Assessment of Peanut (Arachis hypogaea L.) Responses to Heat and Drought Stress Through Physiological, Machine Learning, and Transcriptomic Approaches

TR Number

Date

2026-08-28

Journal Title

Journal ISSN

Volume Title

Publisher

Virginia Tech

Abstract

Heat-exacerbated drought stress is an increasing constraint on peanut (Arachis hypogaea L.) production in the rainfed Virginia–Carolina region, yet the physiological, phenotypic, and molecular consequences of this combined stress remain incompletely characterized. This dissertation integrated growth chamber, multi-year field, high-throughput phenotyping, and transcriptomic approaches across five interconnected studies to address this gap. In growth chamber and field trials, combined heat and drought stress (HS+DS) significantly reduced stomatal conductance and transpiration relative to control or rainfed treatment. Under sustained mid- to late-season HS+DS imposed with rainout shelters across two field seasons, canopy temperature increased by up to 6 °C, accompanied by greater leaf wilting severity and pod yield reductions of 52–73%. Yield, grade, and economic analyses across 12 genotypes showed that HS+DS reduced the percentages of extra-large kernels and total sound mature kernels and eliminated profitability, with losses exceeding $2,000 ha⁻¹ in 2025. However, breeding line 14x009-1-5-1-1 maintained comparatively stable pod yield across both years. Leaf wilting and stomatal conductance were the strongest predictors of yield loss (R² = 0.86–0.89). To scale wilting assessment beyond manual scoring, a machine learning framework combining UAV multispectral imagery and accumulated growing degree days was developed utilizing data across three years and five sites; Extreme Gradient Boosting achieved the highest predictive accuracy (R² = 0.82–0.87), enabling pixel-level mapping of wilting severity. Finally, RNA sequencing of four virginia-type genotypes under field HS+DS identified 1,733 differentially expressed genes, revealing that genotypes with similar physiological responses employed largely distinct transcriptional strategies. Collectively, these findings demonstrate that HS+DS imposes convergent physiological and yield penalties on peanut via distinct molecular pathways, while also establishing scalable phenotyping tools that can support management and breeding strategies for improved resilience to heat and drought in rainfed peanut production.

Description

Keywords

accumulated growing degree days, candidate genes, canopy temperature, extreme gradient boosting, heat and drought, high-throughput phenotyping, kernel grade, leaf wilting, machine learning, multispectral imagery

Citation