Genomic Selection for Fusarium Head Blight Resistance in Durum Wheat
Ragupathy, R., Ruan, Y., Berraies, S., Campbell, H., Meyer, B., Henriquez, M.A., Kumar, S., Burt, A., Zhang, W., Cuthbert, R.D., Fobert, P., and Knox, R.E. 2019. Genomic Selection for Fusarium Head Blight Resistance in Durum Wheat. XXVII International Plant and Animal Genome Conference, San Diego, CA, USA, January 12-16, 2019.
Fusarium head blight (FHB) is a major disease of durum wheat. FHB infection leads to reduced yield, poor end-use quality and accumulation of mycotoxins (deoxynivalenol-DON) in the grain. Several studies suggested that the resistance to FHB is quantitative and involves multiple loci with relatively small effects. Hence, prediction of breeding values using genome wide markers is a promising strategy for selection and accelerating genetic gain. In this study, different statistical models namely ridge regression-best linear unbiased predictor (rrBLUP, Endelman 2011), Bayes B (Meuwissen et al. 2001) and reproducing kernel Hilbert spaces (RKHS, Gianola and van Kaam 2008) regression accounting for simple marker additive effects, large-effect QTL and epistasis, are being evaluated for implementation in breeding populations. A durum diversity panel consisting of 195 individuals, and two doubled haploid populations from the crosses DT707/DT696 and Strongfield/Blackbird were genotyped using the 90K Infinium iSelect SNP array. The diversity panel was evaluated for disease incidence, severity and FHB Index in the replicated field nurseries in 2015, 2016 and 2017 at two Canadian locations, namely Brandon and Morden. Phenotypic data analysis was carried out and least square (ls) means of lines in individual locations combined over years, as well as for individual years combined over locations were estimated. Genotype calling using GenomeStudio software (Illumina) resulted in the identification of 6,899 SNPs among 195 accessions of the durum panel. Genomic prediction using rrBLUP with 80% of the lines as the training set, and the remaining lines as the validation set with 100 iterations, suggested average prediction accuracies of 0.57 and 0.52 for Morden and Brandon, respectively. Similarly, the mean prediction accuracy was found to be 0.48, 0.34 and 0.39 for the years 2015, 2016 and 2017. Development of genomic prediction models using Bayes B and RKHS methods is ongoing.