Abstract
The present study aimed to evaluate the sensory and chemical effects of nine cap management and maceration techniques. Extended maceration (EM) was applied to Vitis vinifera L. cv. Merlot for 0, 1, 2, 4, 6, and 8 wks. In addition, a punch-down treatment and two submerged-cap treatments were also evaluated, giving a total of nine treatments. Descriptive analysis, polyphenolic measures, basic wine parameters, and volatile compound analysis were used to describe these nine treatments. EM defined the astringent texture of the resulting wines, while cap management modified the bitter taste and pepper spice aroma. In total, 27 of 29 measured volatile compounds exhibited a significant treatment effect, thus demonstrating the importance of these enological practices on the final volatile profile. Correlation with EM length was shown for 15 of the significant volatile compounds. In particular, the ester profile of the wines displayed decreased variability after 2 wks of maceration. The results of this study suggest that EM may not be as impactful on astringency as previously thought, due to the fact that EM only significantly affected this attribute after 6 wks. Additionally, the results of the cap management treatments indicated that bitterness can be modified by punch-down versus pump-over techniques.
Cap management refers to a range of techniques used during alcoholic fermentation of red wine to extract color and flavor, and to maintain microbial integrity. A cap is formed during fermentation as carbon dioxide, a by-product of the process, saturates the fermenting juice and produces a buoyant force that causes flotation of grape solids, skins, and seeds. This flotation is problematic, as color, flavor, and volatile compounds are extracted only when skins are in contact with the fermenting juice. Cap formation may also trap generated heat, which may cause fermentation difficulties due to yeast cell stress and death, resulting in stuck or sluggish fermentations (Boulton et al. 1996). Additionally, air-exposed skins are well suited for the growth of aerobic microflora, providing conditions for deleterious aroma production (Boulton et al. 1996).
Three general strategies are employed to keep the cap in contact with the fermenting juice: pump-overs, punch-downs, and cap submersion. With pump-overs, a specified quantity of the fermenting juice is used to irrigate the cap. Generally, an external pump draws juice from the bottom of the ferment, below the cap, and then irrigates it over the cap surface. A high-flow pump-over rate can be used to aggressively break up the cap, whereas a gentle pump-over, utilizing an irrigator attachment, can disperse the juice in a gentle fashion. Punch-downs employ a purpose-built wooden or stainless steel plunger to push the cap into the fermenting juice. Typically, punch-downs are done manually, but mechanical devices using a hydraulically driven plunger are common. These mechanical devices, in general, can be more aggressive than a manual punch-down, although a study using Pinot noir and Dornfelder grapes reported that mechanical punch-downs produce a phenolics profile similar to that of a standard pump-over (Fischer et al. 2000). A submerged cap is achieved by including a barrier below the juice surface, thus preventing the cap from rising above the juice surface. This allows for continuous contact between the grape solids and the fermentation. Two potential issues may arise with this technique: overheating and insufficient mixing. Skins accumulating at the barrier can form an insulating layer that will retain the generated heat (Boulton et al. 1996). In addition, gradient-driven extraction can be inhibited. Punch-downs and pump-overs disrupt the chemical equilibrium between grape skins and the fermenting juice, allowing for additional extraction. In the case of a submerged cap, the extraction of relevant compounds may occur at a slower rate than with the aforementioned other cap management strategies (Amerine 1955).
Once alcoholic fermentation is complete, the pomace (skins and seeds) is generally removed from the wine through draining and pressing. In red wine, extended maceration (EM) occurs when the wine is left in contact with the pomace after the fermentation has reached dryness (Sacchi et al. 2005). Anecdotally, EM is typically applied to high-end red wines such as Nebbiolo or Cabernet Sauvignon, with periods of contact time ranging from a few days to a more typical period of 2 to 3 wks, with some producers allowing EM to proceed for several months.
