juveniles in the nest. Close to the time of juvenile emergence from the natal nest (approximately the first 2 weeks of January), the mother and juveniles were captured by digging up the nest. At least two juveniles (usually one male and one female) were radio-collared from a litter. Juveniles were radio-collared with 15-g mortalitysensing transmitters (Titley Ltd, Ballina, New South Wales, Australia). To increase the sample size of ferrets radio-collared on removal and non-removal areas, some ferrets were captured in traps rather than being collared in the nest. Although the exact birthplace of these individuals was not known, they were still used when measuring individual dispersal distances because they were trapped in January and February, before the period of peak dispersal. Emigration of juvenile ferrets from the birthplace occurred within a very short time period (early to mid-march in both seasons), so by tracking ferrets whose natal den locations were known, it was possible to determine that juvenile ferrets had not yet left their natal ranges. In total, 52 juvenile ferrets were radio-collared during this study. All juvenile ferrets weighed at least 400 g (range 400 950 g) when collared, so radio-transmitters were a maximum of 3 75% of initial body weight. Radiotracking of juvenile ferrets commenced as soon as radio-collars were fitted, well before the peak period of dispersal. Radio-collared ferrets were trapped or dug up approximately every 6 weeks to check the fit of the radio-collar around the neck of the animal, and adjusted for size if necessary.

71 Dispersal of juvenile ferrets in New Zealand Radio-tracking dispersing ferrets by following them on foot, and digging of dens for initial collar placement or subsequent resizing of collars, may affect their behaviour or survival. However, in this study, it was decided to track ferrets on foot because several highvoltage hydro-electric power lines are located in the Mackenzie Basin that previously have distorted radiotransmitter signals, producing significant error in radiofixes obtained by triangulation (Parker et al. 1996), and it was necessary to have accurate radio-fixes to measure dispersal. Most animals were radio-tracked twice per week, and only during the day when they were located in a den, so as to minimize disturbance during the dispersal period. In addition, dens were dug only occasionally. Radio-collared juveniles were generally not trap-shy during dispersal (some were repeatedly trapped during regular population censuses, which is common for ferrets), so it was assumed that observer disturbance had minimal effect on subsequent dispersal behaviour. Juvenile ferrets were radio-tracked on foot at least twice per week using a hand-held Yagi antenna, and their locations fixed using a GPS as for adults above. Survival, timing of dispersal, dispersal distances, movements away from the natal den, use of topographic features during dispersal, and settlement of juvenile ferrets were monitored for several months, until the end of their first winter (late August) in both years. Dispersal was recorded as having occurred when (and if ) a ferret left its mother s home range permanently. Confirmation that juveniles had settled in a new location was taken as the point when > 20 radio-fixes were obtained within an average home range area of an adult ferret (approximately 1 km 2 ; Norbury, Norbury & Heyward 1998) over a 4-month period post-dispersal (May August in each season). Fourteen females and 13 males were radio-collared in the first year and 15 females and 10 males in the second year, giving a total of 29 females and 23 males over the 2 years. The survival of radio-collared ferrets was compared in two ways. First, a semi-parametric model (Cox proportional hazards model; Venables & Ripley 1997) was fitted in order to test whether survival of juvenile ferrets (the baseline hazard function) was dependent on ferret density, rabbit density or year (the covariates). Non-independence of observations within a site was allowed for by including site as a cluster variable in the model. This analysis therefore enabled a comparison of survival rates of ferrets among individual sites. Secondly, survival of radio-collared juveniles was calculated using the Kaplan Meier procedure with a staggered entry design, which takes into account loss of individuals from, or entry of new individuals into, the sample population (Pollock et al. 1989). Survival data obtained from ferrets collared on the three removal areas were pooled, as