A Phylogenetic Analysis of the Caminalcules. IV. Congruence and Character Stability

Abstract

The taxonomic congruence of phenetic and cladistic classifications is tested, as is the closeness of estimated cladograms to the true cladogram as a function of the number of characters employed, using the Caminalcules as an example. This group of “organisms” was generated artificially according to principles believed to resemble those operating in real organisms. Congruence was tested by random bipartitions of character suites. Results show the stability of phenetic classifications as very much higher than that of cladistic estimates. The consensus of cladistic classifications based on bipartitions is higher with the true cladogram than between cladistic classifications based on complementary bipartitions. But phenetic classifications based on these same bipartitions have an even higher consensus with a phenetic standard classification based on all characters, and a moderately higher consensus with the true cladogram. The following three cladistic estimation methods did not produce materially differing results: distance Wagner procedure with midpoint rooting; the same procedure with rooting at the true ancestor; Wagner parsimony based on additive binary coding, analysis of characters with no comparison states, and rooting at the true ancestor. Estimated cladograms based on all 85 characters are closer to the true cladogeny than phenograms computed for these characters. When estimated cladograms are based on fewer characters, they tend to become less close to the true cladogram than are phenograms based on the identical subsuites of characters. The exact relationships differ with the consensus measure employed. Estimates based on shortest Wagner trees are frequently not as close to the true cladogram as longer trees based on the same characters. The implications of the above findings for systematic theory and practice are discussed.