Previous studies have reported the influence of EM with several grape varietals at varying maceration lengths. These range from a three-day period for Shiraz (Reynolds et al. 2001) to an EM of 42 days in Cabernet Sauvignon (Scudamore-Smith et al. 1990) and 90 days with Aglianico di Taurasi (Francesca et al. 2014). The effect of EM on Merlot, the cultivar investigated in the current study, has also been studied to some extent (Harbertson et al. 2009, Casassa et al. 2013a, Daudt and de Oliveira Fogaça 2013). However, these studies generally limited their analyses to empirical measures, including total phenolics, specific proanthocyanidins, and/or anthocyanins (Gómez-Plaza et al. 2001, Gil et al. 2012). In most cases, the authors also analyzed the wines for their basic wine components such as alcohol, pH, and titratable acidity (TA). Therefore, the full chemical implications of EM are not fully understood.
Notably, most of the previous studies on these enological techniques performed limited or no sensory analyses. Reynolds et al. (2001) applied pairwise testing and reported significant differences in red color, body, finish, and herbaceous character between a three day EM treatment and a control. A study undertaken by Francesca et al. (2014) reported a decrease in astringency with maceration, with the lowest astringent sensation after 70 days of maceration. Casassa et al. (2013a) reported a significant increase in astringency only with prolonged maceration. These studies demonstrate a clear need to further evaluate how these winemaking practices affect complex sensory parameters. If the impact of these techniques on flavor is better understood, it will allow for better processing decisions.
Furthermore, a narrow range of studies have been undertaken to help evaluate the effect of EM on aroma chemistry. One study demonstrated that the concentration of free volatiles peaks after five days EM, whereas 15 days gave the highest concentration of bound volatiles (Yilmaztekin et al. 2015). No difference in methoxypyrazine content was reported in Cabernet Sauvignon after 24 hrs compared with free-run juice, whereas another study reported that the evolution of volatile compounds from Cabernet Sauvignon grapes generally increases as fermentation occurs in the presence of grape skins, although this was not consistent for all compounds (Callejón et al. 2012). It was concluded that grape solids can bind some volatile compounds (e.g., β-damascenone) and may have significant effects on volatile compound formation and release. Extended skin contact also seems to favor the occurrence of vanillic acid, higher fusel alcohol concentrations, and fatty acid content (Fischer et al. 2000).
As outlined above, previous investigations on Merlot fermentations have primarily focused on the influence of EM on phenolic extraction. The objectives of this research were to measure both the compositional changes and sensory impact of this technique applied to Merlot ferments for 0, 1, 2, 4, 6, and 8 wks. In addition, a punch-down treatment and two submerged-cap treatments were also evaluated. The current manuscript presents the relevant chemical measures and highlights key relationships with the sensory data.
Materials and Methods
Wine production
Grapes from Vitis vinifera L. cv. Merlot clone 3 were harvested from the UC Davis Oakville Research station vineyard. The fruit (4500 kg) was machine-harvested on 5 Sept 2013, with a total soluble solids (TSS) concentration of 27.4 Brix, and was transported to the UC Davis experimental winery. The grapes were crushed and destemmed with a Bucher Vaslin Delta E4 destemmer, and the must was pumped into 27 jacketed stainless steel 150 kg fermenters. To help achieve homogeneity, crushed fruit was sequentially added, in 10 cm additions, to each of the 27 fermenters. Post-processing 1 g/L tartaric acid additions were made to decrease the must pH to 3.6. Yeast assimilable nitrogen (YAN) was measured and adjusted to 280 mg/L YAN by using a combination of Fermaid K (Lallemand) and diammonium phosphate (American Tartaric Products). All fermenters were inoculated 24 hrs after fruit processing with 200 mg/L of EC1118 yeast (Lallemand), and fermentation began within 72 hrs of crushing.
The following nine treatments, summarized in Table 1, were performed in triplicate: (1) pump-over and pressed at dry, zero maceration (Em0); (2) pump-over and pressed after 1 wk of postdry EM (Em1); (3) pump-over and pressed after 2 wks of postdry EM (Em2); (4) pump-over and pressed after 4 wks of postdry EM (Em4); (5) pump-over and pressed after 6 wks of postdry EM (Em6); (6) pump-over and pressed after 8 wks of postdry EM (Em8); (7) submerged cap and pressed at dry (Su0); (8) submerged cap and pressed after 8 wks postdry maceration (Su8); and (9) punch-down (PuD) in which the fermentation was punched down and then pressed at dry. The submerged-cap treatments were achieved with purpose-designed heavy stainless steel mesh, and punch-downs were performed three times daily with a stainless steel plunger. The pump-overs were completed by a built-in fermenter pump three times daily for 12 min to achieve one fermenter volume of juice to be pumped on each occasion. Cap and juice temperatures were automatically monitored and adjusted every 15 min, with cap temperatures maintained at 25 to 28°C. When the TSS had decreased to 14 Brix, malolactic fermentation (MLF) was initiated with 0.01 g/L Lallemand Alpha Oenococcus oeni (Lallemand). The wines were considered dry when the residual sugar, that is, the sum of glucose and fructose measurements, was <0.1 g/L.
