Detection of Red Wine Faults Over Time with Flash Profiling and The

beverages

Article Detection of Red Wine Faults over Time with Flash Proﬁling and the Electronic Tongue

Victoria D. Paup 1, Tara Cook-Barton 1, Charles Diako 2 , Charles G. Edwards 1 and Carolyn F. Ross 1,*

1 School of Food Science, Washington State University, Pullman, WA 99164, USA; [email protected] (V.D.P.); [email protected] (T.C.-B.); [email protected] (C.G.E.) 2 School of Food and Advanced Technology, Massey University, Auckland 0745, New Zealand; [email protected] * Correspondence: [email protected]; Tel.: +1-(509)-335-2438

Abstract: Wine faults, often caused by spoilage microorganisms, are considered negative sensory attributes, and may result in substantial economic losses. The objective of this study was to use the electronic tongue (e-tongue) and ﬂash sensory proﬁling (FP) to evaluate changes in red wine over time due to the presence of different spoilage microorganisms. Merlot wine was inoculated with one of the following microorganisms: Brettanomyces bruxellensis, Lactobacillus brevis, Pediococcus parvulus, or Acetobacter pasteurianus. These wines were analyzed weekly until Day 42 using the e-tongue and FP, with microbial plate counts. Over time, both FP and e-tongue differentiated the wines. The e-tongue showed a low discrimination among microorganisms up to Day 14 of storage. However, at Day

21 and continuing to Day 42, the e-tongue discriminated among the samples with a discrimination index of 91. From the sensory FP data, assessors discriminated among the wines starting at Day 28.

Citation: Paup, V.D.; Cook-Barton, Non-spoilage terms were used to describe the wines at significantly higher frequency for all time T.; Diako, C.; Edwards, C.G.; Ross, points until Day 42, at which point the use of spoilage terms was significantly higher (p < 0.05). These C.F. Detection of Red Wine Faults results suggest that application of these novel techniques may be the key to detecting and limiting over Time with Flash Profiling and financial losses associated with wine faults. the Electronic Tongue. Beverages 2021, 7, 52. https://doi.org/10.3390/ Keywords: electronic tongue; flash profiling; wine; faults; spoilage beverages7030052

Academic Editors: Stamatina Kallithraka and 1. Introduction Matteo Marangon Wine faults are any off ﬂavors, taste, or mouthfeel attributes that occur during the

Received: 30 April 2021 winemaking process [1]. Signiﬁcant ﬁnancial losses can ensue when wine faults occur, due Accepted: 19 July 2021 to the cost of storage and destruction [2]. Prominent sources of these wine faults are spoilage Published: 21 July 2021 microorganisms, such as those within the genera Lactobacillus, Pediococcus, Acetobacter, and Brettanomyces. These spoilage microorganisms produce unique metabolites that can impart

Publisher’s Note: MDPI stays neutral undesirable flavors or aromas [3]. In particular, Brettanomyces produces volatile phenols that with regard to jurisdictional claims in can be described as smoky, sweaty, and barnyard [4–7]. Pediococcus produces β-D-glucan published maps and institutional affil- and acrolein that produce ropey, viscous, and oily mouthfeel attributes and can increase iations. perceived bitterness [3,8,9]. Additionally, Lactobacillus produces lactic acid, diacetyl, and other secondary metabolites that can produce buttery and vinegar aromas [3,10–12]. Finally, Acetobacter can produce high amounts of acetic acid and ethyl acetate imparting a vinegar aroma on the final wine [3,13].

Copyright: © 2021 by the authors. The inﬂuence that different metabolites have on the wine is dependent on the con- Licensee MDPI, Basel, Switzerland. centration and wine style [14]. At low concentrations, certain metabolites can impart This article is an open access article complexity; however, at high concentrations certain metabolites negatively impact wine distributed under the terms and quality [15–17]. In particular, Brettanomyces’ metabolites, such as 4-ethylphenol and conditions of the Creative Commons 4-ethylguaiacol, at concentrations less than 400 µg/L, increase complexity through an Attribution (CC BY) license (https:// increase in spice, leather, and smoke aromatic notes, but at 620 µg/L, the elicited aromas creativecommons.org/licenses/by/ cause the wine to be considered spoiled [15,16]. Therefore, early detection of metabolites 4.0/).

Beverages 2021, 7, 52. https://doi.org/10.3390/beverages7030052 https://www.mdpi.com/journal/beverages Beverages 2021, 7, 52 2 of 14

from spoilage organisms is crucial to allow early remediation that may inhibit spoilage from further progressing. Due to the high frequency of testing needed to detect a microbial infection, rapid testing methods are critical. Both rapid sensory and analytical methods are available, but limited research has been completed on their effectiveness in detecting and differentiating wine faults caused by spoilage microorganisms. Rapid sensory methods have increased in popularity due to the speed of data collection and analysis, as well as cost effectiveness [18]. One rapid sensory method is Flash Profiling (FP), originally introduced by Sieffermann [19] as a proposed variant to Free Choice Profiling [20]. Assessors are asked to determine sensory attributes that best differentiate all the samples and then rank the samples for each of the attributes they selected [21,22]. The utilization of FP has been used in a variety of products including wines [18,23], and hot served food products [24]. While specific research on wine faults has not been completed, the previous applications of the FP method indicate it may be an effective option for wine fault detection. In addition to rapid sensory methods, analytical methods for fast differentiation among samples are available. Specifically, the electronic tongue (e-tongue) has been used as a rapid testing method for a wide variety of food products such as wines [25–29], dairy products [30–32], sweeteners [33,34], and spicy products [35,36]. The potentiometric e- tongue evaluates non-volatile compounds and similar to the human tongue can provide a “taste fingerprint” based on the sensory profile of the wine [29,35]. The Alpha MOS® e-tongue utilizes seven cross-selective polycarbonate coated sensors that select for different ions or compounds that are associated with the five basic tastes, as well as spicy and metallic [37]. A previous study determined that the e-tongue was able to differentiate among different wines spiked with varying levels of 4-EC, a Brettanomyces metabolite, at concentrations lower than the sensory threshold [29]. Similarly, the e-tongue has successfully been used to determine defects related to acetic acid concentrations in Apulian red wine [38]. These results indicate that the e-tongue may be a valuable tool in differentiating among wine faults and early detection of potential infections. The objective of this study was to use the e-tongue and FP to evaluate changes in wine over time due to the presence of different spoilage microorganisms. Five common spoilage microorganisms were selected to determine if the e-tongue was able to differentiate among different spoilage microorganisms. Analysis on the inoculated red wine was completed weekly to determine the time point at which the e-tongue and the assessors could differentiate among the wine faults. Based on previous research, it was hypothesized that the e-tongue would differentiate among the inoculated wine samples before the human assessors would be able to. Overall, the data obtained in this study will help determine if the e-tongue can be a useful tool in the early detection of sensory changes elicited by spoilage microorganisms.

2. Materials and Methods 2.1. Materials Yeast peptone dextrose and yeast/mold broth were obtained from Becton, Dickinson, and Company (Sparks, MD, USA). Agar was purchased from Acros Organics (Morris, NJ, USA). Fructose, glucose, ethanol, and hydrogen peroxide were purchased from Fisher Sci- entiﬁc, (Waltham, MA, USA). Sodium chloride, hydrochloric acid, and sodium-L-glutamate solutions were acquired from Alpha Mos (Tolouse, France) for e-tongue conditioning, calibration, and diagnostic procedures.

