Introduction

One of the most significant technological advances in winemaking has been the control of the microbiological process by grape must inoculation using selected yeasts [21]. Today, the use of indigenous wine yeasts selected from each winegrowing area is widespread [20]. These local yeasts are presumed to be more competitive than commercial yeasts because they are better adapted to the ecological and technological features of their own winegrowing area [1, 19]. Therefore, they would be more capable of dominating the fermentation and would become the most important biological agent responsible for the winemaking. Additionally, selection of the appropriate local yeasts would assure the production of quality premium wines maintaining the differential properties of their own area and preserving its natural biodiversity.

The selection of wine yeasts is usually carried out within the species Saccharomyces cerevisiae—the most relevant one in winemaking—according to a set of physiological features (criteria) indicative of their potential usefulness for industrial wine production [21]. These criteria include the evaluation of yeast features related to the achievement of an efficient transformation of grape sugars into alcohol and carbon dioxide at a controlled rate and without off-flavor development. They are evaluated for all isolates by carrying out small-scale fermentations in synthetic or semi-synthetic media [17, 31]. Afterwards, another set of enological properties, generally more related to the kind of wine to be produced, is evaluated for a lower number of preselected isolates usually using sterile grape must [4, 10, 14, 15, 27].

Located at 37°5′ and 40°5′ southern latitude, Argentinean North Patagonia is the southernmost winegrowing region of the world. This region has optimal agro-ecological conditions for high quality viticulture and a long winemaking tradition [6]. Both spontaneous and conducted grape juice fermentations are carried out and young dry red (85%) and white wines (12%) from neutral Vitis vinifera varieties are mostly produced. However, indigenous yeast strains for enological use had not been selected until this study. In this work, 32 indigenous North Patagonian S. cerevisiae isolates of enological origin, previously characterized as belonging to different strains by means of molecular patterns and killer biotype [7, 8], were extensively evaluated for their technological and qualitative traits using a two-step selection protocol. Because killer character is widely distributed among native wine yeasts in North Patagonia [24, 26], additional ecological criteria based in killer interactions were included in the protocol and an indigenous yeast isolate with potential application in the elaboration of more aromatic young red wines was selected.

Materials and methods

Yeasts and growth conditions

Thirty-two indigenous S. cerevisiae isolates obtained from North Patagonian natural Malbec wine fermentations were used in this work. These isolates have been previously characterized by mtDNA-RFLP, PCR amplified inter δ sequence DNA polymorphism analysis and killer biotype as belonging to 32 different molecular patterns [7, 8] (Table 1). Two commercial S. cerevisiae starter strains commonly used in North Patagonian red winemaking were used as reference standards.

Table 1 Origin, identity and some technological and qualitative properties of North Patagonian indigenous S. cerevisiae isolates
Full size table

Killer and sensitive strains

Forty-seven killer yeasts and the sensitive S. cerevisiae P351 strain (PROIMI yeast collection) were used in killer assays. Killer yeasts were the K2-type killer reference S. cerevisiae NCYC 738 strain, 10 S. cerevisiae killer commercial starters widely used in North Patagonian winemaking and 35 indigenous killer yeasts including 10 S. cerevisiae and 25 non-Saccharomyces isolates (Debaryomyces hansenii, 1 isolate; Hanseniaspora uvarum, 4 isolates; Metschnikowia pulcherrima, 7 isolates; Pichia anomala, 4 isolates; Pichia kluyveri, 4 isolates; Torulaspora delbrueckii, 2 isolates; Rhodotorula glutinis, 2 isolates; and Zygosaccharomyces rouxii, 1 isolate). All indigenous killer isolates were obtained from Malbec and Merlot grapes and musts as well as cellar surfaces in North Patagonia.

For all assays, the yeast cells were previously growth overnight in GPY Agar medium (0.5% w/v yeast extract, 0.5% w/v peptone, 4% w/v glucose, and 2% w/v agar, pH 4.5).

Killer assays

Killer activity and sensitivity were determined using the seeded-agar-plate technique on YEPD-MB medium (1% w/v yeast extract, 2% w/v glucose, 2% w/v peptone, 2% w/v agar, 0.0003% w/v methylene blue) buffered at pH 4.6 with 0.5 M phosphate–citrate as described by Sangorrín et al. [25].