The fermentations reached dryness 12 days postcrushing, and the three unmacerated treatments (Em0, PuD, and Su0) were pressed using a hydraulic press. The remaining treatments were pressed over the following 8 wks at the maceration intervals indicated. During the maceration period, each fermentation tank was pumped over for 5 min per day, and cap temperatures were maintained at 22°C.
At completion of MLF, wines were racked into 50 L stainless steel containers, and 80 mg/L of sulfur dioxide in the form of potassium bisulfite (American Tartaric Products) was added to achieve a free sulfur dioxide concentration of 30 to 40 mg/L. Wines were then allowed to further settle and cold-stabilize at 0 to −2°C for 3 wks prior to bottling. Each treatment replicate was bottled separately into 750 mL Bordeaux-style dark amber glass bottles and then sealed with Saranex screwcaps. After bottling, the wines were stored at 16°C and 60% relative humidity.
Descriptive sensory analysis
The 27 wines were evaluated by a trained panel of 12 volunteers (eight men and four women, 21 to 72 years old) five months after bottling. All panelists had previous wine tasting experience and were selected on the basis of their availability and interest. Consensus terminology and reference standards (Table 2) were developed over seven training sessions held over 2 wks. Following the training, panelists were required to evaluate the 27 wines during nine tasting sessions. This was completed by tasting nine wines per session using a randomized block design, in which the wines were randomly divided into a flight of four wines and a flight of five wines. The panelists were randomly assigned an individual tasting booth that contained the flight of five wines. After these wines were rated, panelists were presented the second flight of four wines. A 1 min rest was imposed between each sample, and a 5 min rest between flights.
Each tasting session began with the assessment of the 11 aroma standards. Panelists were required to smell each aroma standard labeled with a random three-digit code and indicate the identity of the standard. After all standards were assessed, feedback was provided by indicating any incorrect responses, and the panelists were then instructed to review these aroma standards.
The FIZZ sensory data collection program (ver. 2.50, Bio-systèmes) was used for the testing. The initial screen displayed six attributes, always presented in the same order: Red Fruit, Dark Jam, Prune/Raisin, Citrus/Floral, Bay, and Vegetative, with an adjacent unstructured line. The second page presented Pepper Spice, Earthy, Aldehydic, Alcohol, and Volatile Acidity (VA), and the third page presented the taste/mouthfeel terms Sweet, Sour, Bitter, Hot Mouthfeel, Overall Astringency, Drying, and Astringent Texture. Specific attribute descriptions and reference standards are included in Table 2.
All evaluations were conducted with black ISO glasses and red lighting. Panelists were provided with two unsalted crackers, a cup of water, and an empty cup to expectorate. Panel performance was evaluated with PanelCheck (Tomic et al. 2010) and was found to be acceptable (data not shown).
HS SPME-GC-MS
A total of 29 volatile compounds were measured using headspace solid-phase microextraction-gas chromatography-mass spectrometry (HS SPME-GC-MS) as previously described (Hjelmeland et al. 2013). All 29 compounds were verified by analyzing reference compounds, except vitispirane I and II. Most reference compounds were purchased from Sigma Aldrich with a purity of >80%. The exceptions were ethyl-2-methylbutyrate (SAFISIS), linalool (Alfa Aesar), acetic acid EMD (Merck), 2-methyl butanoic acid (TCI America), hexanoic acid (Acros Organics, Thermo Fisher Scientific), and propionic acid (MP Biomedicals), all with a purity of >80%. All standards were diluted using 100% ethanol (Gold Shield Chemical Co). The retention times of the authentic standards were matched to the experimentally measured compounds. Compounds were also verified with quantifier/qualifier ion ratios and published retention indices. Alkanes C8 to C20 were purchased from Sigma Aldrich. The reported retention indices and references were previously reported (Hjelmeland et al. 2013). Each treatment replication was measured in triplicate.
