Fig 1.
Sampled kill and random sites in Kanha Tiger Reserve, central India.
Kill sites for tiger are shown in orange and for leopard are shown in blue, and random sites are shown in grey.
Fig 2.
Maps showing model predictions of observed attack risk (top, A and B) and livestock owners’ perceptions of attack risk (bottom, C and D) for tiger (left, A and C) and leopard (right, B and D) on livestock. Maps represent relative probabilities of risk; absolute magnitudes of risk are not directly comparable.
Table 1.
Final model output for tiger and leopard observed attack risk for livestock.
Fig 3.
Livestock owners median perceived attack risk from tigers and leopards in different land-use categories.
Tiger risk is shown in orange; leopard risk is shown in blue. Sample size: nvillage-tiger = 19; nvillage-leopard = 44; nfield-tiger = 39; nfield-leopard = 50; nedge-tiger = 51; nedge-leopard = 40; nforest-tiger = 95; nforest-leopard = 60.
Fig 4.
Relationship between perceived and observed risk from tiger (A, orange) and leopard (B, blue) on livestock in each pixel across the Kanha landscape. Kruskal Wallis tests with Dunn’s post-hoc tests indicated that all perceived risk levels within each plot significantly differed from one another (P < 0.001).
Fig 5.
Variation in individual owners’ perceived risk within each land-use category.
Variation was calculated as the proportion of owners that deviated from the median perceived risk level. Larger values represent greater variation between owners’ responses and lower overall alignment in owners' perceptions of risk. Perceived risk from tigers is shown in orange and from leopards is shown in blue.