Phylogeny of xantusiid lizards: concern for data and analysis.

Phylogenetic analyses of new DNA sequence data from the mitochondrial 16S rRNA gene in xantusiid lizards support the intergeneric relationships obtained previously (S.B. Hedges, R.L. Bezy, and L.B. Maxson, 1991, Mol. Biol. Evol. 8:767-780) with data from the 12S rRNA and cytochrome b (cyt b) genes. The total data set now includes 1028 alignable sites, 471 of which are variable and informative for the distance analyses and 281 of which are informative for the parsimony analyses. Crother and Presch (1993), Mol. Phylogenet. Evol. 1:289-294) claim that their reanalyses of our 12S rRNA and cyt b sequence data do not support a robust phylogeny for xantusiid lizards. However, that conclusion is not supported by their own analyses of the combined data from those two genes, which result in the same phylogenetic tree of xantusiid genera that we obtained in the original study with the same method (maximum parsimony). This result was unchanged when Crother and Presch eliminated sites containing insertions/deletions and ambiguities, and when transversions were weighted. The less robust result for the separate cyt b analyses, probably due to the smaller size of the data set, was already noted (Hedges et al., 1991). We believe that the best estimate of relationships, in this case, is obtained by combining the sequence data from these tightly linked mitochondrial genes. We also refute the criticisms by Crother and Presch of the neighbor-joining method. To correct for a higher rate of transitions in mitochondrial sequence data, they weight transversions more heavily (5x) than transitions. We present theoretical criticisms of this weighting method and advocate the use of scaled corrections available with distance methods. Crother and Presch also claim that a robust phylogeny of xantusiid lizards is not obtained when some morphological data (13 informative characters) are combined with the molecular data in a single analysis. However, there are serious problems with their morphological data and methods of analysis. We reevaluate the three pivotal morphological characters in their alternative phylogeny for xantusiid genera and demonstrate that none of the three provides unambiguous support for their alternative arrangement. Crother and Presch implement a new approach whereby the entire morphological data set is weighted equally to the molecular data set in a combined analysis, thus resulting in a very inflated weight assigned to each character in the small morphological data set. Using this rationale, each of the 13 informative morphological characters would receive a greater than million-fold weight if combined with the sequence data for the entire genome of these lizards.(ABSTRACT TRUNCATED AT 400 WORDS)