Biological Conservation 160 (2013) 114–120

Contents lists available at SciVerse ScienceDirect

Biological Conservation journal homepage: www.elsevier.com/locate/biocon

Disentangling the importance of interspecific competition, food availability, and habitat in species occupancy: Recolonization of the endangered Fennoscandian arctic fox S. Hamel ⇑, S.T. Killengreen, J.-A. Henden, N.G. Yoccoz, R.A. Ims Department of Arctic and Marine Biology, Faculty of Biosciences, Fisheries and Economics, University of Tromsø, 9037 Tromsø, Norway

a r t i c l e

i n f o

Article history: Received 29 August 2012 Received in revised form 31 December 2012 Accepted 7 January 2013

Keywords: Conservation Endangered species Lemming Lemmus lemmus Norway Occupancy modeling Red fox Vulpes vulpes

a b s t r a c t Ecosystems alterations can profoundly affect species occurrence and distribution, thereby influencing trophic and/or competitive interactions. Arctic foxes have been fully protected in Fennoscandia following their drastic decline in the early 19th century, and their lack of recovery has been hypothesized to result from increased interspecific competition with red foxes that have colonized the arctic fox’s range and/or changes in prey dynamics due to increased variation in climatic conditions. We used a large-scale spatiotemporal study combined with an extensive red fox removal intervention to test these two hypotheses by evaluating the relative importance of diverse environmental factors affecting the recolonization of arctic foxes in Northern Norway. Arctic foxes were never observed at sites where just a few daily observations of red foxes were recorded, and the probability of recolonization was zero at sites where no red fox was removed. The probability of recolonization increased when lemming abundance was very high, but the relative importance of this variable was half that of red foxes. Thus, our results lend support to both hypotheses, but they clearly highlight the presence of red foxes as the factor most limiting arctic fox recolonization in this part of its distribution. As the abundance of arctic foxes has remained extremely low even after more than 70 years of full protection, direct conservation actions seem to be needed to promote the recovery of arctic fox populations. Our research therefore highlights that conservation actions targeting a major reduction in red fox numbers are needed because the competitive interaction with red foxes is the most important factor limiting recolonization of this species. ! 2013 Elsevier Ltd. All rights reserved.

1. Introduction Species occurrence and distribution can be profoundly affected by changes in the ecosystem, which can result from natural transitions as well as anthropogenic alterations caused by variation in human use of habitats/animals (White and Jentsch, 2001). Although these ecosystem modifications can allow certain species to expand their range and increase in numbers, more sensitive species might be forced to reduce their range as a result of habitat becoming unsuitable or direct competition with new invasive species, sometimes even putting them at risk of extinction (Byers, 2002; Didham et al., 2005; Kallimanis et al., 2005). For example, the successful introduction of feral pigs (Sus scrofa) on Santa Cruz Island provided access to abundant prey for golden eagles (Aquila chrysaetos) nesting on the mainland (Roemer et al., 2002). As a result, the eagles colonized the island and became the apex predator, which indirectly caused a rapid decline in the island fox (Urocyon littoralis) population, and thereby altered the competitive relation⇑ Corresponding author. Tel.: +47 77623169.

E-mail address: [email protected] (S. Hamel).

0006-3207/$ - see front matter ! 2013 Elsevier Ltd. All rights reserved. http://dx.doi.org/10.1016/j.biocon.2013.01.011

ship between foxes and island spotted skunks (Spilogale gracilis amphiala), two endemic predators (Roemer et al., 2002). Similarly to invasion by exotic species, another classic and worldwide example of ecosystem alteration is associated with habitat loss caused by human increase in land use (Vitousek, 1994). For instance, deforestation occurring over most of the range of woodland caribou (Rangifer tarandus caribou) has produced early stage forests that are unsuitable for this species but highly beneficial to other ungulates like moose (Alces alces) and white-tailed deer (Odocoileus virginianus) (Bergerud and Ballard, 1988; Cumming, 1992; Schaefer, 2003). This has resulted in an increase in wolf abundance, and thereby predation on caribou (Bergerud, 1974; Cumming, 1992; Seip, 1991), and now the recovery of this species is unlikely without drastic management actions (Vors et al., 2007; Wittmer et al., 2007). These examples illustrate two of many situations where significant management actions are required to help native species recolonize a habitat. Nonetheless, the mechanisms influencing the capacity of a species to recolonize abandoned sites are often poorly known, and identifying the factors affecting the recolonization of a species would allow implementing more targeted management or conservation actions.

S. Hamel et al. / Biological Conservation 160 (2013) 114–120

The arctic fox (Vulpes lagopus) has a northern circumpolar distribution and is generally considered relatively abundant where it occurs (Audet et al., 2002). At the southern margins of the arctic tundra, however, there is evidence that this species has declined during the last century (Herteinsson and Macdonald, 1992). In Fennoscandinavia, the decline of the arctic fox appears to be particularly acute; the population was considered to be near extinction at the beginning of the 19th century and has remained at very low numbers, despite being fully protected since the 1940s (Angerbjörn et al., 1995; Hersteinsson et al., 1989). Several explanations for the lack of recovery of arctic foxes in this part of its distribution range have been proposed. They can be summarized under two main hypotheses: increased interspecific competition with red foxes (Vulpes vulpes) and changes in prey dynamics (Angerbjörn et al., 1991, 1995; Frafjord et al., 1989; Henden et al., 2010; Hersteinsson et al., 1989; Herteinsson and Macdonald, 1992; Kaikusalo and Angerbjörn, 1995; Tannerfeldt et al., 2002). According to the asymmetric size competition hypothesis, larger species dominate over smaller species in contests (Fedriani, 2000; Macdonald and Sillero-Zubiri, 2004; Nelson, 2007; Palomares and Caro, 1999; Tannerfeldt et al., 2002). Evidence supporting this hypothesis has largely been reported in canids (Fedriani, 2000; Levi and Wilmers, 2012; Nelson, 2007; Palomares and Caro, 1999), so we can expect the larger red fox (!5.5 kg) to be a dominant competitor to the arctic fox (!3.5 kg), possibly monopolizing patches or areas rich in prey or subsidies (Henden et al., 2010; Herteinsson and Macdonald, 1992; Killengreen et al., 2007; Shirley et al., 2009; Tannerfeldt et al., 2002). Furthermore, northward and altitudinal movements of red foxes into arctic foxes range have been documented since 1960 and are likely to increase as the arctic tundra continues to warm up (Herteinsson and Macdonald, 1992; Post et al., 2009). At the same time, the decline in red fox hunting towards the end of the 19th century likely released the harvest pressure that maintained red fox abundance at lower levels in the past (Hersteinsson et al., 1989). In addition to the increase in red fox abundance, the population dynamics of the arctic fox’s main prey have changed. In Fennoscandia, arctic foxes are specialist predators that depend strongly on the cyclic density fluctuations of lemming populations, but lemming cycles have altered in the past decades, passing from large outbreaks every 3–5 years to a dampening of the cycle with irregular amplitudes (Elmhagen et al., 2002; Ims et al., 2008). Increased variation in climatic conditions, especially winter conditions, likely explains the alterations in the rodents cycles (Ims et al., 2008; Kausrud et al., 2008). Given the lack of recovery after more than 70 years of protection in Fennoscandia (Audet et al., 2002), direct conservation actions seem to be required to allow arctic foxes to recolonize this region. All the factors that are likely to have contributed to the populations decline hold a potential to influence the recolonization of this species, but the influence of each of these factors and their relative importance is unknown. Based on the two main hypotheses, previous population dynamic models (Henden et al., 2008, 2010) predicted that the presence of arctic foxes in specific areas would be dependent on the abundance of specific prey in the tundra (lemmings), direct competitors (red foxes), or both, with their relative effect depending on the distance to more productive ecosystems (forest and marine). Using a large-scale spatio-temporal study combined with an extensive red fox removal intervention, we test these predictions by evaluating the relative importance of diverse environmental factors affecting the recolonization of arctic foxes in Finnmark, Northern Norway. Because arctic fox abundance has been extremely low in this area, we expected the initial occurrence to be very low and the recolonization process to be slow. Our study design allowed us to disentangle the importance of the two main hypotheses. We expected to find support for both hypotheses, but because the population status has been very

