Effect of Whole Cluster Fermentation on Phenolics in Cold-Hardy Hybrid Wines

Generally, red wines made from cold-hardy hybrid grapes have much lower tannin content than Vitis vinifera wines and they are often perceived as being of lower quality. This study evaluated the chemistry and sensory perception of whole-cluster (WC) fermented “Frontenac” and “Marquette” red wines from Iowa (F-ISU, M-ISU) and “Marquette” red wines from Minnesota (M-UMN). Three conditions were compared: Control 0% WC (w/w), 25% WC (w/w), and 50% WC (w/w). The 50% WC wines had significantly higher iron-reactive phenolics, tannin, and anthocyanin contents than control wines. The color intensity in 50% WC wines was maintained after 4 months of aging. A tasting evaluation was conducted with the “Marquette” wines from both states and untrained panelists preferred 50% WC “Marquette” wines over control wines. This study showed promising results on the use of 50% (w/w) whole clusters as a pre-fermentative technique to improve “Marquette” and “Frontenac” red wine quality.


Introduction
Cold-hardy hybrid grape varieties are prominent in the wine industry of the Midwest region of the USA because of their resistance to pests, diseases, and the harsh cold winters and late spring frosts. Interspecific (i.e., hybrid) grape varieties are crosses of European Vitis vinifera species with North American Vitis spp. (e.g., V. riparia, V. labrusca, V. rotundafolia, and V. rupestris) (Read et al., 2020). "Frontenac" and "Marquette" are two of the most planted red hybrid varieties in the Midwest region, but the effect of different winemaking techniques on the chemistry of the wines is poorly understood.
Tannins and other flavonoids (e.g., anthocyanins) are important for quality in red wines as they contribute astringency, color stability, and protect against oxidation (Casassa & Harbertson, 2014). Previous studies have shown that V. vinifera red wines with higher phenolic and tannin content are rated as higher quality (Fanzone et al., 2012;Mercurio et al., 2010). Condensed tannins (proanthocyanidins) are oligomers and polymers of flavan-3-ols (e.g., (-)-epicatechin, (+)-catechin) found in grape seeds, skins, and stems and are the primary contributors to astringency in wines (Souquet et al., 2000;Sun et al., 2013). However, as reported by Springer and Sacks (2014), hybrid grapes contain, on average, 1.8-fold lower tannin content than V. vinifera varieties with even larger differences (5.5-fold) when comparing wines, which also indicates a difference in extractability. For example,  found the tannin content in Marquette wines to be below the limit of quantification (140 mg/L) by protein precipitation while Manns et al. (2013) found tannin content in their control Marquette wines ranging from 48-133 mg/L by HPLC-DAD. Nicolle et al. (2019) found tannin content in Frontenac wines to be low (< 100 mg/L) by HPLC with a fluorescence detector. Another report found tannin content on finished red wines to range based on the method of analysis with concentrations of 48-412 mg/L and 200-831 mg/L for Frontenac and Marquette, respectively (Watrelot, 2021). These concentrations are below averages reported for V. vinifera cultivars Pinot Noir (316-382 mg/L and 748 mg/L) and Cabernet Sauvignon (557-701 mg/L and 978 mg/L) measured by protein precipitation and HPLC-DAD, respectively (Harbertson et al., 2008;Watrelot, 2021).
Anthocyanins are extracted from the skins, and sometimes the flesh, of red grapes during the first days of alcoholic fermentation and contribute red color to the wine. In V. vinifera, anthocyanins are present as anthocyanin-3-Oglucosides whereas red hybrid varieties have a high proportion of anthocyanin-3,5-O-diglucosides (Kumar et al., 2022).  reported total monomeric anthocyanin content in Marquette wines to be 319-359 mg/L at bottling and 241-277 mg/L after 5 months of aging. This included ~ 51% malvidin-3,5-O-diglucoside; compared to ~ 15% malvidin-3-O-glucoside, the primary anthocyanin in V. vinifera varieties (Kumar et al., 2022). The formation of pigments between tannins and anthocyanins, induced by high tannin to anthocyanin ratios, in V. vinifera wines has been shown as a positive point for red wine color stability and for a better solubility and retention of tannins during aging (Kumar et al., 2022).
