Application of fluorescence spectroscopy with multivariate analysis for authentication of Shiraz wines from different regions

Aim: To investigate the possibility of utilising simultaneous measurements of absorbance-transmittance and fluorescence excitation-emission matrix (A-TEEM) combined with chemometrics, as a robust method that gives rapid results for classification of wines from different regions of South Australia according to their Geographical Indication (GI), and to gain insight into the effect of terroir on inter regional variation.

Methods and Results: Additionally, to obtaining various colour parameters, the A-TEEM technique enables the “fingerprint” of wine samples to be attained in response to the presence of fluorophoric compounds. This is accomplished by recording a three-dimensional excitation-emission matrix (EEM) over multiple excitation and emission wavelengths, which can then be analysed using multivariate statistical modelling to classify wines. Shiraz wine samples (n = 134) from six different GIs of South Australia (Barossa Valley, Clare Valley, Eden Valley, Langhorne Creek, McLaren Vale, and Riverland) were analysed and absorbance spectra, hue, intensity, CIE L*a*b, CIE 1931, and EEMs were recorded for each sample. EEM data were evaluated according to the cross-validation model built with extreme gradient boost discriminant analysis (XGBDA) using score probability to assess the accuracy of classification according to the region of origin. Preliminary results have shown a high prediction ability and the data extracted from A-TEEM could be used to investigate phenolics as potential chemical markers that may provide effective regional discrimination.

Conclusions: 

The molecular fingerprinting capability and sensitivity of EEM in conjunction with multivariate statistical analysis of the fluorescence data using the XGBDA algorithm provided sufficient chemical/spectral information to facilitate accurate classification of Shiraz wines according to the region of origin. A-TEEM coupled with XGBDA modelling appears to be a promising tool for wine authentication according to its geographical origin.

Significance and Impact of the Study: Having tangible evidence that Australian fine wines may be discriminated on the basis of geographical origin, will help to improve the international reputation of Australian wines and increase global competitiveness. Understanding of the important regional chemical parameters would allow grape growers and winemakers to optimise their viticultural and winemaking practices to preserve these characteristics of their terroir. Moreover, verifying the content in the bottle according to the label descriptions with a rapid method, has the potential to verify product provenance and counteract fraud in cases where wine of inferior/questionable quality or contaminated wine is presented as originating from Australia.