An Experiment in Taxonomic Judgment

Publisher Website
Google Scholar
Abstract
Twenty-two subjects ranging from distinguished professional systematists through graduate students to children were asked to classify the 29 Recent species of Caminalcules by whatever taxonomic principle seemed "natural" to them. The resulting classifications were compared with three standards-two of these are phenetic and based on a numerical taxonomic study of these images by A. J. Boyce and I. Huber, the third is the true cladogram furnished by J. H. Camin. Each experimental classification was coded numerically in two ways to furnish a phenetic and a cladistic interpretation of the classification. Correlations among the classifications reveal substantial variation among them. The average resemblance of the classifications is highest with the phenetic standard based on correlations, which is a better measure of shape; average resemblance is lower for the phenetic standard based on distances, which measure size; and the lowest resemblance is shown to the true cladistic classification. There is no relation between experience or field of expertise and closeness to a standard classification; children do as well as professionals. Errors in judgment about close affinities between pairs of operational taxonomic units are largely restricted to one genus, E. Application of the technique of judgment analysis revealed that four to six out of 112 characters predict the majority of the classificatory decisions of each of the subjects. These characters determine generic structure in the Caminalcules but are not the visually most striking characteristics. Classifications described by their originators as cladistic or phylogenetic were closer to phenetic standards than to the cladistic standard.