Phylogenetic analysis: concepts and methods.

In a series of publications over the past several years, Cavalli-Sforza and Edwards have presented theoretical models and estimation procedures for the analysis of genetic relationships of human populations [1, 2, 3]. Their techniques of phylogenetic analysis have given a new perspective to the study of human variation and its origins. Moreover, their techniques are suitable for the study of microevolutionary relationships among populations of any species for which gene-frequency data are available. However, Cavalli-Sforza and Edwards only outlined the actual methodology involved. Recent studies concerning South American Indian tribes [4, 5] and domestic cattle breeds [6] have utilized their theoretical framework and their techniques; none of these papers discusses the methods thoroughly. Two of the models have been compared in separate studies of human populations [4, 7]; although they used different methodologies, both concluded that the two models led to similar results. One study [7] was based on examination of all possible relationship structures, but involved only the one set of data. The other study [4] based the conclusion on examination of only an undisclosed fraction of the possible relationships. Thus, to date, there has been no systematic comparison of the various models proposed by Cavalli-Sforza and Edwards [3]. Because suitable gene-frequency data are rapidly accumulating and many workers are interested in such analyses, detailed discussion and comparison of the models and methods are in order. Here we present (1) the way in which Cavalli-Sforza and Edwards's theoretical models and estimation procedures can be applied to the study of real populations, (2) comparisons of the results obtained with the various methods used for the various models, and (3) our interpretation of the results of such an analysis. We also review some basic concepts.