Genetic dissection of snow mold tolerance in US Pacific Northwest winter wheat through genome-wide association study and genomic selection

Authors: LOZADA, DENNIS - Washington State University; GODOY, JAYFRED - Washington State University; MURRAY, TIMOTHY - Washington State University; Ward, Brian; CARTER, ARRON - Washington State University

Submitted to: Frontiers in Plant Science
Publication Type: Peer Reviewed Journal
Publication Acceptance Date: 9/25/2019
Publication Date: 10/29/2019
Citation: Lozada, D., Godoy, J.V., Murray, T.D., Ward, B.P., Carter, A.H. 2019. Genetic dissection of snow mold tolerance in US Pacific Northwest winter wheat through genome-wide association study and genomic selection. Frontiers in Plant Science. https://doi.org/10.3389/fpls.2019.01337.
DOI: https://doi.org/10.3389/fpls.2019.01337

Interpretive Summary: Snow mold is a disease of winter wheat which is caused by several different fungi species when plants are exposed to dark, high-humidity conditions underneath snow cover for extended periods of time. It is a yield-limiting disease, and is particularly problematic in the Pacific Northwest, where fields may be snow-covered for most of the winter. This experiment utilized a genome-wide association study (GWAS) to identify genomic regions associated with tolerance to snow mold. The GWAS was performed using an association mapping panel of 458 lines planted in Mansfield, WA and Waterville, WA in 2017 and 2018, and identified 100 significant tolerance regions, located on 17 of the 21 total wheat chromosomes. Regions of snow mold tolerance located on chromosomes 5A and 5B coincided with major freezing tolerance and vernalization (i.e. cold temperature exposure requirement) loci, respectively. Lines with increasing numbers of favorable snow mold tolerance alleles combined together demonstrated higher snow mold tolerance in the field. Due to the high number of significant genomic regions identified, the use of genomic selection (GS) was also investigated for increasing snow mold tolerance. GS is a method which evaluates all genomic regions influencing a trait simultaneously to predict how a line will perform. It utilizes a training population, which is genotyped with genome-wide markers and evaluated for snow mold tolerance, and a testing population, which is genotyped with markers. In this experiment, snow mold tolerance data was collected on the testing population as well, to evaluate the accuracy of predictions generated by the GS model. However, in practice the predictions would be used to make selections from the testing population without performing any in-field data collection. The association mapping panel of 458 lines was used as the training population, while sets of lines evaluated in the field from 2015 to 2018 were used as testing populations. The mean accuracy of predictions across these years was 0.36. Selecting lines based on their model-predicted snow mold tolerance values resulted in a 24% increase in actual observed snow mold tolerance. Taken together, the results of this study show that a) improving snow mold tolerance by combining specific genomic regions associated with tolerance will be difficult, due to the large number of significant regions identified, and b) GS shows promise for increasing snow mold tolerance by leveraging information from across the wheat genome to make predictions.

Technical Abstract: Snow mold is a yield-limiting disease of wheat in the Pacific Northwest (PNW) region of the US, where there is prolonged snow cover. The objectives of this study were to identify genomic regions associated with snow mold tolerance in a diverse panel of PNW winter wheat lines in a genome-wide association study (GWAS) and to evaluate the usefulness of genomic selection (GS) for snow mold tolerance. An association mapping panel (AMP; N = 458 lines) was planted in Mansfield and Waterville, WA in 2017 and 2018 and genotyped using the Illumina® 90K single nucleotide polymorphism (SNP) array. GWAS identified 100 significant markers across 17 chromosomes, where SNPs on chromosomes 5A and 5B coincided with major freezing tolerance and vernalization loci. Increased number of favorable alleles was related to improved snow mold tolerance. Independent predictions using the AMP as a training population (TP) to predict snow mold tolerance of breeding lines evaluated between 2015 and 2018 resulted in a mean accuracy of 0.36 across models and marker sets. Modeling nonadditive effects improved accuracy even in the absence of a close genetic relatedness between the TP and selection candidates. Selecting lines based on genomic estimated breeding values and tolerance scores resulted in a 24% increase in tolerance. The identified genomic regions associated with snow mold tolerance demonstrated the genetic complexity of this trait and the difficulty in selecting tolerant lines using markers. GS was validated and showed potential for use in PNW winter wheat for selecting on complex traits such tolerance to snow mold.