were data from ferrets collared on non-removal areas, because sample sizes were too small for statistical analyses among individual sites (Table 1). Survivorship curves were compared using a log-rank test (Krebs 1989). Only four ferrets were lost during this study. Of these four, two were never heard after placement of the radio-transmitter, so the transmitters may have failed. The other two signals were lost after a few weeks of radio-tracking and had moved several kilometres during this time, so the animals may eventually have dispersed out of tracking range. All lost individuals were censored from the Kaplan Meier analyses at the point after which their radio-signal was no longer heard. Lost individuals were not recorded as deaths, and they were not used in calculations of dispersal distance. When comparing dispersal distances of juvenile ferrets, data were first log-transformed and tested for normality (Sokal & Rohlf 1995). Results POPULATION INDICES OF FERRETS During this study, ferret numbers (± SE) on all sites remained below 22 ± 3 captures per 100 TN (Fig. 2). The numbers of ferrets remained low on predator removal sites during the lethal trapping operation in both years, but increased after trapping ceased in early January in each year because of immigration by juvenile ferrets in late summer. CHARACTERISTICS OF JUVENILE DISPERSAL IN FERAL FERRETS A frequency distribution of dispersal distances of juvenile ferrets showed a negative-exponential shape (Fig. 3). Juvenile ferrets moved a median distance of 5 0 km from their natal site (range 0 5 45 0 km; Fig. 3). In both seasons, survivorship (95% confidence limits, n) of ferrets that dispersed was not significantly different from ferrets that did not [year 1: dispersers 19% (0 42%, n = 13), non-dispersers 34% (0 73%, n = 8), log-rank test, χ 2 = 0 45, d.f. = 1, P = 0 50; year 2: dispersers 67% (33 100%, n = 16), non-dispersers 62% (29 96%, n = 7), χ 2 = 0 03, d.f. = 1, P = 0 50], although the statistical power to detect differences between dispersers and non-dispersers was probably low. After emergence from the natal den in early January, juvenile ferrets that dispersed spent an average (± SE) of 57 4 ± 3 5 days (range 28 84 days) within the natal home range before dispersing, so peak dispersal occurred in approximately late February and early March. This was verified by trapping records (G. Norbury, unpublished data). Regular radio-fixes on each individual showed that, during dispersal, ferrets were often found denning along linear topographic features and habitat edges such as under cover of low-growing shrubs or in overgrown stream channels and river banks, and it was assumed they used these linear habitat features during dispersal. Nine of 16 (56 ± 13%) juvenile ferrets radiocollared on predator removal areas dispersed, whereas 21 of 29 (72 ± 8%) ferrets from non-removal areas dispersed. This difference was not significant (r 2 = 0 06, d.f. = 9, P = 0 47), although statistical power to detect differences between treatments was probably low.

72 A.E. Byrom Fig. 2. Ferrets captured per 100 trap nights (TN) on riverbeds from 1997 to 1999. R = predator removal areas. Error bars are standard errors. Dashed lines represent cessation of predator removal in each season. Fig. 3. Frequency distribution of dispersal distances of 34 juvenile ferrets. Not all ferrets collared originally could be used for this plot because they were road-killed, died or lost before they settled in a new location.

73 Dispersal of juvenile ferrets in New Zealand Fig. 4. (a) Year 1: survival of 14 radio-collared juvenile ferrets on three predator removal sites (after removal had ceased) and 13 ferrets on three control sites after emergence from the natal den in January 1998, through to the end of their first winter in August 1998. (b) Year 2: survival of six radio-collared juvenile ferrets on one predator removal site (after removal had ceased) and 22 ferrets on control sites through to the end of their first winter in August 1999. Ninety-five per cent confidence limits are reported in the text. Fig. 5. Survival of juvenile ferrets (with 95% confidence limits) on predator removal and control sites, as a function of ferret abundance in summer (captures per 100 trap nights; TN) at each site. Survival was estimated using the Kaplan Meier procedure. SURVIVAL OF JUVENILE FERRETS Juvenile ferrets collared on non-removal sites had significantly poorer survival in both years than ferrets from the predator removal sites (Fig. 4), thus ferret density at source had a strong effect on survival of juvenile ferrets (approximate Z = 2 95, P < 0 03; Fig. 5). There was no effect of rabbit density on ferret survival (Z = 0 43, P = 0 67). Even allowing for ferret density, there was some evidence for a year effect on survival of

