Effect of strains from different Saccharomyces species used in different inoculation schemes on chemical composition and sensory characteristics of Sauvignon blanc wine

Enhancing the sensory profile of wine using different microorganisms has always been a challenge in winemaking. The aim of our work was to evaluate the impact of different fermentation schemes by using mixed and pure cultures of three different Saccharomyces species at classic and cold inoculation temperatures on wine chemical composition and sensory profile. All the produced wines were analysed for the main oenological parameters (total acidity, volatile acidity, residual sugars, alcohol, and malic acid), as well as for volatile aromatic compounds, such as higher alcohols, acetate esters, ethyl esters and varietal thiols. In addition, implantation monitoring at the strain level and sensory analyses were performed in all fermentation trials. The wines fermented with S. pastorianus strains were characterised by significant lower acetic acid production and greater malic acid degradation compared to the wines fermented with S. cerevisiae strains. Interestingly, when the inoculation process for all trials was performed at a cold temperature (13 °C), the typical aromas of Sauvignon blanc were stronger than those in the classic inoculation temperature (18 °C) trials. Furthermore, the co-inoculation of S. pastorianus and S. bayanus resulted in wines exhibiting a higher intensity of citrus fruit notes than the wines fermented with the commercial S. cerevisiae strains. These results show that some strains from different Saccharomyces species other than S .cerevisiae, such as S. pastorianus and S. bayanus, could be promising starters for Sauvignon blanc wines and their performance can be modulated by the inoculation conditions.

Besides S. cerevisiae, other Saccharomyces species belonging to the Saccharomyces strict sensu group can be isolated from natural environments and be used in the production of alcoholic beverages (Rainieri et al., 2003). The nomenclature of the Saccharomyces species has been provided by Dujon and Louis (2017). Saccharomyces hybrids, like S. bayanus, are used in winemaking,because of their interesting oenological properties (e.g., high osmotolerance and cryotolerance) and their contribution to sensory attributes (Libkind et al., 2011;Salvadó et al., 2011;López-Malo et al., 2013). Moreover, the common lager yeast, S. pastorianus, a natural hybrid of S. cerevisiae and S. eubayanus (Martini and Martini, 1987), has been used in the production of a broad range of active compounds responsible for aroma in beer fermentation (He et al., 2014). Nevertheless, little research has been carried out on these non-conventional yeast species in wine fermentation, whether in pure or mixed cultures.s Sauvignon blanc is a French grape variety which is planted all over the worldand which expresses varietal aromas of tropical and citrus fruits, as well as green nuances, due to the production of thiols, esters, higher alcohols and methoxypyrazines (Tominaga et al., 1998;Šuklje et al., 2016). The impact of microorganisms on the Sauvignon blanc aromatic profile has been well-documented Masneuf-Pomarède et al., 2006;Sadoudi et al., 2012;Šuklje et al., 2016;Hart et al., 2017). Most of the research has focused on the enzymatic action of Saccharomyces or non-Saccharomyces species in relation to the precursors of volatile thiols, cysteinylated or glutathionylated conjugates, which seems to be strain and culture dependent Zott et al., 2011).
This study evaluated the use of three Saccharomyces species in different fermentation schemes as a meansof modulating Sauvignon blanc varietal typicity. Under-developed strains of S. pastorianus and S. bayanuswere tested in pure and mixed cultures at two inoculation temperatures for the first time. In parallel, commercial S. cerevisiae strains were used for the control conditions. The consequences of all the fermentation schemes on volatile compounds and the sensory profiles of the produced wines were analysed. The use of non-conventional yeast species in winemaking processes can

