Phylogenetic utility of rapidly evolving DNA at high taxonomical levels: Contrasting matK, trnT-F, and rbcL in basal angiosperms

摘要

The prevailing view in molecular systematics is that relationships among distantly related taxa should be inferred using DNA segments with low rates of evolution. However, recent analyses of sequences from the rapidly evolving matK and trnT-trnF regions yielded well resolved and highly supported trees for early diverging angiosperms. We compare here the phylogenetic structure in matK, trnT-F, and rbcL datasets for the same 42, primarily basal angiosperm taxa. Phylogenetic trees based on matK or trnT-F are far more robust than those based on rbcL. Combined analysis of the rapidly evolving regions provides support for higher-level relationships stronger than that derived from analyses of multi-gene datasets of up to several fold the number of characters analyzed here. In addition to displaying a higher percentage of parsimony-informative characters, the average phylogenetic signal per informative character is significantly higher in the datasets from rapidly evolving DNA than in the more slowly evolving rbcL, as detected using resampling of identical numbers of parsimony-informative characters from the data matrices and subjecting different statistics for overall tree robustness and phylogenetic signal to significance tests. Automated via a set of scripts, the method used here should be easily extendable to comparisons of a broader range of genomic regions for varying taxon samplings. The relative performance of markers correlates not only with a lower mean homoplasy in matK and trnT-trnF compared to rbcL, but in particular correlates negatively with the percentage of sites exhibiting maximum or close to maximum homoplasy. A likelihood ratio test confirms that the rapidly evolving gene matK evolves significantly closer to neutrality, which may be one of the underlying factors for lower levels of overall homoplasy. Our results are in line with evidence from simulation studies suggesting that the deleterious effect of multiple hits in using rapidly evolving DNA at rather deep phylogenetic levels may have been overestimated, and thus promote extending the use of rapidly evolving DNA to deeper phylogenetic levels. (c) 2006 Elsevier Inc. All rights reserved.