Abstract
Abstract
Human-wildlife conflict (HWC) is a global challenge for biodiversity conservation, food security, and human well-being. Effective management towards human-wildlife coexistence, recognized as a priority in Target 4 of the Kunming–Montreal Global Biodiversity Framework, requires understanding the drivers of HWC. However, the strong influence of local human and ecological context means that most HWC research focuses on individual case studies, limiting transferability of patterns across contexts. Here, we link local contexts and transferable drivers by testing whether HWC exhibits an intermediate level of organization, in which local predictor profiles are structured by broader ecological, geographic, and taxonomic contexts. After compiling ∼11,000 conflict incidents, we construct and compare random forest models across 29 mammalian HWC scenarios spanning diverse species, conflict types, biomes, and continents to identify contexts in which predictor profiles converge. While predictors varied among individual conflicts, HWC scenarios were organized by shared context: conflicts occurring within the same biome tended to have similar predictor profiles, with continent and taxonomic group providing additional structure. Proximity to protected areas was the most consistently important predictor of HWC globally. The likelihood of conflict increased with human population density across nearly all scales and contexts, while distance to roads was also an important predictor in many contexts. Together, these results demonstrate an intermediate level of organization in HWC that reconciles case-specific variation with broader patterns across biomes, continents, and taxonomic groups. This framework provides a basis for scaling mitigation strategies beyond individual case studies while avoiding one-size-fits-all approaches.