Empirical validation of the nearly neutral theory at divergence and population genomic scale using 144 placental mammals genomes.

By limiting the efficacy of selection, random drift is expected to play a major role in genome evolution. Formalizing this idea, the nearly-neutral theory predicts that the ratio of non-synonymous over synonymous polymorphism π N / π S within populations, and divergence d N / d S between species, should both correlate negatively with N e . This has previously been tested in mammals and other groups. However, most studies have focused on either d N / d S or on π N / π S , thus not addressing the problem across evolutionary scales. In addition, many studies at the macro scale have used life-history traits (LHT) as a proxy of N e , assuming that large-bodied organisms have lower N e than small-bodied species. However, this assumption itself has rarely been validated against more objective measures of N e , such as genetic diversity π S = 4 N e μ , in part because π S estimates are scarce. Here we propose an integrative test of the nearly-neutral predictions on 150 mammalian species, using 6000 orthologous genes, spanning the macro and the micro-evolutionary scale, using for the latter a measure of heterozygosity on each of the assembled diploid genomes. At the micro scale, we observe, for the first time in mammalian nuclear genomes, a relationship between π N / π S and π S . At the macro scale, we confirm the positive correlation between d N / d S and LHT but, more importantly, establish that LHT and d N / d S are correlated with π S , although weakly so. Together, these results provide the first global test of the nearly-neutral theory in mammals across time scales, suggesting all variables are correlated with a single hidden variable: N e .