Genome-Wide Association Mapping and Genomic Predictions for Bacterial Fruit Blotch Resistance in the USDA <i>Citrullus amarus</i> Collection

<i>Acidovorax citrulli</i> infects seedlings, adult plants, and fruits, causing bacterial fruit blotch (BFB) in watermelon. Host resistance would provide an effective and economical management option for BFB, but there are currently no resistant watermelon cultivars. Several resistant accessions were previously identified in the U.S. Department of Agriculture's (USDA) <i>Citrullus amarus</i> collection. Identifying the genetic basis of this resistance would allow the development of BFB-resistant cultivars through introgression from this crop wild relative. Genome-wide association studies (GWAS) are an excellent tool for dissecting the genetic architecture of a trait. The USDA's <i>Citrullus amarus</i> collection (N = 127 accessions) was genotyped with whole genome resequencing, resulting in 2,126,759 single nucleotide polymorphism (SNP) markers, then phenotyped for BFB resistance and used for GWAS of seedling resistance to <i>A. citrulli</i>. Four models were used for GWAS in R with the GAPIT package. Mixed linear model and multilocus mixed linear model analysis did not identify any significant marker associations. Fixed and random circulating probability unification identified three quantitative trait nucleotides (QTN) on chromosomes 2, 4, and 8. Bayesian-information and linkage-disequilibrium iteratively nested keyway identified only one significant QTN on chromosome 8. The three significant QTNs explained 65.1% of the phenotypic variance using a linear regression model. Putative candidate genes within the linkage disequilibrium blocks of significant SNPs code proteins relevant to biotic resistance, such as Patellin-6, macrophage migration inhibitory factor homolog, prenylated Rab acceptor 1 family protein and trichome birefringence-like family proteins. The predictive ability of six genomic prediction models for <i>A. citrulli</i> seedling resistance ranged from 0.45 to 0.75. Along with identifying genomic regions associated with BFB seedling resistance, this study observed moderate to high predictive abilities across genomic prediction models.