Basic wine chemistry panel
pH and TA as tartaric acid equivalents were measured as previously described (Amerine et al. 1980). Enzymatic assays were used to measure malic acid (MA), acetic acid (AA), and reducing sugars (RS). Ethanol was determined with an Alcolyzer Wine M/ME (Anton Paar).
Color analysis
For each wine sample, CIEL*a*b* color space values were measured with a chromameter equipped with a 10 mm wide cell (CR-400 with a pulsed xenon lamp light source and a 2° observer, which closely matches the CIE 1931 Standard Observer and a C* illuminant [Konica Minolta Sensing Americas, Inc.]). Color was expressed as Lightness (ranging from black/white, 0 to 100), a* (red/green, +a*/−a*), and b* (yellow/blue, −b*/b*). Prior to collecting the measurements, the chromameter was calibrated on a white calibration plate.
Polyphenol analysis
Tannin (mg/L catechin equivalents [CE]), total phenols (mg/L CE), anthocyanins (mg/L malvidin-3-glucoside equivalents [M3G]), small polymeric pigments (SPPs), and large polymeric pigments (LPPs) were measured as previously described (Harbertson et al. 2003, 2015). Each fermentation replication was measured in triplicate.
Data analysis
All statistical analyses were performed using R version 3.3.0 “Supposedly Educational” (R Core Team 2016) and the RStudio IDE (R Studio Team 2016). R packages ggplot2 (Wickham 2009) and FactoMineR (Husson et al. 2014) were extensively used. A significance of α = 0.05 was set for all analyses.
Analysis of variance and least significant difference
The overall experimental design had nine treatments, with each treatment performed in triplicate using three individual fermentation vessels. Each fermentation vessel was considered unique, and, therefore, when applicable, the variable of fermentation replicate was nested within the treatment. Analysis of variance (ANOVA) was applied to the descriptive analysis, basic wine chemistry, and polyphenol data sets.
The descriptive analysis data was analyzed using a fixed-effects ANOVA, with fermentation replicate nested within treatment. In the case where treatment was significant and the judge × treatment interaction and/or the treatment × sensory replication was significant, the F-statistic for the given attribute was recalculated using the interaction mean square (Gay 1998). A nested ANOVA model was also applied to the polyphenol measurements. Finally, the basic wine chemistry measures were analyzed with a one-way, fixed-effects model. Significant difference among univariate mean values were determined for each data set with Fisher’s least significant difference (LSD).
Pearson product moment correlation coefficient r
The linear relationship between the six maceration lengths for treatments (Em0 to EM8), and a specific measured parameter, was tested by correlation analysis. All unique values were incorporated for each fermentation replicate.
Partial least squares (PLS)
A PLS2 model was applied by relating the significant SPME-GC-MS measures, combined with measures of total phenol mg/L CE, tannin mg/L EC, and ethanol (%, v/v), to the scaled significant (ANOVA) sensory descriptors. The chemical measures comprised the predictor variables (x-matrix) and the sensory data, the response variables (y-matrix). The model was evaluated with the root mean square error of prediction via “leave one out” cross validation.
Results and Discussion
This investigation evaluated the sensory and chemical changes that resulted from nine different maceration and cap management treatments. Polyphenolics, basic wine chemistry, and volatile compound data sets were collected and then related to the descriptive analysis results. The following discussion describes the key relationships identified between the chemical profile and the perceived sensory profile. In addition to highlighting the variability among treatments, the within-treatment variability is also discussed. Thus, a given treatment is identified by a three-digit alphanumeric code (e.g., Em0 or PuD), whereas the letter A, B, or C distinguishes the specific fermentation replicate (e.g., Em0_A, Em0_B, Em0_C, or PuD_A).
