Machine Learning Integrates Multispectral Phenotyping and Ionic Signatures to Reveal Stage‐Specific Drought Resilience in Cotton

Abstract

Summary statement Cotton ( Gossypium hirsutum ) drought sensitivity depends strongly on flowering stage, but stage‐resolved, non‐destructive detection remains limited. Using controlled short‐term droughts imposed at early, mid, or late flowering, we integrated multispectral and hyperspectral canopy phenotyping with physiology and explainable machine learning to identify spectral predictors of metabolic status and recovery. Early and mid‐flowering drought responses were largely recoverable, whereas late‐flowering drought caused the most potent and least reversible losses in photosynthesis, canopy structure, and fiber quality. These results highlight late flowering as a critical vulnerability window and provide a mechanistically grounded framework for rapid phenotyping of stage‐specific drought resilience.

Plant phenotyping relevance

マルチスペクトル・ハイパースペクトルによる非破壊キャノピー表現型計測と説明可能な機械学習を中核に、乾燥耐性を迅速推定する方法・枠組みを提示している。

Code and data availability

The paper's phenotyping data (PlantEye multispectral, Resonon hyperspectral, physiological measurements) and analysis code are not publicly available; the Data Availability Statement says data are available only on request from the corresponding author. No public URLs or repositories are provided, and no qualifying URL

No evidence-backed public reproduction asset is currently recorded.