Characterisation and classification of Spanish white wines by multivariate data analysis

White wines from two Spanish regions (Aragón and Navarra) were studied. Wines were produced from grapes of three different varieties : Chardonnay, Garnacha and Macabeo. Oenological parameters (pH, rH, total acidity, ethanol, dry extract, ash, alkalinity of ash and colour), major volatile compounds (acetaldehyde, methyl acetate, ethyl acetate, methanol, 1-propanol, isobutanol and isoamylalcohol), 2,3-butanodiol and glycerol were analysed. Multivariate statistical methods were employed in wine differentiation and classification according to varietal and geographical origin. The multifactor ANOVA was employed for a preliminary study of the contribution of each analytical variable to the characterisation of wines according to region and variety. As a consequence, we excluded those variables that did not show a relevant role in the regional and varietal differentiation from the subsequent statistical treatments : rH, total acidity and methyl acetate. The results of the factor analysis shows a varietal factor (F1) and a regional factor (F2). F1 was correlated with 2,3-butanodiol, 1-propanol, ash, pH and colour. Regional factor was basically correlated with methanol, isobutanol isoamyl alcohol and glycerol. Factor analysis provided a good description of the differences between production zones but grape variety characterisation was more difficult. Application of stepwise discriminant analyses to raw data resulted in satisfactory classification of wines according to variety. Chardonnay wines were correctly classified after the first step. The variables with the greatest discriminating power were 1-propanol and ash content. In contrast, isoamyl alcohols and colour played an important role in Garnacha and Macabeo wines classification. The result of discriminating analysis by production zone confirms the importance of alcoholic compounds to regional classification. The analysis of the standardised coefficients shows that the discriminant function correlates mainly with the parameters 2,3-butanodiol, isoamyl alcohols and ethanol content. Ethanol is the variable with the greatest discriminating power to Aragón wines. The classification of Navarra wines is correlated whit 2,3-butanodiol and isoamyl alcohols.