Rate tests for phenotypic evolution using phylogenetically independent contrasts.

Most studies of evolutionary rates, except those concerning laboratory populations, use data from fossils to measure amounts of change occurring over time (e.g., Raup and Crick 1981; Cracraft 1984; Stanley 1985; Futuyma 1986; Bookstein 1987, 1988, 1989; Campbell and Day 1987; Gingerich 1987; Rosenzweig et al. 1987; Forey 1988; MacFadden 1988; McNamara 1990; Taylor and Larwood 1990; Fenster and Sorhannus 1991). For many phenotypic traits and for many kinds of organisms, however, fossil information is either sparse or unavailable (e.g., physiological or behavioral traits, such as maximal running speed or home range area [Garland 1983; Garland et al. 1988; Garland and Losos, in press; Janis, in press]; organisms such as Drosophila [Parsons 1987, 19911). Evolutionary-rate tests based on comparative data from living species may, therefore, greatly extend the types of traits and organisms that can be studied and compared (e.g., Coyne and Orr 1989; Lynch 1990; Vogl and Wagner 1990). Possible explanations for variation in evolutionary rates are discussed elsewhere (e.g., Futuyma 1986; Bookstein 1987, 1988, 1989; Campbell and Day 1987; Coyne and Orr 1989; Vrba 1989; Lynch 1990; Vogl and Wagner 1990; Fenster and Sorhannus 1991 ; Janis, in press) and will not be considered here. Recently, Lynch (1990) has described a test for comparing evolutionary rates within a clade with those predicted under the neutral theory of phenotypic evolution. His test requires estimates of divergence times (as may be derived from paleontological information andlor molecular clocks) and of the expected rate at which mutation adds new phenotypic variance to populations. Here, I consider a different but complementary problem, that of comparing rates of phenotypic evolution among clades. I show that phylogenetically independent contrasts (Felsenstein 1985) allow simple and statistically valid comparisons of evolutionary rates among monophyletic groups of extant (or extinct) species. The proposed test requires the genealogy (or cladogram) for all included species, as well as estimates of phylogenetic branch lengths, but it does not require information on mutation rates. As well, the clades to be compared do not need to be sister clades. Because organisms are related hierarchically, data for living species generally lack biological and hence statistical independence (e.g., Grafen 1989; Harvey and Pagel 1991; Harvey and Purvis 1991; Martins and Garland 1991). As noted by