Effect of the Addition of Non- Saccharomyces at First Alcoholic Fermentation on the Enological Characteristics of Cava Wines

the Addition of Non-Saccharomyces at First Fermentation on the Characteristics of Abstract: Background: Cava is considered to be a high-quality wine internationally. Hence, it has undergone consistent improvement and/or the preservation of its aromatic qualities, bouquet, color, and foamability, throughout its elaboration and aging. Methods: This study investigates the use of different Saccharomyces and non- Saccharomyces yeasts strains ( Torulaspora delbrueckii and Metschnikowia pulcherrima ) in Chardonnay and Xarel.lo cava wines. The usual enological parameters, the volatile composition, the protein contents, and foamability were determined, and sensory analyses were also performed for all of the viniﬁcations (both before tirage and after 18 months of aging on the lees). Results: the protein and foamability results show that there is a direct relationship between both parameters, with better foam persistence achieved in some non- Saccharomyces fermentation. M. pulcherrima base wines showed a high protein content, improving foamability and foaming persistence. In addition, the results of the aromatic composition and the sensory analysis showed that the use of T. delbrueckii at ﬁrst fermentation produced interesting cavas from an aromatic perspective. These cavas showed the highest values of ethyl isovalerate (120–126 µ g/L), providing aromatic fruity notes, especially fresh green apple. Conclusions: the use of non- Saccharomyces yeasts in the base wine fermentation can be an alternative to produce cavas with differentiated aromatic characteristics and interesting foaming ability.


Introduction
In the last century, Saccharomyces was the only genus of yeasts used in wine cellars and, although grapes have a great diversity of species and yeasts strains, the dominance of Saccharomyces cerevisiae during alcoholic fermentation was expected and desired. Traditionally, the use of Saccharomyces yeasts together with non-Saccharomyces species during alcoholic fermentation influences the final composition of the wine. Thus, the contribution to the final quality of the wine of some yeast strains that acted during fermentation can be negative, because in some cases increases in ethyl acetate, volatile acidity, and acetaldehyde are observed, among others [1,2]. Conversely, they may present a positive contribution, increasing some desirable metabolites, such as acetate esters [3], which generally provide greater organoleptic complexity. Different genera and species of non-Saccharomyces yeasts are now emerging to improve the winemaking process and the quality and differentiation of wines. Thus, many research topics are based on the use of non-Saccharomyces species to improve wine quality and aromatic profile or to modulate the composition of the wine [4,5].
Among the non-Saccharomyces yeasts, the most studied genera are Candida, Metschnikowia, Pichia, Torulaspora, Starmerella, and Lachancea, because they help improve the organoleptic characteristics of wines, such as color, aroma, and sensory characteristics. of the musts was as follows: for CH, brix grade 16.9; nitrogen assimilable by yeast 297 mg/L; ammonium 122 mg/L; pH 3.41; total acidity 8.23 g/L expressed as tartaric acid; and malic acid 5.7 mg/L, and for XA, brix grade 16.3; nitrogen assimilable by yeast 193 mg/L; pH 3.15; total acidity 6.45 g/L expressed as tartaric acid; and malic acid 1.9 mg/L. During grape processing, 40 mg/L of SO 2 (Sulphur 18, Agrovin S.A., Ciudad Real, Spain) was added to prevent oxidation and for microbiological control. The vinifications were carried out in duplicate for each variety on a pilot scale in the experimental VITEC winery. For each vinification, 26 L of must obtained from 40 kg of varietal grapes was used (65% yield). The cold settling was carried out with active pectolytic enzymes (Endozym Éclair, AEB Iberica S.A.U., Barcelona, Spain) at 11 • C overnight and the alcoholic fermentation was carried out at a controlled and constant temperature (17 • C) in cold water baths. To carry out the alcoholic fermentation to obtain the base wine, five different commercial yeasts were used: three different strains from S. cerevisiae, Y1 (IOC 18-2007, Institut Oenologique de Champagne, Epernay Cedex, France) indigenous yeast selected from the Champagne vineyards, Y2 (Viacell Rhône 4600, Lallemand SAS, Blagnac Cedex, France) yeast selected by the Inter-Rhône Technical Service in the northern Cotes du Rhône region, and Y3 (Sensy Yseotm Lalvin, Lallemand SAS, Blagnac Cedex, France) indigenous yeasts selected by natural crossing, and two non-Saccharomyces yeasts, Y4 (Flavia MP346, Lallemand SAS, Blagnac Cedex, France) pure culture of Metschnikowia pulcherrima and selected from nature by the University of Santiago de Chile (USACH), and Y5 (Level Biodiva, Lallemand SAS, Blagnac Cedex, France) pure culture of Torulaspora delbrueckii. The addition of each of the yeasts was carried out as recommended by the manufacturer. The course of the fermentations was monitored by the consumption of sugar, and it was considered complete when the residual sugar concentrations were below 0.5 g/L. In all cases, fermentation began 24-48 h after inoculation and lasted about 10 days. After fermentation, base wines were sulphited to reach 30 mg/L of free SO 2 . After 24 h, wines were racked and clarified with the addition of 40 g/hL bentonite (Bentogran, AEB Iberica S.A.U., Barcelona, Spain). After that, and to carry out the second fermentation, base wines were bottled in glass bottles of 750 mL and, finally, cavas were obtained. For this, S. bayanus (Uvaferm PMA, Lallemand SAS, Blagnac Cedex, France) was added in the tirage solution. Tirage also contained 40 g/L adjuvant (Adjuvant 92, Station Oenotechnic of Champagne, France) and 20 g/L sugar (glucose). Finally, after disgorgement, cavas were aged on their lees for 18 months. Analyses were made in base wines (BW) just before tirage, and in cavas after 18 months of aging on lees (18M).
The values of the alcoholic strength and the volatile acidity were obtained using a Fourier transform infrared spectroscopy (FTIR) system (WineScan TM by FOSS, Hilleroed, Denmark), internally calibrated according to OIV [13].

