QTL mapping for seedling and adult plant resistance to stripe and leaf rust in two winter wheat populations

The two recombinant inbred line (RIL) populations developed by crossing Almaly × Avocet S (206 RILs) and Almaly × Anza (162 RILs) were used to detect the novel genomic regions associated with adult plant resistance (APR) and seedling or all-stage resistance (ASR) to yellow rust (YR) and leaf rust (LR). The quantitative trait loci (QTLs) were detected through multi-year phenotypic evaluations (2018–2020) and using high-throughput DArTseq genotyping technology. RILs exhibited significant genetic variation with p < 0.001, and the coefficient of variation ranged from 9.79% to 47.99% for both LR and YR in all Environments and stages of evaluations. The heritability is quite high and ranged between 0.47 and 0.98. We identified nine stable QTLs for YR APR on chromosomes 1B, 2A, 2B, 3D, and 4D and four stable QTLs for LR APR on chromosomes 2B, 3B, 4A, and 5A. Furthermore, in silico analysis revealed that the key putative candidate genes such as cytochrome P450, protein kinase-like domain superfamily, zinc-binding ribosomal protein, SANT/Myb domain, WRKY transcription factor, nucleotide sugar transporter, and NAC domain superfamily were in the QTL regions and probably involved in the regulation of host response toward pathogen infection. The stable QTLs identified in this study are useful for developing rust-resistant varieties through marker-assisted selection (MAS).