115

low for decades, we expected the variables associated with red fox competition to be the strongest factors influencing recolonization. 2. Materials and methods 2.1. Study area The study was conducted in Finnmark in the northeastern part of Norway (70–71"N). Within the study area, camera traps were placed in three different areas: Nordkynn peninsula, Varanger peninsula, and Ifjordfjellet (Fig. 1). All areas have documented presences of arctic fox in the past in terms of breeding dens, but breeding has only been reported on the Varanger peninsula in recent years (Eide et al., 2010). Prey species available for the two fox species during winter are tundra voles (Microtus oeconomus), gray-sided voles (Myodes rufocanus), Norwegian lemmings (Lemmus lemmus), mountain hares (Lepus timidus), willow ptarmigans (Lagopus lagopus), and rock ptarmigans (Lagopus muta). The study area is made up of summer pastures (Varanger and Nordkinn peninsulas) or transition area (Ifjordfjellet) for large populations of semi-domestic reindeer (Rangifer tarandus). In the nearby forests, field voles (Microtus agrestis), red voles (Myodes rutilus) and moose (Alces alces) form additional prey or carcass resources. Another source of alternative prey is subsidies from the marine ecosystem. The neighboring fjords and the Barents Sea are very productive and ice free during winter (Dayton et al., 1994), providing resources for foxes such as marine invertebrates, fish, and carcasses of sea birds and mammals (Killengreen et al., 2011). 2.2. Study design The study took place within a 2-month period in late winter over 6 years (2005–2010). Originally, 48 sites were sampled. At each site, we placed a camera trap, which consisted of one automatic digital camera placed in front of a frozen block of reindeer slaughter remains. The camera took photographs of the carcass and its close surroundings every 10 min. Flash was activated at night. Both batteries and memory sticks were changed approximately every 14 days, and carcasses that had been completely or partly consumed were replaced. This camera-trap design has been shown to provide reliable estimates for occupancy models (Hamel et al., in press). To perform the analyses, we excluded sites where the camera was either removed or had malfunctioned 1 year, which resulted in 25 sites that were sampled in all years. Both fox species are known to scavenge frequently on ungulate carrion (Elmhagen et al., 2002; Killengreen et al., 2007, 2011), and it has been proposed that such food resources are of paramount importance for persistence of foxes in tundra during winter (Linnell and Strand, 2002; Selås and Vik, 2007). Therefore, pictures provided information on the use of carrion, and we assumed that this correlated with habitat use (or asymptotic occupancy, sensu Efford, 2012) of each species at each site. Although resource use is a fundamental component of habitat use, little is known about the link between carrion and habitat use in foxes, and therefore our findings should mostly be interpreted as a result of carrion use. The camera traps were placed to span three environmental gradients: distance to coast, distance to alpine/arctic forest line (mostly birch trees, Betula pubescens), and altitude. Distance to coast and forest represented the distance to more productive neighboring ecosystems, whereas altitude reflected a gradient in local productivity within the tundra (decreasing productivity with increasing altitude, Aunapuu et al., 2008; Karlsen et al., 2008). In addition, distance to road was measured from each camera trap to investigate if proximity to anthropogenic construction influenced the presence of foxes. Roads were defined as county roads with asphalt cover, and forests were defined as areas with a

116

S. Hamel et al. / Biological Conservation 160 (2013) 114–120

Fig. 1. Study area and location of the camera-trap sites (in blue – squares: Nordkynn peninsula, circles: Varanger peninsula, and triangles: Ifjordfjellet) that were available between 2005 and 2010 in Finnmark, Northern Norway. Roads are shown in red and forested areas in black, and white squares represent locations of lemming trapping sites.

continuous aggregation of trees greater than 1 km2. Altitude varied between 50 and 440 m above sea level, while distance from coast, forest, and road varied between 0.7–30.6 km, 1.1–23.5 km, and 0.4–20.3 km, respectively. Lemmings are a major food resource for both fox species (Elmhagen et al., 2000; Killengreen et al., 2011), and the Norwegian lemming abundance peaked in all areas during the study (Ims et al., 2011). To evaluate how lemming fluctuation influenced occupancy of arctic foxes, we used the trappingbased indices of abundance of lemmings (number of lemmings per 100 trap-nights) from an associated rodent monitoring program in the three tundra areas (see Ims et al., 2011 for methodological details for evaluating lemming abundance). In the analyses, we used lemming indices the previous fall, which were spatially aggregated to the scale of four areas (Nordkynn peninsula, Varanger peninsula north, Varanger peninsula south, and Ifjordfjellet; see Fig. 1). Using abundance of lemmings in the spring or abundance of all rodents provided similar results because these abundance indices are correlated, but we kept the abundance of lemmings in the previous fall since model selection (see below) provided greater support for this abundance index (lower AIC, data not shown). As part of a conservation program, an intensive campaign to cull red foxes on Varanger peninsula was initiated in April 2005, after the first camera trap monitoring was completed. The red fox culling continued over the course of the entire study, amounting to a total of 885 red foxes culled. This resulted in a large-scale removal intervention that divided the study area in two: a control region (Nordkynn peninsula and Ifjordfjellet) and a treatment region (Varanger peninsula). We therefore used this treatment effect to evaluate the influence of red foxes on the recolonization of arctic foxes. We also evaluated the influence of the site-specific presence of red foxes, using the number of days red foxes were observed at a site.

