Dynamic Risk Maps Predict Highly Pathogenic Avian Influenza Hotspots Across North America

Highly pathogenic avian influenza (HPAI) poses a major threat to North American food security, wildlife, and public health. It has caused unprecedented mortality in wild birds and poultry, with growing concern for pandemic emergence in humans. Here, we developed a temporally explicit ecological risk-mapping framework to identify where and when North American landscapes become permissive to HPAI circulation and cross-species exposure. We used monthly species distribution models for migratory wild bird species to estimate their probability of occurrence and seasonal aggregation in wetlands and agro-ecological regions. These dynamical bird layers were combined with poultry, cattle, and human population densities, land cover, and proximity to water bodies in a machine-learning model to generate fine-scale, seasonal maps of HPAI. Model performance was evaluated against an independent set of reported H5N1 outbreak detections from the 2021-2025 North American epidemic, yielding high accuracy and capturing nearly all detections within the highest-risk fraction of land area. The resulting risk surfaces revealed recurrent spillover corridors and hotspots in the Prairie Pothole Region, the Mississippi Alluvial Valley, the Delmarva Peninsula, the Georgia-South Carolina coast, and the Pacific Northwest, with risk strongly amplified where dense livestock populations intersect aquatic habitat. Rather than a single dominant carrier, we identified a rotating assemblage of migratory waterfowl species whose relative contributions to risk shift across season and region. Overall, our dynamic risk maps delineate small, mobile pressure zones that concentrate most observed H5N1 actively, providing actionable targets for surveillance, biosecurity, and early intervention at wildlife-livestock-human interface.