Yeast strains used in this study were obtained from the yeast culture collection maintained at EVEGA-AGACAL. The autochthonous yeast strains were isolated from grapes and musts from organic and conventional vineyards in Galicia over several years, identified as explained in [22] and stored at EVEGA. Within this yeast collection, 29 non-Saccharomyces strains belonging to Metschnikowia spp. (10 strains), Starm. bacillaris (10 strains), L. thermotolerans (6 strains) and T. delbrueckii (3 strains), were chosen due to the oenological interest of these species in recent literature [2, 3, 23, 24]. Metschnikowia pulcherrima and Metschnikowia fructicola are two genetically very close species with a very similar oenological potential that are difficult to identify, so they have been considered in the same group [13]. Depending on the availability of indigenous strains isolated from each of the relevant species mentioned above, between 3 and 10 isolates were evaluated (Table 1). The use of strains isolated from different Denominations of Origin (DOs) and from different plots and cultivation systems within the same species allowed to explore different oenological behaviours to select the most appropriate ones. In addition, the S. cerevisiae XG3 strain, previously tested and selected at EVEGA, was used as a control [25, 26].

The original culture of each yeast strain was streaked on WL nutrient agar medium to isolate single colonies. Then, a single colony was used for pre-inoculum growth (for 24 h at 28 °C) with the aim of obtaining sufficient biomass of fresh growth. The number of cells in the preinocula was quantified by measurement of the optical density at 600 nm in a spectrophotometer. Microfermentations were carried out in duplicate, using 75 mL of pasteurized (10 min at 100 °C) white grape must in 100 mL sterilized bottles closed with sterile-venting membrane screw caps. The strains in pure culture were inoculated in flame-sterile conditions at a concentration of 5 $\times$ 10${}^6$ CFU/mL in the pasteurised must. White thawed must (obtained from a mix of traditional white grape cultivars from Galicia) with the following characteristics was used: probable alcohol content: 11.5 %vol., total acidity: 6.3 g tart. L${}^{-1}$, malic acid: 4.1 g·L${}^{-1}$, tartaric acid: 3.1 g·L${}^{-1}$, sugars: 188 g·L${}^{-1}$, probable alcoholic strength: 11.5 %vol., soluble solids: 19.6 °Brix, and total SO${}_2$: 15 mg·L${}^{-1}$. The chemical analyses of the must after pasteurization remained constant; only malic acid (4.2 g·L${}^{-1}$), tartaric acid (3.0 g·L${}^{-1}$) and total SO${}_2$ (8 mg·L${}^{-1}$) parameters varied slightly.

Fermentations were carried out at a controlled temperature of 17 °C $\pm$ 1 °C for 31 days. The fermentative ability was monitored by daily measurement of weight loss caused by CO${}_2$ evolution. At the end of fermentation, the presence of the added strain in each assay was confirmed by spreading 100 $\mu$L of wine on WL medium to assess colony morphology.

Wines were analysed in the chemistry laboratory of EVEGA immediately after the end of fermentation. General parameters (total acidity, volatile acidity, lactic acid, malic acid, glucose + fructose, glycerol, alcohol content and pH) were determined by Fourier transform infrared spectrometry (FTIR) using a Wine Scan FT120 analyser (FOSS Electric, Barcelona, Spain) calibrated according to [27]. In addition, the ethanol yield was calculated as the ethanol production (g·L${}^{-1}$) per sugar consumption (g·L${}^{-1}$).

Volatile compounds including higher alcohols (methanol, propanol, isobutanol, 1-butanol, 2-methyl-1-butanol and 3-methyl-1-butanol), acetaldehyde and ethyl acetate were quantified in an Agilent model 7890A (Palo Alto, CA, USA) chromatograph with flame ionization detector (FID) as described by [28].