2.2. Yeast Strains and Starter Cultures Brettanomyces bruxellensis strains I1A and F3 were acquired from the Washington State University culture collection and grown on yeast peptone dextrose (YPD) agar incubated at 28 ◦C[39]. Starter cultures were prepared by transferring a single colony from YPD agar to 50 mL of yeast and mold (YM) broth. Following incubation for 4 days at 28 ◦C, 1 mL of culture was transferred to 50 mL of YM broth containing 5% (v/v) ethanol. After 4 days Beverages 2021, 7, 52 3 of 14

of incubation under the same conditions, 1 mL of culture was transferred to YM broth containing 10% (v/v) ethanol. Cells were harvested in late exponential growth phase by centrifuging samples at 2000× g for 15 min, washing twice with 0.2 M Na2HPO4 (pH 7.0) buffer, and then resuspending in the same buffer prior to inoculation.

2.3. Bacterial Strains and Starter Cultures Acetobacter pasteurianus ATCC 12873, Lactobacillus plantarum WS-16, and Pediococcus parvulus WS-29A were acquired from the Washington State University culture collection and grown on modiﬁed apple juice Rogosa (MR) agar [4] incubated at 28 ◦C. Starter cultures were prepared by transferring a single colony from MR agar to 50 mL of MR broth. Following incubation for 7 days at 28 ◦C, 1 mL of culture was transferred to 50 mL of MR broth containing 5% (v/v) ethanol. After 7 days of incubation under the same conditions, 1 mL of culture was transferred to MR broth containing 10% (v/v) ethanol. Cells were harvested in late exponential growth phase by centrifuging samples at 2000× g for 15 min, washing twice with 0.2 M Na2HPO4 (pH 7.0) buffer, and then resuspending in the same buffer prior to inoculation.

2.4. Wines Merlot wine (Yakima Valley, WA, USA) from the 2018 vintage was obtained and free SO2 was removed using hydrogen peroxide. Sugars (1 g/L glucose, 1 g/L fructose) were added before the wine was sterile-filtered through 0.45 µm polyvinylidene fluoride cartridges (Millipore Sigma, Bellerica, MA, USA) housed in stainless-steel filter housings (Pall, Port Washington, NY, USA) into sterile 750 mL screw-capped bottles. 740 mL of wine was added to each bottle. Microorganisms were inoculated at 1 × 104 CFU/mL into the bottles (n = 14 bottles per microorganism). Bottles were closed with screw caps, with the exception of bottles containing A. pasteurianus and an uninoculated control, which were sealed and halfway unscrewed to allow air flow. The air flow was important due to the growth requirements of A. pasteurianus [13]. Additionally, the inclusion of oxygen may increase oxidation, therefore a control with the same condition was included. Wines were stored at 23 ◦C for 42 days, with replicate bottles of each treatment (n = 5), in duplicate, and the controls (n = 2) removed for weekly analysis (Table1).

Table 1. Wine treatments evaluated by electronic tongue and Flash Proﬁling. Bottles were fully closed with screw caps (sealed) or partially unscrewed to allow air ﬂow.

Code Inoculated Microorganism Closure Bottle Replicate 1A Brettanomyces I1A Sealed 2 1B Brettanomyces F3 Sealed 1 1C None Sealed 1 1D None Partially unscrewed 1 1E Acetobacter pasteurianus Partially unscrewed 1 2A Pediococcus parvulus Sealed 1 2B Lactobacillus brevis Sealed 1 2C Acetobacter pasteurianus Partially unscrewed 2 3A Pediococcus parvulus Sealed 2 3B Lactobacillus brevis Sealed 2 3C Brettanomyces I1A Sealed 1 3D Brettanomyces F3 Sealed 2 Beverages 2021, 7, 52 4 of 14

2.5. Spoilage Microorganism Culturability Prior to weekly sensory analysis, microorganism culturability was monitored by spiral plating (Autoplate 4000, Spiral Biotech, Bethesda, MD, USA) onto Wallenstein Laboratory agar (WL, Becton, Dickinson, and Company, Franklin Lakes, NJ, USA) for B. bruxellensis or MR for bacteria. When culturability declined, 1 mL of wine was spread plated onto media to detect low populations. All plates were incubated at 28 ◦C for 7 days prior to counting.

2.6. Flash Profiling The sensory panel consisted of assessors (n = 7, 4 females and 3 males) age 22 to 45 with previous trained panel experience, not exclusive to alcoholic beverages. These assessors were recruited from the Washington State University community through electronic announcements. The use of human subjects for this study was approved by the Washington State University Institutional Review Board (IRB # 17595-001) and informed consent was obtained from all assessors prior to participation. All samples (n = 12) were evaluated in each session under white light in partitioned booths in the Washington State University Sensory Evaluation Facility in Pullman, WA., USA. The wines samples (30 mL) were poured 1 h prior to evaluation and covered with a petri dish and presented in ISO glasses with letter codes at 23 ◦C[22]. During the initial evaluation at Day 0, assessors had a brief orientation session to familiarize the assessors with the concept of FP, outline the procedures, and complete a practice evaluation. Each assessor was presented with all of the wine samples (n = 12) simultaneously. They were asked to identify the aroma attributes that best described the differences among the samples, and rank the samples based on those identified attributes from lowest to highest intensity [22]. Ties were allowed between the samples. Assessors were instructed to avoid hedonic terms. To facilitate aroma term generation, the Brettanomyces and red wine aroma wheel were provided during all evaluation sessions, as well as a list of terms generated after the first session [40,41]. FP was completed weekly for 42 days, starting on Day 0 immediately after inoculation, for a total of seven evaluation sessions. To determine the extent of spoilage in the inoculated wines, the terms generated by the FP were categorized into two categories, traditional spoilage or non-spoilage terms. Traditional spoilage terms were identified using the categories with negative connotations (dairy, fermentation, earthy, chemical/solvent, rotten and putrid, animal, savory, and veggie) from the Brettanomyces aroma wheel [41]. Additional terms, such as geranium and mousy, were included based on their association with Lactobacillus and Pediococcus infections [3,12]. The percentage of spoilage-related terms compared to the overall term usage was used to indicate when spoilage occurred in the inoculated wine.

2.7. Electronic Tongue The wines were analyzed using a potentiometric e-tongue (Astree® II electronic tongue unit, Alpha MOS®, Toulouse, France) at each time point immediately after FP was completed. The e-tongue utilizes seven cross-selective taste sensors (salty, sweet, bitter, umami, spicy, metallic, and sour), a reference electrode (Ag/AgCl) and the AlphaSoft software (ver. 12, Alpha MOS®, Toulouse, France). Instrument preparations and sampling parameters were followed according to manufacturers’ procedures and as previously described [29,35]. For each e-tongue run, the analysis sequence contained randomly selected treatments. Each treatment consisted of a group of four 25 mL beakers, in addition to a reference wine sample (Carlo Rossi Burgundy). Samples were separated by MilliQ water for a 10 s sensor cleaning. The reference samples allowed the comparison among samples from different runs. At each time point, three runs were completed to analyze all the samples. Until e-tongue analysis could be completed, opened wine bottles were ﬂushed with nitrogen and kept at 4 ◦C to limit additional microorganism growth, with all samples being analyzed within two days of FP being completed. Beverages 2021, 7, 52 5 of 14

2.8. Statistical Analysis The FP data were analyzed using Generalized Procrustes Analysis (GPA) to produce consensus conﬁgurations for the samples and sensory attributes. Agglomerative Hierar- chical Cluster (AHC) analysis was completed on the FP data to determine if any clusters existed among the samples. The z-test for two proportions was conducted on the frequency of spoilage and non-spoilage related terms (p ≤ 0.05). Statistical analysis was competed with XLSTAT Sensory 19 (Addinsoft, New York, NY., USA). The electronic tongue data were analyzed using the Astree AlphaSoft software (ver. 12, Alpha MOS®, Toulouse, France) to calculate the discrimination index (DI); the software was also used to perform Principal Component Analysis (PCA) in order to visually determine differences among the samples. The discrimination index is a measurement of the overlap among the samples and the distance between the samples, giving an indication of how well the electronic tongue can differentiate among samples. Strong discrimination is indicated by a discrimination index above 80 [29]. The PCA biplots were plotted using R Base Graphics (ver. 4.0.3, R Core Team, Vienna, Austria) based on the loadings and scores from the PCA obtained from the AlphaSoft software.

