One-hundred thirty-five RILs used by Wu et al. (2021) for mapping the QTLs associated with partial resistance to Aphanomyces root rot were included in this study. In brief, the Aphanomyces root rot-resistant pea parent “00-2067” developed by Dr. J. Kraft and V. A. Coffman at the Irrigated Agriculture Research and Extension Center in Prosser, WA, United States (Conner et al., 2013; Wu et al., 2021), was used in genetic crosses with the susceptible parent “Reward” (Bing et al., 2006) to produce F₁ plants, which were then used to develop an F₈ RIL population (Supplementary Figure 1) by the single-seed descent (SSD) method (Brim, 1966).

Five single-spore isolates (SSI), S4C (F. solani), F4A (F. avenaceum), F037 (F. acuminatum), F039 (F. proliferatum), and FG2 (F. graminearum), representing the Fusarium species most frequently recovered from symptomatic pea plants in root rot surveys in Alberta, were used to screen the parental cultivars “00-2067” and “Reward.” Briefly, to obtain the SSIs, surface-sterilized pieces of root tissue with disease lesions were placed on potato dextrose agar (PDA) and incubated at 25°C for 2–3 days and then transferred to the peptone-pentachloronitrobenzene (PCNB) medium for further selection. Mycelial tips of the fungal isolates were cut from selected colonies under a stereomicroscope (Zeiss Axio Scope A1, Carl Zeiss Canada Ltd., Canada), and the water agar (WA) procedure was used to obtain SSI (Zitnick-Anderson et al., 2020). The species designation of each of the SSIs was confirmed based on morphology and evaluation with the PCR primer sets ITS4/ITS5 and EF-1/EF-2, while isolate virulence was confirmed by fulfilling Koch’s postulates (Feng et al., 2010; Chen et al., 2014; Zhou et al., 2014; Wu et al., 2017; Chang et al., 2018; Zitnick-Anderson et al., 2018).

Conidial suspensions of the five isolates were generated following Son et al. (2013). Pure cultures of each Fusarium spp. were grown in Petri dishes on PDA under darkness at room temperature for 4–6 weeks. Sterile-distilled water was added to each Petri dish, and the surface of each colony was gently scraped with a sterile inoculation needle to dislodge the spores (and hyphal fragments), with the resulting suspension decanted into 250-ml Erlenmeyer flasks, containing a 100-ml autoclaved CMC medium (1.5% carboxymethyl cellulose, 0.1% yeast extract, 0.05% MgSO₄⋅7H₂O, 0.1% NH₄NO₃, 0.1% KH₂PO₄, 100-ml H₂O). The flasks were covered in aluminum foil to block light and incubated on a rotary shaker at room temperature for 2 weeks. The suspension was centrifuged to collect conidia. The concentration of conidia was estimated with a hemocytometer and adjusted to a final concentration of 2 × 10⁶ spores ml^–1 with sterile-deionized water.

Plastic cups (9-cm diameter and 10.5-cm depth) were filled with a sterilized potting mixture (Cell-TechTM, Monsanto, Winnipeg, MB, Canada). In the greenhouse tests with each SSI (S4C, F4A, F037, F039, and FG2), the roots of seven 5-day-old seedlings of the partially resistant parent “00-2067” and the susceptible parent “Reward” were immersed in the conidial suspension for 15 min and transplanted into the soilless mixture in a cup. An aliquot (1 ml) of conidial suspension was pipetted onto the roots before they were covered with the potting mix. The plants were kept in a greenhouse at 20–25°C/15–18°C day/night and a 16-h photoperiod with daily watering to maintain the potting mix at saturation conditions conducive for FRR development. Each experiment was repeated two times. After 3 weeks, disease severity was estimated for the parental cultivars on a scale of 0–4, where: 0 = completely healthy; 1 = brown or black spots on the main root; 2 = lesions covering the main root, but the rootlets still healthy; 3 = lesions spread to the entire root system; and 4 = root totally dead (Figures 1a,c).

The most virulent of the Fusarium isolates was used as inoculum to screen the 135 RIL population and parents under greenhouse conditions. The inoculation and the maintenance of the plants were as described above. The pots were arranged in a randomized complete block design (RCBD) with four replicates. The greenhouse experiment was repeated four times. After 3 weeks, plant height (cm) was measured from the base of the stem to the top leaf. Plant vigor was evaluated as a measure of the wilting severity on a scale of 0–4 (4 = plant completely healthy; 3 = thin stem and short height; 2 = brown lesions on stem and yellowing of leaf tips; 1= wilting on stems and leaves; 0 = plant completely dead) (Figure 1b). The plants were then carefully uprooted, washed under standing water, and assessed for disease severity as described above.

