Evolutionary determinants of genome-wide nucleotide composition

Abstract

One of the long-standing mysteries of evolutionary genomics is the source of the wide phylogenetic diversity in genome nucleotide composition (G + C versus A + T), which must be a consequence of interspecific differences in mutation bias, the efficiency of selection for different nucleotides or a combination of the two. We demonstrate that although genomic G + C composition is strongly driven by mutation bias, it is also substantially modified by direct selection and/or as a by-product of biased gene conversion. Moreover, G + C composition at fourfold redundant sites is consistently elevated above the neutral expectation-more so than for any other class of sites.

Figure 1. Relationship between genome-wide nucleotide composition and the neutral expectation

The data are subdivided into three major groups of organisms. The diagonal dashed lines denote agreement with the neutral expectation, with points above the diagonal reflecting conditions in which there is selection for elevated G+C content. For reference, the lower panel provides isoclines of expected genome compositions under selection, with values of the composite parameter S = ϕN_es being equivalent to the ratio of the power of selection in favor of G+C content relative to the power of genetic drift. The neutral equilibrium expectation is calculated from Equation (1), and the observed G+C content is based on direct observation of genome contents. All data can be found in Supplementary Dataset 1: Tables 1–5.

Figure 2. Expected equilibrium levels of within-population nucleotide diversity scaled by the neutral expectation

Derived from Equation (3) in the text, with various strengths of selection (S) color coded as in Figure 1.