Systematic underestimation of the age of selected alleles

A common interpretation of genome-wide selection scans is that the dispersal of anatomically modern humans out of Africa and into diverse environments led to a number of genetic adaptations. If so, patterns of polymorphism from non-African individuals should show the signature of adaptations dating to 40,000–100,000 Kya, coinciding with the main exodus from Africa. However, scans of polymorphism data from a few populations have yielded conflicting results about the chronology of local, population-specific adaptations. In particular, a number of papers report very recent ages for selected alleles in humans, which postdate the development of agriculture 10 Kya, and suggest that adaptive differences among human populations are much more recent. I present an analysis of simulations suggesting a downward bias in methods commonly used to estimate the age of selected alleles. These findings indicate that an estimate of a time to the most recent common ancestor (tMRCA) obtained using standard methods (used as a proxy for the age of an allele) of less than 10 Kya is consistent with an allele that actually became selected before the onset of agriculture and potentially as early as 50 Kya. These findings suggest that the genomic scans for selection may be consistent with selective pressures tied to the Out of Africa expansion of modern human populations.