2.3. Statistical analyses We first summarized the presence/absence data from the pictures taken every 10 min to daily species occurrence. We then

modeled arctic fox occupancy by fitting a hierarchical model of wildlife occurrence based on the model of MacKenzie et al. (2003), which allows the estimation of multi-season occupancy of a species. This model estimates the species initial occupancy (w – proportion of sites occupied at the first survey), the probability of colonization (c – probability that a site that was unoccupied is occupied in the following survey) and the probability of extinction (e – probability that a site that was occupied is unoccupied in the following survey), while accounting for variation in the probability of detection (p). We fitted this model using the function ‘‘colext’’ of the package ‘‘unmarked’’ (Fiske and Chandler, 2011) of the R software (R Development Core Team, 2012). Because surveys spanned over a large time frame, we measured occupancy in an open system, and thereby occupancy estimates must be interpreted as the percentage of sites ‘‘used’’ by a species rather than ‘‘occupied’’ (MacKenzie et al., 2004), and hence (re)colonization and extinction must be interpreted as changes in site use. The different covariates included in this model are described in Table 1. To determine the influence of covariates on arctic fox occupancy, we built models a priori and used the Akaike Information Criterion (AIC) to determine their importance (Burnham and Anderson, 2002). We ranked models according to their AIC scores and weights, and we considered models with a DAIC < 2 to be equivalent (Burnham and Anderson, 2002). Because of the numerous covariates included in the study, we performed model selection in four steps. At each step, we varied the influence of covariates on only one process (w, c, e, or p) at a time, holding the other processes constant – that is modeling these processes using either only the intercept or the best model selected if the process had been evaluated at a previous step. We first modeled the probability of detection, the initial occupancy, the probability of colonization, and then the probability of extinction. We used the best model selected at each step to be used for the next step. For example, if the model with only ‘‘year’’ was selected as the best model describing p, then ‘‘year’’ was always included for modeling p in all of the following steps (see online Appendix, Table A1). For

117

S. Hamel et al. / Biological Conservation 160 (2013) 114–120 Table 1 Description of the different covariate types that are possible to include in multiseason occupancy models, along with the respective covariates available in our study. Covariate type

Description

Covariates available

Site covariate

Fixed to a site and hence does not change at each survey

Yearly site covariate

Fixed to a year but they may vary among sites

Observation covariate

Vary among the presence/ absence observations recorded and can only be used to model the detection probability

Distance to forest Distance to coast Distance to the closest road Altitude Treatment vs. control site Year of the survey (constant among all sites) Total number of days red foxes were observed at a site (varying among all sites) Abundance of lemmings (varying among areas but constant among sites aggregated within each area) Date in Julian days Presence/absence of bait Presence/absence of red fox

parsimony, we considered the best model as the simplest of the equivalent models, except when we modeled p. Because the goal of modeling p is to account for detection bias, we considered any covariates included in equivalent models as having an influence on p, and thus used a model including all these covariates for modeling the other three processes (w, c, and e). We first determined the influence of the covariates on p to account for variation in detection when modeling the other occupancy processes. We evaluated the influence of the year of the survey and the three observation covariates (see Table 1) on the probability of detection, while holding the other three processes constant by including only an intercept. Secondly, we modeled the initial occupancy of arctic foxes, holding constant c and e by including only an intercept, and using the modeling selected at the previous step for p. We evaluated the influence of all site covariates and yearly site covariates (Table 1), except for year of the survey, because initial occupancy is only modeled for the first year, and for treatment, because no treatment occurred in the first year. Because no arctic fox was observed in 2005, and hence the initial occupancy was 0, we expected more variation in the probability of colonization than in the probability of extinction over the years. We therefore modeled colonization first, and then used the best model describing colonization to finally evaluate variation in extinction. For both c and e, we evaluated the influence of all site covariates and yearly site covariates (Table 1), using the model selected at the previous steps for w and p. We never included both ‘‘year of the survey’’ and ‘‘abundance of lemmings’’, or ‘‘distance to road’’ and ‘‘distance to coast’’ in the same model, as these pairs of variables were highly correlated (r > 0.6). We presented estimates and 95% confidence intervals (CI) of variables included in the final model selected as a measure of their influence on occupancy processes. Because we centered and standardized all continuous variables, the size of the parameter estimates can be used to compare the importance of each predictor relative to one another (Schielzeth, 2010). Since our models included binary predictors, we standardized continuous variables by dividing by two standard deviations to allow comparisons with binary variables (Gelman, 2008). Because the interpretation of parameter estimates for variables standardized by dividing by two standard deviations can be more difficult than a normal standardization, we illustrated results based on the non-transformed variables.

3. Results Equivalent models describing the probability of detection of arctic foxes included year, date, presence of bait, and presence of red foxes (online Appendix Table A1), but variation in the probability of detection resulted mainly from variation among years (Table 2). Because arctic foxes had not been observed in 2005, initial occupancy was estimated at 0 and was not affected by any covariates (Tables 2, A1), as expected. The best model describing the probability of colonization had a very strong support (high AIC weight, Table A1) and included the influence of number of red foxes, red fox treatment, abundance of lemmings, distance to forest and road, and altitude (Table 2). As soon as about three daily observations of red foxes were recorded, arctic foxes were never observed to colonize a site (Fig. 2). Arctic foxes colonized some sites where removal of red foxes took place, but their probability of colonizing control sites where no red foxes were removed was zero (Fig. 2). The probability of colonization was zero when lemming abundance was low and only increased at very high lemming abundance (Fig. 2). The probability of colonization was zero in close proximity to forest and roads and at higher altitudes, but it increased with increasing distances and at lower altitudes (Fig. 2). Although the confidence intervals for the probability of colonization were large, the effects clearly show that recolonization of arctic foxes only occurred under specific conditions (Fig. 2). The importance of the effect of the number of daily presence of red foxes, the red fox treatment, and the distance to road were similar, although the influence of the number of red foxes was slightly less precise (Table 2). The importance of each of these variables was two times stronger than that of lemming abundance, distance to forest, and altitude (Table 2). Finally, the best of the equivalent models describing the probability of extinction only included the influence of altitude (Table A1), with its effect being similar to the effect it has on colonization (Table 2). Although other covariates were included in equivalent models, their estimates had very wide confidence intervals that largely included 0, demonstrating very low support for these effects on extinction probability.

