Feasibility Study on the Use of Visible and Near-Infrared Spectroscopy Together with Chemometrics To Discriminate between Commercial White Wines of Different Varietal Origins DANIEL COZZOLINO,* ,² HEATHER EUNICE SMYTH, ²,§ AND MARK GISHEN ² The Australian Wine Research Institute, Waite Road, Urrbrae, P.O. Box 197, Glen Osmond, SA 5064, Australia; The Cooperative Research Centre for Viticulture, P.O. Box 154, Glen Osmond, SA 5064, Australia; and School of Agriculture and Wine, The University of Adelaide, Glen Osmond, SA 5064, Australia The use of visible (vis) and near-infrared spectroscopy (NIR) was explored as a tool to discriminate between samples of Australian commercial white wines of different varietal origins (Chardonnay and Riesling). Discriminant models were developed using principal component analysis (PCA), principal component regression (PCR), and discriminant partial least-squares (DPLS) regression. The samples were randomly split into two sets, one used as a calibration set (n ) 136) and the remaining samples as a validation set (n ) 133). When used to predict the variety of the validation set samples, the DPLS models correctly classified 100% of Riesling and up to 96% of Chardonnay wines. These results showed that vis-NIR might be a suitable and alternative technology that can be easily implemented by the wine industry to discriminate Riesling and Chardonnay commercial wine varieties. However, the relatively limited number of samples and varieties involved in the present work suggests caution in extending the potential of such a technique to other wine varieties. KEYWORDS: Visible; near-infrared spectroscopy; white wine; classification; discrimination; PCA; PCR; DPLS INTRODUCTION Determination of food authenticity is one of the most important issues in food quality control and safety. The authenticity of wine is regulated by strict guidelines laid down by the responsible national authorities, which may include official sensory evaluation, chemical analysis, and examination of the records kept by the wine producer (1). Wine identification, or classification, mainly in terms of variety and geographical region of origin, has received increasing attention during the past 10 years using multivariate statistical techniques (1-3). Recently, the use of multivariate statistical techniques on chemical and sensory data has gained increasing attention as a means to classify wines from different geographical regions and to describe similar and discriminating sensory and chemical characteristics (1). Group classification of wine authenticity has been attempted using several different types of compositional data including volatile compounds (4), aroma components (5), and minerals and trace elements (6), as well as phenolic compounds and amino acids (1). All of these methods require sophisticated and expensive analytical equipment such as high-performance liquid chromatography (HPLC), mass spectrometry (MS), gas-liquid chromatography (GLC), and atomic absorption spectroscopy (AAS). Near-infrared spectroscopy (NIR) was originally developed to provide a rapid measurement of the composition of grains and oilseeds (7). NIR has emerged in the past 30 years as a method to predict the quality of different foods and agricultural products due to the speed of analysis, minimum sample preparation, and low cost (7-9). Most of the established methods have involved the development of NIR calibrations for the quantitative prediction of composition in food (8). This was a rational strategy to pursue during the initial stages of its application, given the type of equipment available, the state of development of the emerging discipline of chemomet- rics, and the overwhelming commercial interest in solving such problems (10-12). One advantage of NIR is that it can record the response of the molecular bonds of its chemical constituents to the near-infrared spectrum (e.g., O-H, N-H, and C-H bonds) and thereby build a characteristic spectrum that behaves as a “fingerprint” of the sample (10, 13). This opens the possibility of using spectra to determine complex attributes of foods including organoleptic scores or even sensory character- istic (10). In addition, the application of multivariate statistical techniques such as principal component analysis (PCA) or discriminant analysis provides the possibility to use and * Corresponding author. Email: Daniel.Cozzolino@awri.com.au. Tele- phone: +61 8 8303 6600; Fax: + 61 8 8303 6601. ² The Australian Wine Research Institute and The Cooperative Research Centre for Viticulture. § The University of Adelaide. J. Agric. Food Chem. 2003, 51, 7703-7708 7703 10.1021/jf034959s CCC: $25.00 © 2003 American Chemical Society Published on Web 11/12/2003