Predicting the global risk of chikungunya virus under climate change using ensemble species distribution models

Climate change is expanding vector-borne disease ranges, yet Chikungunya virus (CHIKV) risk projections remain limited by single-model uncertainty and lack of vector integration. CHIKV, transmitted by Ae. aegypti and Ae. albopictus, threatens 1.3 billion people globally, necessitating robust spatiotemporal risk assessment.

Using hierarchical ensemble modeling in Biomod2, we first projected vector distributions based on 19 bioclimatic variables and elevation, then integrated vector suitability as biological predictors for CHIKV under 16 CMIP6 scenarios (4 SSPs × 4 GCMs, 2021-2100). Eleven algorithms were evaluated and ensembled to minimize uncertainty.

Ensemble models achieved excellent performance (Ae. aegypti: AUC = 0.949, TSS = 0.773; Ae. albopictus: AUC = 0.934, TSS = 0.764; CHIKV: AUC = 0.909, TSS = 0.659). Ae. aegypti distribution was constrained by temperature stability (isothermality, temperature seasonality), while Ae. albopictus responded to both temperature and precipitation. CHIKV distribution was primarily vector-driven (84% explanatory power), further modulated mainly by the mean temperature of wettest quarter. Currently, 21.26% of global land area (139 countries) faces CHIKV risk, concentrated in tropical/subtropical zones. Future projections reveal northward expansion into temperate regions (northeastern North America, central Europe, East Asia), but extreme warming (SSP585) may contract tropical habitats via thermal stress.

Multi-model projections identify region-specific invasion risks, with previously unaffected temperate areas emerging as high-priority surveillance zones by 2100. These findings provide actionable risk maps for targeted vector control and preparedness strategies in 139 at-risk countries, particularly those lacking population immunity. Model heterogeneity underscores the necessity of ensemble approaches for climate-health policy planning.