Re: Molecules and Morphology in Amniote Phylogeny
study reveals the underlying weaknesses of the total evidence approach. To the Editor:
Size of the Data Sets A continuing debate in the systematic literature centers on the analysis of multiple data sets in phylogeny It has been suggested that the phylogenetic signal of estimation. Advocates of the consensus approach (also larger data sets will overwhelm that of smaller data called ''taxonomic congruence'') suggest that different sets in a combined analysis (Miyamoto, 1985). Eernisse data sets should be analyzed separately, as long as they and Kluge (1993) counter that criticism by suggesting have different ''biological properties'' (e.g., Miyamoto that character covariation of the different data sets is and Fitch, 1995). At the other extreme, proponents of more important than the number of characters in each the total evidence approach (also called ''character condata set. However, if characters covary, then they will gruence'') advocate combining all characters in a single support the same topology. It is the characters in two analysis (e.g., Kluge, 1989). Intermediate positions also data sets that support conflicting topologies that are of have been taken (Bull et al., 1993; Chippindale and concern. In such cases, the parsimony criterion dictates Wiens, 1994). However, the possibility that some data that a data set with more characters supporting one sets inherently have greater potential for conveying topology will prevail over a data set with fewer characphylogenetic information than other data sets usually ters supporting an alternative topology. is not an issue that is addressed. Certainly, multiple
Although sequence data are most likely to overmorphological data sets have been combined in the whelm morphological data in a combined analysis, the past, long before the total evidence approach was introopposite situation occurred in Eernisse and Kluge's duced, without generating any controversy. It is the (1993) analysis. In that case, a large number of parsicombining of molecular and morphological data sets in mony-informative morphological characters supporta single analysis that has caused the greatest attention ing a bird-crocodilian grouping (Gauthier et al., 1988) because it implies that each class of data is equally efwere combined with a small number of parsimonyfective in conveying phylogenetic information. We disinformative characters supporting a bird-mammal agree and believe that there is a qualitative distinction grouping (Hedges et al., 1990). It was not surprising between molecular and morphological data that argues that the combined analysis also supported a birdfor their independent treatment in evolutionary studcrocodilian grouping. That outcome could have been ies (Nei