Modeling bonobo (Pan paniscus) occurrence in relation to bushmeat hunting, slash-and-burn agriculture, and timber harvest
Hickey, Jena Renee
MetadataShow full item record
Bushmeat hunting is anticipated to directly influence the distribution of bonobos through mortality of individuals and/or indirectly via bonobo avoidance of areas with higher hunting activity. Fragmentation of lowland rainforest is expected to facilitate hunter access to potential bonobo habitat, thereby reducing bonobo occurrence and reducing the effective habitat afforded by fragmented forests. We calculated four bonobo-specific fragmentation metrics based on remotely sensed data and fit univariate logistic regression models relating each metric to bonobo nest occurrence data collected in 2009. We found strong correlation between all fragmentation metrics and bonobo nest occurrence, with nests less likely to occur as fragmentation increased. We ranked the metrics based on predictive accuracy, with forest edge density (ED) ranking the highest. Using a maximum entropy modeling approach and 10 years of collaboratively compiled bonobo nest data, we built the first spatially explicit multivariate model predicting the rangewide distribution of bonobos. Of the rangewide environmental variables tested, the most important were distance from agriculture, distance from roads, ED, percent forest, and distance from river. Except percent forest, we view these predictors as proxies of hunting impact. Areas closer to agriculture are closer to human populations who tend to hunt in the surrounding forest. Roads and navigable rivers provide human access to areas that would otherwise likely be less vulnerable to hunting. ED distills the information of forest fragmentation from agriculture, logging, major rivers, and roads into a single metric that relates to hunter accessibility. At a finer scale, we fit bonobo site-occupancy models using three landscape-level metrics and three field-derived measures of human activity (machete cuts, traps, and roads). We used an information-theoretic approach to select the best fit models out of 65 potential combinations. ED occurred in all 13 of the top models, machete cuts occurred in 11, and both were negatively correlated with bonobo occupancy. Very likely, it is the poaching associated with these metrics that is the single common threat influencing bonobo occurrence. Our results indicate that forest fragmentation and hunting both negatively influence bonobo occupancy and both landscape- and local-level variables are important considerations in order to conserve this species.