干旱区生态环境知识资源中心

Motivated by an inconclusive debate over implications of resource scarcity for violent conflict, and common reliance on national data and linear models, we investigate the relationship between socio-ecological vulnerability and armed conflict in global drylands on a subnational level. Our study emanates from a global typology of smallholder farmers’ vulnerability to environmental and socioeconomic stresses in drylands. This typology is composed of eight typical value combinations of variables indicating environmental scarcities, resource overuse, and poverty-related factors in a widely subnational spatial resolution. We investigate the relationships between the spatial distribution of these combinations, or vulnerability profiles, and geocoded armed conflicts, and find that conflicts are heterogeneously distributed according to these profiles. Four profiles distributed across low- and middle-income countries comprise all drylands conflicts. Comparing models for conflict incidence using logit regression and receiver operator characteristic analysis based on (1) the set of all seven indicators as independent variables and (2) a single, only vulnerability profile-based variable proves that the nonlinear typology-based variable is the better explanans for conflict incidence. Inspection of the profiles’ value combinations makes this understandable: A systematic explanation of conflict incidence and absence across all degrees of natural resource endowments is only reached through varying importance of poverty and resource overuse depending on the level of endowment. These are nonlinear interactions between the explaining variables. Conflict does not generally increase with resource scarcity or overuse. Comparison with conflict case studies showed both good agreement with our results and promise in expanding the set of indicators. Based on our findings and supporting literature, we argue that part of the debate over implications of resource scarcity for violent conflict in drylands may be resolved by acknowledging and accounting for nonlinear processes.