Abstract
Immunoassays with some dessert wines made from Botrytis-infected grapes (referred to as Botrytis wines) have shown that Botrytis antigens can be detected even at dilutions of 1:64,000, while none or very low levels have been detected in those wines that do not claim to be made from Botrytis-infected grapes. It is commonly accepted that Botrytis wines have a characteristic honeylike aroma, but there are apparently no scientific reports on the differences in sensory perception of dessert wines made from Botrytis-infected grapes and those made from noninfected grapes, such as icewines. The purpose of this project was to determine if it was possible to distinguish, through sensory descriptive analysis, Botrytis and non-Botrytis wines as determined by immunoassay. Despite the high variability in grape variety, wine processes, and wine regions of the wines selected for this study, there is a trend showing that wines with low Botrytis antigen units are high in fresh fruit and citrus aromas. However, wines with high Botrytis antigen levels were not especially high in perceived honey aromas. This result could be an artifact of the high variability in varieties, wine styles and wine regions.
Late-harvest wines, commonly described as dessert wines, are generally characterized by being sweet, having a fruity aroma and a smooth mouthfeel. The high sugar content of these wines is achieved naturally in one of two ways. First, the berries are left on the vine until freezing occurs. The frozen berries are then harvested and pressed to remove only the concentrated juice and not the ice crystals (icewines). Icewines are traditionally made in Germany; however, Canada has been producing well-regarded ice wines since the late 1980s. Second, late-harvest infections of the fungus Botrytis cinerea are encouraged (Botrytis wines). These late-season infections of Botrytis, which should not be accompanied by other fungal and bacterial infections, are known as “noble rot.” The fungus brings about a number of changes, the most important being the removal of water.
In Germany, a range of these dessert wines is produced: Spätlese (with light Botrytis infection) and Auslese, Beerenauslese, and Trockenbeerenauslese made from grapes with increasingly higher levels of Botrytis. In Hungary, Tokaji Aszu wine is made by adding a paste of heavily Botrytis-infected grapes (Aszu grapes) to Tokaji must or wine. The marketing of the Aszu wines as 3, 4, 5, or 6 puttonyos is notionally based on the number of Aszu grape collection tubs added to the barrels of must or wine (Lambert-Grócs 2003). In Bordeaux the more well-known dessert wines are the Sauternes, made from botrytized Semillon and Sauvignon blanc grapes. In California, the very dry climate is not conducive to field infections of grape berries by Botrytis, and methods have been developed to achieve infections artificially by either spraying berries late in the season with a suspension of Botrytis spores or harvesting whole bunches and spraying and incubating them in a controlled humid environment.
There are reports on the changes in the composition of grape juice and wine brought about by infections with Botrytis noble rot. These berries have a higher concentration of tartaric acid and a lower concentration of malic acid than noninfected berries (Carpantié 1954, in Ribéreau-Gayon et al. 1980). They also contain significant concentrations of gluconic acid and higher concentrations of glycerol, erythritol, arabitol, mannito, and trehalose than noninfected berries (Ribéreau-Gayon et al. 1980). Barbe et al. (2001) studied the effect of Botrytis on certain microorganisms on the SO2 binding power of musts and found that acetic acid bacteria were responsible for this increased SO2 binding power by oxidizing the three main sugars from botrytized grapes. Bock et al. (1988a) found that noble rot interferes in the microbial transformation of flavor substances such as geraniol and nerol and in the reduction of cinnamaldehyde and unsaturated acids (Bock et al. 1988b). Flath et al. (1972) found that the addition of the lyophilized mycelium of Botrytis cinerea increased the headspace concentration of a number of volatiles such as acetaldehyde, 1-,1-diethoxyethane, and 2,4,5-trimethyl-1,3-dioxolane. The effect of adding Botrytis-infected grapes on aroma components of Tokaji Aszu wines was studied by Miklósy et al. (2000) and Miklósy and Zoltán (2004). In their study in 2000, the authors performed sensory descriptive analysis on the wines, but the results were only briefly mentioned in the article. Few other studies have been performed in which sensory analysis has been included, although one recent study (Nurgel et al. 2004) characterized sensory properties of Canadian icewines in combination with chemical attributes.
It is commonly accepted that Botrytis wines have a characteristic honeylike aroma, but there are apparently no scientific reports on the differences in sensory perception of dessert wines made from Botrytis-infected grapes and those made from noninfected grapes, such as icewines. That is because, until recently, there has been no method of determining the concentration of Botrytis-specific compounds in various dessert wines. We have now shown (Dewey 2002) that Botrytis cinerea antigens can be quantified in wine using an immunoassay employing a Botrytis-specific monoclonal antibody, BC-12.CA4, originally developed to detect and quantify water-soluble antigens of Botrytis present in grape juice from infected berries (Meyer and Dewey 2000, Dewey and Meyer 2004). Immunoassays with some Botrytis dessert wines have shown that Botrytis antigens can be detected even at dilutions of 1:64,000, while none or very low levels have been detected in those wines that do not claim to be made from Botrytis-infected grapes (Dewey 2002).