Various winemaking techniques have been used to maximize the extraction of phenolics and therefore manage wine quality with V. vinifera red grapes; however, limited research has been dedicated to understanding the chemistry of coldhardy hybrid grapes or ways to optimize the quality of resulting wines. Several winemaking practices are used to manage phenolic content and stabilize color in V. vinifera wines such as the use of the hot press, cold soak (Manns et al., 2013), extended maceration (Casassa et al., 2013), addition of exogenous tannin (Fredrickson et al., 2020;Harbertson et al., 2012), addition of stems (Casassa et al., 2019(Casassa et al., , 2021, or the use of whole clusters (WC) (Casassa et al., 2019(Casassa et al., , 2021Suriano et al., 2015). However, these techniques have varying effects on hybrids or have not yet been used. For example, hot press treatment significantly increased anthocyanin concentration in Marquette must but not in the final wine, while increasing the tannin concentration in both must and wine (Manns et al., 2013). Similarly, cold soak treatment increased extraction of anthocyanins in must from "Maréchal Foch," "Corot Noir," and "Marquette" but the final wines showed no difference. Tannin concentration was unaffected with cold soak treatment (Manns et al., 2013). The addition of exogenous tannin can be unpredictable due to the unknown purity of tannin products (Fredrickson et al., 2020) and often require the addition of more than recommended by the manufacturer at the risk of adverse sensory effects to V. vinifera wines (Harbertson et al., 2012). Further, Frederickson and colleagues (Fredrickson et al., 2020) found that the retention rate of tannin addition is lower in hybrid cvs. Maréchal Foch and Corot Noir than in V. vinifera cv. Cabernet Franc.
Previous studies evaluated the effect of whole clusters on V. vinifera cvs. Pinot Noir and Primitivo. Casassa et al. (2019) evaluated the chemical and sensory effects of 20% (w/w) WC on Pinot Noir and found no effect on tannin or polymeric pigments with WC treatment, most probably due to the low proportion of WC. Higher percentage of whole clusters have been used in Pinot noir, i.e., 50% and 100% WC and both WC conditions increased tannin content relative to the control proportional to the percentage of WC used (Casassa et al., 2021). Suriano et al. (2015) observed an increase in monomeric phenolic compounds and tannin content, as well as a decrease in anthocyanins content in Primitivo wines after using 25% and 50% WC. The additional tannins contributed by stems increased color stability after 12 months of aging compared to the control despite lower levels of anthocyanins.
As previously mentioned, whole cluster fermentation seems a promising technique to improve the extraction of phenolic compounds especially tannin in cold-hardy grape cultivars and therefore enhance wine quality. The objective of this study was to evaluate the effect of different proportions of whole clusters on tannin content, color stability, and tasting perception in Frontenac and Marquette red wines. To our knowledge, this is the first study of whole cluster fermentation using Frontenac or Marquette.

Plant Material
Interspecific grape cultivars Frontenac (F-ISU) and Marquette (M-ISU) were provided by Tassel Ridge Winery (Leighton, IA, USA). The M-ISU grapes were manually harvested from the Meadow Creek South and West vineyards on 30 Aug. 2021 at 6:00 AM, sprayed with 30 mg/L KMBS, and stored in a refrigerated truck until 1 Sept. 2021 when they were delivered to the Iowa State University (ISU) winery for processing. The F-ISU grapes were manually harvested from the Meadow Creak East vineyard on 17 1 3 Sept. 2021 and transported to the ISU winery for processing. Interspecific grape cultivar Marquette from Minnesota (M-UMN) was harvested from the University of Minnesota (UMN) Horticulture Research Center (74.4%, Excelsior, MN, USA) and Sovereign Estate Winery (25.6%, Waconia, MN, USA) and provided by UMN. The M-UMN grapes were manually harvested on 8-9 Sept. 2021 and transported to the grape breeding and enology project facility of UMN for processing following the same procedure as M-ISU and F-ISU.

