Comparative Analysis of High‐Throughput Phenotyping Techniques for Early Detection of Herbicide Stress

Abstract

ABSTRACT Early detection of herbicide‐induced stress is essential for optimising weed control, improving crop safety and advancing precision agriculture practices. While non‐destructive imaging technologies offer great potential for rapid stress diagnosis, direct comparisons of their performance across species, herbicides and application rates remain scarce. This study systematically compared three high‐throughput imaging methods: chlorophyll fluorescence imaging, multispectral imaging and 3D multispectral scanning, for their ability to detect and classify early physiological and morphological responses to bentazone (HRAC 6) and glyphosate (HRAC 9) in a weed (velvetleaf, Abutilon theophrasti ) and a crop species (common bean, Phaseolus vulgaris ). Plants were imaged daily, immediately before treatment and for five consecutive days after herbicide application. Chlorophyll fluorescence parameters, particularly NPQ, q P and F s ′, emerged as the earliest and most sensitive indicators, detecting stress within 24 h of treatment. Stepwise discriminant analysis revealed that intermediate time points (48–72 h after application) achieved the highest classification accuracies, reaching up to 100%, with chlorophyll fluorescence as the dominant trait for selection. Multispectral traits, such as reflection in red and far red, and saturation, together with morphological traits like total leaf area and leaf inclination, provided complementary information and enhanced discrimination accuracy among herbicides. These findings highlight the superiority of chlorophyll fluorescence imaging for detecting early herbicide stress, while demonstrating the added value of integrating spectral and morphological traits. The results support the development of multi‐sensor phenotyping pipelines for rapid, accurate and field‐adaptable diagnostics in both weed and crop management.

Plant phenotyping relevance

複数の高スループット画像計測法を比較・評価し、植物の生理・形態応答を用いた早期ストレス検出性能を検証しているため、フェノタイピング手法が中心です。

Code and data availability

公開論文であることは確認できましたが、現在の公式API・許可済み取得経路では本文を自動取得できませんでした。

No evidence-backed public reproduction asset is currently recorded.