Simple Summary
We examined human-pressure, prey and predator spatial associations in Northeast China Tiger and Leopard National Park. Camera and source-specific patrol observations informed models of animal space use. Four directional working hypotheses and two exploratory comparisons distinguished prey–predator, pressure–prey, direct pressure–predator, and prey-mediated relationships. Adding approximate coefficient uncertainty reduced the number of screened pathways, especially during snow. Higher prey use did not consistently coincide with higher predator use. The retained B–H–M analysis describes four snow-free zoning contexts. These patterns guide field checks of pressure sources and monitoring coverage; they do not establish which interventions would improve predator recovery.
Abstract
Human pressures may be associated with predator space use directly or through prey, with variation by season and zoning context. We combined cameras, source-specific patrol observations, and environmental covariates across 18,243 km grids in Northeast China Tiger and Leopard National Park. Hierarchical models estimated predator spatial-use probabilities and prey use intensities using 2024 monthly inputs. Tigers used cameras and patrols, whereas leopard states used cameras only. Spatial-error piecewise SEMs used 500 independently paired state draws and 400 conditional multivariate coefficient simulations per pair. No direct path and two single-prey chains passed the extended descriptive screening rule during snow, compared with four direct paths and one chain in snow-free conditions. The four directional hypotheses received limited, pressure- and species-specific support; seasonal and B–H–M comparisons remained exploratory. Negative net indirect estimates sometimes arose from component signs contrary to the proposed resource-depletion pathway. Opposing farmland–leopard estimates produced an uncertain total rather than evidence of exact cancellation. Six park-wide alternatives examined estimation, covariate, and network assumptions; the original B–H–M models described four snow-free contexts. These conditional patterns support field assessment rather than causal mediation or intervention prescriptions.