Introduction
Globally, stripe rust or yellow rust (YR) and brown rust or leaf rust (LR) are two important biotic stresses of wheat (Triticum aestivum L.).The YR caused by Puccinia striiformis (Pst) generally causes crop damage in the range of 0.1%-5.0%;however, crop losses can increase to 5%-25% (Wellings, 2011) based on varietal reaction and prevailing environmental conditions, and in severe conditions, crop damage can reach up to 100% (Ali et al., 2014).Most wheat-cultivating areas covering the United States, Eastern and Southern Asia, East Africa, Oceania, the Arabian Peninsula, and Western Europe are vulnerable to Pst incidence.The monoculture of single or closely related cultivars coupled with favorable environmental conditions is ideal for pathogen evolution.Several incidences of YR epidemics have occurred in different parts of Central Asia and Kazakhstan (Yessenbekova et al., 2016;Kokhmetova et al., 2018;Kokhmetova et al., 2021a).The YR incidence in Central and Western Asia has substantially increased between 2001 and 2010 (Morgounov et al., 2013).Recently, Central Asia recorded four YR epidemics between the years 2009and 2014(Ziyaev et al., 2011;;Sharma et al., 2014).Historically, YR used to be confined to cool weather conditions; however, it has slowly moved to non-conventional regions due to race evolution (Muleta et al., 2017;Godoy et al., 2018).The LR caused by Puccinia triticina (Pt) is comparatively less devastating than the other two wheat rusts; however, it causes more crop damage as the frequency of its occurrence is very high and has wide global distribution (Huerta-Espino et al., 2011).The wide adaptability of this rust makes it spread to temperate areas, resulting in approximately 70% of yield losses (Herrera-Foessel et al., 2006;Aktar-Uz-Zaman et al., 2017).North Kazakhstan has reported five leaf rust epidemics during 2001-2009, causing yield damage in the range of 10%-50%, particularly in the susceptible varieties (Kokhmetova et al., 2016;Kokhmetova et al., 2021b).
Rust resistance breeding provides a sustainable solution to protect the wheat from loss of yield and grain quality.Genetically, there are two kinds of rust resistance; one is racespecific seedling resistance or all-stage resistance (ASR) and the other is race-non-specific adult plant resistance (APR) or partial resistance (Chen, 2013).Race-specific resistance is qualitative in nature, is governed by a single gene or oligogenes and only effective against a single or few races, and follows the gene-for-gene hypothesis (Flor, 1971).They express from the seedling to adult plant stage and confer vertical resistance.Generally, race-specific seedling resistance is less durable, which can easily be overcome by race evolution (Jones and Dangl, 2006).In contrast, race-nonspecific resistance genes are more durable, but when used alone, they are unable to provide high levels of resistance; however, when used in combination with other race-specific or race-non-specific genes, they provide adequate resistance (Singh et al., 2000).
At present, there are 86 YR genes that were cataloged (McIntosh et al., 2020;Zhu et al., 2023); however, only a few genes like Yr5 and Yr15 are effective to most of the prevailing Pst races across the globe (Sharma-Poudyal et al., 2013).The Yr gene diversity in commercial cultivars is very important in managing stripe rust epidemics.Additionally, non-race-specific resistance driven by a few Yr genes such as Yr18, which express at the adult plant stage, confers field resistance against the three wheat rusts, which have been widely used for several decades (Randhawa et al., 2012;Krattinger et al., 2016).However, single-gene-based resistance in varieties is not enough to protect the cultivars, particularly under high disease pressure conditions (Zhang et al., 2019).Thus, the combined use of APR genes along with one or a few ASR genes may be ideal to protect the cultivars with durable resistance (Ellis et al., 2014;Liu et al., 2018).Similarly, 83 Lr genes have been identified (McIntosh et al., 2020;Kolmer et al., 2023), and 15 Lr genes exhibited APR response, including Lr34, Lr46, Lr67, Lr68, Lr74, Lr75, Lr77, and Lr78.Among them, seven are race-specific and eight are race-non-specific (McIntosh et al., 2016).Among race-specific APR genes, Lr12, Lr13, Lr22b, Lr35, and Lr37 are qualitative in nature and provide hypersensitive reactions only at the adult plant stage (McIntosh et al., 1995;Singh and Bowden, 2011).Previous reports in Kazakhstan revealed that several Lr genes became ineffective due to pathogen evolution, resulting in new virulent races.Several Lr genes including Lr9, Lr10, Lr19, Lr34, Lr37, and Lr68 are still providing resistance to several races, whereas Lr1 has lost its effectiveness (Koishybaev et al., 2010).Some of the APR genes like Lr1, Lr10, Lr21, Lr22a, Lr34, and Lr67 have been cloned; a few cloned genes like Lr34 and Lr67 were found to be associated with complex loci conferring resistance to multiple biotic stresses.Few pleiotropic gene complexes like Lr19/Sr25, Lr26/Yr9/Sr31/Pm8, Lr37/Yr17/Sr38, Lr67/Sr55/Yr46/Pm46, and Lr34/Yr18/Pm38/Sr57 are widely used in the breeding programs across the globe, including Kazakhstan, that are still providing sufficient resistance (Kokhmetova et al., 2016;Kokhmetova et al., 2021b).
Although several race-specific seedling genes were identified for YR and LR, the genetic dissection of rust resistance in wheat through QTL mapping is equally important in the management of wheat rust as the durability of most of the race-specific seedling genes is very less, particularly under high disease pressure conditions in regions with a wide distribution of single or similar varieties.The evolution of novel races and the breakdown of race-specific genes led wheat breeders toward identifying and utilizing the durable race-nonspecific APR genes and QTLs.The recent advancements in nextgeneration sequencing (NGS) technologies, the development of the wheat reference genome (IWGSC, 2018), and the cost reduction of genotyping made the genetic dissection of QTL regions and candidate genes more precise and effective.Previously, various mapping populations and marker systems were used to locate QTLs for LR resistance (Kolmer, 2015;Li et al., 2016;Kthiri et al., 2019;Zhang et al., 2019;Bokore et al., 2020;Ciechanowskaa et al., 2022;Delfan et al., 2023).Similarly, several QTLs were identified for SR resistance in different genetic backgrounds (Wang et al., 2015;Zhang et al., 2019;Farzand et al., 2021;Rollar et al., 2021;Yuan et al., 2020;Cheng et al., 2022;Rauf et al., 2022;Tehseen et al., 2022).However, very few are effective in providing resistance, and many among them provide race-specific resistance and hence have limited applicability to wide area deployment.
Therefore, we designed our study to identify the genomic regions that confer ASR and APR resistance to leaf and stripe rust resistance to the races prevalent across Central Asia, particularly in Kazakhstan, using two RIL mapping populations derived from Almaly × Anza and Almaly × Avocet S with multienvironment evaluations.We also attempted to provide the putative candidate genes for the identified stable QTLs to assist further validation and gene cloning experiments.