74 A.E. Byrom Table 2. Suspected causes of natural mortality of radio-collared juvenile ferrets on braided riverbeds in the Mackenzie Basin, New Zealand. Ferrets were classified as diseased if they were staggering, dazed, sneezing or coughing, or acting strangely. Ferrets were classified as starving if repeated telemetry fixes found them foraging diurnally (often on stream invertebrates), obviously emaciated, with patchy, dull fur, and gnawing on very old carcasses Mortality cause Year 1 Year 2 Total Starvation 5 2 7 Disease 3 2 5 Cannibalism 1 0 1 Unknown 2 1 3 Total mortalities 11 5 16 juvenile ferrets (Z = 1 86, P = 0 06), which reflected the much higher survival of ferrets on all sites in the second year. Kaplan Meier analyses showed similar results (Fig. 4). Nineteen per cent (0 38%) of ferrets collared on the non-removal sites in year 1 (n = 13) survived their first 4 months, and survival declined to zero during winter 1998. By contrast, 86% (63 100%) of juvenile ferrets collared on the removal sites (n = 14) survived their first 4 months, and 42% (2 81%) survived their first winter. In the second year, survival was much higher on both non-removal and removal sites: 71% (45 97%) of ferrets collared on non-removal sites (n = 22) survived the first 4 months, and by the end of winter 1999 survival declined to 60% (28 90%). On the removal site, 100% survival was recorded (n = 6) to the end of the ferrets first winter. In both years, survival of juvenile ferrets on removal sites was significantly higher than survival on non-removal sites (log-rank test, year 1: χ 2 = 4 41, d.f. = 1, P = 0 036; year 2: χ 2 = 6 57, d.f. = 1, P = 0 010). Most deaths of juvenile ferrets (12 of 13 deaths in which the mortality agent could be determined) were from disease or starvation, and not from predation (Table 2). Three diseased ferrets and one starved ferret were eaten by Australasian harriers Circus approximans Peale, although it could not be determined whether they were scavenged or killed. SEX-SPECIFIC DISPERSAL AND SURVIVAL There were no sex-specific differences in survival and dispersal patterns of juvenile ferrets in this study. Males and females did not differ in the amount of time spent on the natal range prior to dispersal (females: 54 4 ± 3 7 days; males: 60 6 ± 6 1 days; Student s t-test: t = 0 88, d.f. = 22, P = 0 39) and both sexes moved similar distances away from the natal site in each year (females: 11 8 ± 3 4 km; males: 6 7 ± 1 6 km; Student s t-test: t = 0 33, d.f. = 32, P = 0 75; Table 3). Survival of male ferrets [year 1: 25% (0 71%); year 2: 86% (59 100%)] and female ferrets (year 1: 35% (10 65%); year 2: 70% (40 100%)] also did not differ significantly in either year (log-rank test, year 1: χ 2 = 0 46, d.f. = 1, P = 0 499; year 2: χ 2 = 0 002, d.f. = 1, P = 1 00). Discussion This study is the first to quantify the responses of juvenile ferrets to an experimental reduction in density by removal of ferrets during large-scale predator control in New Zealand. Three of the four hypotheses tested were rejected. First, there was no support for the hypothesis of a sex bias in dispersal of juvenile ferrets. Both sexes of ferrets dispersed (not just males). Secondly, there was no evidence of a mortality cost to dispersal in juvenile ferrets. In both years, mortality patterns were more likely to be related to social factors and food availability than to dispersal per se. These results are consistent with observations of dispersal in other solitary carnivores (Storm et al. 1976; Bekoff, Daniels & Gittleman 1984; Waser 1996; Koopman, Cypher & Scrivner 2000) and some herbivores (Efford 1998; White & Bartmann 1998). Dispersal appears more likely to be male-biased in polygynous herbivore species (Holekamp 1984; Wiggett & Boag 1989; Byrom & Krebs 1999) or more social carnivore species (Rood 1987; Keane, Creel & Waser 1996; McNutt 1996; Boyd & Pletscher 1999). Thirdly, there was no support for the hypothesis that dispersal frequency (i.e. proportion dispersing) increased with increasing population density of ferrets. Evidence for a relationship between dispersal and population density is equally diverse in other herbivores (Cowan et al. 1996, 1997; Byrom & Krebs 1999) and carnivores (Storm et al. 