Avian Genomes Revisited: Hidden Genes Uncovered and the Rates versus Traits Paradox in Birds

International audienceAccording to current assemblies, avian genomes differ from those of the other lineages of amniotes in 1) containing a lower number of genes; 2) displaying a high stability of karyotype and recombination map; and 3) lacking any correlation between evolutionary rates (dN/dS) and life-history traits, unlike mammals and nonavian reptiles. We question the reality of the bird missing genes and investigate whether insufficient representation of bird gene content might have biased previous evolutionary analyses. Mining RNAseq data, we show that the vast majority of the genes missing from avian genome assemblies are actually present in most species of birds. These mainly correspond to the GC-rich fraction of the bird genome, which is the most difficult to sequence, assemble and annotate. With the inclusion of these genes in a phylogenomic analysis of high-quality alignments, we uncover a positive and significant correlation between the ratio of nonsynonymous to synonymous substitution rate (dN/dS) and life-history traits in Neoaves. We report a strong effect of GC-biased gene conversion on the dN/dS ratio in birds and a peculiar behavior of Palaeognathae (ostrich and allies) and Galloanserae (chickens, ducks and allies). Avian genomes do not contain fewer genes than mammals or nonavian reptiles. Previous analyses have overlooked $15% of the bird gene complement. GC-rich regions, which are the most difficult to access, are a key component of amniote genomes. They experience peculiar molecular processes and must be included for unbiased functional and comparative genomic analyses in birds