Multi-sensor phenotyping of yield and yield stability for genotype selection in durum wheat

Area of Science:

• Agricultural Science
• Plant Breeding
• Climate Resilience

Background:

• Climate change necessitates wheat varieties with high yield and stability across diverse environments.
• Mediterranean agriculture faces challenges from variable precipitation and rising temperatures.

Purpose of the Study:

• To develop a classification framework for selecting climate-resilient durum wheat genotypes.
• To evaluate high-throughput phenotyping for predicting yield and stability.

Main Methods:

• Assessed 64 durum wheat cultivars under irrigated and rainfed conditions.
• Utilized ground-sensor and UAV-derived vegetation indices (VIs) and thermal-infrared (TIR) data for phenotyping.
• Employed Random Forests and sequential feature selection for yield and stability prediction.

Main Results:

• Developed a robust yield prediction model (R² > 0.74) integrating environmental covariates.
• Achieved moderate but improved stability predictions (R² up to 0.56) using VI stability.
• Identified early vigor and earlier senescence as key traits for desirable genotypes, contrasting with 'stay-green' behavior in undesirable ones.

Conclusions:

• Early vigor and senescence timing are critical for selecting climate-resilient wheat.
• High-throughput phenotyping offers a cost-effective method for joint yield-stability evaluation.
• This approach shifts breeding focus towards sustainable agriculture and climate adaptation.