Insights into Convergent Evolution From Studying Amino Acid Patterns in Independent Lineages of Birds

Vocal learning, the ability to imitate sounds, is a complex convergent trait crucial for spoken language and observed in a few independent lineages of mammals and birds. While convergences in gene expression have been found in vocal learning brain regions, amino acid convergences remain unclear. Here, we investigated whether avian vocal learning clades have amino acid convergences linked to their specialized trait. We developed a tool, Convergent Variant Finder, and applied it to an alignment of 48 species representing nearly all bird orders to identify convergent single amino acid variants among vocal learners and over 8,000 other polyphyletic species combinations. We discovered that the number of convergent variants was associated with the product of branch lengths of the most recent common ancestors of each species combination. The number of convergent variants in vocal learning clades did not exceed that of control species combinations. However, a subset of genes with vocal learner-specific convergent amino acid variants was enriched in the "learning" process, under positive selection, and significantly overlapped with gene sets for FOXP2 targets, singing-induced regulation in vocal learning nuclei, and differentially expressed in vocal learning nuclei. Moreover, we confirmed that the majority of convergent patterns in vocal learners were in the genomes of 363 species densely sampled across the avian tree. We propose that amino acid and nucleotide convergence accumulates at a steady state, with the rate proportional to divergence time. Selection associated with convergent traits, such as vocal learning, then likely acts on a subset of these changes.