Table 2 Summary of the estimates describing variation in the probability of initial occupancy, colonization, extinction, and detection of arctic foxes between 2005 and 2010 in Finnmark, Northern Norway. Variable Initial occupancy Intercept Colonization Intercept Red foxes Lemmings Treatment Distance to forest Distance to road Altitude Extinction Intercept Altitude Detection Year 2005 Year 2006 Year 2007 Year 2008 Year 2009 Year 2010 Bait Date Red foxes

Estimate [95% CI] "10.2 ["73.1, 52.7] "7.9 "10.3 4.5 9.4 4.7 11.7 "5.5

["13.6, "2.2] ["21.9, 1.2] [0.6, 8.4] [0.3, 18.5] ["0.9, 10.3] [3.0, 20.4] ["10.4, "0.7]

"1.9 ["3.7, "0.2] "4.7 ["9.0, "0.4] 0.8 "2.4 "1.0 "1.4 "1.9 "1.5 0.7 "0.1 0.2

["306.9, 308.5] ["3.3, "1.6] ["1.9, "0.2] ["2.5, "0.3] ["2.8, "1.1] ["2.3, "0.7] ["0.1, 1.4] ["0.9, 0.8] ["0.5, 0.9]

Treatment – sites with red foxes removal vs. control sites. Red foxes – presence/ absence for the detection process and total number of presence for the colonization process. Lemmings – lemming abundance. Bait – presence of bait.

118

S. Hamel et al. / Biological Conservation 160 (2013) 114–120

4. Discussion

0.0

0.00

0.02

0.1

0.04

0.2

0.06

0.3

0.08

0.4

0.10

In the first year of the study, the percentage of sites used (or the asymptotic occupancy, sensu Efford, 2012) by arctic foxes in Finnmark was estimated to be null. Over the years, arctic foxes recolonized some sites where red foxes had been removed. The probability of recolonization, which must be interpreted as an increase in site use, remained very low, and this resulted in very wide confidence intervals on the recolonization estimates. The low recolonization rate, however, must be seen in the light of the demography of the arctic fox, which is temporally scaled to the period of the lemming cycle (being 4–5 years in the study region; Ims et al., 2011). This means reproduction, and thus population growth, normally takes place only in the peak phase of the lemming cycle, which only occurred once during the 6-year study period. Nevertheless, our study clearly demonstrates that even over a time span where

the potential for recolonization is low, red fox constitutes the main limiting factor to arctic fox recolonization in this part of its distribution range. Indeed, arctic foxes were never observed at sites where just a few daily observations of red foxes were recorded, and the probability of recolonization was zero at sites where no red foxes were removed. Distance to roads was also a strong determinant, with areas located furthest from roads being the only sites likely to be recolonized. To a lesser extent, the probability of recolonization increased in winters when lemming abundance the preceding fall was very high, in areas located furthest from forests, and at sites at low altitude. The relative importance of lemming abundance, forest, and altitude, however, was two times lower than that of red foxes and roads. Note, however, that the effect size of lemming abundance may have been underestimated because the lemming censuses did not accurately match spatially the 25 camera sites (see Fig. 1).

0

10

0

30

20

2

6

8

10

0.0

0.00

0.2

0.05

0.4

0.10

0.6

0.15

0.8

0.20

Number of red foxes

4

Number of lemmings

CONTROL

TREATMENT

5

10

15

20

1.0 0.0

0.0

0.2

0.2

0.4

0.4

0.6

0.6

0.8

0.8

1.0

Distance to forest (km)

0

5

10

15

Distance to road (km)

20

100

200

300

400

Altitude (m)

Fig. 2. Probability of colonization (95% CI) of arctic foxes between 2005 and 2010 in Finnmark, Northern Norway, according to the number of red foxes observed (i.e. total number of days red foxes were observed at a site), the experimental removal of red foxes, the abundance of lemmings (i.e. number lemmings/100 trap-nights), the distance to the forest, the distance to the closest road, and the altitude.

S. Hamel et al. / Biological Conservation 160 (2013) 114–120

With respect to the two main hypotheses proposed for the decline and subsequent lack of recolonization of the arctic fox in Fennoscandia, i.e. changes in prey availability and increased interspecific competition, our results lend support to both hypotheses. Nevertheless, they clearly highlight the presence of red fox as the factor most limiting arctic fox recolonization. According to the asymmetric size competition hypothesis and based on competitive interactions seen among other closely related canids (Fedriani, 2000; Levi and Wilmers, 2012; Nelson, 2007), competition between arctic fox and the larger red fox are expected to be strong (cf. Henden et al., 2010). Many studies have indicated that the larger size red fox can exclude the arctic fox (Herteinsson and Macdonald, 1992; Killengreen et al., 2007; MacPherson, 1964; Tannerfeldt et al., 2002), and our results support these observations as arctic foxes seem to be confined to the areas furthest away from more productive habitats and human activity (see below). In regions where voles and lemmings display high amplitude cyclic fluctuations, both arctic and red fox prey predominantly on small rodents (Elmhagen et al., 2002). The distribution of small rodents, however, is often highly heterogeneous (Hansson, 1977). In Fennoscandia, voles display higher peak densities in more productive lowland than highland tundra areas (Ekerholm et al., 2001), whereas lemmings show the opposite pattern (Ims et al., 2011; Oksanen et al., 2008) but their peak densities are temporally much more erratic and infrequent (Ims et al., 2011). Accordingly, the impact of competition on the arctic fox were predicted to be especially strong if the dominant red fox monopolizes tundra patches providing most resources in terms of voles and ungulate carrion (Henden et al., 2010) and thereby displace the arctic fox to less productive areas (Killengreen et al., 2007; Tannerfeldt et al., 2002). In Sweden, arctic foxes were shown to shift their habitat use from higher to lower altitudes in winter, so that their ranges became more overlapping with the more productive areas used by red foxes outside of the breeding season (Dalén et al., 2004). Our study, however, showed that very low red fox activity drastically reduced arctic fox presence outside of the breeding season, thereby indicating that such a range shift is likely only possible when red fox density is low. It also suggests that even though our study took place in winter, competition with red fox is likely to be stronger in summer. Of all the habitat covariates evaluated, roads had the strongest negative impact on recolonization. Roads may have a negative influence on species distribution and habitat use (Fahrig and Rytwinski, 2009), especially if they create barriers leading to habitat fragmentation, a major process affecting animal movement and habitat use (Forman and Alexander, 1998). Because of the very low human density in the area, however, it seems unlikely that such low disturbance conditions would strongly prevent movement and habitat use of a relatively large and highly mobile mammal like a fox. More importantly, most roads are found close to the coast and forested areas along human settlements, suggesting that the effect of roads probably integrates the influence of all these habitat covariates in concert. Because the marine and forest ecosystems comprise more productive habitats than the tundra, this suggests that arctic foxes recolonized the areas furthest from more productive habitats at the landscape scale. At a more local level, however, arctic foxes were found at lower altitude, and hence in more productive habitats. Altitude is not correlated with the large-scale variables because lower altitude valleys are found in the middle of the peninsulas, far from roads, coastline, and forests (Fig. 1). Therefore, the presence of arctic foxes at lower altitude at the local scale suggests that they do not avoid more productive habitats per se, but are rather excluded from or avoiding more productive areas at a larger scale because of competition by red foxes and/or human disturbances (as suggested by Frafjord et al., 1989; Herteinsson and Macdonald, 1992; Killengreen et al., 2007;

