European Journal of Sustainable Agroecosystems

Background: Grain protein content (GPC) is an important trait determining wheat quality and nutritional value; however, improving GPC under water-limited conditions is a challenge for breeding programs. Genomic prediction models provide opportunities for improving genetic gain through genomic-assisted selection, especially in harsh agro-climatic environments.
Objective: In this study, we tested the ability of genomic prediction models to predict grain protein content in a diverse panel of wheat germplasm under well-watered and water-limited field conditions. We focused on the identification of genomic regions and candidate genes controlling protein accumulation under drought stress.
Methods: We phenotyped a panel of 287 breeding lines of bread wheat (Triticum aestivum L.) at four locations under two water regimes in two growing seasons. A high density SNP array (35,000 SNPs) was used for genotyping, and genomic prediction models were developed using ridge regression best linear unbiased prediction (rrBLUP), Bayesian sparse linear mixed models (BSLMM) and random forest algorithms. In both environmental conditions, cross-validation was used to evaluate prediction accuracy.
Results: Genomic prediction models predicted GPC with prediction accuracies ranging from 0.64 to 0.78 under well-watered conditions and from 0.58 to 0.71 under water-limited conditions. BSLMM models showed better performance with both treatments. We detected 47 significant SNP markers on chromosomes 1A, 3A, 5A, 6A and 7D which accounted for 52.3% of the phenotypic variance for GPC. Candidate genes related to nitrogen metabolism, amino acid biosynthesis and aquaporin function were identified in target genomic regions. The heritability of GPC was reduced from 0.71 to 0.58 under water stress, but the genomic prediction models were still accurate under water stress conditions.
Conclusions: Genomic prediction enhances breeding efficiency for grain protein content in water-limited environments. The identified genomic regions and candidate genes offer potential targets for marker-assisted selection and functional validation in wheat improvement programmes. Genomic prediction and phenotypic selection accelerates the development of drought-resilient, high-protein wheat varieties