Genomic Prediction of the F1 Performance in Sorghum

Sorghum bicolor (L.) Moench is expected to be the most promising crop for the biofuel production because of the large biomass and the high sugar content in the stalk. The F₁ hybrid is mainly used for biofuel. To efficiently accelerate the breeding of Sorghum, the performance of F₁ needs to be accurately predicted. In this study, we predicted the performance related to the biofuel production in the F₁ Sorghum using genome-wide markers. We used two CMS lines, CMS-A and CMS-B, as the seed parent to make the F₁ series. About 200 Sorghum accessions were crossed with these CMS lines. In 2013, we cultivated the F1 series and the parent lines in Japan and Mexico. We examined seven traits, fresh weight; flowering state; height; stalk number; stalk diameter (only in Japan); stalk juice content; brix value of stalk juice. We applied two methods for the genomic prediction. One is GBLUP, which is based on kinship matrix. The other is PHENIX, which is based on a multiple-phenotype mixed model. Using GBLUP, we could accurately predict the F₁ traits with high marker-based heritability, such as height and stalk diameter. Using PHENIX, the prediction accuracy increased in proportion to the correlation among traits.