|LORENZ, AARON - University Of Nebraska|
|SMITH, KEVIN - University Of Minnesota|
Submitted to: Crop Science
Publication Type: Peer Reviewed Journal
Publication Acceptance Date: 3/15/2012
Publication Date: 7/2/2012
Citation: Lorenz, A.J., Smith, K.P., Jannink, J. 2012. Potential and optimization of genomic selection for fusarium head blight resistance in six-row barley. Crop Science. 52(4):1609-1621. DOI: 10.2135/cropsci2011.09.0503.
Interpretive Summary: Fusarium head blight (FHB) is a devastating disease of barley, causing reductions in yield and quality. Marker-based selection for resistance to FHB and lowered deoxynivalenol (DON) grain concentration would save considerable costs and time associated with phenotyping. An approach called genomic selection (GS) is ideal for traits like FHB that are affected by many genes and uses genome-wide marker information to predict genetic value. We evaluated the potential of GS for genetic improvement of FHB and DON. We found high prediction accuracies, with the correlation between prediction and observation as high as 0.72 and 0.68 for FHB and DON, respectively. Accuracies remained high even when the number of markers used was reduced to 384, or when the population on which the prediction model was built had as few as 200 individuals. We found little to no advantage from increasing the population size by combining observations on lines from different breeding programs. The high accuracies and minimal dataset sizes indicate that genomic selection will be a powerful approach to increase resistance to FHB in barley.
Technical Abstract: Fusarium head blight (FHB) is a devastating disease of barley, causing reductions in yield and quality. Marker-based selection for resistance to FHB and lowered deoxynivalenol (DON) grain concentration would save considerable costs and time associated with phenotyping. A comprehensive marker-based selection approach, called genomic selection (GS), uses genome-wide marker information to predict genetic value. We used a cross-validation approach that separated training sets from validation sets by both entry and environment. We used this framework to test the potential of GS for genetic improvement of FHB and DON, as well as test the effect of different factors on prediction accuracy. Based on our findings, we are confident GS can enhance genetic gain for these traits per unit time and cost. Prediction accuracy for FHB was found to be as high as 0.72, and that for DON was found to be as high as 0.68. Little difference was found between marker effect estimation methods in terms of prediction of entry genetic value. The extensive linkage disequilibrium (LD) present in this population allowed the marker set to be reduced to 384 markers and TP size to be reduced 200 with little effect on prediction accuracy. We found little to no advantage to combining subpopulations in order to increase TP size. Apparently, little genetic information is shared between subpopulations, either because of different marker-QTL linkage phases, different segregating QTL, or non-additive gene action.