Genomewide association study of ionomic traits on diverse soybean populations from germplasm collections

Abstract

The elemental content of a soybean seed is a determined by both genetic and environmental factors and is an important component of its nutritional value. The elemental content is chemically stable, making the samples stored in germplasm repositories an intriguing source of experimental material. To test the efficacy of using samples from germplasm banks for gene discovery, we analyzed the elemental profile of seeds from 1,653 lines in the USDA Soybean Germplasm Collection. We observed large differences in the elemental profiles based on where the lines were grown, which lead us to break up the genetic analysis into multiple small experiments. Despite these challenges, we were able to identify candidate single nucleotide polymorphisms (SNPs) controlling elemental accumulation as well as lines with extreme elemental accumulation phenotypes. Our results suggest that elemental analysis of germplasm samples can identify SNPs in linkage disequilibrium to genes, which can be leveraged to assist in crop improvement efforts.

Keywords: GRIN; GWAS; diversity panel; ionomics; multi‐locus mixed model; soybean.

Figure 1

Molybdenum accumulation in single soybean seeds (mg/kg) across experimental grow‐outs

Figure 2

Distribution of cadmium phenotype (linear model residuals, see Methods) in lines from a single grow‐out: Stoneville, MS, 1999. Lines are ordered by median of between two and eight seed replicates

Figure 3

Principal component analysis of the genotypes of 1,391 soybean lines. Colored by country of origin: China (532), Japan (267), South Korea (200), Russia (61), other, or unknown country of origin (331)

Figure 4

Quantile–quantile plot of the observed p‐values against expected p‐values from the GWAS analysis for sulfur accumulation. The MLMM algorithm includes cofactors that reduce inflation of p‐values (green line). The model without cofactors indicates presence of p‐value inflation (blue line). The expected distribution of p‐values under the null hypothesis (red line)

Figure 5

Confirmation grow‐out of high and low sulfur and phosphorus accumulating lines. (a) Regrow versus original concentration of eight lines selected for high and low phosphorus accumulation. Correlation between GRIN concentration and regrow was 0.24. (b) Regrow versus original concentration of 10 lines selected for high and low sulfur accumulation, regrown in both greenhouse and field environments. Error bars indicate the standard error of the replicate seeds. Correlation (r ²) between GRIN seed concentrations and the regrown high and low varieties grown in the greenhouse and in the fields were 0.61 and 0.84, respectively