Fusarium Head Blight (FHB) Is a major disease caused by F. grammearum, that infects wheat
(Tritfcum aestfvum L) ano other cereals. one Important aspect for managing FHB in wheat Is breeding tor
resistant varieties. However, evaluating FHB within a breeding program takes a large amount of resources.
Marker assisted selection (MAS) has been effective ror traits controlled by major genes. but most olthe
genes controlling FHB resistance are not effectively deployed by conventional MAS. Genomic selection
(GS) is a new rorm of MAS and can facilitate breeding ror complex traits by estimating all marker effects
simunaneousty and predicting the genomic estimated breeding values (GEBVs) that will be used as
selection criteria. GS has the potential to Increase lhe genetic gain per year by decreasing the time per
cycle. The challenge remains now in implementing GS and identifying the model with the highest prediction
accuracy for each trait. we evaluated the predlction accuracy of several GS models in a population of 640
soft winter wheat lines. The populatlon was evaluated for lncklence (INC). severity (SEV). index (IND).
Fusarium damaged kernel (FOK). kernel damage index (ISK). and deoxynlvalenol concentration (DON) in
inoculated FHB nurseries In multiple environments and genotypic data was obtained through genotyping by
sequencing (GBS). Ten-lokl cross valldatlon prediction al>lltties ranged rrom 0.45 (INC) to 0.57 (SEV).
Similar prediction accuracies were obtained within clusters but were much lower when data from one
cluster was used to predict another. Eliminating the top 10-15% less predictable individuals increased
predictlon accuracy by up to 58%. The results from this work will facilitate GS lmplementatlon and the
identification of the best lines for selection and crossing for FHB resistance within this population.