Fermentations

Small-scale fermentations were carried out in 250 ml Erlenmeyer flasks containing 200 ml of semi-synthetic medium (200 g l−1 of sucrose, 10 g l−1 of yeast extract, 1 g l−1 of ammonium sulphate, 1 g l−1 of potassium phosphate, 1 g l−1 of magnesium sulphate). After sterilization, citric acid (10 g l−1) was added to the fermentation flask in order to decrease the medium pH to 3.5, a pH similar to the wine [13]. After this, they were inoculated with 17 ml of each indigenous S. cerevisiae young cultures containing 107 cells ml−1. Flasks were plugged with glass fermentation traps containing sulphuric acid to allow only CO2 to evolve from the system and they were kept at 27 °C without shaking [31]. The fermentation evolution was daily followed by weight loss until constant weight during two consecutive measures. At this time, the fermented media were racked and centrifuged at 4,000 g for 5 min and the clear fermented media were assayed for ethanol production and total and volatile acidity as described below. The fermentations were carried out in triplicate in separate trials.

Microvinifications

Microvinifications were carried out in 5 l flasks containing 3.5 l of pasteurized (90 °C, 3 min) Malbec must (240 g l−1 of sugars, pH 3.4) supplemented with sodium metabisulphite at a final concentration of 150 mg l−1. S. cerevisiae indigenous isolates were inoculated at a cellular density of 107 cells ml−1. The winemaking was conducted at 25 °C and its kinetics was followed by measuring the Baumé degrees (1°Baumé = 17 g l−1 of fermentable sugars). When alcoholic fermentations were completed (<1°Baumé) young wines were racked, centrifuged (4,000 g, 5 min) and their enological parameters determined.

Enological parameters

All enological parameters were determined according to the methods proposed by Ribereau-Gayon et al. [22]. Therefore, ethanol concentration was determined by steam distillation and expressed as fermentative power or Gay Lussac degrees (FP or °GL, % v/v) and specific productivity [SP, ethanol (g) biomass (dry weight g−1) hour (h−1)]. The biomass values were obtained by subtracting the initial yeast culture dry weight from the final one. For dry weight determination 5 ml culture samples taken at initial and end of each fermentation were centrifuged at 10,000 g 5 min, washed twice and resuspended in cold sterile distilled water; these cell suspensions were placed on pre-weighed dishes and dried at 105 °C to constant weight. Volatile acidity (VA) was determined by steam distillation followed by titration with NaOH 0.1 N and was expressed as acetic acid (g l−1). Total acidity was determined at 20 °C by direct titration using NaOH 0.1 N and expressed as tartaric acid (g l−1). Free sulphur dioxide was determined by direct titration with iodine and total sulphur dioxide was determined by sample treatment with KOH (1 N) followed by titration with iodine. Malic acid and glycerol were enzymatically determined using Boehringer Mannheim commercial kits.

Higher alcohols and esters determinations

Aliquots of the cell-free wine samples were analysed by headspace solid-phase-microextraction sampling (SPME) using poly (dimethylsiloxane) (PDMS) fibres (Supelco, Sigma-Aldrich, Barcelona, Spain) and GC according to Rojas et al. [23]. Aliquots of 1.5 ml of the samples were placed in 15 ml vials and 0.3 g of NaCl and 15 ul of 0.1% (v/v) 2-octanol in ethanol (Fluka) were added as internal standard. The vials were closed with screwed caps and 3-mm thick teflon septa. Solutions were stirred for 2 h at 25 °C to get the required headspace–liquid equilibrium. PDMS fibers were injected through the vial septum and exposed to the headspace for 7 min and then desorbed during 4 min in an HP 5890 series II gas chromatograph equipped with an HP Innowax column (Hewlett-Packard) (length, 15 m; inside diameter, 0.25 mm; film thickness, 0.25 um). The injection block and detector (FID) temperatures were kept constant at 220 and 300 °C, respectively. The oven temperature was programmed as follows: 40 (10 min) to 200 °C at 4 °C/min, and 200 to 260 °C at 20 °C/min and kept 2 min at 260 °C.

The following standards were purchased from Fluka: isobutilic alcohol, isoamylic alcohol, 1-hexanol, bencylic alcohol, 2-phenyl ethanol, ethyl acetate, isobutyl acetate, ethyl lactate, isoamyl acetate, hexyl acetate, diethyl succinate, bencyl acetate, ethyl caprylate, ethyl 3-hydroxibutanoate, 2-penylethyl acetate. All standards were of greater than 99% purity.

Ester and higher alcohols concentrations calculated using standard solutions were the average of three independent assays.

Hydrogen sulfide production

Test tubes (16 mm × 16 mm) containing 15 ml of sterile red grape juice, supplemented with sodium metabisulphite at a final concentration of 150 mg l−1, were inoculated with young yeast cultures. At this time, paper saturated with Lead acetate (PbAcO) was suspended from the tube top using its screwed cap. Assays were carried out at 26 °C during 15 days. The qualitative hydrogen sulfide production was detected by the blackening of the PbAcO paper [10] and the isolates were classified as high (+++), medium (++) and low (+) sulphide producers.

