REP provides meaningful measurement of support across datasets

摘要

The relative optimality of the best and next-best hypotheses indicates the strength of support for the optimal hypothesis and may be calculated as either the difference or ratio of their optimality scores. The Goodman-Bremer support measure (GB) calculates the support for a given clade in the most parsimonious tree as the difference between the length of the optimal tree and the optimal tree that lacks that clade. The ratio of explanatory power (REP) support measure calculates support as the ratio of optimality scores, which simplifies to the ratio of observed GB and the maximum possible GB, GB/GB_(max). In this paper we show that REP support provides a logical basis for comparison of support across datasets and that recent claims to the contrary are incorrect.