Meta genetic analysis of melon sweetness

Abstract

Through meta-genetic analysis of Cucumis melo sweetness, we expand the description of the complex genetic architecture of this trait. Integration of extensive new results with published QTL data provides an outline towards construction of a melon sweetness pan-QTLome. An ultimate objective in crop genetics is describing the complete repertoire of genes and alleles that shape the phenotypic variation of a quantitative trait within a species. Flesh sweetness is a primary determinant of fruit quality and consumer acceptance of melons. Cucumis melo is a diverse species that, among other traits, displays extensive variation in total soluble solids (TSS) content in fruit flesh, ranging from 2⁰ Brix in non-sweet to 18⁰ Brix in sweet accessions. We present here meta-genetic analysis of TSS and sugar variation in melon, using six different populations and fruit measurements collected from more than 30,000 open-field and greenhouse-grown plants, integrated with 15 published melon sweetness-related quantitative trait loci (QTL) studies. Starting with characterization of sugar composition variation across 180 diverse accessions that represent 3 subspecies and 12 of their cultivar-groups, we mapped TSS and sugar QTLs, and confirmed that sucrose accumulation is the key variable explaining TSS variation. All modes-of-inheritance for TSS were displayed by multi-season analysis of a broad half-diallel population derived from 20 diverse founders, with significant prevalence of the additive component. Through parallel genetic mapping in four advanced bi-parental populations, we identified common as well as unique TSS QTLs in 12 chromosomal regions. We demonstrate the cumulative less-than-additive nature of favorable TSS QTL alleles and the potential of a QTL-stacking approach. Using our broad dataset, we were additionally able to show that TSS variation displays weak genetic correlations with melon fruit size and ripening behavior, supporting effective breeding for sweetness per se. Our integrated analysis, combined with additional layers of published QTL data, broadens the perspective on the complex genetic landscape of melon sweetness and proposes a scheme towards future construction of a crop community-driven melon sweetness pan-QTLome.

Keywords: Cucumis melo; Cucurbitaceae; BSA-Seq; Fruit-quality; GWAS; Genome; Half-diallel; QTL mapping; QTLome; Sugars; Total soluble solids (TSS).

Fig. 1

Characterization of TSS and sugar composition and GWAS of TSS in the Melo180 diverse collection. a Sugar profile in flesh of mature fruits of 177 diverse accessions. Accessions are ordered by their total sugars. b Manhattan plot of GWA results for TSS. Upper pane is based on GLM analysis and displays the genome-wide distribution of naïve model P values. Lower pane is MLM with population relatedness correction. Dashed horizontal line is the significance threshold. Red arrows indicate significant QTL peaks identified in the MLM analysis with a visible association signal (above the false-discovery noise) also in the GLM analysis. Pink arrow on chr.3 is a significant association identified only by the GLM analysis. c Quantile–quantile (Q-Q) plot for distribution of TSS P values in the MLM model. The negative logarithm of the observed (y-axis) and the expected (x-axis) P values are plotted for each SNP (dot). The gray dashed line indicates the null hypothesis

Fig. 2

Mode-of-inheritance of TSS in 190 half-diallel F₁ hybrids and their 20 parents. a Heat map of the 20 × 20 half-diallel matrix for TSS. Both axes are ordered by parental GCA, which was based on the mean of all 19 crosses for each parent. Diagonal is the performance per se of each of the parents. b Correlation between parental mean (mid-parent) and F1 hybrid for TSS across 190 hybrid groups (HDA20). Red ellipse indicates the bivariate normal density at 95% coverage. c Frequency distribution of additive TSS difference, a, across 190 half-diallel parental pairs; the 90 hybrid groups with a > 2 are highlighted; these were selected for the dominance analysis. d Frequency distribution of TSS dominance effect, d, across 90 selected hybrid groups with a > 2; d is the difference between the F1 and its parental mean. e Frequency distribution of degree of dominance (d/a) across 90 selected hybrid groups with a > 2. f Example of dominant mode-of-inheritance towards low-TSS in the DUL × PI164323 hybrid group. g Example of complete additivity (no dominance deviation) in the ARJ × TOG hybrid group (members of the Core25 set). h Example of over-dominant mode-of-inheritance towards high-TSS in the SAS × QME hybrid group. In f–h, mean diamonds show the trait mean and confidence interval. Mid-horizontal line is the trait mean. The top and bottom of each diamond span the 95% confidence interval for the mean of each group. The lines near the top and bottom of the diamonds are overlap marks for visual statistical difference between group means

