Genome wide association mapping through 90K SNP array against leaf rust pathogen in bread wheat genotypes under field conditions

2.1 Germplasm collection

The total 105 bread wheat genotypes were studied in this experiment. According to the maintaining sources, the germplasm divided into three groups as mentioned in Supplementary Table 1 (Ahmed et al., 2019). In first group the genotypes G-1 to G-20 developed in the Department of Plant Breeding and Genetics, University of Agriculture Faisalabad (PBG-UAF), Pakistan, while second group genotypes G-21 to G-55 were from exotic source and third group genotypes G-56 to G-105 were from indigenous sources.

2.2 Experimental site and sowing conditions

Studied germplasm were sown by hand drill for screening against leaf rust in research area of Department of Plant Breeding and Genetics, The Islamia University of Bahawalpur (29⁰ 24′N latitude, 71⁰ 41′E longitude and 214 m above sea level) Pakistan under randomized complete block design (RCBD) in field with three replications to evaluate the leaf rust susceptible and resistance genotypes during the two seasons 2018–19 and 2019–20. For analysis purpose averaged data based on over years were used. Each genotype was planted in plot size of 1.2 m × 2.5 m and the experimental plots were surrounded by planting three rows of highly susceptible genotype Morocco. Inoculation was done artificially by means of various methods like dusting with talcum powder, rubbing, spraying with distilled water and needle injection methods on Morocco twice in a week at tillering and heading stage for the development of a heavy rust infection pressure (Hussain et al., 2015).

2.3 Data recording of leaf rust under field conditions

Disease severity of leaf rust in percentage and host response were observed by modified Cobb’s scale described by (Peterson et al., 1948). Disease severity was recorded four time with 10 days interval when Morocco showed 40–50% rust severity. Rating of the final disease severity (FDS) was recorded when Morocco variety showed 90–100% disease severity. The values of coefficient of infection (CI) were calculated by the equation described by (Pathan and Park 2006). Area under disease progress curve (AUDPC) was estimated for each genotype by using the following equation (Anwaar et al., 2019). $$\mathit{AUDPC}=d\left[1/2\left(y_1+y_k\right)+\left(y_2+y_3+-----+y_{k-1}\right)\right]$$ where d = days between two consecutive records (time intervals)

• y₁ + y_k = Sum of the 1st and last disease scores

• y₂ + y₃ + − − − − − + y_k-1 = Sum of all in between disease scores.

2.4 Statistical analysis

Recorded data were subjected to analysis of variance (ANOVA) (Steel et al., 1997) using the GenStat (v10) software (Payne et al., 2008). The significance level = 0.01 was used for highly significant effects and = 0.05 was used for significant effects. RADAR-graph were developed for mean value using Excel-Stat in which display values relative to a centre point for examined traits under both environments (Ahmed et al., 2020).

2.5 DNA extraction and genotyping

In green-house, wheat seeds were sown in small plastic trays for healthy seedlings. After three weeks, fresh leaves were collected for DNA according to the CIMMYT Molecular Genetics Manual procedure in 96 well-plates. The concentration and quality of isolated DNA was assessed by Nano-drop (ND1000, Thermo Scientific, USA). The DNA of each genotype (70–100 ng/μl) preserved in 96-well plate and prepared for genotyping with high-density illumina 90 K Infinium SNP array (Dreisigacker et al., 2013). After genotyping, the genome-wide positions of SNPs in terms of genetic distance (cM) situated on chromosomes founded on consensus genetic map of bread wheat (Wang et al., 2014). For further analysis SNPs were filtered and excluded the monomorphic SNPs, more than 20% missing SNPs, minor alleles and allelic frequency < 5% for GWAS in the present study.

2.6 Population STRUCTURE and GWAS analysis

Bayesian clustering technique was used to classify the group of genetically similar population via statistical software STRUCTURE v.2.3 (Pritchard et al., 2000). Burn-in iterations of 10⁴ cycles, followed by a simulation run of 10⁶ cycles and the admixture model selection were used. Web-based analysis “Structure Harvester v0.6.93” was applied to obtain maximum value or peak of ‘‘K’’ for validation to understand the STRUCTURE results which were based on ad-hoc techniques. We selected the K values ranged 1–10 and 6 independent runs to attain reliable effects.