ANOVA was conducted using R software (R Core Team, 2019) for disease severity, vigor, and plant height in four greenhouse environments. The mean and least square mean (LSM) of all traits were calculated for single environments and total data using the package “lsmeans” (Lenth, 2016) in R. To estimate random effects, the best linear unbiased predictors (BLUPs) and heritability were also calculated using the package “Phenotype” (Zhao, 2020) in R. Correlation analysis was conducted within each trait (all variables including means for single environments, LSM, and BLUPs for total data) and among traits (including disease severity, vigor, and plant height) using the package “PerformanceAnalytics” (Peterson et al., 2019) in R, displaying the correlation coefficient, frequency distribution, and dot plot. The P-value of the Shapiro–Wilk test was used to determine the normality for each variable using R (R Core Team, 2019).

The 13.2K SNP markers and 222 SSR markers, the parents, and the RIL population genotyped by Wu et al. (2021) were used in this study. In brief, SNP genotyping was carried out at TraitGenetics GmbH, Gatersleben, Germany, using an SNP array developed from gene-encoding sequences, which are distributed uniformly across the pea genome (Tayeh et al., 2015). The SSR markers were obtained from Loridon et al. (2005). In the case of the SSR markers, the PCR assays, thermal cycling conditions, and genotyping using an ABI PRISM 3730 x l DNA analyzer (Applied Biosystems, Foster City, CA, United States) were as described by Wu et al. (2021). Filtering of the SNP and SSR was carried out to retain highly polymorphic markers and RIL individuals with > 95% genotyping data, as well as markers that exhibited the expected 1:1 segregation ratio.

Linkage analysis was carried out using the filtered SNP and SSR markers, following Wu et al. (2021). This involved the generation of a draft linkage map using the minimum spanning tree map (MSTMap) (Wu et al., 2008) and then refined by MAPMAKER/EXP 3.0 (Lincoln et al., 1992). The Kosambi map function (Kosambi, 1944) was used to calculate the genetic distances (in cM) between the markers. The map construction was carried out with MapChart v. 2.32 (Voorrips, 2002) using the Kosambi map function, of which the linkage groups were assigned to chromosomes based on the consensus SNP map of pea developed by Tayeh et al. (2015). The sequences of the SNP markers flanking the QTLs associated with partial resistance to FRR caused by F. graminearum were used in BlastN (E-value ≤ E-20) searches of the Pulse Crop Database¹ to determine their possible functions.

Additive-effect QTL analysis was first carried out using the genotypic and phenotypic data (disease severity, vigor, and plant height) from the RILs inoculated with F. graminearum (FG2). This was then repeated for the RILs inoculated with F. avenaceum (F4A). The analysis was conducted using means for the four single greenhouse experiments, LSM, and BLUPs of the total data by Composite Interval Mapping (CIM) using WinQTL Cartographer v2.5 (Wang et al., 2012). The program was set at 1-cM walking speed; forward and backward regression method; window size, 10 cM; five background cofactors; 1,000 permutations, and p < 0.05 (Wang et al., 2012). The LOD score threshold was set at 3 for QTL detection. The confidence interval for each QTL was defined by the consensus region bordered by the four environments.

The QTL names were defined according to the QTL detection studies by Coyne et al. (2015, 2019), where the name of the Fusarium isolate was indicated, followed by “Ps” = Pisum sativum, the first number = the pea linkage group (Tayeh et al., 2015), and the second number = the serial number of the QTL on the linkage group; for example, “Fg-Ps4.1” represents the QTL for disease severity caused by F. graminearum located on linkage group IV of the pea genome. The chromosomes and pseudomolecules were named in accordance with Neumann et al. (2002) and Kreplak et al. (2019), respectively. A similar nomenclature was adopted for vigor (e.g., Vig-Ps2.1) and plant height (e.g., Hgt-Ps2.1).

Quantitative trait loci identified in at least two of the four environments were classified as stable. The percentage of variation (R²) was determined for each QTL. Furthermore, QTL with R² > 10%, 5–10%, and <5% were arbitrarily classified as major-, moderate-, or minor-effect QTL, respectively. The origins of favorable alleles for individual traits were assigned to different parents, following Wu et al. (2021). Pairwise epistatic interactions were estimated with IciMapping V.4.1 using the ICIM-EPI method (Meng et al., 2015). The significance threshold for major, moderate, and minor was arbitrarily set at R² > 15%, 7.5–15%, and <7.5%, respectively. Epistatic-effect QTL were named with the prefix “E,” followed by the QTL name and a serial number (e.g., E.FG-Ps1, E.Vig-Ps1, and E.Hgt-Ps1).

