*2.10. Statistical Analysis*

Statgraphics v.5 software (Graphics Software Systems, Rockville, MD, USA) was used to calculate means, standard deviations, analysis of variance (ANOVA), least-significant difference (LSD) test and principal component analysis (PCA). The LSD test was used to detect significant differences between means. Significance was set at *p* < 0.05.

#### **3. Results and Discussion**

#### *3.1. Basic Oenological Parameters*

a

In general, no significant differences were found in the wines fermented by *H. vineae* compared to conventional wines fermented by *S. cerevisiae*, with the exception of the total acidity parameter. The *S. cerevisiae* wines showed 0.5 g/<sup>L</sup> more total acidity than the *H. vineae* wines (Table 2). However, no differences in lactic acid, malic acid and volatile acidity content were found, therefore, the decrease of total acidity may be due to the precipitation of tartaric acid during the alcoholic fermentation. It is important to mention that these differences were not reflected in the pH values, since the pH was similar in all the wines studied.

**Table 2.** Ethanol content (% *v*/*v*), pH, total acidity (g/L) as tartaric acid, volatile acidity as acetic acid (g/L), malic acid (g/L), lactic acid (g/L) and glucose and fructose (g/L) after fermentation process. Mean ± SD for three replicates.


Means with the same letter are not significantly different (*p* < 0.05).

Regarding the residual sugar content, both yeasts have been able to ferment all the sugar, with final concentrations in the wine below 2 g of residual sugar per litre. These results are in line with those obtained by other authors that compared both yeas<sup>t</sup> species in Macabeo and Merlot grape wines [25]; nevertheless, [26] found 0.5 g/<sup>L</sup> of glucose and frutose more in *H. vineae* wines than in *S. cerevisiae* wines before the malolactic fermentation. This fact is linked to the glycolytic power—all wines showed similar ethanol contents around 11.9% *<sup>v</sup>*/*<sup>v</sup>*. These results indicate that both yeas<sup>t</sup> species may produce wines with similar basic oenological parameters.

Targeted NMR analysis allowed the identification and quantification (Table 1) of typical wine metabolites in both *H. vineae* and *S. cerevisiae* samples: furfural (9.64 ppm), formiate (8.41 ppm), shikimic acid (6.87 ppm), fumaric acid (6.4 ppm), β-glucose (4.55 ppm), fructose (4.04 ppm), citrate (2.84 ppm), succinate (2.66 ppm), glutamine (2.25 ppm), acetate (2.01 ppm), proline (2.05 ppm), γ-aminobutyric acid (1.96 ppm), arginine (1.70 ppm), alanine (1.55 ppm), threonine (1.28 ppm), valine (1.1 ppm) and isoleucine (0.91 ppm). With the results obtained by NMR, a principal component analysis (PCA) was performed. Using the 2D-projections (PC1 = 43.1%, PC2 = 24.2%), slight overlaps were observed amongs<sup>t</sup> groups (Figure 1A). The distribution was better explained with the first three components (PC1 = 43.1%, PC2=24.23% and PC3=13.59%). Even though the results were not statistically significant between the two yeasts studied (Table 1), the PCA made it possible to differentiate the wines studied into two independent clusters corresponding with the two target yeasts (Figure 1). Chemical shift loading plots (Figure 1B) show a set of relevant resonances that permits the discrimination between yeasts by PCA: formiate (8.4123 ppm, PC1 [+], PC2 [+]); shikimic (6.8740 ppm, PC1 [−], PC2 [−]); β-glucose (4.5395 ppm, PC1 [−], PC2 [−]); fructose (4.0375 ppm, PC1 [+], PC2 [−]); citrate (2.8415 ppm, PC1 [−], PC2 [−]); succinate (2.6655 ppm, PC1 [+], PC2 [−]); all amino acids present positive PCA 2

(glutamine 2.2465 ppm, PC1 [+], PC2 [+]; alanine 1.551 ppm, PC1 [+], PC2 [+], valine 1.0595 ppm, PC1 [+], PC2 [+] and isoleucine 0.9140 ppm, PC1 [−], PC2 [+]) and acetate (2.0925 ppm, PC1 [−], PC2 [−]). These results allow us to differentiate the metabolism of both yeasts, even though these differences were not quantitatively observed. It is noted that we identified the same separation between the must fermented by *H. vineae* and *S. cerevisiae* when the PCA was done on fermentative volatile compounds (Figure 2).

