Modeling grape taste and mouthfeel from chemical composition. (1st March 2022)

• Record Type: Journal Article
• Authors: Ferrero-del-Teso, Sara; Suárez, Alejandro; Ferreira, Chelo; Perenzoni, Daniele; Arapitsas, Panagiotis; Mattivi, Fulvio; Ferreira, Vicente; Fernández-Zurbano, Purificación; Sáenz-Navajas, María-Pilar
• Is Part Of: Food chemistry. Volume 371(2022)
• Publication Date: 2022-03-01
• Subjects: Sensory analysis -- Sorting task -- Rate-k-attributes -- Astringency sub-qualities -- Tannin activity; Food -- Analysis -- Periodicals; Food -- Composition -- Periodicals; 664
• DOI: 10.1016/j.foodchem.2021.131168
• Languages: English
• ISSNs: 0308-8146

Abstract

Highlights: Mouthfeel and taste of polyphenolic fractions of grapes were characterized. PLS models predicting sensory properties from chemical composition were obtained. Dryness could be satisfactory modeled from tannin and anthocyanin parameters. Stickiness is well predicted from the level of large polymeric anthocyanins. Bitterness can be predicted from flavonol concentration. Abstract: This research aims at predicting sensory properties generated by the phenolic fraction (PF) of grapes from chemical composition. Thirty-one grape extracts of different grape lots were obtained by maceration of grapes in hydroalcoholic solution; afterward they were submitted to solid phase extraction. The recovered PFs were reconstituted in a wine model. Subsequently the wine models, containing the PFs, were sensory (taste, mouthfeel) and chemically characterized. Significant sensory differences among the 31 PFs were identified. Sensory variables were predicted from chemical parameters by PLS-regression. Tannin activity and concentration along with mean degree of polymerization were found to be good predictors of dryness, while the concentration of large polymeric pigments seems to be involved in the "sticky" percept and flavonols in the "bitter" taste. Four fully validated PLS-models predicting sensory properties from chemical variables were obtained. Two out of the three sensory dimensions could be satisfactorily modeled. These results increase knowledge about grape properties and proposes the measurement of chemical variables to infer grape quality.