Stage-specific drought resilience in cotton revealed by integrating machine learning, physiological traits, spectral phenotyping, and ionomic signatures

Cotton is known for its drought tolerance and ability to sustain fiber production under water-limited conditions. However, the specific impact of drought during flowering remains elusive despite its critical influence on cotton physiology, yield, and fiber quality. An experiment was conducted to dissect the flowering stage-specific drought and recovery responses of cotton by imposing 10-day drought treatments at early, mid, and late flowering stages, followed by rewatering and monitoring of recovery. A comprehensive time-series dataset was obtained which encompassed spectral and physiological indices, lint quality metrics, and ionomic profiles, from which we mapped cotton varietal trait trajectories across different flowering phases. The spectral shifts from green and near-infrared dominance toward higher anthocyanin reflectance and senescence-associated bands showed developmental stage-specific patterns. These spectral disruptions mirrored an intense physiological decline in Rubisco activity and net photosynthesis particularly during late-stage stress, with only partial trait recovery post-drought. Based on machine learning classifiers and t-SNE clustering, physiological parameters provided a better homogenous cluster between drought and recovery across stages than spectral imaging. Fiber quality traits such as strength and elongation also showed the lowest resilience scores under late-stage drought. The temporal trait clustering and ternary balance modeling illustrated that recovery dynamics are non-linear and trait dependent. Our results uniquely define a spectral–physiological architecture of cotton drought resilience, offering trait-based targets and predictive frameworks for stage-specific stress adaptation in crops.