Plant material and field experiments
The parental genotypes used to develop RILs are contrasting for both YR and LR; Almaly was the resistant parent, whereas Anza and Avocet were the susceptible parents.The RIL populations were developed by crossing Almaly × Anza (160 RILs) and Almaly × Avocet S (206 RILs) through the single-seed descent method in southeastern Kazakhstan (Supplementary Table S1).Since Almaly is the common parent in both the crosses, hereafter Almaly × Anza will be referred to as the Anza population, whereas Almaly × Avocet S will be referred to as the Avocet population.The RILs were evaluated at the seedling and adult plant-growth stages for LR and YR pathogens.The mapping populations of both the crosses along with parental genotypes were evaluated at the Kazakh Research Institute of Agriculture and Crop Production (KazNIIZiR), Almalybak (43 °13′09″N and 76 °36′17″E) for 2 consecutive years during 2018-19 and 2019-20 for YR and LR APR, respectively.Additionally, the RIL population of Almaly × Anza was evaluated during 2020-21 for LR APR in a randomized complete block design (RCBD) following the two replications.Each RIL was sown in a tworow plot of 1.5 m length and row-to-row spacing of 25 cm.The susceptible check variety, Morocco, was planted at an interval of every 20 plots.The RILs were also evaluated for YR and LR ASR in a greenhouse facility at the All-Russian Institute of Plant Protection (ARIPP), St Petersburg, Russia (59 °73′73″N, 30 °42′47″E) during 2020.Three to five seeds of each genotype were planted in 10-cmdiameter plastic pots in a disease-free area.The RILs were inoculated after 7-10 days under greenhouse conditions with three races of P. striiformis and six races of P. recondita with different levels of virulence to Lr and Yr genes (Supplementary Table S3).All entries were arranged in an RCBD design with three replications.The complete phenotypic data file of two biparental populations is provided in Supplementary Table S1.

Phenotyping Seedling resistance in greenhouse
The P. striiformis races were differentiated in 2020 using a set of 12 wheat lines developed in the Avocet wheat background and on nine supplemental wheat differential lines using a method developed by Johnson et al. (1972).The determination of the type of plant reaction was carried out twice within 14-20 days after infection according to the Gassner and Straib accounting scale (Gassner and Straib, 1932).At the same time, the reactions of 0, 1, and 2 points were assigned to the resistant type R (Resistant), and those of 3 and 4 points were assigned to the susceptible type S (Susceptible).The P. triticina races were also differentiated during 2020 using 20 nearisogenic lines (NILs) developed in the Thatcher background, each carrying one of the LR resistant genes (Kolmer and Ordonez, 2007;Schachtel et al., 2012;Kolmer et al., 2014).The virulence of the phenotypes was determined on these 20 differential lines and encoded with 0 and 1 for avirulence and virulence, respectively (Long and Kolmer, 1989;Kolmer and Ordonez, 2007).The virulence analysis tools (Schachtel et al., 2012) was used for the nomenclature of P. triticina races.The type of response to leaf rust was determined twice within 14-20 days after infection, according to the scale of Mains and Jackson (1926).The reactions of 0, 1, and 2 points were assigned to the resistant type R (Resistant), and those of 3 and 4 points were assigned to the susceptible type S (Susceptible).
The seedlings of the RIL population from Almaly × Avocet S cross along with the parents were inoculated with two races of P. striiformis, i.e., 108E187 (Pst_1) and 110E191 (Pst_2), and two races of P. triticina, i.e., MLTTH and TLTTR, to determine the racespecific resistance.Similarly, the seedlings of the RIL population from Almaly × Anza cross along with parents were inoculated with two races of P. striiformis, i.e., 108E187 (Pst_1) and 101E191 (Pst_3), and four races of P. triticina, i.e., THTTQ, TCTTR, TCPTQ, and THTTR.The plants were infected with spores at a three-leaf stage, and plants were placed in a humid chamber for 24 h.The seedling infection type of the RIL was scored using the same approach as that for races differentiation.The pathotypes used in this study and their virulence reaction to rust genes are provided in Supplementary Table S3.

