DNA characters and cladistics: the optimization of functional history

Abstract

Abstract Cladistic analysis of DNA characters must cope with the issue of functional constraints. We discuss functional considerations in systematically useful DNA data, exemplified by repetitive DNA sequences, ribosomal RNAs, and transcription factors. DNA data may be utilized as characters at several levels, such as (i) nucleotides, (ii) codons and amino acids, and (iii) strings of nucleotides. We consider a cladistic analysis of the rbcL gene in a sample of 40 land plants. Examination of codon usage in the rbcL gene reveals assymetries in codon frequencies suggesting that codon usage is influenced by the relative usage of codon neighbours in the genetic code. This pattern is presumably due to functional conservation of the rbcL gene. Furthermore, we show that branch support in cladograms based on analysis of nucleotides is almost entirely (97.5 per cent) based on silent substitutions or functionally labile amino acid replacements, suggesting considerable functional constraints on the rbcL gene. Hence, the cladograms are to a large extent constrained by functional history, which may or may not be entirely congruent with phylogeny. Branch support on cladograms based on string data is to a lesser extent (70 per cent) based on silent or functionally labile amino acid changes, suggesting that string data, on average, may be less influenced by whole-gene functional constraints. However, the correspondingly greater proportion of presumably functional changes in underlying nucleotides (28 per cent versus 2.5 per cent) indicates a greater likelihood that functional associations among particular nucleotides may bias tree construction.