Watermelon Team | 2024 Progress Report

View the 2024 Watermelon Report including all tables and figures in pages 20 – 36 of the pdf version of this report.

Watermelon Team members:

  • Amnon Levi (USDA, ARS)
  • Sandra Branham (Clemson University)
  • Shaker Kousik (USDA, ARS)
  • Kai-Shu Ling (USDA, ARS)
  • Cecilia McGregor (University of Georgia)
  • Umesh Reddy (West Virginia St University)
  • Pat Wechter (Clemson University)

1.2.1 Seed multiplication of core collections

Cecilia McGregor
University of Georgia

Thirty-nine S2 accessions (31 C. amarus, 7 C. lanatus and 1 C. mucosospermus) have been sequenced and seed are ready to be sent foe increase (waiting for LOU to be executed).

2.1 Map resistances and identify QTL for key cucurbit diseases

The WPop GSB1 (PI 482276 x Crimson Sweet) F2:3 population used for identification of Qgsb5.1 (syn. ClGSB5.1; Gimode et al., 2020) and Qgsb7.1 (syn. ClGSB7.1; Gimode et al., 2020) is being advanced to a RIL population. This RIL population will be used to identify QTL associated with stem resistance to S. citrulli.
F2 population WPop GSB3 from a cross between Crimson Sweet and PI 482379 (resistant) has been phenotyped. Khufu (Korani et al., 2021) will be used to identify QTL associated with resistance in this population. PI 482379 has not been previously used for QTL mapping or breeding.

2.2. Develop and verify markers for MAS

We previously developed and/or validated KASP marker assays for selection of Qgsb5.1 (syn. ClGSB5.1; Gimode et al., 2020), and Qgsb7.1 (syn. ClGSB7.1; Gimode et al., 2020). These QTL are currently being introgressed into Crimson Sweet. We evaluated 8 BC2F3 introgression lines as well as 6 genotypes obtained from North Carolina State University (Rivera-Burgos et al., 2021) in the field. Sugar Baby, Crimson Sweet and Fiesta were included as susceptible controls and UGA11 (selection from PI 482379), UGA81 (selection from PI 189225) and UGA 1081 (selection from PI 482276) were included as resistant controls. Plants were artificially inoculated with an S. citrulli isolate and leaf surfaces were kept wet using a mist system.

Fig.1 (a) Disease severity and Brix of introgression lines (grey) and genotypes form NCSU (blue) inoculated with S. citrulli in the field in summer 2023. Red triangles are average Brix for each line. (b) Disease severity of introgression line 2008_1_1 compared to controls.

Introgression lines 2008_1_1 and 2008_1_2 and NCSU-RIL-117 and NCSU-RIL-204 had significantly lower gummy stem blight disease severity (AUDPC) than susceptible controls. The average brix of these lines was 4.9, 7.4, 4.1 and 7.3, respectively. The BC3F3 for lines 2008_1_1 and 2008_1_2 are complete and will the evaluated in the field in 2024.


  • Gimode, W., K. Bao, Z. Fei and C. McGregor (2021) QTL Associated with Gummy Stem Blight Resistance in Watermelon. Theor Appl Genet 134:573–584.
  • Korani W, O’Connor D, Chu Y, Chavarro C, Ballen C, Guo B, Ozias-Akins P, Wright G, Clevenger J (2021) De novo QTL-seq Identifies Loci Linked to Blanchability in Peanut (Arachis hypogaea) and Refines Previously Identified QTL with Low Coverage Sequence. Agronomy 11:2201
  • Rivera-Burgos, L.A., E. Silverman, N. Sari and T.C. Wehner. (2021) Evaluation of Resistance to Gummy Stem Blight in a Population of Recombinant Inbred Lines of Watermelon × Citron. HortScience 1:1–9.