The purpose of this project was to determine whether, by conducting a sensory analysis of a number of Botrytis and non-Botrytis wines from different vintages and from different geographical areas, it is possible to distinguish between Botrytis and non-Botrytis wines by sensory descriptive analysis and by using the Botrytis antigen test. Additionally, we wanted to determine if there were significant correlations between the values determined by the Botrytis antigen test and the sensory descriptive data.
Materials and Methods
Wines.
Dessert wines from different regions worldwide were donated or purchased. The 23 wines (Table 1⇓) represent a variety of different grapes, vintages, and processing methods and were chosen to span the sensory variation in dessert wines as well as to include wines with low to high levels of Botrytis cinerea infections as determined by immunoassay.
Quantitative immunoassay for Botrytis antigens in wines.
A quantitative plate-trapped antigen enzyme-linked immunosorbent assay (PTA-ELISA) was used based on that developed by Dewey et al. (2000) and Dewey and Meyer (2004). The assay employs the Botrytis-specific monoclonal antibody, BC-12.CA4, raised by Meyer and Dewey (2000). Nunc MaxiSorp 96-well plates (Nalge Nunc, 442404; VWR, San Francisco, CA) were coated for two hours with the dessert wines serially diluted into phosphate-buffered saline (PBS). Wells were then incubated successively with the Botrytis antibody, BC-12.CA4 (hybridoma supernatant diluted 1:4 in PBST [PBS with 0.05% Tween 20]) for 30 min, a commercial goat anti-mouse polyvalent peroxidase conjugate (Sigma-Aldrich, St. Louis, MO) diluted 1:1000 in PBST for 30 min and the peroxidase substrate, tetramethyl benzidine (liquid TMB; Adgen, Auchicruive, UK). The enzyme reaction was stopped by the addition of H2SO4. All incubation steps were performed at ambient temperature (~20°C) and working volumes were 100 μL, and plates were washed four times between each incubation step with PBST. Absorbance values were read at 450 nm with an automated microplate reader (E Max; Molecular Devices, Sunnyvale, CA). All dessert wines were tested in triplicate together with a set of standards. The standards were prepared from a 100 μg/mL PBS extract of freeze-dried mycelium of the fungus, grown in grape juice, serially diluted into PBS. The level of Botrytis antigen units (BAgU) in each wine was determined by inverse prediction from the standard curve of the absorbance values of the standards, using the method of Dewey et al. (2000), where 1.0 μg/mL PBS extract of freeze-dried mycelium is equivalent to 1.0 BAgU.
Descriptive sensory analysis.
Twenty-three wines presented in 18 sessions were evaluated in triplicate. Each session consisted of a set of six or four wines, which were then randomized according to a Latin Square design. Each judge tasted the same six or four wines, but in a different order. This was done to reduce the number of open bottles per session and thereby prevent changes within the bottles. The data were collected by FIZZ for Windows (version 2.00 E; Biosystems, Couternon, France). Wines were evaluated in individual booths, under red light conditions, at ambient temperature, 20°C. The bottles of wines were stored at room temperature the day the wines were tested to ensure that all judges were served the samples at the same temperature.
Wines were evaluated monadically for odor and flavor. A break of 20 seconds followed each wine to allow adequate mouth cleansing by eating a cracker (optional) and rinsing thoroughly with water. After rating three samples in this manner, a 5-minute break was introduced in order to avoid adaptation. The remaining wines were then rated in the same manner as the first three.
A nonstructured, 10-cm line scale was used, anchored on the left end with “low” intensity and on the right end with “high” intensity. The panelist marked each value with a marker on the screen. The values were converted to a 9-point scale and exported to an Excel spreadsheet.
Reference standards (see Table 2⇓) of each of the rated attributes were made in a neutral white wine base. Standards were presented in general wine glasses covered with Petri dishes and smelled by panelists before each session.
Training of sensory panelists.
Panelists (three males and eight females were either students attending the viticulture and enology program at the University of California (UC) Davis or staff from the Pomology Department at UC Davis. All had prior experience participating in wine descriptive analysis panels. The panelists participated in six training sessions, lasting about one hour each. A list of attributes was developed and then discussed until agreement was reached as to their definition of usage. Examples of the dessert wines that were part of the official project were used as training samples. Standards were made before each training session and improved until agreement was reached on the composition and strength of each. These standards represented the final attributes chosen to describe the dessert wines (Table 2⇑).
Data analyses.
