Prediction of harvest-related traits in barley using high-throughput phenotyping data and machine learning

Abstract

Abstract Developing crop varieties that maintain productivity under drought is essential for future food security. Here, we investigated the potential of time-resolved high-throughput phenotyping to predict harvest-related traits and identify drought-stressed plants. Six barley lines ( Hordeum vulgare ) were grown in a greenhouse environment with well-watered and drought treatments, and phenotyped using RGB, thermal infrared, chlorophyll fluorescence and hyperspectral imaging sensors. Temporal phenomic classification model accurately distinguished between drought-treated and control plants, achieving high accuracy (R 2 ≥ 0.97) even when exclusively using predictors only from the early phase after drought induction. Canopy temperature depression at the early stage and RGB-derived plant size estimates at the late stage were identified as key classification features. Temporal phenomic prediction model of harvest-related traits achieved particularly high mean R 2 values for total biomass dry weight (0.97) and total spike weight (0.93), with RGB plant size estimators emerging as important predictors. Prediction accuracy for these traits remained high (R 2 ≥ 0.84) when using only predictors from the first half of the experiment. Models trained on pooled drought and control data outperformed single-treatment models and retained high accuracy when applied across treatments. These findings support the integration of high-throughput phenotyping and temporal modelling to enable timely and more cost-effective selection of drought-resilient genotypes, and illustrate the broader potential of phenomics-driven approaches in accelerating crop improvement under stress-prone conditions.

Plant phenotyping relevance

RGB・熱赤外・蛍光・ハイパースペクトルによる高スループット表現型取得と、時系列モデルによる干ばつ状態および収穫形質の予測が研究の中心である。

Code and data availability

The paper's phenotyping data and analysis are said to be available only in supplementary Files S1-S14 of the preprint itself; no public repository, author-hosted URL, or code deposit is provided in the supplied blocks. No paper-specific public asset with an actionable URL qualifies.

No evidence-backed public reproduction asset is currently recorded.