Genomic Selection in CWRS and CWAD Wheat Breeding

Funding Partners: SeCan

Project Description

Genomic selection which utilizes genome-wide markers to predict the breeding value of lines within a population is not a new concept, having been proposed by Meuwissen et al in 2001. The strategy has been used for some time in animal breeding where the cost of genotyping is relatively small compared to the value of the animals. Genomic selection is more recent in crop breeding with the reduced cost of genotyping now making it feasible to assess individual plants more cheaply than doing phenotypic assessment. Many important traits selected for through breeding, such as yield, grain end use quality, and disease resistance are quantitative and under genetically complex control being based on multiple small effect genes. Genomic selection is a method of selecting traits under such complex genetic control.

The proposed strategy in this project is to apply genomic selection as early as possible within the breeding programs that is cost effective based on the current cost of genotyping and number of lines under consideration. Our strategy to continue to apply genomic selection at the F4 generation is based on a pilot study discussed more in the next section of this application ”˜how does this project build on your previous research or research of others’. By applying genomic selection at an early generation the gene pool of breeding populations will be enriched for desirable alleles in the same way that phenotypic selection is done for agronomic and disease traits such as height and rust resistance. Genomic selection, however, confers the advantage of enriching the gene pool for complex genetic traits, such as FHB resistance, grain yield and quality, that are not selected for phenotypically until later generations due to cost. By applying genomic selection early in the breeding process, expensive multi-location agronomic and specialized disease nursery field trialing is not wasted on breeding lines with a low potential to become commercial cultivars. Genomic selection will be applied after rounds of phenotypic selection in the F2 and F3 generations for highly heritable traits such has height and certain forms of disease resistance. This will reduce the cost of genotyping, by applying genotyping to the most acceptable lines in a population based on phenotypic selection.

Research Results

Key Achievements and Results

Harnessing the cutting-edge technology of high-throughput genotyping and genomic selection early in the breeding process, SCRDC breeders were able to eliminate worthless breeding lines to maximize valuable field resources and shorten cultivar development times.

This technology allowed the breeding program to continue to progress during the pandemic lockdown, by performing selection based on computer modelled predictions rather than physically rating and measuring the plants in the field.

Multi-trait selection based on predictions was especially useful for negatively correlated traits. For instance when selecting germplasm with genetics conferring high protein concentration in wheat, grain yield is adversely affected. As part of the GS modelling, it was possible to analyze multiple traits together in a ‘a pareto front’ where each line in the breeding material was assigned an optimized rank with the highest ranked germplasm containing a good balance of both protein and yield, allowing breeders to carry desirable levels of both traits to subsequent breeding rounds.

Genomic selection was also found to be especially useful for quality and milling traits. These traits are very expensive to measure making it economically unfeasible to test for desirable characteristics in early breeding generations due to the large number of lines. Prior to the advent of GS, breeders evaluated these traits in later generations that were smaller in number after culling lines with undesirable agronomics or disease susceptibility. However, in doing this, breeders risked inadvertently culling material with desirable quality and milling characteristics. By harnessing genomic selection, breeders need only measure quality and milling characteristics of the TRS in order to predict these traits in thousands of lines at early generation breeding cycles, allowing breeders to identify and propagate these traits through the breeding program.

Lessons Learned and Next Steps

Since Genomic Selection models generate predictions based on statistical likelihood, prediction accuracy is important. To mitigate the potential risk of misidentifying, and therefore failing to advance, genetically superior germplasm based on prediction inaccuracy, SCRDC breeders have learned to use GS to ‘cull inferior’ germplasm rather than ‘advance the best’ germplasm. In doing so, GS is applied conservatively as a tool to continually reduce overhead while enriching the gene pool.

GS models can be continually improved by integrating previous years’ data. As a next step, SCRDC scientists will look to further tune existing GS models to improve prediction accuracies.

Stakeholder Significance

Ultimately, Genomic Selection allows for faster development of new wheat varieties with improved genetics translating into less risk and higher profits for Canadian farmers.