Decay analysis of large sets of phylogenetic data

Abstract

SummaryMorgan, D. R.: Decay analysis of large sets of phylogenetic data. – Taxon 46: 509‐517. 1997. – ISSN 0040‐0262.Decay analysis provides an alternative, or supplement, to bootstrap analysis for obtaining estimates of the reliability of groups resulting from a phylogenetic analysis. This approach, which consists of examining trees longer than the shortest and noting the tree lengths at which groups of interest are no longer monophyletic, is used increasingly. One serious shortcoming of decay analysis as often implemented is that the number of trees that need to be saved and examined can be much larger than the capacities of hardware and software that have been available. A data matrix was constructed so that all trees up to five steps longer than the shortest could be saved and examined. A series of decay analyses was then conducted with this data set to determine the effects produced in the results when all trees are not found and to evaluate methods that could be used to conduct decay analyses of large data sets. Limiting the number of trees to be saved resulted in overestimates of support for monophyletic groups, a problem that became substantial when only 33% of all trees were examined. Two methods of decay analysis that use constraint trees to save a subset of all trees were evaluated. Decay index values resulting from both methods were identical to those obtained from examination of all trees, indicating that these approaches should be useful in conducting decay analyses of large data sets.