Winemaking
The M-ISU and F-ISU wines were produced at the ISU winery while the M-UMN wines were produced at the UMN winery, following the same procedure, and transported to ISU after alcoholic fermentation, pressing, and malolactic fermentation (MLF). M-ISU, M-UMN, and F-ISU grapes were processed on 1 Sept. 2021, 10 Sept. 2021, and 17 Sept. 2021, respectively, according to the following protocol. Three treatments were prepared in triplicate for each variety and location: control (0% (w/w) whole cluster (WC), 20 kg crushed/destemmed berries per bucket), 25%WC (25% (w/w) whole cluster, 5 kg WC + 15 kg crushed/destemmed berries), 50%WC (50% (w/w) whole cluster: 10 kg WC + 10 kg crushed/destemmed berries). Whole clusters were placed into each bucket first and the crushed/destemmed berries were added on top. Lalvin ICV D254 yeast (Scott Laboratories, Petaluma, CA, USA) with GoFerm™ yeast rehydration nutrient (Scott Laboratories, Petaluma, CA, USA) was added according to supplier instructions. Punch-downs were carried out after inoculation and twice daily throughout alcoholic fermentation. Lalvin VP41 bacteria (Scott Laboratories, Petaluma, CA, USA) was co-inoculated 48 h after yeast addition. Fermentation temperature, degree Brix, and density (portable density meter DMA 35, Anton Paar, Ashland, VA, USA) were monitored every day for 7 days throughout fermentation after punch down. The wines were pressed after completion of alcoholic fermentation to dryness, after 6 to 8 days, into 11.4 L jugs. The jugs were kept at room temperature until completion of MLF by enzymatic assay for another 7 to 12 days. At this stage, the M-UMN wines were transported to ISU to finish the process. Tartaric acid was added to M-ISU wines to reduce the pH by 0.1 unit (22.5 g dissolved in 30 mL of each wine for each 11.4 L jug for M-ISU). The 11.4 L jugs were racked into 3.8 L jugs and 0.5 mg/L molecular SO 2 was added (as a 6.6% (w/v) solution KMBS) according to each wine's pH. Wines were flushed with argon following SO 2 addition and transferred to cold storage (between 4 and 7.8 °C) for cold stabilization for 4 weeks. The wines were bottled in 750 mL bottles, after adding free SO 2 as a target of 0.8 mg/L molecular SO 2 , flushed with argon, sealed with a cork, and transferred to a cellar (13 °C and 75% humidity) for aging for 4 months in the horizontal position. Samples of each wine were collected at crushing, with pooled samples, bottling for each replicate and treatment, and after 4 months of aging as a pooled of replicate for each treatment, for chemical analysis.

Basic Chemical Properties
The basic chemistry parameters including pH, titratable acidity (TA), degree Brix, color intensity and hue, and free and total SO 2 were measured in all samples from crushing to 4 months aging. The pH was measured using an Orion Star™ A211 Benchtop pH Meter (Thermo Fisher Scientific, Waltham, MA, USA). Titratable acidity was measured in 5 mL of must or wine with 0.1 N sodium hydroxide to an endpoint of pH 8.2 and expressed in tartaric acid equivalents. Degree Brix was measured with a digital refractometer RF153 (FLIR Commercial Systems Inc., Nashua, NH, USA). The color intensity and hue were calculated from absorbance values at three wavelengths (420 nm, 520 nm, and 620 nm) measured in a 1 mm quartz cuvette with a Genesys 150 UV-Vis Spectrophotometer (Thermo Fisher Scientific, Waltham, MA, USA).

