Molecular Evolution in the Drosophila Melanogaster Species Subgroup: Frequent Parameter Fluctuations on the Timescale of Molecular Divergence

Article excerpt


Although mutation, genetic drift, and natural selection are well established as determinants of genome evolution, the importance (frequency and magnitude) of parameter fluctuations in molecular evolution is less understood. DNA sequence comparisons among closely related species allow specific substitutions to be assigned to lineages on a phylogenetic tree. In this study, we compare patterns of codon usage and protein evolution in 22 genes (>11,000 codons) among Drosophila melanogaster and five relatives within the D. melanogaster subgroup. We assign changes to eight lineages using a maximum-likelihood approach to infer ancestral states. Uncertainly in ancestral reconstructions is taken into account, at least to some extent, by weighting reconstructions by their posterior probabilities. Four of the eight lineages show potentially genomewide departures from equilibrium synonymous codon usage; three are decreasing and one is increasing in major codon usage. Several of these departures are consistent with lineage-specific changes in selection intensity (selection coefficients scaled to effective population size) at silent sites. Intron base composition and rates and patterns of protein evolution are also heterogeneous among these lineages. The magnitude of forces governing silent, intron, and protein evolution appears to have varied frequently, and in a lineage-specific manner, within the D. melanogaster subgroup.

UNDERSTANDING the forces governing the origins and evolutionary fates of DNA mutations is central to the study of molecular evolution. A great deal of attention has been focused on determining the relative contributions of genetic drift and natural selection to patterns of divergence among genomes (reviewed in OHTA 2002). However, the magnitude, timescale, and genomic breadth of fluctuations in molecular evolutionary forces remain to be studied systematically. Such knowledge is critical for modeling the causes of molecular evolution and is necessary for designing tests of adaptive and deleterious evolution and methods for phylogenetic inference and ancestral state reconstruction.

Determinants of molecular evolution include mutation rates and patterns, effective population sizes, rates of recombination and biased gene conversion, and the fitness effects of mutations. Strict constancy of all these factors is implausible. However, the timescale of parameter fluctuations determines their relevance to molecular evolution; variability in evolutionary forces cause heterogeneous substitution patterns if parameters changes occur on a similar timescale as molecular evolution (GiLLESPIE 1993, 1994; CUTLER 2000a,b). Although theoretical concerns suggest that appropriately scaled parameter fluctuations should not be common, numerous studies have invoked nonstationarity to explain variable rates of protein evolution at particular loci or in specific lineages. These include fluctuations in neutral mutation rates (FiTCH and MARKOWITZ 1970; FITCH 1971; TAKAHATA 1987), effective population sizes (e.g., OHTA 1987, 1993; MORAN 1996; JOHNSON and SEGER 2001; WOOLFIT and BROMHAM 2003), and distributions of fitness effects (e.g., GILLESPIE 1991; EANES et al. 1993; MESSIER and STEWART 1997; ZHANG et al 2002a).

Synonymous codon usage appears to be particularly amenable to microevolutionary analysis of parameter fluctuations. A relatively simple model of "major codon preference" (MCP) that incorporates mutation pressure, genetic drift, and weak selection favoring translationally preferred codons (Li 1987; BULMER 1991) is supported by both biochemical and population genetic evidence (!KEMURA 1985; reviewed in ANDERSSON and KURLAND 1990; SHARP et al. 1995; AKASHI 2001; DURET 2002). A similar scenario (with variation in parameter values) appears to apply across synonymous families and protein-coding genes within a given genome. MCP predicts fitness classes of silent DNA mutations (translationally preferred and unpreferred changes) (AKASHI 1995) and comparisons of evolutionary patterns between the classes can identify weak selection as well as departures from equilibrium and their causes (AKASHI 1996). …