The GWAS population is composed of 191 wheat genotypes including 89 spring wheat and 102 winter wheat genotypes that consisted of 111 cultivars and breeding lines from Kazakhstan, 17 cultivars from Russia, and 1 cultivar from Brazil, as well as 30 lines sourced from CIMMYT and CIMMYT-ICARDA-IWWIP. Most importantly, the cultivars in the panel are extensively used in breeding programs in Kazakhstan and Central Asian countries.

The GWAS panel was phenotyped under greenhouse conditions for Z. tritici (mix of five isolates, namely, 156-22, 154-22, 1-22, 170 6-22, and 3-22 screened in a single experiment) and P. nodorum (four independent isolates, namely, 149-22_ToxA, 150-22_Tox1, and 118-22_Tox3 from Russia and Sn2K from USA and a purified toxin SnToxA from USA screened in five different independent experiments) during 2023. The P. nodorum isolates 149-22_ToxA, 150-22_Tox1, and 118-22_Tox3, originating from the Tambov and Altay regions of Russia, were identified as producers of toxins ToxA, Tox1, and Tox3, respectively. This identification was confirmed by inoculation experiments with differential wheat genotypes (“Mironovskaya 808” as a susceptible check, “Don Mira” as a resistant check) infiltration assays, and PCR techniques using ToxA-, Tox1-, and Tox3-specific markers (Nuzhnaya et al., 2023; Kovalenko et al., 2023; Zeleneva et al., 2023). All isolates were received from the All-Russian Institute of Conservation, Russia, except isolate Sn2K and purified toxin SnToxA, which are from South Dakota State University (SDSU), USA. Hence, six independent experiments were conducted in greenhouse conditions. The experiments were conducted at the All-Russian Institute of Plant Protection (ARIPP) in St. Petersburg-Pushkin, Russia. However, the experiments using isolate Sn2K and purified toxin SnToxA were conducted at SDSU in Brookings, SD, USA. The wheat genotypes were grown in a completely randomized design with five replications. Ten seeds from each genotype were sown in 20-cm pots, with each pot serving as one replicate. Soil preparation and inoculation followed standard protocols, using a universal substrate (“Terra vita” produced by “Nord Pflp,” Russia). The data generated from greenhouse experiments are given in Supplementary Table S1 .

The P. nodorum isolates were preserved on V8-PDA agar at 21°C in a 12-h light and dark cycle for 2 weeks (Phan et al., 2016). Stock cultures of Z. tritici were grown on yeast sucrose agar (YSA; 10 g L⁻¹ yeast extract, 10 g L⁻¹ sucrose, and 1.2% agar) with kanamycin (50 µg/mL) supplement (Scala et al., 2020). Thirty-day-old cultures were stored in a refrigerator at +4°C temperature for inoculation prior to use (Scala et al., 2020). Inoculation with foliar pathogens involved spraying conidial suspensions (P. nodorum: 10⁶ spores/mL; Z. tritici: 10⁷ spores/mL) containing 0.1% Tween 20 surfactant, as described by Scala et al. (2020) and Fagundes et al. (2020). The inoculum was evenly sprayed on the plants, and the pots containing plants were kept in the climate chamber (Model MLR-352H-PE, “PHCbi”, Tokyo, Japan). A thoroughly cleaned spray gun was set at 2.0 bar pressure to spray the Z. tritici/P. nodorum inoculum on the selected marked leaf sections of each plant. Approximately 15 min were allowed to settle the inoculum on the leaf surface. The pots were then kept in big plastic bags containing approximately a liter of water. To create a congenial environment of high relative humidity (RH), the bags were tightly closed with tape or plastic clips. Furthermore, a greenhouse facility was used to incubate the plants containing bags at approximately 20°C during the day and ~12°C at night with a 12-h day/12-h night cycle. After a 48-h treatment cycle, the pots were taken out from the bags and kept in the trays, ensuring randomized placement. To maintain the RH of 70% to 90%, the plants were given water at frequent intervals. The same growth conditions including light and temperature were used for the wheat plants to grow for 21 days after inoculation (Kolomiets et al., 2017; Fagundes et al., 2020).

