Phylogeny of the Berberidaceae With an Evaluation of Classifications

Abstract

An extended outline of the logical basis of character compatibility analysis is presented. Convexity of taxa on the chosen estimate of phylogenetic his- tory is advocated as a criterion for evaluating classifications. Thirty characters for fifteen genera of the Berberidaceae are analyzed for character compatibility. A final undirected estimate of phylogeny of the Berberidaceae is given. Airy Shaw's Ber- beridaceae s. str. is found to be polyphyletic on the final estimate while all of Janchen's taxa are convex. Character compatibility analysis supports recognition of four subfamilial taxa: 1) Berberis, Mahonia, and Ranzania; 2) Podophyllum, Dysosma, and Diphylleia; 3) Epimedium, Vancouveria, Jeffersonia, Plagiorhegma, and Achlys; and 4) Caulophyllum, Leontice, Bongardia, and Gymnospermium. This technique can be use- ful in the resolution of systematic problems above the generic level. The use of cladistics as an objective means of estimating evolutionary history has received an increasing amount of attention from plant sys- tematists. Most of this attention has been focused on problems below the generic level. A pertinent question is: can cladistic methods be of use in resolving evolutionary relationships among higher plant taxa? To illus- trate the utility of these techniques at higher levels, I have applied char- acter compatibility analysis to the genera of the Berberidaceae. Although the concepts of character compatibility have been elaborated in previous literature (Estabrook 1972; Estabrook et al. 1975, 1976a, b; Estabrook and Landrum 1975; McMorris 1977; Estabrook and Meacham 1979), an outline is presented here in the hope that an understanding of the logical basis of the method can be facilitated by a less technical discussion. Read- ers who wish a more complete, formal exposition should refer to the cited articles. AN OUTLINE OF CHARACTER COMPATIBILITY ANALYSIS All current cladistic methods assume that the evolutionary history of the group of organisms being studied, called the study collection, can be adequately represented by a tree. This condition places some restrictions on the entities that are suitable as objects of study. For example these objects, generally termed evolutionary units or EU's, cannot be individuals of sexual species because each individual has two parents and is thus the product of a reticulation in the evolutionary history of the group. A tree, which by definition has no reticulations, cannot describe this situation. The EU's must be at least populations of individuals, but it is clear that in many cases an assumption that the evolutionary history of plant pop- ulations contains no reticulations is unwarranted. Because of this limi-