The nearly neutral theory contends that the interaction of drift and selection is important and occurs at various levels, including synonymous and nonsynonymous substitutions in protein coding regions and sequence turnover of regulatory elements. Recent progress of the theory is reviewed, and the interaction between drift and selection is suggested to differ at these different levels. Weak selective force on synonymous changes is stable, whereas its consequence on nonsynonymous changes depends on environmental factors. Selection on differentiation of regulatory elements is even more dependent on environmental factors than on amino acid changes. Of particular significance is the role of drift in the evolution of gene regulation that directly participates in morphological evolution. The range of near neutrality depends on the effective size of the population that is influenced by selected linked loci. In addition to the effective population size, molecular chaperones such as heat shock protein 90 have significant effects on the range of near neutrality.A lthough the neutral and the nearly neutral theories have now been recognized as realistic models to apply to evolutionary changes of genes and proteins, their significance for morphological evolution is still unsettled (1). During the last several years, ''slightly advantageous'' as well as ''slightly deleterious'' amino acid substitutions have been noted as important (2-4), and the significance of such weakly selected mutations for morphological evolution needs to be reconsidered. In addition, the molecular basis of gene regulation that is essential for development is being elucidated (5-7). In the evolution of gene regulation, interaction of selection and drift has been suggested to be important (8). The purpose of this article is to review such findings and to expand the concept of near-neutrality.
Synonymous vs. Nonsynonymous SubstitutionsA most efficient way to find how natural selection has worked is to examine the patterns of synonymous and nonsynonymous substitutions. The McDonald and Kreitman test (9) is most popular, and compares the numbers of nonsynonymous and synonymous substitutions within a population and between closely related species. These authors found that the number of fixed amino acid changes at the Adh locus between Drosophila sibling species is greater than the prediction under the neutral theory, and suggested that the fixed amino acid substitutions were selectively advantageous. Many subsequent studies showed that there were various patterns; i.e., most mitochondria data were characterized by an excess of nonsynonymous withinpopulation (polymorphic) changes, indicating weak negative selection against amino acid changes, whereas nuclear gene data showed more diverse patterns (for a review, see ref. 10).Recent studies along this line are attempts to examine collections of data of many loci. Smith and Eyre-Walker (11) did a statistical analysis using data of Drosophila simulans and Drosophila yakuba. Assuming three distinct classes of nonsy...