A Generalized approach to genetic drift and its applications - I. An example of the evolution of ribosomal RNA genes

Abstract

ABSTRACTGenetic drift should be any random forces that change gene frequency stochastically. While the conventions including the WF (Wright-Fisher) model are about sampling errors between generations, the branching process (BC) has been known to be a more general framework. In addition to connecting ecology and population genetics, BC can address genetic drift in systems of multi-copy sequences that include transposons, viruses and multi-copy genes. Here, we apply this general concept to rRNA genes, which have a median of 300 copies per human individual (designated C ∼ 150 copies for haploid). By using BC, we determine the strength of genetic drift in rRNA genes to be ∼ 20 times higher than single-copy genes in human and mouse polymorphisms. The large increases in drift, likely due to the homogenizing forces (such as unbiased gene conversion) within individuals, reduce Ne* to < 10Ne, despite C ∼ 150 (Ne* and Ne being the effective population sizes of rRNA genes and single-copy genes respectively). Significantly, when the divergence between species is analyzed, some variants appear to experience extremely strong drift such that Ne* becomes smaller than Ne, as if C < 1. Therefore, the rapid divergence of rDNAs between mouse species can be easily accounted for by strong drift. Positive selection needs to be invoked only when the strong drift can be convincingly ruled out, as is between human and chimpanzee. In conclusion, the branching process reveals many stochastic forces in molecular evolution that are mistakenly attributed to selection in the conventional models.