Total Protein Concentration Determined by UV Spectrophotometric Method
Protein quantification was performed using a UV-visible spectrophotometer (BioDrop µLite, from Thermo Fisher Scientific, Waltham, MA, USA) for micro sample volumes. For measurement and quantification, all of the protein powder extracted from 16 mL of wine sample according to the preparation described by Silvestri et al. [14] was used. Thus, 40 µL of 0.16 M TRIS-HCL (pH 6.8) was added to the protein powder in a 1.5 mL tube and then placed in an ultrasound bath for 60 min at 50 • C. The measurement of the absorbance of the proteins was carried out at 280 nm using 3 µL of sample. Previously, 3 µL of 0.16 M TRIS-HCL (pH 6.8) was used as a blank.

Wine Protein Composition Evaluated by SDS-PAGE
Samples of base wines were analyzed by means of sodium dodecyl sulfate polyacrylamide gel electrophoresis (SDS-PAGE) ( Figure S1). A previous sample extraction treatment was carried out, which was used both for its quantification by UV and for gel electrophoresis, as described by Silvestri et al. [14]. The proteins of the extracted wine samples were separated using SDS-PAGE (15% resolving gel and 5% stacking gel) in a vertical electrophoresis unit with an applied voltage of 100 V. The marker used was Precision Plus Protein™ Dual color Standards (Bio-Rad, Hercules, CA, USA). The gels were stained with Coomassie Brilliant Blue R-250 (0.05%, w/v) in methanol/acetic acid/water (25:10:65, v/v/v) and decolorized in the same solution without the colorant.

Foamability
The foamability of the different cavas was measured using the Mosalux procedure [15]. Briefly, before making the first measurement, it was necessary to regulate CO 2 flow at 7 L/h and the pressure at 1 atm for 15 min. Then, the gas injection was stopped. To measure the foamability of the cavas, 100 mL of sample was used, which was previously filtered through a membrane with a porosity of 0.65 µm and warmed to 18 • C for 12 h. The sample was then placed in the Mosalux cylinder and CO 2 was injected through the glass frit. The maximum height reached by the foam (maximum) was first measured and then, when the foam sample showed stable persistence, the height was measured in a steady state (minimum). Both measurements were performed in duplicate for each sample.