3. Results 3.1. Culturability of Inoculated Microorganisms After inoculation (~5 × 105 CFU/mL) of Brettanomyces I1A, Brettanomyces F3, P. parvulus, A. pasteurianus, and L. brevis into the wine, populations of Brettanomyces I1A, Bret- tanomyces F3, and A. pasteurianus declined immediately by about 1 log (Figure1). For both strains of Brettanomyces, the populations steadily increased for 21 days before reaching a relatively consistent population at roughly 1 × 106 CFU/mL for the remaining storage time. Similar growth trends and CFU/ mL were observed in a previous study examining the inﬂuence of temperature and sulfur dioxide additions [42]. Previous studies have observed 4 Beverages 2021, 7, x FOR PEER REVIEWthat as few as 10 CFU/mL of the Brettanomyces organism can cause medicinal, horsey,6 of 14 or barnyard-like aromas in red wine, indicating that the populations observed in this study would be sufﬁcient to evoke sensory changes [43–45].

Figure 1. Culturability (CFU/mL) of Brettanomyces I1A (orange), Brettanomyces F3 (green), Pediococcus Figure 1. Culturability (CFU/mL) of Brettanomyces I1A (orange), Brettanomyces F3 (green), Pediococ- cusparvulus parvulus(grey), (grey),Acetobacter Acetobacter pasteurianus pasteurianus(yellow), (yellow), and andLactobacillus Lactobacillus brevis brevis(blue) (blue) in Merlot in Merlot wine wine stored ◦ storedat 23 atC 23 over °C42 over days. 42 Adays. dashed A dashed line indicates line indicates that populations that populations were below were the below detection the detection limit. limit. The remaining spoilage microorganisms displayed different growth trends com- 3.2.pared Electronic to the TongueBrettanomyces strains. P. parvulus maintained the initial inoculation levels for 21 days, before experiencing a slight decrease in CFU/ mL (Figure1). The last two strains, A. pasteurianusPCA was performedand L. brevis to ,determine had similar the growth ability trends.of the e-tongue After inoculation, to discriminateA. pasteurianus among theand inoculatedL. brevis had wine a sharpsamples decrease at each in time population point. At until Day reaching 14, the e-tongue “too few had to count” a DI of at − Day16% 7 when discriminating among the inoculated samples (Figure 2). The negative DI indicated overlap among some of the samples and therefore an inability to discriminate among the samples. In particular, L. brevis (Bottle 1 and 2), P. parvulus (Bottle 2), and the sealed control had substantial overlap. The PCA described 99.9% of the variation, with PC1 being associated with the response to the salty sensor and PC2 being associated with the bitter, metallic, and umami sensors. Similar results were observed for Day 0 and 7, with the e- tongue having various levels of negative DI.

Figure 2. Principal component analysis of Merlot wine samples on Day 14 of storage at 23 °C as analyzed by the electronic tongue. Samples are labeled with names as described in Table 1. The discrimination index was −16.

Beverages 2021, 7, x FOR PEER REVIEW 6 of 14

Beverages 2021, 7, 52 6 of 14

and 21, respectively. The decrease in population may be attributed to unfavorable environmental conditions. L. brevis prefers environments below 12.8% ethanol before growth Figureis restricted 1. Culturability and most red (CFU/mL) wine exceeds of Brettanomyces that concentration I1A (orange), [46]. Similarly, BrettanomycesA. pasteurianus F3 (green), Pediococ- cusprefers parvulus lower (grey), ethanol Acetobacter content wines pasteurianus and requires (yellow), oxygen and to Lactobacillus fully grow [brevis13]. To (blue) allow in Merlot wine storedoxygen at migration,23 °C over the 42 bottles days. wereA dashed loosely line sealed; indicates however, that this populations may not have were allowed below the detection limit.sufﬁcient oxygen to encourage growth, as demonstrated by the sharp population decrease initially observed. After the sharp initial decrease, both A. pasteurianus and L. brevis increased in population after 14 to 7 days, respectively; however, the population never 3.2.exceeded Electronic the initial Tongue inoculation population. Despite not exceeding the initial inoculation population,PCA was the microorganismsperformed to determine were still growing the ability and producing of the e-tongue metabolites to that discriminate may among thehave inoculated inﬂuenced wine the sensory samples characteristics at each time of the po wineint. [ 13At]. Day 14, the e-tongue had a DI of −16% when3.2. Electronic discriminating Tongue among the inoculated samples (Figure 2). The negative DI indicated overlapPCA among was performed some of to the determine samples the and ability therefore of the e-tongue an inability to discriminate to discriminate among among the samples.the inoculated In particular, wine samples L. brevis at each (Bottle time 1 point. and 2) At, P. Day parvulus 14, the e-tongue(Bottle 2) had, and a DI the of sealed control had−16% substantial when discriminating overlap. The among PCA the described inoculated samples99.9% of (Figure the variation,2). The negative with DIPC1 being asso- ciatedindicated with overlap the response among some to the of the salty samples sensor and and therefore PC2 anbeing inability associated to discriminate with the bitter, me- among the samples. In particular, L. brevis (Bottle 1 and 2), P. parvulus (Bottle 2), and the tallic,sealed and control umami had substantial sensors. overlap.Similar Theresults PCA describedwere observed 99.9% of for the variation,Day 0 and with 7, with the e- tonguePC1 being having associated various with levels the response of negative to the salty DI. sensor and PC2 being associated with the bitter, metallic, and umami sensors. Similar results were observed for Day 0 and 7, with the e-tongue having various levels of negative DI.

Figure 2. Principal component analysis of Merlot wine samples on Day 14 of storage at 23 ◦C as analyzed by the electronic Figuretongue. 2. Principal Samples component are labeled with analysis names of as Merlot described wine in Table samples1. The discriminationon Day 14 of index storage was at− 16.23 °C as analyzed by the electronic tongue. Samples are labeled with names as described in Table 1. The discrimination index was −16. Starting at Day 21, the e-tongue discriminated among the different inoculated wine samples with a strong discrimination index of 91 (Figure3). Most of the variation among the samples was described by PC1 and was associated with the sweet sensor, describing 88.4% of the variation. Samples that were more similar to each other were placed closer to each other on the PCA, therefore the replicate bottles that were tested should be in similar areas on the PCA. Unexpectedly, the replicate bottles for the Brettanomyces samples were not closely related to each other on the PCA, displaying an almost contrasting relationship along PC1. This variation may have been caused by slightly varying amounts of available Beverages 2021, 7, x FOR PEER REVIEW 7 of 14

Starting at Day 21, the e-tongue discriminated among the different inoculated wine samples with a strong discrimination index of 91 (Figure 3). Most of the variation among the samples was described by PC1 and was associated with the sweet sensor, describing 88.4% of the variation. Samples that were more similar to each other were placed closer to each other on the PCA, therefore the replicate bottles that were tested should be in similar areas on the PCA. Unexpectedly, the replicate bottles for the Brettanomyces samples were not closely related to each other on the PCA, displaying an almost contrasting relationship Beverages 2021, 7, 52 along PC1. This variation may have been caused by slightly varying amounts7 of 14 of available oxygen in the individual wine bottles, or by the natural variation found in Brettanomyces growth [13,45]. However, as the storage continued, the replicates were more closely re- oxygen in the individual wine bottles, or by the natural variation found in Brettanomyces lated. growth [13,45]. However, as the storage continued, the replicates were more closely related.