Foam production

Foam height was measured every day throughout the fermentation assays described above. Yeasts were classified into three categories based on the maximum foam height reached: lower (foaming lower than 2 mm), middle (foaming between 2 and 4 mm) and higher (foaming greater than 4 mm) according to Martínez-Rodríguez et al. [10].

Ethanol tolerance

Ethanol tolerance was determined using fermentation assays proposed by Kurtzman and Fell [5]. Five, 10 and 15% v/v ethanol were added into the fermentation media before the yeast inocula. Fermentation tubes without ethanol were used as controls.

Statistical analysis

ANOVA and Tukey honest significant difference tests (HSD) with α = 0.05 were performed by mean comparison. The data normality and variance homogeneity in the residuals were verified by Lilliefors and Bartlet tests respectively.

Results and Discussion

Yeast preselection

Technological and qualitative properties

In this first step, local yeasts were evaluated in their key technological and qualitative properties using both small-scale fermentations carried out on semi-synthetic media (fermentative power, FP; volatile acidity, VA; and foam production) and specific assays (SH2 production, ethanol tolerance, killer behaviour). For comparative purposes, two commercial S. cerevisiae strains, (BC and F10), were also evaluated.

The results obtained from small-scale fermentations showed little foam production as well as similar FP and VA values in most of the isolates (Table 1, Fig. 1). Statistical analysis of FP data evidenced that: (a) there was not significant difference between commercial starters, (b) the isolates MIi4, MIf4 and MMf11 showed significantly lower values than BC (black symbols in Fig. 1) and (c) the isolates MIi8, MIi20, MIm17, MMf1, MMf8, MMf9, MIf16 and MMf17 showed values significantly higher than F10 (p < 0.01) as well as than other indigenous isolates (grey symbols in Fig. 1). When ethanol production was standardized by biomass and time (specific productivity, SP) significant differences were only detected among some indigenous isolates. Isolates MMf9 and MIm16 showed the highest SP values (data not shown). With regards to VA production, the commercial F10 strain as well as 16 indigenous isolates—indicated as triangles in Fig. 1—showed significantly lower VA values than BC starter (p < 0.01).

Fig. 1
figure 1

Correlation between fermentative power (FP) and volatile acidity (VA) values obtained with all studied indigenous isolates and both BC and F10 commercial starters. Black shadowed symbols FP values significantly lower than BC ones. Grey shadowed symbols FP significantly higher than F10 ones. Triangles VA values significantly lower than BC ones (p < 0.01, n = 3)

Full size image

On the other hand, the results obtained from the specific assays evidenced that all the yeasts including commercial starters have capability for sulphide production, however, most of them (51.7%) showed an intermediate production level (Table 1). On the other hand, all the isolates were capable to ferment sugars in the presence of 5 and 10% of ethanol in the media. However, only MIf8, MMf3 and MMf9 isolates were capable to ferment with 15% of ethanol.

With regards to the local yeasts killer properties, the technological aspects of this phenomenon were firstly evaluated. Several killer toxin types has been described in S. cerevisiae, however, K2-type toxin has proved to be the most important one in winemaking because it is active under the wine pH conditions [30]. Evaluation of the killer phenotype using sensitive and K2-type killer reference S. cerevisiae strains showed that all indigenous isolates but two, MMf9 (killer isolate, K+R+) and MIi13 (neutral isolate, K−R+), exhibited the sensitive (K−R−) phenotype (Table 1).

Additional ecological criteria

The evaluation of killer phenotype using K2-type killer and sensitive reference strains is a frequently used technological criterion in yeast selection programmes [16, 29]. However, ecological aspects of this phenomenon, such as killer interactions of the studied isolates against other indigenous non-Saccharomyces and Saccharomyces yeasts from the same winemaking area, become relevant in areas where killer yeasts are widespread such as the North Patagonian region [24, 26]. There is evidence that in guided fermentations the dominance of the starter can be subordinated to the specific winemaking conditions settle down during the initial stages [18]. Indigenous and/or commercial killer yeasts present at the beginning of fermentation, originated from grapes or winery surfaces, could hinder the implantation of S. cerevisiae starters inducing stacking fermentations [16, 30]. In this context, killer sensitivity of the studied isolates was tested against 10 S. cerevisiae killer commercial starters, and 10 S. cerevisiae and 25 non-Saccharomyces killer indigenous isolates.

The isolates response against indigenous S. cerevisiae killer yeasts was the same than that observed against the K2-type killer reference strain showed in Table 1. However, the 50% of the indigenous isolates was resistant to, at least, one commercial killer starter. Under assays conditions, MMf9 and MIi13 isolates showed resistance phenotype against all commercial killer strains whereas the isolates MMf17 and MIf6 as well as MIi8, MIi20, MMf4, MIm17, MMf7 and MIi6 showed resistance to 50 and 20% the commercial strains, respectively. These results evidence the relevance of studying additional regionally significant yeast killer interactions, different from the classical technological evaluation using K2-type reference strains.