Between the parental cultivars, “00-2067” developed lower root rot severity than “Reward” in response to each of the five isolates (Supplementary Table 1), confirming that “00-2067” was tolerant, while “Reward” was susceptible. There were significant differences (p < 0.001) between the mean root rot values of the tolerant parent “00-2067” and the susceptible parent “Reward”, following inoculation with F. graminearum isolate FG2 (Figure 1a) and F. avenaceum isolate F4A, while no significant differences were detected following inoculation with the F. solani, F. acuminatum, and F. proliferatum isolates S4C, F037, and F039, respectively (Supplementary Table 1). Therefore, FG2 and F4A were selected to screen the 135 F₈ RIL population for QTL identification associated with resistance to FRR.

The mean root rot severity, vigor, and plant height of the RIL population inoculated with FG2 and F4A are presented in Tables 1, 2. ANOVA indicated that the genotypic effect of disease severity, vigor, and plant height was significant (p < 0.001) (Supplementary Tables 2a,b). This suggested that a high proportion of genetic variance was transmitted from parental cultivars to the progenies. Heritability values of 92 and 86% for disease severity and vigor were obtained for plants inoculated with FG2 and F4A, respectively, while heritability values for plant height ranged from 79 to 91% (Supplementary Tables 2a,b). The G × E interactions were significant for disease severity, vigor, and plant height for F4A but not for FG2, while differences among the four greenhouse experiments were significant for both FG2 and F4A (p < 0.001) (Supplementary Tables 2a,b).

Estimated disease severity values (±SE) on the parental cultivar “00-2067” inoculated with FG2 were 1.5 ± 0.7, 1.3 ± 0.6, 2 ± 1.2 and 1.5 ± 0.7 for the four greenhouse experiments, 1.6 ± 0.8 for LSM and 1.2 for the BLUPs. This was comparable with the estimated mean of 1.1 ± 0.4 obtained in the preliminary screening of the parents (Table 1 and Supplementary Table 1). On the other hand, the estimated disease severity values (±SE) for “Reward” were 3.3 ± 0.5, 3.3 ± 0.5, 3.5 ± 0.6, and 3.0 ± 0.0 for the four greenhouse experiments, 3.3 ± 0.5 for LSM and 4.1 for the BLUPs; these values were also comparable to the estimated mean of 3.3 ± 0.4 obtained in the preliminary screening (Table 1 and Supplementary Table 1). A t-test indicated a significant difference between the parents for disease severity in all four experiments. Frequency distribution (Figure 2A) indicated that the disease severity data of the RILs in the four experiments were continuous, but only DSGH3 and DSGHC followed a normal distribution based on the Shapiro–Wilk test (Table 1). High correlation coefficients, ranging from 68 to 99%, were found for disease severity among the single experiments, pooled, and BLUPs (Figure 2A). The differences in vigor between the parents inoculated with FG2 were significant, except for VGH4. The parental cultivar, “00-2067” had estimated means (±SE) of 4.0 ± 0.0, 4.0 ± 0.0, 3.0 ± 1.2 and 3.5 ± 0.7 for the four greenhouse experiments and 3.6 ± 0.8 for the pooled data. In the case of “Reward”, the estimated means (±SE) were 2.0 ± 0.8, 2.5 ± 0.6, 1.5 ± 0.6 and 2.7 ± 0.6 for the four greenhouse experiments and 2.2 ± 0.7 for the pooled data. The BLUPs for the parental cultivars “00-2067” and “Reward” were 4.2 and 1.6, respectively (Table 1). The Shapiro–Wilk test indicated that the RIL population vigor data for the four greenhouse experiments did not follow a normal distribution, except for VGHC (Figure 2B). A significant correlation (0.34 < r < 0.96, p < 0.001) existed among the single experiments, pooled, and BLUPs for vigor (Figure 2B). The height of “00-2067” plants inoculated with FG2 was greater than plants of “Reward” for the means in the single environments, LSM, and BLUPs, although the differences were not significant based on a t-test. The estimated means in single conditions, LSM, and BLUP for plant height (±SE) of “00-2067” were 234.5 ± 54.6 cm, 157.3 ± 50.6 cm, 155.5 ± 59.5 cm, 159.7 ± 6.7 cm, 176.7 ± 56.6 cm and 158.9 cm, respectively. For “Reward”, the plant heights were 177.5 ± 36.1 cm, 120.7 ± 31.5 cm, 129.5 ± 26. cm, 178.5 ± 34.6 cm, 151.5 ± 36.9 cm, and 100.8 cm, respectively. The frequency distribution of plant height of the RIL population for all six variables was not normal and slightly skewed (Table 1 and Figure 2C). A high correlation (0.42 < r < 0.95, p < 0.001) was found for plant height among the single experiments, pooled, and BLUPs data (Figure 2C).

