A new method for locating changes in a tree reveals distinct nucleotide polymorphism vs divergence patterns in mouse mitochondrial control region

A new, model-based method was devised to locate nucleotide changes in a giv en phylogenetic tree. For each site, the posterior probability of any possi ble change in each branch of the tree is computed. This probabilistic metho d is a valuable alternative to the maximum parsimony method when base compo sition is skewed (i.e., different from 25% A, 25% C, 25% G, 25% T): compute r simulations showed that parsimony misses more rare --> common than common --> rare changes, resulting in biased inferred change matrices, whereas th e new method appeared unbiased. The probabilistic method was applied to the analysis of the mutation and substitution processes in the mitochondrial c ontrol region of mouse. Distinct change patterns were found at the polymorp hism (within species) and divergence (between species) levels, rejecting th e hypothesis of a neutral evolution of base composition in mitochondrial DN A.