Ecological interactions can strongly shape mammal communities. The “fear of predation” itself can influence spatial and temporal selection of predators and prey. Yet, not all co-occurrence patterns arise from interactions, as species may also co-occur or not given the same or differential habitat use. Here we describe the spatio-temporal dynamics of medium and large mammals in the Brazilian Atlantic Forest from a 2011 camera-trap survey. We tested how species occurrences relate to environmental and spatial covariates; whether variation among species activity patterns leads to temporal segregation; and whether the residual variation in species occurrences shows signals of interspecific interactions. For that we fitted a Hierarchical Model of Species Communities (HMSC) to partition their spatio-temporal distribution into components reflecting environmental preferences and interspecific interactions across space and time. For the 39 camera-trap stations we recorded three environmental covariates (fixed effects): altitude, distance to forest edge, and whether the camera-trap station was on road or on game trail. We also included as fixed effects the presence of domestic dogs and poachers. We included the locations of the camera-trap stations as a spatially explicit random effect, the day of observation as a temporally explicit random effect, and the station-day pair as a random effect to identify co-activity among the species.
Results/Conclusions
We had 951 occurrences of 18 native mammal species from 3473 camera-trap stations day. The fixed effects explained 87% of the variation, the random effects accounted for the remaining 13%. Our results indicate that the spatial segregation between predators and preys is mediated by the avoidance of preys for their predators, which supports the “landscape of fear” hypothesis. We found that the most important environmental factor explaining mammal’s community structure was whether the camera-trap stations were located on roads or game trails (25% of the explained variation), followed by the presence of poachers (22%). All species that were positively associated with roads (ocelot, jaguar and puma) were predators, and many of the species that were negatively associated with roads (e.g. spotted paca, armadillo, crab-eating raccoon and tayra) were preys of such predators. All species avoided poachers and domestic dogs, which further reinforces the “landscape of fear” hypothesis, as both domestic dogs and poachers are key predators from the viewpoint of the target species. Our results thus suggest that co-occurrence patterns among studied mammals reflect predominantly their similarities and differences on their activity and habitat use, and in particular their responses to human activities.