Livestock activity shifts large herbivore temporal distributions to their crepuscular edges

Wildlife species are transitioning to greater crepuscular and nocturnal activity in response to high human densities. This plasticity in temporal niches may partially mitigate the impacts of human activity but may also result in underestimating human effects on species foraging, predator-prey relationships, and community level interactions. We deployed remote cameras to characterize shifts in herbivore diel activity in protected habitat vs pastoralist landscapes. We then compared species traits including body mass, dietary preferences, and behavioral characteristics as potential predictors of species sensitivity to livestock. Our data capture a significant temporal shift away from core cattle activity for nearly every herbivore species in our study, leading to more crepuscular activity patterns. As livestock were primarily diurnal and predators primarily nocturnal in pastoralist habitat, species that decreased their overlap with livestock were more likely to increase their overlap with potential predators. Other than species’ typical daytime activity levels, we found no evidence that any particular trait significantly predicted temporal shifts in response to livestock. Instead, species generally trended toward greater activity levels at dawn, suggesting that cattle have a homogenizing effect on community-wide activity patterns. Our findings highlight how cohabitation with livestock can profoundly alter the temporal niches of wild herbivores. Shifts in diel activity patterns may reduce herbivore foraging time or efficiency and potentially have cascading shifts on predator-prey dynamics. Given that species traits could not predict responses to livestock, our analysis suggests that conservation strategies should consider each species separately when designing interventions for wildlife management.

Methods
Camera-Trap Data Collection

This study was a comparative assessment of species activity patterns in the presence (Mara conservancies) and absence (Serengeti National Park) of cattle. Species activity patterns were inferred from camera image data. Serengeti camera data was previously published on dataDryad (doi: 10.1038/sdata.2015.26). For Mara conservancy data, we deployed unpaired, unbaited remote cameras along systematic grids in accrodance with Snapshot Serengeti methodology. In the conservancies, a total of 63 Cuddeback cameras (Models 1347 and 1279; Cuddeback Inc., Green Bay, WI, USA) were spaced at ~1.15 km intervals. Enonkishu housed 21 cameras from Nov. 2018 to 2022, and the grid was expanded into Ol Choro and Lemek conservancies in Aug. 2021. For optimal species detection and camera efficiency, cameras captured 3 images per trigger during the day and 1 image per trigger at night. Each trigger was considered an individual “capture event”, remaining dormant for 1-minute in between captures. To insure temporal independence. capture events were filtered to remove repeats of the same species at the same camera within 30-minute intervals.