INTRODUCTION
Saccharomyces cerevisiae is a commonly used wine yeast species which drives alcoholic fermentation,has a high fermentation rate and contributes to the enrichment of wine with positive and attractive aromas. The use of selected strains of S. cerevisiae to drive alcoholic fermentation is a technique which has been applied for years to assure fermentation flow and sugar depletion. More recently, it has been observed that, due to global climate change, grapes are starting to accumulate more sugar at harvest -when their technological maturity has been reached -than in the past. This phenomenon could lead to winemaking problems, such as stuck or delayed fermentation, as the yeasts are not yet well-adapted to the new stressful wine environment (Fleet, 2008;Mira de Orduña, 2010). Furthermore, the wine market's demand for wines with increased sensory properties, such as aromatic complexity and varietal typicity, has turned interest towards non-Saccharomyces (nSc) and other nonconventional yeast species (Fleet, 2008;Sadoudi et al., 2012;Albertin et al., 2019). The nSc yeasts are characterised by high genetic diversity,and some species and strains can be detrimental to wine quality, while others can be beneficial (Renault et al., 2009;Tempère et al., 2018). Representative genera of the latter category detected in early fermentations are Metschnikowia, Candida, Hanseniaspora, Rhodotorula and Pichia (König and Berkelmann-Löhnertz, 2017). In spontaneous fermentation, wild yeasts are metabolically active at the beginning of the fermentation process,and their population starts to decline as the ethanol level increases. In order to mimic the natural evolution of this yeast species during the winemaking process, as well as to ensure the quality of the end-product, the selected strains of nSc yeast are used in co-inoculation or sequential inoculation with the S. cerevisiae strain (Bely et al., 2008;Ciani et al., 2010;Comitini et al., 2011;Domizio et al., 2011;Benito et al., 2015). Nevertheless, some interactions, including cell to cell conduct, production of antimicrobial molecules and antagonism for nutrients, have been observed; therefore, the compatible co-existence of nSc species with S. cerevisiae in winemaking is not easy (Sadoudi et al., 2012;Branco et al., 2015;Englezos et al., 2018, Englezos et al., 2019Petitgonnet et al., 2019).

Wine fermentations and vinification protocols
80 kg of white grapes (Sauvignon blanc variety, 2018) was crushed and pressed, and 80 mg/L of potassium metabisulfite was added to the resulting must. The must then underwent clarification by cold/gravity settling for 12 hours, after which it was decanted. Almost 50 L of clarified must was prepared and split into 22 bottle fermentors with almost 2 L per bottle. Twenty-two experimental fermentations (11 duplicated trials) were conducted by adding different strains in pure and mixed cultures and using two different inoculation temperatures. For the cold conditions, yeasts were directly inoculated when the temperature was 13 o C, while for the classic conditions, yeasts were inoculated when temperature reached 18 o C. In all cases, the yeasts were inoculated in the must at 1.10 6 CFU/mL. In the case of mixed cultures, a proportion of 70/30 (S. pastorianus/S. bayanus) was chosen as S. pastorianus cells tend to flocculate. Twenty-four hours after yeast addition, 200 mg/L of SpringFerm™ (Fermentis, France) was added. Upon completion of one third of fermentation, 240 mg/L of diammonium phosphate (DAP) and 200 mg/l SpringFerm™ (Fermentis, France) were added. All the strains used were produced by Fermentis, France. The strains and the corresponding codes are shown in Table 1.

Microbial growth
Samples for all the fermentation trials were taken at the time of inoculation, after 48 hours, and when two thirds of the sugars were consumed. 1 ml of each sample was transferred to 1.5 ml tubes and the cells were spun down at 4000 rpm for 3 min. The supernatants were discarded and the cell pellet was washed three times with 1 ml sterile distilled water. Finally, the cells were suspended in 1 ml water and serial dilutions were performed on Wallerstein Laboratory Nutrient agar (WLN) and Lysine medium agar for the numeration of total yeast and wild yeast respectively. The incubation of the plates was carried out at 28 o C and 37 o C, as S. pastorianus cannot grow above 30 o C. The plates of all samples were incubated at both temperatures in order to determine the cultivability of all tested species.

