Agricultural and geographic factors shaped the North American 2015 highly pathogenic avian influenza H5N2 outbreak
(A) Maximum clade credibility tree representing the ancestral reconstruction of county groups across the evolutionary history of the outbreak. The ancestral reconstruction was based on an EBSP coalescent and strict molecular clock evolutionary model. Tree branches are colored based on the most probable county group of the descendant node. Thin gray node bars represent the 95% highest posterior density (HPD) of the node height (i.e., the time at which that ancestor is estimated to have existed). (B) Diffusion rate summary among county groups. County groups were defined based on state and composition of host type within the county. Counties with only turkey cases (turkey exclusive; T) were grouped separately from counties with at least one layer chicken case (mixed poultry; C). Arrows represent transition rates with strong support (Bayes factor > 10) with arrow thickness proportional to the magnitude of transition rate. (C) Conditional effect size of environmental and geographic covariates within the generalized linear model (GLM). Conditional effect size represents the effect size of the variable coefficient given inclusion in the GLM. Supported covariates (Bayes factor > 3) are bolded. Covariates are ordered by Bayes factor. The dashed gray line represents a conditional effect size of 0, signifying little impact of the covariate on viral dispersal.