OTU Stability and Factors Determining Taxonomic Stability: Examples from the Caminalcules and the Leptopodomorpha

Publisher Website
Google Scholar
Abstract
Tests of OTU [operational taxonomic unit] stability were carried out by random selection of subsets of OTU from the Recent Caminalcules. From these samples UPGMA [unweighed pair-group methods with arithmetic averages] phenograms were computed and cladograms were estimated based on Wagner parsimony and compatibility analysis. These were compared with pruned versions of the trees based on 29 OTU representing their standards and the true cladogram known for this artificial data set. A similar analysis was carried out for a complete sampling of sets of OTU from a data set based on the Leptopodomorpha (Hemiptera). Only the 1st type of comparison could be carried out for this data set. In the Caminalcules, when classifications based on samples of OTU are compared with their own standards (classifications based on all OTU), which would be the typical situation in real organisms where true cladogenies are not known, stability seems to be a function of OTU sample size. For low OTU numbers, stability is higher in classifications based on estimated cladograms; but for high OTU numbers, it is greater in classifications based on phenograms. While there is a trend towards increasing stability in the phenograms as OTU sample size increases, no such trend is evident in esimated cladograms of the standards or of the true cladogram. By contrast, estimated cladograms have a higher consensus with the true cladogram than do UPGMA phenograms regardless of number of OTU employed. In the Leptopodomorpha, classifications based on UPGMA and Wagner parsimony methods are about equally stable with a suggestion of slightly better performance for Wagner parsimony at low OTU numbers and a reversal of this trend for higher OTU numbers. There is no evidence of marked superiority of Wagner parsimony as had been claimed in an earlier study of this group. The outcome of studies of both character and OTU stability is approximately related to the ratio n/(2t-3), where n is the number of characters and t is the number of OTU. When this ratio is low, phenograms are more stable than estimated cladograms; when it is high the reverse relation holds. Also, when the ratio is low, phenograms are better estimates of the true cladogeny than are estimated cladograms. The reverse relation holds for high values of the ratio. To confirm these conclusions, combined studies of character and of OTU stability applied to both data sets were carried out. As character number decreased, or OTU number increased, classifications based on phenograms became more stable than those based on estimated cladograms, and became better for reconstructing the true cladogeny than estimated cladograms in the Caminalcules. Character stability and OTU stability may in fact be 2 separate aspects of the same phenomenon.sbd.taxonomic stability. The merits of various designs suitable for testing OTU stability in numerical taxonomy are discussed. All are based on the results of sampling subsets of OTU for the entire taxon.