Navigating multi-predator dynamic risk landscapes in anthropogenic environments: spatio-temporal plasticity in roe deer habitat selection

Context Non-lethal and lethal human activities such as hunting or agricultural practices strongly shape the movement and behaviour of wildlife populations. The recent return of natural predators in parts of the world complexifies the risk landscape for prey in anthropogenic ecosystems. A large-scale analysis is strongly needed to disentangle how hunted species manage to navigate the dynamic landscape of risk between human and natural predators to persist within human-dominated landscapes. Objectives We investigated how roe deer may flexibly adjust the selection for open and forest habitats to trade-off between hunting and natural predation risk. We used a large dataset across Europe with areas with and without risk from hunting and lynx, while controlling for non-lethal human disturbance using the Human Footprint Index. Methods We used integrated step selection analysis at diel and seasonal temporal scales to look at selection for open vs. forest habitats across sites differing in the occurrence and timing of the hunting season, presence of lynx, human disturbance while controlling for resource availability. Results We found only weak evidence that deer shifted habitat selection in presence of hunting, while lynx predation risk was associated with greater preference for open habitat at night. When both hunting and lynx co-occurred, deer selection for open and forest was driven by the presence of lynx rather than hunters. In absence of any lethal risk, deer habitat selection was associated with non-lethal human disturbance. Time of day and season were important determinants of deer habitat selection. Conclusions We found a consistent response to predation risk by lynx, while hunting induced a more variable response, probably because being less predictable. Adjusting habitat selection and activity at different temporal scales as a function of these relative risks allows ungulates to navigate in the dynamic multi-risk landscape of Europe’s anthropogenic environments. Our results set the empirical baseline which will help to disentangle habitat selection and activity shifts in future hunter-predator-prey studies, as the large predator guild in Europe further re-establishes, and human impact on natural habitats increases.