Multi-model genome-wide association study on key organic naked barley agronomic, phenological, diseases, and grain quality traits

Abstract

The study objective was to assess the potential benefits of using genomic tools in organic plant breeding programs to enhance selection efficiency. A diversity panel of 247 spring naked barley accessions was characterized under Irish organic conditions over 3 years. Genome-wide association studies (GWAS) were performed on 19 traits related to agronomy, phenology, diseases, and grain quality, using the information on 50 K Single Nucleotide Polymorphisms (SNP). Four models (EMMA, G model, BLINK, 3VMrMLM) were applied to 5 types of Best Linear Unbiased Predictors (BLUP): within-year, mean, aggregated within-year). 1653 Marker-Trait-Associations (MTA) were identified, with 259 discovered in at least two analyses. 3VMrMLM was the best-performing model with significant MTA together explaining the largest proportion of the additive variance for most traits and BLUP types (from 1.4 to 50%). This study proposed a methodology to prioritize main effect MTA from different models’ outputs, using multi-marker regression analyses with markers fitted as fixed or random factors. 36 QTL, considered major, explained more than 5% of the trait variance on each BLUP type. A candidate gene or known QTL was found for 18 of them, with 13 discovered with 3VMrMLM. Multi-model GWAS was useful for validating additional QTL, including 8 only discovered with BLINK or G model, thus allowing a broader understanding of the traits’ genetic architecture. In addition, results highlighted a correlation between the trait value and the number of favorable major QTL exhibited by accessions. We suggest inputting this number in a multi-trait index for a more efficient Marker-Assisted Selection (MAS) of accessions best balancing multiple quantitative traits.