One step ahead to predict potential poaching hotspots: Modeling occupancy and detectability of poachers in a neotropical rainforest

Poaching is a common threat to vertebrates even within protected areas, yet it is difficult to predict and prevent due to a lack of information on its spatial distribution. We apply occupancy modeling to produce a spatially-explicit diagnostic of potential poaching areas in a protected area of the Atlantic Forest biodiversity hotspot, in Brazil. We used camera trapping along a 13-month period (April 2013 to May 2014) on 39 sampling sites selected using a systematic random design stratified by vegetation type. Using a single-species, single-season occupancy model, we evaluated seven covariates that might influence occupancy and detectability, to identify and compare sites selected by poachers. A total of 7020 trap-days was conducted during the study. Occupancy by poachers was higher near water resources and forest edges. Detectability of poachers was higher near water resources, forest edges and human settlements, in areas with higher abundance of game species, and in periods of higher lunar light intensity. Occupancy-based estimates of poaching matched well historical poaching records in the Reserve, and indicated that poaching pressure is not homogeneous across the Reserve; rather, there are clear poaching hotspots in areas with higher accessibility to poachers. Our results provide subsidies for increasing knowledge about this illegal practice, and points out for future strategies of conservation and management of game species. In addition, our methodological approach may be used in other Reserves to identify poaching hotspots, thus assisting managers in predicting and avoiding this illegal activity.