Work package 1 - WP Leader: AGE
Multi-environment trials of a set of 400 advanced breeding lines
The field trials (five locations, two in Austria and three in Germany each year; Table 1; Annex1) will form the basis for phenotyping to calibrate marker-based selection models in WP 2 and WP 3. The trials will include 400 advanced inbred lines evenly derived from each project partner. The lines should be unselected for PHS and represent material from European wheat breeding programs involving multiple crosses. Based on the trial layout, the lines will be grouped according to maturity and replicated once in each location because a spatial influence on the investigated traits is negligible. A set of 200 lines will be evaluated in both years to form the calibration set. Each year will contain 100 lines for validation of prediction models.
Work Package 2: WP Leader: AGES
Phenotypic assessment of PHS tolerance
The success of the marker-based selection methods in WP 3 heavily depends on the accurate evaluation of the phenotype of interest. In this work package, it is intended to evaluate breeding material consisting of 400 lines for PHS, a trait showing different heritabilities in multiple environments across years. Improved methods will be applied to enhance the assessment of PHS tolerance in unselected material.
Work Package 3: WP Leader: LfL
Calculation of genetic values with genomic prediction methods and implementation in breeding programs Within this work package, all lines will be genotyped with a high-density SNP chip and a marker array built in-house including 24 descriptive markers. These descriptive markers are associated to PHS and other quality related traits. Based on this genotypic information and the phenotypic evaluation in WP2, different marker-based selection methods will be developed within the calibration set comprising 200 lines. The marker-based selection methods will be the basis for comparing predictions of genetic values based on low- and high-density marker panels. To predict genetic values, prediction models will be implemented assuming homogeneous or marker-specific shrinkage of regression coefficients. For the implementation of marker-based selection strategies into breeding programs, the optimal stage for developing a calibration set will be evaluated.
Work Package 4: WP Leader: LfL
Validation and comparison of marker-based selection strategies for PHS tolerance
Predicted genetic values will be validated with a statistical approach, i.e. cross-validation, and on the validation set within the first and second year. For Cross-validation, different sampling procedures based on the family structure will be applied to divide the data set into an estimation and test set. For validation, the correlation between predicted and observed genetic values as well as the ranking of lines within each year will be evaluated.