Significant differences between samples were tested by variance analysis (ANOVA) and Tukey tests (p 0.05) using SPSS software (version 18.0, PASW Statistics, Chicago). The data obtained from the basic chemical and aromatic analyses of the wines were assessed in a quantification matrix, evaluating each of the oenological characteristics.

Strain evaluation for each species was based on the weighted scoring method of the different parameters among all wines made with all different strains (of all species equally). The reference values established by the regulatory councils of the Galician Denominations of Origin for quality white wines were used. Therefore, the scores are comparable between strains of different species; however, it is more appropriate to compare strains within a given species because there are parameters such as the presence of reducing sugars that will penalise species with lower fermentative capacity. A maximum value of ten points was established for the positive range which was calculated by applying the following function: f = 10 $\times$ value of the parameter/highest value of the parameter found among all the strains. All the wines ranged total acidity concentrations between 4 and 9 g·L${}^{-1}$ (in accordance with the requirements of the DO Ribeiro); a higher total acidity value was considered positive (within these limits), i.e., it was awarded a higher score. Conversely, the altering parameters or those whose lower concentration was considered more positive were weighted inversely, i.e., with respect to a minimum score of 0 points: volatile acidity, ethyl acetate, acetaldehyde and presence of reducing sugars. For volatile acidity, the lowest value was considered the best value = 0.36 g·L${}^{-1}$, with none of the wines exceeding the highest concentration allowed by the DO Ribeiro (0.8 g·L${}^{-1}$). However, wines that exceeded 0.6 g·L${}^{-1}$ of volatile acidity were penalised. Ethyl acetate was penalised above 120 mg·L${}^{-1}$ (maximum value allowed by DO Rías Baixas). Acetaldehyde was penalised for amounts above 40 mg·L${}^{-1}$. In the case of pH, the optimum value was pH = 3.3. The minimum alcoholic strength was 9° (according to DO Ribeiro). The total score was calculated as the arithmetic sum of the scores assigned to each parameter. In addition, a weighting was added for each parameter, giving a higher relative score according to its relative importance depending on the standards required by these regulatory councils. The best rated native strain for each species was selected and its oenological potential was further evaluated in sequential fermentation with autochthonous strain S. cerevisiae XG3 and with a commercial strain (S. cerevisiae EC1118) as reported in [29, 30]. In addition, Principal Component Analysis (PCA) was used to separate the wines according to basic chemical and volatile composition. Pearson’s correlation analysis (r coefficient) was applied to determine the correlation of chemical compounds between different wines fermented with native non-Saccharomyces strains in pure and sequential fermentations, using Past software (version 4.07b, Øyvind Hammer, Oslo). Previously, data were standardized using the function f = (x-mean)/standard deviation, to guarantee the equity in those chemical compounds with different values or units.

The mean value of the estimated number of cells in the pre-inoculum for all strains was 10${}^{5.5\pm 0.1}$ cells mL${}^{-1}$ (Fig. 1). These values justify a correct and homogeneous inoculation of each strain ensuring the correct fermentation implantation in pure culture [31].

Fig. 1.

Estimated number of cells in the pre-inoculum. Number of cells in preinocula of some strains checked by measurement of the optical density at 600 nm.

Fermentation kinetics is shown in Fig. 2. As expected, S. cerevisiae XG3 began to ferment within the first 2–3 days and showed a higher fermentation speed than non-Saccharomyces vinifications. The fermentative ability of autochthonous non-Saccharomyces strains in pure culture varied among species and strains, with L. thermotolerans and T. delbrueckii standing out for their high fermentation rates, but lower speed than S. cerevisiae, as reported by other works [8, 9, 10, 12, 32, 33]. Starm. bacillaris strains showed a lower fermentation ability than strains from L. thermotolerans and T. delbrueckii, although this species has been described as tolerant to relative high ethanol contents and can survive up to the middle-end phase of the fermentation process [12, 34]. Regarding Metschnikowia spp. strains, they exhibited low to moderate fermentative power, as this species is not able to tolerate ethanol concentrations over 4–5% (v/v), so it naturally disappears when this content of ethanol is exceeded during alcoholic fermentation [13]. M. fructicola and M. pulcherrima presented similar fermentation kinetics and fermentation yields over time. The poor or slower fermentative activity of non-Saccharomyces species confirmed the need to add a S. cerevisiae strain to successfully complete fermentation. However, the slower fermentation kinetics could be positive for a better retention of volatile compounds and for energy savings during the tanks cooling [3, 35]. Unsurprisingly, the control (pasteurised must sample w/o inoculum) showed no evidence of fermentation or contamination.