119

Tannerfeldt et al., 2002). This is further supported from findings on Svalbard, where red foxes are absent and arctic foxes use coastal areas more than inland areas (Eide et al., 2012). Because red foxes are the most limiting factor for recolonization and they are abundant in adjacent forests and coastline areas, it seems unlikely that arctic foxes will recolonize the most productive habitats at a larger scale unless red fox abundance is greatly reduced in these habitats. As the abundance of arctic foxes has remained extremely low in Fennoscandia even after more than 70 years of full protection, direct conservation actions seem to be needed to promote the recovery of arctic fox populations. Our research highlights the most important factors limiting recolonization and range use of this species, and thereby where conservation actions should be directed. To enable the recolonization of arctic foxes in the Fennoscandian tundra, conservation actions targeting a major reduction in red fox numbers in the tundra and adjacent ecosystems must be undertaken, and the development of new roads and human settlements should be avoided whenever possible. Nonetheless, as climate change will likely lead to increased presence of red foxes in the tundra (Post et al., 2009), as well as affect the regularity of lemming cycles (Ims et al., 2008), arctic fox recolonization will probably become even more limited in the future. Hence, rapid management intervention is warranted if we hope to see the arctic fox populations recover in Scandinavia. Because the study covers only the very first stage of population recovery, i.e. from almost none to only few individuals, it is likely that other factors such as abundance of prey might become more important at a more advanced stage of population recovery. Thus, continued monitoring of the influence of these factors will be required as conservation actions will be taken to ensure adopting the best conservation strategy for the arctic fox. Acknowledgements We are grateful to Å. Bye, E. Isakson, B.H. Kristoffersen, A.P. Sarre, and A. Ørjebu for tending the camera study and their general help in the field, and to D. Ehrich for help with the habitat figure. We thank A. Angerbjörn and two anonymous reviewers for constructive comments. This study was part of the project ‘Arctic fox in Finnmark’, which was funded by the Norwegian Directorate of Nature Management. Appendix A. Supplementary material Supplementary data associated with this article can be found, in the online version, at http://dx.doi.org/10.1016/j.biocon.2013.01. 011. References Angerbjörn, A., Arvidson, B., Norén, E., Strömgren, L., 1991. The effect of winter food on reproduction in the arctic fox, Alopex lagopus: a field experiment. J. Anim. Ecol. 60, 705–714. Angerbjörn, A., Tannerfeldt, M., Bjärvall, A., Ericson, M., From, J., Norén, E., 1995. Dynamics of the arctic fox population in Sweden. Ann. Zool. Fennici 32, 55–68. Audet, A.M., Robbins, C.B., Larivière, S., 2002. Alopex lagopus. Mamm. Species 713, 1–10. Aunapuu, M., Dahlgren, T., Oksanen, T., Grellmann, D., Oksanen, L., Olofsson, J., Rammul, U., Schneider, M., Johansen, B., Hygen, H.O., 2008. Spatial patterns and dynamic responses of arctic food webs corroborate the exploitation ecosystems hypothesis (EEH). Am. Nat. 171, 249–262. Bergerud, A.T., 1974. Decline of caribou in North America following settlement. J. Wildl. Manage. 38, 757–770. Bergerud, A.T., Ballard, W.B., 1988. Wolf predation on caribou: the Nelchina herd case history, a different interpretation. J. Wildl. Manage. 52, 344–357. Burnham, K.P., Anderson, D.R., 2002. Model Selection and Multimodel Inference. a Practical Information-Theoretic Approach, second ed. Springer, New York. Byers, J.E., 2002. Impact of non-indigenous species on natives enhanced by anthropogenic alteration of selection regimes. Oikos 97, 449–458.

