Testing for equality of molecular evolutionary rates: A comparison between a relative-rate test and a likelihood ratio test

In this study, we examined the following: (1) the power of a relative-rate test (RRT) and a Likelihood ratio test (LRT), (2) the effect of the number of ingroup taxa on the power of the RRT and the LRT, and (3) the use of the molecular-rate tests and divergence times in discerning variability in nucleotide substitution rates between two clades. The results indicate that the RRT was more powerful than the LRT at detecting differences in total branch lengths when one set of descendant lineages of the ingroup have evolved, on average, faster than the other set since they diverged from the common ancestor. The LRT was more powerful than the RRT when the total branch length was substantially longer (or shorter) in half of the lineages within one set of the descendants of the common ancestor. Only in this case was the power of the LRT improved as the number of terminals increased, whereas for the RRT, the power increased somewhat in all the simulations when taxa were added. It was also concluded that in studies in which interclade rate differences are of interest, the results of the RRT and the LRT are difficult to interpret without information from the fossil record.