Utility of Quantitative Genetic Parameters in Character Weighting

Abstract

-A fundamental assumption of any phylogenetic analysis is that are evolutionarily independent. This suggests the genetic correlation as a potentially useful weighting tool for quantitative characters. The quantitative genetic framework used to measure genetic correlations and heritabilities is first briefly described. Several hypothetical examples are then discussed where purely phenotypic values of lead to incorrect phylogenetic conclusions. These may occur either because are presumed to be genetically independent when they are not, or by misinterpreting environmentally induced variation as under genetic control. Genetic correlations and heritability estimates for eight from a quantitative genetic study of a population of Mexican ambystomatid salamanders are presented. On the of these data, gill lengths have low heritabilities and are rejected as useful characters. Head width, head length, and interorbital distance have very high genetic correlations, and are best considered as a single character. In the final section, some of the difficulties in interpreting quantitative genetic data are discussed; these include extrapolating from laboratory to field situations, variability in genetic parameters during phylogeny, and using intermediate genetic correlations as character weights. [Quantitative genetics; genetic correlation; character weighting; phylogeny reconstruction; Ambystomatidae.] The decision of whether or not to weight is fundamental to any systematic or geographic variation study. There are at least three solutions to the problem of character weighting: (1) use information and inferences drawn during the course of a systematic study to weight (termed a posteriori weighting); (2) use information from outside the study to evaluate (termed a priori because it is done before the systematic study and, thus, is independent of it); (3) avoid the issue entirely and count all as equal (still a form of character weighting, but all weights are arbitrarily set at one). If character weighting is attempted at all, a posteriori weighting seems to be generally preferred. Thus, Mayr's (1969: 217) claim that neither function nor conspicuousness nor any other known aspects I Based on presentation given as part of symposium on Character Weighting, Cladistics, and Classification, 30 December 1984, Annual Meeting of the Society of Systematic Zoology, Denver, Colorado. 2Present address: Department of Ecology and Evolutionary Biology, University of California, Irvine, California 92717. of a character gives it a priori a greater weight than other characters has been generally followed (Eldredge and Cracraft, 1980; but see Hecht and Edwards, 1976). Several authors have advocated a posteriori weighting, particularly Kluge and Farris (1969), Farris (1969) and Felsenstein (1981b), who argued that should be weighted by the inverse of their within-population standard deviation. In this paper I explore the potential use of parameters estimated from quantitative genetics as a priori weighting factors. Many workers have suggested that some understanding of the genetic of systematic would be useful in phylogenetic reconstruction (Inger, 1958; Farris, 1969; Mayr, 1969; Felsenstein, 1973, 1981a, b, 1985; Mickevich and Johnson, 1976; Arnold, 1983; Atchley, 1983; Simon, 1983). There appears to be general agreement that with a strong genetic basis are better indicators of relationship than those with an overriding environmental influence (Mayr, 1969; Felsenstein, 1973; Atchley, 1983). In addition, should be evolutionarily independent, since,