Evaluating drought tolerance stability in soybean by the response of irrigation change captured from time-series multispectral data

This study investigated a method to evaluate the drought tolerance stability of a genotype in a single environmental trial by capturing the plant response to irrigation changes. Genotypes that exhibit stable phenotypes under various drought stress conditions are required for stable crop production. However, considerable time and money are required to evaluate the environmental stability of a genotype through multiple environmental trials. As an index of drought tolerance stability, we calculated the coefficient of variation (CV) of shoot fresh weight of 178 soybean ( Glycine max (L.) Merr.) accessions in a total of nine types of drought treatments, including changing irrigation treatments (every five or ten days) over 3-year trials. To capture the plant responses to irrigation changes, time-series multispectral (MS) data were collected, including the timings of the irrigation/non-irrigation switch in the changing irrigation treatments. We built a random regression model (RRM) for each of the nine treatments using the time-series MS data. We built a genomic prediction model (MTRRM model) using the genetic random regression coefficients of RRM as secondary traits and evaluated the accuracy of each model for predicting CV. In two out of the three years, the prediction accuracy of MTRRM models built in the changing irrigation treatment was higher than that in the continuous drought treatment in the same year. When the CV was predicted using the MTRRM model across years in the changing irrigation treatment, the prediction accuracy was 61% higher than that of the simple genomic prediction model. These results suggest that drought tolerance stability can be evaluated in a single environmental trial, which may reduce the time and cost of selecting genotypes with high drought tolerance stability. ### Competing Interest Statement The authors have declared no competing interest. * CV : Coefficient of variation RRM : Random regression model HTP : High throughput phenotyping HS : Hyperspectral MS : Multispectral NDVI : Normalized difference vegetation index NDRE : Normalized difference red-edge VIs : Vegetation indices GWAS : Genome wide association study MTM : Multi-trait model UAV : Unmanned aerial vehicle P4M : Phantom 4 Multispectral SNP : Single-nucleotide polymorphism MVN : Multivariate normal AIC : Akaike’s information criterion