Abstract: A predictive model for rapid quantification of moisture and reducing sugar in fresh sweet potato was established by near-infrared (NIR) spectroscopy, which could facilitate the quality analysis and germplasm screening of sweet potato. In this study, 146 samples of different lines of sweet potato were selected, out of which 109 samples were used as calibration samples while the rest were used as validation samples. After different spectral pretreatments and optimal wavelength selection by synergy interval partial least squares, a principal component regression model and a partial least squares model were developed for predicting the contents of moisture and reducing sugar in sweet potato, respectively. The results showed that the coefficient of determination, root mean square error of prediction, and standard deviation ratio of the optimal model for moisture and reducing sugar contents were 0.974 and 0.885, 1.154 and 0.270, and 6.334 and 3.148, respectively, indicating that the two models have good prediction performance, and the NIR model predicted values had a good correlation with the corresponding chemical values. Therefore, the predictive model is suitable for rapid quantification of water and reducing sugar in large-scale sweet potato breeding.

Key words: near-infrared spectroscopy, sweet potato, moisture, reducing sugar, synergy interval partial least squares