Organic Acids, Residual Sugars, and Alcohols
Reversed-phase high-performance liquid chromatography (RP-HPLC, 1200 series, Agilent Technologies, Santa Clara, CA, USA) with a diode array detector (DAD) and refractive index detector (RID) was used to evaluate the concentrations of organic acids, sugars, and alcohols in the must and wine, as previously published . Bio-Rad Aminex HPX-87H and Bio-Rad fermentation monitoring columns (Bio-Rad, Hercules, CA, USA) with a H + guard cartridge were used. A 10 µl sample previously filtered at 0.45 µm using a PTFE filter was injected with a 5-mM sulfuric acid mobile phase at a flow rate of 0.65 mL/min for 35 min. Malic acid was detected at 210 nm with the DAD while all other acids, sugars, and alcohols were detected with the RID (cell temperature of 55 °C). A calibration curve of all compounds, including tartaric acid, malic acid, lactic acid, glucose, fructose, glycerol, and ethanol was used to calculate their concentrations.

Total Iron-Reactive Phenolics Content
The total iron-reactive phenolics (IRP) content in all samples was evaluated using the Harbertson-Adams Assay (Harbertson et al., 2002). Briefly, 75 µL must/wine was added to 800 µL of a buffer of SDS/TEA, and the absorbance at 510 nm was recorded after a 10 min incubation as "background." Then, 125 µL ferric chloride reagent was added and the absorbance at 510 nm was recorded after a 10 min incubation as "final." The absorbance values were evaluated against a blank of the same buffer and ferric chloride solutions without a sample. Total iron-reactive phenolics were expressed as (+)-catechin equivalents.

Tasting
A tasting evaluation was conducted only on M-ISU and M-UMN, during the 2022 Minnesota Grape Grower Association annual conference in Rochester, MN. During the 60-min session, participants received a brief presentation of whole-cluster winemaking and the effect of stem inclusion on phenolic content followed by a tasting session. Participants were asked to take a sip of wine, evaluate the sensory attributes on an anonymous paper ballot, and spit out the wine. There were three rounds of three wines each, for a total of 9 wines. The first round of wine was the three whole clusters conditions of M-UMN in a random order. The second round was the three whole cluster conditions of M-ISU in a random order. The third round was M-ISU 50%WC, M-UMN 50% WC, and M-UMN CTL in a random order. For each round, the participant went from left to right and evaluated the wines which had a random, 3-digit number assigned. The participants were asked to rate each wine from 1 to 10 for each attribute. The attributes were fruit intensity, red fruit, dark fruit, dried fruit/jam, vegetal, sweetness, acidity, bitterness, saltiness, body, texture, length, hue, and color intensity. After rating the wines, the participants were asked to indicate if they could detect a difference between the 3 wines of each round and which wine they preferred from each round. At the end of the tasting, participants were asked to answer socio-demographic questions: age, gender, race, and occupation. This sensory evaluation was approved by the Institutional Review Board of Iowa State University prior to being conducted (IRB ID: 21-449).

Statistical Analysis
Analysis of Variance (ANOVA) and Tukey's honestly significantly different (HSD) test (α = 0.05) were performed using JMP 15.2 software to detect significant differences between biological or analytical triplicate.