Collectively, the correlation analysis among traits indicated that root rot caused by FG2 was negatively correlated with vigor and plant height. High correlation coefficients were detected between disease severity and vigor in all conditions (–0.65 < r < –0.90, p < 0.001), indicating the adverse effect of FG2 on root and aboveground growth. Plant height showed low to moderate correlation with disease severity (–0.22 < r < –0.35, p < 0.05) and vigor (0.19 < r < 0.38, p < 0.05).

The estimated means (±SE) of disease severity for “00-2067” were 1.0 ± 0.0, 1.0 ± 0.8, 1.3 ± 0.5, 1.0 ± 0.0, 1.1 ± 0.4, and 1.0, while, for “Reward”, they were 3.3 ± 0.5, 3.3 ± 0.5, 3.5 ± 0.6, 3.0 ± 0.0, 3.3 ± 0.4, and 3.0 for DSGH1, DSGH2, DSGH3, DSGH4, LSM of pooled data, and BLUPs, respectively (Table 2). These values were comparable to the estimated means (±SE) of 1.8 ± 0.5 and 2.8 ± 0.2 for disease severity obtained in the preliminary screening of the parents (Table 2 and Supplementary Table 1). t-tests indicated significant differences between estimated means of the parental cultivars “00-2067” and “Reward” inoculated with F4A. The Shapiro–Wilk test indicated that only the root rot data of the RIL population for DSGH4 and DSGH Pooled followed a normal distribution (Table 2), although the data for the four greenhouse experiments were continuous (Figure 3A). The correlation coefficient between the experiments ranged from 0.44 to 0.93 (p < 0.001) (Figure 3A). Based on the t-tests, the parental cultivar “00-2067” inoculated with F4A had significantly greater vigor than “Reward.” The estimated vigor values (±SE) for “00-2067” were 4.0 ± 0.0, 3.7 ± 0.5, 3.5 ± 0.6, and 4 ± 0 for the four individual greenhouse experiments, 3.8 ± 0.5 for LSM for the pooled data and 4.0 for BULPs of the pooled greenhouse experiments (Table 2). The estimated vigor values (±SE) for “Reward” were 1.7 ± 0.5, 2.0 ± 1.4, 1.2 ± 1.5, and 2.5 ± 0.6 for the individual greenhouse experiments, 1.9 ± 1.1 for LSM, and 1.9 for the BULPs of the pooled greenhouse experiments. All vigor variables for the RIL population were continuous with slight left skewness (–0.4∼–0.9) (Figure 3B). Additionally, the data did not follow a normal distribution based on the Shapiro–Wilk test (Table 2). The correlation coefficient between the experiments ranged from 0.54 to 0.98 (p < 0.001) (Figure 3B).

In contrast to vigor, the difference in plant height of the parental cultivars inoculated with F4A was not significant based on the t-test. The estimated plant height for “00-2067” for the individual experiments, LSM, and BLUP was 210.8 ± 128.2 cm, 174.5 ± 104.8 cm, 159.5 ± 13.5 cm, 210.0 ± 53.1 cm, 188.7 ± 82.8 cm, and 208.0 cm, respectively. The estimated plant height for “Reward” was 118.3 ± 100.2 cm, 193.5 ± 104.5 cm, 125. ± 98.9 cm, and 194.0 ± 38.4 cm for the individual experiments, 157.7 ± 88.5 cm for LSM and 133.1 cm for the BLUP. The frequency distribution for the RIL population was continuous and slightly skewed to the right. In addition, HGH2, HGH3, HGH Pooled, and HGH BLUPs followed a normal distribution (Table 2 and Figure 3C). Plant height variables were also significantly correlated (0.28 < r < 0.97, p < 0.01) (Figure 3C).

