Location: Sunflower and Plant Biology Research
Project Number: 3060-21220-034-019-S
Project Type: Non-Assistance Cooperative Agreement
Start Date: Jul 1, 2021
End Date: Dec 31, 2024
1. To evaluate a genomic prediction model for white mold disease in a dry bean training population using genotype-by-sequencing (GBS); 2. To optimize the training population for different dry bean market classes in relation to the testing population, in order to increase genomic prediction accuracy for white mold disease; and 3. To evaluate a multi-trait genomic prediction model for white mold resistance, yield, and canning quality.
For Objective 1, a total of 250 advanced breeding lines from the Michigan State dry bean breeding and genetics program that have been previously phenotyped in the National White Mold trials (N=96) and including additional lines from the breeding program (N=154) will be used to start the initial training population (TR). Additional lines not previously phenotyped will be evaluated at Montcalm Research Farm near Entrican, MI, in a field naturally infested with white mold and under pivot irrigation to induce white mold pressure. White mold disease will be rated as a combination of incidence and severity on a scale of 1 to 9 and grain yield and canning quality traits will also be collected. Phenotypic data will be analyzed using linear mixed models that account for either experimental design features such as incomplete blocks, rows, and columns or spatial variation in the field using ASReml-R and the SpATS R package in the R software. All breeding lines will be genotyped using the GBS assay and a GBLUP model for white mold and yield will be fitted individually using the rrBLUP package in R to determine predictive ability. For Objective 2, a total of (N=100) breeding lines from a preliminary yield trial (PYT) will be used as the testing set. These lines will be genotyped following protocols in Obj. 1, but not phenotyped. Using the TR developed in Obj. 1 and the TE developed in this objective, we will perform an optimization method between both the TR and TE using a weighted relationship matrix with stratified sampling algorithm and protocol. Genomic prediction abilities and accuracies will be estimated similarly as in Obj. 1 within the TR and using the TE to validate genomic prediction. For Objective 3, single trait (ST) and multi-trait (MT) models for genomic prediction of yield, canning quality, and white mold disease resistance will be developed. For the ST model, the same procedure from Obj. 1 will be used for model development and cross validation. For the MT model, a Bayesian multivariate Gaussian model will be used. Predictions will be obtained using the ‘MTM’ package in R and genomic prediction abilities will be evaluated within the TR and using the TR to predict TE. Finally, the best MT model combined with the best strategy from Obj. 2 will be used to select individuals for white mold resistance from the TE. A group of 5 lines will be selected and validated for white mold resistance under greenhouse and field conditions for white mold.