Abstract
BACKGROUND: Prey species need to adjust their habitat selection either temporally or spatially to reduce predation risk or human disturbance. We tested the risk allocation hypothesis that poses that foraging is higher in low-risk habitats and during low-risk periods for impala (Aepyceros melampus) in the Serengeti Ecosystem. We expected impalas to select ranges that balance long-term predation risk and forage availability and adjust their habitat utilization within their ranges temporally to balance short-term predation risk and energy intake. METHODS: We modelled multi-scale resource selection for 36 GPS-tracked impala (2016-2018) to disentangle spatial and temporal trade-offs between forage acquisition and predation risk using resource selection functions at the landscape and within-home-range level. We contrasted resource selection inside and outside of Serengeti NP (SNP) for woody cover, forage availability (vegetation heterogeneity, NDVI) and risky places (proximity to water, terrain ruggedness). We also modelled responses to diel and lunar cycles in space use. RESULTS: Impalas attuned their resource selection both towards maximizing forage availability and minimizing risk exposure, especially at the landscape level. While home-ranges were generally placed in more homogeneous woodland close to water sources, impalas preferred to use more heterogeneous, open and rugged patches within their home-ranges. They adjusted their responses outside SNP by placing their home-ranges in less rugged terrain and farther from water sources. Impalas balanced foraging and risk temporally by adjusting their preferences to circalunar and diel patterns. CONCLUSIONS: Impala attuned their home-range placement and utilization according to the 'landscape of fear' hypothesis inhabiting a 'squeezed' landscape. Here, prey to a functionally diverse carnivore guild clearly balanced their space use between food acquisition and spatio-temporal risk avoidance. Humans, however, acted as a 'super-predator' eliciting a stronger behavioural response compared to natural predators. Following the risk allocation hypothesis, the presence of humans was of such an intensity and predictability that it changed the anti-predator responses in impala compared to the adjacent natural system with carnivores only. A functional response in habitat selection indicated that impala avoided long-term spatial risk in home-range placement, while within their home-ranges short-term temporal risk mimicked the diel patterns observed within the protected area with natural predators only.