The ongoing global spread of highly pathogenic avian influenza (HPAI) continues causing major impacts on poultry, wildlife, and public health. Among the most affected countries in Europe, Italy experienced three major epidemic waves in 2016–2023, caused by H5N8 and H5N1 subtypes. Yet, the underlying spatio-temporal dispersal dynamics and drivers of these epidemics remain to be elucidated. Here, we combined viral genome sequencing with continuous phylogeographic inference to reconstruct the evolutionary and spatio-temporal history of these epidemics. By combining genetic data with landscape, poultry-related variables, and wild bird abundances, we explored how the environment was associated with viral diffusion in space and time.

We identified distinct diffusion patterns and environmental associations across the three epidemics. The first two waves (2016–2017 and 2021–2022) were dominated by local poultry-to-poultry transmission in high-density farming areas, with agricultural areas being significantly associated with a relatively higher viral lineage dispersal velocity in 2021–2022. In contrast, the most recent wave (2022–2023) showed evidence of mid- to long-distance dispersal events positively associated with wetlands, waterbodies, and gull abundances, pointing to a larger role of wild birds in HPAI spatial dynamics. The estimated weighted diffusion coefficients and spatial wavefront distances revealed a shift from localised poultry-driven to long-distance spread, likely related to wild birds.

Our findings highlight a progressive shift from poultry-driven epidemics to more complex environmental transmission dynamics involving wild birds and natural habitats. These changes underscore the need to adapt surveillance and control strategies to an evolving and ecologically diverse viral landscape.