1976; Allen & Sargeant 1993; Arthur, Paragi & Krohn 1993; Beier 1995; Boyd & Pletscher 1999; Hayes & Harestad 2000; Koopman, Cypher & Scrivner 2000). It was not possible to reject the hypothesis that survival of juvenile ferrets was inversely related to popu- Table 3. Mean distances moved from the natal site, and (for ferrets classified as dispersers) mean timing of dispersal for radiocollared juvenile ferrets in two consecutive years. Numbers in parentheses are sample sizes. Not all ferrets collared originally could be used in these analyses because some ferrets did not disperse, or were road-killed, died or lost before they settled in a new location Distance moved from natal site (km ± SE) Timing of dispersal (days after emergence from natal den ± SE) Males Females Males Females Year 1 5 6 ± 1 1 (8) 3 9 ± 0 8 (6) 64 9 ± 8 9 (8) 64 0 ± 9 4 (5) Year 2 7 8 ± 3 1 (8) 15 8 ± 4 7 (12) 55 0 ± 8 5 (6) 49 6 ± 2 1 (10) Both years 6 7 ± 1 6 (16) 11 8 ± 3 4 (18) 60 6 ± 6 1 (14) 54 4 ± 3 7 (15)

75 Dispersal of juvenile ferrets in New Zealand lation density. In both years, juvenile ferrets showed density-dependent survival in response to removal of predators (including ferrets). In year 1 that response was particularly pronounced, and no juvenile ferrets survived their first winter on non-removal sites. On those sites, adult ferrets and possibly other predators such as feral cats may have excluded younger ferrets from settling and establishing territories effectively. In year 2, survival of juvenile ferrets was much higher on both non-removal and removal sites, but was still dependent on population density. Other researchers have reported a correlation between population density and survival in New Zealand mustelids (stoats Mustela erminea L., Powell & King 1997; ferrets, Caley and Morriss 2001) but few studies have attempted to radiotrack small-bodied (< 5 kg) carnivores during dispersal (Bekoff, Daniels & Gittleman 1984), so the results of this study augment a relatively sparse literature on dispersal and survival of mammalian carnivores. Density-dependent population responses in pest populations after control are becoming a serious problem world-wide (Van Vuren & Smallwood 1996; Leirs et al. 1997; Krebs 1999). Control operations like the predator removal in this study often reduce populations to below carrying capacity (Caughley, Pech & Grice 1992; Sinclair 1997). Yet control operations will probably only work if the reduction in density exceeds any potential density-dependent response in the population (Sinclair & Pech 1996; Barlow, Kean & Briggs 1997; Sinclair 1997). Barlow & Norbury (2001) modelled the effects of removal of adults on ferret populations in New Zealand and found that removal of at least 50% of the population each year would be needed to effect a suppression of 50% in long-term average ferret density, because of strong density-dependent responses in the population, estimated from annual changes in ferret abundance indices. The present study provides empirical support and a mechanistic basis for that model and is the first to report a density-dependent survival response in juvenile ferrets in New Zealand following removal trapping. The consequences of this result for conservation and for the spread of TB should not be underestimated. Predation on native wildlife, including ground-nesting birds, invertebrates and reptiles, is an important problem when attempting to conserve these species. Introduced predators can thwart this conservation goal by having severe impacts on native wildlife through the hyperpredation effect (Courchamp, Langlais & Sugihara 2000). A frequency distribution of dispersal distances of juvenile ferrets showed a typical negative-exponential shape (Fig. 3), and dispersal of juvenile ferrets was not related to ferret population density at the natal site. Effective ferret control may therefore be compromised by rapid immigration of juvenile ferrets (sometimes termed a vacuum effect ; Efford, Warburton & Spencer 2000). The median dispersal distance of juvenile ferrets in this study was 5 km, so buffers designed to prevent reinvasion of an area should be at least 5 km wide to reduce immigration by at least 50% of dispersing juvenile ferrets. Wider buffer widths would be required if the goal was, for example, to reduce reinvasion by 90% of dispersing juveniles. This is in contrast to possum populations in New Zealand (which are also responsible for the spread of TB); possums are unlikely to regain pre-control numbers through immigration after