**Figure 1.** Principal component analysis (PCA) score plots comprising the 67.33% variance (**A**) and 80.92% variance (**C**) and chemical shift loading plots (**B**) obtained by a variable NMR bucketing procedure) of the data-reduced NMR fingerprints of Albillo white wines fermented at two different conditions. Red and blue ovals (89% confidence intervals) represent respectively *H. vineae* and *S. cerevisiae* fermentation groups, each analyzed in triplicate.

**Figure 2.** Principal component analysis (PCA) of the fermentative volatile compounds.

#### *3.2. Volatile Compounds from the Alcoholic Fermentation*

Considering the total volatile compounds identified, *S. cerevisiae* produced a larger amount of volatile compounds (Table 3) with around 1200 mg/L. In this regard the concentration of acetaldehyde and 2,3-butanediol have a special importance. The amount of these compounds was significantly higher in the wines from *S. cerevisiae*. Similar results were obtained after the fermentation of artificial red must [27].

Both yeas<sup>t</sup> species did not show significant differences in the sum of higher alcohols. It interesting to point out that other authors reported a decrease in higher alcohols after the fermentation of the Chardonnay grapes must have with *H. vineae* compared with that of *S. cerevisiae* [5].

The fermentation with *H. vineae* resulted in increases in acetate esters and some ethyl esters, like ethyl acetate with concentrations around 79 mg/L. These results are similar to the results obtained by [5].

2-Phenylethyl acetate is an ester with strong aromatic power and its perception threshold reported is 250 μg/<sup>L</sup> [28]. This compound is associated with fruity, floral and honey aromas [25]. The 2-phenylethyl acetate concentration was significantly higher in *H. vineae* wines than in *S. cerevisiae* wines (Table 3). This fact has been reported by several authors [25,29] who identified up to 50 times more abundance of this compound in wines fermented by *H. vineae*. However, no significant differences in 2-phenylethanol content were found. This can be due to the fact that there are significant differences between these two yeas<sup>t</sup> species in the acetylation step due to an increase in the copy number of the acetyl transferases genes in *H. vineae* [29].

In addition, the "odour activity values" (OAV) were calculated (see Table 3). It allows us to estimate the contribution of a specific compound to the aroma of the wine [30]. Among the compounds that have been identified, only ethyl acetate, 2-methyl-1-butanol, 2.3-butanediol, isoamyl acetate, hexanol and 2-phenylethyl acetate have obtained an OAV greater than one. It must again be emphasized the importance of the 2-phenylethyl acetate. This compound had 31.84 OAV and statistically higher concentrations in *H. vineae* than in *S. cerevisiae* wines. In this regard, the concentration identified as 2-phenylethyl acetate had an important organoleptic repercussion in the wines obtained by the fermentation of *H. vineae*, providing fruity, floral and honey aromas to these wines.

A principal component analysis (PCA) was done for the 15 volatile compounds identified after the fermentation process (Figure 2) and it allowed to differentiate the aromatic profile between the yeasts studied. The distribution was explained with the first two components. The compounds 2-phenylethyl acetate, ethyl acetate, 3-methyl-1-butanol, 1-propanol, hexanol, isoamyl acetate and methanol are associated positively with the PC1. A cluster including the wines fermented by *H. vineae* was found in the positive values of the PC1 with the highest concentration of these volatiles. It is noteworthy the contribution of the 2-phenylethyl acetate produced by the metabolism of this yeas<sup>t</sup> species; on the contrary, on the negative values of the principal component PC1, a cluster composed of the wines fermented by *S. cerevisiae* was identified, including the contribution of 2-phenyl ethanol and indicating the difference between the two yeas<sup>t</sup> species in the acetylation of this compound.



