Determinants of occupancy
In winter, local road density was the most important predictor of occupancy for wolf, with higher road density associated with a lower probability of wolf occupancy (Fig 2C). This corroborates findings from Jedrzejewskiet al. (2004) in northern Poland where wolf had higher occupancy in less disturbed or less fragmented forests. In our study area, the proportion of forest was not an important predictor of wolf occupancy in either season, even though multiple studies have found it to be an important habitat characteristic for wolf (Jedrzejewski et al., 2004; Zlatanova & Popova, 2013) This may be due to the characteristics of our study area which is heavily forested (mean proportion forest = 0.78 and 0.75 for winter and autumn monitoring sessions, respectively), thus forest cover is not a limitation to wolf occurrence. In autumn, terrain ruggedness was the most important predictor of wolf occupancy; when terrain ruggedness index was >200 (moderately to highly rugged areas the probability of wolf occupancy declined steeply (Figure 2F). This can be explained by the fact that wolf’s main prey source in Romania, wild boar ((Promberger–Fürpass, 2004; Sin et al., 2019) was documented to prefer less fragmented areas with large beech forest stands in autumn and early winter (Fonseca, 2008). Additionally, red and roe deer, which are also important prey for wolves, are known to move after the rut season (November - December) to more marginal, less topographically-fragmented areas that provide connectivity to the lower winter grounds (Zweifel-Schielly et al., 2009). Proportion of forest was a positive predictor of lynx occupancy in autumn, which corroborates other studies that found that lynx occurrence in the Carpathians decreased at low levels of forest cover (Rozylowicz et al., 2010). Local road density was also an important predictor of lynx occupancy in winter, with lynx occupancy positively associated with road density (Figure 2A). While not heavily documented within the Lynx genus, other felid species have been known to use roads as travel corridors and for hunting and movement within their home range (Bailey, 1993; Bragin, 1986; Gordon & Stewart, 2007; Kerley et al., 2002; Matyushkin, 1977; Rabinowitz et al., 1987). Our results suggest that, in winter, Eurasian lynx are more likely to occupy areas with higher densities of local logging roads; these roads, which in our area are mostly unpaved, dirt roads, may provide easier access to resources within lynx home ranges due to decreased complexity of terrain and decreased snow depth/harder snowpack from vehicle travel. We did not observe this relationship with wildcat, however. Rather, there was a slightly negative relationship between density of local roads and wildcat occupancy in autumn (Figure 2E) which could be an artifact of body size; most documented examples of felids utilizing roads for movement within their home ranges was with larger bodied species (>11 kg). We also did not observe this relationship in winter however, this is likely an outcome of the importance of altitude for wildcat occupancy, which has a strong negative relationship (Figure 2B). Higher altitudes are associated with greater snow depth, and while lynx are well adapted to move in deep snow and altitude was not important for lynx occupancy, wildcats have physical limitations that make travel through deep snow more difficult. A study in Switzerland had similar findings whereby wildcats moved to areas free of snow in winter and spring and moved back to high elevations in summer (Mermod & Liberek, 2002). Similarly, in North America, the relationship between Canadian lynx (Lynx canadensis ) and bobcat (Lynx rufus ), is mediated by snowpack, with the distribution of the smaller-bodied species, the bobcat, being limited by snow depth at the northern edge of its range (Morin et al., 2020; Reed et al., 2017). Our results for marginal occupancy of lynx, wildcats, and wolf provide insights into both habitat selection and spatial relations for these elusive carnivores in Romania. Our results suggest lynx may use roads for movement and hunting, a practice common for other felids of similar body size, but not described in this species. Additionally, we provide further support for previous findings on habitat selection and occupancy for these three European terrestrial predators.