GC-MS Analysis
The volatile compounds of base wines (BW) and cavas (18 M) were extracted using SPME (DVB/CAR/PDMS) and analyzed using a GC 7890A gas chromatograph (Agilent Technologies, Santa Clara, CA, USA) equipped with an inert mass spectrometer 5975C MSD (Electronic Shock Source Triple Axis Detector) according to the method developed by Torrens et al. [16]. The chromatographic conditions were those previously optimized and described [17]. The results of the volatile compounds were semi-quantitative data and expressed as µg/L in relation to the response provided by the internal standard (2-octanol). All analyses were performed in duplicate.

Sensory Analysis
The quantitative descriptive analysis (QDA) was performed by a trained tasting panel (by ISO 8586: 2012) in a sensory standardization room (ISO 8589: 2007). A total of ten cavas were tasted after 18 months of aging in the bottle on lees (18M). The analysis was carried out over two sessions, one session to taste Chardonnay cavas and the other session to taste Xarel.lo cavas. Prior to the sensory analysis, seven descriptors were selected to guide to the panelists when performing the evaluation. These were: fresh, candied, and dried fruit, floral, cake shop, honey, and evolution. The perceived aroma intensity was rated by each panelist from 0 to 5 and the data obtained were processed with the FIZZ software (V.2.47B Biosystems, Barcelona, Spain). A specific tasting sheet was designed according to the aforementioned olfactory descriptors.

Statistical Analysis
A simple analysis of variance (ANOVA) was carried out using the StatGraphics Centurion XVI (Manugistics Inc., Rockville, MD, USA) program. Tukey's procedure was used and the differences at p-values < 0.05 were considered significant.

Enological Analysis
The values of glucose and fructose, malic acid, total acidity, and pH were determined for both Chardonnay and Xarel.lo of all base wines (BW) were analyzed before tirage (Table S1). In addition, the alcoholic strength, volatile acidity, and color characteristics were analyzed of all base wines and also of all cavas after 18 months of aging on lees (18M) ( Table 1). Table 1. Basic analysis of Chardonnay (CH) and Xarel.lo (XA) base wines (BW) and cavas after 18 months of aging on lees (18M), with the addition of five different S. cerevisiae (Y1, Y2, Y3) and non-Saccharomyces (Y4 and Y5) yeasts. Different letters indicate significant differences between samples with different yeasts for each grape variety, and in each of the analysis times. Units: alcoholic strength (%vol) and volatile acidity (g/L). Samples (n = 2).
Regarding the effects of the different yeasts on the wines, the results showed that the CH base wines did not present significant differences in terms of their alcoholic strength and color intensity values, regardless of the yeast used. However, the volatile acidity parameter showed significant differences for all of the yeasts used, with Y3 yeast providing the highest concentration, and non-Saccharomyces yeasts (Y4 and Y5) provided the lowest concentrations. This behavior is consistent with previous studies that also found that some non-Saccharomyces, such as S. pombe, T. delbrueckii, and L. thermotolerans, produced lower concentrations of volatile acidity than some S. cerevisiae (approximately 0.1 g/L) [4,11]. The XA base wines also presented a similar trend, with wines fermented with Y3 presenting the highest volatile acidity, and the lowest concentration was found in wines fermented with the addition of M. pulcherrima (Y4). Wine fermented with the addition of T. delbrueckii (Y5) also showed low values on the volatile acidity content.
In addition, these same parameters were analyzed in the cavas after 18 months of aging on the lees (18M). All of them presented significant differences. In the CH variety, the cavas elaborated with Y3 presented the highest values of alcoholic degree and volatile acidity, while the cavas with M. Pulcherrima presented the lowest values, as already happened in BW. Additionally, the cavas elaborated with addition to T. delbrueckii presented very low concentrations of volatile acidity. In the XA variety, as in CH, the cavas with the addition of Y3 presented the highest values of alcoholic strength and volatile acidity. However, the cavas with Y4 present the lowest values of both parameters. Finally, regarding color intensity, few differences were observed. The highest value of color intensity was obtained in CH cava with Y1.
Previous studies [18,19] already pointed out that the use of yeasts other than Saccharomyces can reduce the alcoholic content of wine, which is in agreement with our data. Furthermore, it is well known that volatile acidity is one of the important parameters that influence the quality of wine or cava and also depends to a great extent on the type of yeast that undergoes alcoholic fermentation. As has been observed in recent studies [20] there are different non-Saccharomyces yeast strains related to producing a low content of volatile acidity, as occurs in our study, with the yeast M. Pulcherrima (Y4), which shows that they may have or generate desirable oenological properties in wines and cavas.

