Genome-Wide Association Study and QTL Meta-Analysis Identified Novel Genomic Loci Controlling Potassium Use Efficiency and Agronomic Traits in Bread Wheat

Abstract

Potassium use efficiency, a complex trait, directly impacts the yield potential of crop plants. Low potassium efficiency leads to a high use of fertilizers, which is not only farmer unfriendly but also deteriorates the environment. Genome-wide association studies (GWAS) are widely used to dissect complex traits. However, most studies use singlelocus one-dimensional GWAS models which do not provide true information about complex traits that are controlled by multiple loci. Here, both single-locus GWAS (MLM) and multi-locus GWAS (pLARmEB, FASTmrMLM, mrMLM, FASTmrEMMA) models were used with genotyping from 90 K Infinium SNP array and phenotype derived from four normal and potassium-stress environments, which identified 534 significant marker-trait associations (MTA) for agronomic and potassium related traits: pLARmEB = 279, FASTmrMLM = 213, mrMLM = 35, MLM = 6, FASTmrEMMA = 1. Further screening of these MTA led to the detection of eleven stable loci: q1A, q1D, q2B-1, q2B-2, q2D, q4D, q5B-1, q5B-2, q5B-3, q6D, and q7A. Moreover, Meta-QTL (MQTL) analysis of four independent QTL studies for potassium deficiency in bread wheat located 16 MQTL on 13 chromosomes. One locus identified in this study (q5B-1) colocalized with an MQTL (MQTL _11 ), while the other ten loci were novel associations. Gene ontology of these loci identified 20 putative candidate genes encoding functional proteins involved in key pathways related to stress tolerance, sugar metabolism, and nutrient transport. These findings provide potential targets for breeding potassium stress resistant wheat cultivars and advocate the advantages of multi-locus GWAS models for studying complex traits.

Identification of genetic factors controlling phosphorus utilization efficiency in wheat by genome-wide association study with principal component analysis

“…A spring wheat diversity panel of 150 genotypes including landraces, pre-green revolution, post-green revolution, and elite varieties was used; the details of the panel were published in our earlier paper (Safdar et al 2020). The phenotypic experiment was carried out using a previously published cigar roll method (Zhu et al 2005).…”

“…The phenotypic experiment was carried out using a previously published cigar roll method (Zhu et al 2005). This experiment was planted in 3 replicates and the growth conditions were the same as published earlier (Safdar et al 2020), except for P concentration. The experiment was divided into two treatments: normal P supply; and, reduced P supply to evaluate the phenotypic variation under P deficiency stress.…”

“…Population structure (Q) was estimated using STRUCTURE software, whereas a relative kinship matrix (K) and linkage disequilibrium (LD) were estimated using TASSEL software. The information about genotype, Q and LD has been provided in our previous studies (Ain et al 2015;Safdar et al 2020). GWAS was performed for P, P utilization efficiency, PC1 and PC2 using the commonly used mixed linear model in TASSEL software (Yu and Buckler 2006).…”

“…Population linkage disequilibrium decay was 800 kb in sub-genome B, 500 kb in sub-genome D, and 300 kb in sub-genome A. Details about each dataset, population structure, and linkage disequilibrium are reported in our earlier studies (Ain et al 2015;Safdar et al 2020) -readers are encouraged to see those papers for further details.…”

“…Therefore, we conducted a 21-day experiment in controlled lab conditions using a cigar roll method to check the phenotypic variation of seedling traits, P uptake and P utilization efficiency. For the genomic variation, we used a wheat 90K SNP data published in our earlier study for this diversity panel (Safdar et al 2020). To correlate the phenotypic and genotypic variation, we followed a method recently proposed by Yano et al, where they performed principal component analysis for the phenotypic data to bring the correlated variable into a smaller number of uncorrelated principal components (PCs), which captured the most important information from the original data (Yano et al 2019).…”