Different letters indicate values with statistical significant differences (*p* < 0.05). 1 In wines; 2 In hydroalcoholic solution 10% *v*/*v*; 3 In beer; 4 from [38]; 5 [39]; 6 [34]; 7 [34]; 8 [32].

#### *3.3. Intracellular Components and Polysaccharides Content Measured in the Ageing on Lees*

The relative measurement of the intracellular components release has been done by the UV absorbance at 260 and 280 nm [40,41]. These measurements correspond to the relative amount of nucleic acids and proteins, respectively [42].

Regarding the monitoring at 260 nm, the samples with HV yeas<sup>t</sup> lees showed the highest values during the entire ageing period. However, the SCG37 samples showed the lowest absorbance values without significant di fferences with SP938 through the AOL stage. It is also interesting to note the di fference between the two *Saccharomyces* strains studied, the SC7VA samples showed absorbance values around 0.4–1 AU, while the lees of the yeas<sup>t</sup> SCG37 resulted in lower values, around 0.1–0.2 AU. These results may indicate that the same yeas<sup>t</sup> species can show di fferent capacities for releasing cellular compounds depending on the strain used.

Similar results were obtained in the monitoring of 280 nm absorbance, but in this case no significant di fferences were obtained between the HV and SC7VA samples during the 91 days of ageing. These values indicated that both yeasts could be used to accelerate the release of cellular compounds. Therefore, the use of HV and SCVA yeas<sup>t</sup> strains could be indicated to perform an AOL process.

The polysaccharides released after the action of glucanases are a good indicator of the autolysis process, being the parietal mannoproteins the majority of these polysaccharides [12]. After 156 days of ageing, the samples on SP938 lees have shown the highest content of polysaccharides with values around 23.5 mg/L. This quick releasing of compounds from the *Schizosaccharomyces* cell wall has already been observed by other authors [12]. It is interesting to stress the fact that the SP938 samples did not show the greatest absorbance values at 260 and 280 nm (Figure 3). This is possibly due to the fact that the high molecular weight polysaccharides do not have absorbance at these wavelengths as nucleic acids and proteins.

The HV samples showed a polysaccharides content of around 11 mg/L; this concentration was not statistically significant with respect to samples aged on the lees of the two *Saccharomyces* yeas<sup>t</sup> strains (Figure 4). In the same way, it was not significantly di fferent from the results obtained in L31 samples. The results obtained in the hydroalcoholic solution of these three yeas<sup>t</sup> species were similar to the result of other assays with *Saccharomyces* previously done [13]. In other words, the yeas<sup>t</sup> *H. vineae* could be an alternative to replace *S. cerevisiae* yeas<sup>t</sup> in an AOL process after the alcoholic fermentation.

**Figure 3.** Evolution of the absorbance at 260 nm (**a**) and at 280 nm (**b**) in hydroalcoholic solutions, throughout 156 days of ageing on lees. HV (*Hanseniaspora vineae*); SP938 (*Schizosaccharomyces pombe* strain 938); SCG37 (*Saccharomyces cerevisiae* strain G37); L31 (*Lachancea thermotolerans* strain L31); SC7VA (*Saccharomyces cerevisiae* strain 7VA). Mean ± standard deviation of three replicates. Different letters in the same day indicate values with statistically significant differences (*p* < 0.05).

**Figure 4.** Polysaccharides content (mg/L) after 156 days of ageing on lees in hydroalcoholic solution. HV (*Hanseniaspora vineae*); SP938 (*Schizosaccharomyces pombe* 938 strain); SCG37 (*Saccharomyces cerevisiae* G37 strain); L31 (*Lachancea thermotolerans* L31 strain); SC7VA (*Saccharomyces cerevisiae* 7VA strain). Mean ± standard deviation of three replicates. Different letters indicate values with statistically significant differences (*p* < 0.05).