Basic Chemical Properties
Fermentation temperature and degree Brix curves showed good reproducibility between the three whole cluster (WC) conditions (CTL, 25%WC, and 50%WC) for F-ISU, M-ISU, and M-UMN (Fig. S1). The fermentation temperature increased up to 20.8 °C, 21.3 °C, and 22.4 °C for F-ISU, M-ISU, and M-UMN, respectively, after 4 days and decreased until the end of fermentation for all varieties (Fig. S1). The average degree Brix at harvest was 28.5, 26.1, and 25.7 for F-ISU, M-ISU, and M-UMN, respectively. Fermentation occurred for 6 to 8 days until dryness, ending with a degree Brix below 3.0. The F-ISU-CTL wine showed lower temperature and degree Brix than the WC wines for the last 2 days of fermentation suggesting the fermentation was slowed by the inclusion of whole clusters. This hypothesis is supported by the M-ISU fermentation as the degree Brix for the M-ISU-CTL wine was also lower than the WC wines on day five, but the M-UMN fermentation showed differently. Both M-UMN-WC wines had lower degree Brix than the M-UMN-CTL for the first 2 days of fermentation (F-ISU and M-ISU showed no differences here), but on days three and four, only M-UMN-25%WC had lower degree Brix than the M-UMN-CTL.
At crushing, the pH was 3.63, 3.41, and 2.91 for F-ISU, M-ISU, and M-UMN, respectively. The pH for all samples increased significantly from crushing to bottling to 4 months of aging with a higher increase from bottling to 4 months of aging than from crushing to bottling (Table 1). As expected, the TA had the opposite trend as pH with significant decreases from crushing to bottling for all samples and further significant decreases for each M-UMN condition after 4 months of aging. No significant differences between whole clusters conditions of pH and TA at crushing or bottling were observed. However, after 4 months of aging, the pH was significantly higher in F-ISU-50%WC (4.00) than F-ISU-25%WC (3.96) and F-ISU-CTL (3.96) wines, as well as higher in the M-UMN-CTL (3.67) wine than M-UMN-25%WC (3.65) and 50%WC (3.66) wines. Stems have been shown to increase the potassium content in wines (Hashizume et al., 1998) which can increase the precipitation of potassium bitartrate leading to changes in the pH and decreased titratable acidity of the wine (Waterhouse et al., 2016). However, this pH increase with the use of 50% WC was only observed in Frontenac wines, suggesting that other chemical parameters such as calcium content might also have an effect on wine pH (Hashizume et al., 1998). Higher pH has been shown to decrease anthocyanin extraction but increased the extractability of (+)-catechin and (-)-epicatechin by 28% and 20%, respectively (Forino  et al., 2020). Stem inclusion can also have a direct impact on reducing anthocyanin retention by adsorbing a proportion which is then removed at pressing (Spranger et al., 2004;Suriano et al., 2015). Hue increased significantly from crushing (0.5) to bottling (0.7) for each F-ISU condition but did not change significantly after 4 months of aging. The hue did not change significantly for any M-ISU wine over time while the hue for all M-UMN wines increased between each time point from 0.3 to 0.4 to 0.5-0.6 at crushing, bottling, and aging, respectively (Table 1). The increase in hue for F-ISU and M-UMN may be due to oxygen exposure during fermentation that caused more yellow-red pigments to form  while the M-ISU wines had a relatively high hue (0.9-1.0) and low color intensity (≤ 8.2), so oxygen exposure did not have an effect. The color intensity for the F-ISU wines increased over time, but the differences observed were not statistically significant. Color intensity for the M-ISU wines increased significantly from crushing (2.4) to bottling (5.6-6.1) and then remained statistically unchanged from bottling to 4 months of aging. The color intensity of the M-UMN wines increased from 3.9 to 17.6-20.2 from crushing to bottling and remained statistically unchanged after 4 months of aging (Table 1). Increased color intensity from crushing to bottling was expected because maceration on the skins during fermentation extracts anthocyanins with peak concentrations around the middle of fermentation (Casassa et al., 2013) and decreases with post-fermentation maceration (Scudamore-Smith et al. 1990, Harbertson et al. 2009). Overall, the M-ISU and M-UMN-WC wines had higher color intensity than the CTL wines which was consistent with the results of Suriano et al. (2015) in wines made from Vitis vinifera cv. Primitivo. Stem inclusion may better aerate the must leading to oxygen-stabilized color compared to completely destemmed must. However, Frontenac wines showed the opposite, where F-ISU-CTL had higher color intensity than the F-ISU-WC wines. This difference between the Frontenac and Marquette wines may be because the two varieties have a different anthocyanin profile. Both are rich in anthocyanin diglucosides but Frontenac wines tend to be richer in delphinidin-3,5-O-diglucoside while Marquette is richer in malvidin-3,5-O-diglucoside (Burtch et al., 2017). The hue was much higher and color intensity lower (i.e., more yellowness to redness, paler wines) for the M-ISU wines (hue ≥ 0.9, intensity ≤ 8.2) compared to the M-UMN wines (hue ≤ 0.6, intensity ≥ 17.6) at bottling. The difference in color between the Marquette wines from different states can be explained by a difference in starting pH and/or final ethanol content (3.41, 13.5-14.4 vol% and 2.91, 15.1-15.6 vol% for M-ISU and M-UMN, respectively, Table 1). Forino et al. (2020) found more anthocyanins extracted in wines at pH 3.2 than at pH 3.7. Additionally, Librán et al. (2013) observed more anthocyanins extracted from grape pomace at higher ethanol concentration (50-75 vol%) for acid pH conditions; however, Casassa et al. (2013) found that small differences (1-2 vol%) had no effect on the extraction of anthocyanin. This suggested that M-UMN had more color intensity because of lower pH while the difference in ethanol content had little effect on color. Other factors of terroir may also contribute such as temperature throughout the growing season with cooler areas leading to more anthocyanin accumulation during ripening (Downey et al., 2006) and higher berry temperature leading to increased content of anthocyanins (Scharfetter et al., 2019).
At crushing, tartaric acid content was 3.8, 3.3, and 7.8 g/L for F-ISU, M-ISU, and M-UMN, respectively. Tartaric acid content decreased from crushing to bottling for every sample as expected from potassium bitartrate precipitation during cold stabilization (Table 1). Between bottling and 4 months aging, tartaric acid content ranged from 2.7-3.6, 2.4-2.8, and 2.36-3.75 g/L for F-ISU, M-ISU, and M-UMN, respectively. These ranges were consistent with previously published values of 2.3-4.0 g/L for Frontenac and Marquette (Riesterer-Loper et al., 2019) and 3.0-4.9 tartaric for Marquette . Following fermentation, malic and acetic acids were below the limit of detection (LOD) for all samples while lactic acid was detected only after MLF (data not shown).
As expected, ethanol was not detected at crushing for all samples (Table 1). Ethanol content increased from bottling to 4 months aging in each of the M-ISU wines, but the difference was only significant in the M-ISU-25%WC where ethanol content increased from 14.0 to 14.3% (v/v). This was most likely due to the relatively small standard deviation compared to M-ISU-CTL and M-ISU-50% WC and not because of any practical difference. Ethanol content decreased in all M-UMN wines during the aging period, but these changes were not statistically significant. Glycerol content was minimally affected by the WC condition (Table S1).

