Abstract
The global-scale epidemiology and genome-wide evolutionary dynamics of influenza B remain poorly understood compared with influenza A viruses. We compiled a spatio-temporally comprehensive dataset of influenza B viruses, comprising over 2,500 genomes sampled worldwide between 1987 and 2015, including 382 newly-sequenced genomes that fill substantial gaps in previous molecular surveillance studies. Our contributed data increase the number of available influenza B virus genomes in Europe, Africa and Central Asia, improving the global context to study influenza B viruses. We reveal Yamagata-lineage diversity results from co-circulation of two antigenically-distinct groups that also segregate genetically across the entire genome, without evidence of intra-lineage reassortment. In contrast, Victoria-lineage diversity stems from geographic segregation of different genetic clades, with variability in the degree of geographic spread among clades. Differences between the lineages are reflected in their antigenic dynamics, as Yamagata-lineage viruses show alternating dominance between antigenic groups, while Victoria-lineage viruses show antigenic drift of a single lineage. Structural mapping of amino acid substitutions on trunk branches of influenza B gene phylogenies further supports these antigenic differences and highlights two potential mechanisms of adaptation for polymerase activity. Our study provides new insights into the epidemiological and molecular processes shaping influenza B virus evolution globally.
Author summary
Influenza B viruses cause roughly one third of the global influenza disease burden. However, many important questions regarding the global-scale molecular epidemiology and evolutionary dynamics of influenza B virus have yet to be comprehensively addressed compared to influenza A virus. This is in part due to limited globally-sampled genomic data. We improved the availability of influenza B virus data by sequencing over 350 full genomes, fillings gaps from under-sampled regions by as much as 12-fold. Using a dataset of over 2,500 influenza B virus genomes, we show major differences in the genome-wide evolution, molecular adaptation, and geographic spread between the two major influenza B lineages. These findings have implications for vaccine design and improve our understanding of 
influenza virus evolution.