Fig. 2.

Fermentative yield of 30 autochthonous strains. Fermentative ability in pure culture of indigenous non-Saccharomyces strains and S. cerevisiae XG3. Lt, L. thermotolerans; Mf, M. fructicola; Mp, M. pulcherrima; Sb, Starm. bacillaris; Td, T. delbrueckii; Sc, S. cerevisiae.

Fermentations with different non-Saccharomyces strains influenced the chemical characteristics of wines. Basic chemical parameters and aromatic composition of wines are summarized in Tables 2,3, respectively. The results evidenced a significant effect of species and strains in almost all the parameters determined. T. delbrueckii and L. thermotolerans increased the total acidity and lowered the volatile acidity of wines compared with the control wines (fermented with S. cerevisiae XG3) (Fig. 3A). This ability has been widely reported for L. thermotolerans [36, 37, 38], but the results are less clear for T. delbrueckii [39]. To a lesser extent Metschnikowia spp. and Starm. bacillaris also decreased the volatile acidity as found in other studies [13, 36]. L. thermotolerans, especially Lt93 strain, also increased the lactic acid ratio, as well as Metschnikowia spp. but the M. pulcherrima strains did so to a lesser extent than L. thermotolerans. Accordingly, the pH was lower in the L. thermotolerans wines but not in all the wines fermented with Metschnikowia spp. Besides, L. thermotolerans strains reduced the amount of malic acid in the wine compared to S. cerevisiae. This ability is of special interest in warm regions to mitigate the acidity reduction in wines because of climate change [10, 29, 40].

Fig. 3.

Total acidity, alcohol content and glycerol. Box and whisker chart of three relevant basic chemical compounds: (A) alcohol content, (B) total acidity, and (C) glycerol of non-Saccharomyces strains compared to S. cerevisiae XG3.

Moreover, the alcoholic degree of non-Saccharomyces wines was lower than control in all species and strains (Fig. 3B). For instance, a reduction of 1.25% v/v in the case of L. thermotolerans, 1.69% v/v with T. delbrueckii, and 2.90% v/v with Starm. bacillaris was observed, although in most cases there was residual sugars in the wines. Accordingly, the wines from these species presented lower ethanol yields than control wines obtained with ScXG3. Concerning to alcoholic strength and sugar consumption by the different non-Saccharomyces in single fermentations, results are consistent with the literature, but some strains such as Lt93 reached higher ethanol yield (ethanol production (g·L${}^{-1}$)/sugar consumption (g·L${}^{-1}$), g/g) than most of the studies consulted [36, 41, 42]. As expected, Metschnikowia spp. did not complete fermentation either; in this case, an average alcohol content of only 3.8% v/v (ethanol yield 0.30 g/g) was achieved and a high glucose + fructose content was remained in wine as reported by other authors [18, 43]. L. thermotolerans and T. delbrueckii produced wines with a higher ethanol concentration and consumed more sugars in the must compared to Starm. bacillaris and M. pulcherrima. Most of the strains of Starm. bacillaris has a strong fructophilic character, adapted to sweet wine fermentations and during fermentation ethanol yield from sugar consumed is lower as compared with S. cerevisiae [12].