Total Iron-Reactive Phenolic Content
Total iron-reactive phenolic content (IRP) increased from 623, 556, and 333 mg/L (+)-catechin equivalent (CE) at crushing to 1107, 882, and 1068 mg/L CE at bottling in F-ISU, M-ISU, and M-UMN control wines, respectively ( Fig. 1). At bottling, the IRP content of the F-ISU-25% WC (1134 mg/L CE) and F-ISU-50% WC (1246 mg/L CE) wines were higher than the F-ISU-CTL (1107 mg/L CE), but the difference was only significant between the F-ISU-50% WC and CTL. The IRP content of the M-ISU wines at bottling was not significantly different although an increase was observed from CTL (882 mg/L CE) to 25% WC (933 mg/L CE) to 50% WC (972 mg/L CE). The IRP content was not significantly different between the M-UMN wines at bottling. Similar to the results of Casassa et al. (2021), whole clusters contributed to increased phenolic content in Vitis vinifera cv. Pinot noir wines, however, they found greater increases than the current study with wine from the hybrid cv. Marquette. This was most likely due to the overall lower phenolic content of cold-hardy hybrid grape varietals. IRP content also increased in the M-ISU-25% WC and M-ISU-50% WC wines and decreased in the M-ISU-CTL wine after aging, but these changes were not significant. After 4 months of aging, the IRP content of the F-ISU-50% WC (1568 mg/L CE) wine was significantly higher than both the F-ISU-CTL (1377 mg/L CE) and F-ISU-25% WC (1430 mg/L CE) with no significant difference between the latter two conditions. The IRP content of the M-ISU-CTL wine decreased to 831 mg/L CE while the IRP content of the M-ISU-25% WC and M-ISU-50% WC conditions increased to 1101 mg/L CE and 1145 mg/L CE after aging, respectively, which were both significantly higher than the M-ISU-CTL though not different from each other. After the 4-month aging period, the IRP content of the M-UMN-50% WC (1425 mg/L CE) wine was significantly higher than the M-UMN-CTL (1278 mg/L CE) while the M-UMN-25% WC (1357 mg/L CE) was also higher but not significantly so. Ethanol disrupts hydrophobic interactions between the phenolics and cell wall material by decreasing solvent polarity as discussed by Hanlin et al. (2010), which could explain the increases observed after aging. Overall, the 50% WC wines had significantly higher IRP content after aging which was probably due to phenolic compounds contributed by the stems during alcoholic fermentation.

