The evaluation of Fourier transform infrared spectroscopy (FT-IR) for the determination of total phenolics and total anthocyanins concentrations of grapes
The assessments of grape and wine quality are complex issues and the wine industry needs more objective analysis of grape and wine quality. The standard quality assessment protocol for grading grapes at most wine cellars in South Africa is based on viticultural practices and the determination of chemical parameters such as ºBrix, pH and titratable acidity (TA). Grape juice indices calculated by formulae such as ºBrix/pH, TA/pH, ºBrix/TA, ºBrix x (pH)2 have been used in the past but these approaches have had limited success. It was shown that the total anthocyanins and total phenolics of red grapes correlate with wine quality and provide additional objective measures of grape quality. Most methods for the quantification of total anthocyanins and total phenolics are complex and time consuming and therefore not easily implemented in the routine laboratory environment. Fourier transform infrared spectroscopy (FT-IR) is widely used in South African laboratories for the routine quantification of wine and grape parameters but the commercial calibration models supplied for the quantification of grape total anthocyanins and phenolics are not satisfactory. The focus of this study was to develop new FT-IR calibration models for the quantification of total anthocyanins and phenolics of grapes and to use the generated data during a preliminary evaluation of the implementation of these parameters as part of the grape quality control protocol at a commercial winery in South Africa. The potential of Fourier transform infrared spectroscopy (FT-IR) for the rapid quantification of total anthocyanins and total phenolics in red grapes was investigated and evaluated for prediction accuracy with independent validation sets. The design of calibration sets aimed at capturing most of the variation due to vintage (2004 and 2005), cultivar (Cabernet Sauvignon, Merlot, Pinotage, and Shiraz) and sugar concentration. Best prediction accuracies were obtained for calibration sets using grapes from a single vintage or cultivar or approximately the same sugar concentration. The highest prediction accuracies were obtained for total anthocyanins calibration sets of grapes with sugar concentrations ≥ 23.5ºBrix (SEP = 0.13 mg/g; R2 validation set = 0.77) and for total phenolics calibration sets of grapes with sugar concentrations < 23.5ºBrix (SEP = 0.13 OD280/g; R2 validation set = 0.74). Strong correlations were found between the spectral data and the total anthocyanins (SEP = 0.12 mg/g; R2 validation set = 0.84) and total phenolics concentration data (SEP = 0.10 OD280/g; R2 validation set = 0.76) for 2005 Merlot calibration sets indicating that the FT-IR spectra captured most of the variation. Overall the RPD (ratio of the standard deviation of the reference data to the standard error of prediction) values of all calibration models were below 3 indicating that calibration models are fit for screening purposes. Spectroscopic absorbance at 280 nm is not specific enough for the quantification of total phenolics and the use of an alternative reference method such as high performance liquid chromatography (HPLC) will be considered in the future. Principal component analysis (PCA) revealed that the major sources of variation in the FT-IR spectra of grapes could be ascribed to vintage and grape sugar concentration and this had an effect on the accuracy of the analytical data generated when using FT-IR spectroscopy. This report is the first to our knowledge where FT-IR has been used for the quantification of total anthocyanins and phenolics of grapes. The evaluation of the reference laboratory protocol for the quantification of total anthocyanins and total phenolics in grapes were evaluated in Chapter 4 and emphasized the importance of meticulous laboratory practices to obtain reliable reference data for calibration purposes. This large scale investigation of the total anthocyanins and phenolics concentrations in grapes is the first of its kind in South Africa and a quantitative database containing analytical data of the anthocyanins and total phenolics concentrations of 692 grape samples representing a wide range of grape maturities of Vitis vinifera cultivars Cabernet Sauvignon, Merlot, Pinotage and Shiraz from the 2004 and 2005 vintages was established based on the reference values. The data were used in a preliminary investigation of the implementation of total anthocyanins and total phenolics concentrations as part of grape quality classification at a commercial South African winery (Chapter 5). The results showed that the total anthocyanins and total phenolics concentration in grapes increased with increasing grape maturity (measured as sugar concentration). ANOVA and post-hoc analysis (Bonferroni testing) revealed significant differences between the total anthocyanins and total phenolics concentrations of the four Vitis vinifera cultivars investigated. Grapes harvested earlier in the season had significantly higher (p≤0.05) total anthocyanins and total phenolics concentrations than grapes harvested later in the season. This implies that grapes harvested earlier in the harvest season could produce wines with higher quality. Grapes from regularly irrigated vineyards had lower total anthocyanins and total phenolics concentrations compared to dryland vineyards. The current grape grading system in use at the industrial cellar did not correlate well with the total anthocyanins and total phenolics concentrations of grapes which highlighted the need for the inclusion of more objective measures during grape grading. The information captured in the database can be used as a starting point to establish profiles of the typical anthocyanins and total phenolics of South African grapes and data from more vintages should be included and continually updated. These findings highlight the important contribution of the results obtained in this preliminary study for the incorporation of total anthocyanins and phenolics concentrations as objective parameters of grape quality. Finally multivariate data analysis of the FT-IR spectra revealed important information regarding factors (both physical and chemical) that contribute to the variation of the spectra. The main variation between the 2004 and 2005 samples can probably be interpreted in terms of the water content of the samples.