Introduction

The microbiology of wine fermentation has been extensively studied, revealing the intricate complexity of fermentation ecology (Romano et al., 2019.). Traditional wine production relies on spontaneous fermentation, with indigenous yeast strains naturally present on grape surfaces and winery equipment driving the process and imparting unique organoleptic characteristics. However, due to challenges linked to reproducibility, quality control, and traceability, spontaneous fermentation has declined in modern wineries (Pinto et al., 2020).

In light of the aforementioned issues, many winemakers use selected Saccharomyces cerevisiae strains as commercial starter cultures, thus ensuring more controlled and predictable fermentations. While these starter cultures offer practical benefits, their widespread use across different regions leads to the standardisation of fermentative flora. This homogenisation reduces the complexity of wine, masking the distinctive aromas and terroir-specific attributes that define regional identity (Binati et al., 2019; Rainieri & Pretorius, 2000). In recent decades, there has been a renewed interest in the study and use of indigenous yeast strains to preserve the uniqueness of the terroir, and thus produce wines that truly reflect their place of origin (Lappa et al., 2020). The connection between a wine's microbiota and its terroir highlights the relationship between the native microorganisms and the environmental conditions of a vineyard. Numerous studies highlight that, when co-fermented with Saccharomyces cerevisiae, non-Saccharomyces (NS) strains enhance fermentation dynamics and enrich the aromatic profiles of wine, thus reflecting a wine's regional identity and terroir (Jolly et al., 2014; Padilla et al., 2016). There are many reports on the properties and advantages of NS yeasts; however, these yeasts are generally developed within a microvinification framework at the laboratory scale, and the results are rarely validated at an industrial or semi-industrial scale, thus calling into question their applicability (Jolly et al., 2014). Finding new Saccharomyces and NS strains with promising oenological properties for use as starter cultures represents a major challenge today. Yeasts selected for their specific fermentative and technological capabilities play a vital role in optimising winemaking processes, enhancing the expression of terroir, and crafting wines with unique sensory profiles. Validating these strains at a semi-industrial scale is essential to ensure their effectiveness and applicability in producing high-quality wines.

In previous studies, we isolated, identified, and conducted physiological and biochemical characterisations of native yeasts from Tannat grapes in the Maldonado vineyard (Morera et al., 2022). Three isolated strains; Saccharomyces cerevisiae T19-3FS, Saturnispora diversa T19-1FS, and Starmerella bacillaris T19-3MS demonstrated high performance in laboratory conditions. In this study, we assessed the oenological potential and reproducibility of these native strains at a pilot scale in 200-L stainless tanks during the vinification of a Tannat must with an expected ethanol content of 12.9 %; we then explored the influence of the volatile compounds on the resulting wines.

Materials and methods

1. Yeast strains

Saccharomyces cerevisiae T19-3FS (S. cerevisiae 3FS), Starmerella bacillaris T19-3MS (S. bacillaris 3MS), and Saturnispora diversa T19-1FS (S. diversa 1FS) strains had been previously isolated from grapes of Tannat variety in the vineyard of Maldonado-Uruguay (Morera et al., 2022). The commercial S. cerevisiae yeast strain used as a reference was Zymaflore™ RX60 from Laffort (France). All the strains were maintained at 4 °C for short-term storage on a YPDA medium (20 g L^-1 bacteriological peptone, 10 g/L yeast extract, 20 g/L glucose, 15 g/L agar, Oxoid) in Petri dishes and preserved at 20 % glycerol in YPD (10 g/L yeast extract, 20 g/L peptone, 20 g/L dextrose) at –80 °C.

2. Obtaining biomass for semi-industrial-scale purposes

The yeast cultures were grown in the laboratory by refreshing each strain on YPDA plates and then they were incubated at 28 °C for 48 h. Then, 500 mL flasks were filled with 50 mL of YPD broth, which was inoculated with about 2 × 10⁶ cell/mL of each strain for 12 h at 28 °C and 180 rpm (preculture). For the fermentation trials in the Uruguayan winery Bodega Garzón, the starter culture for fermentation was prepared using the preculture in 5 L of YPD medium and incubated for 48 h at 28 °C and 180 RPM. The cells of each strain were then collected by centrifugation (5000 g, 15 min) and exhaustively washed with distilled water. Cell concentration in the pellet was determined by microscope in the Neubauer chamber. The collected biomass was preserved at 4°C for a restricted duration until utilised.

