WINEMAKINGAWRI REPORT

In vino veritas – investigating technologies to fight fraud

By Eric Wilkes, Martin Day, Markus Herderich and Dan Johnson Managing director The Research Institute, PO BOX 197, Glen Osmond, South Australia 5064 Dan Johnson

Many of the great of the world have been subject to claims of substitution or counterfeiting, damaging the reputation of the affected brands and countries of origin. This article describes recent work at the AWRI looking at analytical approaches to determine the origin of wine that could form an important part of global efforts to fight wine fraud.

he Grocery Manufacturers’ Association estimates that fraud AT A GLANCE: may cost the global food industry • Fraud in the global food industry is estimated to cost between $10-15 Tbetween $US10 billion and $15 billion per billion per year year, affecting approximately 10% of all • Wine fraud can involve product substitution or alteration, mislabelling, commercially sold food products (Johnson counterfeiting or sale of stolen goods 2014). Wine is an attractive target for fraud • High profile cases of fraud have considerable potential to damage the both for high value iconic products where reputation of wine producers demand outstrips supply, and in the high • A recent AWRI project investigated the potential for a combination of volume section of the market. Wine fraud analytical techniques to predict whether or not a wine was of Australian can involve a variety of different activities, origin including: • A combination of strontium isotope ratios and trace metal concentrations • dilution or partial replacement gave excellent results in classifying wines as Australian or non-Australian, with an alternative wine or water to as well as indications that it could be used to identify the region of origin increase the overall volume of Australian wines • substitution with an alternate • Greater sample numbers and additional analytes will be needed to develop (cheaper) wine these promising results into a robust tool to fight wine fraud. • artificial enhancement of wine through addition of an unapproved additive to improve colour or aroma Attempts to establish robust analytical can provide increasingly robust results • mislabelling, for example, methods to verify a wine’s authenticity have for the identification of provenance. The intentional misrepresentation of been under way for many years, particularly AWRI recently completed a proof-of- country of origin, region, , in Europe, with mixed success. Some cases concept project to investigate whether variety, or production techniques of adulteration and counterfeiting can be a combination of different analytical • misrepresentation of the geographic easily detected through obvious typographic techniques could be used to determine origin of a product through false errors on labels or the poor taste of whether or not a wine was made from customs documents, particularly for bulk shipments wine, but others require sophisticated Australian grapes. A secondary aim was analytical approaches. Given the diverse to find out if the chosen parameters • counterfeiting of rare, collectible or range of methods used to commit wine could also identify the region of origin of well-known wines fraud, no single solution exists. This AWRI Australian wine. • wine theft and resale through report focuses on recent investigations of For the study, 231 commercial wines unapproved channels. analytical approaches to authenticate the from major wine regions of Australia In almost all cases, wine fraud is purely inherent attributes of a wine. were sourced along with 37 international economic and does not pose a health risk to consumers. wines from a selection of the major bulk COMBINING TESTS FOR MORE wine-producing countries. The samples TESTING THE AUTHENTICITY OF WINES ROBUST RESULTS were evenly split between red and white Efforts to fight wine fraud can be broadly wines, specifically split into two areas: One possible avenue to produce more and , with the exception of • developing packaging solutions or robust tools for determining the origin the red wines from Tasmania where Pinot other product attributes that can of a wine (or any food or beverage) is to Noir wines were chosen. The analytical be used during manufacturing or at use more than one unrelated parameter. methods chosen were strontium isotope the point of sale to establish product Each parameter be in some ratios (namely the ratio of 87Sr/86Sr), the authenticity manner influenced by the product’s concentration of a range of trace metals • development of analytical tests origin, but not necessarily sufficient on found in wine and the wines’ spectra which can positively identify the its own to provide definitive confirmation. in the mid-infrared, near-infrared, source of a given product. When used in combination, however, they ultraviolet and visible regions.

36 www.winetitles.com.au WINE & JOURNAL MARCH/APRIL 2016 V31N2 WINEMAKINGAWRI REPORT

DOES HAVE A BIG INFLUENCE ON THE IS THIS WINE AUSTRALIAN? ANALYTICAL PARAMETERS? For the primary question of the study, ‘Is it possible to Isotope ratios are generally not affected by winemaking differentiate wine produced in Australia from those produced processes. The relative amounts of trace metals are, in other countries?’, the first step was to review the strontium however, more likely to be influenced by winemaking. Hence, ratio data in isolation. Using a simple one-way ANOVA test, a it was important to understand if the chosen parameters distinct difference was found between the 231 wines produced were influenced by winemaking practices. Or, to put it in Australia and the 37 wines from other major wine-producing another way, to find out if changes in winemaking practice countries (Figure 1). could cloud the provenance determined using the analytical However, while the means were statistically significantly data. different, the range of results showed a degree of overlap (Figure

