Facilitating hands‐on plant breeding training: A case study in genetic gain and genomic selection in a student‐led barley breeding program

Abstract

The University of Nebraska Barley ( Hordeum vulgare L.) Breeding Program is led by doctoral students who aim to develop high‐yielding winter barley varieties. The student breeders are responsible for crossing, data collection, and advancement decisions, which provides firsthand career preparation experience. However, the impact of student leadership on the success of the program is unknown. This study used a historical yield dataset of 302 unique advanced lines grown in Nebraska from 2002 to 2022 to evaluate the realized genetic gain of the breeding program and as a training population for genomic selection (GS). For yield GS, entries from a new observation nursery were used as the testing set. GS models included variations of genomic best linear unbiased prediction, four Bayesian models (Bayes A, Bayes B, Bayes C, Bayes LASSO [Bayesian least absolute shrinkage and selection operator]), and two machine learning approaches (random forest and support vector machine). The realized genetic gain for yield was 62.4 kg/ha/year or 2.2% annual increase. The Bayes A model trained on a single location (Lincoln, NE) achieved the highest predictive accuracy for yield ( r  = 0.420). Implementing GS for yield was successful in the Nebraska barley breeding program and can aid future student barley breeders by providing selection continuity between leadership transitions.