Phenotypic variation
Continuous variation for percentage of symptomatic spikelets (PSS) was observed at the GWAS panel in both 2017 and 2018 growing seasons, from highly resistant (PSS<25%) to highly susceptible (PSS>75%)(Fig. 1). The disease symptom was more severe in 2018 growing season (Fig. 2a). Wheat cultivars from different provinces of China exhibited different levels of resistance to FHB(Fig. 2b). Cultivars from Hunan and Jiangsu provinces exhibited consistently highly resistant to FHB in two seasons, whereas cultivars from Shandong province showed the highest susceptibility.
Population structure analysis
To estimate the sub-populations of the 171 wheat cultivars, population structure analysis was performed using 1,676 polymorphic SNP markers distributing on 21 wheat chromosomes with r2 values >0.2. The results indicated that the cultivars could be separated into two sub-populations (K=2) (Fig. 3a,3b). Subgroup 1 consists of 99 cultivars, mainly comprising varieties from Anhui, Jiangsu, Henan, Shaanxi and Hunan; subgroup 2 consists 72 cultivars (Table S1), most of which were from Henan, Jiangsu, Shandong, Shanxi. Wheat cultivars from Anhui and Hunan were all clustered into subgroup 1.
Linkage disequilibrium (LD) analysis
The filtered markers from the 90K SNP genotyping arrays were used to calculated LD decay for the A, B, and D sub-genomes separately as well as the whole genome. 38.9% of all pairs of loci had significant LD (P < 0.001) with an average r2 of 0.281 from 23,556 polymorphic SNPs which distributed at the genome-wide level. The B sub-genome contained the largest number of significant markers (50.0%), followed by A (39.7%) and D (24.0%) sub-genomes. The highest LD decay distance was present in the D sub-genome and the lowest was found in the B sub-genome. The average LD decay distance was ~10.5 Mb for the whole genome and 10, 9.5, and 12 Mb for A, B, and D sub-genomes, respectively (Fig. 3c).
Marker-trait associations
Association analysis was conducted using PSS data across two years and 23,556 filtered markers. Altogether, 26 loci (88 MTAs, P < 10−3) with phenotypic variances explained (R2) ranging from 6.64-14.18% were identified across all of the chromosomes except for 2D, 6A, 6D and 7D (Fig. 4a, 4b). Among these, 41, 32 and 15 significant markers were located on the A, D and B sub-genomes, respectively (Fig. 4c; Table S2). More FHB MTAs were found on chromosomes 1A, 1D, 1B, 2A, 3B, 4A and 5D.
Twenty-eight MTAs located on chromosomes 1B (1), 4A (7), 5D (19) and 7A (1) were consistently identified in both seasons and could be considered as stable QTL ( Table 1 ), SNP GENE-0293_154 located on 1B explained 6.91-7.18% of phenotypic variances (R2). Seven and 19 SNPs located on 4AL and 5DL chromosomes explained phenotypic variances (R2) ranging from 9.36-11.63% and 8.11-14.18%, respectively. The SNP BobWhite_c22875_239 located on 7A could explain 8.12-8.53% of phenotypic variances (R2).
Due to the high level of LD in wheat, the SNP clusters identified on chromosomes 4AL (QFhb-4AL) from 621.85 Mb to 622.24 Mb and 5DL (QFhb-5DL) from 546.09 Mb to 547.27 Mb most likely represented chromosome regions containing significant FHB associated loci, respectively. Haplotype analyses of the associated markers revealed three haplotype groups (Fig. 5a), Haplotype 1 consisted of 149 cultivars with an average PSS of 48.92% over two years, in which 24 were resistant, 55 were moderately resistant, and 70 were susceptible. Haplotype 2 consisted of 19 cultivars with an average PSS of 19.94% over two years, and 12 of them were resistant and 7 were moderately resistant. Haplotype 3 comprised three resistant cultivars with an average PSS of 11.52%. The results indicated that other resistant genes also existed in the cultivars of Haplotype 1 (Table 2, S3). Interestingly, each haplotype contains wheat cultivars with same associated SNPs on both QFhb-4AL and QFhb-5DL simultaneously (Fig. 5b).