GAPIT (genome association and prediction integrated tool) was also applied with the model selection preference to test the reliability of the results. It was advanced in R package which offer maximum likelihood precision and run in a computationally effective method. GAPIT implements unconventional statistical approaches containing the compressed mixed linear model (CMLM) and CMLM-based genomic prediction and selection. P-values, R² and marker effects were extracted from GWAS results (Lipka et al., 2012). The threshold level for significant marker-trait associations (MTA) was 10⁻³ (log10p) or above (Rahimi et al., 2019) after applying the false discovery rate (FDR) < 0.05 correction using Bonferroni adjustment (P = 1/n, n = total number of SNPs) (Benjamini and Hochberg 1995). Mixed linear model (MLM) estimated from newly developed GWAS. To define the spurious associations derived from population structure, covariates from either STRUCTURE or principal components (PCs) were considered as fixed effects. The relationships among individuals were calculated using a kinship matrix and incorporated MLM. Overall 18,500 SNPs from functional iSelect bead chip analyses visually showed polymorphism; to locate them on the published genetic map in the studied genotypes.

3.1 Field evaluation of disease severity parameters against leaf rust

A field evaluation was conducted of 105 wheat varieties over two growing seasons 2018–19 and 2019–20 against leaf rust disease using disease severity Coefficient of infection (CI), Area under disease progress curve (AUDPC) and Final Disease Severity (FDS) parameters. Mean squares results from analysis of variances (ANOVA) showed highly significant (P ≤ 0.001) effects among investigated wheat genotypes for coefficient of infection under field conditions (Table 1). The mean responses of studied wheat genotypes to leaf rust under field conditions are shown in Fig. 1. In 2014–15 growing season, all investigated wheat genotypes showed the significant values of coefficient of infection. Data analysis and mean comparison indicated that wheat cultivars were considerably different based on the coefficient of infection values. The values 10–20, 21–30, 31–45 and 46–80 were categorized as high, moderate, low and very low levels of resistant, respectively. During this study, an effort was also done to elucidate the association between described parameters. Among the studied germplasm, G-36 and G-27 and G-102 showed highest values of coefficient of infection 77.6, 77.6 and 76.0 followed by G-6 and G-42 having values of 63.6 and 63.0 respectively, thus, these were classified as susceptible wheat genotypes. However, G-28 showed the minimum value of coefficient of infection 12.0 and followed by G-30, G-35, G-38, G-2, G-11, G-23, G-16 and G-48 having values of 14.6, so were ranked as resistant genotypes against leaf rust pathogen (Fig. 1). The described results are in harmony with the results of (Anwaar et al., 2019). (Ali et al., 2009) investigated 20 wheat genotypes and ‘Morocco’ as susceptible check at NIFA for describing variability for field based partial resistance to rust and lines with coefficient of infection values of 0–20, 21–40 and 41–60 were considered as possessing better, moderate and low levels of partial resistance, respectively. Only Morocco (a susceptible check) was possessing 100 CI value.

Close

Fig. 1

Performance of 105 wheat genotypes against leaf rust pathogen using Final Disease severity (FDS), Area under disease progress curve (AUDPC) and Coefficient of Infection (CI) parameters based on data averaged over two seasons 2018–19 and 2019–20 under field conditions.

Export to PPT

Analysis of variances results showed highly significant (P ≤ 0.001) effects among studied wheat genotypes for area under disease progress curve under field conditions (Table 1). Data analysis and mean comparison indicated that leaf rust infection was well established apparently in all the tested wheat cultivars screened for the disease. The results of the mean values of area under disease progress curve of different wheat varieties to leaf rust is shown in Fig. 1. The genotypes G-102, G-36, G-27 and G-6 were ranked as susceptible to area under disease progress curve (AUDPC) with the values of 1600, 1550, 1350 and 1300 respectively, whereas G-42, G-11 and G-2 were ranked as resistant having the minimum value 300 of area under disease progress curve. The rest of the cultivars had the range of value from 450 to 850 were classified as moderately resistant (MR) to moderately susceptible (MS) genotypes. Results of current study showed that AUDPC run in a parallel manner to rust severity. It is obvious from the previous findings by Safavi (2015), Taye et al. (2015), Anwaar et al. (2019) that a very strong positive relation found between AUDPC and final rust severity that means severely infected cultivar showed higher AUDPC values. And the second thing was disease progress rates and AUDPC values were positively correlated and highly significant.

