Prediction of Oleic Acid Content in Brassica napus L. Seeds Based on Hyperspectral Parameters at Seedling Stage: A New Method for Rapidly Screening Germplasm with Different Oleic Acid Content at Early Growth Stage of Rapeseed

Abstract Background: Remote sensing prediction techniques are new methods to assist crop production and screen breeding materials, mainly to predict crop conditions. In rapeseed production and breeding, remote sensing can predict growth status and harvest quality through spectral reflection at seedling stage. Results: We bulit a spectral early-prediction system (SEP) containing multiple feedback formulas and predicted the oleic acid content of rapeseeds at harvest stage through reverse validation. We established a recognition model for early prediction of oleic acid in range of view of 56-85% with a resolution of 1%. The Root Mean Square Error (RMSE) between prediction result and ideal model below 1 was found for verification measurements. Conclusion: Besides other practical features such as nondestructive collection of material information, simple operation, and the ability to accurate prediction of oleic acid per plant, the SEP system is a promising tool for rapid screening of different oleic acid breeding materials, offering an easy setup to process in breeding work or field production.