Spatiotemporal analysis of plague risk in Tibet: Multi-source data-driven ensemble model development and validation

Plague, caused by the highly lethal bacterium Yersinia pestis , remains a critical public health threat. Tibet's unique alpine environment fosters sensitive ecosystems, yet plague dynamics there are understudied. This study reveals environmental drivers and distribution evolution of plague foci in Tibet. Using Tibet plague surveillance data (2000−2021), we integrated eight environmental variables (maximum temperature, precipitation, normalized difference vegetation index, land use and land cover, soil moisture, slope, aspect, and population spatial distribution). Ten algorithms via BIOMOD2 modeled plague risk across four phases (2000–2004, 2005–2009, 2010–2015, 2016–2021), validated with 2022–2023 data. We simulated risk distribution, calculated area/centroid changes, and elucidated spatiotemporal evolution. The ensemble model (EM) showed excellent external validation (2022: AUC = 0.98, TSS = 0.92, Kappa = 0.88; 2023: AUC = 0.99, TSS = 0.99, Kappa = 0.99). Key drivers were population, precipitation, max temperature, and NDVI. The EM revealed stage-wise centroid migration. High-risk areas concentrated in Lhasa, eastern Shigatse, and eastern Nagqu during 2000–2004, contracting spatially 2005–2009 with northeast centroid shift. Post-2010, the centroid shifted southeast then northwest, linked to anthropogenic activities and climate variations. Multi-model integration and spatiotemporal simulation revealed plague foci distribution and centroid evolution in Tibet, capturing human-climate impacts, providing a scientific basis for targeted prevention in high-risk areas. • First fine-scale plague risk analysis for Tibet's sensitive alpine ecosystems. • BIOMOD2 ensemble model (10 SDMs) achieves AUC ≥ 0.98, offers replicable zoonotic framework. • Key drivers POP, PRCP, TMAX, NDVI; plague foci migrate, link humans with climate.