Figure 3. Principal component analysis of Merlot wine samples on Day 21 of storage at 23 ◦C as analyzed by the electronic Figure 3. tongue.Principal Samples component are labeled analysis with names of Merlot as described wine in samples Table1. The on discriminationDay 21 of storage index wasat 23 91. °C as analyzed by the electronic tongue. Samples are labeled with namesThe abilityas describ to discriminateed in Table among 1. The thediscrimination different inoculated index was wines 91. after Day 21 of storage continued for the rest of the testing period, with discrimination indices exceeding 80%. TheseThe ability results indicateto discriminate that at Day among 21, the the spoilage differe microorganismsnt inoculated produced wines after sufﬁcient Day 21 of storage continuedmetabolites tofor signiﬁcantly the rest of affect the testing the sensory peri characteristicsod, with discrimination of the wines. Previous indices studies exceeding 80%. Thesehave results found thatindicate the e-tongue that at is ableDay to 21, detect the subthreshold spoilage microorganisms concentrations of varying produced chem- sufficient icals, including 4-ethylcatechol in Merlot wines, and spicy compounds in water [29,31,35]. metabolitesTherefore, to the significantly sensory changes affect in the the inoculated sensory wines characteristics at Day 21 are of likely the wines. to be present Previous at studies havehuman found threshold that the or subthresholde-tongue is concentrations. able to detect subthreshold concentrations of varying chemicals,The including ability to discriminate 4-ethylcatechol among samplesin Merlot based wines, on by-products and spicy produced compounds by com- in water [29,31,35].mon spoilage Therefore, microorganisms the sensory using changes rapid analytical in the instrumentsinoculated was wines also foundat Day in other21 are likely to be presentstudies [at29 ,human38,47]. A threshold previous study or subthreshold found that the electronic concentrations. nose (e-nose), another rapid analytical instrument, detected differences in acetic acid perception caused by natural spoilageThe ability at a single to discriminate timepoint [47 among]. A similar samples study usedbased a combination on by-products of e-tongue produced and by com- mone-nose spoilage methods microorganisms to evaluate wines us exposeding rapid to the analytical air at multiple instruments timepoints, was reporting also found that in other studiesthese [29,38,47]. methods could A previous discriminate study among found samples that from the different electronic timepoints nose (e-nose), [48]. However, another rapid analyticalin both ofinstrument, the previously detected mentioned differences studies, complementary in acetic acid sensory perception and microbial caused work by natural were not undertaken. spoilageAdditionally, at a single timepoint the difference [47]. from A the similar control study to the individualused a combination wine spoilage of microor- e-tongue and e- noseganisms methods was to also evaluate evaluated. wines When exposed compared to to the the controlair at multiple wine at Day timepoints, 14, the e-tongue reporting that theseresults methods for P. parvuluscould discriminateand L. brevis showed among low samp DI, withles from overlapping different occurring timepoints with the [48]. How- ever,control. in both The of discrimination the previously index mentioned for A. pasteurianus studies,was complementary also negative, indicating sensory an inabil- and microbial ity to discriminate among the samples, but the overlap was observed among the replicates work were not undertaken. instead of with the control. However, the e-tongue discriminated among the Brettanomyces strains.Additionally, After Day the 21, thedifference e-tongue differentiatedfrom the contro amongl to the the spoilage individual microorganisms wine spoilage and micro- organismsthe control was across also allevaluated. the microorganisms When compared tested. As to previously the control discussed, wine at the Day variation 14, the e-tongue resultsbetween for P. replicates parvulus may and be L. caused brevis by showed environmental low DI, factors, with asoverlapping well as by the occurring natural with the growth variation between bottles [13,45].

Beverages 2021, 7, x FOR PEER REVIEW 8 of 14

control. The discrimination index for A. pasteurianus was also negative, indicating an inability to discriminate among the samples, but the overlap was observed among the replicates instead of with the control. However, the e-tongue discriminated among the Brettan- omyces strains. After Day 21, the e-tongue differentiated among the spoilage microorganisms and the control across all the microorganisms tested. As previously discussed, the variation between replicates may be caused by environmental factors, as well as by the natural growth variation between bottles [13,45]. Beverages 2021, 7, 52 8 of 14 3.3. Flash Profiling At each the seven time points, GPA was completed on the FP data. To determine when3.3. the Flash assessors Profiling could detect differences among the inoculated wine samples, AHC was completedAt each the on seven the GPA time points,data. The GPA first was completedinstance of on clustering the FP data. was To determineobserved at Day 28 andwhen three the clusters assessors were could identified detect differences (Figure among4). Cluster the inoculated one included wine samples, Brettanomyces AHC I1A (Bot- was completed on the GPA data. The first instance of clustering was observed at Day tle 2),28 andBrettanomyces three clusters F3 were (Bottle identified 1), sealed (Figure control,4). Cluster partially one included unscrewedBrettanomyces control,I1A and Aceto- bacter(Bottle pasteurianus 2), Brettanomyces (BottleF3 1). (Bottle Cluster 1), sealedtwo included control, partially Pediococcus unscrewed parvulus control, (Bottle and 1), Lactoba- cillusAcetobacter brevis (Bottle pasteurianus 1), and(Bottle Acetobacter 1). Cluster Pasteurianus two included (BottlePediococcus 2). Cluster parvulus three(Bottle included 1), Pedio- coccusLactobacillus parvulus brevis (Bottle(Bottle 2), 1), Lactobacillus and Acetobacter brevis Pasteurianus (Bottle(Bottle 2), Brettanomyces 2). Cluster three I1A included (Bottle 1), and Pediococcus parvulus (Bottle 2), Lactobacillus brevis (Bottle 2), Brettanomyces I1A (Bottle 1), Brettanomyces F3 (Bottle 2). and Brettanomyces F3 (Bottle 2).

Figure 4. Dendrogram of the agglomerative hierarchical cluster analysis of the Flash Proﬁling data generated through

