Abstract
Frost risk is increasingly occurring in winter wheat. Quantitative assessment of frost risk can facilitate the analysis of key genetic factors related to wheat resistance to abiotic stress. We collected 491 wheat accessions and selected four image-based descriptors (BLUE band, RED band, NDVI, and GNDVI) to quantitatively assess their frost risk. Image descriptors can complement the visual estimation of frost risk. Combined with GWAS, a total of 107 quantitative trait loci (QTL) (r2 ranging from 0.75% to 9.48%) were identified, including the well-known frost-resistant locus Frost Resistance (Fr)-1/ Vernalization (Vrn)-1. Additionally, by utilizing published RNA-Seq data, we identified two other frost resistance candidate genes TraesCS2A03G1077800 and TraesCS5B03G1008500. Furthermore, when combined with genome selection (GS), image-based descriptors can predict frost risk with high accuracy (r≤0.84). In conclusion, our research confirms the accuracy of image-based high-throughput acquisition of frost risk, thereby supplementing the exploration of the genetic structure of frost resistance in wheat within complex field environments.