A predictive framework for identifying drought-tolerant sugarcane: discriminant modeling under mannitol-induced stress

Abstract

[EN] Drought stress hinders plant development by limiting water availability, disrupting osmotic equilibrium, and impairing nutrient uptake and photosynthetic efficiency. To counter these effects, temporary immersion bioreactors (TIBs) provide a scalable platform for rapid shoot proliferation and in vitro screening of sugarcane genotypes subjected to mannitol-induced osmotic stress. These controlled responses often parallel field performance, enabling efficient early-stage selection. In this study, apical meristems of sugarcane (cultivar C-1051-73) were cultured in vitro, transferred to TIBs, and exposed to mannitol concentrations ranging from 0 to 200 mM. Key physiological parameters and biochemical stress markers were quantified. Discriminant analysis, implemented via Python-based machine learning, classified stress responses using Fisher's linear function. Control (0 mM) and high-stress (200 mM) treatments served as training sets, while intermediate concentrations (50-150 mM) were used for model validation. The analysis identified 150 mM as the threshold for stress manifestation, with lower concentrations classified as non-stressed. This analytical framework strengthens genetic improvement strategies-including selection, hybridization, mutagenesis, and transgenesis-by enabling early identification of drought-tolerant genotypes prior to field deployment. The approach reduces costs, shortens breeding cycles, and enhances decision-making in cultivar development. Moreover, the methodology is transferable to other crops, provided that discriminant functions are recalibrated to accommodate species-specific physiological profiles. Altogether, the integration of TIBs with discriminant analysis offers a robust, cost-efficient solution for drought tolerance screening, advancing precision agriculture and promoting crop resilience under water-limited conditions.