Disease damage or lesions covered on host tissues or organ described in percentage is called disease severity. Severity results from the number and size of the lesions. Data analysis and mean comparison indicated that 105 wheat genotypes were significantly different based on the final disease severity (FDS) parameter. ANOVA Results showed highly significant (P ≤ 0.001) effects among investigated wheat genotypes for disease severity under field conditions (Table 1). Mean results of the response of different studied wheat genotypes against leaf rust disease under field conditions are shown in Fig. 1. All the studied wheat genotypes exhibited different disease severity ranged from 30 to 80%. The genotypes of G-102, G-27 and G-36 showed the highest final rust severity of 80% followed by G-6 and G-42 which had the value of 70% rust severity and classified as leaf rust susceptible genotypes. While the minimum final disease severity value was 30 % showed by the genotypes G-30, G-37, G-25 and some other cultivars, thus, were more resistant to the pathogen of the leaf rust as indicated in Fig. 1. Current results are in a harmony with results reported by Macharia and Wanyera (2012), Taye et al. (2015). A study by Hasan et al. (2012) was aimed to estimate of losses of grain yield because of brown rust disease on the five local commercial susceptible wheat cultivars in field conditions during 2011–2012 growing seasons at Gemmeiza Agriculture Research Station, Egypt. The investigated cultivars exhibited 5–80% disease severity. And was depicted a relationship between rust severity and yield components which resultantly converted to financial losses.

3.2 Population STRUCTURE

Bayesian method execute in statistical software package STRUCTURE used to estimate the genetic structure of 105 bread wheat genotypes. The results from this technique showed that highest (peak) number of K = 4 which demonstrating the germplasm distributed into four sub-population (Fig. 2A). Different types of coloured in Fig. 2B exhibits the distinct group and overall germplasm allocated into four sub-groups. This technique has been applied in wheat breeding scheme by many scientists and were obtained the explanatory outcomes (Wang et al., 2021). In current experiment, existence of different groups clearly showed the genetic variations between 105 bread wheat genotypes and genetically diverse to one another. Fundamentally, this is the indication of genetic divergence between the clusters or groups and resultantly the presence of more genetic diversity in studied germplasm. Several wheat breeders evaluated the genetic diversity (Rahimi et al., 2019; Leonova et al., 2020) using the similar techniques which were applied in current study and some extent they got the similar results. Development of novel bread wheat genotypes should be attaining the significance level of genetic diversity. Presence of more variation in 105 bread wheat genotypes which indicate the maximum genetic diversity, fearlessly, that the studied germplasm introduced from different sources or assumable mechanical mixing.

Close

Fig. 2

(A) This result achieved of 105 wheat genotypes using 90 k SNP Array from Structure Harvester analysis. It's based on the second order derivation on the variance of the maximum likelihood estimation of model given a specific K. Delta K shows only the uppermost clustering level and number of subpopulations in main population. (B) Population structure of 105 wheat genotypes based on Bayesian method analysed with 90 k SNP Array detecting 4 groups. The dissimilar colours in this figure demonstrating the different group.

Export to PPT

According to the provided pedigree record there are three groups of 105 bread wheat genotypes as shown in Supplementary Table 1. In first group, genotypes G-1 to G-20 which was developed in PBG-UAF, while in second group the genotypes G21 to G-55 were from exotic source, and in third group, genotypes G-56 to G-105 was from indigenous sources. But according to molecular analysis these genotypes divided into four populations. The maximum genetic distance between groups exhibited indicating genetic similarity within groups and genetic dissimilarity between the groups (Lipka et al., 2012; Zegeye et al., 2014). Particularly, results were useable conferring to the previously known pedigree record and origin of wheat genotypes. Genetic diversity evaluation could be helpful to identify the different genotypes for the advancement and improve the future wheat breeding scheme. The genotypes with different genetic makeup can be selected for desirable combinations to develop complex and significant attributes to obtaining maximum yield. Discrimination of wheat genotypes based on their genetic basis would be useful for effective and early selection of desired genotypes in wheat breeding scheme for developing promising wheat genotypes (Juliana et al., 2018; Leonova et al., 2020).

