The Cladistics of Matrix Representation with Parsimony Analysis

Abstract

The construction of supertrees using matrix representation with parsimony (MRP) is equivalent operationally to the construction of cladograms using cladistic analysis of character data. However, the validity of MRP as a phylogenetic method has been questioned because the data used to construct MRP supertrees are the topologies of trees rather than character data. The consistency of MRP analysis with the following cladistic principles is evaluated: 1) only synapomorphies provide evidence for cladistic relationships, 2) ad hoc hypotheses are to be minimized in the generation of cladistic hypotheses, and 3) data used in the inference of cladistic relationships must be independent of each other. To be consistent with these principles, MRP analysis must 1) be based on source trees that were generated using cladistic analyses of character data, 2) weight the input data to account for the relative support for individual nodes on source trees and to eliminate inappropriate biases associated with variation in tree size, 3) be based on source trees with high consistency indices, and 4) be based on source trees that provide independent evidence for relationships. Achieving these criteria is extremely difficult, and all published MRP analyses fail to meet one or more of these conditions. Although MRP supertrees might be justified on pragmatic grounds, these trees should be considered a heuristic synthesis of hierarchical information, rather than a rigorous phylogenetic analysis of the included taxa.