Genetic architecture of adult-plant resistance to stripe rust in bread wheat (Triticum aestivum L.) association panel

Abstract

Stripe rust, caused by Puccinia striiformis f. sp. tritici, is a severe disease in wheat worldwide, including Ethiopia, causing up to 100% wheat yield loss in the worst season. The use of resistant cultivars is considered to be the most effective and durable management technique for controlling the disease. Therefore, the present study targeted the genetic architecture of adult plant resistance to yellow rust in 178 wheat association panels. The panel was phenotyped for yellow rust adult-plant resistance at three locations. Phonological, yield, yield-related, and agro-morphological traits were recorded. The association panel was fingerprinted using the genotyping-by-sequencing (GBS) platform, and a total of 6,788 polymorphic single nucleotide polymorphisms (SNPs) were used for genome-wide association analysis to identify effective yellow rust resistance genes. The marker-trait association analysis was conducted using the Genome Association and Prediction Integrated Tool (GAPIT). The broad-sense heritability for the considered traits ranged from 74.52% to 88.64%, implying the presence of promising yellow rust resistance alleles in the association panel that could be deployed to improve wheat resistance to the disease. The overall linkage disequilibrium (LD) declined within an average physical distance of 31.44 Mbp at r² = 0.2. Marker-trait association (MTA) analysis identified 148 loci significantly (p = 0.001) associated with yellow rust adult-plant resistance. Most of the detected resistance quantitative trait loci (QTLs) were located on the same chromosomes as previously reported QTLs for yellow rust resistance and mapped on chromosomes 1A, 1B, 1D, 2A, 2B, 2D, 3A, 3B, 3D, 4A, 4B, 4D, 5A, 5B, 6A, 6B, 7A, and 7D. However, 12 of the discovered MTAs were not previously documented in the wheat literature, suggesting that they could represent novel loci for stripe rust resistance. Zooming into the QTL regions in IWGSC RefSeq Annotation v1 identified crucial disease resistance-associated genes that are key in plants' defense mechanisms against pathogen infections. The detected QTLs will be helpful for marker-assisted breeding of wheat to increase resistance to stripe rust. Generally, the present study identified putative QTLs for field resistance to yellow rust and some important agronomic traits. Most of the discovered QTLs have been reported previously, indicating the potential to improve wheat resistance to yellow rust by deploying the QTLs discovered by marker-assisted selection.

Keywords: Puccinia striiformis; genome wide association study; linkage disequilibrium; marker assisted breeding; novel loci; quantitative trait loci; yellow rust.

Figure 1

Yellow rust disease severity (DS) of 178 bread wheat genotypes at three locations. Clustering involved resistant (0≤DS ≤ 10), intermediate (10

Figure 2

Field reaction response (FR) of 178 bread wheat germplasms to yellow rust: Resistant (0≤FR ≤ 0.2), intermediate (0.2

Figure 3

CI of yellow rust for 178 bread wheat genotypes obtained from field experiments in three environments. The genotypes were clustered as resistant (0≤CI ≤ 2), intermediate (2< CI<48), and susceptible (48≤ CI ≤ 90).

Figure 4

Frequency distribution of some yellow rust resistance traits for the combined data from three locations. The right and left ends of the bars indicate the uppermost and lowermost infection classes, respectively. The disease resistance at the pre-heading, heading, and mid-maturity stages and combined followed a virtually pseudo-normal distribution. YRPH, coefficient of infection at pre-heading; YRH, coefficient of infection at heading; YRF, coefficient of infection at flowering; YRMM, coefficient of infection at mid-maturity; YRM, coefficient of infection at maturity; and Kulmsa, Kulumsa.

Figure 5

Distribution of DArTSeq SNPs on 21 bread wheat chromosomes.

