Roe deer at risk: teasing apart habitat selection and landscape constraints in risk exposure at multiple scales.

Non-consumptive effects of predators result from the cost of responses to perceived risk. Prey modulate risk exposure through flexible habitat selection at multiple scales which, in interaction with landscape constraints, determines their use of risky habitats. Identifying the relative contributions of landscape constraints and habitat selection to risk exposure is a critical first step towards a mechanistic understanding of non-consumptive effects. Here, we provide an integrative multi-scale study of roe deer spatial responses to variable hunting pressure along a landscape gradient of open habitats and dispersed refuges. Between low-risk and high-risk periods, we investigated shifts in 1) home-range location, 2) probability of using risky habitats (between-habitat scale) and 3) distance to the nearest refuge (within-habitat scale). For 2) and 3), we disentangled the contributions of landscape constraints and habitat selection to risky habitat use. We found that when risk was high, roe deer did not shift their home-range, but generally decreased their use of risky habitats, and sometimes reduced their distance to cover (particularly older animals). There was a functional response in between-habitat selection, with animals living in more open landscapes responding more than those living in landscapes with more refuges. However, individuals living in more open landscapes avoided open risky habitat less. Finally, we found that among-individual variation in risk exposure was generally, but not always, minimized by habitat selection across gradients of landscape constraints. To our knowledge, this is the first study simultaneously documenting prey responses to risk at the within-habitat, between-habitat and home-range scales. Our results support the view that between-habitat selection acts at a higher hierarchical level than within-habitat selection, and provide a framework for disentangling the contributions of habitat selection and landscape constraints to risk exposure. Selection cannot always compensate for landscape constraints, indicating a need for further investigation of the processes underlying habitat selection. [ABSTRACT FROM AUTHOR]