Sensory data were exported from the FIZZ data collection program into an Excel spreadsheet. A mixed-model analysis of variance (using judge*wines interaction as the error term) and Fisher’s protected LSD-values were calculated for each sample when wines were significantly different for an attribute. Botrytis antigen data were analyzed by one-way analysis of variance, and a Fisher’s protected LSD-value was calculated. The correlation between the sensory variables and the Botrytis antigen units was determined. All data analyses were performed using SAS statistical software (version 8e; SAS Institute, Inc., Cary, NC).
Results and Discussion
Quantitative measurements of Botrytis cinerea.
The amount of Botrytis cinerea in the 23 dessert wines as measured by the PTA-ELISA technique is shown in Figure 1⇓. The diluted samples ranged from greater than 50 BAgU/mL at a dilution of 1:1000 to less than 1.0 BAgU/mL at a dilution of 1:1000. For purposes of this study, wines with less than 1.0 BAgU/mL at a dilution of 1:1000 were defined as wines without Botrytis antigens. These wines were Dessert (2003), Late harvest (2000), Late harvest3 (2002), and, interestingly Spätlese (2002). Wines particularly high in BAgU were as expected: the four botrytized wines—Botrytis (2001), Botrytis1 (1999), Botrytis2 (1999), and Late harvest2 (2002); Auslese (2002); Late harvest3 (2002); and Tokaji (1988) with 5 puttonyos. Additionally, the Sauternes and the Icewine had some BAgU/mL, but less than the above-mentioned wines.
Sensory descriptions of wines.
The analysis of variance of the sensory attributes indicated that all attributes differed significantly across the wines. Means and least significant difference (LSD) values for all attributes for all wines are found in Table 3⇓. A few wines had a distinct ethanol aroma, namely Botrytis1 (1999), Botrytis (2001), the two Sauternes, and the Tokaji (1990). Wines particularly high in perceived floral aroma were Botrytis2 (1999), Dessert (2003), Late harvest2 (2002), and Late harvest4 (2002). Wines particularly high in perceived citrus aroma were Dessert (2003), Late harvest4 (2002), and Spätlese (2002). Wines with high perceived tropical fruit aromas were again Dessert (2003), Icewine (2002), Late harvest1 (2002), and Late harvest2 (2002). Trockenbeerenauslese (TBA) (1992), which was low in BAgU/mL, and the Tokaji (1988), which was quite high in BAgU/mL, were both high in perceived dried fruit aroma. Stone fruit aromas were high in Icewine (2002) and Late harvest4 (2002). Trockenbeerenauslese (1992) was notably high in perceived honey aroma, yet it was low in BAgU/mL. The only wines with significant oak and toasty/smoky aromas were the two Tokajis (1988 Tokajis (1990) and the Sweet white (2000). The Tokajis were also the only wines with high caramel and nutty aromas. Sauterne (1997) had a rubbery aroma and Late harvest3 (2002) was high in earthy aroma. Grassy aromas were perceived in Dessert (2003), Late harvest (2000), and Spätlese (2002). Wines high in perceived bitterness tended to also be high in perceived hotness; exceptions were Kabinett (2000), Late harvest1 (2002), and Spätlese (2002). The wines ranged from perceptibly not sweet (Kabinett and Spätlese) to intensely sweet (Icewine, Late harvest1, and Trockenbeerenauslese [1992]). Trockenbeerenauslese (1992) was also the most viscous, while the Spätlese and Kabinett were the least viscous. These latter two wines were also perceived to be the most sour.
Relationship between sensory descriptive analysis and Botrytis antigens.
Correlation coefficients between the individual sensory attributes and the Botrytis antigen units (BAgU) are shown in Table 4⇓. The floral citrus and grassy aromas were significantly negatively correlated with BAgU/mL. Oak and toasted/smoky aromas as well as hotness were significantly positively correlated with BAgU/mL. Interestingly, BAgU/mL was not correlated with sensory attributes traditionally ascribed to botrytized wines, namely honey and dried fruit.
Conclusions
Despite the high variability in grape variety, wine processes, and wine regions of the wines selected for this study, it was possible to determine the BAgU/mL in all of these wines. There did not seem to be a correlation between level of BAgU/mL and honey or dried fruit aroma in these wines, but that could have been an artifact of the wide range of varieties, regions, and wine styles. It would be interesting to use this technique as a screening tool to determine Botrytis levels in a greater range of dessert wines as well as in table wines, as it would allow one to determine the levels of Botrytis in the wines. In future studies, it would be interesting to combine sensory descriptive analysis, Botrytis antigen levels, and GC-MS in the study of selected wines in order to explore which specific components contribute to the Botrytis flavor in these wine types.
Footnotes
Acknowledgments: We would like to thank the following wineries for donating wines: Far Niente, Beringer, Freemark Abbey (in California), and Inniskillin (in Canada). We also thank Dr. Ulrich Fischer for the donation of the German wines. All other wines were purchased at retail.
- Received December 2004.
- Revision received June 2005.
- Copyright © 2005 by the American Society for Enology and Viticulture