Phenotyping for adult plant resistance in the field
The field phenotyping for YR and LR APR was carried out during 2018-2019 for both the populations and also during 2020 for LR APR for the Anza population at Kazakh Research Institute of Agriculture and Crop Production (KazNIIZiR), Almalybak.Pathogen racial mixtures from the local population were used to inoculate the mapping populations.The method proposed by Roelfs et al. (1992) was followed for spore sampling, storage, and propagation.The pathogen was propagated in a greenhouse on the susceptible wheat variety, Morocco.The experimental wheat material was inoculated with a mixture of spores and talc in the ratio of 1:100 by spraying with an aqueous suspension of spores with 0.001% Tween-80 at the stem elongation stages (Z21-32).After infection, the plots were wrapped with a plastic cover for 16-18 h to create high humidity.After the manifestation of diseases on susceptible control varieties, an assessment (2-3 times) of rust resistance was carried out.Leaf and yellow rust resistance of wheat accessions was evaluated using the modified Cobb scale (Peterson et al., 1948;McIntosh et al., 1995).The scoring was based both on disease severity (proportion of the leaf area infected) and on the plant response to infection (reaction type).Plant responses were recorded as resistant (R), moderately resistant (MR), moderately susceptible (MS), and susceptible (S) reactions.

Phenotypic analysis
The phenotypic analysis was done in multi environment trial analysis with R (META-R) version 6.0 software.The best linear unbiased predictors (BLUPs) for each year and across year were used for QTL analysis.Furthermore, genetic correlation among traits and between environments, heritability, and ANOVA was done using META-R.The details of the analysis are provided in Rathan et al. (2023).Past V 3.01 was used to generate frequency distribution graphs.

Genotyping
The genomic DNA was extracted from the parents, and each RIL was extracted from both the populations following the modified cetyltrimethylammonium bromide (CTAB) method (Dreisigacker et al., 2012).The DArTseq technology was used for genotyping of both the RILs in the Genetic Analysis and Service for Agriculture (SAGA) lab based in Mexico (Edet et al., 2018).Briefly, the sequencing of mapping populations was carried out at 192plexing on Illumina HiSeq2500 with 1 × 77-bp reads.Allele calls for SNPs were generated through the proprietary analytical pipeline developed by DArT P/L (Sansaloni et al., 2011).Furthermore, the genetic locations of the SNPs were identified by using a 100 K consensus map given by SAGA (Sansaloni et al. unpublished).The complete genotypic data for the two biparental populations are provided in Supplementary Table S2.