3.3 Genome-wide markers–traits associations for disease severity parameters

Marker-trait association (MTA) study recognized the connection between particular morphological and genetic variation within a genome, which ultimately perceived locus underpinning related characters at the end. In this study, 18,500 high density SNP markers from the 90 K Illumina iSelect SNP array were evaluated to perceive SNPs associated with disease severity for leaf rust related indices. Before analyses of GWAS and genomic prediction, scientists should validate and maintain genotype quality. The GAPIT provides a series of diagnostic tools to help users perform quality control on genotypes (Lipka et al., 2012). Total 56 significant SNPs were correlated with observed characters at or above –log 10 (P < 0.0001) threshold level using MLM (mixed linear model) for studied traits as mentioned in Table 3. The significance level for p values were measured using Bonferroni adjustment at P ≤ 10⁻³ after crossed FDR < 0.05 threshold, some scientists previously used this criteria with small population size as in current study (Qaseem et al., 2018; Rahimi et al., 2019). Manhattan plots (Figs. 3–5) presentation the site of significant SNPs at (P < 0.0001) −log10(p) which significantly linked with the desired characters under studied conditions. The red horizontal line on Manhattan plot entitles the threshold level (P < 0.0001) of significance for SNPs with specific traits. Current results were supported with the findings of (Rahimi et al., 2019; Leonova et al., 2020). In this study, 18,500 high density, polymorphic SNP markers from 90 K Illumina iSelect SNPs array (Wang et al., 2014) were examined to notice SNPs associated with disease severity for leaf rust related associated indices.

Close

Fig. 3

Manhattan plots showing the location of significant SNPs and −log10(p) associated with Coefficient of Infection (CI) in 105 wheat genotypes against leaf rust pathogen under field conditions. The red horizontal line on Manhattan plot entitles the threshold level at P ≤ 10⁻³ of significance for SNPs with specific traits.

Export to PPT

Close

Fig. 4

Manhattan plots showing the location of significant SNPs and −log10(p) associated with Area under disease progress curve (AUDPC) in 105 wheat genotypes against leaf rust pathogen under field conditions. The red horizontal line on Manhattan plot entitles the threshold level at P ≤ 10⁻³ of significance for SNPs with specific traits.

Export to PPT

Close

Fig. 5

Manhattan plots showing the location of significant SNPs and −log10(p) associated with Final Disease severity (FDS) in 105 wheat genotypes against leaf rust pathogen under field conditions. The red horizontal line on Manhattan plot entitles the threshold level at P ≤ 10⁻³ of significance for SNPs with specific traits.

Export to PPT

A total of 17 significantly marker trait associations were found with coefficient of infection, Including 5 markers located on chromosomes 4A, 4 on chromosome 2A, 3 on 7B and the other on 1A, 2B, 3A, 5A and 7A (Fig. 3). In GWAS analysis, wsnp_Ex_c5412_9564046 on chromosome 2A at position 261.05 cM and Kukri_c4709_53 on chromosome 4A were significantly correlated with this trait covering 20.49% and 16.60% of the phenotypic variation respectively. Reliable MTAs were detected for disease severity parameters which located on the 5A, 6D, 6A and 6A chromosomes by wheat scientists (Singh et al., 2009). Associations with SNPs of chromosome 5A were also established a link with current study. In addition, association mapping revealed a Genomic region on chromosome 5A, which determines the presence of resistance loci against leaf rust. Based on the location of QTLs, as well as their origin, the locus on 5A found in this study cannot be attributed to any of the known genes (Singh et al., 2009; Maccaferri et al., 2010; Leonova et al., 2020). This suggests the presence of a previously unknown QTL for resistance to leaf rust in the genomes of studied germplasm.

