Linking hosts, landscapes, and climate to advance zoonotic arbovirus forecasting

Forecasting zoonotic mosquito-borne viruses remains a critical challenge because transmission depends on dynamic, multitrophic interactions among vectors, hosts, pathogens, and the environment. Here, we integrate long-term sentinel chicken surveillance across much of Florida with environmental data to build a predictive framework for eastern equine encephalitis virus (EEEV), a zoonotic mosquito borne disease of concern to human and equine health. Our models captured both environmental drivers and latent spatiotemporal structure, achieving strong predictive accuracy. Models revealed strong nonlinear effects of moderate precipitation a year prior to sampling and higher minimum temperature 1 month prior to sampling, as well as moderate and high percentages of forest and wetland cover on increased EEEV seroconversion. Retrospective predictions showed shifting virus activity across regions, consistent with Culiseta melanura mosquito vector ecology. We also calculated associations between EEEV and abundance estimates for key bird species that are suspected virus hosts using eBird data. Seasonal shifts among migratory and resident birds with predicted virus activity for key species suspected of being important EEEV hosts suggests spring migrants play a role in amplification, residents in summer persistence, and overwintering groups as potential reservoirs. These results demonstrate ecological forecasting of arboviruses is feasible at management-relevant scales, with broad potential to extend to other arbovirus systems. By integrating traditional surveillance with community science, our framework advances both predictive capacity and ecological understanding of zoonotic arboviruses.