120

S. Hamel et al. / Biological Conservation 160 (2013) 114–120

Cumming, H.G., 1992. Woodland caribou: facts for forest managers. For. Chron. 68, 481–491. Dalén, L., Elmhagen, B., Angerbjörn, A., 2004. DNA analysis on fox faeces and competition induced niche shifts. Mol. Ecol. 13, 2389–2392. Dayton, P.K., Mordida, B.J., Bacon, F., 1994. Polar marine communities. Am. Zool. 34, 90–99. Didham, R.K., Tylianakia, J.M., Hutchison, M.A., Ewers, R.M., Gemmell, N.J., 2005. Are invasive species the drivers of ecological change? Trends Ecol Evol. 20, 470– 474. Efford, 2012. Occupancy in continuous habitat. Ecosphere 3 (Art. 32). Eide, N.E., Flagstad, Ø., Andersen, R., Landa, A., 2010. Fjellrev i Norge 2009: Resultater fra det nationale overvakningsprogrammet for fjellrev. NINA Rapport 519, 43. Eide, N.E., Stien, A., Prestrud, P., Yoccoz, N.G., Fuglei, E., 2012. Reproductive responses to spatial and temporal prey availability in a coastal Arctic fox population. J. Anim. Ecol. 81, 640–648. Ekerholm, P., Oksanen, L., Oksanen, T., 2001. Long-term dynamics of voles and lemmings at the timberline and above the willow limit as a test of hypotheses on trophic interactions. Ecography 24, 555–568. Elmhagen, B., Tannerfeldt, M., Verucci, P., Angerbjörn, A., 2000. The arctic fox (Alopex lagopus): an opportunistic specialist. J. Zool. 251, 139–149. Elmhagen, B., Tannerfeldt, M., Angerbjörn, A., 2002. Food-niche overlap between Arctic and red foxes. Can. J. Zool. 80, 1274–1285. Fahrig, L., Rytwinski, T., 2009. Effects of roads on animal abundance: an empirical review and synthesis. Ecol. Soc. 14 (Art. 21). Fedriani, J.M., 2000. Competition and intraguild predation among three sympatric carnivores. Oecologia 125, 258–270. Fiske, I.J., Chandler, R.B., 2011. Unmarked: an R package for fitting hierarchical models of wildlife occurrence and abundance. J. Stat. Softw. 43, 1–23. Forman, R.T.T., Alexander, L.E., 1998. Roads and their major ecological effects. Annu. Rev. Ecol. Syst. 29, 207–231. Frafjord, K., Becker, D., Angerbjörn, A., 1989. Interactions between arctic and red foxes in Scandinavia: predation and aggression. Arctic 42, 354–356. Gelman, A., 2008. Scaling regression inputs by dividing by two standard deviations. Statist. Med. 27, 2865–2873. Hamel, S., Killengreen, S.T., Henden, J.-A., Eide, N.E., Roed-Eriksen, L., Ims, R.A., Yoccoz, N.G., in press. Towards good practice guidance in using camera traps in ecology: influence of sampling design on validity of ecological inferences. Method. Ecol. Evol. http://dx.doi.org/10.1111/j.2041-210x.2012.00262.x. Hansson, L., 1977. Spatial dynamics of field voles Microtus agrestis in heterogeneous landscapes. Oikos 29, 539–544. Henden, J.-A., Bårdsen, B.-J., Yoccoz, N.G., Ims, R.A., 2008. Impacts of differential prey dynamics on the potential recovery of endangered arctic fox populations. J. Appl. Ecol. 45, 1086–1093. Henden, J.-A., Ims, R.A., Yoccoz, N.G., Hellström, P., Angerbjörn, A., 2010. Strength of asymmetric competition between predators in food webs ruled by fluctuating prey: the case of foxes in tundra. Oikos 119, 27–34. Hersteinsson, P., Angerbjörn, A., Frafjord, K., Kaikusalo, A., 1989. The arctic fox in Fennoscandia and Iceland: management problems. Biol. Conserv. 49, 67–81. Herteinsson, P., Macdonald, D.W., 1992. Interspecific competition and the geographical distribution of red and arctic foxes, Vulpes vulpes and Alopex lagopus. Oikos 64, 505–515. Ims, R.A., Henden, J.-A., Killengreen, S.T., 2008. Collapsing population cycles. Trends Ecol. Evol. 23, 79–86. Ims, R.A., Yoccoz, N.G., Killengreen, S.T., 2011. Determinants of lemming outbreaks. Proc. Natl. Acad. Sci. 108, 1970–1974. Kaikusalo, A., Angerbjörn, A., 1995. The arctic fox population in Finnish Lapland during 30 years, 1964–9193. Ann. Zool. Fennici 32, 69–77. Kallimanis, A.S., Kunin, W.E., Halley, J.M., Sgardelis, S.P., 2005. Metapopulation extinction risk under spatially autocorrelated disturbance. Cons. Biol. 19, 534–546. Karlsen, S.R., Tolvanen, A., Kubin, E., Poikolainen, J., Høgda, K.A., Johansen, B., Danks, F.S., Aspholm, P., Wielgolaski, F.E., Makarova, O., 2008. MODIS-NDVI-based mapping of the length of the growing season in northern Fennoscandia. Int. J. Appl. Earth Obs. Geoinformation 10, 253–266.

Kausrud, K.L., Mysterud, A., Steen, H., Vik, J.O., Østbye, E., Cazelles, B., Framstad, E., Eikeset, A.M., Mysterud, I., Solhøy, T., Stenseth, N.C., 2008. Linking climate change to lemming cycles. Nature 456, 93–98. Killengreen, S.T., Ims, R.A., Yoccoz, N.G., Bråthen, K.A., Henden, J.-A., Schott, T., 2007. Structural characteristics of a low Arctic tundra ecosystem and the retreat of the Arctic fox. Biol. Conserv. 135, 459–472. Killengreen, S.T., Lecomte, N., Ehrich, D., Schott, T., Yoccoz, N.G., Ims, R.A., 2011. The importance of marine vs. human-induced subsidies in the maintenance of an expanding mesocarnivore in the arctic tundra. J. Anim. Ecol. 80, 1049–1060. Levi, T., Wilmers, C.C., 2012. Wolves-coyotes-foxes: a cascade among carnivores. Ecology 93, 921–929. Linnell, J.C.D., Strand, O., 2002. Do arctic foxes Alopex lagopus depend on kills made by large predators? Wildl. Biol. 8, 69–75. Macdonald, D.W., Sillero-Zubiri, C., 2004. Biology and Conservation of Wild Canids. Oxford University Press, Oxford. MacKenzie, D.I., Nichols, J.D., Hines, J.E., Knutson, M.G., Franklin, A.B., 2003. Estimating site occupancy, colonization, and local extinction when a species is detected imperfectly. Ecology 84, 2200–2207. MacKenzie, D.I., Bailey, L.L., Nichols, J.D., 2004. Investigating species co-occurrence patterns when species are detected imperfectly. J. Anim. Ecol. 73, 546–555. MacPherson, A.H., 1964. A northward range extension of the red fox in the eastern Canadian Arctic. J. Mammal. 45, 138–140. Nelson, J.L., 2007. Effects of habitat on competition between kit foxes and coyotes. J. Wildl. Manage. 71, 1467–1475. Oksanen, T., Oksanen, L., Dahlgren, J., Olofsson, J., 2008. Arctic lemmings, Lemmus spp. and Dicrostonyx spp.: integrating ecological and evolutionary perspectives. Evol. Ecol. Res. 10, 415–434. Palomares, F., Caro, T.M., 1999. Interspecific killing among mammalian carnivores. Am. Nat. 153, 492–508. Post, E., Forchhammer, M.C., Bret-Harte, M.S., Callaghan, T.V., Christensen, T.R., Elberling, B., Fox, A.D., Gilg, O., Hik, D.S., Høye, T.T., Ims, R.A., Jeppesen, E., Klein, D.R., Madsen, J., McGuire, A.D., Rysgaard, S., Schindler, D.E., Stirling, I., Tamstorf, M.P., Tyler, N.J.C., van der Wal, R., Welker, J., Wooker, P.A., Schmidt, N.M., Aastrup, P., 2009. Ecological dynamics across the Arctic associated with recent climate change. Science 325, 1355–1358. R Development Core Team, 2012. R: A Language and Environment for Statistical Computing. R Foundation for Statistical Computing, Vienna, Austria. Roemer, G.W., Donlan, C.J., Courchamp, F., 2002. Golden eagles, feral pigs, and insular carnivores: how exotic species turn native predators into prey. Proc. Natl. Acad. Sci. 99, 791–796. Schaefer, J.A., 2003. Long-term range recession and the persistence of caribou in the taiga. Cons. Biol. 17, 1435–1439. Schielzeth, H., 2010. Simple means to improve the interpretability of regression coefficients. Method. Ecol. Evol. 1, 103–113. Seip, D.R., 1991. Predation and caribou populations. Rangifer 11, 46–52. Selås, V., Vik, J.O., 2007. The arctic fox Alopex lagopus in Fennoscandia: a victim of human-induced changes in interspecific competition and predation? Biodivers. Conserv. 16, 3575–3583. Shirley, M.D.F., Elmhagen, B., Lurz, P.W.W., Rushton, S.P., Angerbjörn, A., 2009. Modelling the spatial population dynamics of arctic foxes: the effects of red foxes and microtine cycles. Can. J. Zool. 87, 1170–1183. Tannerfeldt, M., Elmhagen, B., Angerbjörn, A., 2002. Exclusion by interference competition? The relationship between red and Arctic foxes. Oecologia 132, 213–220. Vitousek, P.M., 1994. Beyond global warming: ecology and global change. Ecology 75, 1861–1876. Vors, L.S., Schaefer, J.A., Pond, B.A., Rodgers, A.R., Patterson, B.R., 2007. Woodland caribou extirpation and anthropogenic landscape disturbance in Ontario. J. Wildl. Manage. 71, 1249–1256. White, P.S., Jentsch, A., 2001. The search for generality in studies of disturbance and ecosystem dynamics. Prog. Bot. 62, 399–449. Wittmer, H.U., McLellan, B.N., Serrouya, R., Apps, C.D., 2007. Changes in landscape composition influence the decline of a threatened woodland caribou population. J. Anim. Ecol. 76, 568–579.

