Phylogenetic inference from molecular data

Reconstruction of phylogenetic trees of genes or populations is an important subject in the study of evolution. However, statistical methods for reconstructing phylogenetic relationships and for estimating the times of divergence between genes or populations are not well established. I therefore studied three problems concerning the methods for constructing phylogenetic trees and estimating the times of divergence between genes or populations by using molecular sequences and allele frequency data. They are as follows. (1) It is well known that the maximum parsimony method may choose an incorrect tree when the evolutionary rate varies from branch to branch extensively, even if an infinite number of nucleotides are examined. However, this statistical inconsistency of the parsimony method can occur even when the sequences evolve with a constant rate. I have therefore studied the reason and condition for this to happen by considering the cases of five to seven DNA sequences. (2) Molecular data are useful for estimating the divergence time of species. In practice, however, the evolutionary rate may vary with evolutionary lineage. Therefore, I have developed methods for testing the molecular clock and for estimating the times of divergence of species or species groups after elimination of excessively fast or slowly evolving sequences. These methods can also be applied to phylogenetic trees for different populations with allele frequency data. (3) Microsatellite DNA loci are highly polymorphic and thus useful for studying the evolutionary relationships of closely related populations or species. Some new genetic distance measures have been proposed specifically for these loci, but their efficiencies of constructing reliable phylogenetic trees are unclear. I have therefore investigated these efficiencies by computer simulation. The results have indicated that the new distance measures designed for microsatellite DNA loci have a large variance and are generally less efficient than traditional distance measures that are used for classical genetic markers such as electrophoretic and blood group loci.