Finally, all non-Saccharomyces strains increased the concentration of glycerol in the wine, especially Starm. bacillaris and Metschnikowia spp. (Fig. 3C). Particularly, with some strains of Starm. bacillaris the content of glycerol was very high (17.0 in strain Sb474), as described by other authors [2, 11, 12, 36]. This compound contributes unctuousness, volume, slight sweetness and silky mouthfeel to wines [12, 44, 45].

Yeast strains modulate the aroma of wine during fermentation. In particular, higher alcohols are the major compounds that also contribute to the generation of secondary metabolites whose biosynthesis is species dependent but not always is strain dependent [46]. Accordingly, all the yeast strains/species used in this study influenced the fermentative aroma composition of wines. Other major volatile compounds include acetaldehyde and ethyl acetate, both are undesirable at high concentrations imparting aroma like pungent and nail-polish, respectively. However, below 0.5 mg·L${}^{-1}$, acetaldehyde may contribute pleasant notes of bruised apple, sherry and nutty; in the case of ethyl acetate below 7.5 mg·L${}^{-1}$ imparts pineapple and fruity notes [47]. The concentration of major volatile compounds of wines fermented with pure cultures of different yeast strains is shown in Table 3. The results showed significant differences among species for almost all volatile compounds analysed, including the sum of higher alcohols. However, at strains level within a given species, significant differences were only found among Metschnikowia spp. strains for all parameters (except 1-butanol) and among T. delbrueckii strains for propanol and ethyl acetate. The content of higher alcohols ranged between 45 and 185 mg·L${}^{-1}$ in wines from Sb326 and ScXG3, respectively (Fig. 4). These concentrations are lower than 300 mg·L${}^{-1}$, above which the aromas become undesirable, although the wines from pure culture fermentations achieved lower content of higher alcohols than those obtained for the same strains tested in sequential fermentations [29, 30]. In addition, all the yeast species evaluated produced lower concentrations of these volatiles in the wines than S. cerevisiae, being L. thermotolerans the non-Saccharomyces species that produced the highest concentration of higher alcohols, in some cases doubling the concentration generated by Starm. bacillaris. Some studies corroborate the high production of higher alcohols in fermentations performed with L. thermotolerans, M. pulcherrima and Starm. bacillaris; however, variations up to $\pm$ 30% (increments or reductions) compared to the S. cerevisiae controls are found [3, 13, 19]. These results suggest that the production of higher alcohols and other chemical compounds is not only influenced by the strain effect, but may also be influenced by factors such as interactions with S. cerevisiae, other non-Saccharomyces or grape must characteristics [34, 48]. In our study, the differences were mainly due to the concentration of 3-methyl-1-butanol and isobutanol. S. cerevisiae was the species with the highest concentration of 3-methyl-1-butanol. However, the content of isobutanol was, compared to S. cerevisiae, two-fold higher in the wines made with Metschnikowia spp. and L. thermotolerans wines and increased by more than 50% with Starm. bacillaris in most strains as reported in some studies [11, 43].

Fig. 4.

Higher alcohols content. Box and whisker chart of higher alcohols content comparing the non-Saccharomyces strains to S. cerevisiae XG3. The sum of higher alcohols included 1-butanol, 2-methyl-1-butanol, 3-methyl-1-butanol, isobutanol (2-methyl-1-propanol) and propanol.

On the other hand, the high levels of ethyl acetate in wines fermented with some strains of Metschnikowia spp. are noteworthy. However, other strains obtained very low non-altering concentrations of this compound, which reinforces the argument that its production is strain-dependent; therefore, the strain selection is very important [23, 36].

