Effects of harvest timing and technological approaches on volatile compounds and sensory profiles of lower alcohol wines

Over the past 20 years, the level of alcohol in wine has increased in most winemaking regions, raising a number of issues related to consumer health including increased calorie intake and risk from alcohol-related illness and disease, tax policy interventions and wine sensorial quality. The adoption of lower alcohol wines (broadly defined as containing 5.5-11% v/v alcohol) is one solution to addressing these challenges, thus still allowing the enjoyment of wine. However, these wines have not been well received by consumers, in part because the flavour of wine is negatively affected when alcohol is removed, and green-grassy characters may dominate when the wines are produced from grapes harvested at an early stage of ripening. To narrow the sensory gap between lower alcohol and standard alcohol wines and producing more balanced wines with lower alcohol levels, three experiments were planned and conducted in the 2014-15 and 2015-16 growing seasons.
Prior to this investigation, there was little information, if any, in the literature on the chemical and sensory effects of blending finished wines to produce lower alcohol wines.
The initial experiment was conducted in the 2014-15 growing season on Verdelho and Petit Verdot grapes from the Mudgee Region (NSW). To produce finished wines containing less than 11% v/v alcohol, sequential harvests of grapes were undertaken two weeks apart respectively, and equivalent volumes of the resulting wines were blended. For a better understanding of the effect of the practice adopted, the volatile composition of wines was analysed using an established headspace solid-phase microextraction (HS-SPME) method combined to gas chromatography-mass spectrometry (GC-MS). The sensory profile of wines was evaluated using a descriptive analysis. The GC-MS results indicated that blending had an averaging effect on most of the analytes. In addition, it was observed that the blended Verdelho treatment (8.8% v/v) did not significantly differ from the mature fruit wine (10.3% v/v) for the majority of sensory descriptors while no significant differences were found between the blended treatment (11% v/v) and the mature Petit Verdot wine (12.6% v/v). Partial least square regression (PLS2) showed positive correlations between sensory attributes and compositional data. Overall, wine blending allowed the production of lower alcohol wines while exhibiting similar sensory profiles to wines produced from the mature grapes.
A second experiment was conducted in the 2015-16 growing season. The aim of this study was to assess the changes in volatile composition and sensory attributes of Verdelho and Petit Verdot wines produced from early harvest and late harvest grapes and dealcoholisation process. A laboratory scale combined reverse osmosis-evaporative perstraction process was used for dealcoholisation. Early harvest and dealcoholised wines had 9% and 10.5% v/v alcohol for Verdelho and Petit Verdot respectively. GC-MS analyses showed significant differences between early harvest and dealcoholised treatments. In particular, the concentration of ethyl esters was significantly lower in dealcoholised than early harvest wines. This was mainly due to the significant loss of these compounds, which were removed from the original wines during dealcoholisation. For both varieties, the sensory attribute ratings for ‘overall aroma intensity’ and ‘alcohol’ mouthfeel also decreased following dealcoholisation. Despite this, the biggest differences were found between the sensory attributes of early harvest and late harvest wines. Dealcoholisation appeared to be more suitable for red than for white wine, however, dealcoholised wines were distinctively different from both early harvest and late harvest wines even though these wines had similar alcohol level to early harvest wines.
The findings obtained from the first and second experiment were further expanded in the last experiment in which a different grape variety and alcohol ranges were involved. Wines were produced from Shiraz grapes from a single vineyard (Gundagai, NSW) using three methodologies: (i) timing of harvesting; (ii) blending finished wines produced from fruit harvested several weeks apart;
(iii) dealcoholisation (–3% and –6% v/v alcohol).
Two alcohol levels were targeted, 10.5% and 13.5% v/v, both of which are considerably lower concentrations than the commercial wine (up to 17% v/v) produced from this vineyard site. Aside from performing univariate analyses, the ComDim approach was used to identify the analytes that brought about the perceived sensory differences in wines. One of the main results was that dealcoholisation caused a significant loss of several esters and alcohols (different from ethanol), while blending was confirmed to have an averaging effect on many analytes. The sensory analysis of 10.5% v/v Shiraz wines indicated that the perception of a ‘grassy’ attribute was significantly lower in the dealcoholised treatments compared to the early harvest wine, while that of ‘dark fruit’, ‘raisin/prune’, ‘astringency’ and ‘alcohol’ attributes were higher. On the other hand, no significant differences were found among the 13.5% v/v wines, except for ‘alcohol’. Importantly from a sensory perspective, it was observed that ‘alcohol’ was the only sensory attribute significantly different between dealcoholised and original wines (16.3% v/v) produced from overripe grapes. These findings suggested that a 3% v/v dealcoholisation did not negatively affect the sensory quality of finished wines.
The objective of this body of research was to evaluate the changes on volatile composition and sensory attributes in lower alcohol wines produced by grape harvest options, wine blending and dealcoholisation strategies. Insights from this research will provide knowledge that may be applied to control or moderate the dearth of ripe fruit attributes in wines and narrow the sensory gap between lower alcohol and standard alcohol wines by producing more balanced wines with lower alcohol levels.