BACKGROUND

Crop yields (along with biotic and abiotic stress tolerance) have been a consistent primary target in plant breeding. Non-destructive sensing can be used to assay priority traits that breeders seek to improve, and/or intermediate traits that provide insight into the accumulation and retention of yield and quality throughout the season. Simultaneous sensing and crop modeling capabilities are not yet deployed in leafy greens or legume breeding programs, particularly for leaf biochemical and physiological traits. However, with advances in hyperspectral imaging, we can now predict a variety of traits related to plant physiology, nutrient uptake, water/carbon use, and canopy architecture. Leveraging AI methods in the development of these capabilities can improve our quantification of priority and intermediate traits as well as improve the predictive ability of yield-and-quality crop models that are built from remote sensing data. Further, the integration of sensing, modeling, and genomic prediction methods will enable us to account for and predict main and interaction effects of genotype, environment, and management.

GOALS

• Non-destructive and simultaneous sensing of leaf biochemical and physiological traits at field and population scales
• Develop mechanistic and AI-enabled models of crop growth and development in leafy greens
• Develop mechanistic and AI-enabled models of broadscale crop structural properties in legumes with a focus on simulation and prediction of yield and quality

IMPACT

• Dissecting relationships between canopy development, yield, and quality of both leafy greens and legumes will enable non-destructive scoring of additional priority traits for use in selection and potentially enhance prediction of crop yield and quality