Generalized Procrustes Analysis at Day 28 of storage of the Merlot wine at 23 ◦C. The three clusters are distinguished by Figure 4.color Dendrogram and based on of similarity. the agglomerative Samples are labeled hierarchical with names cluster as described analysis in Table of1 the. The Flash broken Profiling line indicates data the generated point at through Generalizedwhich Procrustes truncation occurred. Analysis at Day 28 of storage of the Merlot wine at 23 °C. The three clusters are distinguished by color and based on similarity. Samples are labeled with names as described in Table 1. The broken line indicates the point These clusters were different than what would be expected as the replicate bottles were at which truncation occurred. placed into different clusters, with the exception of the uninoculated control and uninoculated partially unscrewed control. However, these clusters were similar to the groupings observedThese clusters with the e-tongue,were different indicating than that therewhat were would slight be differences expected between as the the replicate dif- bottles wereferent placed replicates. into different These differences clusters, diminished with th overe exception time as the of identified the uninoculated clusters were control and uninoculatedprimarily based partially on the inoculatedunscrewed wine control. microorganism However, at Day these 42 of storageclusters (Figure were5 ). similar At to the L. brevis P. parvulus groupingsDay 42, Cluster observed one contained with the e-tongue,(Bottle indicati 1 and 2),ng the unsealedthat there control were and slight differences be- (Bottle 2), Cluster two contained Brettanomyces I1A (Bottle 1) and Brettanomyces F3 (Bottle tween1), andtheCluster different three replicates. contained BrettanomycesThese differenceI1A (Bottles diminished 2), Brettanomyces over F3time (Bottle as the 2), identified clustersA. pasteurianus were primarily(Bottle 1 andbased 2), and on thethe sealed inoculated control. Differenceswine microorganism in the aromatic at profile Day of42 of storage (Figurethe controls 5). At wereDay observed,42, Cluster indicating one contained that some L. oxidation brevis (Bottle may have 1 occurred.and 2), the However, unsealed control andthe P. unsealedparvulus control (Bottle was 2), significantly Cluster two different contained from the BrettanomycesA. pasteurianus treatmentsI1A (Bottle (Bottle 1) and Brettan- 1 and 2), which were also left unsealed. Therefore, A. pasteurianus significantly influenced omycessensory F3 (Bottle characteristics 1), and beyond Cluster oxidation. three contained The influence Brettanomyces of time suggested I1A (Bottle that the 2), dif- Brettanomyces F3 (Bottleferences 2), among A. pasteurianus samples observed (Bottle at the 1 beginningand 2), and of the the storage sealed time control. may have Differences been in the aromaticcaused profile by different of the growth controls rates ofwere the microorganisms observed, indicating in the replicate that bottles.some oxidation However, may have occurred.over time, However, the broader the effect unsealed of the spoilage control microorganisms was significantly was increasingly different based from on the the A. pasteuri- identity of the microorganism. anus treatments (Bottle 1 and 2), which were also left unsealed. Therefore, A. pasteurianus significantly influenced sensory characteristics beyond oxidation. The influence of time suggested that the differences among samples observed at the beginning of the storage

Beverages 2021, 7, x FOR PEER REVIEW 9 of 14 Beverages 2021, 7, x FOR PEER REVIEW 9 of 14

time may have been caused by different growth rates of the microorganisms in the repli- Beverages 2021, 7, 52 time may have been caused by different growth rates of the microorganisms in the 9repli- of 14 catecate bottles.bottles. However,However, overover time,time, thethe broaderbroader effecteffect ofof thethe spoilagespoilage microorganismsmicroorganisms waswas increasinglyincreasingly basedbased onon thethe identityidentity ofof thethe microorganism.microorganism.

FigureFigure 5.5. DendrogramDendrogram ofof thethe agglomerativeagglomerative hierarchicalhierarchical clustercluster analysisanalysis ofof thethe FlashFlash ProﬁlingProfiling datadata Figure 5. Dendrogram of the agglomerative hierarchical cluster analysis of the Flash Profiling data generated through Generalized Procrustes Analysis at Day 42 of storage of the Merlot wine. The generated through Generalized Procrustes Analysis at Day 42 ofof storagestorage ofof thethe MerlotMerlot wine.wine. TheThe three clusters are distinguished by color and based on similarity. Samples are labeled with names three clustersclusters areare distinguisheddistinguished by by color color and and based based on on similarity. similarity. Samples Samples are are labeled labeled with with names names as as described in Table 1. The broken line indicates the point at which truncation occurred. describedas described in Tablein Table1. The 1. The broken broken line line indicates indicates the the point point at which at which truncation truncation occurred. occurred.

TheThe FPFP datadata atat DayDay 2828 werewere analyzedanalyzed usingusinusingg GPAGPA tototo visualizevisualize thethethe assessorassessor termterm usageusage inin aa consensusconsensusconsensus mapmapmap (Figure(Figure(Figure6 6).6).). Each EachEach assessor assessorassessor generated gegeneratednerated three threethree to toto seven sevenseven terms termsterms at atat each eacheach time timetime point,point, with with thethe the overalloverall overall numbernumber number ofof of termsterms terms usedused used varyingvarying varying amongamong among thethe thetimetime time points.points. points. TheThe FPFP The map-map- FP mappingpingping explainedexplained explained 72.3%72.3% 72.3% ofof thethe of variation thevariation variation amongamong among thethe thesamples.samples. samples. PC1PC1 PC1 describeddescribed described 46.3%46.3% 46.3% ofof thethe of thevariationvariation variation andand and waswas was describeddescribed described asas thethe as relationshrelationsh the relationshipipip betweenbetween between sweaty,sweaty, sweaty, lactic,lactic, lactic, smokey,smokey, smokey, andand spicyspicy and Brettanomyces spicycontrastedcontrasted contrasted toto floral,floral, to ﬂoral,dusty,dusty, dusty,andand baba andrnyardrnyard barnyard notes.notes. BothBoth notes. replicatesreplicates Both replicates ofof BrettanomycesBrettanomyces of I1AI1A andand I1A and F3 were located in the same quadrant and were described as smoky, spicy, ﬂoral, F3F3 werewere locatedlocated inin thethe samesame quadrantquadrant andand wewerere describeddescribed asas smoky,smoky, spicy,spicy, floral,floral, chemi-chemi- chemical, and cherry. These sensory characteristics are often associated with a Brettanomyces cal,cal, andand cherry.cherry. TheseThese sensorysensory characteristicscharacteristics areare oftenoften associatedassociated withwith aa BrettanomycesBrettanomyces infection, especially at the earlier stages when the metabolites are at concentrations that infection,infection, especiallyespecially atat thethe earlierearlier stagesstages whwhenen thethe metabolitesmetabolites areare atat concentrationsconcentrations thatthat may contribute to wine complexity [41]. maymay contributecontribute toto winewine complexitycomplexity [41].[41].

Figure 6. Generalized Procrustes Analysis of the Flash Proﬁle data taken at Day 28 of 23 ◦C storage of the inoculated Merlot wine samples. Wine sample location is based on consensus mapping. Samples are abbreviated based on cluster and are labeled with names as described in Table1. Displayed are the aroma terms used to describe the wines. Sensory terms are presented in a unique color for each assessor.

Beverages 2021, 7, x FOR PEER REVIEW 10 of 14

Figure 6. Generalized Procrustes Analysis of the Flash Profile data taken at Day 28 of 23 °C storage of the inoculated Merlot wine samples. Wine sample location is based on consensus mapping. Sam- ples are abbreviated based on cluster and are labeled with names as described in Table 1. Displayed Beverages 2021, 7, 52 are the aroma terms used to describe the wines. Sensory terms are presented in a unique color10 of for 14 each assessor.