Total 23 significant SNP markers were strongly linked with Area under disease progress curve (AUDPC) including 6 markers located on chromosomes 3B, 2 on each chromosome 2B, 5A, 6B, 7B, 7D and the other on 1B, 6A, 6D and 7A (Fig. 4). These markers explained 11.64% to 23.03% of the phenotypic variation in AUDPC under field conditions. The marker (D_contig29746_525) on chromosome 7D at 436.21 cM explained maximum variation 23.03% while the marker (BS00037898_51) on 3B at 274.8 cM explained (Table 3) minimum variation (11.64%) in this parameter. In the study of (Wang et al., 2021) 32 markers located on seven chromosomes were identified which similar to current study results, as significantly associated (P < 0.00o1) with FDS and AUDPC. The identified resistant accessions and resistance loci will be useful in the ongoing effort to develop new wheat cultivars with strong resistance to leaf rust (Yao et al., 2019; Wang et al., 2021).

Final Disease Severity (%) was highly associated with sixteen QTL regions. Four QTLs associated regions were located on chromosomes 5A, three on 1A, two on each 1B, 2A and 7A, one on each 3B and 3 D (Fig. 5). These MTAs explained 12.97 % to 19.39 % of the variation in FDS (Table 3). The marker (Ku_c10756_1197) explained maximum phenotypic trait variability (19.39%) on chromosome 5A at 690.36 cM while the marker (BobWhite_c15453_678) on chromosome 2D at 208.94 cM explained minimum value (12.97 %) of phenotypic variability. MTA for FDS were distributed across 16 chromosomes, including, 11 SNPs at A-genome, 3 at B-genome and 2 at B-genome. Leaf rust resistance genes designated Lr1 to Lr60 have been have been characterized in common (Bolton et al., 2008). These resistance genes are widely distributed across all the chromosomes of the wheat genome. In present study, significant SNPs were observed and matched with previously identified location of genes by several scientists. Genes Lr2a, Lr2b and Lr2c were mapped to a locus on chromosome 2D, Chromosomes 6B exhibited the Lr3a, Lr3ka and Lr3g genes against leaf rust (Li et al., 2016) The other genes Lr17a and Lr17b are at a locus on chromosome 2A and Lr22a and Lr22b at a locus on chromosome 2D. Genes Lr14a and Lr14b are extremely tightly linked on chromosome 7B and are considered as alleles for all practical purposes (Bolton et al., 2008; McIntosh et al., 2014).

Based on the disease severity parameters of FDS and AUDPC (Macharia and Wanyera 2012), reported infection responses of the wheat cultivars and classified them into four discrete categories: a) S-susceptible) MS-moderately susceptible c) MR-moderately resistant d) R-resistant which was previously reported by several scientists. Resistant wheat varieties are compromised by the continuously evolving races of rust pathogens, which causes leaf rust of wheat and appears to be the most damaging pathogen that threatens global food security by inducing yield reductions in wheat (Li et al., 2016). Diseased symptoms are prevalent on leaf blades, leaf sheaths and glumes. Pathogenic attack in the early crop stages may result in increased yield losses up to 30% (Anwaar et al., 2019).

In the current study total 105 genotypes were investigated, among them 34 genotypes showed resistant results against leaf rust pathogen. Moderately resistance showed by 30 genotypes against this disease in studied germplasm. Total 32 genotypes exhibited moderately susceptible performance to leaf rust pathogen while 9 genotypes showed susceptible behaviour against this pathogen (Table 2). Leaf rust resistant accessions recognised in this experiment verified to be additional fruitful in preliminary field environments against leaf rust pathogen. Thus, these resistant genotypes could be helpful as parental genotypes in breeding schemes for leaf rust resistance. The locus (Kukri_c55051_414) showed pleotropic effects for leaf rust identified parameters on chromosome 5A at position of 68.2 cM under field conditions\. The main objectives of this study were to identify sources of resistance, and to map genomic loci associated with leaf rust resistance using genome wide association study (GWAS) approach. The new identified sources of resistance and MTAs/QTL could be used in wheat breeding programs to improve leaf rust resistance for sustainable food security.