Relationship between astringency and phenolic composition of commercial Uruguayan Tannat wines: Application of boosted regression trees.

Item request has been placed! ×
Item request cannot be made. ×
loading   Processing Request
Read More Add to Saved list
  • Additional Information
    • Source:
      Publisher: Published on behalf of the Canadian Institute of Food Science and Technology by Elsevier Applied Science Country of Publication: Canada NLM ID: 9210143 Publication Model: Print-Electronic Cited Medium: Internet ISSN: 1873-7145 (Electronic) Linking ISSN: 09639969 NLM ISO Abbreviation: Food Res Int Subsets: MEDLINE
    • Publication Information:
      Original Publication: Ottawa, Ontario, Canada : Published on behalf of the Canadian Institute of Food Science and Technology by Elsevier Applied Science, c1992-
    • Subject Terms:
      Regression Analysis* ; Taste*; Fruit/*metabolism ; Phenols/*metabolism ; Vitis/*metabolism ; Wine/*analysis; Adult ; Algorithms ; Chromatography, High Pressure Liquid ; Female ; Food Handling/methods ; Fruit/growth & development ; Humans ; Hydrogen-Ion Concentration ; Judgment ; Male ; Middle Aged ; Tandem Mass Spectrometry ; Taste Perception ; Uruguay ; Vitis/growth & development
    • Abstract:
      Phenolic compounds play a major role in the intensity and characteristics of wine astringency. However, studies involving commercial wine samples are still scarce. The aim of the present work was to study the relationship between astringency and phenolic composition of commercial Uruguayan Tannat wines using boosted regression trees (BRT), a novel predictive method. Forty commercial Tannat wines were evaluated by a trained sensory panel (9 members), who assessed their total astringency intensity using time-intensity (TI) and described their astringency sub-qualities using a check-all-that-apply (CATA) question composed of sixteen terms. The polyphenolic profiles of the wines were determined by HPLC-MS and conventional oenological parameters were also obtained. Fifty BRT models with different partitions of the data in training and test sets were built for astringency maximum intensity (Imax) and for the frequency of use of the 16 astringency sub-qualities considered in the CATA question. As predictor variables, 84 phenolic compounds and oenological parameters were considered for all BRT models. Both strong and weak predictive models were obtained for each response variable. Predictive accuracy was much higher for astringency intensity than for the frequency of mention of astringency sub-qualities. Still, the BRT models allowed to point out to some compositional variables most likely involved in wine astringency perception. Total tannin concentration (chemically determined) was the most relevant explanatory variable for sensory astringency, while flavan-3-ols were the individual phenolic compounds with the highest contribution to astringency, particularly some dimers, trimers and the sum of non-galloylated tetramers. However, the effect of these predictors differed according to the astringency sub-quality considered as response. As expected, non-linear relationships between phenolic compounds and astringency were found. These results contribute to the understanding of the influence of phenolic composition on wine astringency and stress the potential of BRT models for identifying the compounds responsible for this complex sensory characteristic.
      (Copyright © 2018 Elsevier Ltd. All rights reserved.)
    • Contributed Indexing:
      Keywords: Astringency; CATA questions; Phenolic profile; Tannat; Time intensity; Wine
    • Accession Number:
      0 (Phenols)
    • Publication Date:
      Date Created: 20180823 Date Completed: 20191017 Latest Revision: 20191017
    • Publication Date:
      20221213
    • Accession Number:
      10.1016/j.foodres.2018.06.024
    • Accession Number:
      30131135