Linkage mapping and QTL detection
The linkage maps were constructed separately for Anza and Avocet RIL populations using DArTseq SNP markers.The procedure followed for linkage map construction and QTL detection is the same for both populations.The markers were filtered, and the monomorphic markers, markers with >30% missing data, high heterozygosity percentage (>30%) and low allele frequency (<5%) were removed.The BIN functionality in IciMapping 4.2 QTL software was used to remove redundant markers.A filtered set of 1,293 and 1,127 high-quality SNPs were finally used for QTL analysis in Anza and Avocet populations, respectively.The linkage map construction and QTL mapping was done in IciMapping 4.2 QTL software (Wang et al., 2012;Li et al., 2016).The Kosambi mapping function was used to construct linkage groups, using a threshold logarithm of odds (LOD) score of 3.0 (Kosambi, 1943).Within each linkage group, the marker order was carried out with the 2-opt algorithm, and rippling was carried out by maintaining a window size of 5 cM.QTL mapping was done using complex composite interval additive functionality mapping (ICIM-ADD) (Li et al., 2008).Additive QTLs were detected using a 1.0 cM incremental scan.The LOD log confidence for QTL mapping was chosen as 3.0.Then, the QTLs were localized on the respective chromosomes.One-LOD drop from the estimated QTL position was considered the confidence interval.

In silico analysis
Stable QTLs with high phenotypic variation were used for the identification of candidate genes.The genes were identified in the RefSeq v1.0 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 molecular functions of the probable candidate genes found in the overlapping regions and within the 0.1 Mb flanking regions were identified.The role of the genes in governing leaf and yellow rust resistance was validated by comparing with the published literatures.

Genetic parameters and trait associations
Genetic parameters of both the RIL populations derived from Anza and Avocet crosses are presented in Table 1.Wide variability exists for both YR and LR resistance in both the RIL populations for all the races, as evidenced by the presence of a highly significant genotypic variance (Table 1).The frequency distributions of YR and LR severity in the field for RILs from both populations and in the seedling stages exhibited continuous variation (Figure 1; Supplementary Figures S1, S2).The interaction between genotype and location was significant for the pooled mean of YR and LR APR for both the RIL populations.Both populations exhibited a high broad sense heritability (≥0.8) for all the traits, except the APR of yellow rust (0.71) pooled data in the Anza population and APR of leaf rust (0.47) and yellow rust (0.79) pooled data in Avocet RIL population.The CV ranged from 12.9% (YR_ASR_Pst1 in 2020) to 63.3% (LR_APR in 2020) for the Anza population.Similarly, the CV ranged from 9.79% (YR_ASR_Pst1 in 2020) to 43.49% (LR_ASR_ TLTTR in 2020) for the Avocet population.Broad sense heritability estimates (h2) for leaf and yellow rust across years and different infection backgrounds were high (0.82-0.98), indicating that rust resistance can be improved by breeding (Table 1).
The year-wise genetic correlations between YR APR and LR APR in Anza and Avocet populations are presented in Table 2.A significant correlation was found between YR and LR APR during 2019 (p < 0.05) and the overall mean (p < 0.01) in Anza population; however, no correlation was observed in 2018.Furthermore, in the Avocet population, no correlation was observed between the traits.

Marker statistics
Genotyping of both Anza and Avocet populations was carried out using next-generation sequencing technology DArTseq ™ (http:// www.diversityarrays.com/dart-application-dartseq).A filtered set of 1,293 and 1,127 high-quality SNPs were used for linkage map construction and QTL identification in Anza and Avocet populations, respectively (Table 3).In Anza population, 539 SNPs were mapped on A subgenome, 491 SNPs on B subgenome, and only 263 SNPs on D subgenome in Anza population, whereas in Avocet cross, 482 SNPs were mapped on the B subgenome, 423 SNPs on A subgenome, and 222 SNPs on D subgenome.

QTL analysis
The QTLs identified for APR and ASR for both the rusts are presented in Tables 4,5 and illustrated in Figures 2, 3.A set of 51 QTLs were identified, out of which 28 QTLs included LR APR (6 QTLs), YR APR (10), and six each for LR and YR ASR in the Anza population.The remaining 23 QTLs were identified in the Avocet population including LR APR (3 QTLs), YR APR (12 QTLs), LR ASR (5 QTLs), and YR ASR (3 QTLs).Subgenome-wise 17 QTLs were identified on each subgenome of A, B, and D considering both the populations.Furthermore, the information about the favorable alleles of consistent QTLs is provided in Supplementary Table S5.