The disease reaction to P. nodorum isolates from Russia was assessed at 20–22 days post-inoculation through a lesion-based scale of 0–4, where 0–1 indicates resistance, 2 indicates moderate susceptibility, 3 indicates susceptibility, and 4 indicates high susceptibility (Tan et al., 2012). For the screening of wheat with the P. nodorum isolate Sn2K from the USA, inoculation was performed using conidia as described by Liu et al. (2004). Additionally, for phenotyping sensitivity to SnToxA, lines were screened for their reaction to the purified toxin Ptr ToxA, which is equivalent to SnToxA, at a concentration of 10 µg/mL. Four leaves from each genotype (with the second leaf fully expanded) were infiltrated with pure SnToxA culture filtrate as detailed in Faris et al. (1996). Post-infiltration, the plants were kept at 21°C during the day and 18°C at night, with a 16-h photoperiod in the growth chamber. Leaves were scored as insensitive (−) or sensitive (+) after 4 days of infiltration.

Now, on P. nodorum isolates, 149-22_ToxA, 150-22_Tox1, 118-22_Tox3, Sn2K isolate, and SnToxA will be quoted as Pn ToxA_Russia, Pn Tox1_Russia, Pn Tox3_Russia, Pn Sn2K_USA, and Pn ToxA_USA, respectively.

The phenotyping data of three cropping seasons under natural field conditions and two cropping seasons under artificial field inoculation conditions are given in Supplementary Table S2 . The GWAS population was evaluated at the Kazakh Research Institute of Agriculture and Plant Growing (KRIAPG), Almalybak (43°1300900 N, 76°3601700 E) in Southeast Kazakhstan, Almaty, during the 2019–2020, 2020–2021, and 2021–2022 cropping seasons with a plot size of 1 m². The experimental material was given a fertilizer dose of 60 kg/ha N and 30 kg/ha P₂O₅ and standard crop management practices were followed (Dospekhov, 1985). The planting material was sown during mid-September and was harvested in mid-August of the succeeding year during all three years of the testing period. The region receives 400 mm of rainfall annually; hence, only three irrigations were given during crop growth. The field phenotyping was done under natural field disease incidence conditions for three consecutive years during 2019–2020, 2020–2021, and 2021–2022 crop seasons, whereas phenotyping was done under artificial field inoculation conditions for two consecutive years i.e., 2020–2021 and 2021–2022. Field plots were inoculated with a mixed inoculum of Z. tritici, derived from 80 to 100 randomly selected infected leaf samples collected from major spring wheat-producing regions in southeastern Kazakhstan. The diseased straw and stubbles were added to the soil at the rate of 1 kg/m² before sowing.

The Zadoks scale was used to score the disease incidence in the field; the disease severity was scored on the first and flag leaves when all the lines were near or at Zadoks growth stage Z69 (complete flowering stage) and Z75 (medium milking stage) (Zadoks et al., 1974). The STB score was determined by calculating the percentage of infection on individual leaves and averaging the multiple scorings. A double-digit scale of 00–99, which was modified from Saari and Prescot (Saari and Prescott, 1975), was used to classify host reactions to STB. Based on the degree of infection, the genotypes were divided into the following categories: 0%–10% rated as highly resistant (HR: infection-free or some scattered lesions on the lower leaves); 11%–20% rated as resistant (R: low-intensity infection on first leaves and isolated lesions on the second set of leaves); 21%–40% rated as moderately susceptible (MS: lower leaves’ infection is mild to severe and isolated to the low infection spreading to the leaf below the mid portion of plant); 41%–70% rated as susceptible (S: high-intensity lesions on leaves present at the low and middle portion of the plant and mild to high infection of the upper third of plant; flag leaf infection is higher than the traces); 71%–100% rated as highly susceptible (HS: infections spread to spikes and very high infections on all the leaves). The phenotype ratings for STB resistance were calculated as an area under the disease progress curve (AUDPC).

The field (natural infection and artificial infection) resistance for STB was calculated as AUDPC scores. The AUDPC was estimated yearly by cumulating the progress of diseases severity. AUDPC values from double-digit and AUDPC from flag leaf (F) and penultimate leaf (F−1) were separately estimated using the formula defined by Wilcoxson et al. (1974). The AUDPC was calculated based on three STB severity scores taken at 7-day intervals during plant growth. The formula used to calculate the AUDPC was as follows:

y_i is an evaluation of disease at the ith observation;

t_i is time (in days) at the ith observation;

n is the total number of observations.