control (Cowan et al. 1996, 1997; Efford 1998). Predator removal was probably useful in the short term to reduce predation on ground-nesting bird nests during the breeding season. Ferret control may also be useful to prevent the spread of bovine TB by infected juvenile ferrets. The data in Fig. 3 therefore provide the necessary information for a spatial model (Barlow 1993) of ferret dynamics and management. It is difficult to quantify the effect that the introduction of RHD in August 1997 had on dispersal and survival of juvenile ferrets (through changes in rabbit numbers) in the Mackenzie Basin, and the consequent effects on secondary prey. Small changes in primary prey availability (in this case rabbits) may have important ecological consequences for secondary prey (indigenous wildlife) in the community (Holt 1977; Abrams 1987; Courchamp, Langlais & Sugihara 1999, 2000; Norbury 1999) because even a small number of ferrets may do a disproportionate amount of damage to indigenous wildlife if primary prey availability is low (Barlow & Kean 1996). From this study it was not possible to determine whether widespread declines in rabbit densities on adjacent pastureland (but not on riverbeds) had any effect on dispersal patterns of juvenile ferrets. Results from this research were obtained in part because of a management regime imposed quite suddenly in response to the introduction of RHD in August 1997, with its assumed effects on secondary prey as rabbit densities declined. Therefore it was not possible to choose which study sites were kill-trapped, which resulted in three non-treatment sites being located on the Tekapo riverbed in year 1 (Fig. 1). Ideally, treatments would have been assigned randomly to avoid site-specific anomalies. Two lines of evidence suggest that the results obtained were not site-specific anomalies. First, the study sites were very large (the smallest site on the Tekapo riverbed was 330 ha) so observed patterns of mortality were unlikely to be small localized effects. Secondly, although survival of juvenile ferrets was low on all sites in year 1, survival of radiocollared adult females was very high during the same period (100% survival of eight radio-collared adult females on the Tekapo sites was recorded from January to August 1998, and 100% survival of six radio-collared adult females on those sites from January to August 1999). These results suggest two possible improvements in predator management in order to restrict dispersal for management of disease and for conservation. First, trapping of ferrets for control purposes may prove to be more effective if it is done in late autumn, after

76 A.E. Byrom dispersal and settlement of juveniles, and not before dispersal, so as to minimize a vacuum effect (reinvasion of an area by juvenile ferrets). This would reduce ferret density in critical areas for at least one season, and would possibly also reduce the need for ongoing annual control of ferrets. Depending on the type of control used (e.g. cage traps or leg-hold traps), other predators (such as stoats and cats) could be targeted concurrently. Efficacy of seasonal timing of ferret control (late winter spring compared with autumn control operations) could be tested by measuring rates of population recovery after experimentally reducing ferret populations. Secondly, it may be necessary to create wide buffer zones around key wildlife habitat or areas with a high prevalence of TB. Buffer zones of 5 km (the median dispersal distance of juvenile ferrets in this study) would reduce potential immigrants by about 50%. The use of wide buffers may increase benefits to native wildlife, and possibly would also minimize the spread of bovine TB, as suggested by Barlow & Norbury (2001). Buffers have been used to exclude problem predators in Australia (e.g. foxes, Banks, Dickman & Newsome 1998, and dingoes Canis lupus dingo, Allen & Sparkes 2001). Wider buffer widths would be required if it was deemed necessary to exclude more than 50% of immigrant ferret dispersers. Use of such measures may help to mitigate the effects of problem predators world-wide (Simberloff et al. 2000), including potential compensatory effects (Barlow 2000; Courchamp & Cornell 2000). Acknowledgements I thank Andy Kliskey for helping with intensive field work trapping, radio-tracking and taking GPS locations for ferrets, as well as for extensive searches for mobile ferrets. I thank Chelsea Brown, Richard Heyward, Dale Norbury, Grant Norbury, Caroline Thomson and Ivor Yockney for trapping