NUMERICAL TAXONOMY OF IBERIAN PEA LANDRACES BASED ON QUANTITATIVE ANDQUALITATIVE CHARACTERS

A multivariate statistical method was applied on 19 quantitative and 1 1 qualitative traits in order to classify 105 pea landraces from the N orthwest of the Iberian Peninsula. Ten elite cultivars were used as co ntrols. The eleven qualitative characters were transformed by special coding. A procedure of data analyzed based on the combined resemblance matrix as a result of the joint resemblance matrix for quantitative a nd qualitative traits was constructed and described. The 105 pea landr aces display a wide diversity at the infraspecific level for the 19 qu antitative traits, however, just ten of them plus one qualitative trai t could be considered as good descriptors for the pea landraces studie d. The grouping patterns of the landraces did not reflect geographical origin, but provided taxonomically useful results, since it shows the possible use of six groups, vis. Group I (12 landraces), includes acc essions which could be used for green and canned consumption, Group IV (18 landraces) includes edible-podded types and Group V (12 landraces ) could be used for field pea (dry seed cultivars), Groups VI and Vm ( each with 26 accessions) are also extremely interesting because of the ir heterogeneity.