Abstract

Reanalysis of Mayden's data on distribution and relationships of North American freshwater fishes reveals weaknesses of the analytical protocol termed Brooks' Parsimony Analysis (BPA). Standard assumptions 2, 1, and 0 are explored with reference to suites of 3-area statements contained in cladograms of species. Component analysis proved effective for all assumptions for the freshwater fish data whereas BPA yielded results not optimal for any assumption. For Rosen's data on Heterandria and Xiphophoms, component analysis proved effective for assumption 2 whereas results from other methods proved effective for assumptions 1 and 0. Analysis of widespread species indicates that, when they are used to build area cladograms, they cause spurious results. Assumptions 1 and 0 are sensitive to these spurious effects.