Descriptive analysis summary
The descriptive analysis and temporal dominance of sensation results have been previously reported (Frost et al. 2018). Table 3A contains the treatment means for each significant attribute. In brief, descriptive analysis showed significant treatment effects for red fruit, pepper spice, aldehydic, and alcohol aromas, in addition to bitter taste, hot mouthfeel, and astringent texture. The highest intensity of astringent texture was reported for treatments with the longest maceration time, that is, Em8 and Su8. Treatments with short or no maceration (PuD, Em0, Em1, and Su0) had the lowest perceived astringent texture (Table 3A). The unmacerated treatment Em0 had significantly lower bitterness than the EM treatments (Em1 to Em8); however, no further significant differences for bitter taste were identified among the maceration treatments (Em1 to Em8). Pepper spice aroma was significantly higher in the pressed-at-dry treatments Em0, Su0, and PuD, but decreased with EM. In contrast, red fruit aroma was increased with EM, with the highest intensity found in Em6. Mean attribute intensity, coefficient of variation, and Fisher’s protected LSD values are presented in Table 3A.
Basic wine chemistry
Calculated means, coefficients of variation, and LSD values are displayed in Table 3A. TA, pH, AA, and ethanol concentration each showed significant treatment effects. Mean TA values ranged from 5.45 g/L in Em2 to 5.93 g/L in Su0. A significant correlation between the EM treatments Em0 to Em8 and TA was not shown, but pH did show a significant positive correlation with longer macerations. Of the basic chemistry measures, only pH was correlated with EM, which has been previously reported (Casassa et al. 2013a, 2013b). It is possible that this was an effect of grape skin degradation; as maceration occurs, additional grape cell wall material is exposed, allowing for the exchange of skin-bound potassium for hydronium, thereby increasing pH (Harbertson et al. 2009).
AA concentration ranged from 0.36 mg/L to 0.41 mg/L, but no correlation was found with maceration length. Lastly, RS and MA concentrations exhibited little variability apart from Em8. MA concentrations were lower than the limit of detection (30 mg/L) for all fermentations with the exception of Em8_A (404 mg/L) and Em8_B (138 mg/L). Em8_A also contained 1.50 g/L RS, with all other fermentations yielding less than 0.41 g/L RS. These results indicate that EM did not affect AA and MA concentrations after fermentation.
Polyphenolics
With longer EM, wine tannin and total phenol concentrations increased (Table 3A). Tannin concentration increased with 2 wks of maceration, but no significant difference was measured among 2 (Em2), 4 (Em4), and 6 (Em6) wks of maceration. A second increase in tannin concentration was observed at 8 wks (Em8), but notable variability, indicated by a coefficient of variation (CV) of 12% within the Em8 treatment, was present. The punch-down PuD treatment exhibited significantly less tannin than the pump-over (Em0) or the submerged-cap (Su0) treatments, highlighting the ability of the different cap management techniques to alter tannin concentration. This was consistent with previous work reporting that pump-overs yield higher tannin concentrations (Fischer et al. 2000).
An evaluation of the ratio of tannin to total phenols (phenolic ratio, Table 3A) provided insight into the qualitative polyphenol difference as related to cap management and length of EM. The three treatments without EM (PuD, Em0, and Su0) had a significantly (p < 0.05) lower phenolic ratio, indicating that the overall polyphenolics content comprised a lower proportion of precipitable tannin. The phenolic ratio of the Em1 treatment was significantly higher than those in all unmacerated treatments, and a significant correlation was measured between the phenolic ratio and the length of EM (Em0 to Em8). A maximum ratio was reached after 2 wks of maceration (Em2), with similar values observed for the remainder of the maceration period investigated. This indicated that the ratio of precipitable to nonprecipitable polyphenols was determined primarily during the early maceration period. The similar ratio among the Em1 to Em8 maceration treatments could have resulted from precipitable and total polyphenols maintaining the same proportion by nonspecific bulk extraction. This was supported by the significant correlation between tannin and total phenols (r = 0.89, df = 72). If polyphenol extraction were specific, meaning that precipitable and nonprecipitable polyphenols extracted at significantly different rates, the phenolic ratio would have increased or decreased with maceration. These results imply that during fermentation, nonprecipitable polyphenols were extracted at a higher proportion to precipitable polyphenols, but during the maceration, the two pools of phenols increased at similar rates. Cap management also significantly affected the phenolic ratio. The submerged-cap treatment (Su0) had the lowest ratio, as compared with Em0 and PuD. The 8 wk submerged treatment (Su8) also had a lower ratio than the 8 wk maceration (Em8).
