Although the high ecological similarity of sympatric red and Arctic foxes could induce strong exploitation competition, the fine-scale mechanisms driving den selection may differ enough to promote co-existence, at least at low densities. Red foxes, contrary to our expectations, did not switch strategies for selecting dens at drastically different prey abundances across seasons. Both species selected dens to occupy based on physical criteria rather than their proximity to landscape features associated with specific prey types, but the physical variables driving den selection differed by species. Furthermore, a large proportion of the dens favored by red foxes remained available each year, suggesting that red foxes did not exclude Arctic foxes from these specific dens. Finally, both species tolerated each other in close proximity, further evidence that interference competition between these species is low in the Canadian Arctic, unlike in Eurasia (Tannerfeldt et al. 2002; Rodnikova et al. 2011; Lai et al. 2022).
Carnivore spatial and social strategies depend on resource availability (MacDonald 1983; Eide et al. 2004). Tundra summers are characterized by a sudden short-term burst of resource availability (Tannerfeldt et al. 1998; Eide et al. 2004): over 200 species of migratory birds nest in our study area, and a high density of nests is widely distributed throughout the landscape (Ballantyne and Nol 2011; McDonald et al. 2017). Although foxes mainly feed on arvicoline rodents, they respond functionally to the high abundance of migratory birds (McDonald et al. 2017). In summer, food is homogeneously distributed and abundant in our study area, it is not a limiting resource, and no hotspot of resource could create heterogeneous attractivity of dens. Thus, as expected (P2) food proximity did not drive den selection in August for either species. Instead, foxes favored shelter characteristics, which we also expected given vulnerable pups are present on dens throughout summer (Tannerfeldt et al. 2002), as well as diverse predators (eagles, three bear species, wolves, wolverines) and pests.
However, at the beginning of the fox breeding season, food sources are scarce on the tundra. On Herschel Island, Yukon, red foxes selected dens with better tundra food access in spring, and Arctic foxes always favored physical shelter characteristics (Gallant et al. 2014). Differences in prey abundance and distribution between Herschel and our study area may explain the difference in red foxes' den selection criteria. On Herschel Island, brown lemmings inhabiting the low-elevation humid habitats can peak at 60 lemming/ha, thus creating rodent hot spots for foxes because these habitats represent only 1% of the island (Lai et al. 2022). Such hot spots of spring resources are unlikely to occur in our study area, where rodent density is continuously low (Table 2) and where rodents occupy the most abundant habitats (coastal fen) in spring (Dobroski 2022). The low Arctic is subjected to more melt-freeze cycles, which are highly detrimental to small mammals, rendering winter conditions in our study area even harsher than in the high Arctic. Thus, tundra food access may become secondary to shelter quality if most dens provide similar access to prey.
When primary-prey abundance is low, Arctic foxes may use the sea ice, despite higher mortality rates associated with leaving a home range (Roth 2002; Lai et al. 2017; Warret Rodrigues and Roth submitted). Availability of marine resources is highly unpredictable (Roth 2002), and Arctic foxes can commute up to 40 km to the sea ice in short extraterritorial trips to access marine-mammal carcasses without risking losing their home ranges (Lai et al. 2015). In addition, foxes may share concentrated abundant food resources (Tsukada 1997; Eide et al. 2004; Lai et al. 2015). The low performance of coastal-food models in explaining den selection is thus unsurprising.
Like in Scandinavia, our fox populations are close to treeline, and red foxes tended to select dens closer to forested habitats more often than Arctic foxes (Linnell et al. 1999; Dalerum et al. 2002; see Table S1). Tree patches should constitute a prized resource for red foxes: they are home to taiga prey such as snowshoe hares and red squirrels and offer milder conditions in this otherwise harsh region (Ponomarenko et al. 2014). However, the proximity of treeline did not drive den selection for red foxes. These red foxes are an edge population and are thus challenged by scarce availability of preferred habitat, the low availability of food, and their tolerance limits for abiotic conditions (Romeo et al. 2010; Niedzielski and Bowman 2016). Intraspecific competition in red foxes is likely strong, thus red foxes inhabiting tundra dens may be the losers of that intraspecific competition, relegated into sub-optimal habitats, forcing them to prioritize the shelter characteristics of the dens, especially in April when winter conditions remain on the tundra. Our treeless study area is dominated by the low hypoarctic tundra and is colder than the other bioclimatic zones in the region because it is strongly influenced by the cooling effect of Hudson Bay (Ponomarenko et al. 2014). The odds of a red fox using a den increased by 30 for each additional burrow oriented to the south or southeast (burrows likely thawing first when obstructed by ice), supporting the idea that red foxes likely chose dens that can shelter them from the north winds sweeping in from Hudson Bay.
Tundra food access was always secondary to shelter quality for red foxes and Arctic foxes in both seasons based on AIC, but the AUC values (logistic component) suggested the tundra food model performed slightly better than shelter quality for Arctic foxes in April and August. These results suggest that Arctic foxes may select dens using multiple criteria (for example, their choice may be equally influenced by shelter quality and food availability), or that we did not identify crucial variables that drive den selection in this species. For example, we could not differentiate between settled resident and transient foxes, which could have different needs and priorities for den use. Foxes in our study area disperse from their home ranges in spring in search of better conditions more frequently than in the high Arctic (Warret Rodrigues and Roth submitted; Lai et al. 2017), and Arctic fox occupancy in April, at the beginning of the breeding season, decreased over time, suggesting the low hypoartic tundra zone is not a favorable zone to settle in spring and Arctic foxes may leave the area to find other sources of food. Red fox den occupancy increased over time only in summer, not at the beginning of the breeding season, suggesting red foxes living at the edge of their distribution may be reluctant to settle in the area in spring.