Total Protein Concentration Determined by UV-Visible Spectrophotometry
The results obtained sowed that the quantity of protein was more influenced by variety than by the yeast strain added. Specifically, whereas the total concentrations ranged from 78.2 to 44.5 mg/L in the CH samples, these ranged from 36.3 to 21.3 mg/L in the XA base wines. This relationship between the variety and the amount of protein was previously verified by Cilindre et al. who found a higher content of protein in Chardonnay than in Pinot Meunier base wines [21].
In addition, Dambrouck et al. also observed that a large portion of the wine protein came directly from grapes, which in turn was influenced by the variety; however, it should be taken into account that other proteins were produced by yeasts during the fermentation process [22]. This last consideration explains why, in the present study, non-Saccharomyces yeasts provided similar or even higher protein values than some Saccharomyces yeasts for the two grape varieties. The protein content obtained with M. pulcherrima (Y4) in XA and CH was as high as the highest value obtained by one of the S. cerevisiae (Figure 1).

Wine Protein Composition Evaluated by SDS-PAGE
Proteins released by yeast have been reported as exerting a positive effect on the foaming capacity of sparkling wines [11]. Soluble protein fractions of CH and XA base wines were evaluated to study their relationship with foam characteristics. For CH variety, the MW values ranged between 15.9 and 97.7 KDa, and between 18.2 and 92.2 KDa for the XA variety (Table 2). In general, SDS-PAGE did not show notable differences in the distribution and intensity of protein bands between proteins extracted with Saccharomyces (Y1, Y2, Y3) and non-Saccharomyces yeasts (Y4, Y5), as was observed in previous studies [23].

