Quantitative trait loci, trait correlations, and accuracy of genotypic value predictions for maize stover cell wall composition and glucose release for cellulosic ethanol

The efficiency of converting maize (Zea mays L.) stover into fermentable sugars for cellulosic ethanol production partly depends on stover cell wall composition. Quantitative trait loci (QTL) information may guide marker-assisted selection strategies in breeding for improved stover quality, but limited QTL studies have been published for stover cell wall components and glucose release. The first two objectives of this study were to examine the relationships among stover cell wall components and glucose release, and to identify QTL with major effects influencing these traits. Testcrosses of 223 intermated B73 x Mo17 recombinant inbreds were analyzed for stover cell wall composition and glucose release. Genetic and phenotypic correlations were generally favorable and reflected the complexity of stover cell wall composition. Most of the 152 QTL found for glucose release and cell wall components had small effects.;Because there were no major QTL, methods that increase favorable QTL allele frequencies or predict performance, such as marker-assisted recurrent selection or genomewide selection, would be appropriate in marker-assisted breeding for stover quality. However, published information is limited on the accuracy of genotypic value predictions with empirical data in plants. Further objectives were therefore to assess the accuracy of genotypic value predictions from multiple linear regression (MLR) and genomewide selection via best linear unbiased prediction (BLUP) in biparental plant populations; assess the accuracy of predictions for different numbers of markers ( NM) and progeny (NP) in estimation; and determine if an empirical Bayes approach for modeling variances of individual markers and epistatic effects leads to more accurate predictions in empirical data. Accuracy of predictions was assessed by cross validation using the data for the 223 testcrosses, and data for three other maize datasets, one Arabidopsis thaliana (L.) Heynh. dataset, and two barley ( Hordeum vulgare L.) datasets. Accuracy of predictions was higher with BLUP than with MLR and increased as NP and NM increased. Modeling marker variances led to better predictions for some traits, but for many traits BLUP provided comparable or greater accuracy. Including epistasis did not improve the predictions. The simple BLUP approach is sufficient for predicting genotypic value in biparental plant populations.