Character States, Morphological Variation, and Phylogenetic Analysis: A Review

Abstract

The step in cladistic analysis that has received least attention is delimitation of char- acter states, there usually being little justification for their delimitation. It is generally assumed that states of cladistic characters are discrete, even when variation is quantitative. I show here that a majority of the character states of obviously quantitative characters used in lower-level cladistic studies in botany over the last generation are ambiguous even when ingroup variation alone is analyzed. Consideration of variation in the outgroup may compromize either the states recognized in the ingroup and/or the polarity that they are subsequently assigned. Furthermore, many so- called qualitative characters are based on a quantitative phenomenological base filtered through the reified semantic discontinuities of botanical terminology; such characters face the problems of their more obviously quantitative relatives. Methods for delimiting states within quantitative char- acters are examined. Some produce gaps in the variation by redefining the character, scoring the intermediates in a distinctive fashion, performing phylogenetic analyses within the terminal taxa, or changing the hierarchical level at which the variation is evaluated. Others produce states by manipulation of the statistical properties of the variation of the ensemble of taxa being studied. These latter methods often allow greater resolution of the phylogeny, but at the cost of lowering the significance of the most parsimonious tree. The underlying assumptions of the two sets of methods are briefly analyzed. Problems manifest in the division of continuous variation into char- acter states suggest a reappraisal of the early steps of cladistic analysis; in practice, character states often seem to be delimited in conjunction with developing ideas of the phylogeny, rather than in a step prior to a phylogenetic analysis. It is recommended that character states be delimited by carefully analyzed discontinuities (not necessarily absolute gaps) in the variation, attention having been paid to variation in the outgroup, and that morphological characters in general are assumed to be quantitative unless demonstrated otherwise. Explicit justification for the delimitation of char- acter states should be given as a matter of course in all phylogenetic studies.