Table A1. Summary of the results of the model selection process to evaluate the influence of environmental covariates and the presence of red foxes on occupancy estimates of arctic foxes between 2005 to 2010, in Finnmark, Northern Norway. The multi-season occupancy model estimates four processes: the initial occupancy (ψ: psi), the probability of colonization (γ: gam), the probability of extinction (ε: eps), and the probability of detection (p), which are expressed as “psi( )gam( )eps( )p( )” in the table. The variables affecting each process are located in the parentheses of each process, and the abbreviations are simply the first letter of the name of each variable (i.e. Y: Years, T: Treatment, R: Red foxes, L: Lemming, B: Bait, D: Date, A: Altitude; see Footnote for more details), except for the distances which are indicated by two letters (DC: Distance to the coast, DF: Distance to the forest, DR: Distance to roads). As we evaluated the influence of these variables on only one process at a time, only one of the four parentheses varies in each section of the table (the other processes are kept constant with only the intercept or with the best model of this process if it was evaluated previously; see Methods for more details). Equivalent models selected at each process are highlighted in bold, and the final model selected is highlighted in blue.

k

AIC

ΔAIC

AIC weight

Cumulative AIC weight

Detection - p psi(.)gam(.)eps(.)p(Y) psi(.)gam(.)eps(.)p(Y.B.D) psi(.)gam(.)eps(.)p(Y.B.D.R) psi(.)gam(.)eps(.)p(B) psi(.)gam(.)eps(.)p(D) psi(.)gam(.)eps(.)p(.) psi(.)gam(.)eps(.)p(R)

9 11 12 5 5 4 5

657.93 657.95 659.31 666.76 669.55 669.88 670.94

0.00 0.02 1.38 8.83 11.62 11.94 13.00

0.40 0.39 0.20 0.00 0.00 0.00 0.00

0.40 0.79 0.99 1.00 1.00 1.00 1.00

Initial occupancy - psi psi(.)gam(.)eps(.)p(Y.B.D.R) psi(L)gam(.)eps(.)p(Y.B.D.R) psi(DC)gam(.)eps(.)p(Y.B.D.R)

12 13 13

659.31 661.31 661.31

0.00 2.00 2.00

0.29 0.11 0.11

0.29 0.40 0.50

MODELS

psi(DR)gam(.)eps(.)p(Y.B.D.R) psi(DF)gam(.)eps(.)p(Y.B.D.R) psi(R)gam(.)eps(.)p(Y.B.D.R) psi(A)gam(.)eps(.)p(Y.B.D.R) psi(R.L)gam(.)eps(.)p(Y.B.D.R) psi(DR.DF.A)gam(.)eps(.)p(Y.B.D.R) psi(DC.DF.A)gam(.)eps(.)p(Y.B.D.R) Colonization - gam psi(.)gam(R.L.T.DF.DR.A)eps(.)p(Y.B.D.R) psi(.)gam(R.L.T.DR.A)eps(.)p(Y.B.D.R) psi(.)gam(R.L.T.DF.DC.A)eps(.)p(Y.B.D.R) psi(.)gam(Y.R.DC.A)eps(.)p(Y.B.D.R) psi(.)gam(R.L.DC.A)eps(.)p(Y.B.D.R) psi(.)gam(R.L.T.DC.A)eps(.)p(Y.B.D.R) psi(.)gam(Y.DC.A)eps(.)p(Y.B.D.R) psi(.)gam(Y.DF)eps(.)p(Y.B.D.R) psi(.)gam(Y.R.DR.A)eps(.)p(Y.B.D.R) psi(.)gam(Y.T)eps(.)p(Y.B.D.R) psi(.)gam(Y.R)eps(.)p(Y.B.D.R) psi(.)gam(Y.R.DF.A)eps(.)p(Y.B.D.R) psi(.)gam(Y.DC)eps(.)p(Y.B.D.R) psi(.)gam(Y.T.R)eps(.)p(Y.B.D.R) psi(.)gam(Y)eps(.)p(Y.B.D.R) psi(.)gam(Y.A)eps(.)p(Y.B.D.R) psi(.)gam(L.T.R)eps(.)p(Y.B.D.R) psi(.)gam(L.T)eps(.)p(Y.B.D.R) psi(.)gam(DF)eps(.)p(Y.B.D.R) psi(.)gam(Y.DF.A)eps(.)p(Y.B.D.R) psi(.)gam(R)eps(.)p(Y.B.D.R) psi(.)gam(T)eps(.)p(Y.B.D.R) psi(.)gam(R.L.DR.A)eps(.)p(Y.B.D.R) psi(.)gam(DC)eps(.)p(Y.B.D.R) psi(.)gam(L.R)eps(.)p(Y.B.D.R) psi(.)gam(R.L.T.DF.A)eps(.)p(Y.B.D.R) psi(.)gam(Y.DR.A)eps(.)p(Y.B.D.R)

13 13 13 13 14 15 15

661.31 661.31 661.31 661.31 663.31 665.31 665.31

2.00 2.00 2.00 2.00 4.00 6.00 6.00

0.11 0.11 0.11 0.11 0.04 0.01 0.01

0.61 0.72 0.82 0.93 0.97 0.99 1.00

18 17 18 19 16 17 18 17 19 17 17 19 17 18 16 17 15 14 13 18 13 13 16 13 14 17 18

643.91 647.13 647.73 651.85 652.94 653.29 653.61 655.80 655.84 656.04 656.17 656.38 656.45 656.73 657.13 657.44 657.44 657.46 657.49 657.56 657.67 658.00 658.20 658.31 658.58 658.71 658.83

