Patterns of Sequence Evolution and Implications for Parsimony Analysis of Chloroplast DNA

One of the advantages frequently cited for the use of DNA sequence data in phylogeny reconstruction is that explicit models of character evolution, based on our understanding of sequence evolution, may be incorporated into the analysis. In this way the method for recovering the pattern of evolutionary divergence can be tied more closely to the processes that produce the variation (Clegg and Zurawski, 1992; Simon et al., 1994; Moritz and Hillis, 1996). Within the context of parsimony analysis, it has been assumed that the reliability of information for phylogeny reconstruction can be inferred from estimates of substitution rate for different sites, with the most slowly evolving sites being the most reliable and, therefore, most heavily weighted (Farris, 1969, 1977; Felsenstein 1981a; Swofford et al. 1996). However, parsimony analyses based on chloroplast DNA (cpDNA) typically are conducted with equal weight given to all changes (e.g., Kim et al., 1992; Olmstead et al., 1993; Price and Palmer, 1993; Hoot et al., 1995; Johnson and Soltis, 1995; Gadek et al., 1996). The main reasons for this may be historical, stemming from a reluctance to weight characters in morphological cladistic analyses, or operational, due to the difficulty of implementing the more complicated analyses necessary for incorporating weights in phylogenetic analysis. In addition, there is a paucity of statistical analyses of cpDNA sequence substitution patterns (Clegg, 1993), which results in an inability to provide a firm basis for establishing weights.