Comparative analysis of cultivated melon groups (Cucumis melo L.) using random amplified polymorphic DNA and simple sequence repeat markers

Random amplified polymorphic DNA (RAPD) and simple sequence repeat (SSR) markers were used to characterize genetic relationships among 46 accessions in two C. melo L. subsp. melo (Cantalupensis, Inodorus) and subsp. agrestis (Conomon, and Flexuosus) groups. Genetic distance (GD) estimates were made among and between accessions in four melon market classes [Galia, Ogen, Charentais, and Shipper (European and U.S. types)] of Cantalupensis, one market class of Inodorus (Cassaba and Honey Dew), one accession of Conomon, and one accession of Flexuosus by employing three GD estimators; simple matching coefficient, Jaccard's coefficient, and Nei's distance-D. Differences detected among 135 RAPD bands and 54 SSR bands (products of 17 SSR primers) were used to calculate GD. Band polymorphisms observed with 21 RAPD primers and 7 SSR primers were important (p = 0.01) in the detection of genetic differences. Estimators of GD were highly correlated (p > 0.0001; r(s) = 0.64 to 0.99) when comparisons were made between estimation methods within a particular marker system. Lower correlations (r(s) = 0.17 to 0.40) were detected (P > 0.001) between marker systems using any one estimator. The GD of the Conomon and Flexuosus accessions was significantly different (p > 0.001) from the mean GD of all the market classes examined. The mean GD (Jaccard's coefficient) among accessions of Ogen, Galia, Cassaba, Charentais, European shipper, and U.S. shipper groups was 0.11 +/- 0.04, 0.33 +/- 0.09, 0.21 +/- 0.04, 0.26 +/- 0.10, 0.17 +/- 0.05 and 0.22 +/- 0.08, respectively. Market classes were distinct (p > 0.001), such that GDs between Galia and other accessions were the largest (mean GD 0.34 to 0.35), and GDs between Ogen and other accessions were the smallest (mean GD 0.29 to 0.30). Contrasts between the U.S. shipper cultivar Top Mark and accessions within any market class was relatively large (mean GD = 0.42 +/- 0.06). Empirical estimations of variances associated with each marker type in the accessions examined indicated that, per band, lower coefficients of variation can be attained in the estimation of GD when using RAPDs compared to SSRs. Nevertheless, the genetic relationships identified using these markers were generally similar. The disparity between the analyses of the two markers made may be related to the amount of genome coverage which is characteristic of a particular marker system and/or its efficiency in sampling variation in a population. Results of RAPD marker analysis suggest that 80 marker bands were adequate for assessing the genetic variation present in the accessions examined.