Files
Abstract
Yield is the primary agronomic trait of soybean (Glycine max (L.) Merr.) and the key determinant of a given cultivar’s value. This research focused on methods which will enable improvement of soybean yield through genomic selection, cross prediction, and modification of growth habit. Genomic selection is a predictive breeding tool. This technique uses genetic information to determine genotype values for agronomic and seed composition traits. A comparison of models indicated that extended genomic BLUP provided the greatest prediction accuracy for yield and seed composition traits (protein, oil, methionine, cysteine). Additionally, inter-environment validation methods led to statistically reduced accuracies compared to cross validation. In an empirical evaluation, prediction accuracies for yield were similar to those of inter-environmental evaluation, with the same relative performance of models. Of the SNP sets tested the SoySNP3k was preferable, acting as the most resource efficient SNP array while maintaining prediction accuracies. Prediction accuracies for all traits were comparable to those found in more structured populations. The genomic selection methods developed here could aid breeders in their selection decisions.
An additional predictive method studied in this research is optimal cross selection. Optimal cross selection evaluates the genetics of parental candidates and determines which pairs would result in the most productive populations. Of the prediction models tested, extended genomic BLUP achieved the highest prediction accuracies. The relatedness of the training set to the crosses predicted, as well as marker density, were two other significant factors determining prediction accuracy. Predictive ability of optimal cross selection was sufficiently high to be useful in breeding efforts.
Growth habit is a major developmental trait in soybean, regulating transition from vegetative to reproductive growth. Prior research has indicated that introgression of a semi-indeterminate growth habit into determinate growth habit cultivars could lead to increases in yield. Genetic marker assays are needed for growth habit introgression. Six KASP assays were developed and tested using two breeding populations and a diversity panel of soybean accessions. A high level of concordance was found between phenotypic and genotypic growth habit assessment. The assays can be utilized by soybean breeding programs to enable growth habit introgression.