Characterization of genetic relationship of dragon fruit accessions (Hylocereus spp.) by morphological traits and ISSR markers

- Genetic diversity of dragon fruit was revealed by morphological and ISSR makers.
- Coincidence between the two maker systems was investigated in dragon fruit germplasm.
- ISSR fingerprints of dragon fruit germplasm were firstly constructed herein.

Dragon fruit, a kind of fruit belonging to Hylocereus spp., has brought about interest in China not only due to its important economic value with high contents of nutrients, but also due to the striking tolerance to drought stress. Fifteen morphological traits and 111 Inter-Simple Sequence Repeat (ISSR) markers generated from 16 primers were firstly employed to discriminate 50 accessions recently selected in China, as well as to evaluate their genetic relatedness. According to univariate statistical analysis of morphological traits, high variations were observed between or/and within the wild and cultivated lines. The dendrogram of 50 accessions was constructed based on morphological traits. Taking 7.50 as a threshold, three major clusters could be observed, i.e. the first included nine genotypes, which mainly belonged to wild accessions; the second clusters standing for the red pulp genotypes; the third consisting of the white pulp accessions. Each ISSR primer generated 4-10 obvious DNA bands ranging from 100 bp to 1500 bp, with the average of 6.9 per primer. The percentage of polymorphic bands (PPB) varied from 25.00% to 100% with an average of 66.12%, and DNA markers generated from primers UBC824, UBC891and UBC900 could efficiently fingerprint 50 genotypes. Polymorphism information content (PIC) among the tested genotypes varied from 0.49 to 0.93 with an average of 0.85, suggesting a high genetic diversity among the tested genotypes. Based on UPGMA method, 50 genotypes could be grouped into two major clusters at a genetic distance of 0.23. Although there existed discrepancy in genetic relatedness between morphological and ISSR loci, a significant correlation between the data of two evaluation methods was observed using Mantel correspondence test.