The data from the chemical analyses (Tables 2,3) of the wines were included in a quantification matrix (Table 4), evaluating each of the desired oenological characteristics. Lt93 was characterised by high total acidity (+1.2 g·L${}^{-1}$; +17.6%), lactic acid (+2.5 g·L${}^{-1}$), fermentation yield (only 0.8 g glucose + fructose and alcohol production –0.93 %vol.) and higher alcohols production (159.4 mg·L${}^{-1}$; –16.0%) compared to the control ScXG3 as found in similar studies [9, 36, 38]. Within the group, Lt93 stood out from the other L. thermotolerans as the strain with the best fermentation performance and the highest production of lactic acid and glycerol. Td315 was characterized by a better balance of positive parameter values, higher glycerol concentration than ScXG3 (+1.9 g·L${}^{-1}$; +150%) and lower acetaldehyde concentration than the other T. delbrueckii strains. These are desirable values for a strain among the most widespread commercial non-Saccharomyces species [8, 49]. Sb474 highlighted because its wines had the lowest volatile acidity and highest glycerol concentration of all Starm. bacillaris wines. Furthermore, Sb474 showed very little variation between the two replicates in terms of fermentation development (CO${}_2$ evolution). This homogeneity can be favourable in fermentation with S. cerevisiae in aspects such as chemical-aromatic modulation, cell interaction, temperature, or oxygen availability [48, 50, 51]. Mf278 was noted for a high glycerol concentration, higher lactic acid and alcohol content as reported by other authors [3, 13] and it was the only Metschnikowia strain with very low (and acceptable) ethyl acetate concentration. The high concentration of ethyl acetate can be a limiting attribute, as it can spoil the quality of the resulting wine.

According to these values, the best rated strains within L. thermotolerans, T. delbrueckii, Starm. bacillaris and Metschnikowia spp. species were Lt93, Td315, Sb474 and Mf278, respectively. These strains were evaluated in sequential fermentation with the autochthonous S. cerevisiae strain XG3 [29] and with the commercial strain S. cerevisiae EC1118 [30].

The PCA using all chemical compounds analysed allowed a clear separation of wines elaborated with different strains of non-Saccharomyces species in pure culture (Fig. 5A). The first two components, PC1 and PC2, explained 60.5% of the variance. L. thermotolerans strains were located in the positive first quadrant, characterized by higher concentrations of 1-butanol, 2-methyl-1-butanol, 3-methyl-1-butanol, total acidity and lactic acid. In contrast, wines made with Starm. bacillaris were plotted on the negative side of PCA (except for the Sb474 and Sb304 strains), with a high content of glycerol. Torulaspora wines appeared close to the centre, between Lt and Sb strains, due to their higher alcohol content. In the opposite, on the negative side of PC2 and in the positive side of PC1 Metschnikowia wines were characterized by higher content of sugars, ethyl acetate and isobutanol.

Fig. 5.

Principal component analysis (PCA). (A) Pure cultures of autochthonous non-Saccharomyces strains: Metschnikowia spp. (square), L. thermotolerans (dot), Starm. bacillaris (diamond) and T. delbrueckii (triangle). Chemical parameter: %vol., alcohol content; G+F, reducing sugars (glucose + fructose); Gly, glycerol; LA, lactic acid; MA, malic acid; TA, total acidity; VA, volatile acidity. (B) Pure cultures of autochthonous non-Saccharomyces strains (NS), sequential fermentation with S. cerevisiae autochthonous XG3 (x-cross) and commercial EC1118 (plus). Biplot for the first two components (PC) for basic parameters and volatile compounds: ScXG3-fermentation with S. cerevisiae ScXG3; Lt93, Sb474, Td315 and Mf278-fermentations with each of these non-Saccharomyces strains and sequential inoculation with S. cerevisiae ScXG3 (NS+XG3) and S. cerevisiae EC1118 (NS+EC1118).

