Integrating genomics and enviromics data for grain yield prediction in Southern US oat germplasm

Authors: ADEWALE, SAMUEL - University Of Florida; BABER, MUHAMMAD ALI - University Of Florida; JARQUIN, DIEGO - University Of Florida; KHAN, NAEEM - University Of Florida; HARRISON, STEPHEN - Louisiana State University; DEWITT, NOAH - Louisiana State University; BOYLES, RICK - Clemson University; LIU, SHUYU - Texas A&M Agrilife; MELSON, ELLEN - Texas A&M Agrilife; HATHCOAT, DANIEL - Texas A&M Agrilife; Fiedler, Jason; Nandety, Raja Sekhar

Submitted to: The Plant Genome
Publication Type: Peer Reviewed Journal
Publication Acceptance Date: 8/24/2026
Publication Date: N/A
Citation: N/A

Interpretive Summary: Oat is an important small grain cereal crop cultivated globally due to its health benefits. Development of commercially superior oat varieties that meet market needs requires the use of different modern crop improvement tools and strategies. In this study we evaluated the combination of environmental and genomic information with an artificial intelligence-based method to predict performance of grain yield and other traits. The predictive ability of performance based on the artificial intelligence-based model was lower than simpler models, but the addition of environmental data resulted in improved predictive ability. This work showcases next-generation prediction models that enable breeders to efficiently select improved lines, leading to a faster rate of releasing new varieties for American farmers.

Technical Abstract: Genetic gains of oat grain yield have been historically low compared to other major cereal crops. The use of machine learning models to capture complex interactions and leveraging data types other than genomic information in prediction models has great potential for improving complex traits in oat breeding programs. The objectives of this study were to assess the performance of a deep learning model for genomic prediction compared to other statistical models, determine the optimal training set size for grain yield prediction, and investigate the potential of incorporating environmental covariates in enhancing the prediction of oat grain yield. A total of 463 oat lines were evaluated in five environments in the southern United States and genotyped using 12,657 single-nucleotide polymorphism (SNP) markers. Our results showed that a training set size of 200-300 was the optimal size, beyond which there was no significant improvement in predictive abilities, indicating the possibility of reducing phenotyping costs by reducing the size of the oat panel tested. The deep learning model was less superior to genomic best linear unbiased prediction (GBLUP) and other models for grain yield, test weight, and heading days in the different environments. Incorporating environmental effects (G × E or G × W) into the multikernel prediction models across environments resulted in 43% to 91% increase in predictive abilities for grain yield and other traits compared to using genomic data alone. This reveals the potential of incorporating weather data in genomic prediction models. Our findings provide important information for improving genetic gains in oat breeding by integrating genomics and environmental data.