Best linear unbiased prediction of triticale hybrid performance

Predicting single-cross performance is of special interest in hybrid breeding of triticale. We used molecular and phenotypic data of factorial triticale crosses and compared several approaches to predict their single-cross performance. Twenty-three inbred lines and their 76 incomplete factorial crosses were field evaluated for grain yield, plant height, and heading time at five locations in Central Europe. In addition, the parental lines were genotyped with 52 SSR markers. Plant height and heading time were predicted with high accuracy based on mid-parent performance. In contrast, prediction of hybrid performance based on mid-parent value was not accurate for grain yield. Using general combining ability effects led to an enhanced prediction accuracy of hybrid grain yield performance. This accuracy could be slightly improved using best linear unbiased prediction approaches. The prediction accuracy was considerably high even if the number of tested hybrids was small. Consequently, best linear unbiased prediction of hybrid performance is a promising tool for hybrid triticale breeding programs.