Cadmium (Cd) is a toxic heavy metal found naturally in the soil. In humans it primarily enters the body through food consumption, posing risks to organs like the lungs, liver, kidneys, and bones. Wheat genotypes might accumulate different levels of Cd in the grain and therefore, breeding for low grain-Cd accumulation is important to ensure human health.

This project aims to identify winter wheat breeding lines with low grain-Cd content and develop breeding tools for a more precise and affordable selection strategy. For this purpose, 324 winter wheat breeding lines will be evaluated for grain-Cd content across three locations in south of Sweden (Bollerup, Svalöv, Kölbäck) and three years. It has been documented that Cd content in the soil can also influence the Cd accumulation in the grain. Therefore, 25 soil samples per location were evaluated for soil-Cd and ten for pH, clay, and soil organic matter (SOM).

In addition, all fields were scanned with an EM38 gamma ray sensor. Using this soil information a model was developed to predict the soil variables of each plot. The grain-Cd, soil and phenotypic and genotypic data were used to develop linear mixed model genomic prediction models and perform genome-wide association studies. The highest grain-Cd content was found in the lines in the location with the highest soil-Cd level. The Cullis broad sense heritability for grain-Cd across the three locations was high (H² = 0.86).

Using the GBLUP model, the predictive ability was moderate to high across locations (r=0.54), in Bollerup (r=0.56) and Svalöv (r=0.45) whereas a lower predictive ability was seen for Kölbäck (r=0.37). Incorporating either single or multiple soil variables as covariates doesn’t change the predictive ability of the models both across and at specific locations.

Because of low genotype by environment interaction and high predictive ability and heritability in Bollerup, breeders can screen their material in this location and use genomic selection. GWAS across locations resulted in four marker-trait associations on chromosomes 3D, 4A, 5D, and 6D explaining between two and six percent of the variance.

The developed genomic prediction models and identified markers will help breeders to develop simple and affordable prediction models and diagnostic markers for selection of low grain-Cd accumulating wheat genotypes.