and radio-collaring ferrets and adjusting collars. Thanks also to Phil Stenton for many cold nights counting rabbits, and the staff at the Department of Conservation in Twizel, particularly Kerry Brown, Richard Maloney and Mark Sanders, for support and encouragement throughout this study. Phil Cowan, Murray Efford, Chris Frampton, Henrik Moller, Grant Norbury, John Parkes, Oliver Sutherland, Bruce Warburton and one anonymous referee read and commented on earlier drafts of the manuscript. Ray Webster provided invaluable statistical advice. This research was funded through a New Zealand Foundation for Research, Science and Technology Postdoctoral Fellowship (No. C09702) and a New Zealand Lotteries Science Grant (No. SR 86061). Predator and rabbit population assessments were carried out for the project Ecological impacts of removal of rabbits from wildlands/ conservation estate (funded by the Foundation for Research, Science and Techno-logy; contract nos C09640 and C09801). Field work was approved by the Animal Ethics Committee of Landcare Research. References Abrams, P. (1987) Indirect interactions between prey species. Predation: Direct and Indirect Impacts on Aquatic Communities (eds W.C. Kerfoot & A. Sih), pp. 38 54. University Press of New England, Hanover, New England. Allen, L.R. & Sparkes, E.C. (2001) The effect of dingo control on sheep and beef cattle in Queensland. Ecology, 38, 76 87. Allen, S.H. & Sargeant, A.B. (1993) Dispersal patterns of red foxes relative to population density. Journal of Wildlife Management, 57, 526 533. Alterio, N., Moller, H. & Ratz, H. (1998) Movements and habitat use of feral house cats Felis catus, stoats Mustela erminea and ferrets Mustela furo, in grassland surrounding yellow-eyed penguin Megadyptes antipodes breeding areas in spring. Biological Conservation, 83, 187 194. Arthur, S.M., Paragi, T.F. & Krohn, W.B. (1993) Dispersal of juvenile fishers in Maine. Journal of Wildlife Management, 57, 868 874. Banks, P.B., Dickman, C.R. & Newsome, A.E. (1998) Ecological costs of feral predator control: foxes and rabbits. Journal of Wildlife Management, 62, 766 772. Barlow, N.D. (1993) A model for the spread of bovine Tb in New Zealand possum populations. Ecology, 30, 156 164. Barlow, N.D. (2000) The ecological challenge of immunocontraception: editor s introduction. Ecology, 37, 897 902. Barlow, N.D. & Kean, J.M. (1996) Modelling the interaction between ferrets and rabbits. Ferrets as Vectors of Tuberculosis and Threats to Conservation, pp. 38 45. Miscellaneous Series No. 36. Royal Society of New Zealand, Wellington, New Zealand. Barlow, N.D. & Norbury, G.L. (2001) A simple model for ferret population dynamics and control in semi-arid New Zealand habitats. Wildlife Research, 28, 87 94. Barlow, N.D., Kean, J.M. & Briggs, C.J. (1997) Modelling the relative efficacy of culling and sterilisation for controlling populations. Wildlife Research, 24, 129 141. Beier, P. (1995) Dispersal of juvenile cougars in fragmented habitat. Journal of Wildlife Management, 59, 228 237. Bekoff, M., Daniels, T.J. & Gittleman, J.L. (1984) Life history patterns and the comparative social ecology of carnivores. Annual Review of Ecology and Systematics, 15, 191 232. Biggins, D.E., Vargas, A., Godbey, J.L. & Anderson, S.H. (1999) Influence of pre-release experience on reintroduced blackfooted ferrets (Mustela nigripes). Biological Conservation, 89, 121 129. Blandford, P. (1987) Biology of the polecat Mustela putorius: a literature review. Mammal Review, 17, 155 198. Boyce, M.S., Sinclair, A.R.E. & White, G.C. (1999) Seasonal compensation of predation and harvesting. Oikos, 87, 419 426. Boyd, D.K. & Pletscher, D.H. (1999) Characteristics of dispersal in a colonising wolf population in the central rocky mountains. Journal of Wildlife Management, 63, 1094 1108. Brown, K. & Keedwell, R. (1998) Predator Control to Mitigate the Effects of RCD-induced Prey-switching in the Mackenzie Basin. Conservancy Technical Report Series No. 8. Department of Conservation, Canterbury, New Zealand. Byrom, A.E. & Krebs, C.J. (1999) Natal dispersal of juvenile arctic ground squirrels in the boreal forest. Canadian Journal of Zoology, 77, 1048 1059. Caley, P. & Morely, C. (2001) How bad are spotlight transect counts for assessing changes in abundance of European rabbits (Oryctolagus cuniculus)? Proceedings of the Australasian Vertebrate Pest Conference, 12, 373 377. Caley, P. & Morriss, G. (2001) Summer/autumn movements, mortality rates and density of feral ferrets (Mustela furo) at