Obj 2. Map and develop markers for disease resistance

Sandra Branham, Patrick Wechter and Amnon Levi
Clemson University and USDA, ARS U.S. Vegetable Laboratory

2.1: Developing populations (P), phenotyping (Ph), QTL mapping (Q), Fine mapping (F) -Fon race 2, Ph/Q

  • Completed QTL mapping in the USVL246-FR2xUSVL114 RIL population and narrowed the QTL intervals found in the F2:3 population from the same cross. Developed KASP markers for four QTL and validated them in an independent interspecific (USVL246x’Sugar Baby’) population. Manuscript accepted with revision in Theoretical and Applied Genetics.
  • Completed disease screening (two replicated tests) of the C. amarus core collection for response to inoculation with Fon race 2 and used the phenotypes for GWAS and genomic predictions. Ganaparthi et al. 2023 Plant Disease.

2.1: Developing populations (P), phenotyping (Ph), QTL mapping (Q), Fine mapping (F) -Downy mildew, Ph/Q

  • Completed disease screening (two replicated tests) of the C. amarus core collection for DM resistance and used the phenotypes for GWAS. Katuuramu et al. 2022.

2.2: Develop marker (M), verify (V)

Fon race 2, M/V

  • KASP markers for Fon race 2 resistance were developed in the C. amarus 246×114 RIL population. An F2:3 interspecific population of USVL246-FR2 by ‘Sugar Baby’ was evaluated for response to inoculation with Fon race 2 in two replicated tests and the phenotypes used for QTL mapping with the KASP markers. Manuscript accepted with revision in Theoretical and Applied Genetics.

Powdery mildew race 2w, M/V

  • XP-GWAS of powdery mildew race 2 resistance was completed for the USDA Citrullus core collection using historical data. KASP markers were designed for three regions of the genome with a significant signal. They were validated in two hundred accessions from the extremes of the distribution. Manuscript in preparation.

Obj. 3A. Introgress, pyramid/stack resistances into advanced breeding lines

Develop breeding lines (B), introgress into cultivated (I), advanced lines (A), release to breeders (R)

Fon races 1 and 2


  • Phenotypic and marker-based selections for Fon race 2 resistance were made from the interspecific population of USVL246-FR2 by ‘Sugar Baby’ and selfed and backcrossed to ‘Sugar Baby’. We are currently making seed for the F5 and BC2F2 generations from this cross. We have also crossed USVL246 to ‘All Sweet’ and ‘Crimson Sweet’ to begin introgression into a variety of elite backgrounds (Table 1).

Figure: Strategy for introgression of 246 resistance alleles into “Sugarbaby” utilizing backcross and genomic selection

Table 1. Progress of introgression of Fon race 2 resistance into elite breeding lines

Figure 1. A BC1F1 genetic population (Charleston Gray) Charleston Gray] showing resistance to Fusarium wilt race 2.

Figure 2. A KASP marker for a QTL associated with Fusarium wilt race 2-resistance. Dual color scatter plot of a KASP marker differentiating genotypes into three clear clusters.

Obj. 3B. Using ‘genomic selection’ approach to incorporate Fusarium wilt race 2 resistance  into watermelon cultivars


Constructing and utilizing training populations for ‘genomic selection’ experiments

  • USVL246 x USVL114 (RIL), USVL252 x USVL119 (F3)  
  • USVL246 x Sugar Baby (F3), USVL 252 x Sugar Baby (F3; F4; BC1;F2;F3) 

Fig. 3. Genomic selection experiments for Fusarium wilt race 2-resistance and fruit quality traits at the USDA, ARS, U.S. Vegetable Laboratory and Clemson-Costal Research and Education Center (CREC), Charleston, SC (Summer 2023).

Seven genomic selection (GS) models were evaluated. G-BLUP and Random Forest performed best, achieving prediction accuracies. Genomic estimated breeding values (GEBV) effectively identified superior families at different selection intensities and captured all Fon race 2- associated QTLs at 30% intensity.

Figure: Developing small-seed lines with multiple disease resistance (FW races 1, 2 and Potyviruses, including ZYMV and PRSV) (BC6F1 and BC5F2 lines).

