Prediction of anthocyanin content and variety in plum extracts using ATR-FTIR spectroscopy and chemometrics

Abstract

The high anthocyanin content of plums is one major contributor toward their well-known health-benefiting effects. Consequently, rapid methods for measuring anthocyanin content would be greatly beneficial for quality assurance purposes. However, there are no applications of infrared spectroscopy for the prediction of anthocyanin content in solvent extracts from this crop. This study investigated the use of FTIR spectroscopy for the quantification of total anthocyanin content in ethanolic plum extracts. Partial least squares regression (PLS-R) was able to predict the anthocyanin content with acceptable accuracy in a dependent test set (R2pred = 0.84, RMSEP = 13.8 mg/L, RPD = 2.62); however, the use of support vector regression (SVR) gave superior results (R2pred = 0.93, RMSEP = 10.9 mg/L, RPD = 3.32). In addition, the plum variety could be predicted from the FTIR spectra using a support vector machine (SVM) or partial least squares discriminant analysis (PLS-DA) with a high degree of accuracy (97–100 % correct classification rate). This demonstrates the broad-ranging potential of FTIR spectroscopy for the quality assurance of plum samples based on their solvent extracts.