3. Semi-industrial fermentations

S. cerevisiae 3FS, S. bacillaris 3MS, and S. diversa 1FS strains were used in pilot scale fermentations in the Bodega Garzon Winery, Departamento of Maldonado-Uruguay. Each strain biomass was suspended and mixed in 1 L of Tannat grape must at room temperature, and then inoculated at a density of about 2 × 10⁶ cell/mL in a container with 200 L of Tannat grape must (200 g/L sugars, pH of 3.46, 7.2 g/L tartaric acid, and 120 mg/L yeast assimilable nitrogen ) at 21 ± 2 °C. Grape must was supplemented with 50 mg/L of sulfur dioxide in potassium metabisulfite form and 20 g/L of Thiazote (Laffort, France) once the fermentation had started. The inoculum of each strain in the tanks was carried out in duplicate. The fermentation was monitored daily by measuring density, temperature, and pH.

4. HPLC analysis of fermentation metabolites

Fermentable sugars (fructose and glucose) and fermentation metabolites (ethanol and glycerol) were quantified by High-Pressure Liquid Chromatography (HPLC Waters 510). A volume of 20 µL of sample diluted 1/10 (or 1/100 for must) was injected into a ROA-Organic Acid H+ 8 % column (300 × 7.8 mm) (Phenomenex) with a flow of 0.5 mL/min using 1.5 mM sulfuric acid at 25 °C. A coupled infrared (IR) (WATERS R401) detector was used and data processing was performed using the Empower software. All data are reported as the average of three replicates ± standard deviation. Quantification of compounds was carried out by the external standard method using the pure standard (HPLC quality) of each.

5. Analytical determination of volatile compounds

The extraction of free fraction (volatile) was carried out by solid phase extraction (SPE) in Manifold equipment using commercial cartridges Isolute ENV+ (Biotage) (Boido et al., 2003). For the quantification, 0.1 mL of internal standard (1-heptanol at 230 mg/L in a 50 % hydroalcoholic solution) was added to each sample before extraction.

Quantification and identification of isolated volatiles were performed using a gas chromatograph (GC) coupled to a Shimadzu QP 5050 mass spectrometer (GC-MS) equipped with reference libraries (Adams, 2001; Marais et al., 1992; Strauss et al., 1987). The chromatographic column used was Carbowax (25 mx 0.25 mm id) coated with polyethylene glycol (0.25 µm phase thickness). The programme was 60 °C (1 min) with a ramp from 60 °C to 240 °C (from 3 °C/min) and maintenance at 240 °C for 1 min. The injector temperature was 250 °C, the injection mode was splitless and the injected volume was 1.0 μL using He as carrier gas, at 92.6 kPa (55.9 cm/s), an interface temperature of 250 °C at 70 eV, and a masses range of 40–400 amu. Compounds were identified by their MS spectra and confirmed by the determination of their Kovats indices. The amounts were calculated as equivalents of the internal standard (Boido et al., 2003).

Results and discussion

1.Fermentation kinetics and major metabolites

The strains S. cerevisiae 3FS, S. bacillaris 3MS, and S. diversa 1FS, which had shown desirable analytical profiles in laboratory fermentations, were used in pilot scale fermentations of naturally processed grape Tannat must at the premises of a commercial winery. From previous experience and preliminary internal trials conducted on this must we knew that the applied conditions prevent spontaneous fermentation. The aim of our study was to evaluate the fermentation performance of selected native yeast strains and compare them under conditions that reflect standard winery practices, using the commercial S. cerevisiae strain RX60 as a reference. The results of this pilot scale experiment show that these strains are not only capable of completing the fermentation process within a similar time frame (11 days) to the commercial S. cerevisiae strain RX60, but they could also provide insights into their fermentative characteristics and potential for producing high-quality wines. As was expected and consistent with Morera et al. (2022), the three native strains showed similar fermentation behaviour to the reference strain S. cerevisiae RX60 (Figure 1). The kinetic similarities between the native strains and the reference strain S. cerevisiae RX60 are noteworthy, as they suggest that the native strains possess the robust fermentative power essential for successful fermentation in a commercial setting, and that they can thus be used independently for fermentation. The most surprising finding is the remarkable fermentative capacity of the NS strains S. bacillaris 3MS and S. diversa 1FS, which can complete fermentations independently.

