A composite likelihood is defined as a sum of related log-likelihoods. Since all pairwise distances in a distance matrix have correlations due to the phylogenetic relationships among the sequences, the sum of their log-likelihoods is a composite likelihood. Tamura et al. (2004) showed that pairwise distances and the related substitution parameters are accurately estimated by maximizing the composite likelihood. They also found that, unlike the cases of ordinary independent estimation of each pairwise distance, a complicated model had virtually no disadvantage in the composite likelihood method for phylogenetic analyses. Therefore, only the Tamura-Nei (1993) model is available for this method in MEGA4 (see related Tamura-Nei distance). It assumes equality of substitution pattern among lineages and of substitution rates among sites (see related gamma model and heterogeneous patterns).