Characters as groups: a new approach to morphological characters in phylogenetic analysis

Morphological characters remain an essential tool for phylogenetic studies, even in the current age of molecular systematics (Scotland & al., 2003; Jenner, 2004; Wiens 2004; Smith & Turner, 2005; Lee, 2006). Consequently, the reliability and validity of the methods used to identify characters and define character states is highly relevant to systematics research. Although identifying appropriate characters has been a longstanding issue in systematics, advances in phylogenetic methods have not been fully matched by improvements in methods for selecting and evaluating characters (Neff, 1986; Kemp, 1999: 63; Judd & al., 2002: 46–51; Richards, 2003; Mishler, 2005). This paper explores a solution to the problem of delimiting morphological characters by applying insights from cognitive psychology. We do not deal with the many theoretical aspects of the character concept, which have been presented by others (see Wagner, 2001), but suggest new ways of thinking about the terms character and character state, and propose operational definitions of these terms. We view characters as arising through an interaction between the objective object of study (the organism) and the investigator. Characters depend not only on the structure of the organism, but on the ability of the investigator to reliably dismember the organism into meaningful units, and on his or her ability to assess the similarity of these units across study taxa (Richards, 2003). The first process produces characters, the second character states. At a minimum, characters and character states should be consistent from investigator to investigator (they should be reliable), and they should be good estimates of the process being investigated (they should be valid). Reliability can be investigated by statistical Characters as groups: a new approach to morphological characters in phylogenetic analysis