77 Dispersal of juvenile ferrets in New Zealand a farmland site in North Canterbury, New Zealand. New Zealand Journal of Ecology, 25, 53 60. Caughley, G., Pech, R. & Grice, D. (1992) Effect of fertility control on a population s productivity. Wildlife Research, 19, 623 627. Courchamp, F. & Cornell, S.J. (2000) Virus-vectored immunocontraception to control feral cats on islands: a mathematical model. Ecology, 37, 903 913. Courchamp, F., Langlais, M. & Sugihara, G. (1999) Cats protecting birds: modelling the mesopredator release effect. Journal of Animal Ecology, 68, 282 292. Courchamp, F., Langlais, M. & Sugihara, G. (2000) Rabbits killing birds: modelling the hyperpredation process. Journal of Animal Ecology, 69, 154 164. Cowan, P.E., Brockie, R.E., Smith, R.N. & Hearfield, M.E. (1997) Dispersal of juvenile brushtail possums, Trichosurus vulpecula, after a control operation. Wildlife Research, 24, 279 288. Cowan, P.E., Brockie, R.E., Ward, G.D. & Efford, M.G. (1996) Long-distance movements of juvenile brushtail possums (Trichosurus vulpecula) on farmland, Hawkes Bay, New Zealand. Wildlife Research, 23, 237 244. Efford, M. (1998) Demographic consequences of sex-biased dispersal in a population of brushtail possums. Journal of Animal Ecology, 67, 503 517. Efford, M., Warburton, B. & Spencer, N. (2000) Home-range changes by brushtail possums in response to control. Wildlife Research, 27, 117 127. Fletcher, D., Moller, H. & Clapperton, K. (1999) Spotlight counts for assessing abundance of rabbits (Oryctolagus cuniculus L.). Wildlife Research, 26, 609 620. Flux, J.E.C. (1998) Status, ecology and control by predation of rabbits and hares in New Zealand. Rabbit Control, RCD: Dilemmas and Implications (ed. B.D.W. Jarvis), pp. 4 9. Miscellaneous Series No. 55. Royal Society of New Zealand, Wellington, New Zealand. Gaines, M.S. & McClenaghan, L.R. Jr (1980) Dispersal in small mammals. Annual Review of Ecology and Systematics, 11, 163 196. Greenwood, P.J. (1980) Mating systems, philopatry, and dispersal in birds and mammals. Animal Behaviour, 28, 1140 1162. Hayes, R.D. & Harestad, A.S. (2000) Demography of a recovering wolf population in the Yukon. Canadian Journal of Zoology, 78, 36 48. Holekamp, K.E. (1984) Dispersal in ground-dwelling sciurids. The Biology of Ground-Dwelling Squirrels (eds J.O. Murie & G.R. Michener), pp. 297 320. University of Nebraska Press, Lincoln, NB. Holt, R.D. (1977) Predation, apparent competition, and the structure of prey communities. Theoretical Population Biology, 12, 197 229. Howard, W.E. (1960) Innate and environmental dispersal of individual vertebrates. American Midland Naturalist, 63, 152 161. Keane, B., Creel, S.R. & Waser, P.M. (1996) No evidence of inbreeding avoidance or inbreeding depression in a social carnivore. Behavioral Ecology, 7, 480 489. Koopman, M.E., Cypher, B.L. & Scrivner, J.H. (2000) Dispersal patterns of San Joaquin kit foxes (Vulpes macrotis mutica). Journal of Mammalogy, 81, 213 222. Krebs, C.J. (1989) Ecological Methodology. Harper Collins, New York, NY. Krebs, C.J. (1999) Current paradigms of rodent population dynamics what are we missing? Ecologically-Based Rodent Management (eds G. Singleton, L. Hinds, H. Leirs & Z. Zhang), pp. 33 48. Australian Centre for International Agricultural Research, Canberra, Australia. Lavers, R.B. & Clapperton, B.K. (1990) Ferret. The Handbook of New Zealand Mammals (ed. C.M. King), pp. 320 330. Oxford University Press, Auckland, New Zealand. Leirs, H., Stenseth, N.Chr., Nichols, J.D., Hines, J.E., Verhagen, R. & Verheyen, W. (1997) Stochastic seasonality and nonlinear density-dependent factors regulate population size in an African rodent. Nature, 189, 176 180. Lugton, I.W., Wobeser, G., Morris, R.S. & Caley, P. (1997) Epidemiology of Mycobacterium bovis infection in feral ferrets (Mustela furo) in New Zealand. I. Pathology and diagnosis. New Zealand Veterinary Journal, 45, 140 150. McNutt, J.W. (1996) Sex-biased dispersal in African wild dogs, Lycaon pictus. Animal Behaviour, 52, 1067 1077. Newsome, A.E. (1995) Socio-ecological models for red fox populations subject to fertility control in Australia. Annales Zoologici Fennici, 32, 99 110. Norbury, G. (1999) Ecological consequences of rabbit haemorrhagic disease. Rabbit Control, RCD: Dilemmas and Implications, 10 14. Miscellaneous Series No. 55. Royal Society of New Zealand, Wellington, New Zealand. Norbury, G., Norbury, D. & Heyward, R. (1998) Behavioral responses of two predator species to sudden declines in primary prey. Journal of Wildlife