The partially unscrewed and sealed uninoculated control were closely related to each The partially unscrewed and sealed uninoculated control were closely related to each other and were described by ethanol, dusty, and caramelized aromas (Figure 6). The other and were described by ethanol, dusty, and caramelized aromas (Figure6). The groupings on the consensus map may not directlydirectly correlate to thethe clustersclusters determineddetermined from AHC due to the inclusioninclusion ofof allall thethe differentdifferent principalprincipal components.components. TheThe consensusconsensus mapping determined the ability of the experiencedexperienced assessors to discriminate among the samples, as well as providing a sensory proﬁleprofile ofof thethe inoculatedinoculated wines.wines. Based on thesethese results, the assessors differentiated among th thee aromas of the inoculated wine samples at Day 28. These results suggest that that the storage time at which these changes occurred would be between Day 21 and Day 28. The consensus consensus map map visualized visualized the the assessors’ assessors’ term term usage usage at Day at Day 42 from 42 from the FP the data FP (Figuredata (Figure 7). The7). FP The mapping FP mapping explained explained 68.1% 68.1% of the of variation the variation among among the samples. the samples. PC1 describedPC1 described 42.3% 42.3% of the of variation the variation and was and desc wasribed described as the asrelationship the relationship between between fruity, cherry,fruity, cherry,and ethanol and ethanol contrasted contrasted to leather to spicy, leather and spicy, barnyard. and barnyard. Both replicates Both replicates of Brettano- of mycesBrettanomyces I1A and I1AF3 were and in F3 the were bottom in the portion bottom of the portion consensus of the map consensus and were map described and were as honey,described butterscotch, as honey, butterscotch, and cherry. andThe cherry.P. parvulus, The P. L. parvulus, bacillus, L.and bacillus, A. pasteurianusand A. pasteurianus (Bottle 1) were(Bottle clustered 1) were clusteredtogether and together were and described were described by notes bysuch notes as lactic, such as metallic, lactic, metallic, and horsey. and Betweenhorsey. Between Day 28 Dayand 28Day and 42, Day there 42, was there a decrease was a decrease in cited in sensory cited sensory termsterms associated associated with withthe Brettanomyces the Brettanomyces samples.samples. Previous Previous studies studies have found have found that as that the asBrettanomyces the Brettanomyces by-prod-by- uctsproducts (particularly (particularly ethyl-phenols) ethyl-phenols) increase, increase, the perception the perception of fruity of fruity and andvarietal varietal aromas aromas decrease,decrease, even even when when these these by-products by-products are are presen presentt at at sub-threshold sub-threshold concentrations concentrations [49,50]. [49,50]. This effect may explain the resultsresults observed in this study. Similar to whatwhat waswas observedobserved from AHC, AHC, most most of of the the microorganism microorganism bottle bottle replicates replicates were were in similar in similar areas areas on the on con- the sensusconsensus map. map. While While the Brettanomyces the Brettanomyces samplessamples were were not all not present all present in the in same the same cluster, cluster, the samplesthe samples were were characterized characterized by similar by similar terms. terms. This This result result suggests suggests that that clustering clustering may may be basedbe based on the on theother other principal principal components components not displayed not displayed on the on consensus the consensus map but map other- but wiseotherwise included included in the in AHC. the AHC. These These results results were were also alsoobserved observed at the at Day the Day28 time 28 time point. point.

◦ Figure 7. Generalized Procrustes AnalysisAnalysis of of the the Flash Flash Proﬁle Profile data data taken taken at at Day Day 42 42 of of 23 23C °C storage storage of ofthe the inoculated inoculated Merlot Merlot wine wine samples. samples. Wine Wine sample sample location location is based is based on consensuson consensus mapping. mapping. Samples Sam- plesare abbreviatedare abbreviated based based on clusteron cluster and and are are labeled labele withd with names names as as described described in in Table Table1. 1. Displayed Displayed are the aroma terms used to describe the wines. Sensory terms are presented in a unique color for each assessor.

In addition to utilizing the FP data for assessor consensus mapping, the term usage

was visualized, with terms separated as being either spoilage-related or non-spoilage related (Figure8). From Day 0 to Day 35, the assessors used non-spoilage related terms at signiﬁcantly higher frequencies (p < 0.0001). However, at Day 42 of storage, the use Beverages 2021, 7, x FOR PEER REVIEW 11 of 14

are the aroma terms used to describe the wines. Sensory terms are presented in a unique color for each assessor.

In addition to utilizing the FP data for assessor consensus mapping, the term usage Beverages 2021, 7, 52 was visualized, with terms separated as being either spoilage-related or non-spoilage11 ofre- 14 lated (Figure 8). From Day 0 to Day 35, the assessors used non-spoilage related terms at significantly higher frequencies (p < 0.0001). However, at Day 42 of storage, the use of traditionalof traditional spoilage spoilage terms terms was was significantly significantly higher higher than than the the use use of ofnon-spoilage non-spoilage terms, terms, indicatingindicating a ashift shift in in sensory sensory profile. profile. The The percentage percentage of of traditional traditional spoilage spoilage terms, terms, such such as as barnyardbarnyard and and medicinal, medicinal, compared compared to to the the overal overalll term term usage, usage, was was relatively relatively consistent consistent for for thethe first first 35 35 days days of of storage storage (p (p >> 0.05). 0.05). Day Day 42 42 was was the the only only time time point point that that had had a asignificant significant differencedifference in in traditional traditional spoilage spoilage term usage based onon storagestorage (p(p< <0.05). 0.05). AtAt Day Day 42, 42, 58% 58% of ofthe the terms terms utilized utilized to to differentiate differentiate the the inoculated inoculated wine wine samples samples were were spoilage spoilage related, related, which whichwas significantly was significantly higher higher (15%) than(15%) any than other any time other point. time The point. sharp The increase sharp inincrease traditional in traditionalspoilage terms spoilage indicated terms thatindicated between that Day between 35 and Day Day 35 42, and the Day inoculated 42, the inoculated wines reached wines the reachedpoint at the which point they at hadwhich a significant they had increasea significant in negative increase sensory in negative characteristics, sensory character- suggesting istics,wine suggesting spoilage. wine spoilage.

FigureFigure 8. 8. TheThe percentage percentage of of spoilage spoilage and and non-spoilage non-spoilage related related terms terms utilized utilized by by the the experienced experienced assessorsassessors at ateach each time time point point evaluations evaluations from from Day Day 0 to 0 Day to Day 42. Evaluations 42. Evaluations were were completed completed on the on spoilagethe spoilage organism-inoculated organism-inoculated Merlot Merlot wines. wines. Traditional Traditional spoilage spoilage terms terms were werearomas aromas that were that were associatedassociated with with negative negative sensory sensory connotations connotations based based on on the the BrettanomycesBrettanomyces aromaaroma wheel. wheel. A A * indicates * indicates significant differences (p < 0.001) between the usage of spoilage terms and non-spoilage terms. signiﬁcant differences (p < 0.001) between the usage of spoilage terms and non-spoilage terms.

3.4.3.4. Study Study Limitations Limitations InIn this this study, study, only only the the aromatic aromatic profile profile of of the the wines wines was was evaluated evaluated by by the the assessors. assessors. TheseThese aromatic aromatic changes changes are are likely likely to to be be insu insufficientfficient for for a afull full explanation explanation of of the the sensory sensory changeschanges in in the the wine wine conferred conferred by thethe spoilagespoilage microorganisms. microorganisms. Similarly, Similarly, this this study study did did not notevaluate evaluate the the different different enzymatic enzymatic reactions reactions that that can can occur occur during during aging, aging, from from natural natural and andmicrobiological microbiological sources, sources, which which have have been been shown shown to influence to influence the sensory the sensory characteristics character- of isticsthe wine.of the Another wine. Another limitation limitation was the was sampling the sampling schedule schedule and the frequencyand the frequency of sampling. of sampling.In the present In the study, present wines study, were wines sampled were weekly.sampled Moreweekly. frequent More samplingfrequent sampling would be wouldneeded be to needed determine to determine exact times exact at which times sensoryat which changes sensory occurred. changes Evenoccurred. though Even this thoughstudy usedthis study weekly used sampling, weekly sampling, it does provide it does information provide information regarding regarding sampling sampling frequency frequencyfor future for studies. future studies. Additionally,Additionally, this this study study only only utilized utilized a arapid rapid sensory sensory method method instead instead of of traditional traditional descriptivedescriptive analysis analysis which which may may have have provided provided a amore more robust robust examination examination of of the the sensory sensory changeschanges occurring occurring over over storage. storage. Future studies should also explore additional experimental conditions to expand generalizability of the results. In the present study, only one wine varietal and storage temperature were evaluated. Given that different spoilage microorganisms have different growth characteristics and metabolism, a greater range of storage temperatures would provide this additional information. The present study used Merlot, but future studies should explore other wine varietals to see if the relationships in the present study can be extended. Beverages 2021, 7, 52 12 of 14

4. Conclusions The ability of the e-tongue and FP to discriminate and detect wine faults caused by spoilage microorganisms was determined in Merlot wine. The e-tongue differentiated among the different wine faults starting at Day 21 of storage at 23 ◦C. From FP, assessors were able to start discriminating among the faulted wines starting at Day 28; however, a significant increase in traditional spoilage terms was not observed until Day 42. These results indicated that the e-tongue discriminated among faulted wines based on their non-volatile profile before the experienced assessors detected sensory changes in the volatile profile. These results suggest that the e-tongue is a useful tool for early detection of wine faults and may provide winemakers with an opportunity for remediation. The application of these novel techniques (e-tongue partnered with FP) may be the key to limiting financial losses associated with wine faults by early detection and remediation.