In addition to confirming the high fermentative power of the native strains S. cerevisiae 3FS, S. bacillaris 3MS, and S. diversa 1FS in pilot-scale fermentations, it is crucial to assess the metabolites generated after fermentation and their potential benefits for commercial winemaking. Fermentation metabolites can directly impact the organoleptic characteristics and quality of wine; therefore, their evaluation in wine research and optimisation is key to providing insights into their applicability in larger-scale vinification processes. Table 1 provides the chemical compositions of the wines produced at a commercial winery using the three selected native strains at the end of each alcoholic fermentation and 11 days after inoculation with the yeast strains.

We assessed the oenological potential of these strains at a pilot scale during the vinification of a Tannat must with an expected ethanol content of 12.9 %. All three native strains were able to consume nearly all the fermentable sugars (glucose and fructose) present in the must after 11 days of alcoholic fermentation. S. bacillaris 3MS, S. diversa 1FS, and S. cerevisiae 3FS consumed around 99 % of the must sugars, resulting in robust ethanol production in the range of 9.65 to 11.58 % (Table 1). The S. bacillaris 3MS and S. diversa 1FS strains exhibited markedly different behaviour to other non-Saccharomyces (NS) strains as previously reported, highlighting the novelty and relevance of our findings - particularly considering that NS species are typically employed in co-fermentation with Saccharomyces to ensure successful fermentation (Binati et al., 2019; Canonico et al., 2016). Nevertheless, our results are consistent with previous reports, further indicating the robustness and fermentative potential of certain NS strains under specific conditions (Englezos et al., 2015; Bagheri et al., 2018).

Most of the analysed strains exhibited a preference for glucose (G) over fructose (F), confirming the glucophilic character of S. cerevisiae 3FS and S. diversa 1FS, except for S. bacillaris 3MS. The G/F ratios values were < 1 for S. diversa 1FS (0.50) and both of the S. cerevisiae strains (0.40-0.41) (Table 1). The exception was S. bacillaris 3MS (1.83), showing a significant preference for fructose rather than glucose; this is consistent with Magyar and Tóth (2011). The present study highlights the strong fructophilic character of this strain, which left very low levels of fructose residues (0.06 ± 0.05 g/L) in comparison with the other species studied. Glucose and fructose consumption profiles were considerably influenced by the yeast strains used as starters, and selecting strains with different glucose/fructose preferences could be very useful for the wine industry. Yeasts with a higher capacity for fructose consumption are of interest to the industry, as they help avoid undesirable sweetness in dry wines–given that fructose is approximately twice as sweet as glucose–and could also improve wine stability by minimizing residual fructose and reducing the risk of secondary fermentation (Rossouw & Bauer, 2016; Magyar & Tóth, 2011).

Climate change is driving an increase in the alcohol content of wines due to the rapid accumulation of sugars in grapes, posing significant challenges for winegrowers. To mitigate these effects, producers are adopting viticultural, technological, and microbiological strategies for producing balanced wines in the new climatic conditions, thus maintaining their quality and ensuring consumer satisfaction. NS yeasts in mixed or sequential fermentations with Saccharomyces cerevisiae are therefore gaining in popularity due to their ability to reduce ethanol production in wines besides providing aromatic compounds (Basso et al., 2016; Englezos et al., 2015). Some authors have proposed the potential use of S. bacillaris strains in combination with S. cerevisiae in wine fermentations (Capece et al., 2022); for instance, Giaramida et al. (2013) and Zara et al. (2014) demonstrated an increase in glycerol content and a decrease in alcoholic degree when S. Bacillaris was used in combination with S. cerevisiae at a pilot scale. Fermentation trials showed that S. bacillaris 3MS produced a significantly lower ethanol concentration (9.65 ± 0.18 %) than S. cerevisiae 3FS (11.91 ± 0.57 %) and S. cerevisiae RX60 (11.58 ± 0.33 %), confirming its lower fermentative capacity. Inoculation of Tannat must with S. bacillaris 3MS not only highlighted its marked fructophilic behaviour but also resulted in an approximate 2 % (v/v) ethanol reduction compared to the evaluated S. cerevisiae strains - which is of interest for the production of reduced-alcohol wines. The S. bacillaris strain produced a significantly higher glycerol concentration (9.15 ± 0.18 g/L) than S. diversa 1FS (8.19 ± 0.18 g/L) and S. cerevisiae RX60 (8.49 ± 0.42 g/L).

