159 wheat varieties derived from ICARDA were obtained and designated as ICARDA001-ICARDA159 ( Supplementary Table 1 ). CYR32, CYR33, and CYR34 were provided by the wheat stripe rust breeding laboratory of Northwest A&F University.

To assess disease resistance in wheat seedling stage, 10 seeds of a wheat variety were sown in a container and grown under suitable conditions. Once the plants reached the two-leaf stage, they were uniformly sprayed with a 1:100 ratio of Tween to water. Next, each physiological race of stripe rust was mixed with talc in a 1:20 ratio and shaken onto the test varieties. After the wheat leaves were completely diseased, the infection type was rated according to Line and Qayoum’s 9-level reaction pattern.

The stripe rust resistance of adult wheat was evaluated between 2022 and 2023 in experimental fields located in Xining and Guide of Qinghai Province. In 2022, trials in Xining and Guide were designated E1 and E2, respectively, and were inoculated with a mixture of CYR32 and CYR33. In 2023, trials in Xining and Guide were designated as E3 and E4, respectively, and were inoculated with a mixture of CYR32 and CYR34. Wheat lines were scored by recording the number of rust lesions on each leaf during the height of infection.

The genome of 159 wheat varieties was analyzed using a 55K SNP chip developed by the Affymetrix Axiom platform, provided by ZhongYuJin Marker (Beijing) Biotechnology Co., Ltd. Affymetrix Axiom Analysis Suite software (Thermo Fisher) was employed for preliminary screening and genotyping. A total of 53,063 loci were obtained from the raw data, following criteria of DQC (Data Quality Control) of ≥0.82 and CR (Call Rate) of ≥0.95 (https://datadryad.org/stash/share/aphTKrewmqNQnQgdj7vgUlw7I8jPMCT0GYiCyDNneSw). VCFtools software (Danecek et al., 2011) was used for site filtering with a missing rate of >20% and an MAF (Minor Allele Frequency) <0.05, resulting in 24,151 high-quality SNP sites. The marker density map for the obtained high-quality SNP loci was plotted using the CMplot package in the R suite.

The genetic diversity was calculated with PowerMarker V3.25 software (Liu and Muse, 2005) using the obtained 24,151 high-quality SNP sites. These calculations adhered to the PIC (Polymorphic Information Content) parameter, represented by PIC=1-Σ(P_ij)² (P_ij represents the frequency of the j^th allele at the i locus). Plink 1.9 software (Purcell et al., 2007) was used to perform LD filtering on SNP sites, resulting in 2,821 markers. Admixture software (Alexander et al., 2009) was used to analyze the population structure. The range of K values for the subgroups was preset to be 2-15, and the average value was calculated based on six repetitions. The average Cross-Validation (CV) value was calculated for each K value, and the K value associated with the lowest CV value was selected as the optimal subgroup number. Finally, a subgroup map was drawn using R software.

A GWAS was conducted on the 159 wheat varieties using 24,151 high-quality SNPs, phenotypic stripe rust resistance data collected at both the seedling and adult plant stages, a kinship matrix (K-matrix) generated by Tassel 5.0 software (Bradbury et al., 2007) and a Q-matrix created by Admixture.

During disease resistance screening of wheat seedlings, 21 (13.2%) varieties showed immunity to CYR32, while 125 (78.6%) varieties showed high resistance to CYR32, the remaining 13 (8.2%) lines showed susceptibility to rust stripe race CYR32, including IR002, IR017, IR037, IR053, IR055, IR061, IR073, R077, IR078, IR125, IR128, IR135, and IR149. For CYR33, 18 (11.3%) varieties showed immunity, 128 (80.5%) wheat materials showed high resistance, 1 (0.6%, IR133) varieties showed moderate resistance, and 12 (7.6%) varieties showed susceptibility (except for IR135 among the 13 materials mentioned above). For CYR34, 11 (6.9%) materials showed immunity, 68 (42.8%) materials showed high resistance, 1 (0.6%, IR107) material showed moderate resistance, and 79 (49.7%, including the 13 materials aforementioned) materials showed susceptibility, and most of them are highly susceptible to diseases ( Supplementary Table 2 ; Figure 1 ). With the pervalent race of CYR34, most wheat varieties lost resistance at the seedling stage.

