The Human Shield in Time but Not in Space: Scale-Dependent Responses of Small Indian Civet-Prey Interactions to Anthropogenic Disturbance.

Despite growing evidence of the widespread impacts of human activities on carnivores and their prey, it remains unclear how different types and intensities of human disturbance reshape predator-prey interactions. In this study, we conducted a systematic camera-trapping survey on a threatened carnivore, the small Indian civet (Viverricula indica). This species forages on prey with contrasting diel patterns, including nocturnal rats and diurnal species such as Pallas's squirrel (Callosciurus erythraeus) and Chinese bamboo partridge (Bambusicola thoracica) in the southern Wuyi Mountains of southeastern China. Based on data from an extensive sampling effort (60,901 trap days at 180 camera stations), we used kernel density estimation and Pianka's index to examine whether and how different types and intensities of human activity (human presence, roads, and settlements), as well as diverse altitudes and different seasons, affect the spatiotemporal interactions between small Indian civets and their potential prey. We found that all studied species adjusted their activity patterns, either advancing or delaying their peaks, to achieve temporal segregation under varying types and intensities of human disturbance and different altitudes and seasons. At the temporal scale, interactions between small Indian civets and their potential prey supported the human shield hypothesis, suggesting that increased human disturbance provides diurnal prey with refuge from predation pressure. Conversely, at both spatial and spatiotemporal scales, higher levels of human disturbance increased the overlap between small Indian civets and their prey species. These findings highlight that human impacts on wildlife interactions are scale-dependent: temporal refuge for prey does not necessarily reduce spatial or spatiotemporal overlap, which may still increase encounter rates and predation risk. Because our sampling relied on ground-level cameras, our inferences are limited to terrestrial interactions; arboreal interactions remain unquantified and require combined ground-canopy sampling in future work. Effective conservation management thus requires considering these scale-dependent effects of human activities on wildlife interactions.