Assessment of the Genomic System of Classification in the Triticeae

We have made an assessment of the genomic system of classification recently proposed for the Triticeae as a whole by Löve and for perennial Triticeae by Dewey. Limitation of genomes as a measure of similarity were discussed using examples where pairing may fail even between highly similar genomes and where morphology is a better indicator of evolutionary relationships. We found that only two of the 41 characters extracted from Löve's descriptions were described for 36 of the 37 genera, and most characters were described for less than half of the genera. Thus the descriptions were incomplete. This was further illustrated when a data matrix was created using 15 most frequent characters. A majority rule consensus tree was computed using the program BOOTM in Felsenstein's PHYLIP package, and we found that many operational taxonomic units or OTUs (the genomically defined genera) could not be resolved on the tree; homoplasy occurs throughout the tree, synapomorphies were parallel in many branches, and only one character (flowers, perfect/rudimentary) was an apomorphy leading to Elymus. We recommended that all the attributes be considered a priori with equal weight without attributing supremacy to genome, thus enabling the genera to be described, circumscribed, and arranged in a meaningful system of classification.