Closing the phenotyping gap: non-invasive electromagnetic induction enables high-throughput screening of crop root function in the field

Crop root systems play a critical role in water uptake and yield improvement under drought. However, a lack of methods to measure root system function at scale, particularly in the field, poses significant challenges to explore them in crop improvement programs. To address this bottleneck, we introduce a novel approach to directly screen water uptake and root activity in the field through integrating electromagnetic induction (EMI) sensing, inversion modelling and machine learning. This study was conducted as an on-farm field experiment comprising a factorial combination of 13 sorghum genotypes and three row spacings with three replications. Time-lapse EMI surveys were conducted from flowering to maturity using a customised DualEM-21S sensing system. Artificial neural network models were developed to predict 3D soil moisture (θ) using depth-specific true soil electrical conductivity estimated from EMI inversion algorithms, and the prediction accuracy was evaluated. Crop water use and functional root traits were derived from changes in θ profiles between successive EMI surveys using cubic splines fitted within a linear mixed model framework. Results indicate that dynamics of θ can be predicted with high accuracy (R2 = 0.83). There was a significant positive correlation between the EMI-derived crop water use and root length density at the depths without water stress. The functional root traits were significantly affected by growth stages, genotypes, row spacings, and their interactions. In general, narrower row spacing resulted in deeper rooting depth by flowering, which was associated with a larger canopy and higher grain number. In contrast, wider row spacing promoted greater post-flowering root growth, leading to increased post-flowering water use and, consequently, larger grain weight. Overall, our approach enables rapid characterisation of variation in crop water use, rooting depth and root growth in the field. It provides a valuable framework for understanding the dynamic interactions between root function and yield traits across variable field environments, with potential to accelerate genotype selection and optimise crop design in breeding and agronomy programs.