Phenotyping the hidden half: Combining UAV phenotyping and machine learning to predict barley root traits in the field

Abstract

Improving crop root systems for enhanced adaptation and productivity remains challenging due to limitations in scalable non-destructive phenotyping approaches, inconsistent translation of root phenotypes from controlled environment to the field, and a lack of understanding of the genetic controls. This study serves as a proof of concept, evaluating a panel of Australian barley breeding lines and cultivars (Hordeum vulgare L) in two field experiments. Integrated ground-based root and shoot phenotyping was performed at key growth stages. UAV-captured vegetation indices (VIs) were explored for their potential to predict root distribution and above-ground biomass. Machine learning models, trained on a subset of 20 diverse lines, with the most accurate model applied to predict traits across a broader panel of 395 lines. Unlike previous studies focusing on above-ground traits or indirect proxies, this research directly predicts root traits in field conditions using VIs, machine learning and root phenotyping. Root trait predictions for the broader panel enabled genomic analysis using a haplotype-based approach, identifying key genetic drivers, including EGT1 and EGT2 which regulate root gravitropism. This approach offers the potential to advance root research across various crops and integrate root traits into breeding programs, fostering the development of varieties adapted to future environments. HighlightIntegrating UAV phenotyping and machine learning can be used to predict RSA traits non-destructively and offers a new approach to support root research and crop improvement.