Modeling the drivers of large herbivore distribution in human‐dominated southern African savannas

Abstract

AbstractAfrican savanna ecosystems are home to the world's richest large herbivore (LH) assemblages. However, its landscapes are changing faster than any other region on Earth due to human activities and natural events. Therefore, understanding the factors influencing the distribution of LH in these human‐dominated environments is crucial for decision‐making on wildlife and habitat management. We combined ecological aerial surveys, camera trap, and dung count data to investigate how ecological (habitat types, perennial rivers, and rainfall) and anthropogenic (human settlements and cattle grazing areas) factors influence the distribution of LH species in Limpopo National Park (LNP). We used generalized linear models fitting binary logistic regression models to distinguish 25‐km2 cells occupied (where the species were detected) by elephants (Loxodonta africana), buffalos (Syncerus caffer), zebras (Equus quagga), kudus (Tragelaphus strepsiceros), nyalas (Tragelaphus angasii), and impalas (Aepyceros melampus) from unoccupied regions (where the species were not detected) in the LNP landscape. We found that habitat type and rainfall were the most influential factors shaping the pattern of LH distribution in the LNP, except for elephants, whose prevalence was not associated with rainfall. The prevalence of zebras was positively associated with the proximity to perennial rivers, while kudus avoided these areas. While some species (zebras, kudus, and impalas) tended to avoid settlements, others (elephants, buffalos, and nyala) seemed attracted to settlements. Cattle grazing areas were the worst predictors of the distribution of all study species. Our results disclosed the role of ecological factors for the distribution of LH and showed that anthropogenic disturbances seemed to either (partially) prevent the occurrence of LH or show the potential for human–wildlife conflict risk in the study area. Therefore, the results highlight the need to investigate/quantify the potential human–wildlife conflict risk at finer spatial scales to improve future management.