Genetic and genomic tools for improving end-use quality in wheat

Abstract

Wheat accounts for 20% of daily caloric intake of the world population and has one of the widest cultivation distributions of any crop. With increasing demand for both quantity and quality, wheat yields must increase while also maintaining acceptable end-use quality. However, measuring end-use quality is complex, requires large volumes grain and significant effort. The overarching goal of this dissertation research was to develop genetic and genomic tools to facilitate breeding for end-use quality in wheat. Building on initial work with genomic prediction of wheat quality, we continued application of genomic prediction models to the International Maize and Wheat Improvement Center (CIMMYT) wheat breeding program. For practical application in the breeding program to advance selection, we focused on forward prediction in each cycle of the bread wheat program. Models were built on 12 years of past data including over 18,000 entries with quality data. Predictions for 10,000 yield trial lines were generated each year for selection, with forward prediction accuracies of 0.40 to 0.73, and approached heritability. This is one of the largest scale applications of genomic selection. We also studied the interaction of climate change and the important quality genes, high-molecular weight glutenins (HMW-GS) and low-molecular weight glutenins (HMW-GS). A diverse panel of 54 CIMMYT wheat varieties were grown in 2 levels of drought stress, heat stress and optimal growth conditions. Quality traits, HMW-GS and LMW-GS alleles were measured. We fit a mixed linear model for each quality trait with HMW-GS, LMW-GS, environment, and the interactions of those as predictors. Overall, the superior glutenin alleles either maintained or increased quality in stressful environments. This work confirmed that superior alleles should always be selected for, regardless of target environment. To increase the genetic diversity for wheat quality, we analyzed Glu-D1 gene diversity on the wheat D genome donor, Aegilops tauschii. We constructed Glu-D1 molecular haplotypes from sequence data of 234 Ae. tauschii accessions and found 15 subclades and over 45 haplotypes, representing immense gene diversity. We found evidence that the 5+10 allele originated from a newly described Lineage 3 of Ae. tauschii, further supporting that this unique lineage contributed to modern bread wheat. We also observed rare recombinant haplotypes between the x and y subunits of any HMW-GS locus. This work will facilitate incorporation of Ae. tauschii Glu-D1 alleles into modern wheat. Given that certain HMW-GS alleles are highly desirable, we set out to develop a high-throughput, high resolution genotyping method for HMW-GS alleles that would fit within genotyping already done for genomic prediction models. This ‘sequence based genotyping’ approach uses diagnostic k-mers developed to predict alleles in skim-sequenced breeding material. Prediction accuracies for Glu-D1 and Glu-A1 were very good, but lower for the Glu-B1 alleles where many alleles are highly related. Overall, SBG offers a high throughput method to call alleles from existing data. These genetic and genomic tools developed and implemented for end-use quality selection in wheat offer promising resources for continued improvement of both yield and quality in wheat breeding.