Different behaviour in the fermentation of NS and S. cerevisiae strains could significantly affect subtle differences in aromatic profiles and the overall complexity of the wine. As discussed extensively by several authors, native yeasts could enhance the expression of terroir characteristics by contributing to unique aromatic compounds, thus enriching the regional identity of the wine (Morera et al., 2022; Jolly et al., 2014). In the present study, volatile compounds were identified and quantified after the alcoholic fermentation of Tannat must from Uruguay. The Tannat variety, emblematic of the country, produces a robust wine known for its strong character, consistently standing out for producing high-quality wines with a unique identity. Thus, an SPE/GC-MS analysis was conducted on Tannat wines fermented with S. cerevisiae 3FS, S. diversa 1FS, and S. bacillaris 3MS, which were compared to a commercial yeast strain usually used in the winery, thus allowing volatile metabolites produced during yeast metabolism to be evaluated.

2. Volatile aromatic compounds

A total of 53 volatile compounds were identified and quantified in Tannat red wine following alcoholic fermentation. (Table 2). The volatile fraction was composed of acids (9 %), aliphatic and aromatic alcohols, including C6 (15 %), esters (23 %), lactones (11 %), phenols (8 %), norisoprenoids (4 %), terpenes (21 %) and other identified compounds (9 %). Regarding the acid fraction, it is well known that acids are formed enzymatically during fermentation and can contribute to floral, cheesy, fatty, and rancid notes. In the resulting Tannat wines, aliphatic acids (C4–C8) were identified and significant differences were found only between the S. diversa 1FS fermentation (530 ± 64 µg/L) and the other strains (< 358 ± 32 µg/L), indicating that S. diversa 1FS produces the highest amount of acids. Aliphatic and aromatic alcohols are key compounds that enhance the sweet, floral, and fruity notes in the aromas of wines. Regarding these alcohols, significant differences in total alcohols were found between S. diversa 1FS (1030 ± 272 µg/L) and the commercial and other native strains. The greatest significant difference was found in the production of 2,3-butanediol, with S. diversa 1FS producing the highest amounts (379 ± 102 µg/L). 2,3-butanediol is important as it can enrich the aromatic profile of red wine by adding a layer of sweet and creamy notes. The Saccharomyces strains were prominent in the production of C6 alcohols, exhibiting a profile closely resembling this group of compounds, which can evoke vegetal sensations. Esters confer pleasant fruit notes, and therefore these kinds of compounds are desired in wines. The strain that stood out for its total ester production was S. bacillaris 3MS (1133 ± 166 µg/L), with highly significant differences found for the ethyl hexanoate (21 ± 3 µg/L), ethyl octanoate (39 ± 2 µg/L), hexyl methoxyacetate (36 ± 3 µg/L) and diethyl succinate (482 ± 104 µg/L) relative to the commercial strain and other native strains. Additionally, fermentation with S. bacillaris 3MS led to the formation of ethyl 2-hydroxypropanoate (441 ± 49 µg/L) and ethyl octadecanoate (43 ± 4 µg/L); such concentrations were not detected in fermentations with the other native or commercial strains. Diethyl succinate is typically formed during alcoholic fermentation and is associated with aromatic notes reminiscent of wine, toffee, and fruits (Mendes et al., 2022). The origin of this compound may be linked to the catabolic pathway of glutamate, commonly known as the GABA shunt. In this pathway, glutamate is decarboxylated to γ-aminobutyric acid (GABA), which is then transaminated to succinate semialdehyde (SSA), and subsequently oxidised to succinate. When SSA is not oxidised, it can instead be reduced to γ-hydroxybutyric acid (GHB), a precursor for lactonisation. This step may explain why S. bacillaris also stands out for the production of γ-butyrolactone. The final product of the GABA shunt, succinate, can undergo double esterification with ethanol to form diethyl succinate (Pérez et al., 2022). Another strain that showed a notable production of diethyl succinate was the native S. cerevisiae, although it was the lowest producer of γ-butyrolactone. This therefore makes it a promising strain for enhancing these aromas in wines, as these compounds play a crucial role in the aromatic profile of red wines, contributing to their complexity, freshness, and balance. Lactones are among the most important compounds, as they contribute to the sensory characteristics of wines aged in oak wood (Pérez-Olivero et al., 2014). Except for the native S. cerevisiae, all of the strains stood out for producing these compounds, with the highest concentrations observed for S. diversa 1FS (260 ± 39 µg L^-1). Phenols play a crucial role in the aromatic profile of Tannat, a variety known for its intensity, structure, and complexity. S. bacillaris 3MS t produced the highest concentration of these compounds (129 ± 10 µg L^-1), and S. diversa 1FS and S. cerevisiae RX60 produced similar amounts (58 vs 63 µg L^-1, respectively). By contrast, the native S. cerevisiae 3FS did not produce any phenols during Tannat must fermentation. In Tannat, volatile phenols (4-vinyl guaiacol) and phenolic compounds related to vanillin are key contributors to specific aromas, such as spicy, smoky, and sweet vanilla notes. Norisoprenoids and terpenes are part of the glycosylated reserve of aromas. These compounds play a key role in wine aroma due to their very low olfactory thresholds and their ability to impart pleasant floral and fruity notes, even at low concentrations. S. bacillaris 3MS was the highest producer of norisoprenoids (9 ± 0.3 µg L^-1) and terpenes (52 ± 8 µg L^-1), underscoring the potential of this strain for enhancing these valuable aromatic compounds. Among the norisopreoinds compounds, 3-Oxo-7,8-dihydro-alpha-ionol sis a key compound in red wines due in particular to contribution to floral and fruity notes. Its presence can enrich the wine's aromatic profile, adding freshness, complexity, and elegance, and it plays a key role in the sensory perception of young and aged wines. Fermentation using S. cerevisiae 3FS was the only one to produce alpha-ionone (0.4 ± 0.1 µg L^-1), a highly significant compound in red wines due to its contribution to floral and fruity notes, which enrich the wine's aromatic profile. Significant differences in norisoprenoid concentrations were observed between the use of commercial and native strains of Cerevisiae in fermentations (0.4 and 6 µg/L, respectively); however, no significant differences were observed between their terpene fractions (26 and 21 µg/L, respectively).

