Is faster-X adaptation due to large-effect mutations? An empirical test of a new theory

A widespread observation in molecular evolution is that X-linked genes adapt faster than autosomal genes--a pattern known as "faster-X" adaptation. Yet the classical explanation for faster-X adaptation--that partially recessive beneficial mutations experience more efficient selection in hemizygous males--conflicts with theories of dominance, which predict that beneficial mutations should be partially dominant for fitness. Recently, a new theory for faster-X adaptation that does not invoke partial recessivity of beneficial mutations has been proposed, in which mutations with large phenotypic effects experience more positive selection on the X. Here, we tested this theory by estimating rates of adaptation of nonsynonymous mutations in three lineages with a well-documented faster-X: Drosophila melanogaster, Mus musculus, and Homo sapiens. We used three proxies for the phenotypic effects of mutations: amino-acid dissimilarity, sequence conservation, and gene age. As expected, all proxies for scaled phenotypic effects correlated negatively with measures of the efficiency of purifying selection and with adaptive substitution rates on both chromosome types. However, we found no evidence that faster-X adaptation was enriched among large-effect mutations, as predicted by the new theory. We discuss why this could be the case, including challenges in measuring scaled phenotypic effects, in modelling faster-X adaptation, and in estimating rates of adaptation using McDonald-Kreitman tests. Overall, our results highlight that faster-X adaptation is a major unsolved puzzle in evolutionary genetics.