For the stripe rust evaluation of the adult plant stage in Xining, under the E1 environment, 6 materials (3.8%, IR017, IR055, IR125, IR128, IR135, and IR149) showed susceptibility, 111 materials (69.8%) showed high resistance, and 42 materials (26.4%) showed moderate resistance; in the E3 environment, 74 (46.5%) materials showed susceptibility, of which 68 showed resistance in 2022, 72 (45.3%) showed high resistance, and 13 (8.2%) showed moderate resistance. Compared with the E1and E3 environment, there were 50 materials (31.4%) with enhanced disease resistance in the E3 environment, 12 materials (7.5%) with unchanged resistance, and 23 materials (14.5%) with reduced resistance ( Supplementary Table 3 ).

For the identification results of the adult plant stage in Guide, under E2 environment, 6 (3.8%, the same as E1 environment) materials showed susceptibility, 113 (71.1%) materials showed high resistance, and 40 (25.2%) materials showed moderate resistance; in the E4 environment, 74 (45.3%, similar to the E3 environment) materials showed susceptibility, of which 68 materials showed resistance in 2022, 66 (41.5%) materials showed high resistance, and 19 (11.9%) materials showed moderate resistance. Compared with E2 and E4 environment, there were 49 materials (30.8%) with enhanced resistance in the E4 environment, 12 materials (7.5%) with unchanged resistance, and 24 materials (15.1%) with reduced resistance ( Supplementary Table 3 ).

The SNP density map illustrated a non-uniform distribution of 24,151 high-quality SNP loci across three genomes and 21 chromosomes, there was a portion of fragments without SNP markers, and the density was almost entirely concentrated at 5-15 loci/Mb. Among them, the SNP density on chromosome 3D was the highest at 1.7424/Mb, followed by 4D chromosome at 1.4263/Mb, and chromosome 6A has the lowest density at 0.3982/Mb; the number of SNPs on chromosome 4B was the highest, with 1675, followed by chromosome 5B with 1652, and the smallest was chromosome 4D with only 357. There were 9,338 (38.7%), 9,865 (40.8%), and 4,948 (20.5%) polymorphisms identified in genomes A, B, and D respectively. The overall polymorphism ratio was 45.5% (2,4151/53,063), with the highest occurrence in genome B, followed by A and then D. Among them, chromosome 2A possessed the highest PIC value of 0.3526, followed by chromosome 4A with a PIC value of 0.3467. The lowest PIC value was chromosome 3B with 0.2433; the PIC value of the genome A was the highest at 0.3264, followed by the B at 0.2999, and D chromosome at 0.2937. The average PIC value of all chromosomes is 0.3067 ( Table 1 ; Figure 2 ).

The analysis of genetic diversity (GD) revealed an average GD of 0.3744 among the 159 wheat lines, with variations ranging from 0.284 3to 0.4387. Genome A demonstrated the highest GD, followed by B and D. Among all chromosomes, the 2A chromosome had the highest GD value of 0.4387, followed by the 4A chromosome with a GD value of 0.4385, and the 3B chromosome had the lowest value of only 0.2843.

The population structure analysis of the 159 wheat materials revealed that at the minimum CV value, K=11, suggesting that the materials could be divided into 11 subgroups. The Q-matrix generated when K=11 was employed as the covariate in the subsequent association study. The subgroups were comprised of the following: subgroup I with 7 (4.4%) varieties; subgroup II with 21 (13.2%) varieties; subgroup III with 16 (10.1%) varieties; subgroup IV with 11 (6.9%) varieties; subgroup V with 12 (7.6%) varieties; subgroup VI with 18 (11.3%) varieties; subgroup VII with 15 (9.4%) varieties; subgroup VIII with 22 (13.8%) varieties, subgroup IX with 4 (2.5%) varieties; subgroup X with 10 (6.3%) varieties; and subgroup XI with 23 (14.5%) varieties ( Figure 3 ).

The association study of ICARDA wheat varieties using 24,151 high-quality SNP loci and phenotypic data identified 593 loci associated with stripe rust resistance, contributing rates ranging from 11.1% to 18.7%. During the seedling stage, 71 loci were found to be related to stripe rust race CYR32. These loci were distributed across chromosomes 1A, 1D, 2A, 2B, 2D, 3A, 3B, 4A, 5A, 5B, 6B, 6D, 7B, and 7D, with contribution rates ranging from 12.7% to17.2%. With the exception of AX-109949596, AX-108783340, and AX-111778082, these loci were also involved in resistance against CRY33 with a contribution rate of 12.6%-17.2%, suggesting that they were multi-effect loci. An additional 518 loci were related to CYR33 and were distributed across all chromosomes with contribution rates of 12.6%-18.7%. In the association study of adult plants, we identified one stable locus, AX-109318462, on chromosome 2B with contribution rates of 14.4%-17.0% in E1 and E2 ( Figures 4 – 6 ; Supplementary Table 4 ). An additional three loci on chromosomes 2B, 4A, and 7B were detected in E2, with contribution rates ranging from 11.1% to 16.9%. No significant loci were detected in E3 and E4. According to the results of the QQ plot, there were loci that were significantly associated with the stripe rust CYR32 and CYR33, E1 and E2 environments in all loci. Although there was a deviation between the observed P value and the expected P value in the QQ plot, which may be due to errors in statistics during the seedling and adult stage phenotype periods by manpower, it can have a certain impact on the results, but overall it can correspond to the results in the Manhattan plot.