The correlation among the traits for plants inoculated with F4A was similar to that of plants inoculated with FG2. Disease severity was highly correlated with vigor (–0.88 < r < –0.95, p < 0.001) and with plant height (–0.48 < r < –0.63, p < 0.001). Plant height was positively correlated with vigor (0.57 < r < 0.62, p < 0.001).

Linkage grouping, the distribution of markers, map length, and marker density of 2999 (2978 SNP + 21 SSR) retained markers were as described by Wu et al. (2021). The marker distribution in this study was compared with the seven chromosomes of pea as determined by Neumann et al. (2002), linkage groups as determined by Tayeh et al. (2015), and pseudomolecules of pea (Kreplak et al., 2019). The genetic map spanned 1704.1 cM and contained an average marker density of 1.8 markers/cM (Wu et al., 2021). The QTL analysis was conducted with 1,422 unique markers, which represented 10.5% of the markers used for genotyping (Wu et al., 2021).

No significant QTL (LOD > 3.0) for disease severity, vigor, and plant height were detected for the RILs inoculated with F. avenaceum isolate F4A. As such, no QTL likelihood profiles are shown. In the case of RILs inoculated with F. graminearum isolate FG2, a total of 11 QTL were detected for the three parameters and six variables (i.e., GH1, GH2, GH3, GH4, LSM, and BLUPs) by the CIM using Win QTL Cartographer v2.5 (Wang et al., 2012; Table 3). Five of the 11 QTL were identified for disease severity, whereas three QTL each were detected for vigor and plant height. The QTL had LOD scores ranging from 3.0 to 14.4 and the percentage of phenotypic variation (R²) values ranging from 4.05 to 36.35% (Table 3). Based on the R² values, two, six, and three of the QTL were considered major, moderate, or minor effect, respectively. Six of the 11 QTL were identified in two or more environments and hence could be considered stable, while the remaining five QTL were detected in single experiments and hence could be considered unstable.

The most stable QTL for partial resistance to F. graminearum isolate FG2, Fg-Ps4.1, and Fg-Ps4.2 were located in the middle of Chrom4/LGIV at positions 59.3–74.4 cM and 74.0–85.2 cM, respectively (Figure 4B). The 15.1-cM and 11.2-cM genomic regions delimiting these two QTL were flanked by the SNP markers PsCam048871_31524_450 and PsCam001381_1152_437 and the SSR marker AA239 and SNP marker PsCam057281_37909_2940, respectively (Figure 4B). Both Fg-Ps4.1 and Fg-Ps4.2 exhibited a moderate effect, with the percentage variance ranging from 9.1 to 15.4% (Table 3). Two other moderate-effect but unstable QTLs, Fg-Ps3.1 (located on the bottom segment (307.9–316.5 cM) of Chrom5/LGIII and with flanking markers of AA5 and PsCam036163_21311_1095) and Fg-Ps3.2 (located distal to Fg-Ps3.1 and with flanking markers PsCam036163_21311_1095 and PsCam042783_26826_1395) explained 9.62–9.88% of the total variance (Figure 4A). Another unstable QTL, Fg-Ps5.1 [detected on the top part (0.9–9.2 cM) of Chrom3/LGV and flanked by the SNP markers PsCam059449_39630_321 and PsCam011153_7569_125] explained 14.2% of the total variance in greenhouse Experiment 1 (Figure 4C). Four of the QTL for disease severity (with the exception of Fg-Ps5.1) had a negative additive effect, indicating that genomic regions for resistance in Fg-Ps4.1, Fg-Ps4.2, Fg-Ps3.1, and Fg-Ps3.2 originated from “00-2067,” while Fg-ps5.1 derived its resistance from “Reward” (Table 3).

The stability of the QTL for vigor was in the order Vig-Ps4.1 on Chrom4/LGIV (GH1, GH2, and GH4, R² = 9.19 to 13.5%) > Vig-Ps3.2 (GH2 and GH3, R² = 9.53% to 12.13%) > Vig-Ps3.1 (GH4, R² = 4.05%) both on Chrom5/LGIII (Table 3). The QTL Vig-Ps4.1 was located on Chrom4/LGIV from 58.0 cM to 73.2 cm between the SNP marker PsCam000712_620_237 and the SSR marker AA239 (Figure 5B). Vig-Ps3.2, which was located 307.8-316.5 cM on the bottom of Chrom5/LGIII, was flanked by the SSR marker AA5 and the SNP marker PsCam036163_21311_1095; Vig-Ps3.1, which was located on the top segment (67.1–70.5 cM) of the same chromosome or linkage group, was flanked by the SNP marker PsCam013763_9362_423 and the SSR marker AD270 (Figure 5). The two stable QTL, Vig-Ps4.1 and Vig-Ps3.2, had a positive additive effect, indicating that the alleles for vigor originated from “00-2067.” In contrast, Vig-Ps3.1 has a negative additive effect, indicating that the alleles originated from “Reward” (Table 3).