Although measured differences in tannin concentration were found, the sensory results showed no difference in astringent texture among the unmacerated treatments (Em0, PuD, and Su0). Descriptive analysis also revealed no difference in the astringent texture among Em0 to Em6. Differences in astringent texture were only significant at 8 wks of maceration. However, the PuD treatment showed significantly less astringent texture than the Em2, Em4, and Em8 treatments. These results confirm previous literature reports indicating a relationship between perceptible tannin and astringency (Casassa et al. 2013a). The measured tannin difference between Em0 and Su8 was 323 mg/L CE. This was the smallest difference between two samples that showed a significant difference in perceived astringent texture. This finding was largely in agreement with that of a study reporting that a minimum difference of 265 mg/L gallic acid equivalents was required to elicit a difference in the perception of astringency (Hopfer et al. 2012).
In contrast to astringent texture, the perceived intensity of bitterness was altered by the cap management treatments applied. The Em0 treatment resulted in wines that were significantly less bitter than those of the PuD treatment. These treatments each had identical phenolic ratios, calculated from similar total phenol and tannin measures. This result highlights the complexity of bitterness and the need for advanced analytical methods that will capture compound-specific changes relating to bitter taste in wine.
Pigments: Anthocyanin, SPP, LPP, and color L*a*b*
Previous studies have shown that anthocyanin concentration peaks approximately five to 10 days after crushing, followed by a decline that obeys second-order kinetics (Scudamore-Smith et al. 1990, Casassa et al. 2013b). Cap management and winemaking techniques both have been shown to alter peak pigment concentration (Fischer et al. 2000, Casassa et al. 2013a, 2013b), but the effect of EM on the rate of anthocyanin decay has not been fully studied. Casassa et al. (2013b) measured the rate of anthocyanin decay in a treatment including 30 days of total skin contact, of which ~10 days were during primary alcoholic fermentation. The same study did not report a change in the rate of anthocyanin decay after pressing, and it is unclear how, or whether, varying maceration length would affect the decay of anthocyanin after pressing.
In the current study, cap management and maceration altered the wine pigment profile. Anthocyanin measures, collected 11 mos after fermentation, exhibited significant treatment differences. The maceration treatments (Em0 to Em8) showed significantly decreased anthocyanin concentration; each increase in maceration length produced a significant decrease in anthocyanin concentration. This trend continued through Em6, but no difference was detected between Em6 and Em8 (Table 3A). One explanation for this decrease may be sorption of pigments by yeast cells. Although it has been shown that yeast cell walls preferentially bind to larger proanthocyanidins (Bindon et al. 2010), anthocyanins can also be retained by the polymer structure of the yeast cell wall or are absorbed by the yeast cell (Vasserot et al. 1997, Morata et al. 2003). Several studies have reported a rapid decay of anthocyanins early in the fermentation, followed by a reduced decay rate (Nagel et al. 1979, Gao et al. 1997, Yokotsuka et al. 2000). It is also possible that the increased tannin concentration resulting from a longer maceration is affecting the decay of anthocyanin through LPP and SPP formation during maturation (Harbertson et al. 2003).
SPP concentration discriminated the nine treatments in a similar fashion as the anthocyanins concentration, but the LPP concentration showed a different discrimination pattern. Longer maceration produced wines with less SPP, whereas the pressed-at-dry submerged-cap treatment (Su0) yielded the highest SPP concentration. These results imply that SPP and anthocyanins degrade at a similar rate and also highlight the substantial effect of grape solids on the final pigment concentration.
The nine treatments gave a different effect on LPP. With maceration, LPP concentration was the highest in Em2, followed by decreasing amounts in Em4, Em6, and Em8. The three pressed-at-dry treatments had lower LPP levels than did Em2, and the PuD and Su8 treatments had the lowest amount of measured LPP. The availability and concentration of proanthocyanidin could be a limiting factor for LPP formation. The pressed-at-dry treatments (Em0, PuD, and Su0) measured significantly less tannin than the macerated treatments, which could partially explain their low LPP. These results emphasize the substantial effect of maceration on anthocyanin concentration.
