Fig 1.
Interaction of viral evolution, host ecology and the environment.
Viral genetic sequences contain information regarding virus evolution and diversity (A). Because their evolution occurs at a rapid pace, evolutionary patterns can be used in conjunction with location and species data to infer rates of viral dispersal among sampled geographic regions and host species. Many factors may influence observed virus transmission and spread. For instance, host factors (B) such as relatedness of host species and overlap of habitat distributions may be associated with viral transitions between host species. Further, environmental factors (C) may also play a role in the spatial diffusion of the virus. By incorporating viral, host and environmental information into computational models, the impact of host and environmental characteristics on virus spread can be estimated. Public domain map of North America was accessed from The World Factbook 2021, Central Intelligence Agency (https://www.cia.gov/the-world-factbook/static/09f14262b8d528e66631646c85e0edc0/north_america_pol.pdf).
Fig 2.
Discrete trait diffusion models of North American avian influenza using a sample of genetic sequences based on phylogenetic diversity.
Host models (left) are presented for combined internal gene segment (A), hemagglutinin gene subtype (B), and neuraminidase gene subtype (C) models. Source host species on the left of the chord diagrams contribute viral diversity to sink host species on the right. The magnitude of the viral transition rate is proportional to the width of the band, and statistically supported rates are darkened. Bands are colored by the host order of the source species (Charadriiformes–red; Anseriformes–blue). Similarly, geographic models (right) are summarized for combined internal gene segment (D), hemagglutinin gene subtype (E), and neuraminidase gene subtype (F) models. Arrow width is proportional to the magnitude of the transition rate, and only statistically supported rates are displayed. (AK–Alaska, AB–Alberta, BC–British Columbia, GT–Guatemala, MW–Midwest, NB–New Brunswick, NE–Northeast, NL–Newfoundland and Labrador, NRP–Northern Rockies and Plains, NS–Nova Scotia, NW–Northwest, OV–Ohio Valley, PE–Prince Edward Island, QC–Quebec, S–South, SE–Southeast, SON–Sonora, SW–Southwest, W–West). Public domain maps of United States, Canada, and Mexico states, and provinces were accessed from Wikimedia Commons, author Alex Covarrubias (https://commons.wikimedia.org/wiki/File:North_America_second_level_political_division.svg). Public domain map of Guatemala was accessed from The World Factbook 2021, Central Intelligence Agency (https://www.cia.gov/the-world-factbook/static/09f14262b8d528e66631646c85e0edc0/north_america_pol.pdf).
Fig 3.
Viral and host phylogenetic diversity of North American AIV.
(A) Estimation of the phylogenetic history of the PB2 AIV gene segment within North American wild birds. Color bands at the tips of the tree denote the host species distribution. This is contrasted with the phylogenetic history of the avian host species included in this analysis (B). Avian host phylogenetic history was summarized from 1,000 phylogenetic trees previously published by Jetz, et al. Light gray node bars represent the 95% highest posterior density of the node height. The redhead species was not categorized in the internal gene segment models and is therefore not included.
Table 1.
Summary of host and geographic variables used to inform the Bayesian discrete diffusion generalized linear model describing avian influenza virus dispersal among North American wild birds.
Fig 4.
Heat map of conditional coefficient values for host and region generalized linear models of North American avian influenza discrete trait diffusion models.
Conditional coefficient effect sizes are presented for each supported ecological variable across all gene segment and subtype datasets and both subsampling strategies (phylogenetic diversity analyzer (PDA) vs. stratified random sample). Only supported coefficients are displayed. Color darkness is proportional to the magnitude of the effect. Orange represents a negative correlation and blue represents a positive correlation.