Determining levels of steviol glycosides in the leaves of Stevia rebaudiana by near infrared reflectance spectroscopy

The sweet diterpene glycosides extracted from the leaves of Stevia rebaudiana (Bert) Bertoni are of increasing interest in recent years as possible sugar substitutes. Our objectives were to investigate the application of near infrared reflectance spectroscopy (NIRS) in analysis of steviol glycosides (SGs) concentration (w/w) dry basis in S. rebaudiana leaves, using high-pressure liquid chromatography (HPLC) as the reference method. Separate calibration models were developed, using partial least squares regression (PLSR) for stevioside and the total combined concentration for the three main SGs in 33 samples. Using full cross validation the performance of calibration models was assessed in terms of the calibration correlation coefficient ( R c 2 ), root mean squared error of cross-validated prediction (RMSECV), standard deviation ratio (SDR) = (SD/RMSECV), slope and bias of each calibration set. The PLSR model performance was significantly improved by using the total combined concentration for the three main steviol glycosides suggesting this value should be used as the constituent for future development of larger NIR calibration and validation sets.