Spatial & evolutionary dynamics of H9N2

Spatial Phylodynamics of the Internal Gene Segments

Initial (and preliminary) results of a spatial phylogenetic analysis of the nearest relatives to the H7N9 outbreak in the internal gene segments. An outline description of the methodology is given below.

We gratefully acknowledge the authors, originating and submitting laboratories of the sequences from GISAID’s EpiFlu™ Database on which this analysis is based. A list of the sequences, their submitting labs and GISAID links is provided on this page. The background sequences are from NCBI/Genbank but we also acknowledge all the laboratories involved in collecting and curating these sequences.

A spatial reconstruction rendered in GoogleEarth. Each gene segment is denoted in a different colour. Use the time line at the top to animate the trees through time.

Most strongly supported connections between locations estimated jointly across all internal genes


A map showing the sampled locations of avian H9N2 virus with most highly supported connections (blue lines) with Bayes factor (BF) support > 3.0. The locations are shown at the capital or largest city within each country or province.

Joint analysis of avian host across internal gene segments


Joint reconstruction of the phylogeny and host for each of the internal gene segments of the genetically closest viruses in NCBI Genbank. Lineages are coloured by reconstructed host type. Human and environmental samples are allowed to estimate the original host and in the outbreak are identified as chicken in origin.


Data was collected by using building alignments and trees of available viruses that are genetically closely related to the 2013 outbreak in each of the 6 internal gene segments. Initial trees were used to identify the nearest clades to the outbreak including viruses predominently circulating in China and neighboring countries. Dates of isolation and country or province in the case of China were noted as well as the host animal.

A joint discrete phylogeographical analysis was performed using BEAST 1.7.5 allowing and independent tree for each gene segment. On each tree a discrete spatial model was used to reconstruct the movement of viruses amonst the locations with the model sharing a joing migration rate matrix (and BSSVS indicator matrix - Lemey et al 2009). Rate of nucleotide evolution amongst the different gene segments was modelled with a hierarchical phylogenetic model (HPM - Suchard et al 2003).