Using landscape connectivity to predict human-wildlife conflict

Abstract

Human-wildlife conflict has serious conservation consequences, both for populations of wildlife and for the people who live alongside them. Connectivity analyses can incorporate species-specific landscape resistance, and therefore have the potential to be used to understand where wildlife moves and causes conflict with people. We used circuit theory to develop connectivity models for the African savanna elephant in northwestern Botswana based on step-selection functions of movement data for 15 elephants and tested whether areas of high connectivity were correlated with occurrences of crop raiding. We used government records and field assessments of crop-raiding incidents between 2010 and 2016 to quantify conflict, which we predicted would correlate with landscape connectivity. The step-selection model revealed that linear boundaries such as rivers, fences, and dune crests were barriers to movement that impacted connectivity, while high vegetation index values and distance from villages were strong positive predictors of movement. Connectivity values were positively and significantly correlated with frequency of conflict incidents (p < 1.5e−06) over a six-year time span. However, connectivity had no predictive value for whether fields were raided (p < .54) or how frequently a field was raided (p < .77) during a single growing season. This study shows that connectivity may be a useful metric for predicting patterns of conflict occurrence over broad temporal scales, but may have limited predictive power at shorter time scales. It is crucial that conservation and conflict mitigation efforts recognize that methods may be appropriate for different purposes at different scales.