Figure 6

Population structures of 178 bread wheat genotypes representing eight populations. (A) Best delta K value estimated, and the pick at k = 3 indicates the number of sub-populations in the wheat panel. (B) Estimated population structure for K = 3 according to the breeding materials. The different colors (blue, orange, and black) represent genetic groups or sub-populations: the x-axis represents individual samples and the y-axis represents the proportion of ancestry to each cluster. Population abbreviations are IBWSN, International Bread Wheat Screening Nursery; ISEPTON, International Septoria Observation Nursery; HRWYT, High Rain Wheat Yield Trial; HRWSN, High Rain Wheat Screening Nursery; ADAPT, Adaptation trial; NVT, National Verification Trial; and PVT, Preliminary Verification trial.

Figure 7

Principal component and familiar relatedness analysis of 178 wheat genotypes using 6,788 SNP markers. (A) Scatter plot and (B) 3D plots of the principal components. (C) Kinship displayed through a heat map and a tree out of the heat map. The kinship values showed a normal distribution (turquoise curve), and orange represents a weak correlation between pairs of individuals in the panel while red shows a high correlation. The resulting clustering tree is indicated outside of the matrix.

Figure 8

Genome-wide LD decay plot over physical distance based on 6,788 SNP markers. The yellow curve represents the model fits to LD decay. The horizontal magenta dash-line represents the arbitrary threshold for no LD (r² = 0.2). The vertical blue line indicates the intersection between the critical r² value and the average map distance (31.44 bp) to determine QTL confidence intervals.

Figure 9

Some examples of Manhattan plots for the coefficient of infection and GWAS scans resulting in significant associations. Each dot represents an SNP. On the x-axis is the genomic position of the SNPs on the corresponding chromosomes indicated in different colors. On the y-axis is the -log10 of the p-value depicting the significance of the association test. The horizontal orange line is the nominal p-value 0.001 significance threshold used in the association analysis for YRPH, coefficient of infection at pre-heading; YRH, coefficient of infection at heading; YRF, coefficient of infection at flowering; YRMM, coefficient of infection at mid-maturity; and YRM, coefficient of infection at maturity. The quantile-quantile (Q-Q) plots at the right side of the Manhattan plots indicate how well the used BLINK model accounted for population structure and kinship for each of the disease traits. In each plot, the observed –log (P values) from the fitted GWAS models (y-axis) are compared with their expected value (x-axis) under the null hypothesis of no association with the trait. Each blue dot represents a single nucleotide polymorphism; the red line is the model for no association.

Figure 10

Some examples of genomic positions of detected putative QTLs effective for the coefficient of infection. Significant DArTSeq SNPs are presented according to their physical positions on chromosomes in million base pairs. The putative QTLs identified in this study for the MTAs are indicated on the right side of the bars. YRPH, coefficient of infection at pre-heading; YRH, coefficient of infection at heading; YRF, coefficient of infection at flowering; YRMM, coefficient of infection at mid-maturity; YRM, coefficient of infection at maturity; HD, days to heading; FD, days to flowering; DM, days to maturity; TKW, thousand kernel weight; LA, leaf area; PH, plant height; SL, spike length; NSs/S, number of spikelets per spike; NK/S, number of kernels per spike; NK/Ss, number of kernels per spikelets; SW, spike weight; and GYPP, grain yield per plot.

Figure 11

Some examples of Manhattan plots for agronomic and yield-related traits in each environment and combined data. HD, days to heading; FD, days to flowering; DM, days to maturity; TKW, thousand kernel weight; LA, leaf area; PH, plant height; SL, spike length; NSs/S, number of spikelets per spike; NK/S, number of kernels per spike; NK/Ss, number of kernels per spikelets; SW, spike weight; and GYPP, grain yield per plot. The quantile-quantile plots at the right side of the Manhattan plots indicate how well the GWAS model accounted for population structure and kinship for each of the disease traits. In each plot, the observed –log (p-values) from the fitted GWAS models (y-axis) are compared with their expected value (x-axis) under the null hypothesis of no association with the trait. Each blue dot represents a single nucleotide polymorphism; the orange line is the model for no association.