Modeling grain biochemical composition traits of commercial sorghum hybrids under diverse management practices

Authors: GANO, BOUBACA - Donald Danforth Plant Science Center; COQUEREL, MARIE - Donald Danforth Plant Science Center; SAXTON, JOCELYN - Donald Danforth Plant Science Center; ECK, NATHANIEL - Donald Danforth Plant Science Center; Peiris, Kamaranga; Bean, Scott; SHAKOOR, NADIA - Donald Danforth Plant Science Center

Submitted to: Plant Phenomics
Publication Type: Peer Reviewed Journal
Publication Acceptance Date: 1/26/2026
Publication Date: 2/16/2026
Citation: Gano, B., Coquerel, M.D., Saxton, J., Eck, N., Peiris, K.H., Bean, S.R., Shakoor, N. 2026. Modeling grain biochemical composition traits of commercial sorghum hybrids under diverse management practices. Plant Phenomics. Volume 17 - 2026. https://doi.org/10.3389/fpls.2026.1768456.
DOI: https://doi.org/10.3389/fpls.2026.1768456

Interpretive Summary: Sorghum is a versatile cereal crop grown the central U.S. especially in Kansas and Texas and is an important grain for feed, biofuel, and food markets. The growing demand for sorghum has increased the need for enhanced hybrids with superior grain composition, such as high protein and starch content, which are key determinants of its nutritional and economic value. The composition of sorghum grain is highly influenced by genetics, growth conditions, and crop management practices, all of which influence the final yield and quality of the crop. To develop new varieties of sorghum with improved grain traits, grain composition must be measured on large sample sets grown at multiple locations and often with different crop practices, which is currently a labor-intensive and time-consuming process. Thus, this research investigated the ability of machine learning to predict plant growth features and grain composition collected by high throughput techniques and determine relationships between crop management practices and grain composition and ultimately end-use value. By identifying optimal variety-management combinations and leveraging non-invasive, high-throughput plant and grain analysis, this study offers a scalable framework for real-time decision-making and targeted field interventions to improve sorghum varieties.

Technical Abstract: Sorghum (Sorghum bicolor L. Moench) is a vital cereal crop for food, feed, and biofuel production. Predicting its seed biochemical composition—crude protein (CP), lysine (Lys), starch (SC), amylose (AML), and crude fat (CF)—is crucial for improving breeding and management strategies. This study used machine learning (ML) models to predict these traits in commercial sorghum hybrids under different management practices, including precision nitrogen application, cover cropping, and no-till methods. Multi-year field trials (2023–2024) in St. Charles, Missouri, integrated agronomic, physiological, UAV-based, and environmental data for model training and validation. Phenotypic analysis showed that seed composition traits varied significantly by year and management practices, with amylose content on a starch (AMLS) and grain weight basis (AMLG) and the control treatments displaying greater stability across seasons. Among ML models, Lasso and ElasticNet achieved the highest predictive accuracy for crude protein (R² = 0.90) and amylose content (AMLS, R² = 0.99; AMLG, R² = 0.92). Bayesian Ridge was most effective for lysine from protein (R² = 0.64), while Partial Least Squares (PLS) excelled in starch content prediction (R² = 0.80). Linear relationships between crude protein, starch content, and traits such as PhiPS2 and nitrogen rate suggested that improvements in photosynthesis or nitrogen uptake directly enhance their accumulation. However, Partial Dependence Plots (PDPs) revealed strong non-linear effects for amylose, where small variations in leaf temperature (Tleaf) and stomatal conductance (gsw) triggered significant shifts in composition. These findings emphasize the importance of precision management strategies to mitigate environmental fluctuations. This study highlights the role of genotype × management interactions in sorghum breeding and demonstrates the value of integrating UAV-based phenotyping with ML-driven predictions to enhance seed quality and precision agriculture strategies.