Simultaneous Improvement of Grain Yield and Grain Protein Concentration in Durum Wheat by Using Association Tests and weighted GBLUP

Despite the importance of grain protein concentration (GPC) in determining wheat quality, its negative correlation with grain yield (GY) is still one of the major challenges for breeders. Here, a durum wheat panel of 200 genotypes was evaluated for GY, GPC, and their derived indices (GPD and GYD), under eight different agronomic management. The plant material was genotyped with the Illumina 25k iSelect array, and a genome-wide association study was performed. Two statistical models revealed dozens of marker-trait associations (MTAs), each explaining up to 30%. phenotypic variance. Two markers on chromosomes 2A and 6B were consistently identified by both models and were found to be significantly associated with GY and GPC. MTAs identified for phenological traits co-mapped to well-known genes (i.e., Ppd-1, Vrn-1). The significance value (p-values) that measure the strength of the association of each SNP marker with the target traits were used to perform genomic prediction by using a weighted genomic best linear unbiased prediction (WGPLUP) model. This statistical model outperformed conventional GBLUP for all traits (prediction accuracy increase up to 70%). The trained models were ultimately used to predict the agronomic performances of an independent durum wheat panel, confirming the utility of genomic prediction, although environmental conditions and genetic backgrounds may still be a challenge to overcome. The results generated through our study confirmed the utility of GPD and GYD to mitigate the inverse GY and GPC relationship in wheat, provided novel markers for marker-assisted selection and opened new ways to develop cultivars through genomic prediction approaches.