Application of non-supervised pattern recognition techniques to classify Cabernet Sauvignon wines from the Balkan region based on individual phenolic compounds

Abstract

Phenolic compounds in sixteen Cabernet Sauvignon wines from different wine-growing sub-regions in the Balkan region were investigated using HPLC with DAD and fluorescence detector and spectroscopic analysis, as well as statistical PC/F and cluster analysis. The HPLC analysis of investigated red wines showed that the content of total hydroxybenzoic acids, detected at 280 nm, was the highest in wines from Tikves wine-growing subregion, Macedonia (127-140 mg L-1). Total hydroxycinnamic acids, detected at 320 nm, were the highest in wines from Zupa wine-growing subregion, Serbia (43-45 mg L-1). The concentration of total flavonoids (flavan-3-ols, flavonols, flavons and flavanon), detected at 280, 360 and 322/275 nm, respectively, was the highest in wine from Katarzyna Estate wine-growing subregion, Bulgaria (167 mg L-1). Finally, the concentration of total anthocyanins, detected at 520 nm, was the highest in wine from Sumadija wine-growing subregion, Serbia (1463 mg L-1). The results of PCA and cluster analysis together confirmed that the content of phenolic compounds in Cabernet Sauvignon wines depends on agro-climatic factors, oenological practice in different wineries and the growing season in the Balkan region that were investigated. The areas in the Balkan region in this study with similar agro-climatic characteristics showed shorter clustering distance, indicating similar phenol profiling in the red wines tested. (C) 2016 Elsevier Inc. All rights reserved.