Predictive Models for Identifying Drought Tolerance Markers in Cotton Varieties

Abstract

Global cotton production is greatly influenced by drought, which can cause yield losses of up to 58 percent during critical growth stages. Improved drought tolerance in cotton varieties is important for agricultural productivity in water scarce areas. Previous studies have investigated physiological, morphological, and biochemical responses to drought but have little integration of these factors into predictive models for variety selection. To fill these gaps this study utilizes a multi-variate approach to assess the drought tolerance indices, namely Stress Tolerance Index (STI), and Mean Productivity (MP) across four cotton varieties in well watered and water deficit conditions. Using partial least squares regression, Pearson’s correlation and multiple linear regression, key physiological (stomata conductance), morphological (root to shoot ratio) and biochemical (proline and malondialdehyde) marker data were analyzed to identify control measures for plant biomass. Results indicated that STI was dependent on Δproline, Δmalondialdehyde and Δboll weight, and MP was strongly determined by Δchlorophyll a/b ratio, Δstomatal conductance and Δroot/shoot ratio. The developed predictive models had robust accuracy in explaining 95% and 89% of the variability of STI and MP, respectively. We identify these results as vital marker genes for drought tolerance and provide a cost effective strategy for early stage phenotyping in cotton breeding programs. The work advances theoretical knowledge as it helps understand the interaction between physiological and biochemical trait within the context of water stress. Future research should expand upon these models to include molecular level analyses and explore combination of abiotic stressors, in order to further increase the robustness and applicability of these models to various agroclimatic settings.