Phylogenetic inference with not-so-rare mutations and tiny organisms

A common assumption in mathematical models of molecular evolution is that mutations are rare. One example is the boundary mutation assumption, which posits that mutations occur so infrequently that, by the time a new one arises, the previous mutation has already been fixed or lost in the population. This assumption, however, ignores recurrent mutations and can be problematic for highly diverse organisms such as bacteria and viruses. In this study, we challenge the assumption of infrequent mutation in phylogenetic inference. To do so, we compare two mutation models: one that incorporates recurrent mutations and another that allows only boundary mutations. Our results show that while tree topologies remain mostly unaffected, branch lengths and estimates of mutation bias and selection are substantially compromised. These patterns hold across both simulations and empirical case studies, including human immunodeficiency (HIV), hepatitis C (HCV), and influenza A (IAV) viruses. Overall, this study highlights the importance of accounting for recurrent mutations in phylogenetic analyses of highly diverse organisms, many of which have significant epidemiological and medical relevance.