Molecular Phylogeny and Genetic Diversity Analysis

Phylogenetic trees: Recent advances in statistical methods for phylogenetic reconstruction and genetic diversity analysis were discussed. At the present time, three different methods, i.e., neighbor-joining, parsimony, and likelihood methods, are commonly used for constructing phylogenetic trees from DNA or protein sequences. Neighbor-joining is simplest and produces reasonable trees in most cases as long as sufficient amounts of data are used. Likelihood is the most sophisticated method and requires rather stringent conditions. In practice, however, all three methods generally give similar trees particularly when bootstrap tests are used for examining the reliability of the trees obtained (Nei 1996). That is, phylogenetic clusters that are supported by bootstrap tests are almost always the same for the trees constructed by the three methods. The primary factor for determining the reliability of phylogenetic trees is not the statistical method used but the quality and amount of data used (Russo et al. 1996). It is recommended that a phylogenetic tree always be presented with proper branch lengths and bootstrap values. Linearized trees: In recent years it has become popular to estimate the time of divergence between two different groups of organisms (e.g., monocotyledons and dicotyledons). In this case it is important first to test the hypotheses of molecular clock and then eliminate the evolutionary lineages (or genes) that do not obey the molecular clock. Once these lineages (or genes) are eliminated, it is possible to construct a linearized tree under the assumption of a constant rate of evolution (Takezaki et al. 1995). Gene diversity analysis: The extent of genetic diversity within populations is usually measured by gene diversity or heterozygosity (H) for allele frequency data and by nucleotide diversity (7t) for nucleotide sequence data (Nei 1987). The statistical methods for measuring these quantities are well established. By contrast, there is a controversy about the measure of genetic diversity among subpopulations. The extent of this diversity for allele frequency data is measured by Wright's (1951) Fs-r, Cockerham's (1969) 9 orNei's (1973, 1977) G^ The first two measures depend on the assumption that all populations diverged at the same time and the effective population size is the same for all of them (Figure 1A). Pons and Petit (1995) extended Nei's mathematical formula, but conceptually their formulation depends on the same assumption as Wright's. However, the assumption used by Wright is never satisfied in nature; all natural populations have a phylogenetic structure and the effective population size varies from population to population (Figure 1B). The phylogenetic relationship is unique for each set of subpopulations. For this reason, Nei proposed a coefficient of genetic differentiation (Ggy), which is independent of the population history and measures the extent of genetic variation for a given set of existing populations. Therefore, this is applicable to any group of populations. In this approach, Gyr is defined by