All phenotypic analysis was done in multi-environment trial analysis in R (META-R) version 6.0 (Alvarado et al., 2015). In brief, the single year Best Linear Unbiased Estimators (BLUEs) were estimated for the greenhouse evaluations for isolates from Z. tritici and P. nodorum, namely, Zt_Russia (mixed isolates), Pn ToxA_Russia, Pn Tox1_Russia, Pn Tox3_Russia, Pn Sn2K_USA, and Pn ToxA_USA. While calculating BLUEs, the genotypes are considered as fixed effects. Furthermore, genetic and residual variances, broad sense heritability (H2), coefficient of variance (CV), mean, and correlation coefficients were estimated in META-R. These generated AUDPC scores from field evaluations and BLUEs from greenhouse evaluations were used for GWAS analysis.

The details of DNA extraction and genotyping is provided in Kokhmetova et al. (2021). Briefly, DNA was extracted from fresh young leaves following the modified CTAB method as described in Dreisigacker et al. (2016). The genotyping was done with DArTseqTM technology by the Genetic Analysis and Service for Agriculture (SAGA) laboratory in Mexico. Furthermore, the DNA libraries were sequenced with 192-plexing on Illumina HiSeq2500 with 1 × 77-bp reads. The allele calls were generated by a proprietary analytical pipeline developed by DArT P/L (Sansaloni et al., 2011). The monomorphic markers, markers with minor allele frequency less than 5%, markers with > 20% missing allele calls, and markers with > 25% heterozygote frequency were removed to get the high-quality informative markers. The final filtered set of 8,154 markers were further used in GWAS analysis for the marker–trait association (MTA) identification. The genotyping data for 191 wheat entries are given in Supplementary Table S3 .

The estimation of pairwise linkage disequilibrium (LD) values (r ²) between the SNPs, construction of LD decay plots, Principal component analysis (PCA), and construction of Neighbor – Joining (N-J) tree for understanding population structure is done as described in Rathan et al., 2022. Briefly, the LD values (r ²) were generated using TASSEL version 5.2.94 and the LD decay was visualized in R Studio by following the method given by Remington et al. (2001). The point where the LD values drop to half of their maximum value is used to define the extent of LD decay at the genome and subgenome level. The distance matrix was generated in TASSEL version 5.2.94 and exported in Newick format, and the same matrix was used to generate N-J tree in iTOL version 7 tool (iTOL: Interactive Tree Of Life). The PCA was done using Genome Association and Prediction Integrated Tool (GAPIT) version 3.4.

The information about the procedures followed to perform GWAS analysis, generate QQ plots, and fix the Bonferroni correction factor and R ² values is provided in Khan et al. (2022). Briefly, AUDPC scores from field evaluations and BLUEs from greenhouse evaluations were used for GWAS analysis. The BLINK (Bayesian-information and Linkage-disequilibrium Iteratively Nested Keyway) model (Huang et al., 2019) from GAPIT version 3.0 (Wang and Zhang, 2021) was employed to identify MTAs. The Bonferroni correction has been employed to adjust the threshold for statistical significance (α) at 0.05 and subsequently dividing this value by the total number of markers under consideration. The R ² was used to describe the percentage variation explained (PVE) by significant MTAs. The allelic difference of significant MTA was estimated as the difference between the mean value of genotypes with and without favorable alleles for disease scores and was presented in box plots.

The in silico analysis was done as described in our previous study (Kokhmetova et al., 2023). In brief, the putative candidate genes were identified in RefSeq v2.1 assembly from the International Wheat Genome Sequencing Consortium (IWGSC) integrated in the Ensembl Plant database (https://plants.ensembl.org/index.html) using the basic local alignment search tool (BLAST). The 100-kb region overlapping and flanking the associated SNP was mined to identify the putative candidate genes.