Tannin Content
Tannin content at crushing was 30.75, 22.43, and 19.05 mg/L ECE for F-ISU, M-ISU, and M-UMN, respectively (Fig. 2). Tannin content for all samples increased from crushing to bottling due to extraction during fermentation (Berrueta et al., 2020). At bottling, the F-ISU-25% WC and F-ISU-50% WC had significantly higher tannin content than F-ISU-CTL but were not significantly different from each other. The M-ISU-25% WC and M-ISU-50% WC had significantly higher tannin content than M-ISU-CTL but were not significantly different from each other. The M-UMN-50% WC had significantly higher tannin content than M-UMN-CTL and M-UMN-25% WC at bottling with no significant difference between the latter two conditions. After 4 months of aging, the tannin content of F-ISU-CTL increased while it decreased in F-ISU-25% WC negating the significance of the difference observed at bottling. The tannin content of F-ISU-50% WC did not change after aging and remained significantly higher than F-ISU-CTL. Tannin content increased for all three M-ISU conditions after 4 months of aging with the highest tannin content in M-ISU-50% WC (159 mg/L ECE) followed by M-ISU-25% WC (135 mg/L ECE) and M-ISU-CTL (103 mg/L ECE) with significant differences between each condition. The tannin content of M-UMN-CTL and M-UMN-50% WC decreased after 4 months of aging while tannin content increased slightly in M-UMN-25% WC. M-UMN-50% WC (207 mg/L ECE) had the highest tannin content, followed by M-UMN-25% WC (190 mg/L ECE) and M-UMN-CTL (138 mg/L ECE) with significant differences between each condition. Overall, the 50% WC condition had the highest tannin content among the three conditions from each wine with M-UMN-50% WC having the highest followed by F-ISU-50% WC and M-ISU-50% WC (207,192,and 159 mg/L ECE,respectively). Higher ethanol content in wine was expected to be positively correlated with the tannin content in those wines, which was the case when comparing M-UMN vs M-ISU, but the 50% WC wines had lower, though not significantly, ethanol content than the 25% WC wines in each instance. Whole clusters contributed to increased tannin content in the present study with mixed results at 25% WC showing 4-38% more tannin than control wines after aging. The 50% WC wines always showed significantly higher tannin content than control wines after aging with increases ranging from 17 to 55%. The F-ISU wines showed the smallest increase in tannin (4-17%), followed by the M-UMN (38-50%) and M-ISU (31-55%) wines. These results were in agreement with Wimalasiri et al. (2021) where they found a linear relationship between the increasing proportion of whole bunch addition and increased total phenolic and tannin content. In their study, 30% WC addition did not significantly increase tannin content while 60% or more did in wine made from Vitis vinifera cv. Pinot Noir. Further, Casassa et al. (2019) reported no impact on tannin content with 20% WC addition to Pinot Noir wine but an increase of tannin content with 50% and 100% WC (Casassa et al., 2021).

