Inheritance of DM resistance in cucumber was complex. Angelov showed that the resistance of DM in cucumber was controlled by a recessive gene [36] and this gene was linked with dull green fruit skin color related gene D [37]. Most studies have reported that DM was controlled by multiple genes in cucumber, such as a dominant susceptible gene and a recessive gene [38], a pair of dominant and recessive interacting genes [39], three recessive genes [9]. In our study, the results showed that DM resistance was controlled by multiple genes. The different results could be due to different materials and different environments. The core germplasm we used possesses the DM resistance with most variation.
In recent years, GWAS is widely used to detect genes in many species. Our study firstly used the GWAS to detect loci and genes for DM resistance in cucumber. Several loci distributed on all 7 chromosomes have been identified by different populations. Based on previous studies, two loci were mapped on chr1, chr2, chr3, and chr6, respectively, one loci on chr4 and chr7, three loci (dm5.1, dm5.2 and dm5.3) on chr5 [10, 13, 14, 28, 40–45]. In our study, a total of thirteen loci (dmG1.1, dmG1.2, dmG2.1, dmG2.2, dmG3.1, dmG4.1, dmG4.2, dmG5.1, dmG5.2, dmG6.1, dmG6.2, dmG7.1 and dmG7.2) distributed on 7 chromosomes were detected by CG lines. Compared with the previous QTLs, the loci on the Chr1, Chr3, Chr4, Chr5 and Chr6 had been previously reported. Two loci on the Chr2, Chr7 was new reported, respectively. (Table 1). Based on the loci listed in Table 1, eleven of thirteen loci were also identified in QTL mapping. The candidate gene, Csa5G606470, mapped in dmG5.2, had a SNP in the CDS. And the SNP variation result in amino acid changes. The candidate gene was the same as dm5.2 based on BSA-seq analysis in Zhang et al ’s study [41]. The variation was effect on the 3’ UTR, which could be due to different materials. Three loci could be identified in two years, which showed that the results in our study were consistent. Ten of thirteen loci could be identified in one year, which could be due to different environments. In our study, thirteen loci were consistent with previous studies, which showed that GWAS could detect more variations [46]. So, GWAS was efficient to detect DM resistance related QTLs.
DM resistance is a complex trait that involves in multiple genes and metabolic pathways. Some of the major loci were interacted with each other. Three major QTLs in Chr5 were negative additive effects [42]. Wang et al. (2016) detected the effect of four QTLs (dm2.1, dm4.1, dm5.1 and dm6.1). dm6.1 could be interacted with dm2.1, dm4.1 and dm5.1. And dm2.1 could be affected by dm5.1. In our study, the peak SNPs in dmG1.2, dmG2.1, dmG5.2 and dmG7.1 were applied to detect the effect code of each other. dmG1.2 interacted with dmG7.1 in the two experiments. dmG5.2 had effect on dmG1.2 and dmG7.1 in one season. dmG1.2 was independent of dmG2.1 and dmG5.2, and dmG2.1 was independent with dmG7.2.(Supplementary Fig. 2) These results showed that the DM resistance is a complex trait and with multiple gene interactions, which was consistent with previous studies [42, 47]. Therefore, GWAS is an effective strategy to identify complex traits.