CHARACTER CONGRUENCE BETWEEN A MOLECULAR AND A MORPHOLOGICAL PHYLOGENY FOR DABBLING DUCKS (ANAS)

This paper presents an analysis of data from two published phylogenies of North American dabbling ducks (Anns), one based on mitochondrial D NA restriction fragments (Kessler and Avise, 1984, Syst. Zool. 33:370- 380) and one based on morphology (Livezey, 1991, Auk 108: 471-508). Al though the published phylogenies differ greatly in topology, most of t he apparent disagreement stems from the phenetic rooting method used i n the molecular study. Rooting by outgroup and using the same taxa fro m both data sets revealed that the most-parsimonious trees for the two data sets differ only in the placement of one clade. Furthermore, sev eral lines of evidence show that there is greater character congruence between the molecular and morphological data sets for these ducks tha n for most other taxa for which data are published: (1) an incongruenc e index (I-MF; Mickevich and Farris, 1981, Syst. Zool. 30:351-370) dem onstrates that only 1.35% of the character incongruence in the combine d data is due to incongruence between the data sets; (2) for each data set, trees only a few steps longer than the shortest tree match the m ost-parsimonious tree from the other data set; and (3) a way of applyi ng correlation coefficients to branch lengths reveals a high correlati on in branch lengths (r(BL) = 0.78) and total evolutionary changes (r( TE) = 0.68) between matching phylograms from the two data sets. These correlations also show that the two character types evolved at similar relative rates in any given lineage, even though there were real rate differences between lineages. The high degree of character congruence means that the discrepancy between the most-parsimonious trees are du e to insufficient data, not real disagreement. This situation warrants combining the data sets, and the resulting shortest cladogram is more strongly supported than is either of the shortest trees from the indi vidual data sets. The correspondence between the data sets validates t he soundness of the inferred trees, showing that they are a reliable b asis for classification and studies of character evolution.