Can scale-dependent landcover relationships explain canid community composition independent of intraguild occupancy?

Niche theory is frequently used as a framework to integrate environmental variables and species interactions to describe species geographic distribution. Yet, the scale at which species respond to the environment and other species is rarely considered in species distribution modeling. Here we examined the effect of spatial scale on species distribution modeling for a multi-species canid community across a complex landscape. We examined the relationships between the occurrence of three canid species and landcover variables at different spatial scales and tested the effect of intraguild co-occurrence of dominant species (coyote and red fox) on the distribution of the smallest and subordinate species (swift fox), a species of conservation concern in North America’s grasslands. We predicted the geographical distribution of the canid species and their responses to variation in the landscape by identifying how species associate with habitat and at what scales. We modeled occupancy of each species with a Bayesian approach and incorporating Bayesian Latent Indicator Scale Selection to identify ‘scales of effect’ independently by species and land cover type. We demonstrated that each species responded uniquely to the landcover features examined at different scales and expressed distinct habitat relationships. We obtained low detection rates for all canid species (0.10–0.14). Mean occupancy probabilities for both fox species were low (  0.60).We found no evidence of intraguild co-occurrence affecting the subordinate species. We demonstrate substantial variation in the spatial scales of the species-habitat relationships. The distribution of swift fox appears more limited by resource availability that is reflected by larger landscape attributes than the presence or absence of other guild members per se.