Sensory analyses
The sensory assessment was carried out by a group of eight trained and experienced judges. The panelists attended four training sessions over a period of two weeks. The training period included a general initial session in which the panelists were served samples for assessment and asked to identify the sensations and aromas perceived. The selected attributes were grouped into two categories: olfactory descriptors (citrus Maria Dimopoulou et al. whole reaction volume was loaded onto 2 % agarose gel containing 0.01 % Midori Green in 1xSB buffer (5 mM Sodium borate decahydrate) in order to separate the PCR products. The agarose gel was run at 120 Volt for 35 min in 1xSB buffer. DNA profiles obtained from the analysis of Ty1 retrotransposon elements can be used as genetic markers for polymorphism identification, as these DNA elements differ in number and location depending on the Saccharomyces strain (Legras and Karst, 2003). As long as there are interdelta retrotransposons in the genome inherited from S. cerevisiae, this method can amplify other species like S. bayanus and S. pastorianus and thus be used to determine if the strains concerned were present or not. In this study, a Ty1 analysis was performed to distinguish the different strains used, rather than to extract data regarding the origin of each strain.

Analysis of oenological parameters
The must was immediately analysed for the following parameters: glucose/fructose, total acidity, volatile acidity, pH, L-malic acid, Yeast Assimilable Nitrogen (YAN), free and total SO 2 and turbidity using OIV methods (OIV, 2015).

Quantification of higher alcohols and esters
Fermentation volatile aromas were analysed by GC-MS using the HeadSpace Solid-Phase Microextraction (HS-SPME) methodology. A volume of 25 ml of wine spiked with the internal standard 25 μg of 3-octanol, was placed in a 40 ml vial, to which 3 g of NaCl and a magnetic stir bar were added. The vial was then sealed with a screw-top cap with a silicon septum and placed on a heated stir plate. The solution was equilibrated by magnetic stirring at 750 rpm for 5 min at 30 ºC. The SPME needle was then manually inserted into the vial septum, and the fiber (DVB/CAR/PDMS, 75 μm) was exposed to the headspace of the sample for 30 min at 30 ºC. The fiber was then retracted, and the SPME device was removed from the vial and inserted into the GC injector for thermal desorption for 10 min. Analysis was performed using an Agilent 7890A GC, equipped with an Agilent 5873C MS detector. An HP-5 capillary column (30 m × 0.25 mm i.d., 0.25 μm film thickness) was used and the gas carrier was helium with a flow rate of 1 mL/min. The injector and MStransfer line were maintained at 250 ºC and  fruit, herbaceous, tropical fruit, esters and reduction) and gustative descriptors (acidity, balance, body, after-taste and quality level). During the subsequent sessions, the panelists were trained using appropriate solutions. The tests were conducted in individual booths and each sample was served in random order. The panelists were provided with 30 ml of wine in ISO wine glasses, coded with random three-digit numbers at room temperature (18-20 °C). The intensity of the sensory attributes examined was evaluated using a 10-point scale (1: null; 10: very strong).

Statistical analyses
Volatile compound concentrations (micrograms per liter) were expressed as mean ± standard deviation. For each experiment, a first one-way ANOVA was performed using R (R Development Core Team 2010). ANOVA was followed by a Tukey post hoc test to identify differences between groups, using a 95 % confidence interval.

RESULTS
At the beginning of the study, one yeast strain of each species in pure and mixed cultures, was inoculated at two inoculation temperatures: the classic inoculation temperature (18 ºC) and the cold inoculation temperature (14 ºC). As the cold inoculation gave more promising results for the strain of S. pastorianus, a second strain of the species was added to test the effect of cold inoculation. In order to present the findings in a more comprehensible way, the Results Section is divided according to the inoculation temperature.