In the case of plant height, the most stable QTL, Hgt-Ps3.1, was detected in three of the four experiments (GH1, GH2, and GH4; R² = 9.94–36.35%). This QTL was located on the bottom segment of Chrom5/LGIII (Figure 6) and was flanked by the SNP marker PsCam020937_11699_2576 and the SSR marker AA5 (Figure 6A). The second most stable QTL, Hgt-Ps7.2, was detected across two (GH2 and GH4) of four greenhouse experiments (R² = 7.04–20.04%). These QTL were located 142.3–168.0 cM on Chrom7/LGVII and were flanked by the SNP markers PsCam002756_2184_427 and PsCam045262_28962_162 (Table 3 and Figure 6B). Hgt-Ps7.1, which was flanked by the SNP markers PsCam035831_20992_561 and PsCam021891_12310_347 (81.2–115.3 cM) (Figure 6B), was detected in only one environment (GH1) on the same chromosome (R² = 13.63–14.06). The additive effect was negative for Hgt-Ps3.1, but positive for Hgt-Ps7.1 and Hgt-Ps7.2 (Table 3). This suggested that the QTL for height on Chrom5/LGIII was derived from “Reward”, while the QTL on Chrom7/LGVII originated from “00-2067.”

Two hundred eight putative digenic epistatic pairs were identified using all variables for disease severity, vigor, and plant height. These comprised 65 (12-24) for disease severity, 57 (10–21) for vigor, and 86 (15–28) for plant height. The 208 putative digenic interactions consisted of one major epistatic effect (PVE ≥ 15%), 13 moderate epistatic effects (7.5% ≤ PVE ≤ 15%), and 194 minor epistatic effects (PVE ≤ 7.5%). BLUPs for disease severity, vigor, and plant height detected 20, 18, and 21 putative digenic interactions, respectively. Within the 59 digenic interactions among BLUPs of all traits, epistatic analysis identified one major QTL pair, three moderate QTL pairs, and 55 minor QTL pairs (Table 4). In contrast, LSM of the pooled data detected 23 digenic interactions for disease severity, 14 for vigor, and 19 for plant height. The total 56 pairs included seven moderate epistatic-effect QTLs and 49 minor-effect QTLs.

Twenty-five digenic epistatic interactions with major and moderate effects were identified by 33 flanking markers, of which 10 epistatic-effect QTL with 14 flanking markers were linked to three additive-effect QTL (Fg-Ps3.1, Fg-Ps3.2, and Vig-Ps3.1). The remaining 15 epistatic QTL were not related to any of the additive-effect QTL (Table 4). Eight of the 10 epistatic-effect QTL were linked to Fg-Ps3.2, including the most significant QTL pairs, E.Hgt-Ps1 (R² = 31.2%), followed by E.Hgt-Ps7 (R² = 19.1%) and E.Hgt-Ps4 (R² = 13.5%). The fourth was E.Hgt-Ps3 (R² = 13.5%), which was linked to Fg-Ps3.2 and Vig-Ps3.1. Only E.Fg-Ps7 and E.Vig-Ps7 were linked to Fg-Ps3.1, showing moderate epistatic effect (R² = 9.5% and R² = 12.6%, respectively).

The QTL associated with partial resistance to F. graminearum on Chrom5/LGIII and Chrom4/LGVI flanked four and 74 candidate genes, respectively (Supplementary Table 3). Fifteen of the 74 genes were related to plant defense mechanisms. These included UDP formation and transportation, the integral component of membrane proteins, histone-lysine N-methyltransferase, phospholipid transport, actin cytoskeleton, calcium-ion binding, methyltransferase, UBQ-conjugating enzyme/RWD, AP-4 adaptor complex, oxidoreductase, acyl group transferases, hydrolases, G protein-coupled receptors, and protein involved in phosphorylation and proteolysis. Some of the genes were involved in pathways related to plant defense mechanisms. Psat4g125440 is involved in cellulose biosynthesis, while Psat4g111280, Psat4g110800, Psat4g108480, and Psat4g102720 are involved in protein ubiquitination.