Fine-scale movement decisions
The global model, including all covariates for the SSFs, revealed step selection for higher NDVI, lakes, roads, areas of verified livestock predation, and areas of participatory mapped livestock predation, and selection against rivers, steep slopes, elevation, boundaries, and areas of participatory mapped risk from hyenas (Table 3). Of these, selection against boundaries showed the strongest effect (β = 0.273, p < 0.01), yet there was marked individual variation in selection for all covariates (Figure 4). When comparing models containing combinations of ecological, infrastructure, and human experience/perception factors, the two models within Δ AIC ≤2 were the infrastructure-only model and the model containing infrastructure and human experience/perception covariates.
Seasonal SSFs comparing all covariates across the rainy and dry seasons showed that in the dry season hyenas exhibit an increase in fine-scale selection for elevation, lakes, and roads, and selection against boundaries (Figure 2b).
When comparing global models across land management types, Soysambu hyenas differed from LNNP hyenas in that they showed selection for roads (β = -0.19, p < 0.001). In contrast, LNNP hyenas differed from Soysambu in that they showed selection for lower NDVI (β = -0.077, p < 0.001) and areas of verified livestock predation (β = -0.552, p < 0.001) and against boundaries (β = 0.62, p < 0.001, Figure 3b). A case study on a single hyena that was known to frequently cross between the two protected areas (Supporting Information) showed selection against boundaries during the dry season (β = 0.186, p < 0.001) and for boundaries in the rainy season (β = -0.068, p < 0.01), as well as an increase in speed of movement (i.e., log of step length) in relation to boundary selection during the rainy season (β = 0.062, p < 0.001) as compared to the dry season (β = 0.035, p < 0.001).