On the other hand, all analyzed isolates including those showing resistant phenotype (MMf9 and MIi13 S. cerevisiae isolates) were sensitive to four P. anomala and two P. kluyveri indigenous killer isolates (data not shown) evidencing a different target molecule in S. cerevisiae cell wall. The relevance of these killer interactions in winemaking is currently been evaluated in our laboratory.

Among all evaluated properties, FP values and VA production as well as ecological criteria were prioritized in the preselection stage. The indigenous isolates MIi20 and MMf9 showing appropriate combinations of FP and VA features were selected to continue the study. Moreover, MMf9 isolate showed killer character against sensitive collection strains and resistant phenotype against K2 killer collection strain, all the indigenous and commercial S. cerevisiae killer yeasts (Table 1) whereas MIi20 showed resistant phenotype against some of the later. Both isolates additionally showed other remarkable qualitative traits such as low SH2 production—malodorous compound in wines [12], little foam production (Table 1) and capability to ferment sugars in medium containing 10–15% of ethanol, an important feature for ensure the dryness of must.

Finally, it is interesting to note that the results obtained from semi-synthetic medium assays shows a good correlation with those obtained from sterile must (Table 2). Comparison between PF and VA values obtained from both media (Table 2) shows that even when absolute values were significantly modified, differences among isolates were kept (ANOVA with no significant interaction, p: 0.5321).

Table 2 Comparison among some strain properties evaluated in semisynthetic medium and grape must
Full size table

Selection

In young wines obtained from neutral V. vinifera varieties the final aromatic quality is strongly influenced by the fermentative flavour originated from yeast metabolism. In the selection stage, small-scale red grape musts fermentations using the preselected indigenous isolates (MIi20 and MMf9) were carried out in order to evaluate both the aromatic profiles (Higher alcohols, HA; and Esters) and the physicochemical characteristics of the obtained wines. Additionally, an isolate (MMf8) representing the most frequent molecular pattern (pattern IIIa in Lopes et al. [7]) was also included in this selection stage. Fermentations carried out with both BC and F10 commercial strains were used as controls.

As a whole, the results indicate that all Malbec grape must fermentations took place properly and that the all wines had a composition considered normal for this winemaking scale. However, significant differences were observed among some physicochemical characteristics as well as the volatile compound concentrations of the wines produced by different yeast strains (Tables 3, 4).

Table 3 Enological features of Malbec wine from 5 l fermentations with two commercial strains and three selected indigenous isolates
Full size table
Table 4 Volatile compounds (mg l− 1) produced in Malbec wine fermentations (5 l) using two commercial strains and three selected indigenous isolates
Full size table

It is interesting to note that MMf8 and MMf9 indigenous isolates were the lowest acetic acid and total higher alcohol producers. Elevated concentrations of both acetic acid (more than 0.8 g l−1) and higher alcohols (more than 300 mg l−1) are related to defective wines [28]. Additionally, isoamyl alcohol and 2-phenyl ethanol, the major constituents of the higher alcohols fraction; have been related to caustic notes and floral flavour notes in wines, respectively. MMf9 wines showed the lowest levels of isoamyl alcohol as well as the highest proportion of 2-phenyl ethanol with regard to the total higher alcohols values (30.7% of the total HA) (Table 4).

The fresh, fruity aroma of young wines is largely derived from a mixture of esters produced during fermentation [3]. However, high levels of ethyl acetate (EA), the main ester in wine, can spoil the wine bouquet with an unpleasant aroma [28] whereas long chain esters (esters different to ethyl acetate) contribute positively to wine complexity and quality [2, 28]. In this context, some authors have proposed that a relationship as [long chain esters]/[HA]+[EA] can be related to the olfactory quality of wine [11]. This relation (R) was regarded as good selection criterion in this stage; wines obtained with MMf9 isolate showed the highest values for this relationship (Table 4). Similar behaviour was observed using the same strain in both Malbec and Merlot type musts (data not shown).

This poly-phasic selection study allowed us to have a complete and real picture of the indigenous yeast strain physiology and ecology. As a result, one indigenous S. cerevisiae strain (MMf9) was selected in order to be developed as a regional North Patagonian wine starter culture. Results obtained from implantation studies carried out with this indigenous strain at industrial level (10,000 l fermentations) in two regional cellars [9] enabled us confirmed its aptitude for enological use evidencing that the methodology proposed for the yeast selection was adequate.