0.00 3.22 3.82 7.94 9.03 9.37 9.70 11.89 11.92 12.13 12.26 12.47 12.54 12.82 13.22 13.52 13.53 13.55 13.57 13.65 13.75 14.09 14.28 14.40 14.66 14.79 14.92

0.70 0.14 0.10 0.01 0.01 0.01 0.01 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00

0.70 0.84 0.95 0.96 0.97 0.97 0.98 0.98 0.98 0.98 0.99 0.99 0.99 0.99 0.99 0.99 0.99 0.99 0.99 0.99 1.00 1.00 1.00 1.00 1.00 1.00 1.00

psi(.)gam(Y.DR)eps(.)p(Y.B.D.R) psi(.)gam(.)eps(.)p(Y.B.D.R) psi(.)gam(R.L.DF.A)eps(.)p(Y.B.D.R) psi(.)gam(A)eps(.)p(Y.B.D.R) psi(.)gam(Y.DF.DR)eps(.)p(Y.B.D.R) psi(.)gam(L)eps(.)p(Y.B.D.R) psi(.)gam(DR)eps(.)p(Y.B.D.R) Extinction - eps psi(.)gam(R.L.T.DF.DR.A)eps(R.L.DC.A)p(Y.B.D.R) psi(.)gam(R.L.T.DF.DR.A)eps(R.L.DF.A)p(Y.B.D.R) psi(.)gam(R.L.T.DF.DR.A)eps(A)p(Y.B.D.R) psi(.)gam(R.L.T.DF.DR.A)eps(R.L.T.DC.A)p(Y.B.D.R) psi(.)gam(R.L.T.DF.DR.A)eps(R.L.T.DF.A)p(Y.B.D.R) psi(.)gam(R.L.T.DF.DR.A)eps(Y.R.DC.A)p(Y.B.D.R) psi(.)gam(R.L.T.DF.DR.A)eps(DF)p(Y.B.D.R) psi(.)gam(R.L.T.DF.DR.A)eps(DC)p(Y.B.D.R) psi(.)gam(R.L.T.DF.DR.A)eps(Y.R.DF.A)p(Y.B.D.R) psi(.)gam(R.L.T.DF.DR.A)eps(.)p(Y.B.D.R) psi(.)gam(R.L.T.DF.DR.A)eps(T)p(Y.B.D.R) psi(.)gam(R.L.T.DF.DR.A)eps(R)p(Y.B.D.R) psi(.)gam(R.L.T.DF.DR.A)eps(L)p(Y.B.D.R) psi(.)gam(R.L.T.DF.DR.A)eps(L.T)p(Y.B.D.R) psi(.)gam(R.L.T.DF.DR.A)eps(Y.A)p(Y.B.D.R) psi(.)gam(R.L.T.DF.DR.A)eps(L.R)p(Y.B.D.R) psi(.)gam(R.L.T.DF.DR.A)eps(L.T.R)p(Y.B.D.R) psi(.)gam(R.L.T.DF.DR.A)eps(Y.DC)p(Y.B.D.R) psi(.)gam(R.L.T.DF.DR.A)eps(Y.DC.A)p(Y.B.D.R) psi(.)gam(R.L.T.DF.DR.A)eps(Y.DF.A)p(Y.B.D.R) psi(.)gam(R.L.T.DF.DR.A)eps(Y.DF)p(Y.B.D.R) psi(.)gam(R.L.T.DF.DR.A)eps(R.L.DR.A)p(Y.B.D.R) psi(.)gam(R.L.T.DF.DR.A)eps(Y)p(Y.B.D.R) psi(.)gam(R.L.T.DF.DR.A)eps(Y.T)p(Y.B.D.R) psi(.)gam(R.L.T.DF.DR.A)eps(R.L.T.DR.A)p(Y.B.D.R) psi(.)gam(R.L.T.DF.DR.A)eps(Y.R)p(Y.B.D.R) psi(.)gam(R.L.T.DF.DR.A)eps(Y.DR.A)p(Y.B.D.R)

17 12 16 13 18 13 13

659.03 659.31 659.54 659.87 660.57 661.03 661.10

15.11 15.40 15.62 15.95 16.65 17.12 17.19

0.00 0.00 0.00 0.00 0.00 0.00 0.00

1.00 1.00 1.00 1.00 1.00 1.00 1.00

22 22 19 23 23 25 19 19 25 18 19 19 19 20 23 20 21 23 24 24 23 22 22 23 23 23 24

638.16 639.14 639.99 640.16 641.13 641.26 641.80 642.63 643.63 643.91 644.54 645.44 645.64 646.48 646.72 647.44 648.31 648.47 648.60 648.64 649.08 650.23 650.59 651.90 652.23 652.58 653.69

0.00 0.98 1.83 2.00 2.97 3.10 3.64 4.47 5.47 5.75 6.38 7.28 7.48 8.32 8.56 9.28 10.15 10.32 10.44 10.48 10.92 12.07 12.43 13.74 14.07 14.42 15.53

0.30 0.18 0.12 0.11 0.07 0.06 0.05 0.03 0.02 0.02 0.01 0.01 0.01 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00

0.30 0.48 0.60 0.71 0.77 0.84 0.88 0.91 0.93 0.95 0.96 0.97 0.98 0.98 0.99 0.99 0.99 0.99 0.99 1.00 1.00 1.00 1.00 1.00 1.00 1.00 1.00

psi(.)gam(R.L.T.DF.DR.A)eps(Y.T.R)p(Y.B.D.R) psi(.)gam(R.L.T.DF.DR.A)eps(R.L.T.DF.DC.A)p(Y.B.D.R) psi(.)gam(R.L.T.DF.DR.A)eps(R.L.T.DF.DR.A)p(Y.B.D.R) psi(.)gam(R.L.T.DF.DR.A)eps(Y.DF.DR)p(Y.B.D.R) psi(.)gam(R.L.T.DF.DR.A)eps(DR)p(Y.B.D.R) psi(.)gam(R.L.T.DF.DR.A)eps(Y.R.DR.A)p(Y.B.D.R) psi(.)gam(R.L.T.DF.DR.A)eps(Y.DR)p(Y.B.D.R)

24 24 24 24 19 25 23

653.87 654.19 654.22 654.40 654.70 656.46 661.85

15.71 16.03 16.06 16.24 16.54 18.30 23.69

0.00 0.00 0.00 0.00 0.00 0.00 0.00

(.) Intercept only. (Y) Years in categories from 2005 to 2010. (T) Treatment – sites with red foxes removal vs. control sites. (R) Red foxes – presence/absence for the detection process and total number of presence for the other processes. (L) Lemming abundance. (B) Bait presence. (D) Date. (DC) Distance to the coast. (DF) Distance to the forest. (DR) Distance to the closest road. (A) Altitude.

1.00 1.00 1.00 1.00 1.00 1.00 1.00