Phylogenetic Nomenclature, Hierarchical Information, and Testability

In a continuing debate about the usefulness of phylogenetic nomenclature, Platnick (2012) has apparently conceded the two main points of our previous paper (de Queiroz and Donoghue 2011) by providing no counter-arguments. First, contrary to Platnick’s previous assertions, when appropriate comparisons are made (i.e., between taxonomies consisting of the same-named groups), there are no differences in information content, as measured by implied three-taxon statements, between hierarchical taxonomies whose names are governed by rank-based versus phylogenetic nomenclature. The reason is that three-taxon informativeness is a property of the clades that are recognized (the taxonomy) rather than of the rules governing their names (the nomenclatural system). Second, phylogenetic nomenclature outperforms its rank-based counterpart when the approaches are compared using a nomenclaturally relevant criterion. Specifically, phylogenetically defined names result in fewer unnecessary name changes in the context of new phylogenetic hypotheses. Having abandoned his criticism of phylogenetic nomenclature based on three-taxon informativeness, Platnick (2012) focused his rebuttal more or less entirely on the mutual exclusivity of taxa whose names bear the same rank-signifying endings. He contended that information about mutual exclusivity gives rank-based names more predictive power and testability than their phylogenetically defined counterparts. Consequently, rather than favoring the replacement of rank-based definitions by phylogenetic ones, as advocated by ourselves and others, Platnick advocated extending the use of standard, rank-signifying endings and, by implication, rank-based definitions to the names of clades at all hierarchical levels (currently they extend only up to the rank of superfamily in zoology, and are used inconsistently above the rank of family in botany). Contrary to the impression given by Platnick, we argue here that the information in rank-signifying endings is limited, so that any loss of such information necessitated by the current version of the International Code of Phylogenetic Nomenclature (hereafter, the PhyloCode) would not be a great hindrance to science. In addition, incorporating that information comes at a cost in the form of unnecessary and inappropriate name changes. We also argue that the application of phylogenetic definitions leads to very explicit hypotheses about taxon composition that are no less testable than those associated with traditional taxa, the testability of which has nothing to do with rank-signifying endings or rankbased definitions. Moreover, phylogenetic definitions can be formulated so that names designate particular phylogenetic hypotheses and are therefore rejected when the hypotheses that they represent are considered falsified, a possibility that does not exist with rankbased definitions. Finally, we argue that the use of categorical ranks is not logically necessary to convey the hierarchical information associated with what are commonly thought of as rank-signifying endings. Such endings can be interpreted as signifying only relative rather than absolute (categorical) ranks, in which case they are compatible with phylogenetic nomenclature. This observation highlights the generality of the phylogenetic approach to nomenclature as well as a distinction between that general approach and the specific rules and recommendations adopted in the PhyloCode.