Generation time and the rate of molecular evolution.

Abstract

The importance of generation time in determining the rate of molecular evolution has long been at issue. Resolution of this question has important consequences both for the validity of the neutral theory (as formulated by Kimura 1983) and for the use of molecular data in systematics. Wu and Li (1985) performed two tests based on comparative analyses of the DNA sequences of a variety of mammalian genes, the results of which, they argue, show clear evidence of a generation-time effect. In the first “relative rate” test the numbers of nucleotide substitutions were compared between homologous genes in humans (a long generation-time species) and “reference-species” with the numbers in the same genes between mice or rats (short generation-time species) and the same reference species. The reference species used were dogs, rabbits, pigs, cows, and goats. The results (summarized in table 1) show that for all categories of nucleotide sites the number of substitutions per site, averaged over 12 genes, was greater in the mousereference comparison than in the human-reference comparison. It is argued that this shows that there has been a faster rate of substitution in the shorter-generation-time rodent lineage, than in the longer-generation-time human lineage. In conducting the test, Wu and Li assumed that rodents and humans are more closely related to each other than either is, on average, to the reference species. This pattern of relationship (fig. 1 a) implies the following approximate numbers of substitutions (at the fourfold degenerate sites) in the three branches of the phylogenetic tree: reference species, 0.2; humans, 0.2; rodents, 0.4. Since the number of substitutions in the human and reference-species branches are approximately the same but the reference-species branch is longer than the human branch, this implies that the order of substitution rates among the three lineages has been: rodents > humans > reference. (The fourfold degenerate sites are used here to illustrate the point, which applies equally to the other kinds of sites.) The relative order of generation times among the three lineages, on the other hand, is: rodents < reference < human, there being approximately an order of magnitude difference between each adjacent pair. Thus, when all three lineages under this phylogenetic scheme are considered, there appears to be variation in substitution rate, but this variation is not adequately explained by variation in generation time. Wu and Li (1985) suggest that this test does not depend on any knowledge of the divergence times of the species. This is true of absolute divergence times, but the test depends critically on a knowledge of relative divergence times. The phylogeny in figure la is by no means firmly supported by the fossil record. Most of the major mammalian divergences occurred during the late Cretaceous and the Paleocene. However, the mammalian fossil record during this period is poor, and the relative order of divergences is not clear. An alternative pattern of relationship among the species (figure lb), one that places humans closer to the reference species than to rodents, can explain the variation in substitution rate among the three branches as resulting from variation in the lengths of the branches. In other words, if the data are analyzed in the context of a different, equally feasible, phylogenetic scheme, they can be taken as evidence that substitutions have occurred at a constant rate that is independent of generation time.