The seed of wine

The delectable sensory qualities of wine all begin with the grape seed. This is because grape seeds contain many of the phenolic compounds, such as flavonoids and proanthocyanidins, that determine the subsequent taste of the wine. Know the seed and you’ll know a lot about how the wine will turn out.

At the moment, knowing the seed requires analysing it with chromatography-based techniques, which are fairly time consuming as they require the phenolic compounds to be physically extracted from the seed. But now chemists at the University of Turin in Italy have shown that NIR spectroscopy can make an acceptable and much faster alternative.

Taking seeds from red Nebbiolo grapes harvested from vines grown in several growing regions in Northern Italy in 2010 and 2011, the chemists first determined the concentration of various phenolic compounds in the seeds using high performance liquid chromatography (HPLC). Next, they analysed the seeds with NIR spectroscopy and used partial least squares (PLS) to compare the NIR data with the phenolic composition determined by HPLC.

As they reveal in the Journal of Agriculture and Food Chemistry, the resultant PLS models could predict the phenolic composition from NIR data, including the concentration of flavonoids, proanthocyanidins, catechin and galloylation, but only for a limited selection of seeds. So models generated for specific years or growing regions would only be accurate for seeds from those years and growing regions. Similarly, the chemists also showed that these models couldn’t be applied to seeds from different grape varieties. However, they did find that combining the NIR data for both years improved matters, producing more robust models.

Despite the restricted scope of the models, the chemists assert that they could still provide useful information for vineyards, especially when making decisions about harvesting and processing grapes, and managing phenol extraction during the winemaking process. Furthermore, the chemists are now looking to increase the robustness and accuracy of these NIR models by considering seeds from additional years, growing regions and varieties.

Blog tags: