High water productivity phenotype identification in soybean via canopy features

Overview

Drought is the leading cause of soybean yield loss, yet breeding programs typically select for yield alone rather than water productivity (WP), the ratio of yield to evapotranspiration. High-yielding varieties often perform poorly under water-limited conditions, and traditional WP phenotyping methods are slow, costly, and rarely applied at scale. This solution proposes a rapid, image-based approach to identify high WP soybean phenotypes by linking canopy structure features captured through hyperspectral and thermal imaging to drought-tolerant genotypes, even when drought stress is imposed early in the season. The approach enables breeders and agronomists to screen varieties for water-efficient performance, supporting the development of more resilient soybean varieties for dry and variable climates.

Technical specifications

Core approach:

• Phenotyping pipeline that associates canopy spectral and thermal reflectance features with water productivity outcomes
• Greenhouse experiments with five soybean varieties of contrasting genetic backgrounds, yield potential, and drought tolerance
• Two water regimes: ideal evapotranspiration (ETmax, no drought) and 50% of ETmax (drought), with drought imposed early in the season through harvest
• Mini-lysimeters for continuous, individual plant evapotranspiration measurement and automated water replacement
• Daily hyperspectral and thermal imaging at midday to capture canopy reflectance dynamics
• Biometric assessment of crop performance at harvest
• Predictive modeling using decision tree and elastic net algorithms to associate canopy features with higher WP phenotypes

Key features:

• Rapid, non-destructive canopy imaging replaces labor-intensive traditional WP assessment
• Captures morpho-physiological adaptations associated with drought severity and timing
• Applicable across genotypes with different drought tolerance profiles
• Designed to bridge ground-truth drought stress with image-based plant adaptation metrics

Technology readiness level

The approach is at an early-to-mid research stage. Validation experiments are planned in greenhouses at the College of Agriculture, University of Sao Paulo, Brazil, combining controlled drought treatments with continuous lysimeter-based ET monitoring and daily hyperspectral and thermal imaging. Predictive models (decision tree and elastic net) will be trained and tested to associate canopy features with high WP phenotypes. Further field validation across environments and genotypes will be needed before operational deployment in breeding pipelines.