Numerical taxonomic studies in the generaTriticum L. andPisum L.

Summary

Up till now numerical taxonomy has been applied to cultivated plants only to a limited extent though this method seems to be especially suitable for handling the great diversity of these plants. Moreover the theory of numerical taxonomy has not been studied sufficiently.

Therefore studies with cultivated plants were begun with the aim to investigate the fitness of different numerical taxonomic techniques for their classification. Materials used: 1) 24 species ofTriticum and additionally 24 species ofAegilops 2) 1225Triticum taxa from Iran (mostlyT. aestivum) 3) 312 taxa ofPisum (mostlyP. sativum).

Using 10 coefficients and 7 clustering methods inTriticum andAegilops in most cases a clear separation of both genera resulted as well as a species arrangement withinTriticum according to the genomes.

The main aim when dealing with the wheat from Iran was the investigation of the influence of different character sets. At this the coefficient ofGower and the unweighted pair-group method using arithmetic averages (UPGMA) were applied. The separation of the different species is satisfying clear while the representatives of the different varieties ofT. aestivum could not always be found in one cluster. This effect increased when adding further characters sets.

InPisum the influence of 11 similarity coefficients to the dendrogram formation has been investigated. With two-state characters the simple matching coefficient and with mainly multistate characters the Canberra metric gave best results (using UPGMA and information measures SUMRAT and SAMRAT).

The hitherto existing results show the complicatedness of the evaluation of numerical taxonomic methods in their application to cultivated plants. Therefore generalizations are not yet possible.