A Framework for the Quantitative Study of Evolutionary Trees

摘要: A direct method for calculating expected data from an evolutionary model two state characters is described. The uses four vectors p, q, r and s. p q are the probabilities of a character change on 2n - 3 edges tree T (n number taxa). s properties data, independent any have 2n-1 entries. For given T, with specified (p or q), we determine r, then s, each possible partitions taxa. relationship can be inverted. This allows tree, to estimated directly observed (r s). These relationships been used analyse behaviour building algorithms under conditions when there sufficient data. (This does not as more collected, i.e., convergence single tree.) With equal rates evolution (i.e., molecular clock), show that n = 4 taxa, parsimony will always converge correct but give examples 5 where incorrect even evolution. further example 6 shows arbitrarily small change. We interpret basic difficulty 'long attract.' If additional taxa intersect long this effect reduced. Some distance methods may also tree. (Evolu- tionary trees, evolution, parsimony.) Reconstructing trees events occurred many millions years ago major intellectual challenge. Two recurring problems been: choice optimality criteria estimation