Genotyping-by-sequencing of a melon (Cucumis melo L.) germplasm collection from a secondary center of diversity highlights patterns of genetic variation and genomic features of different gene pools

Abstract

Background: Melon (Cucumis melo L.) is one of the most important horticultural species, which includes several taxonomic groups. With the advent of next-generation sequencing, single nucleotide polymorphism (SNP) markers are widely used in the study of genetic diversity and genomics.

Results: We report the first successful application of genotyping-by-sequencing (GBS) technology in melon. We detected 25,422 SNPs by the analysis of 72 accessions collected in Apulia, a secondary centre of diversity in Southern Italy. Analyses of genetic structure, principal components, and hierarchical clustering support the identification of three distinct subpopulations. One of them includes accessions known with the folk name of 'carosello', referable to the chate taxonomic group. This is one of the oldest domesticated forms of C. melo, once widespread in Europe and now exposed to the risk of genetic erosion. The second subpopulation contains landraces of 'barattiere', a regional vegetable production that was never characterized at the DNA level and we show was erroneously considered another form of chate melon. The third subpopulation includes genotypes of winter melon (C. melo var. inodorus). Genetic analysis within each subpopulation revealed patterns of diversity associated with fruit phenotype and geographical origin. We used SNP data to describe, for each subpopulation, the average linkage disequilibrium (LD) decay, and to highlight genomic regions possibly resulting from directional selection and associated with phenotypic variation.

Conclusions: We used GBS to characterize patterns of genetic diversity and genomic features within C. melo. We provide useful information to preserve endangered gene pools and to guide the use of germplasm in breeding. Finally, our findings lay a foundation for molecular breeding approaches and the identification of genes underlying phenotypic traits.

Keywords: Cucumis melo; Genetic diversity; Genomics; Genotyping-by-sequencing; Molecular breeding.

Fig. 1

Population structure analysis. Results are shown for the minimum number of subpopulations (K) which sufficiently define genetic variation, as inferred by the estimation of the ΔK parameter. Each individual is represented by a vertical line, which is partitioned into coloured segments whose length depends on the estimated membership fraction (q) in each subpopulation. Individuals are assigned to a specific subpopulation when the highest q is higher than 0.6 (a) Genetic structure of the C. melo germplasm collection used in this study. The subpopulations I, C, and B, which refer to the types inodorus, ‘carosello’ and ‘barattiere’, respectively, are indicated. b Genetic structure of the subpopulation I. The three subgroups are named Y, G and S as they contain accessions with yellow, green and speckled fruit rind, respectively

Fig. 2

Typical fruit phenotype of a ‘carosello’, ‘barattiere’ (b) and ‘inodorus’ (c) accessions cultivated in the Apulian center of diversity

Fig. 3

Genetic relationships within the germplasm collection used in this study. a Principal component analysis (PCA) plot. Different colors represent accessions phenotypically classified as inodorus (orange), ‘carosello’ (green) and ‘barattiere’ (blue). Circles delimit groups corresponding to the subpopulations C, B and I identified by structure analysis (b) Neighbour-Joining cladogram. Accessions assigned to the C, B and I subpopulations are highlighted (red lines). Admixed accessions are indicated with dots. Shaded areas indicate the I genetic clusters including accessions with green and speckled fruit rind and the C genetic cluster encompassing all the accessions collected in Southern Apulia

Fig. 4

Average LD decay (r²) estimated in the C. melo subpopulations C (a), B (b) and I (c)

Fig. 5

Analysis of single-loci pairwise F_ST estimates among the subpopulations C, B and I. a Genomic distribution of 8,012 SNP loci identified in this study in function of pairwise F_ST estimates. The horizontal line on each graph defines highly polymorphic loci associated with F_ST > 0.9 (b) Venn diagram of highly polymorphic loci (F_ST > 0.9)