FIGURE 2
Identified QTLs and their genetic position from the Almaly × Anza population.

Putative candidate genes
The putative candidate genes were identified for consistent QTLs with high PVE for leaf and yellow rust APR through in silico analysis and are presented in Table 6, and additional details are provided in Supplementary Table S4.The QTL, i.e., QLR-APR-4A, located at

Discussion
The present rate of genetic gain is approximately 0.8%-1.2%for the major food crops, including wheat, and in recent years this progress has plateaued.The current rate of annual progress is too short of the 2.4% required to feed approximately 9.5 billion people by 2050 (Krishnappa et al., 2021a;Ray et al., 2012;Ray et al., 2013).Although the genetic progress of crop plants is a continuous process, protection of crop yield from biotic and abiotic stresses is also very important to minimize the crop losses and to have sustainable crop production.Rust (yellow, leaf, and stem) diseases are very important biotic stresses in wheat, which cause substantial crop damage across the globe.Genetic dissection of complex traits through QTL mapping will be helpful in designing the appropriate breeding strategies through MAS (Krishnappa et al., 2021b;Khan et al., 2022).Many of the race-specific/seedling resistance genes identified for all three rusts are from wild relatives, and their direct utilization in breeding programs is hindered due to an undesirable linkage drag associated with resistance locus.Furthermore, the durability of the seedling resistance genes is less compared to APR genes.Hence, to avoid linkage drag and resistance breakdown, plant breeders showed much interest in molecular studies in elite genetic backgrounds (Tehseen et al., 2022).
Phenotyping of 206 RILs from the Avocet and 160 RILs from Anza population suggests the presence of a wide variability of resistance to both the rusts, APR and ASR.Previously, a similar kind of broad variability was observed for wheat leaf rust (Rollar et al., 2021).A high broad sense heritability of approximately 0.85 and above was recorded for all the traits including leaf and yellow rust APR and leaf and yellow rust ASR in both populations.A similar range of broad sense heritability was also observed in previous reports (Rollar et al., 2021;Gao et al., 2016).The magnitude of correlations between leaf and yellow rust APR was relatively low, although a slight positive correlation was observed in the Anza population.

Conclusion
The study with two RIL populations derived from a cross between Almaly × Anza (160 RILs) and Almaly × Avocet S (206 RILs) suggested the presence of wide variability for yellow and leaf rust APR and ASR.We identified a set of 13 consistent QTLs including yellow rust APR (9 QTLs) and leaf rust APR (4 QTLs).Among them, QLR-APR-2B and QYR-APR-4D.2 from the Anza population and QLR-APR-5A.2,QYR-APR-4D.1, QYR-APR-4D.2, and QYR-APR-3D from the Avocet population are important candidates to target for further validation and deployment in LR and YR resistance breeding.Several putative candidate genes were identified in this study; mainly, zinc finger proteins, DNA-binding pseudobarrel domain superfamily, and NAC domain superfamily with the associated functions in the resistance mechanism of leaf and yellow rust were identified.The functional characterization of these candidate genes will provide greater applicability of this study in rust resistance breeding.

TABLE 4
QTLs identified for yellow and leaf rust resistance in Almaly × Anza RIL population locations for 3 years.

TABLE 4 (
Continued) QTLs identified for yellow and leaf rust resistance in Almaly × Anza RIL population locations for 3 years.

TABLE 5
QTLs identified yellow and leaf rust resistance in Almaly × Avocet RIL population locations for 3 years.

TABLE 5 (
Continued) QTLs identified yellow and leaf rust resistance in Almaly × Avocet RIL population locations for 3 years.

TABLE 6
Putative candidate genes for leaf and yellow rust adult plant resistance.