Testing the phylogenetic stability of early tetrapods

Clack (1998) described Eucritta melanolimnetes, an Early Carboniferous tetrapod, and used parsimony to investigate its evolutionary relationships to other early tetrapods. Through comparison of a most parsimonious tree and nine trees of one additional step in length, Clack concluded that &&The phylogeny from the current data set is not particularly robust'' and claimed that &&This instability undoubtedly results from the extreme degree of character con#ict that Eucritta presents.'' We have developed measures of leaf and overall phylogenetic stability that allow such claims about the impact of individual leaves to be tested. Phylogenetic trees represent hypotheses of evolutionary relationships for a set of terminal taxa or leaves. Most studies of the robustness of phylogenetic trees focus on the support for, or stability of, clades. However, trees can also be thought of as collections of less inclusive hypotheses of phylogenetic relationships, such as 3-taxon statements, with implications for the assessment of support (Wilkinson, 1994). For example, a single unstable leaf can result in clades with minimum robustness despite strong support for the phylogenetic relationships of the remaining leaves (Wilkinson, 1996). Measures of leaf stability allow the contribution of each leaf to overall phylogenetic stability to be determined. The rationale for the measures is very simple. The phylogenetic relationships of a set of leaves are a function of the relationships among each subset of three leaves (triplet). The support for the relationships within each triplet provides a measure of the stability of that triplet. Stable and unstable leaves will tend to occur in stable and unstable triplets, respectively. The average stability of the triplets including a leaf provides a measure of the stability of that leaf. Similarly, the