In general, a composite likelihood is defined as a sum of log-likelihoods for related estimates. In MEGA4, the maximum composite likelihood is used for describing the sum of log-likelihoods for all pairwise distances in a distance matrix (Tamura et al. 2004) estimated by using the Tamura-Nei (1993) model (see related Tamura-Nei distance). Further information is in the Maximum Composite Likelihood Method.