Our results support the hypothesis that more homogenous landscapes increase the space use requirements of animals, which leads to increased probability of road crossings (after accounting for road density). In DK and SG, where landscape composition was more heterogeneous compared to NG, hare home ranges had a lower road density than in the surrounding area (whereas habitat structure was unchanged), suggesting that hares avoided establishing home ranges in areas with comparatively higher road densities. Conversely, in NG, where agricultural fields were substantially larger, hares established home ranges in areas with similar road density than the surrounding areas, likely because space use requirements prevented them from avoiding roads. This emphasizes that simple landscapes with seasonally varying resource availability increase space use requirements by hares compared to more complex landscapes (Ullmann et al. 2018), which might lead to increased energy expenditure due to increased movements and increased mortality risk due to frequent road crossings.
Broadly, our results suggest that hares generally avoid main roads (or show no selection), and that they select for minor roads during the vegetation growth seasons, when crops were of poor food quality for hares and cultivated fields generally avoided (Mayer et al. 2019; Mayer et al. 2018). This effect was more pronounced in areas with comparatively more homogenous habitat structure (NG and DK). Thus, road type, as well as habitat structure, are important to consider when investigating effects of roads on animal movement (Meisingset et al. 2013; Ouédraogo et al. 2020; Roedenbeck and Voser 2008). In SG, seasonal changes had little effect on the selection for proximity to roads (with close proximity generally being avoided), probably because hares had smaller home ranges compared to the other two study areas, providing sufficient resources year-round (Mayer et al. 2019; Ullmann et al. 2018). Moreover, as hares in SG avoided establishing home ranges in areas that intersected main roads, this variable had comparatively little effect on their habitat selection.
Apart from inhibiting movement and causing direct mortality via vehicle collisions (Hell et al. 2005), main roads might also be a disturbance factor for hares and other wildlife, due to high traffic volumes, noise and light pollution. Additionally, it is possible that hares can evaluate the relative risk in relation to road type, and that proximity to main roads did not offer foraging opportunities. Independent of the cause and mechanism of road avoidance, our findings indicate that main roads potentially create barrier effects, as shown in other studies on hares (Roedenbeck and Voser 2008) and other mammals (Carricondo-Sanchez et al. 2020; Prokopenko et al. 2017; Scrafford et al. 2018). This might have consequences for the long-term viability of animal populations, if roads create dispersal barriers leading to population fragmentation and genetic isolation (Dyer et al. 2002). However, this is likely not the case for hares and other small mammals that can adapt to areas with high road densities (Fey et al. 2016; Mayer and Sunde 2020).
The selection for proximity to minor roads during spring and pre-harvest by hares in DK and NG was likely related to the availability of high-quality forage, such as grasses and herbs along road verges (Jakobsson et al. 2018), whereas agricultural fields provide little high-quality food pre-harvest (Mayer et al. 2019). In line with the hypothesis that minor roads provide favorable foraging conditions, Roedenbeck and Voser (2008) showed that the density of unpaved field tracks had a positive effect on hare abundance in Switzerland. In addition, minor roads might also serve as movement corridors during spring and pre-harvest, when vegetation on agricultural fields is high. Similar to our study, Barbary macaques (Macaca sylvanus) showed seasonal selection for roads, avoiding roads in spring when natural food was abundant, and approaching roads in fall and winter, as natural food availability declined and food provisioning by tourists increased (Waterman et al. 2020). This shows that animals can seasonally exploit human infrastructure to meet their biological needs. The selection/avoidance of larger distances to both main and minor roads is harder to explain in biological terms, and was probably related to other factors, such as correlation with land cover variables at large distances from roads (which we did not detect due to comparatively few GPS positions being very far from roads).
Concerning habitat structure, hares generally selected for cropland and comparatively smaller fields, and avoided forests and built up areas. A discussion of these patterns can be found elsewhere (Mayer et al. 2018; Petrovan et al. 2013; Roedenbeck and Voser 2008; Tapper and Barnes 1986).
Main road crossings
Hares mainly crossed roads during their usual activity time, during the night, dusk and dawn, which coincided with typically lower traffic volumes (Mayer et al. 2021b), and lower human activity in general. Unsurprisingly, the road density within an individuals’ home range positively correlated with the number of road crossings in all study areas, which might affect the probability of animal road mortality. Importantly, within home range increases in road density led to much stronger increases in road crossings by hares in NG compared to hares in DK (this could not be assessed for SG due to a limited sample size). More generally, after accounting for road density, hares in the comparatively more homogenous landscapes of NG and DK crossed roads more often than in SG (agricultural field sizes were smallest in SG and largest in NG, with DK being intermediate, but having the highest road density). This suggests that animals in comparatively more heterogeneous landscapes are better able to adjust their habitat selection to avoid main roads than animals inhabiting simpler landscapes.
Consequently, simple landscapes might exacerbate potentially negative effects of main roads, if more road crossings translate to increased risk of road mortality. For example, in DK, where road density was highest, a third of the hare mortality was caused by collisions with a vehicle, providing some evidence that road mortality constitutes a significant mortality cause. In a Slovakian study, road-killed hares represented on average 15.5% of the annual hunting bag, and in some instances as much as 75% (Hell et al. 2005). However, Roedenbeck and Voser (2008) could not show an effect of road density on road mortality rates, a pattern that might be mitigated by fine-scale habitat selection by hares. Although increased road density might have negative consequences for some animal populations (Fahrig and Rytwinski 2009), this is not necessarily the case (Kroeger et al. 2021), and road density can actually have positive impacts on species abundance in some cases due to reduced predation (Rytwinski and Fahrig 2013).
Main road crossings also changed seasonally, with hares in NG crossing roads more often during spring. This pattern was related to increased movement distances, which was likely linked to the peak mating season during spring. Conversely, hares in DK crossed roads less often pre-harvest compared to the rest of the year, which was in contrast to our prediction that hares move more between fragments of shorter vegetation (such as roadsides) while being excluded from areas with higher vegetation (e.g. cereals) pre-harvest (Mayer et al. 2018). While hares likely still were excluded from areas with high vegetation pre-harvest, this probably created a barrier effect, leading to fewer road crossings. These findings further emphasize how the interplay of seasonal changes and habitat structure can affect animal space use in complex ways.
Overall, our findings suggest that the presence of high quality habitat (here measured as comparatively smaller field size), providing resources throughout the year, will allow animals to establish smaller home ranges that do not intersect main roads, and reduce the number of road crossings, consequently reducing the risk of road mortality. Nevertheless, hares in DK, were road density was highest, conducted comparatively more road crossings than in the two German areas, emphasizing that increasing road density will affect the likelihood of animals crossing roads and consequently, the risk of being killed by vehicles. Our findings further highlight that it is important to distinguish between road types, as different roads can have different impacts on animals (e.g. small roads providing foraging opportunities via roadside vegetation and large roads being consistently avoided with the potential to act as barriers). Finally, we showed that the selection or avoidance of roads can change seasonally, depending on the habitat structure and resource availability in the surrounding landscape. This emphasizes the need to account for landscape composition and seasonal variation when studying habitat selection in relation to roads.