The BLUEs generated for different Septoria isolates tested under greenhouse conditions during 2023 and the summary statistics from the study are presented in Table 1 . Heritability was high for all the studied isolates, ranging from 0.90 to 0.98. The genetic variance was highly significant for all phenotypes ( Table 1 ). The coefficient of variation (CV) ranged from 12.89 (Pn ToxA_Russia) to 28.5 (Pn ToxA_USA). Genotypic and phenotypic correlation coefficients for various Septoria reactions tested under greenhouse conditions are presented in Table 2 . The reaction of Zt_Russia mixed isolates had a significant and positive genetic correlation with isolates Pn ToxA_Russia (0.16*), Pn Tox1_Russia (0.3**), and Pn Sn2K_USA (0.28**). Pn ToxA_Russia isolate reaction had a significant and positive genetic correlation with Pn Tox1_Russia (0.35**), Pn Tox3_Russia (0.15*), and toxin Pn ToxA_USA (0.15*). The Pn Tox3_Russia isolate had a significant and positive genetic correlation with the Pn Sn2K_USA (0.15*) isolate. The Pn Sn2K_USA isolate had a significant and positive genetic correlation with the reaction to the toxin Pn ToxA_USA (0.69**). Similarly, Zt_Russia mixed isolates had a significant and positive phenotypic correlation with isolates Pn ToxA_Russia (0.14*) and Pn Tox1_Russia (0.29**). The Pn ToxA_Russia isolate has a significant and positive phenotypic correlation with Pn Tox1_Russia (0.34**), Pn Tox3_Russia 0.15*), and Pn ToxA_USA (0.16*). Since trait heritability is high, the phenotypic and genotypic correlations were very close to each other.

The GWAS analysis was conducted with 8,154 high-quality SNP markers. The subgenome and chromosome level distribution is provided in Table 3 . A, B, and D subgenomes had 3,298, 3,941, and 915 markers, respectively. At the chromosome level, the 4D chromosome harbored only 28 markers, whereas chromosome 2B harbored the maximum number of 870 markers.

The PCA plot-based population structure is presented in Figure 1A . The heat map of the pairwise kinship matrix is presented in Figure 1B . Through Neighbor-Joining (NJ) analysis, the population was divided into three subgroups ( Figure 1C ). The r ² values for all the SNPs were estimated and plotted against the genetic distance (cM) to calculate the LD values ( Figure 2 ). The LD decay was rapid in the B subgenome (0.40 cM) followed by the A subgenome (0.62 cM) and the whole genome (0.66 cM). However, LD decay was much slower in the D subgenome (4.28 cM) as compared to the A and B subgenomes.

The SNPs linked to STB and SNB resistance were further used to identify the putative genes using the annotated wheat reference sequence (IWGSC RefSeq v2.1) and are given in Table 5 . The genes falling in the 100-kb region flanking on either side of the marker were used to identify putative candidate genes. The region of MTA 100021621 associated with the Pn ToxA_Russia isolate possibly encodes a winged helix DNA-binding domain superfamily, an F-box-like domain superfamily, and an LRR domain superfamily. Similarly, an MTA, 2275733, associated with the Pn ToxA_Russia isolate encodes a protein response to low-sulfur, glycine–arginine–phenylalanine (GRF)-type zinc fingers (GRF-ZFs). An SNP 1234457 linked with Pn ToxA_Russia encodes Cytochrome P450. Another MTA, 5971516, for the Pn Tox3_Russia isolate encodes an RNA-binding S4 domain superfamily. Similarly, 1070935 associated with the Pn Tox3_Russia isolate encodes the haem peroxidase superfamily, peroxidases haem-ligand binding site pleiotropic MTA 100023665 associated with Pn Sn2K_USA and Pn ToxA_USA isolates, and encodes potassium transporter, palmitoyltransferase, nucleotide-binding alpha-beta plait domain superfamily, and RNA-binding domain superfamily. SNP 1230893 for AUDPC 2020_Natural encodes the DNA-binding domain superfamily, disease resistance protein, NB-ARC, LRR domain superfamily, P-loop containing nucleoside triphosphate hydrolase, virus X resistance protein, O-methyltransferase domain, and S-adenosyl-L-methionine-dependent methyltransferase superfamily. Similarly, 1202459 encodes for the LRR domain superfamily, NB-ARC, and virus X resistance protein. For AUDPC 2021_Artificial, SNP 1212480 on chromosome 3A encodes protein of unknown function DUF247.