Spatio-temporal Bayesian evolutionary analysis

To perform spatial-temporal Bayesian evolutionary analysis, the database was sub-divided into three subsets including HA (1,878 records), NA (1,933 recodes), and PB2 (1,827 recodes) according to the sequence quality. The HA, NA, and PB2 gene sequences were aligned using ClustalW implemented in MEGA 6.06 (Tamura et al., 2013). The spatial-temporal Bayesian evolution analysis was performed using the BEAST 1.10.0 (Drummond and Rambaut, 2007; Suchard et al., 2018) with the following settings: the HKY85 nucleotide substitution model was used to describe the process of one nucleotide being substituted for another; the uncorrelated relaxed clock with a lognormal distribution, which indicated no a priori correlation between a lineage’s rate and that of its ancestor, was applied to model rate heterogenicity; and the expansion growth with growth rate tree prior was applied for the analysis based on the occurrence dynamics of the H7N9 virus. Other parameters were set to default values as recommended by the BEAST (Shi et al., 2017). A Markov Chain Monte Carlo (MCMC) chain was selected with 10,000,000 steps and sampled every 1,000 steps. The first 10% of samples were cut-off as burn-in by the TreeAnnotator program in the BEAST package. The generated maximum clade credibility (MCC) trees were viewed in Figtree 1.4.3.⁷ Spatio-temporal evolutionary analysis based on the MCC trees with continuous traits of HA, NA, and PB2 was performed using Spread D3 (Bielejec et al., 2016).