In the screenings of 159 wheat varieties at the seedling stage and during a two-year, two-environment adult field trial, we identified 146 (91.8%) lines that were resistant to both CYR32 and CYR33, while only 80 (50.3%) were resistant to CYR34. These results are consistent with previous studies and suggest that most of the introduced wheat varieties had good resistance to CYR32 and CYR33, indicating that these varieties contained genes, gene combinations and even new genes that could effectively resist CYR32 and CYR33, which could effectively prevent the epidemic of these two epidemic races in the future, and the current materials did not have enough resistance to the new toxic race CYR34 (Zhang et al., 2022). This may be attributed to the early application of CYR32 and CYR33 promoting resistance in wheat (Zeng et al., 2015). Yao conducted a preliminary study on the overwintering conditions of wheat stripe rust in the eastern wheat area of Qinghai, and the results showed that the amount of bacteria before winter was the primary factor affecting the overwintering of wheat stripe rust (Yao et al., 2014). Therefore, the utilization of disease resistance during the whole growth period is of great significance for the prevention and control of stripe rust fungus in Qinghai wheat. Subsequently, these 159 materials can be used for molecular testing to obtain the disease resistance genes that may be contained in them, so as to provide a basis for the development of new varieties.

The PIC variation of the 159 materials ranged from 0.243 to 0.353, with an average of 0.307. This indicates that the SNP markers were moderately polymorphic, which is consistent with previous studies (Kumar et al., 2020; Liu YK. et al., 2020). Additionally, the genetic diversity ranged from 0.284 to 0.439, with an average of 0.374. This suggests a relatively concentrated genetic background between the breeds, aligning with previous research (Cao et al., 2015; Li et al., 2019). An analysis of the 55K SNP chip data from 159 ICARDA wheat varieties combined with the phenotypic stripe rust resistance data revealed that genome B contained the highest quantity of polymorphisms, while genome D exhibited the least. These findings reflect the works of Zhang (Zhang et al., 2016) and Tehseen (Tehseen et al., 2021). A total of 593 significant loci were detected, 68 of which were significant in conferring resistance against both CYR32 and CYR33. These loci are distributed across chromosomes 1A, 1D, 2A, 2B, 2D, 3A, 3B, 4A, 5A, 5B, 6B, 6D, 7B, and 7D. The analysis of stripe rust resistance at the adult-plant stage identified locus AX-109318462 on chromosome 2B as the sole locus associated with resistance in both E1 and E2. Subsequently, referring to the IWGSC_RefSeq_v1.0, the loci within the 5Mb interval on the physical map were considered as one significant locus. Using the QTL genetic map of wheat stripe rust resistance constructed by Cheng (Cheng et al., 2019) and the comparison diagram by Yao (2022), the 69 loci were compared and analyzed. It was found that 9 of these 69 loci may be newly discovered loci, namely AX-110974432 (3A), AX-110447030 (3A) AX-108736767 (3A), AX-89776892 (3A), AX-110653920 (5B), AX-109997800 (5B), AX-95252437 (6D), AX-86163952 (7B), AX-110962394 (7B) (the chromosome where the locus is located is indicated in parentheses); the remaining loci are the same or similar to the reported QTL or Yr gene positions. The two loci located on chromosome 5B partially coincided with Qyr.pd-5B.1 between the markers Xbarc275 and XIWA2095, and were very close to the reported QTLs related to stripe rust disease type and QTLs associated with infection type, respectively, so it was speculated that there may be genes related to stripe rust type or reactive type or both in the vicinity of these two loci, and two loci located on chromosome 7B were close to the Yr63 gene on this chromosome, Yr63 is located at a hot spot known for against pests and diseases in plants and animals, which was enriched with multi-nucleotide-binding and leucine rich repeat (NLR) and kinase domain encoding genes (Mackenzie et al., 2023), it is speculated that it is possible to discover new defense genes near this site.