Yeast population dynamics and fermentative behaviour of pure and mixed cultures of different Saccharomyces species in classic inoculation conditions
Fermentations of clarified and slightly sulfited Sauvignon blanc must were conducted at laboratory scale using two different Saccharomyces species in pure and mixed cultures. A fermentation trial conducted with a pure culture of Saccharomyces cerevisiae was used as a comparative control. The grape must was inoculated with pure cultures of Saccharomyces pastorianus strain 1 (Sp1), with pure cultures of Saccharomyces bayanus strain 1 (Sb1), with a mixed culture of S. pastorianus strain 1 and S. bayanus strain 1 in a ratio of 70/30w/w (Sp1Sb1), and with the pure culture of S. cerevisiae strain1 (Sc1) ( Table 1). All the inoculums of the dry inactivated yeast were first weighed in order to achieve exact ratios of the  concentration of glucose and fructose was calculated at the end of the fermentation process in order to confirm the completion of each fermentation trial and to calculate the duration of the process. Oenological parameters were analysed in all final wines ( Table 2).
The wild yeast population level was high at the beginning of all fermentation trials (4±0.5 Log CFU/mL) but started to decrease during the fermentation process (Figure 1). The wild yeast could not be detected after five days in the control culture (Sc1), and after seven days in the other cases, showing that the inoculated strain(s) dominated the fermentation process. The highest population level (9 ± 0.35 Log CFU/mL) was reached by Sb1 after 72 h. The control entered the stationary phase within the first 38 h,in contrast to the other schemes which reached the stationary phase after 72 h. The Ty1 sequence analysis ( Figure 2) for the S. bayanus strain in the Sb1Sp1 mode showed that its population was three times higher than the S. pastorianus population after 48 h and was slightly elevated at the end of the fermentation, as resulted by the colonies tested. Given that there are more than twice the living cells perg in the dry S. bayanus product than in the S. pastorianus (data not shown), the initial ratio of inoculation in terms of living cells was about 50 % S. bayanus and 50 % S. pastorianus; this means that Sb1 took the lead during the fermentation, whereas Sp1 remained until the end of the fermentation. Maria Dimopoulou et al. strains. They were then activated and the fermentation processes, as well as the population dynamics of the inoculated yeast and the wild yeast, were monitored. The fermentation kinetics was evaluated by monitoring the density of each culture, and the population dynamics was estimated until the first week of fermentation, when at least 70 % of the total alcohol was produced (Figure 1). In addition, the Fermentative behaviour was different for all schemes (Table 2). After eight days of fermentation, the remaining sugars were 16 ± 1 g/L for Sp1Sb1, 72 ± 0.4 g/L for Sp1, 5 ± 0.3 g/L for Sb1 and 19.5 ± 0.5 g/L for Sc1. In all cases, the remaining sugars after the first week of fermentation comprised fructose in a proportion of more than 80 % compared to glucose. The duration of the fermentation process was 9 days for Sb1and 11 days for Sp1Sb1. Both cases were significantly (p < 0.05) shorter than the fermentation of Sp1 (16 days).
In the control conditions (Sc1), the fermented sugar was consumed within 13 days, showing an intermediate fermentation capacity compared to the tested conditions. All the trials achieved dryness, but differed in fermentation rates. The presence of S. pastorianus in either mixed or pure culture significantly decreased the fermentation rate compared to S. cerevisiae (control).
In both pure or mixed cultures, the two Saccharomyces species, S. pastorianus and S. bayanus, produced less (p < 0.05) acetic acid than the pure culture of S. cerevisiae. Even if the level of volatile acidity was not high (0.33 ± 0.01) for the control condition, the tested fermentation schemes with the other Saccharomyces species expressed lower levels of production, almost close to zero. Furthermore, these three inoculation modes (Sp1Sb1, Sp1 and Sb1) attained higher levels (p < 0.05) of malic acid degradation than in the control conditions (Sc1).

