Abstract
Human-wildlife conflict is a major conservation issue, particularly in lower income countries, where it affects marginalized people and leads to the extirpation of threatened species. Managers increasingly use fences to reduce this conflict but lack evidence on the effectiveness of these barriers, especially on whether this reduces the number of incidents or only shifts the problem elsewhere. We adapted an approach designed to measure how individual animals respond to barriers (barrier behavior analysis) to evaluate a human-wildlife conflict intervention. We used movement data from 20 GPS-collared elephants to assess the extent to which their behavior was influenced by community-managed beehive fences around Gorongosa National Park in Mozambique. We measured the number of times elephants were stopped by the fences and compared this with the number of times elephants were stopped by a natural barrier formed by a major river. Beehive fences blocked elephant movement in 69.3% of encounters, whereas the river barrier blocked 35.9%. Human-elephant conflict levels were lower after construction of the fence, dropping from a mean of 566 crop and infrastructure damage incidents per year in 2018 and 2019 to a mean of 117.5 incidents per year in 2020 and 2021. The mean distance of crop and infrastructure damage incidents from the park boundary increased from 0.98 to 1.97 km, and the number of human injuries and deaths increased from 1 to 8. Our results showed that community-run beehive fences can be effective barriers and reduce overall levels of human-elephant conflict in agricultural landscapes. They also showed how fencing can change the spatial pattern of conflict. This highlights the benefits of understanding how conflict mitigation methods change individual animal behavior and of measuring intervention effectiveness at a landscape scale.