We compare cladistic, phenetic, stratophenetic, and typological methods of recognizing species within a single well-sampled, long-lived Ordovician bryozoan genus, PeronoporaNicholson, 1881. A consensus of 11 methods recognizes 14 species within Peronopora, each of which has both cladistic and phenetic support, and support by at least four different methods. Comparison of methods was based on: 1) the number of species recognized; 2) the degree of splitting of consensus groups; 3) consistency in determining first and last appearance datums, and stratigraphic ranges; 4) consistency in the number of specimens assigned to species; 5) stability in recognizing geographic distributions; and 6) correlations with character heritability. The five methods most closely approaching consensus were: 1) Ward clustering on Euclidean distance; 2) K-means splitting; 3) Ward clustering on correlation coefficients; 4) cladistic parsimony; and 5) cladistic parsimony plus iterative discriminant analysis. The three methods farthest from consensus were: 1) stratophenetics; 2) average linkage on Euclidean distance; and 3) conventional typology. Cladistic parsimony is the only method that can recognize all 14 consensus clusters. We argue that the overall advantages of parsimony analysis outweigh the merits of the various phenetic approaches in recognizing paleontological species. We also argue that given sufficient allochrony in sampling, cladistic structure is detectable both within and among species. Recognizing species based on fixed character states would require at least 72 species in our material.
The crown groups of Pachut and Anstey (2002) are herein recognized as monophyletic species in the sense of Mishler and Theriot (2000), one of which we designate as a new species, P. browni. Six of our eight nonmonophyletic stem groups are recognized in this paper as metaspecies in the sense of Donoghue (1985) and Gauthier et al. (1988). Two of our stem groups are phenetically indistinguishable from phyletically contiguous crown groups, and we attribute the failure of parsimony analysis to recognize the monophyly of these groups to homoplasy.