Overall production of volatile compounds from pure and mixed cultures of Saccharomyces species at classic inoculation temperature
All the tested schemes were significantly different regarding their capacity to release volatile thiols. The best producer was Sp1 which released almost three times more than Sb1 (Table 3). Out of all the tested thiols, 3-mercaptohexanol-1 was found in the highest concentrations,with Sp1 containing the highest concentration (877.77± 8.14 ng/L). SP1 also produced the highest amounts of 4-methyl-4mercaptopentan-2-one (9.32 ± 0.5 ng/L). The control (Sc1) was the second best producer for the three tested volatile thiols. Regarding total higher alcohols, Sp1 contained the most (26.07 ± 0.5 ng/L) and was significantly different from the control wine. 2-phenylethanol was the alcohol that discriminated the four tested schemes, in which Sp1 produced 4 times more than Sc1. The Sp1Sb1 and Sb1 schemes showed an intermediate production capacity of higher alcohols. The last group of aromatic compounds analysed was the esters. The wine fermented with the mixed culture of S. pastorianus and S. bayanus contained the greatest amounts of tested esters (9.30 ± 0.75 mg/L in total). This predominance was mainly due to the high level of isoamyl acetate (7.14 ± 0.57 mg/L), the The results of the Ty1 sequence analysis (Figure 2) show that in both the CldSp1Sb1 and CldSp2Sb1 schemes there was 30-40 % of S. pastorianus at the beginning and after 48 h of inoculation, and around 30 % at the end of fermentation; i.e.,it was slightly above, but similar to classic inoculation populations, as resulted by microbiological analysis The tested inoculation schemes showed different fermentation behaviour. Fermentation lasted 11 days in all schemes, except for the two cases of inoculation with pure culture of S. pastorianus (CldSp1 and CldSp2) for which the duration was significantly (p < 0.05) longer (16 and 17 ± 1 days). In both the CldSp1 and CldSp2 schemes, after 13 days of fermentation the sugar of the medium was only fructose at a concentration of 10 ± 3g/L, while glucose had been depleted. The first plateau of the Saccharomyces yeast was reached by CldSp1 within 48 hours of inoculation, while the highest yeast population was recorded by CldSp1Sb1 (8 ± 1.12 Log CFU/mL). The Saccharomyces yeast dynamics were not really different in the tested schemes. On the other hand, the wild yeast of the initial Sauvignon blanc must followed a different population kinetics depending on the inoculation mode. When the two S. pastorianus Maria Dimopoulou et al. compound representing around 75 % of the total examined esters. S. pastorianus performed the least well regarding the production of volatile esters, although it produced two to three times more 2-phenylethyl acetate in our fermentation trials of Sauvignon blanc,which is most certainly linked to its very high production of 2-phenylethanol.

Yeast population dynamics and fermentative behaviour of pure and mixed cultures of different Saccharomyces species under cold inoculation conditions
The laboratory fermentations of Sauvignon blanc must were carried out and analysed as previously described. For this experiment, the inoculation temperature was 13 °C (cold condition), and extra Saccharomyces strains were tested in order to examine possible intra-species differentiation. More precisely, a second S. pastorianus strain was tested in pure culture (Sp2) and in mixed culture with 30 % S. bayanus (Sp2Sb1). A second strain of S. cerevisiae in pure culture (Sc2) ( Table 1) was used as well. The wild yeast population kinetics was estimated in all the trials. The population dynamics and the fermentative behaviour of the examined schemes are shown in Figure 3 and Table 4.
strains were present, either in pure or mixed cultures, the wild yeast was able to maintain its high population (average 5 ± 1.15 Log CFU/mL) during the first 48 h of inoculation of the Saccharomyces strains, while in the other cases their population lost more than 1 Log CFU/mL. In all cases, the wild yeast completely depleted after one week into the fermentation process.
A ''blank''fermentation (not inoculated) was also included in this study. The yeast cells responsible for the fermentation of the inoculated sample were mostly either non-Saccharomyces species that did not reveal any DNA profile, or Saccharomyces species different to the inoculated species, which gave completely different DNA profiles after analysis of their Ty1 sequences (data not shown).
The analysis of the oenological parameters in the final wines revealed important differences between the tested inoculation schemes ( Table 4). The wines fermented with the two pure cultures of S. pastorianus (CldSp1 and CldSp2) had low volatile acidity, with concentrations going as low as zero (acetic acid g/L) and significantly lower (p < 0.05) values than the other schemes, especially the two control conditions (CldSc1 and CldSc2). In addition, the levels of glycerol in the final wines increased to as much as 6.1 ± 0.14 g/L in the CldSp1 and CldSp2 trials. The degradation of malic acid was highest (39 %) in CldSp2 and lowest (p < 0.05) in CldSc1 and CldSc2 (20 % and 28 % respectively). The fermentation with CldSc2 produced the most SO 2 (36±1.41 mg/L), four times that in the CldSb1 and CldSp2Sb1 trials.