A significant treatment effect was shown for the color measures of lightness (L) and blue/yellow (b*). Increased maceration showed an increase in the yellow spectrum as indicated by increased b* values. Lightness also increased with maceration, but significant differences were not observed for measures of the green/red a*. Although not evaluated, visual differences would likely not be observed, given the small differences in the L*a*b* measures.
Volatile compound analysis
The present study measured 29 volatile compounds with SPME-GC-MS, of which 27 showed significant treatment effects. The volatile measurements were performed 12 mos after harvest and 4 mos after the descriptive analysis. To evaluate the effect of EM on the measured volatiles, the correlation among all fermentation replicates of the six EM treatments (Em0 to Em8) was examined. Of the 27 significant compounds, 15 were positively correlated with maceration length. Previous studies have commonly concluded that maceration generally increases overall volatile compound concentration (Álvarez et al. 2006, Francesca et al. 2014, Petropulos et al. 2014, Yilmaztekin et al. 2015). However, these studies reported on few compounds and did not indicate a maximum length of maceration that is relevant for aroma compound extraction.
The volatile ester profile displayed a decrease of within-treatment variation with longer maceration. This was best demonstrated with the CV (in %) for each of the nine significant esters, which decreased with maceration. The pressed-at-dry treatment (Em0) and 1 wk maceration (Em1) each had % CV values ranging from 92 to 20% for each of the nine significant esters. This was in contrast to the % CVs for each ester with 2, 4, 6, and 8 wks of maceration, which show a drastic decrease in % CV. This trend implied an effect of maceration on the volatile ester profile that apparently has not been reported in the literature. A similar effect is not repeated for the remaining compound classes.
With the exception of phenethyl alcohol, the yeast-derived volatile alcohols increased with maceration. Benzyl alcohol exhibited the strongest correlation with EM (r = 0.88, df = 52) of the five alcohols, whereas hexanol was not correlated with maceration, but cap management (Em0, PuD, and Su0) significantly altered the concentration of hexanol (Table 3A and 3B), with punch-downs producing the greatest effect. The general increase of volatile alcohols could have resulted from yeast metabolism continuing to occur after the alcoholic fermentation had finished, which would then end with pressing and potassium bisulfite addition.
Three terpenes were negatively correlated with length of maceration: α-terpinene (r = −0.59, df = 52), linalool (r = −0.47, df = 52), and nerolidol (r = −0.77, df = 52). A positive relationship was shown with β-citronellol (r = 0.27, df = 52). The grape-derived β-damascenone (r = −0.90, df = 52), along with methionol (r = −0.83, df = 562), negatively correlated with maceration.
Relating sensory to chemistry PLS-R
The PLS regression (Figure 1) revealed relationships between sensory and chemical compositional data. The PLS-R was developed using the sensory attributes that differed significantly across the treatments in conjunction with the significant chemical measures of tannin, total phenols, LPP, SPP, anthocyanin, ethanol, pH, TA, and 27 significant volatile measures. The sensory and chemical measures were related to the fermentation replicates to visually display the within-treatment variation. Each fermentation replicate is indicated as A, B, or C in Figure 1. For example, the three submerged-cap, pressed-at-dry treatments (Su0) are indicated as Su0_A, Su0_B, and Su0_C.
The pepper spice aroma showed a positive relationship with the three unmacerated treatments (PuD, Su0, and Em0), and with eight of the volatile GC measures (di-acetyl, α-terpinene, linalool, methionol, phenyl acetate, β-damascenone, γ-nonalactone, and nerolidol). This relationship was also observed in the univariate analysis, as each of the eight volatiles measured high concentrations in the pressed-at-dry treatments (Table 3B). The aroma of black pepper is commonly associated with measures of rotundone, as has been shown for Syrah (Mattivi et al. 2011). Although not measured in the present study, rotundone is almost certainly not the causal compound, as Merlot lacks the enzymatic ability to express rotundone in appreciable quantities (Takase et al. 2016). The PLS-R indicated that the combination of methionol and β-damascenone influenced the pepper spice character. In contrast, a previous study using model wine concluded that methionol has a negligible impact on aroma (de-la-Fuente-Blanco et al. 2016). Additionally, β-damascenone has been reported to modify wine aroma in an indirect manner (Pineau et al. 2007), as well as being associated with a prune aroma (Pons et al. 2008). These conflicting results highlight the difficulties in capturing drivers of wine aroma and the confounding effect of the wine matrix.
