Spatiotemporal Transmission Networks of Plague in China and Their Driving Factors: Historical Insights for the Sustainable Prevention and Control of Zoonotic Diseases

The Third Plague Pandemic caused severe mortality in China, yet its transmission routes, spatial patterns, and multi-factor driving mechanisms have not been comprehensively quantified. This study reconstructs plague transmission in China from 1772 to 1949 using a newly compiled transmission network dataset, machine learning models, and structural equation modeling. The results are as follows: (1) Spatially, plague transmission followed a three-tier pattern: initial occurrence in natural foci, diffusion along transport corridors, and secondary transmission in towns. The network was dominated by short-distance contagious spread, supplemented by long-distance jump dispersal. Across five regions, networks exhibited five topological forms (single-core radial, multi-core parallel, single-core fan-shaped, dual-core, and small-core with large periphery), with core nodes consistently located at historical transport hubs and administrative centers. (2) Lag effects varied by factor: floods peaked at a 3-year lag, droughts showed cumulative effects up to 5 years, temperature acted at a 2-year lag, and precipitation had the strongest effect at 1 year. Famine, war, and population density acted contemporaneously. (3) Factor importance ranked in descending order: temperature anomaly, flood, famine, population density, drought, war, and precipitation. All factors exhibited nonlinear threshold effects, including U-shaped (temperature), inverted U-shaped (precipitation and drought), threshold-jump (floods), J-shaped (famine), N-shaped (war), and shallow U-shaped (population density) patterns. (4) Structural equation modeling identified famine and drought as the most direct predictors (path coefficients 0.739 *** and 0.680 ***). Climatic factors acted mainly indirectly through the disaster chain of “climatic anomaly → frequent disasters → famine amplification → epidemic outbreak.” These findings provide three entry points for sustainable zoonotic disease prevention: hierarchical sentinel deployment based on network topology, climate–disaster-chain early warning with 1–5 year lead times, and tiered responses keyed to climate intervals. The results align with SDG 3 (good health and well-being) and SDG 11 (sustainable cities and communities).