Genome-wide association mapping for yield and other agronomic traits in an elite breeding population of tropical rice (Oryza sativa)

Abstract

Genome-wide association mapping studies (GWAS) are frequently used to detect QTL in diverse collections of crop germplasm, based on historic recombination events and linkage disequilibrium across the genome. Generally, diversity panels genotyped with high density SNP panels are utilized in order to assay a wide range of alleles and haplotypes and to monitor recombination breakpoints across the genome. By contrast, GWAS have not generally been performed in breeding populations. In this study we performed association mapping for 19 agronomic traits including yield and yield components in a breeding population of elite irrigated tropical rice breeding lines so that the results would be more directly applicable to breeding than those from a diversity panel. The population was genotyped with 71,710 SNPs using genotyping-by-sequencing (GBS), and GWAS performed with the explicit goal of expediting selection in the breeding program. Using this breeding panel we identified 52 QTL for 11 agronomic traits, including large effect QTLs for flowering time and grain length/grain width/grain-length-breadth ratio. We also identified haplotypes that can be used to select plants in our population for short stature (plant height), early flowering time, and high yield, and thus demonstrate the utility of association mapping in breeding populations for informing breeding decisions. We conclude by exploring how the newly identified significant SNPs and insights into the genetic architecture of these quantitative traits can be leveraged to build genomic-assisted selection models.

Fig 1. Selected Manhattan plots for flowering time (FLW, top), length-breadth ratio (LBR, top middle), plant height (PH, bottom middle), and grain yield (YLD, bottom).

Dashed line shows the 0.1 FDR significance threshold.

Fig 2. Physical map of significant GWAS QTL.

Black points—jittered GBS SNPs, red triangles—physical position of the most significant SNP for a given peak, blue rectangles—physical position of candidate flowering time genes.

Fig 3. Graphic representation of the correlation matrices of phenotype values for the (A) dry season, and (B) wet season.

Fig 4. Phenotypic distributions for flowering time in the dry season (top), plant height in the dry season (middle), and yield in the wet season (bottom), showing the most desirable 5% of individuals (red) and least desirable 5% of individuals (blue).

The most significant haplotypes associated with each end of the distribution are shown to the left and right of the histogram with the number of individuals in the best or worst 5% that carry the respective haplotypes. "Confirmed other" refers to individuals that were known NOT to carry the most significant haplotype. Individuals that were neither confirmed to carry the significant haplotype or confirmed to carry other haplotypes had missing data at one or more SNPs that did not preclude the possibility of the individual carrying the significant haplotype.