Author Contributions: Conceptualization, V.D.P., C.G.E., and C.F.R.; methodology, V.D.P., T.C.-B., C.G.E., and C.F.R.; formal analysis, V.D.P. and C.D.; investigation, V.D.P., and T.C.-B.; writing-original draft preparation, V.D.P., and C.F.R.; writing-review and editing, V.D.P., C.D., C.G.E., T.C.-B., and C.F.R.; supervision, C.F.R.; funding acquisition, C.G.E. and C.F.R. All authors have read and agreed to the published version of the manuscript. Funding: These authors acknowledge the financial support of the Northwest Center for Small Fruit Research. This work was partially supported by the USDA National Institute of Food and Agriculture, Hatch project with Accession# 1016366. Institutional Review Board Statement: The study was conducted according to the guidelines of the Declaration of Helsinki, and approved by the Institutional Review Board of Washington State University (IRB# 17595-001; approved April 2019). Informed Consent Statement: Informed consent was obtained from all subjects prior to their participation in the study. Data Availability Statement: Data available on reasonable request due to restrictions e.g., privacy. Conflicts of Interest: The authors declare no conflict of interest. The funders had no role in the design of the study; in the collection, analyses, or interpretation of data; in the writing of the manuscript, or in the decision to publish the results.

References 1. AWRI. Wine Faults, Taints and Flavours. Australian Wine Research Institute. Available online: https://www.awri.com.au/ industry_support/winemaking_resources/sensory_assessment/recognition-of-wine-faults-and-taints/ (accessed on 8 July 2020). 2. Alcohol and Tobacco Tax and Trade Bureau (TTB). TTB: TAD: Wine-Common Audit Issues. TTBGov-Wine-Common Audit Issues. Available online: https://www.ttb.gov/wine/wine-common-audit-issues (accessed on 19 May 2015). 3. Bartowsky, E.J. Bacterial spoilage of wine and approaches to minimize it. Lett. Appl. Microbiol. 2009, 48, 149–156. [CrossRef] 4. Fugelsang, K.C.; Edwards, C.G. Wine Microbiology: Practical Applications and Procedures; Springer: New York, NY, USA, 2006. 5. Suárez, R.; Suárez-Lepe, J.A.; Morata, A.; Calderón, F. The production of ethylphenols in wine by yeasts of the genera Bret- tanomyces and Dekkera: A review. Food Chem. 2007, 102, 10–21. [CrossRef] 6. Malfeito-Ferreira, M.; Barata, A.; Loureiro, V. Wine spoilage by fungal metabolites. In Wine Chemistry and Biochemistry; Polo, C., Moreno-Arribas, M.V., Eds.; Springer: New York, NY, USA, 2009; pp. 615–645. 7. Schumaker, M.R.; Chandr, M.; Malfeito-Ferreira, M.; Ross, C.F. Influence of Brettanomyces ethylphenols on red wine aroma evaluated by consumers in the United States and Portugal. Food Res. Int. 2017, 100, 161–167. [CrossRef] 8. Davis, E.R.; Wibowo, D.; Fleet, G.H.; Lee, T.H. Properties of wine lactic acid bacteria: Their potential enological significance. AJEV 1988, 39, 137–142. 9. Lonvaud-Funel, A.; Guilloux, Y.; Joyeux, A. Isolation of a DNA probe for identification of glucan-producing Pediococcus damnosus in wines. J. Appl. Bacteriol. 1993, 74, 41–47. [CrossRef] 10. Sponholz, W.R. Wine spoilage by microorganisms. In Wine Microbiology and Biotechnology; Fleet, G.H., Ed.; Harwood Academic Publishers: New York, NY, USA, 1993; pp. 395–420. 11. Liu, S.Q.; Pilone, G.J. A review: Arginine metabolism in wine lactic acid bacteria and its practical significance. J. Appl. Microbiol. 1998, 84, 315–327. [CrossRef] 12. Lonvaud-Funel, A. Lactic acid bacteria in the quality improvement and depreciation of wine. Antonie Leeuwenhoek 1999, 76, 317–331. [CrossRef][PubMed] Beverages 2021, 7, 52 13 of 14

