Predicting spatial aspects of human–elephant conflict

Summary Human–elephant conflict (HEC) in Africa occurs wherever these two species coincide, and poses serious challenges to wildlife managers, local communities and elephants alike. Mitigation requires a detailed understanding of underlying patterns and processes. Although temporal patterns of HEC are relatively predictable, spatial variation has shown few universal trends, making it difficult to predict where conflict will take place. While this may be due to unpredictability in male elephant foraging behaviour (the male behaviour hypothesis) it may also be due to variations in the data resolution of earlier studies. This study tested the male behaviour and data resolution hypotheses using HEC data from a 1000‐km 2 unprotected elephant range adjacent to the Masai Mara National Reserve in Kenya. HEC incidents were divided into crop raiding and human deaths or injuries. Crop raiding was further subdivided into incidents involving only male elephants or family groups. A relatively fine‐resolution, systematic, grid‐based method was used to assign the locations of conflict incidents, and spatial relations with underlying variables were explored using correlation analysis and logistic regression. Crop raiding was clustered into distinct conflict zones. Both occurrence and intensity could be predicted on the basis of the area under cultivation and, for male elephant groups, proximity to major settlements. Conversely, incidents of elephant‐induced human injury and death were less predictable but were correlated with proximity to roads. A grid‐based geographical information system (GIS) with a 25‐km 2 resolution utilizing cost‐effective data sources, combined with simple statistical tools, was capable of identifying spatial predictors of HEC. At finer resolutions spatial autocorrelation compromised the analyses. Synthesis and applications . These results suggest that spatial correlates of HEC can be identified, regardless of the sex of the elephants involved. Moreover, the method described here is fully transferable to other sites for comparative analysis of HEC. Using these results to map vulnerability will enable the development and deployment of appropriate conflict mitigation strategies, such as guarding, early warning systems, barriers and deterrents. The utility of such methods and their strategic deployment should be assessed alongside alternative land‐use and livelihood strategies that limit cultivation within the elephant range.