Conclusion

The results of the present study align with those of growing research, which highlights the potential of using NS yeasts to improve fermentation outcomes while preserving wine typicity. Selected native yeast strains were evaluated through pilot-scale fermentations, resulting in finished wines that exhibited desirable oenological properties, including distinctive aromatic profiles and an appropriate chemical composition, thereby underscoring their potential for commercial application in regional winemaking. Furthermore, the use of these NS strains as a pure inoculum in must fermentation - without the need for co-fermentation with Saccharomyces cerevisiae represents a significant advancement in modern winemaking, offering an alternative to traditional methods that rely solely on commercial starters for their robustness and fermentative power. This approach could enable winemakers to fully harness the unique characteristics of NS yeasts, such as Starmerella bacillaris 3MS, and Saturnispora diversa 1FS, to enrich the aroma of wines. Not only does this approach promote biodiversity but it also fosters greater innovation and differentiation in winemaking, resulting in wines with enhanced typicity and authenticity. Furthermore, Starmerella bacillaris 3MS offers a promising strategy for producing wines with lower alcohol content while enhancing sensory quality. Its ability to reduce ethanol yield, increase glycerol production, and enhance aromatic complexity makes it a valuable tool, particularly in regions with rising grape sugar levels. Although this strain can successfully ferment independently, it would be useful to carry out future research on semi-pilot co-fermentation using Starmerella bacillaris and native S. cerevisiae 3FS, in order to compare alcohol levels and aromatic profiles.

Acknowledgements

The authors express their gratitude to Andrés González Ritzel from the Laboratorio de Ecología Qumica, Facultad de Química, Universidad de la República (Montevideo, Uruguay) for granting access to the GC-MS equipment. They also extend their thanks to Ana Ramón from Sección Bioquímica y Biología Molecular de la Facultad de Ciencias Universidad de la República for providing access to their equipment.