Ecosystem Interactions Underlie the Spread of Avian Influenza A Viruses with Pandemic Potential

Abstract

Despite evidence for avian influenza A virus (AIV) transmission between wild and domestic ecosystems, the roles of bird migration and poultry trade in the spread of viruses remain enigmatic. In this study, we integrate ecosystem interactions into a phylogeographic model to assess the contribution of wild and domestic hosts to AIV distribution and persistence. Analysis of globally sampled AIV datasets shows frequent two-way transmission between wild and domestic ecosystems. In general, viral flow from domestic to wild bird populations was restricted to within a geographic region. In contrast, spillover from wild to domestic populations occurred both within and between regions. Wild birds mediated long-distance dispersal at intercontinental scales whereas viral spread among poultry populations was a major driver of regional spread. Viral spread between poultry flocks frequently originated from persistent lineages circulating in regions of intensive poultry production. Our analysis of long-term surveillance data demonstrates that meaningful insights can be inferred from integrating ecosystem into phylogeographic reconstructions that may be consequential for pandemic preparedness and livestock protection.

Fig 1. Poultry production, global trade intensity and location of viral sampling.

(A) Map showing global chicken density (millions of chickens/km²) (B) Reported trade intensity of domestic poultry 1995 to 2011 between regions (C) Locations of available viral sequences within discrete regions defined by sampling effort, poultry production and sequence data.

Fig 2. Phylogenetic estimation of ecological interactions and geographic spread.

(A) H9-HA phylogenetic tree for global isolates where branches are coloured according to discrete geographic region and thick and thin lines indicate ancestral transitions between natural and agricultural ecosystems respectively (B) Graph showing the proportional Markov jump counts between ecosystems over time (C) Heat maps showing mean H9 migrations estimated to and from East China per year. Heat maps for other regions are shown in S4 Fig.

Fig 3. Inferred migration rates and patterns.

(A) Map showing statistically supported transitions between geographic regions by ecosystem. Line thickness corresponds to viral flow rates shown in Table 2 (thinnest<0.5; ≥0.5<1; ≥1<2;≥ 2 thickest). (B) Density distribution of statistically supported mean transition rates between ecosystems. *Domestic-to-domestic rates are significantly faster than domestic-to-wild (BF>100), wild-to-domestic (BF = 62.3), and wild-to-wild (BF = 39.4). (C) Statistically supported mean migration rates per MCMC step of wild-to-wild avian transitions versus domestic-to-domestic avian transitions.