Cowpea heat and drought stress phenotyping platform using handheld sensors and machine learning

Overview

Cowpea is a fast-growing, highly nutritious legume that is sensitive to heat and drought, two abiotic stresses that threaten crop productivity worldwide. This project develops a phenotyping and predictive modeling platform that links early-stage photosynthetic responses of cowpea varieties to final biomass and grain yield under heat stress. By combining handheld sensor measurements with machine-learning models, the platform aims to enable rapid identification of heat-tolerant and heat-sensitive cowpea genotypes, supporting breeding decisions and climate-resilient crop development.

Technical specifications

Phenotyping approach:

• Uses the MultispeQ handheld device connected to the open-source PhotosynQ platform for measuring multiple photosynthetic parameters in the field or growth chamber
• Screens cowpea varieties at the 9-leaf stage under controlled heat stress conditions (45–50°C) over single-day, 3-day, or intermittent durations
• Captures parameters such as maximal photosystem II quantum efficiency at defined intervals during and after stress

Predictive modeling:

• Applies machine-learning algorithms including Linear Regression and Autoregressive Integrated Moving Average (ARIMA) to correlate photosynthetic capacity and growth dynamics with final dry weight and grain yield
• Builds on prior work using deep-learning models to classify organelle morphology abnormalities from confocal microscopy images of Arabidopsis with over 97% accuracy
• Designed to scale to larger field datasets from farmers for prediction across additional crops

Scope of screening:

• Initial validation covered 20 cowpea varieties (5 plants per genotype) under a 3-day heat stress protocol
• Future validation focuses on 8 selected varieties representing tolerant and sensitive responses, tracked through a 2-month life cycle

Technology readiness level

The platform is at an early-to-mid research and validation stage. Proof-of-concept data has been collected on 20 cowpea varieties showing correlations between photosynthetic efficiency decline under heat stress and grain yield loss. A related machine-learning framework has demonstrated high accuracy in classifying organelle morphology from microscopy images. Ongoing and future validation will refine stress protocols, expand predictive modeling, and explore scaling to field-scale, multi-crop applications. The technology is not yet commercially deployed but is positioned for collaborative development with breeding programs, agricultural technology companies, and research partners.