Confirmation of a Portion of the Sibley-Ahlquist "Tapestry"

DNA-DNA hybridization was used to compare seven taxa from five avian orders, with an alligator as outgroup. Complete matrices ofT5oH andNPH (both sym- metrized and unsymmetrized) gave the same FITCH topology, which was supported in 100% of bootstrapped and jackknifed trees. The outgroup alligator rooted the tree between anser- iform-galliform and coliiform-strigiform-columbiform clades, and resolution within the latter favored a strigiform-columbiform association. In contrast, Tm gave differing and more poorly supported FITCH resolutions for deeper nodes because the distances were compressed due to greatly reduced NPHs. An F-ratio test between FITCH and KITSCH trees based on sym- metrized Jukes-Cantor-correctedT/s indicated significant rate variation among the lin- eages. Despite this result, the UPGMA algorithm applied to symmetrized data gave a topology identical to theToH andNPH FITCH trees, whether or not the outgroup alligator was included. However, phenograms calculated from unsymmetrized ds of all three indices as- sociated Bubo and Colius, as did the FITCH tree based on a completed matrix reconstructed from Sibley and Ahlquist's original data. Thus, our results support Sibley and Ahlquist's use ofT5oH to assess ordinal patterns in avian phylogeny, replicate a portion of their "tapestry" based on the same DNA-DNA hybridization technique, and show that for these taxa least- squares and phenetic algorithms generate much the same topology. Received 31 August 1993, accepted 21 November 1993. SIBLEY AND AHLQUIST'S (1990) summary pub- lication of their long series of avian DNA-DNA hybridization experiments provoked a number of critical reviews, many challenging the au- thors' assertion that numerous aspects of high- er-category phylogeny had been resolved. In particular, critics questioned whether the tech- nique has sufficient range or resolution, wheth- er the correct measure of thermal stability had been used, and whether the experimental de- sign (utilizing relatively few labeled taxa) was adequate to support Sibley and Ahlquist's claims for the structure of avian phylogeny and clas- sification (e.g. Krajewski 1991, O'Hara 1991, Rai- kow 1991, Lanyon 1992). Thus, the implications of Sibley and Ahlquist's work for systematic ornithology remain unclear pending resolution of these issues. We believe that much of the debate surround- ing Sibley and Ahlquist's work can be settled empirically. The traditional view of corrobo- ration of phylogenetic hypotheses requires comparison of results from independent char- acters. However, a central question for the de- bate about Sibley and Ahlquist's results remains whether their findings can be replicated with the same technique. While strict replication would entail data production and analysis fol- lowing protocols used by Sibley and Ahlquist, the criticism about the validity of these methods mandates a second level of replication in which Sibley and Ahlquist's supposed errors in data collection and analysis are avoided. If both lev- els of replication should yield the same results, one could at least conclude that Sibley and Ahlquist's trees are the ones given by DNA- DNA hybridization and are robust to depar- tures from ideal design and analysis. Differ- ences, however, would call Sibley and Ahl- quist's results into question with the very same technique. We view reconciliation of any dis- crepancies between trees produced by DNA- DNA hybridization and those produced by oth- er methods (e.g. DNA sequencing) as a separate issue.