Over the centuries, winemaking has followed an empirical approach, with the “know-how” being passed down from generation to generation. Due to the romanticism associated with wine and the belief that wine is an expression of place (hence the French term “terroir”), there has traditionally been a minimalistic approach to winemaking.

However, as wine has become an increasingly important commodity worldwide – with 24.67 billion liters produced globally in 2017 (Decanter News, 2017) – the development of innovative technologies is becoming a crucial factor in determining the success of the industry as a whole (Pretorius & Høj, 2005). Due to the deep-rooted and rich history of wine, compared to other industries, the acceptance of technological advances and innovation has been rather slow (Bisson et al., 2002; Pretorius & Høj, 2005; Dambergs et al., 2015).


The consistency that is expected from a particular wine producer refers to a certain level of quality rather than a consistency in flavor. Thus, innovation in the wine industry is shifting towards optimizing quality (Pretorius & Høj, 2005). What constitutes a “quality wine” has become, however, an increasingly controversial subject of debate (Pettigrew & Charters, 2006; Johnson & Bruwer, 2007; Lockshin & Corsi, 2012). Regardless, as the grape juice provides the nutrients required by the yeast during fermentation, it has been identified as the primary determinant of the quality of the final wine (Fleet, 2003).

As an example, the concentration and composition of Yeast Assimilable Nitrogen (or YAN) in grape must have frequently been reported to influence the quality of the grape, and consequently, of the final product (Mendes-Ferreira, Barbosa, Lage, & Mendes-Faia, 2011). During fermentation, yeast is only able to make use of a certain portion of the nitrogen contained by the grape berry and this is commonly referred to as Yeast Assimilable Nitrogen (YAN). The most significant impact that YAN has on wine flavor and aroma is by providing amino acid substrates for the Ehrlich pathway (Hazelwood et al., 2008), which results in the formation of higher alcohols, and through subsequent reactions, various esters, and volatile acids (Styger et al., 2011). Additionally, nitrogen deficiency has been highlighted as the primary cause for stuck/sluggish fermentations (Bisson, 1999).

It is no surprise that the role of YAN in the fermentation of grape juice to wine has been an area of research that has received increasing attention in the past three decades. Most of the work, however, has taken a descriptive format, presenting the state of the nitrogen content of different cultivars, vintages, and geographical origins in terms of average, maximum, minimum, and median values. Some attempts at data mining were met with success but were limited in scope.


The work presented in Petrovic et al (2019) began as an unsupervised survey of the YAN status in the main wine producing area of South Africa. In addition to the descriptive aspects linked to the results, the large number of samples included allowed for a more in-depth look at the data. Given the unsupervised nature of the survey, care had to be taken in the exploratory part of the work. Thus, the article focused on the statistical methods that would be appropriate to elucidate the roles of the cultivar and geographical origin of the grapes in determining the concentration and composition of YAN.

A correspondence analysis (CA) was used to identify which cultivars associated with specific levels of YAN. “Very low,” “low,” “high,” and “very high” levels were designated by conducting a cluster analysis on the cases. Viognier samples were associated with “very high” levels of YAN and Cabernet Franc, Merlot, Cabernet Sauvignon, Roussanne, Chenin Blanc, Sémillon, and Cinsaut with “very low” levels of YAN. Pinotage, the South African red cultivar, also associated with “very high” levels of YAN. Chardonnay, Grenache Blanc, and Sauvignon Blanc grouped together with “high” levels of YAN. As such, the correspondence analysis was able to give a concise overview of the structure of the observations and variables in the dataset, substantiating the trends found thus far.

A Classification and Regression Tree (CART) analysis was also performed on the data. This technique is useful in identifying the most important variables, in terms of explanatory power, in a particular dataset (Barlin et al., 2013). In most cases, cultivar was found to play a larger role than origin. Previous studies, investigating the effect of the environment on grapevine gene expression patterns, found that genes involved in amino acid metabolism were less affected by the changing environmental conditions (Santo et al., 2013). This is in support of the hypothesis that cultivar plays an overriding role in determining the concentration and composition of YAN and that growing environment plays a subordinate role in the modulation of YAN.


The results of the CART analysis also allowed the results of both of the tested variables, cultivar and origin, to be viewed simultaneously. There were some districts found to repeatedly be associated with either higher or lower levels of YAN, even though not always statistically significant. Furthermore, certain cultivars were associated with either very high or very low concentrations of YAN, regardless of the district.

This study showed that YAN is indeed an inherent trait of a cultivar and although exact YAN concentrations cannot be predicted purely based on the cultivar, a certain level of discernment is possible. The results also showed which cultivars are more likely to require nitrogen additions to avoid the occurrence of stuck fermentations and those which could potentially run the risk of having excess nitrogen at the end of fermentation.

Despite the unsupervised nature of the survey, through careful statistical exploration, the information extracted from the data could be increased and, as such, the value of the work. In addition to informing the South African wine industry of the grape nitrogen status, this study can be seen as hypothesis-forming. In other words, it can constitute the basis for further experiments, specifically designed to answer specific questions related to factors such as the role climate and genetics of the grapevine play in determining YAN levels.

These findings are described in the article entitled A statistical exploration of data to identify the role of cultivar and origin in the concentration and composition of yeast assimilable nitrogen, recently published in the journal Food Chemistry as part of a wider study focused on the N content of South African wine grapes. The study delved further into the statistical classification of cultivars based on amino acid content and calibration of IR instrumentation for high throughput measurement of N content. This work was conducted by Gabriella Petrovic, Martin Kidd, and Astrid Buica from Stellenbosch University.

About The Author

Astrid is working as a Assistant Professor in the field of Agri Sciences in the department of Soil Science at Stellenbosch University, South Africa.

Gabriella is a research scientist at the Stellenbosch University Institute for Wine Biotechnology.