Cooperative Research and Development Agreement (CRADA)-Partnerships for constructing MAGIC Populations and Speed the Use of CucCAP Germplasm and Genomic Data by Seed Companies (Users)

CRADA was established and signed between USDA-ARS and BASF/Nunhems

Additional CRADAs for constructing the MAGIC populations are underway with Origene Seeds, Enza Zaden, Rijk Zwaan, Takii Seeds, and Sakata Seeds.

Cucurbits project summary report (2023- 2024). Watermelon resistance to CGMMV

Kai-Shu Ling, Bazgha Zia, and Amnon Levi
USDA, ARS, U.S. Vegetable Laboratory

Cucumber green mottle mosaic virus (CGMMV) is an emerging tobamovirus in North America. The control of this virus through breeding for natural resistance requires the identification of a new source of genetic resistance. In screening of the USDA watermelon germplasm, we have identified a source of resistance to CGMMV in a wild watermelon relative (Citrullus colocynthis L.). A segregating population of F2 libraries was generated through a cross between resistance (USVL#157) and susceptible (USVL#138) C. colocynthis lines. Phenotypic analysis through mechanical inoculation of the F2 population revealed a genetic segregation, suggest the existence of two gene model controlling the CGMMV resistance. Bulked segregant analysis (BSA) analysis was conducted by bulking two extreme phenotypes of the F2 generation as an effort to identify the SNPs associated with CGMMV resistance. A total of four SNPs and several candidate genes have been identified to be associated with CGMMV resistance in watermelon. These SNPs are now being validated by using the genome of the C. colocynthisline PI 537300. The mapping percentage ranged from 68% to 98% when mapped to the PI 537300 genome and 80% – 85% for the reference genome ‘Charleston gray’. The identified SNPs spanned the region at chromosome 3, 5, 6 and 10. Upon validation using the DEEPBSA tool, the SNP at chromosome 10 dropped below threshold. Therefore, to confirm our results we mapped it to a second genome, PI 537300. The mapping quality was improved. The BSA sequence analysis revealed a total of 27 genes surrounding these identified SNP regions. Four of these genes were potential candidate genes based on the genomic position and functional annotation. These genes encoded for the U-box domain-containing protein, NB-ARC domain containing protein, Serine hydroxymethyltransferase and ABC transporter D family member, respectively.

Figure 1. Genomic position of significant SNPs associated with resistance to the CGMMV pathogen in Watermelon.

Table 1 Summary of sequencing data

Table 2. Identified SNPs identified with Reference genome Charleston Gray associated with resistance to the CGMMV infection in Watermelon.

Powdery mildew of watermelon.

Shaker Kousik, Rahul Kumar, Amnon Levi; USDA, ARS, U.S. vegetable Laboratory (USVL), Charleston, SC

  • Completed evaluation of commercial seedless varieties for resistance to powdery mildew in 2021 and 2022. Several resistant seedless watermelon lines were identified, and detailed results are presented in Extension section of the report in 2023.
  • Publicly released USVL531-MDR. Manuscript was published in HortScience. The cover Page for the April 2023 HortScience issue displays resistance to Powdery mildew in USVL531-MDR compared to USVL677-PMS.
  • A F2 population (USVL608-PMR X USVL677-PMS) was used for the QTLs mapping using the bulk segregant analysis method and a genomic region was mapped on the chromosome 2 and KASP markers were designed to find linked markers.
  • Three F2 populations (USVL608-PMR X USVL677-PMS, USVL608-PMR X Sugar baby, USVL608-PMR X Dixie Lee) were phenotyped for resistance to powdery mildew in a growth chamber after inoculation. Data on powdery mildew development on each F2 progeny was collected.
  • KASP markers were used for the genotyping of F2 progenies (96 for each population) and homozygous resistant progenies were selected.
  • Different KASP Markers worked well with the different susceptible parent.
  • The KASP markers were tightly linked with the Powdery mildew phenotypic rection.
  • F3 progenies were produced from the selected homozygous resistant plants for the confirmation linkage with markers and phenotype.
  • All the F3 progenies of three populations showed resistance to powdery mildew and genotyping of plants confirmed the presence of homozygous resistant alleles in all progenies.
  • Advanced lines are being developed for the public release from each of the three crosses.
  • Evaluated and collected data on powdery mildew development on RIL lines in the field (summer 2022, Summer and Fall 2023). Five plants of each RIL line were planted per plot and each RIL line had two replications. Data has been collected over three field season and is being analyzed.
  • DNA was extracted from all the 187 RIL Lines. DNA bulk of powdery mildew resistant and susceptible lines were sent for sequencing to North Carolina State University Genomics Center. DNA of all the RIL Lines will be sequenced in collaboration with University of Ilinois.
  • Advanced RIL lines with PM resistance and red flesh to develop useable resistant germplasm lines. The advanced RIL lines had the KASP marker based on the NBS-LRR gene in watermelon Chr02, ClaPMR2 that is tightly linked with PM resistance.
  • Publicly released USVL531-MDR. Manuscript submitted to HortScience has been accepted and was published in the April 2023 issue. The cover Page for the April 2023 issue displays resistance to Powdery mildew in USVL531-MDR compared to USVL677- PMS.

Figure 1. Genotyping of three F2 populations USVL608-PMR (resistant) X USVL677-PMS (susceptible), USVL608-PMR X Sugar baby (susceptible), USVL608-PMR X Dixie Lee (susceptible) using KASP markers. KASP markers were developed from one major QTL in watermelon Chr02. There was a common KASP marker for USVL608-PMR X Sugar baby, USVL608-PMR X Dixie Lee population and another marker for USVL608-PMR X USVL677-PMS population.

Figure 2. Genotyping of three F3 populations of USVL608-PMR X Sugar baby (susceptible), USVL608-PMR X Dixie Lee (susceptible) using KASP markers. Phenotyping and genotyping confirmed that all the progenies were homozygous resistant.

Figure 3. Phenotypic reaction of parents used in the crosses to develop powdery mildew resistant watermelon lines with high quality. The varieties Sugar Baby and Dixie Lee with high Brix and good fruit quality were crossed with USVL608-PMR to develop F2 and F3 lines.

Phytophthora fruit rot of watermelon

Figure 4. Phytophthora fruit rot resistant RIL Lines with red flesh selected from the cross of 677 type USVL531-MDR X USVL677-PMS based on evaluation in Fort Pierce, FL. June 2023. Fruit on left from USVL-677-PMS displays significant development of Phytophthora fruit rot compared to the fruit on right which is resistant.

  • Three advanced lines (F11) lines developed from the cross between USVL531-MDR (resistant) and USVL677-PMS (susceptible) were evaluated at in Fort Pierce, FL for phytophthora fruit rot resistance and horticultural traits (Brix value, flesh color, rind thickness, fruit color, and weight.
  • One hundred eight seven advanced RIL lines were evaluated at the two different locations 531 type in Charleston for the phytophthora fruit rot resistance and horticultural traits during summer and fall of 2023.
  • Fruits were harvested at maturity and envaulted for phytophthora fruit rot resistance under control conditions in a large walk-in-growth chamber.
  • Fruits were evaluated for fruit color, rind thickness, flesh color, fruit firmness, taste (bitter or sweet), brix value, seed color, and seed type (egusi or non-egusi).
  • Under field conditions, RILs were evaluated for leaf type, flowering date, and fruiting date.
  • DNA was extracted from 187 advanced RIL lines for whole genome sequencing.
  • Bulks were made from the phytophthora fruit rot and powdery mildew resistant and susceptible progenies for the QTLs mapping. Bulks have been sequenced and QTLs mapping analysis is being done.
  • KASP markers will be developed for phytophthora fruit rot and powdery mildew resistance.
  • Evaluated F2 and F2:3 population of USVL003-MDR (Citrullus mucosospermus) X Dixie Lee (C. lanatus, cultivated type with good horticultural traits). QTLseq analysis indicated significantly associated QTLs with Phytophthora fruit rot resistance in Chr04, Chr07 and Chr10.
  • Advanced, red-fleshed resistant Phytophthora fruit rot resistant lines (USVL003-MDR x Dixie Lee) after screening and selection.

