Chimpanzee non-avoidance of hyper-proximity to humans

A species, especially when it is endangered and surrounded by anthropogenic elements, can be threatened by habitat fragmentation. Food resource availability in the species' usual or surrogate habitats may reinforce or decrease its use of certain areas. Our objective was to to determine the influence of natural and anthropogenic variables on spatial distribution of eastern chimpanzees (Pan troglodytes schweinfurthii). We first determined the home range of a wild chimpanzee community (hereafter Sebitoli Chimpanzee Community [SCC]) based on global positioning system (GPS) point locations (n = 2,586 direct observations and feces locations) collected between 2009 and 2013 in Sebitoli, Kibale National Park, Uganda. We described SCC home range using grid cells (2,959 cells of 100 m x 100 m) referencing environmental, spatial, and topographical variables (n = 15). We then determined diet and food species abundance within their territory (n = 63 vegetation plots and 18 months phenological survey) and predicted distribution of the 10 most foraged fruit species within SCC home range, using environmental, spatial, and topographical variables (n = 10) by applying a maximum entropy model (maxent). We then predicted chimpanzee presence as a function of environmental, spatial, and topographical variables (n = 15) using the maxent model and assessed its truthfulness with the kernel model, based only on GPS point locations. Chimpanzees in Sebitoli were mostly observed in a core area of 5.42 km2 within the 25-km2 home range. They did not avoid forest edges in contact with human populations, especially males who used larger core areas than females. Factors with the greatest positive impact on the chimpanzee distribution model (maxent area under the curve [AUC] = 0.907) were related to 3 different food resources: 1) proximity to forest edges considered as attractive because of presence of crops cultivated by local farmers; 2) proximity to a tarmac road that crosses the SCC home range with shoulders covered by attractive terrestrial herbaceous vegetation; and 3) presence of wild forest fruiting resources. The results obtained with the maxent predictive model, applied on a fine scale, were consistent with the kernel model, based on real observations. Therefore, such an approach may be recommended for surveys or action plans interested in sustainable management of wildlife in an anthropogenic environment and may be a useful tool to better understand parameters of a prefered zone