Anthocyanin Content
Anthocyanin content was not significantly different between WC conditions at either time point in F-ISU and ranged from 621-660 mg/L which was the highest of any fruit source. M-UMN had higher anthocyanin content (428-569 mg/L) overall than M-ISU (301-402 mg/L) which explains the higher color intensity in M-UMN (Fig. 3, Table 1). These ranges are both higher overall than those previously reported for Marquette (233-359 mg/L) . Both M-ISU WC and M-UMN-50% WC wines showed significantly higher anthocyanin content than the CTL wines after aging. This was unexpected because stems are known to adsorb anthocyanins, thus decreasing initial anthocyanin content (Suriano et al., 2015). Additionally, the WC Marquette wines had higher tannin content than the CTL wines (Fig. 2) which might promote the formation of polymeric pigments during aging, thus decreasing monomeric anthocyanins (Picariello et al., 2017). However, anthocyanin content decreased significantly in every Marquette wine except M-ISU-CTL after aging, but the color intensity did not change which further supports the formation of polymeric pigments (Fig. 3, Table 1).

Tasting Evaluation
The goal of this tasting evaluation was to gather preliminary data to indicate large differences in sensory attributes (if present) and gauge preferences between all wine treatments. Participants (n = 30) rated 14 sensory attributes for each of the M-ISU and M-UMN wines after 4 months of aging. Participants were 37.5% female, 62.5% male, and 50% each younger and older than 40 years. Forty-six percent of participants self-reported as winemakers. The results of Rounds 1 (n = 26) and 2 (n = 30) in which M-ISU and M-UMN wines were compared, respectively, (Table S2) showed no significant differences between conditions for any of the attributes, This was most likely because the participants were not trained, or calibrated, which led to a large standard deviation. However, a majority of participants marked the M-ISU-50% WC and both cluster-UMN WC wines as preferred in their respective rounds (Fig. 4). This most likely indicated that a difference existed between conditions, but because of a lack of training, the participant responses could not detect it. A similar conclusion was drawn by Ciccone et al. (2021) when they compared the ability of trained and untrained panels to identify descriptors for cooked potatoes in a High Identity Traits (HITS) method. The untrained panel was more variable in its descriptors. Additionally, participants preferred M-UMN-CTL and M-UMN-50% WC wines to the M-ISU-50% WC wine presented in Round 3. The participants rated the M-ISU-50% WC wine significantly lower for color intensity, red fruit, fruit intensity, and length than both M-UMN wines (data not shown). They also rated the M-ISU-50% WC wine lower in the body than the M-UMN-50% WC and higher in hue and vegetal than both M-UMN wines. There were not any differences detected for taste attributes of acidity, bitterness, sweetness, or saltiness. This may indicate a preference for darker wines and the influence of color on aroma attributes similar to the findings of Jeesan and Seo (2020) in different rice preparations or Wang and Spence (2019) in comparing white, rosé, and white-dyed-rosé wines.

Conclusions
Overall, the 50% WC wines had significantly higher ironreactive phenolics, tannin, and anthocyanin contents than the control wines. Additionally, color intensity in 50% WC wines was maintained after 4 months aging and sensory analysis found untrained panelists preferred 50% WC "Marquette" wines over control wines. This study showed promising results on the use of 50% (w/w) whole clusters as a pre-fermentative technique to manage phenolics and improve red wine quality in Marquette and Frontenac.