In addition, the results obtained with the best rated strains in pure fermentations were compared with those obtained with the same non-Saccharomyces strains in sequential fermentations. PCA results also shown a proper separation of wines elaborated with non-Saccharomyces strains in pure culture and in sequential fermentations (Fig. 5B). In this case, the first two components, PC1 and PC2, explained 62.1% of the variance. Wines made with non-Saccharomyces in pure culture (NS) were in left side. On the opposite side were grouped the wines produced by sequential fermentation with ScXG3. However, wines fermented in sequential fermentation with the commercial ScEC1118 strain (NS+EC1118) were plotted in the middle of the PCA. In addition, strains from the sequential fermentations, especially those from the NS+EC1118 group, clustered more closely than those from the pure culture, indicating greater homogeneity due to the influence of S. cerevisiae strains on the resulting wines while retaining their differentiation. Some authors have proposed different mechanisms in the relationships between S. cerevisiae and non-Saccharomyces [18, 51, 52]. In this sense, the NS+XG3 group showed the highest concentration of volatile compounds, which have confirmed that native strains can improve the chemical and sensory characteristics, and differentiation of wines in accordance with numerous recent studies [3, 30, 53, 54, 55]. In contrast, compounds accounting for acidity in the biplot are concentrated in the area of NS+EC1118 wines. On the other hand, some chemical compounds such as glycerol do not point to the species that produce the highest concentration in the biplot, which indicates that they do not depend on the type of fermentation but rather depends on the species and strain.

A correlation (Pearson) study (Table 5) was performed to show possible correlations between chemical compounds, strains, and fermentation groups (pure and sequential cultures). Regarding correlations between strains in single fermentations, the results showed some differences. As expected, the correlations between S. cerevisiae control strains in pure culture fermentations showed high r-values with each other (r $>$ 0.9; p 0.05 in all compounds, except for those related to acidity where no significant differences were found). Moreover, the correlations were higher between the same strain ScXG3 under different conditions (pure and sequential fermentation controls with different musts, temperatures, and fermentation volumes). Furthermore, comparing all the strains with each other in pure culture, the highest correlations were found in the higher alcohols, except for Mf278. However, in the case of basic chemical compounds the opposite occurred, especially in acidity-related compounds, Mf278 obtained the highest correlations.

On the other hand, the main correlations with significant differences (p 0.05) between the groups of sequential fermentations (NS+ScXG3 and NS+EC1118) occurred in the following pairwise: in all overall compounds and in volatile acidity (r = 0.885), 3-methyl-1-butanol (r = 0.893), ethyl acetate (r = 0.880) and glycerol (r = 0.985). Furthermore, the compounds 2-methyl-1-propanol (r = 0.868; p = 0.056) and reducing sugars (r = 0.852; p = 0.067) showed correlation with significance close to p = 0.05. Similarly, some correlations with significant differences (p 0.05) were found between pure culture fermentations and sequential fermentations in the overall major volatile compounds, glycerol, and ethyl acetate. Lt93 and Td315 are the non-Saccharomyces strains that showed the highest correlation in all tests performed (r $>$ 0.8; p 0.05) in most of the comparisons between pure and sequential culture. Sb474 showed medium correlation (Sb474+ScXG3: r = 0.56, p = 0.03; Sb474+EC1118: r = 0.50, p = 0.17). In contrast, Mf278 barely showed any correlation (r = 0.33–0.46; p $>$ 0.05) with sequential fermentations. This suggests that the contribution of the selected non-Saccharomyces strains can be stable regardless of other variables such as must type (chemical composition, variety), winemaking techniques (temperature, fermentation volume) or S. cerevisiae strain (indigenous or commercial) but not for all parameters and it depends on species and strain. Furthermore, in the overall comparisons (in the compound groups), many correlations were observed between the sequential fermentations with each other including Mf278 and Sb474 strains, indicating a stronger influence of the S. cerevisiae strain (Supplementary Figs. 1,2,3 of the supplementary material). Moreover, the number of these correlations was observed more frequently in the sequential fermentations with ScXG3, suggesting a greater influence of the indigenous strains on the preservation of the chemical profile, especially in the aromatic parameters analysed. However, correlations of the different native non-Saccharomyces strains between pure culture and sequential fermentation showed some singularities for some specific parameters (data not shown). Lt93 showed a higher correlation with the native strain ScXG3 in sequential fermentation, although no significant differences were found for some basic chemical parameters such as acidity or sugar content. As for Td315, the pure culture showed correlation with both sequential fermentations. Moreover, for volatile compounds, correlation was also observed between the two sequential fermentations with Td315 (r $>$ 0.9; p = 0.038), although in the comparison with Td315+ScXG3 no correlation was found for some basic chemical parameters, such as sugar content. The influence of the pure culture with Sb474 showed a correlation mainly between the sequential fermentation with ScXG3, but no significant differences were found between both sequential fermentations (except for acidity) and surprisingly neither between glycerol (parameter with especially high content in Starm. bacillaris species). That is, the amount of glycerol is higher in pure culture than in sequential fermentations. This did not happen with the Mf278 strain in which the highest correlations between pure and sequential culture were found only with ScXG3 (not in Mf278+EC1118) in glycerol, acidity, and aromatic compounds, but not for the rest of the non-volatile compounds together. Furthermore, high correlations were found between both sequential fermentations. In this sense, as expected according to the PCA results (Fig. 5B), the correlation between the chemical composition of the pure culture and the sequential fermentations is lower as the effect of the S. cerevisiae strain on the chemical composition is significant. Therefore, these autochthonous strains could be used as starters in sequential fermentations to obtain wines with singular properties. The results of the Pearson’s test showed some correlations between the strain in pure culture and the sequential fermentation despite the influence of other factors, such as S. cerevisiae strain or grape juice characteristics. So, the results suggested that some properties are going to remain stable in the selected strains during sequential fermentations, which maintain the chemical properties.