Willows on the den surface played apparently contradictory roles in the shelter quality model for red foxes in April. When willows were present, the odds of a red fox using a den were 89 times higher than when willows were absent, but among dens used, presence of willows on top lowered the odds of frequent use by 35%. Although the willows retain snow, which could improve insulation and provide better lemming habitat, the odds of a red fox using a den decreased by nearly 100% per meter increase of snow depth. Foxes are ecosystem engineers: through their activities (digging, excreting, bringing prey carcasses) they concentrate soil nutrients and modify the vegetation composition on dens, notably favoring the growth of shrubs like willows (Gharajehdaghipour et al. 2016; Fafard et al. 2020). The growth of willows on specific dens likely results from their intensive use for decades, possibly over centuries (McPherson 1969). As a result, this variable in the logistic component of our models may simply reflect the general quality and attraction level of particular dens. As the willows grow, they accumulate more snow during winter, which may make these dens less accessible and therefore less attractive for red foxes over time.
Several studies have suggested that red and Arctic foxes behave with flexibility regarding territoriality (Tsukada 1997; Strand et al. 2000; Goszczyński 2002; Eide et al. 2004). Elsewhere in the Arctic, territoriality was often a key determinant of den selection by Arctic foxes, which preferred isolated dens (Linnell et al. 1999; Dalerum et al. 2002; Szor et al. 2008; Gallant et al. 2014). Clumped and random-spacing occupancy patterns have also been observed, likely due to topographic features reducing the availability of good denning sites (Fine 1980; Prestrud 1992; Anthony 1996). In our area, Arctic foxes preferred clumped dens, but red foxes disliked dens in aggregations despite favoring a shorter distance to the nearest den. Clumps of dens, although they may attract competitors, allow foxes to split and relocate litters, a practice that may have evolved as an anti-predatory behavior and to lower parasitic loads (Kilgore 1969; Prestrud 1992; Anthony 1996). As mentioned above, larger predators are numerous in our area, likely driving our foxes to prioritize anti-predatory strategies. Red foxes may have adopted an intermediate strategy, avoiding clumped dens to reduce interactions with possible competitors, but still favoring sites that facilitate litter split and relocation to avoid predation and parasitic overload.
Denning distances between heterospecific-fox pairs, both average and minimum, were lower than in Scandinavia (Tannerfeldt et al. 2002) and similar to observations in the Canadian high Arctic, where the minimal distance between breeding red and Arctic foxes was also 2.3 km (Lai et al. 2022). Furthermore, like in the high Arctic (Gallant et al. 2014), previous occupation by a red fox does not deter Arctic foxes, even during the same season (Johnson-Bice et al. in press). Finally, only a small proportion of dens favored by red foxes are in use each year, suggesting that Arctic foxes are not excluded from high-quality dens. Contrary to our predictions (P3 and P4), season did not affect average distance between neighbors, and although the distance between homospecific Arctic fox neighbors was shorter than between heterospecific neighbors (as we expected), distance between homospecific red fox neighbors was not. This spacing pattern suggests competition for space and territoriality, but not interference. Red foxes cope with the harsh winter weather by increasing their basal metabolic rate to expand their thermoneutral zone, which also increases their energetic requirements (Fuglesteg et al. 2006; Careau et al. 2007). Over-wintering on the low-Arctic tundra presents challenges for red foxes, as most red foxes had to leave the area altogether in search of milder conditions, while the remaining residents ranged nearly twice as much in winter compared to summer (Warret Rodrigues & Roth submitted). Resident Arctic foxes, in contrast, maintained a similar ranging behavior year-round. The spacing patterns of den occupancy we observed seem to be driven by the red fox need for space: the average distance between dens occupied by red foxes and neighbors of both species were similar, corresponding approximately to the diameter of a red fox winter home range (~ 6.7 km; Warret Rodrigues and Roth submitted). Red foxes seem to be more sensitive to the nearest neighbors, regardless of species, than Arctic foxes. The lack of seasonal effect on neighbor distance could be caused by increased territoriality, and generally lower tolerance for congeners and conspecifics, due to the presence of pups (Tannerfeldt et al. 2002).
Red and Arctic foxes differ in their fine-scale den-selection criteria, which likely help relax exploitation competition over denning sites and we did not find evidence of interference competition. In partial agreement with P5, red fox occupancy has increased in August, while Arctic fox occupancy at the end of the breeding season has remained stable, suggesting that currently red foxes are not excluding Arctic foxes. Coexistence mechanisms are widespread and usually involve some sort of segregation, including resource segregation (Ritchie 2002). Our study demonstrates that, under some favorable circumstances, resource segregation can result from fine-scale differences in competitors’ realized niches. Taiga species settling on the tundra (“newcomers”), could thus coexist with their tundra competitors, at given densities of both competitors. Under the current Arctic conditions, the abiotic factors and low density of the settling species may still induce stronger intraspecific competition compared to interspecific competition, thus favoring the long-term coexistence of newcomers and native tundra species (Amarasekare 2003; Banks et al. 2013). Milder abiotic conditions resulting from Arctic warming may relax intraspecific competition over key resources for newcomers and thus increase interspecific competition (especially if the abundance of newcomers increases) thus disrupting the current balance that favors species coexistence. In that context, our study provides a starting point to monitor the changes in species relationships that could result from climate change.