Models based on ultraviolet spectroscopy, polyphenols, oligosaccharides and polysaccharides for prediction of wine astringency

Astringency elicited by tannins is usually assessed by tasting. Alternative methods involving tannin precipitation have been proposed, but they remain time-consuming. Our goal was to propose a faster method and investigate the links between wine composition and astringency. Red wines covering a wide range of astringency intensities, assessed by sensory analysis, were selected. Prediction models based on multiple linear regression (MLR) were built using UV spectrophotometry (190-400 nm) and chemical analysis (enological analysis, polyphenols, oligosaccharides and polysaccharides). Astringency intensity was strongly correlated (R2=0.825) with tannin precipitation by bovine serum albumin (BSA). Wine absorbances at 230 nm (A230) proved more suitable for astringency prediction (R2=0.705) than A280 (R2=0.56) or tannin concentration estimated by phloroglucinolysis (R2=0.59). Three variable models built with A230, oligosaccharides and polysaccharides presented high R2 and low errors of cross-validation. These models confirmed that polysaccharides decrease astringency perception and indicated a positive relationship between oligosaccharides and astringency.