Overall production of volatile compounds of pure and mixed cultures of Saccharomyces species under cold inoculation conditions
The seven fermentation schemes under the cold inoculation conditions were also examined for the three categories of aromatic compounds that have an impact on wine sensory characteristics ( Table 5). The produced amounts of varietal thiols significantly discriminated all the tested inoculation schemes. The best and the worst producer were the two schemes with S. cerevisiae, Sc2 and Sc1, which produced 1744.95 ± 25.22 ng/L and 748.95 ± 18.12 ng/L respectively. The mixed culture with the S. pastorianus 2 strain produced 256 ng/L more of total thiol compounds in the final wine than the mixed culture with S. pastorianus strain 1, whilet he two pure independent cultures of S. pastorianus, Sp1 and Sp2, did not producesignificantly different total amounts.
The concentration of higher alcohols in the final wines ranged from 24.75 ± 016 mg/L (CldSc1) to 29.08 ± 1 mg/L (CldSp1). The schemes of pure cultures of S. pastorianus strains 1 and 2, as well as those of S. cerevisiae strains 1 and 2, were significantly different. Conversely, when both strains of S. pastorianus were inoculated with the strain of S. bayanus, the produced wines did not differ in alcohol level. No differentiation was shown between the different fermentation schemes for both2-methyl-1-butanol and hexanol-1.
The seven fermentation schemes could be differentiated regarding the concentration of fermentation esters in the final wines. When the two strains of S. pastorianus were inoculated in pure cultures (Sp1 and Sp2), the concentration of the ten ester compounds was significantly lower than the corresponding mixed cultures with S. bayanus. More precisely, the strain S. pastorianus 1 produced 2.07 mg/Lmore total esters in the CldSp1Sb1 scheme than in CldSp1, and S. pastorianus 2 produced 2.36 mg/L more aromatic esters in the CldSp2Sb1 scheme than in CldSp2. The fermentation of wines with both S. cerevisiae strains resulted in significantly different concentrations of fermentation esters being produced.

Specific trends in wine sensory properties under different inoculation schemes
The Sauvignon blanc wines produced using the different fermentation schemes described above were all evaluated by the sensory expert panel. Figure 4 shows the mean scores of ten sensory Values with different superscript roman letters (a-e) in the same row are significantly different according to Tukey's post hoc test (p < 0.05). characteristics, with a maximum score of 10, of the wines produced with cold and classic inoculation schemes, either in mixed or pure cultures. The statistical analysis (ANOVA) indicated a significant difference (p < 0.05) between the two inoculation schemes for citrus and tropical fruit aromas, as well as for the reduction level of the produced wines. Specifically, when the inoculation temperature was 13 °C (cold), the produced wines fermented using the Saccharomyces species expressed the fruity aromas of citrus and tropical fruits more. Furthermore, this fermentation scheme minimised the undesirable notes of wine reduction. Figure 4 shows the effects of the different fermentation schemes in the cold inoculation conditions on the sensory profile of Sauvignon blanc wines, which were evaluated using ten descriptors. All the fermentation schemes expressed similar profiles, except for the mixed culture of S. pastorianus strain 1 co-inoculated with S. bayanus strain1. According to the ANOVA, the wine inoculated with CldSp1Sb1 showed a significant increase (p < 0.05) in notes of citrus fruits and decrease in herbaceous notes in comparison with all the other fermentation schemes.