Wine Protein Composition Evaluated by SDS-PAGE
Proteins released by yeast have been reported as exerting a positive effect on the foaming capacity of sparkling wines [11]. Soluble protein fractions of CH and XA base wines were evaluated to study their relationship with foam characteristics. For CH variety, the MW values ranged between 15.9 and 97.7 KDa, and between 18.2 and 92.2 KDa for the XA variety (Table 2). In general, SDS-PAGE did not show notable differences in the distribution and intensity of protein bands between proteins extracted with Saccharomyces (Y1, Y2, Y3) and non-Saccharomyces yeasts (Y4, Y5), as was observed in previous studies [23].    As shown in Table 2, Chardonnay samples showed the most intense bands around molecular weights between 21 and 24 KDa. When observing the yeast effect, CH base wines fermented with the addition of Y1 presented up to 63% of their total content with an MW of 22.8 KDa. A similar trend was observed in wines with the addition of Y2, which showed 48.4% protein of 22.1 KDa and 25% of 24.9 KDa of MW. The wine with the addition of Y3 showed differences with respect to the previous wines, because although it presented 30% protein with MW of 21 KDa, this sample also presented around 30% of the proteins with higher molecular weights of 66.7 and 57.4 KDa. Wines fermented with the addition of non-Saccharomyces yeasts (Y4 and Y5) presented the highest content in proteins with high MW (28.8% of 74.8 KDa in the case of Y4 and 20.2% of 68.5 KDa in the case of Y5). Regarding base wines produced by Xarel.lo, we can observe that the Y1 and Y2 wines presented the highest MW percentages at 68.4 and 26.9 KDa values for Y1, and 73.2 and 29.5 KDa values for Y2. Y3 wine showed up to 70% of its proteins with MW of 25.3 KDa. However, and as happened in the CH variety, the wines with the Y4 and Y5 yeasts presented between 30% and 50% protein with MW around 30 KDa, and between 20% and 30% protein with the highest MW, 91.6 and 92.2 KDa, respectively. Therefore, it can be highlighted that the Chardonnay variety presented a greater number of bands (protein fractions) compared to the Xarel.lo variety, which could be influenced by the different quantity of proteins extracted from the grape. González-Royo et al. [11] observed that wines elaborated with sequential inoculation with T. delbrueckii and M. pulcherrima had higher values of lower molecular weight (LMW, molecular weight < 60 kDa) compared to the control wines (wines without addition of non-Saccharomyces yeasts), which related to the improved foam parameters observed in wines produced with non-Saccharomyces yeasts. In our study, this trend was observed in XA cavas produced from base wines of Y4 (M. pulcherrima), which presented higher values of LMW (molecular weight <60 kDa). This fact could explain the height of foam stability (HS) value observed in Y4 (Figure 2). By comparison, when the possible influence of the different yeasts on foamability was studied, it was observed that the highest HM and HS measurements were obtained when using Y1 and Y4 for the Chardonnay variety and Y3 and Y4 for the Xarel.lo variety. As can be seen, the samples that provided the highest HM values coincided with those with the highest values of protein content, so it can be concluded that there is a direct relationship between both parameters. These results agree with previous studies which reported that proteins and mannoproteins released by yeasts exert a positive effect on the foam [26], and that an increase in the amount of protein could lead to an increase in foamability [26] and lower velocities of foam dissipation [27]. Therefore, in this study, the results indicate that the wines examined with the addition of M. pulcherrima at first fermentation presented good foamability. Furthermore, this may be related to the high protein content, which can stabilize the bubble film due to its surface properties [26].  Figure 2 shows the values of the parameters HM (maximum foam height) and HS (height of foam stability) for each of the studied cavas (CH and XA). In general, as can be seen, CH cava presented slightly higher HM values than XA, as was observed in the previous studies by Andrés-Lacueva et al. [24] and Vanrell et al. [25], in which Chardonnay cavas showed higher HM values than Macabeo, Xarel.lo, and Parellada cava wines. However, the HS values were similar in all varieties. By comparison, when the possible influence of the different yeasts on foamability was studied, it was observed that the highest HM and HS measurements were obtained when using Y1 and Y4 for the Chardonnay variety and Y3 and Y4 for the Xarel.lo variety. As can be seen, the samples that provided the highest HM values coincided with those with the highest values of protein content, so it can be concluded that there is a direct relationship between both parameters. These results agree with previous studies which reported that proteins and mannoproteins released by yeasts exert a positive effect on the foam [26], and that an increase in the amount of protein could lead to an increase in foamability [26] and lower velocities of foam dissipation [27]. Therefore, in this study, the results indicate that the wines examined with the addition of M. pulcherrima at first fermentation presented good foamability. Furthermore, this may be related to the high protein content, which can stabilize the bubble film due to its surface properties [26].

