The logical basis for coding ontologically dependent characters.

Abstract

The coding of dependent morphological characters represents a major methodological problem in phylogenetics. Based on a distinction of semantic and ontological logical character dependency, I suggest how inapplicables can be treated properly and introduce rules of mutually dependent character states, which specify how character states of one character determine character states in its ontologically dependent characters. Using various examples, I discuss a set of general rules that applies independently of whether the ontological dependency results from property instantiation, parthood or subsumption. When implemented in a matrix editor, these rules would significantly facilitate the coding procedure, speed up coding of large matrices and increase overall consistency. If implemented in algorithms for tree reconstruction, the rules would prevent inconsistent reconstructions of ancestral states, which is a common problem with inapplicables. In the second part of the paper I set out to explore the potential of using a semantic framework and semantic techniques for automatically detecting instances of ontological dependency and specific cases of semantic dependency and how they can be applied for automatically coding character state values for ontologically dependent characters using the general rules discussed in the first part of the paper. This approach utilizes graph-based semantic representations of instance anatomy, which represent ontology-based descriptions of the anatomical organization of individual organisms.