DISCUSSION
Sauvignon blanc has become an international variety with distinctive sensory attributes, which is cultivated all over the world and makes great financial profits for the wine industry. Obtaining high quality wines which will keep their varietal typicity could be directly associated with the microorganisms and the fermentation processes used in winemaking. As most wild yeasts are not able to ferment all the sugars in the must, and as their co-existence with S. cerevisiae strains is not always compatible, the use of other species has become much sought-after. Our work investigated the use of non-conventional yeast strains from different Saccharomyces species in the production of Sauvignon blanc wine. At the beginning, one strain of S. pastorianus and one of S. bayanus were chosen to ferment Sauvignon blanc must at the classic inoculation temperature (18 °C). However, the quality of the wines produced using the different fermentation schemes (including a mixture of both strains) was found to compete well with the wine fermented using the commercial S. cerevisiae strain; the experiment was therefore expanded to include an additional strain of S. pastorianus in pure culture and in co-inoculation with S. pastorianus, as well as a second control strain of S. cerevisiae. Moreover, the inoculation temperature for all the trials was decreased to 13 °C to assess theimpactofcold conditions on the aromatic intensity of wine.
Under classic inoculation conditions (18 °C), the mixed culture of S. pastorianus and S. bayanus were of intermediate oenological interest compared to the wines fermented with the pure cultures of the species. The values for the analysed parameters of the mixed culture were closer to the ones of the pure culture of S. bayanus. This was expected as a DNA fingerprint analysis has revealed S. bayanus to be predominant in the co-inoculation scheme. For instance, the inoculum ratio has been shown to have a decisive influence on wine characteristics in the case of mixed cultures of Saccharomyces and non-Saccharomyces species Satora et al., 2018).
Compared to the tested commercial S. cerevisiae strain in all the fermentation trials, the presence of both of these non-conventional yeast strains, either in pure or in mixed cultures, contributed interesting properties to the produced wines, such as significantly lower volatile acidity and medium de-acidification via malic acid degradation. Torulasporadel brueckii is the most reported species able to reduce the volatile acidity of wine in mixed cultures with S. cerevisiae (Bely et al., 2008;Renault et al.;2009, Ciani et al., 2010Ciani and Comitini, 2015). Saccharomyces stricto sensu strains have been reported be able to degrade malic acid via the maloethanolic pathway (Redzepovic et al.;Morata et al., 2019). S. cerevisiae can hardly metabolise malic acid, especially compared to the relative enzymatic activity of the species of Schizosaccharomyces pombe. In contrast with our results, it has been previously reported that S. bayanus strain EC1118 can only degrade 8 % of the malic acid during alcoholic fermentation, while S. pastorianus can mainly synthesize malic acid rather than degrade it (Redzepovic et al., 2003). Substantially more strains of each species should be tested under various vinification conditions for their ability to perform maloethanolic fermentation.
As far as the control trials were concerned, both S. cerevisiae strains exhibited high intra-species variability for almost all the performed analyses, a phenomenon not observed with either of the   S. pastorianus strains. Even if our results come from a very low number of tested strains, this observation could be explained by the high genome plasticity of S. cerevisiae reflected in the genotypic and phenotypic differentiation between the species (Albertin et al., 2015). Indeed, the species has a remarkable capacity for adaptation to selective pressures deriving mainly from environmental factors (Legras et al., 2018). Masneuf-Pomarède et al. (2006) showed that fermentation temperature plays an essential role in the concentration of released thiols, which for all tested S. cerevisiae strain swas found to be higher at a fermentation temperature of 20 °C than that of 13 °C. Moreover, when the development of S. cerevisiae was delayed and the wild yeast were dominant in the must of Sauvignon blanc, the concentration of thiols increased, especially that of 3-mercaptohexan-1ol (3MH) (Zott et al., 2011). However, the effect of inoculation temperature had never been tested before. In our work, we have shown for the first time that a cold inoculation temperature can increase the production of the examined thiols in all tested fermentation schemes with strains from different Saccharomyces species. This significant difference between the two inoculation modes was also confirmed by the sensory analyses, with all wines inoculated at cold temperature containing more citrus and tropical fruits notes than the wine inoculated at classic temperature. Additionally, this work is also the first to show that non-conventional yeast species can increase the aromatic typicity of Sauvignon blanc. It would be interesting in the future to study the activity of β-lyase (one of the enzymes involved in the release of thiols (Belda et al., 2016) at different inoculation and fermentation temperatures.

CONCLUSIONS
In the present study, we investigated the impact of the two non-conventional yeast species, S. pastorianus and S. bayanus, in mixed and/or pure cultures on the fermentation and sensory profiles of a Sauvignon blanc wine. We evidenced that the applied temperature at the inoculation phase can significantly affect the varietal notes of Sauvignon blanc for all tested fermentation schemes at both the chemical and sensory levels. Furthermore, when the Sauvignon blanc must was co-inoculated with S. pastorianus and S. bayanus cultures, the produced wines expressed higher citrus fruit notes. Our results show the promising oenological potential of the non-conventional yeast strains. In the future, it would be interesting to study more strains of S. bayanus and S. pastorianus in pure or mixed cultures with different inoculum support.