Of all descriptors, pepper spice exhibited the highest % CV (Table 3B), with eight of the nine treatments >100%. This variability of the pepper spice descriptor was driven by the variation among the fermentation replicates of Em0, Em1, and PuD, and is visualized by the position of the individual fermentation replicates in the PLS-R (Figure 1). Em0_C was less associated with pepper spice than Em0_A and Em0_B, as indicated by its position in Figure 1. One fermentation replicate of the 1 wk maceration treatment, Em1_B, was positively positioned with pepper spice. The remaining two fermentation replicates, Em1_A and Em1_B, were plotted in a separate quadrant.
The maceration treatments Em2 and Em4 were positively positioned with alcohol aroma and hot mouthfeel. The fermentation replicates of these two treatments were positioned adjacently, implying low within-treatment variability. The low variability of alcohol aroma and hot mouthfeel was also expressed in the individual treatment % CV values. Hot mouthfeel levels had the lowest % CV of all treatments. The measured ethanol concentration was positioned with alcohol aroma and hot mouthfeel, indicating a relationship between ethanol content and sensory perception. In addition, nine volatile compounds also showed a strong positive relationship with alcohol aroma and hot mouthfeel; this compound group comprises limonene, geraniol, isobutanol, isoamyl alcohol, ethyl isobutyrate, ethyl butyrate, ethyl-2-methylbutrate, ethyl isovalerate, and ethyl hexanoate. This mix of volatile compounds likely drove the alcohol aroma found in the wines. Red fruit aroma was associated with ethyl acetate, β-citronellol, and AA measures. Ethyl acetate is often associated with lifting the fruit aroma of red wine. The measured VA was below the threshold, indicating that ethyl acetate would not be perceived as “nail polish”, but as a fruity character. Furfural and benzyl alcohol were associated with aldehydic aroma. These compounds are typically described as having the aroma of “almond, caramel, or woody.”
Bitterness was associated with the macerated treatments. In general, the fermentation replicates for Em6, Em8, and Su8 were positioned near each other, indicating that these treatments were characterized by the same chemistry. The three Su8 fermentation replicates were also strongly associated with total phenol, tannin, and astringent texture.
Conclusion
The cap management and EM practices applied here altered the chemical and sensory profiles of the resulting Merlot wines. The highest intensity of astringent texture was reported for treatments with the longest maceration time, whereas the treatments with short or no maceration had the lowest perceived astringent texture. Additionally, the control treatment had a significantly lower perceived bitterness than all wines produced by using pump-overs and then subjected to EM. Overall, differences in astringent texture were realized with >6 wk maceration, but bitterness was best manipulated with cap management. These results are largely consistent with the tannin and total phenol concentration measures, where significant tannin increases were observed after both 2 and 8 wks of maceration. The capacity of the different cap management techniques to alter tannin concentration was also illustrated, with the punch-down treatment producing significantly less tannin than the pump-over or the submerged-cap treatment. These results are important considerations in the production of wine, as different treatments can produce wines with varying mouthfeel and overall style. The pepper spice aroma was significantly higher in all treatments without maceration, regardless of cap management technique, whereas increased maceration length led to a higher perception of red fruit character. The influence of maceration on prominent aroma characteristics further underlines the importance to the winemaking industry of understanding the effects of this technique on wine sensory characteristics.
Acknowledgments
Funding was provided through the Stephen Sinclair Scott Endowment, the Ray Rossi Chair, and the American Vineyard Foundation.
- Received June 2017.
- Revision received January 2018.
- Revision received April 2018.
- Accepted May 2018.
- Published online October 2018
- ©2018 by the American Society for Enology and Viticulture
Literature Cited
Sign in for ASEV members
ASEV Members, please sign in at ASEV to access the journal online.
Sign in for Institutional and Non-member Subscribers
Log in using your username and password
Pay Per Article - You may access this article (from the computer you are currently using) for 2 day for US$10.00
Regain Access - You can regain access to a recent Pay per Article purchase if your access period has not yet expired.