Amino Acid Versus Nucleotide Characters for Phylogenetic Inference of the “Basal” Angiosperms

When nucleotide and amino acid characters for phylogenetic inference are compared, six factors need to be considered. In this study, an empirical example based on eight protein-coding genes from the “basal” angiosperms was used to investigate how these six factors contribute to the relative performance of nucleotide and amino acid characters for phylogenetic inference. Nucleotide characters were generally found to outperform amino acid characters with respect to the tree-based measures examined. Of the 167 clades resolved on the amino acid jackknife trees for each of the eight protein-coding genes analyzed separately, 22 conflicted with clades resolved by the nucleotide jackknife trees for the respective genes. Independent evidence supports the nucleotide resolution in 15 cases and the amino acid resolution in four. Of the 15 cases for which independent evidence supported nucleotide resolution, five could be explained by composite amino acid characters and/or convergent amino acid character states. Although the nucleotide characters generally outperformed the amino acid characters, a wide diversity in their relative performance is evident on comparing the eight genes relative to one another. Therefore, one cannot make blanket statements regarding which type of character coding is the better of the two.