Establishment of a predictive model to identify suitable soybean genotypes for oil processing

Abstract Various characteristics of soybeans play important roles in soybean oil quality. However, the relationship between soybean chemical traits and soybean oil processing quality characteristics is unclear. In this study, the chemical traits of soybean and ten characteristic parameters of soybean oil from 58 genotypes were determined. An evaluation model for the overall soybean oil quality was established according to ten parameters, based on which soybean oil of 58 genotypes was classified into three quality grades by cluster analysis. A prediction model for the quality of soybean oil based on the chemical traits of soybeans was constructed by stepwise regression. The relative errors between the actual values and the predicted values of the validation genotypes were less than 10%. The predicted quality for 11 of the 14 tested soybean varieties corresponded to the actual oil quality, suggesting that the model is reliable. Our study provides a method for selecting ideal soybean genotypes for producing high-quality soybean oil.