Mammalian predator and prey responses to recreation and land use across multiple scales provide limited support for the human shield hypothesis