Figure 5. Extreme variability in fruit flesh color, seed type (Egusi or non-egusi), rind thickness, flesh firmness and fruit size, and leaf type were observed among the RIL lines (F11) developed from the cross of USVL531-MDR and USVL677-PMS. Data has been collected on all these traits and is being analyzed along with the genotypic data. Geneotype data is being developed whole genome sequencing of all 187 RIL Lines.

Umesh Reddy

West Virginia State University

Whole genome re-sequencing and analysis of 161 diverse watermelon accessions

The watermelon whole genome re-sequencing project, conducted on 161 watermelon accessions, has produced a vast dataset of high-density genetic variants by employing state-of-the art genomic techniques. Each accession was subjected to deep sequencing, achieving over 30x coverage, which ensures a comprehensive representation of the genomic diversity within the sample population. Leveraging high-throughput sequencing technologies, the study aimed to capture a broad spectrum of genetic variation by mapping sequenced reads to two reference genomes. This genomic insight is pivotal for genome-wide association studies (GWAS) and marker-assisted breeding programs.

Methodology: The experiment began with the germination of seeds in a controlled greenhouse environment, leading to the collection of leaf material that was snap-frozen in liquid nitrogen. This immediate preservation is crucial to halt any enzymatic processes that could degrade the nucleic acids. Genomic DNA was subsequently extracted using the DNeasy Plant Pro kit (Qiagen, USA). Agarose gel electrophoresis and Qubit fluorimeter were used to assess the quality and quantity of the genomic DNA, respectively, ensuring that only high-quality genomic DNA proceeded to the library preparation stage for Illumina sequencing. The Illumina DNA sequencing libraries prepared from the genome DNA were sequenced on the NovaSeq platform (Illumina, USA), yielding a massive data trove of raw reads. For each accession, the depth of sequencing exceeded 30x coverage, far surpassing the standard required for accurate variant calling. The raw reads were subjected to stringent quality control before being aligned to two distinct reference genomes: Charleston Grey (Citrullus lanatus) and USVL246 (Citrullus amarus), using the BWA-MEM algorithm. SAMtools facilitated the subsequent sorting and indexing of reads, and duplicate reads were marked using Picard tools.

Variant calling was accomplished using the GATK Haplotype Caller, which is widely recognized for its robustness in detecting genomic variants. Following this, a series of quality filters were applied, tailored to exclude any variant not meeting the stringent quality parameters as follows: for SNPs, a Quality by Depth (QD) less than 2.0, Fisher Strand (FS) greater than 60.0, Mapping Quality (MQ) less than 40.0, MQRankSum less than -12.5, and ReadPosRankSum less than -8.0; for indels, a QD less than 2.0, FS greater than 200.0, and ReadPosRankSum less than –
20.0. These thresholds are critical in ensuring that only the most reliable variants are carried forward in the analysis. Following the hard filtering step, the SNPs underwent further filtration to retain only biallelic forms. These biallelic SNPs were then subject to additional filtering based on minor allele frequency (MAF) of 0.05 and a call rate of 70%, which are criteria indicative of reliable and informative genetic markers for downstream analysis.

The project’s sequencing efforts resulted in a staggering number of raw reads (11.9 billion), specifically 11,970,459,092 in total for 161 accessions, which after quality filtering, retained an impressive total of 11,423,749,213 reads (as shown in Table 1). Remarkably, the percentage of these reads uniquely mapped to the reference genomes was more than 99% for both, indicating the high fidelity of the data and reflecting the high precision of the alignment process.

Table 1 presents a comprehensive summary of the sequencing and variant calling efforts. In the table, it is noteworthy that the USVL246 genome mapping revealed a greater number of variants compared to the Charleston Grey genome, which may indicate a broader genetic diversity within the Citrullus amarus species or reflect the genetic diversity of the accession set.

Table 1. Summary of watermelon re-sequencing, genome mapping and genomic variants.