Management, 62, 45 58. Parker, N., Pascoe, A., Moller, H. & Maloney, R. (1996) Inaccuracy of a radio-tracking system for small mammals: the effect of electromagnetic interference. Journal of Zoology (London), 239, 401 406. Parkes, J.P., Norbury, G.L. & Heyward, R.P. (1999) Has rabbit haemorrhagic disease worked in New Zealand? Proceedings of the New Zealand Society of Animal Production, 59, 68 70. Pierce, R.J. (1986) Differences in susceptibility to predation during nesting between pied and black stilts (Himantopus spp.). Auk, 103, 273 280. Pollock, K.H., Winterstein, S.R., Bunck, C.M. & Curtis, P.D. (1989) Survival analysis in telemetry studies: the staggered entry design. Journal of Wildlife Management, 53, 7 15. Powell, R.A. & King, C.M. (1997) Variation in body size, sexual dimorphism and age-specific survival in stoats, Mustela erminea (Mammalia: Carnivora), with fluctuating food supplies. Biological Journal of the Linnean Society, 62, 165 194. Ragg, J.R., Moller, H. & Waldrup, K.A. (1995) The prevalence of bovine tuberculosis (Mycobacterium bovis) infections in feral populations of cats (Felis catus), ferrets (Mustela furo) and stoats (Mustela erminea) in Otago and Southland, New Zealand. New Zealand Veterinary Journal, 43, 333 337. Ragg, J.R., Waldrup, K.A. & Moller, H. (1995) The distribution of gross lesions of tuberculosis caused by Mycobacterium bovis in feral ferrets (Mustela furo) from Otago, New Zealand. New Zealand Veterinary Journal, 43, 338 341. Rebergen, A., Keedwell, R., Moller, H. & Maloney, R. (1998) Breeding success and predation at nests of banded dotterel (Charadrius bicinctus) on braided riverbeds in the central South Island, New Zealand. New Zealand Journal of Ecology, 22, 33 41. Rood, J.P. (1987) Dispersal and intergroup transfer in the dwarf mongoose. Mammalian Dispersal Patterns: The Effects of Social Structure on Population Genetics (eds B.D. Chepko- Sade & Z.T. Halpin), pp. 85 103. University of Chicago Press, London, UK. Simberloff, D., Dayan, T., Jones, C. & Ogura, G. (2000) Character displacement and release in the small Indian mongoose, Herpestes javanicus. Ecology, 81, 2086 2099. Sinclair, A.R.E. (1997) Fertility control of mammal pests and the conservation of endangered marsupials. Reproduction, Fertility and Development, 9, 1 16. Sinclair, A.R.E. & Pech, R.P. (1996) Density dependence, stochasticity, compensation and predator regulation. Oikos, 75, 164 173. Sinclair, A.R.E., Pech, R.P., Dickman, C.R., Hik, D., Mahon, P. & Newsome, A.E. (1998) Predicting effects of predation on conservation of endangered prey. Conservation Biology, 12, 564 575.

78 A.E. Byrom Sokal, R.R. & Rohlf, F.J. (1995) Biometry, 3rd edn. W. H. Freeman, New York, NY. Staines, B.W. (1978) The dynamics and performance of a declining population of red deer (Cervus elaphus). Journal of Zoology, 184, 403 419. Storm, G.L., Andrews, R.D., Phillips, R.L., Bishop, R.A., Siniff, D.B. & Tester, J.R. (1976) Morphology, reproduction, dispersal and mortality of midwestern red fox populations. Wildlife Monographs, 49. Thomson, P.C., Marlow, N.J., Rose, K. & Kok, N.E. (2000) The effectiveness of a large-scale baiting campaign and evaluation of a buffer zone strategy for fox control. Wildlife Research, 27, 465 472. Van Vuren, D. & Smallwood, K.S. (1996) Ecological management of vertebrate pests in agricultural systems. Biological Agriculture and Horticulture, 13, 39 62. Venables, W.N. & Ripley, B.D. (1997) Modern Applied Statistics with S-Plus, 2nd edn. Springer-Verlag, New York, NY. Warrick, G.D., Scrivner, J.H. & O Farrell, T.P. (1999) Demographic responses of kit foxes to supplemental feeding. Southwestern Naturalist, 44, 367 374. Waser, P. (1996) Patterns and consequences of dispersal in gregarious carnivores. Carnivore Behaviour, Ecology and Evolution, Vol. 2. (ed. J.L. Gittleman), pp. 267 295. Cornell University Press, Ithaca, NY. Waser, P.M. & Jones, W.T. (1983) Natal philopatry among solitary mammals. Quarterly Review of Biology, 58, 355 390. Waser, P.M., Creel, S.R. & Lucas, J.R. (1994) Death and disappearance: estimating mortality risks associated with philopatry and dispersal. Behavioral Ecology, 5, 135 141. White, G.C. & Bartmann, R.M. (1998) Effect of density reduction on overwinter survival of free-ranging mule deer fawns. Journal of Wildlife Management, 62, 214 225. Wiggett, D.R. & Boag, D.A. (1989) Intercolony natal dispersal in the Columbian ground squirrel. Canadian Journal of Zoology, 67, 42 50. Received 3 July 2000; final copy received 25 September 2001