Volatile Composition
Among the different volatile compounds, a total of 18 were selected because they represent the four most abundant families (esters, acetates, alcohols, and fatty acids) of wines after carrying out the alcoholic fermentation. As shown in Table 3, these were analyzed in base wines (BW) and cavas, after 18 months of aging bottled on lees (18M). Table 3 shows the total volatile concentration of the base wines and cavas. In general, and as expected, there was an aromatic decrease between the analysis times, before tirage in base wines (BW) and after 18 months of aging on lees (18M) as previously observed in other studies [28]. When observing BW values, a higher content of volatiles was found with the XA variety. However, when comparing the different yeasts, both varieties of wine showed the same trend. Thus, the volatile content was similar regardless of the yeast used, except when Y3 was used, which presented significant differences with respect to the other yeasts and provided the lowest volatile concentration.
Regarding the values after 18 months of aging on the lees, the aromatic content values were similar between varieties; thus, considering its higher volatile contents in base wine, XA suffered a greater decrease. When observing each of the varieties independently, greater significant differences were observed between the five yeasts used with respect to BW. In addition, the yeast T. delbrueckii (Y5) stood out in both varieties, with which high values of aroma concentrations were obtained, similar and/or higher than those obtained by some Saccharomyces yeasts.
To better study the results obtained, the different chemical families were separately studied. Thus, in relation to esters, in general these showed the same trend as the one described above for total aromas (Table 3). This behavior is due to the fact that the esters are the most abundant chemicals found in wine aroma and, therefore, are those that contribute the most to the total aroma. Furthermore, this family is of great importance because it provides the two important descriptors, fruit and flowers [29,30]. The base wines with non-Saccharomyces yeasts (Y4 and Y5) presented the highest concentrations of esters in CH, and values equal to or higher than the other Saccharomyces yeasts in XA. After 18 months of aging on the lees, higher amounts of esters were found when using Y2 and Y5 and, when dealing with XA cavas, also Y1. With reference to the different esters found, the highest concentrations of ethyl butyrate, ethyl isovalerate, ethyl hexanoate, ethyl octanoate, and ethyl decanoate were observed in Y2 and Y5 in CH at 18 M (Table S2). Y5 also showed the highest value of ethyl isovalerate in XA cavas. These compounds are directly related to the fruity notes of pineapple, apple, pear, anise, and flowers (Table S2). Furthermore, in general, all of these compounds presented concentrations well above their sensory limit (Table S2), with the exception of ethyl butyrate, which only exceeded it in some cases. Therefore, non-Saccharomyces yeasts, such as T. delbrueckii (Y5), presented and conserved the same or higher concentrations of esters than some Saccharomyces yeasts. Table 3. Families of volatile compounds (µg/L) from Chardonnay and Xarel.lo base wines (BW) and cavas, after 18 months of aging on the lees (18M), with the addition of five different yeasts in the alcoholic fermentation (Y1, Y2, Y3, Y4, Y5). Different letters indicate significant differences between samples with different yeasts for each grape variety, and in each of the analysis times. Samples (n = 2). Regarding acetates, their highest concentration in CH base wine was shown with yeast Y4 but, in contrast, in XA this yeast provided the lowest concentration. In general, the main compounds that showed significant differences were isoamyl acetate, hexyl acetate, and 2-phenylethyl acetate (Table 3), which provide aromas of banana, pear, and green tea, respectively (Table S2) [31,32]. Regarding the 18M analysis time, the highest concentrations of acetates were observed in the Chardonnay variety with the yeasts Y2, Y3, and Y5. It should be noted that, for cavas with Y2 and Y5 yeasts, the main compounds responsible for these high concentrations were isoamyl acetate and hexyl acetate. In the case of cava with yeast Y3, the increase was due to the ethyl acetate. This compound should be carefully considered because it provides positive aromas such as sweet fruits when it appears at low concentrations, but its presence at high concentrations implies undesirable aromas, such as glue or solvent. In this case, the cava with the Y3 yeast presented a high concentration of this compound, with a concentration of 8496 µg/L (Table 3), and was the only sample to exceed the sensory limit (LS of ethyl acetate 7500 µg/L) [33]. This indicates that there are some species of S. cerevisiae which provide certain aromatic deviations in cavas, unlike other non-Saccharomyces yeasts which do not produce them. The Xarel.lo cavas presented the highest concentrations of acetates with the three yeasts Y1, Y2, and Y5. These high values were mainly due to the isoamyl acetate compound, with values well above its sensory limit (670 µg/L) [32]. This could give the cavas a marked tropical aromatic profile because banana is their main aromatic descriptor. In addition, in cavas with Y2 and Y5, higher concentrations of isoamyl acetate and 2-phenylethyl acetate were observed. All of these results show that the yeast T. delbrueckii (Y5) is also capable of providing a more marked tropical character in cavas, unlike others in which only Saccharomyces yeasts were used.
