Genomewide Selection and Prospective Targeted Recombination in Elite Maize Biparental Crosses

Abstract

Genomewide selection is now routine in maize (Zea mays L.). The first two studies in this dissertation investigated advanced aspects of genomewide selection, whereas the last study investigated the potential of targeted recombination in elite biparental crosses. The three studies utilized data on 969 maize populations that were phenotyped for yield, moisture, and test weight at multiple years and locations and genotyped with 2911 single nucleotide polymorphism markers. The first study showed that prior selection in the training population reduced the response to selection and predictive ability, but it did not increase the similarity among the best lines. Including a small number of the poorest lines in the training population nearly restored the predictive ability to its original level. The second study showed that in genomewide selection, it is better to use a smaller ad hoc training population than a single, large training population. In particular, the response to selection and predictive ability were lower in a global training population with about 50,000 lines than in a set of about 4500 lines chosen to maximize relatedness with the population undergoing genomewide selection. The third study showed that on average, targeted recombination doubled the predicted gains. For each trait, the gains with targeted recombination were 60% to 400% of the gains from nontargeted recombination. Targeted recombination did not increase gains in only around 4% of the populations. The results indicated that targeted recombination is a potentially powerful sequel to genomewide selection in maize.