The results of the PCA and Pearson test showed that the different non-Saccharomyces strains influenced the chemical profile of wines, through the formation of different compounds. Sequential inoculation of strain Mf278 increased the production of positive aroma compounds with respect to the indigenous control strain S. cerevisiae ScXG3 alone; it should be taken into consideration that ScXG3 has been selected in previous studies for its optimal winemaking behaviour [26, 56, 57]. Differences could be due to the fermentation process, and it should be considered for the use of these strains in future studies. T. delbrueckii and Starm. bacillaris strains were the worst qualified; however, several authors obtained better wines than their controls made with pure S. cerevisiae following a co-inoculated strategy with these species and S. cerevisiae [1, 18, 42, 58]. In fact, there are already commercial strains of T. delbrueckii and to a lesser extent of Starm. bacillaris that can be used due to their desirable oenological characteristics such as increased glycerol concentration, acidity, low ethanol and esters production [12, 59].

All these results are consistent with those found in the previous literature and allow us to affirm that some indigenous strains of non-Saccharomyces yeasts, used by sequential inoculation, can contribute to the typicity and differentiation of wines and even increase the contribution of aromatic compounds with positive sensory evaluation [1, 3, 55]. In this sense, the differentiation and correlations obtained between the different strains and fermentations show that the choice of yeast strain affects the chemical profile of the wines, but certain distinctive parameters are maintained in sequential fermentation.

Therefore, we suggest further research, with different strains and even by applying more than one non-Saccharomyces together (in multistarter yeast inoculum or sequentially before inoculation of S. cerevisiae) which has been proposed to compensate the shortage of one non-Saccharomyces species and further enhance the wine quality [18]. This will make it possible to evaluate the positive synergistic effects on wine quality and typicity, the interaction between species and ultimately increase the knowledge of interspecific relationships. Accordingly, combinations of T. delbrueckii, Hanseniaspora vineae, Hanseniaspora uvarum, M. pulcherrima, Zygosaccharomyces bailii or L. thermotolerans were reported as a multistarter or co-inoculation fermentations improving the chemical and aromatic compositions of wines fermented only with S. cerevisiae [7, 18, 39, 60].