Abstract

The aim of the study was to compared the genetic similarity coefficients (Jaccard, Dice, Simple Matching) and different clustering methods (UPGMA, WPGMA, Single Linkage and Complete Linkage) combinations for cultivated olives. A total of 12 samples, Gemlik, Manzanilla, Edremit, Domat, Uslu and Memecik cultivars were screened with RAPD-PCR analysis by Operon random primers OP-Q kit. The closest samples based on their genetic similarity values have been found as ‘Edremit and Gemlik 5’ and the most distant ones have been found as ‘Manzanilla 1 and Gemlik 3’ in both of Dice and Jaccard coefficients; whereas, in simple-matching coefficient, the closest samples based on their genetic similarities have been determined as Edremit and Gemlik 5 and the most distant ones have been determined as Manzanilla 1 and Gemlik 3 and also Manzanilla 1 and Gemlik 4. The results from Mantel test of original matrices show that the correlation between Jaccard and Dice similarity matrices was high and significant (0.9971). UPGMA clustering for Dice coefficient was given a highest cophenetic correlation as 0.9571 and Complete linkage clustering for Simple Matching was yielded a lowest correlation value as 0.8992. The results obtained from consensus indices shown that Consensus fork index was found (CI_C =0, 9000) in Jaccard and Dice coefficients. Simple Matching coefficient had very low values with the Dice and Jaccard coefficients (CI_C =0.1000). PCO analysis provided results matching up one-to-one with the data obtained from Dice and Jaccard coefficient UPGMAs.

Key words: Clustering methods, genetic similarity coefficients, PCO, randomly amplified polymorphic DNA, olive.