WHEN MOLECULES AND MORPHOLOGY CLASH - A PHYLOGENETIC ANALYSIS OF THE NORTH-AMERICAN AMBYSTOMATID SALAMANDERS (CAUDATA, AMBYSTOMATIDAE)

Extremely conflicting phylogenetic conclusions for the same taxa derived from different data sets are rare. When conflicts do occur, they may result from homoplasy as well as nonindependence of characters. We present an analysis of the phylogenetic relationships of the salamander family Ambystomatidae, based on variation at 26 allozyme loci. Our results conflict strongly with those derived from Kraus's (1988, Syst. Zool. 37:106-141) morphological analysis of the same taxa. The major conflict involves five species traditionally placed in the Ambystoma subgenus "Linguaelapsus." This group, and the clades within it, are each diagnosed by many morphological synapomorphies, providing unambiguous evidence for the monophyly of the subgenus. However, our allozyme data provide equally compelling evidence that "Linguaelapsus" is polyphyletic. We jackknifed and bootstrapped our allozyme data in a likelihood and a cladistic parsimony framework, respectively; all analyses point to the same conclusions. Subdividing the allozyme data into rapidly and slowly evolving loci, and analyzing each separately by maximum likelihood, does not change our conclusions. Our results demonstrate that even apparently "well-corroborated" nodes of a cladogram may be suspect if the assumption of character independence is violated.