The total number of variants identified was higher when aligned to the USVL246 genome, with 18,585,670 variants, compared to 12,991,397 for the Charleston Grey genome. Notably, the number of SNPs called from the USVL246 genome was approximately 4.7 million more than from the Charleston Grey genome. Similarly, indels identified in the USVL246 genome exceeded those in the Charleston Grey by nearly 750,000. Quality filtration of these variants resulted in 8,455,555 high-confidence SNPs for Charleston Grey and 9,512,704 for USVL246. The number of quality
filtered indels was 2,104,199 for Charleston Grey and 2,746,997 for USVL246. Post filtration, 8,249,596 biallelic SNPs for Charleston Grey and 9,134,054 for USVL246 were retained. When considering biallelic SNPs that met the stringent MAF and call rate thresholds, 47,220 were identified in the Charleston Grey genome and an impressive 484,314 in the USVL246 genome.

The transition/transversion ratio (ts/tv), an indicator of the genetic variation quality, was marginally higher in the USVL246 mapping (2.06) compared to the Charleston Grey genome (2.05). The number of transitions (ts) stood at 7,390,561 for Charleston Grey and significantly increased to 10,990,744 for USVL246. Conversely, the number of transversions (tv) was also higher in the USVL246 genome, with 5,332,990 compared to 3,611,101 for Charleston Grey.

Future Directions: The variants identified through this comprehensive and high-coverage approach will be further utilized for genome-wide association studies (GWAS) in conjunction with various phenotypic data. This will enable the investigation of the genetic basis of important traits in watermelon, facilitating the identification of candidate genes and markers for breeding programs.

Developing a Multi-Parent Advanced Generation Intercross (MAGIC) Population Useful for Enhancing the Watermelon Germplasm and for identification of gene loci associated with Disease Resistance

Watermelon CucCAP2 Team in Collaboration with Seed Companies

Amnon Levi, Shaker Kousik, Cecilia McGregor, Sandra Branham, Patrick Wechter, Zhangjun Fei, Umesh Reddy, and Kai-shu Ling

Table 1. MAGIC Population-Founder Lines (United States Plant Introduction; PIs and USVL lines) with disease, potyvirus (PRSV, ZYMV, SqVYV) or root-knot nematode resistance. R-resistant (tolerant), MR-moderate resistance (tolerance) S-susceptible

Two MAGIC populations are under construction

MAGIC populations are at F2 stage and will be continued to F8/F9 generations in collaboration with seed companies, with the objective to have 500 F8/F9 RIL lines for each population.

MAGIC-1 (22005) Pedigree
21171 x 21159
21058-1♂ x 21122-1♀ 21171
21061-2♂ x 21015-1♀ 21159
[USVL 531 x PI 269677]-2♂ x [Crimson Sweet x UGA 1081]-5♀ 21058 X
[USVL 246 x Sugar Baby]-6♂ x [Calhoun Gray x PI 595203]-4♀ 21122
[Hungarian x USVL252]-4♂ x [PI 392291 x Mickylee]-3♀ 21061 X
[Klondike Black Seeded x NH Midget]-3♂ x [Dixie Lee x PI 244019]-1♀ 21015

MAGIC-2 (22002) Pedigree
21154 x 21168
21090-1♂ x 21086-3♀ 21154
21056-3♂ x 21047-1♀ 21168
[PI 392291 x Mickylee]-3♂ x [Crimson Sweet x PI 244019]-3♀ 21090 X
[NH Midget x Calhoun Gray]-1♂ x [Hungarian x USVL252]-3♀ 21086

[USVL 246 x Sugar Baby]-2♂ x [USVL 531 x PI 269677]-1♀ X 21056

[Jenny x PI 595203]-1♂ x [PI 189225 x PI 279461]-1♀ 21047

Collaboration With Seed Companies in Developing the MAGIC Populations for Watermelon

Cooperative research and development agreement (CRADA)-Partnerships to speed transfer of germplasm and technology to Users (Seed Companies)

Cooperative Research and Development Agreement (CRADA) was signed between USDA, ARS and BASF- Nunhems (December 2023).

Additional CRADAs are underway with Origene Seeds, Enza Zaden, Rijk Zwaan, Takii Seeds, and Sakata Seeds.