A Flavoromics Approach To The Characterisation Of Pinot Noir Wine Sensory Properties

Relationships between wine composition and sensory properties are well recognised. The number of volatile and non-volatile compounds detected in wines is vast and continues to grow. Connections with aroma, taste and mouthfeel properties have been discovered for many; however, the complexity of wine composition hinders the discovery of new relationships. Interactions, including synergistic and masking effects, make the nature of the contributions of individual compounds difficult to establish. This complexity also makes the drivers of wine sensory attributes difficult to ascertain. In this project, predictive models were constructed for Pinot noir wine sensory properties using a flavoromics platform optimised for the profiling of volatile and non-volatile wine compounds. These models allowed the untargeted assessment of composition drivers of aromas, tastes and mouthfeel, and Pinot noir wine quality as perceived by expert panellists. A large number of volatile and non-volatile compounds were identified as strong contributors to the predictive abilities of the models, demonstrating that Pinot noir wine sensory properties are driven by the complex interaction of many different compounds. These results were replicated in a study of Merlot wines, and the strong influence of ethanol concentration on wine sensory was reiterated. Standardised winemaking techniques were applied to determine the origin of Pinot noir wine flavours in the vineyard. Relationships were explored between grape and pre-ferment volatile composition, wine volatile profiles and sensory properties. While several flavour markers were identified, it was noted that connections between pre- and post-ferment volatile composition with respect to aroma were complex and difficult to unravel. Flavoromics studies generate large data sets; the ability of modern analytical instruments to measure the composition of wines has developed so that thousands of molecular features can be detected. To aid the construction of parsimonious predictive models, a novel variable selection technique was applied using the volatile profiles of sensory similes used to describe wine orthoand retronasal aromas (e.g. black currant, cinnamon, vegetal). Several compounds detected in both wines and sensory references were found to be strong predictors of wine aromas. Familiarity of the panellists with the references was important for the success of the method, nevertheless these results indicate similar chemistries may be driving the corresponding aromas. Several compounds were identified in this research that may contribute to wine quality, as well as specific aroma and mouthfeel attributes, providing opportunities for manipulations to design wine style in the winery and vineyard.