An automated root phenotype platform enables nondestructive high-throughput root system architecture dissection in wheat.

This article presents the development and application of an automated, nondestructive, high-throughput root phenotyping platform (Root-HTP) designed to characterize root system architecture (RSA) in wheat (Triticum aestivum L.) throughout all developmental stages. Utilizing deep learning-based image segmentation (D-LinkNet) combined with RootNav software, the platform extracts 47 RSA traits—including 33 novel traits in wheat—and enables large-scale genome-wide association studies (GWAS). Analysis of 155 wheat accessions identified 2,650 significant single nucleotide polymorphisms (SNPs) and 233 quantitative trait loci (QTLs) linked to root traits, with some overlapping yield-related QTLs; notably, the candidate gene TaMYB93 was validated as influencing root tortuosity and grain yield. Furthermore, the study developed a predictive model for wheat yield based on 18 RSA traits and proposed a parsimonious RSA ideotype associated with high yield, highlighting the platform’s potential to accelerate genetic dissection and breeding for improved root traits and crop productivity.