Fig. 3

Bulk-sequencing analysis (BSA-Seq) of three sweet × non-sweet melon populations. a-c Whole-genome view of BSA-Seq results of the high- versus low-TSS bulks in the SAS × DOYA F5 population (a), TAD × QME F6 (b) and TAD × PI164323 F6 (c). The 12 melon chromosomes are presented. Physical positions are based on the DHL92 v4.0 genome. Trend-lines are the running average of SNP-Index and Δ-SNP Index values. Blue line is the SNP-index in the low-TSS bulk. Red line is the SNP-index in the high-TSS bulk. Green line is the Δ-SNP Index. Dashed red horizontal line is the 0.4 threshold for significance of the Δ-SNP Index. d-f Zoom in on chromosome 3 for BSA-Seq results of the high- versus low-TSS bulks in the SAS × DOYA F5 population (d), TAD × QME F6 (e) and TAD × PI164323 F6 (f). Background dots are the SNP Index and Δ-SNP Index at each of the more than ~ 250,000 informative SNPs along the chromosome. Red = High-TSS bulk, Blue = Low-TSS bulk, Green = Δ-SNP Index. Dashed red horizontal line is the 0.4 threshold for significance of the Δ-SNP Index. Dashed rectangle is the QTL interval in each populations. g-i Allelic effect analysis at the qTSS3.1 QTL-peak marker in 2022 and 2023. Homozygote genotypes are presented. Each point represents the mean of an F5 or F6 line

Fig. 4

TSS QTL mapping in the TAD × DUL RIL population over 3 years. a whole-genome QTL analysis over three years (2016–2018). Dashed horizontal lines are the genome-wide QTL significance threshold. Red arrows indicate qTSS3.1 over 3 years. Blue arrows are non-consistent QTLs. b Chromosome 3 QTL plot in 2016 and 2018. The dashed rectangle indicates the QTL borders based on a 1.5 LOD confidence interval. c Allelic effect analysis at the qTSS3.1 QTL-peak marker in 2016–2018. Homozygote genotypes are presented. Each point represents the mean of a RIL in the TAD × DUL population

Fig. 5

Sweetness-QTLome: Integrative TSS QTL mapping across 5 populations in the current study alongside sugars and TSS QTLs mapped in 15 previous experiments in melon. a Comparative map plotting sweetness-related QTLs across 20 experiments. Each row represents an experiment (A–T, details on each population and experiment are provided in Sup. Table S4). The 12 melon chromosomes with their physical map coordinates (Mbp) are plotted on the bottom. Rectangles represent the positions of QTLs. Very thin rectangles mostly represent significant GWAS markers without defined QTL confidence intervals. Rows P–T in the upper part are QTLs mapped in the current study. Red rectangles are significant QTLs where the favorable (high TSS) allele is contributed by the high-TSS parent. Blue rectangles are significant QTLs where the favorable (high TSS) allele is contributed by the low-TSS parent. Gray horizontal rectangles in rows A–O are a collection of sweetness QTLs reported in 15 published experiments. Light colored rectangles (gray and pink) are large QTLs that span more than 15 Mbps and excluded from the bins analysis. Dashed vertical rectangles represent chromosomal bins with QTLs in 6 or more studies. b Overview of the diverse melon germplasm used for the integrated TSS QTLome. Each circle represents a cultivar-group within Cucumis melo. Letters attached to dashed lines or circle represent the corresponding population/experiment as listed in Sup. Table S4. c Bar chart for the number of QTLs per population (across the 20 studies described in Sup. Table S4). Gray bars represent previous studies. Blue bars are populations in the current study. d Heatmap for the number of QTLs within 5 Mb chromosome bins across the 20 studies

Fig. 6

Cumulative nature of TSS QTLs. a–d Analyses of QTL-pair interactions across three populations. Values not connected by a common letter are significantly different at P < 0.05. e–f triple-QTL stacking effect in the SAS × DOYA F5 population (e) and TAD × QME F6 population (f). g–h Allelic effect analysis for qTSS1.1 and qTSS4.1 where the favorable, TSS-increasing allele is contributed by the low-TSS parent. i–j Correlation between expected and observed TSS of QTL pairs across three populations in 2022 and 2023. Dashed diagonal is the expected = observed line. Continuous regression line is shown for both years and its linear function is presented. k–l Comparison of allelic effects of single QTLs versus QTL-pairs in 2022 and 2023. Mean diamonds show the trait mean and confidence interval. Mid-horizontal line is the trait mean. The top and bottom of each diamond span the 95% confidence interval for the mean of each group. The lines near the top and bottom of the diamonds are overlap marks for visual statistical difference between group means