13. Bartowsky, E.J.; Xia, D.; Gibson, R.L.; Feet, G.H.; Henschke, P.A. Spoilage of bottled red wine by acetic acid bacteria. Lett. Appl. Microbiol. 2003, 36, 307–314. [CrossRef] 14. Francis, I.L.; Newton, J.L. Determining wine aroma from compositional data. Aust. J. Grape Wine 2005, 11, 114–126. [CrossRef] 15. Chatonnet, P.; Dubourdie, D.; Boidron, J.; Pons, M. The origins of ethylphenols in wines. J. Sci. Food Agric. 1992, 60, 165–178. [CrossRef] 16. Loureiro, V.; Malfeito-Ferreira, M. Spoilage yeasts in the wine industry. Int. J. Food Microbiol. 2003, 86, 23–50. [CrossRef] 17. Bartowsky, E.J.; Henschke, P.A. The ‘buttery’ attribute of wine- diacetyl- desirability, spoilage and beyond. Int. J. Food Microbiol. 2004, 96, 235–252. [CrossRef] 18. Liu, J.; Gronbeck, M.S.; Di Monaco, R.; Giacalone, D.; Bredie, W.L.P. Performance of flash profile and napping with and without training for describing small sensory differences in a model wine. Food Qual. Prefer. 2016, 48, 41–49. [CrossRef] 19. Sieffermann, J.M. Le profil Flash: Un Outil Rapide et Innovant d’e’ Valuation Sensorielle Descriptive [Conference Paper Presentation]; The L’innovation: De l’ide’eau succe’s-Douzie’mes rencontres AGORAL: Paris, France, 2000. 20. Williams, A.A.; Langron, S.P. The use of free-choice profiling for the evaluation of commercial ports. J. Sci. Food Agric. 1984, 35, 558–568. [CrossRef] 21. Dairo, V.; Sieffermann, J.M. A comparison of 14 jams characterized by conventional profile and a quick original method, the flash profile. J. Food Sci. 2002, 67, 826–834. [CrossRef] 22. Delarue, J. Flash profile, its evolution and uses in sensory and consumer science. In Rapid Sensory Profiling Techniques: Applications in New Product Development and Consumer Research; Delarue, J., Lawlor, J.B., Rogeaux, M., Eds.; Elsevier: New York, NY, USA, 2015; pp. 119–148. 23. Liu, J.; Bredie, W.L.O.; Sherman, E.; Harbertson, J.F.; Heymann, H. Comparison of rapid descriptive sensory methodologies: Free-choice profiling, flash profile and modified flash profile. Food Res. Int. 2018, 106, 892–900. [CrossRef] 24. Albert, A.; Varela, P.; Salvador, A.; Hough, G.; Fiszman, S. Overcoming the issues in the sensory description of hot served foods with a complex texture. Application of QDA, flash profiling and projective mapping using panels with different degrees of training. Food Qual. Prefer. 2011, 22, 463–473. [CrossRef] 25. Legin, A.; Rudnitskaya, A.; Clapham, D.; Seleznev, B.; Lord, K.; Vlasov, Y. Electronic tongue for pharmaceutical analytics: Quantification of tastes and masking effects. Anal. Bioanal. Chem. 2004, 380, 36–45. [CrossRef][PubMed] 26. Rudinitskaya, A.; Rocha, S.M.; Legin, A.; Pereira, V.; Marques, J.C. Evaluation of the feasibility of the electronic tongue as a rapid analytical tool for wine age prediction and quantification of the organic acids and phenolic compounds. The case-study of Madeira wine. Anal. Chim. Acta 2010, 662, 82–89. [CrossRef][PubMed] 27. Campos, I.; Bataller, R.; Armero, R.; Gandia, J.M.; Soto, J.; Martinez-Manez, R.; Gil-Sanchez, L. Monitoring grape ripeness using a voltametric electronic tongue. Food Res. Int. 2013, 54, 1369–1375. [CrossRef] 28. Diako, C.; McMahon, K.; Mattinson, S.; Evans, M.; Ross, C.F. Alcohol, tannins, and mannoprotein and their interactions influence the sensory properties of selected commercial merlot wines: A preliminary study. J. Food Sci. 2016, 81, S2039–S2048. [CrossRef] 29. Diako, C.; Vixie, B.; Weller, K.M.; Dycus, D.A.; Ross, C.F. Determination of 4-ethylcatechol in a merlot wine using sensory evaluation and the electronic tongue. Int. J. Food Sci. Technol. 2017, 52, 2489–2496. [CrossRef] 30. Newman, J.; O’Riordan, D.; Jacquier, J.C.; O’Sullivan, M. Masking of bitterness in dairy protein hydrolysates: Comparison of an electronic tongue and a trained sensory panel as means of directing the masking strategy. LWT 2015, 63, 751–757. [CrossRef] 31. Lipkowitz, J.B.; Ross, C.F.; Diako, C.; Smith, D.M. Discriminating aging and protein-to-fat ration in cheddar cheese using sensory analysis and a potentiometric electronic tongue. J. Dairy Sci. 2018, 101, 1990–2004. [CrossRef][PubMed] 32. Walsh, E.A.; Diako, C.; Smith, D.M.; Ross, C.F. Influence of storage time and elevated ripening temperature on the chemical and sensory properties of white cheddar cheese. J. Food Sci. 2020, 85, 268–278. [CrossRef][PubMed] 33. Dyminski, D.S.; Paterno, L.G.; Takeda, H.H.; Bolini, H.; Mattoso, L.H.C.; Candido, L.M.B. Correlation between human and electronic tongues responses on the analysis of commercial sweeteners. Sens. Lett. 2006, 4, 403–408. [CrossRef] 34. Waldrop, M.E.; Ross, C.F. Sweetener blend optimization using mixture design methodology and the electronic tongue. J. Food Sci. 2014, 79, S1782–S1794. [CrossRef][PubMed] 35. Paup, V.D.; Barnett, S.M.; Diako, C.; Ross, C.F. Detection of spicy compounds using the electronic tongue. J. Food Sci. 2019, 84, 2619–2627. [CrossRef] 36. Schlossareck, C.; Ross, C.F. Electronic tongue and consumer sensory evaluation of spicy paneer cheese. J. Food Sci. 2019, 8, 1563–1569. [CrossRef] 37. Nery, E.W.; Kubota, L.T. Integrated, paper-based potentiometric electronic tongue for the analysis of beer and wine. Anal. Chim. Acta 2016, 918, 60–68. [CrossRef] 38. Lvova, L.; Yaroshenko, I.; Kirsanov, D.; Natale, C.D.; Paolesse, R.; Legin, A. Electronic tongue for brand uniformity control: A case study of Apulian red wines recognition and defects evaluation. Sensors 2018, 18, 2584. [CrossRef][PubMed] 39. Jensen, S.L.; Umiker, N.L.; Arneborg, N.; Edwards, C.G. Identification and characterization of Dekkera bruxellensis, Candida pararugosa, and Pichia guilliermondii isolated from commercial red wines. Food Microbiol. 2009, 26, 915–921. [CrossRef] 40. Noble, A.C.; Arnold, R.A.; Buechsenstein, J.; Leach, E.J.; Schmidt, J.O.; Stern, P.M. Modification of a standardized system of wine aroma terminology. AJEV 1987, 38, 143–146. Available online: https://www.ajevonline.org/content/ajev/38/2/143.full.pdf (accessed on 12 February 2021). 41. Joseph, L.; Albino, E.; Bisson, L.F. Creation and use of a Brettanomyces aroma wheel. Catal. Discov. Pract. 2017, 1, 12–20. [CrossRef] Beverages 2021, 7, 52 14 of 14

42. Zuehlke, J.M.; Edwards, C.G. Impact of sulfur dioxide and temperature on Culturability and viability of Brettanomyces bruxellensis in wine. J. Food Prot. 2013, 76, 2024–2030. [CrossRef][PubMed] 43. du Toit, M.; Pretorius, I.S. Microbial spoilage and preservation of wine: Using weapons for nature’s own arsenal. S. Afr. J. Enol. Vitic. 2000, 21, 74–96. [CrossRef] 44. Loureiro, V.; Malfeito-Ferreira, M. Dekkera/Brettanomyces spp. In Food Spoilage Microorganisms; Blackburn, C.D.W., Ed.; Woodhead Publishing: Cambridge, UK, 2006; pp. 354–388. 45. Wedral, D.; Shewfelt, R.; Frank, J. The challenge of Brettanomyces in wine. LWT 2010, 43, 1474–1479. [CrossRef] 46. Jia, K.; Zhang, Y.; Li, Y. Systematic engineering of microorganisms to improve alcohol tolerance. Eng. Life Sci. 2010, 10, 422–429. [CrossRef] 47. Gamboa, J.C.R.; Albarracin, E.S.; Silva, A.J.; Lima, L.L.A.; Ferreira, T.A. Wine quality rapid detection using a compact electronic nose system: Application focused on spoilage thresholds by acetic acid. LWT 2019, 108, 377–384. [CrossRef] 48. Gil-Sanchez, L.; Soto, J.; Martinez-Manez, R.; Garcia-Breijo, E.; Ibanez, I.; Llobet, E. A novel humid electronic nose combined with an electronic tongue for assessing deterioration of wine. Sens. Actuators A Phys. 2011, 171, 152–158. [CrossRef] 49. Filipe-Ribeiro, L.; Cosme, F.; Nunes, F.M. Reducing the negative sensory impact of volatile phenols in red wine with different chitosans: Effect of structure on efﬁciency. Food Chem. 2018, 242, 591–600. [CrossRef][PubMed] 50. Tempere, S.; Schaaper, M.H.; Cuzange, E.; de Lescar, R.; de Revel, G.; Sicardd, G. The olfactory masking effect of ethylphenols: Characterization and elucidation of its origin. Food Qual. Prefer. 2016, 50, 135–144. [CrossRef]