Evolutionary Polarity of Character States

Abstract

Various criteria have been used in recent literature to decide whether a given character state is primitive or derived. The problems are similar in botany and zoology. The present review emphasizes the botanical literature, in which there has been no general examination of the validity of the criteria since Spome's excellent review (149). Many of the criteria that Spome rejected as being inadequate have since crept back in to use; the criterion that he favored, correlation of characters, is of doubtful utility; and a criterion that he did not discuss, character state distribution, has since become widely used, but in two very different ways. Many of the criteria are interrelated, and several depend on further, generally unstated criteria; criteria are often used in combinations, so that an author's actual procedure becomes unclear. Finally, a criterion rejected in one place in an article may be accepted elsewhere (sometimes in the same paragraph). Hence a reexam­ ination of the validity of the various proposed criteria is overdue. There is now considerable interest in various forms of cladistic or phylo· genetic classification in botanical circles [(22, 72, 91, 166) and references]. Although these approaches to classification are thought beneficial because of their explicit assumptions and procedures, it is ironic that the recent literature hardly mentions the crucial early step: the assignment of evolu­ tionary polarity (101) to character states-i.e. the decision about which character states in the taxa being examined are derived or apomorphous and which are primitive or plesiomorphous (the terms are defined below). That estimates of evolutionary relationships depend on the decision about which character states are derived was shown by Solbrig (145), who obtained different phylogenies when he used three different criteria for this decision.