Regarding aromatic alcohols, a different trend between the two grape varieties was observed. Whereas Xarel.lo presented a slight decrease in alcohol concentrations during aging, the Chardonnay variety showed a slight increase in the concentration of alcohol. However, despite these different trends, after 18 months of aging on lees, it was observed that both varieties present similar values of total alcohols, with a concentration range between 4964 and 6353 µg/L. In reference to the analyses of alcohols in Chardonnay BW, all of the base wines presented similar values, obtaining the highest value with yeast Y4 and the lowest concentration with Y2. For the Xarel.lo variety, the high concentrations of alcohols produced by Y3 and Y4 yeasts stand out. These high values were mainly due to two of the studied compounds: isoamyl alcohol and 2-phenylethyl alcohol. These compounds provide floral aromas of roses, pollen, and perfume [1,34]. Regarding the 18M time, in the Chardonnay variety, the cava obtained with Y2 had the highest concentration followed by Y5, Y3, and Y1. This is mainly due to the obtained concentrations of the isoamyl alcohol compound (Table 3). In XA, all of the cavas generally presented similar values, with the exception of the cava with yeast Y5, which presented a slightly lower value. Therefore, with respect to the chemical family of alcohols, it could be noted that the yeast M. pulcherrima (Y4) implied the highest alcohol content of BW but, after aging on lees, these values were generally similar to those obtained by the Saccharomyces yeasts, as observed in the XA variety.
Finally, in the family of fatty acids, the cava obtained with the Y3 yeast presented the lowest concentrations of fatty acids (in all studied cases), with significant differences compared to the other yeasts. In Chardonnay, four base wines presented similar values (Y1, Y2, Y4, Y5). However, for the Xarel.lo variety, the highest concentrations of fatty acids obtained when using Y2 and Y5 stood out. It should be noted that T. delbrueckii (Y5) presented the highest fatty acid values. These high values were mainly due to the concentrations obtained for hexanoic acid and octanoic acid (Table 3). These compounds give aromas of soap, cheese, and yogurt (Table S2). By comparison, after 18 months of aging on the lees, the cava with the Y2 yeast presented the highest concentrations of fatty acids in both varieties. However, it could be noted that non-Saccharomyces yeasts (Y4 and Y5) presented similar or higher values than those produced by other Saccharomyces yeasts (Y1 and Y3). Figure 3 shows the results of sensory analysis of the cavas after 18 months of aging in the bottle on lees. The Chardonnay variety showed the highest values of the fresh fruit descriptor when the Y2 and Y5 yeasts were used. This behavior may be directly related to the high concentrations of esters and acetates that were observed in the analytical analysis. Furthermore, it was observed how cava obtained with Y2 and Y4 presented the highest scores for bakery and floral descriptors, in addition to Y5 for the latter. These aromatic descriptors are related to the high concentrations of fatty acids and alcohols, which are related to lactic and rose aromas, respectively. In comparison, it was shown that the cava obtained with the yeast Y3 was the most evolved sensorial and also showed the highest scores of candied fruit and evolution descriptors, which could be due to its high content of ethyl acetate. In the Xarel.lo variety, it was observed that the cavas with the highest fresh fruit descriptors were those made with Y4 and Y5 yeasts, followed by those obtained with Y2. This can be directly related to the high concentrations of esters and acetates found in gas chromatography analysis. As happened with the Chardonnay variety, Xarel.lo cava obtained with Y3 was the most sensorially evolved. Therefore, it can be emphasized that non-Saccharomyces yeasts provided more fruity and fresh aromatic profiles, especially in the case of T. delbrueckii, and much less evolved than most Saccharomyces yeasts.

Conclusions
The use of different Saccharomyces and non-Saccharomyces yeasts strains for the production of cavas allows products with certain distinctive chemical-physical and sensory attributes to be obtained. In addition, the different aromatic profiles of the wines not only appear after the moment of vinification with the addition of different yeasts, but certain aromatic, taste, and visual distinctions with respect to foamability also appear, after a long time of aging on the lees. In this study, it was possible to corroborate that the use of non-Saccharomyces yeasts allowed cavas with a similar or even better organoleptic quality to be obtained than those made with Saccharomyces. This was the case of the base wine made with M. pulcherrima yeast (Y4), which provided cavas with better persistence in foaminess. Moreover, the cavas made with T. delbrueckii preserved a higher concentration of aromas after 18 months of aging on the lees, with the subsequently more complex sensory properties, than those obtained with respect to S. cerevisiae. These results open the door to new studies that allow for longer aging periods to be evaluated, testing the behavior of other varieties, and verifying the effect of the combination of yeasts other than Saccharomyces.