Statistical analysis of the combined trace metals and 2). This suggests that in itself, the strontium isotope ratio of any strontium isotope data showed that once the influence of given wine is not enough to confidently classify a wine as being ▶ elements known to be influenced by winemaking practices Australian or not. (e.g. copper) had been removed, there was essentially no difference between the red and samples. This 0.712 gives high confidence that wine production practices are 0.720 having little effect on the chosen analytical measures 0.711 0.715 Sr Sr 86 because red and white wines are made by very different 86 0.710 Sr / Sr / processes. 87 87 0.710 The various spectral techniques investigated were, 0.709 however, much more dependent on the organic chemistry 0.705 0.708 of the wine in question and, hence, more open to influence from winemaking style and practice. While the spectral data Australian Non-Australian Aus-87/86Australian Sr Non-Aus_87/86Non-Australian Sr showed some differentiation based on provenance, it was Figure 1. Comparison of the Figure 2. Box plot showing much less robust than that supplied by the isotopic ratios mean value for strontium isotope the median and distribution of or trace metal concentrations. As such, the use of spectral ratios and 95% confidence the strontium isotope ratios measures was not pursued further. intervals for the Australian and for Australian (n=231) and international wines. international wines (n=37). Leave It To Us The team at Portavin makes your life easier when it comes to wine bottling and services. With ten working across five sites, your wine is in the hands of a family owned company that cares. From bottling to packing and dispatching export containers, quality assessment and warehousing – there’s no task too large or too small. And it’s reassuring to know that we offer all major accreditations and our sites are close to key transport hubs, reducing environmental impact. Call Portavin today for an obligation free chat about your business needs.

Portavin – caring for your wine from tank to shelf.

Adelaide Margaret River Melbourne Perth Sydney (08) 8447 7555 (08) 9755 0500 (03) 9584 7344 (08) 9437 1033 (02) 9722 9400

www.portavin.com.au [email protected]

V31N2 WINE & VITICULTURE JOURNAL MARCH/APRIL 2016 www.winetitles.com.au 37 VITICULTUREAWRI REPORT

To further strengthen the ability to WHERE IN AUSTRALIA IS IT FROM? WHERE TO FROM HERE? differentiate the wines, the trace metal concentration data was combined with As an extension of the above study the It is clear that the approach of using the strontium ratio data. Statistical data was next used to determine if it was a combination of unrelated analytes analysis showed 11 trace elements (As, possible to identify the Australian wines’ such as stable isotope ratios and trace Be, Cr, Cu, Li, Ni, Pb, Rb, Sn, V and Zn) region of origin using similar techniques. metal concentrations has potential for were the most promising. The data for Australia has 67 defined geographical determining the provenance of wines, copper and tin were removed due to the indicators. However, the limited nature of both at the country and regional level. The possible influences from winemaking the dataset meant that it was not practical project also suggests that there are unique and packaging processes. The remaining to try to achieve this level differentiation characteristics derived from the various data, in combination with the strontium and, instead, analysis was conducted based Australian wine regions, and that they differ isotope data, were analysed using partial on the eight production zones represented from the rest of the world. least squares discriminant analysis (PLS- by the wines in the set. After statistical This study provides the basis for a DA), a statistical technique commonly analysis it could be seen that there was database and methodology to evaluate the used to understand the relationships significant grouping of the different provenance of wine labelled as Australian. between different sets of data. The rate production zones (Figure 3) in the PLS-DA However, to provide a truly robust tool the of correct classification from this analysis data. dataset needs to be extended significantly was 94.7%, with 187 out of 194 Australian Further inspection of these results to provide the levels of confidence required wines correctly assigned, and 30 out suggested they were intimately linked to in the case of disputes. This extension will of 37 wines being correctly assigned to the underlying geology of the regions being need to include a much larger group of the ‘other countries’ group. While not studied. This was confirmed by overlaying international wines and a range of other yet ‘foolproof’, this is a strong result. geological and soil data on the zones and analytes such as other stable isotope It is expected that if the data set were combining those with similar underlying ratios, to help further differentiate wine increased to include a much larger set geology. Using this information, it was from regions around the world and within of international wines and an increased possible to allocate 85% of wines to their Australia. It will also be important to range of Australian wines, then a robust correct region. The strong influence of investigate the effect of and vintage classification tool could be developed, geology is promising as it suggests that the changes. A further target will be to develop particularly if further unrelated analytical dataset is likely to become more robust and analytical methods to identify the variety of measures (such as the stable isotope allow for even more detailed differentiation grapes a wine has been made from. This ratios of other elements such as boron or as the number of wines and regions will require different analytical approaches lithium) were included. analysed increases and other geologically- from those outlined here, with DNA and based parameters are introduced. protein analysis techniques likely to be investigated.

ACKNOWLEDGEMENTS

1 This work was supported by Australia’s 0.8 grapegrowers and winemakers through their investment body Wine Australia, 0.6 with matching funds from the Australian Government. The AWRI is a member of the 0.4 Wine Innovation Cluster, in Adelaide. The

0.2 authors thank Wine Australia for assistance %) 9 with sourcing samples, and Ella Robinson 0 for her editorial assistance. 2 (11%, 2 (11%, - to r -0.2 Fac REFERENCES AND FURTHER READING

-0.4 Everstine, K.; Spink, J. and Kennedy, S. (2013) Economically motivated adulteration (EMA) of food: common characteristics of EMA incidents. J. Food -0.6 Protection 76(4):723-735. Frank, M. (2015) Counterfeit wine goes global. -0.8 40(11):39-41. Johnson, R. (2014) Food fraud and ‘economically -3 -2 -1 0 1 2 3 4 motivated adulteration’ of food and food ingredients. Factor-1 (57%, 8%) Congressional Research Service Report https://www. fas.org/sgp/crs/misc/R43358.pdf Margaret River Central Ranges Tasmania

Yarra Valley Big Rivers McLaren Vale Hunter Valley

Figure 3. Grouping of Australian wines by production zone after statistical analysis of trace metal and strontium isotope data. Dotted ellipses are for clarity of groupings only and are not statistically derived. WVJ

38 www.winetitles.com.au WINE & VITICULTURE JOURNAL MARCH/APRIL 2016 V31N2