**1. Introduction**

Grapes hold a diverse microbial population consisting of bacteria and yeasts that meet the microorganisms located in the winery facilities after the harvest. During the initial stages of the spontaneous alcoholic fermentation (AF), this pool of microbes achieves a balance until *Saccharomyces* (*S*.) *cerevisiae* becomes the main yeas<sup>t</sup> in the fermentative process.

Early AF is characterized by a diverse yeas<sup>t</sup> population, with low frequency of detection of *S. cerevisiae*, but with a high presence of non-*Saccharomyces* yeasts. The presence of unknown microbiota makes it a risky and unpredictable practice. Therefore, the inoculation of commercial *S. cerevisiae* strains has been widespread in the modern wine industry all over the World. Indeed, the non-*Saccharomyces* yeasts have not been well-regarded by oenologists and these have tended to make efforts to avoid their involvement in AF [1]. These traditional and conservative oenological practices have led to a homogenization and globalization of winemaking, a sameness in the taste and flavours of finished wines [2].

A general strategy to increase the diversification of wines has made oenology return to its origins of natural and diverse microbial populations. For this purpose, the employment of non-*Saccharomyces* yeas<sup>t</sup> species has shown promising results. This new trend has triggered the studies and published results of non-*Saccharomyces* yeasts which has led to some of them being used as commercial culture starters [1].

The use of mixed starter cultures of selected non-*Saccharomyces* combined with *S. cerevisiae* to avoid any stuck fermentations is thought to be a solution for ensuring AF completion, while various organoleptic characteristics involved in the quality of the final products are improved [2,3]. Furthermore, mixed cultures composed of more than one non-*Saccharomyces* species in combination with *S. cerevisiae* have been employed with the aim of simulating this complex yeas<sup>t</sup> community present in spontaneous AF [4,5]. In general terms, the early inoculation of *Metschnikowia* (*M*.) *pulcherrima* has been aimed to improve flavour of wines [6]. In the case of *Lachancea* (*L*.) *thermotolerans*, the objective is the increase of lactic acid that would have also an impact in the aromatic profile of wines [7]. Moreover, *Torulaspora* (*T*.) *delbrueckii* has been initially employed for reducing the alcohol after the AF and for improving the aroma profile of wines [8].

The current study aims to describe the oenological e ffects of the sequential early inoculation of a pure culture of *M*. *pulcherrima* and a mixed culture of *L*. *thermotolerans* and *T*. *delbrueckii* in the vinification of Tempranillo, Grenache and Graciano grape varieties. With this purpose, the impact dependent on the specific grape variety in semi-industrial conditions was analysed. To this end, the kinetics of AF, implantation rate, variation of the oenological, colour and aromatic parameters after AF and clustering after malolactic fermentation (FML) were individually performed for each grape variety.

#### **2. Material and Methods**

#### *2.1. Grapes and Initial Must Samples of the Three Varieties*

Grapes of the three red grape varieties from the D.O.Ca. Rioja, Tempranillo, Grenache and Graciano were employed to perform this study. These grape varieties were chosen for being important in the region where this study was developed but also, they are very present in international winemaking areas. When the grapes had reached an average probable alcohol by volume (APBV) of approximately 13%, around 225 k of each one were individually harvested, crushed and destemmed (Figure S1).

Samples of the three must were physicochemical characterized. APBV, pH and total acidity were analysed according to o fficial ECC methods [9]. Malic acid was determined also by the o fficial method [9]. by an enzymatic method carried out with an automated clinical chemistry analyser (Miura One, TDI, Madrid, Spain). The yeas<sup>t</sup> assailable nitrogen (YAN) was measured following the protocol described by Aerny [10].

The three musts were also microbiologically characterised by plating the appropriate dilution on Chloramphenicol Glucose Agar (CGA 05% yeas<sup>t</sup> extract, 20% glucose, 005% chloramphenicol, 17% agar,) plates, incubated at 28 ◦C for 48 h. Ten yeas<sup>t</sup> colonies were isolated from each plate containing between 30 and 300 colony forming units per millilitre (CFU/mL). DNA was then extracted from fresh culture following the protocol determined by <sup>L</sup>ópez et al. [11]. Then, a partial region of the 26S rDNA gene was amplified with PCR using the primers and conditions established by Cocolin et al. [12]. PCR amplicons were purified and sequenced by Macrogen Inc. (Seoul, South Korea). The sequences were compared to the GenBank nucleotide database using the Basic Local Alignment Search Tool (BLAST) [13]. The identification was considered appropriate if gene sequences showed identities of at least 98%.

## *2.2. Yeast Species*

This study was performed with four oenological yeas<sup>t</sup> species, *M*. *pulcherrima*, *L*. *thermotolerans*, *T*. *delbrueckii* and *S. cerevisiae* (VRB commercial yeas<sup>t</sup> from Lallemand Bio S.L., Toronto, Canada). *M. pulcherrima* and *S. cerevisiae* were pure cultures while *L. thermotolerans* and *T. delbrueckii* (*L*& *T*) were combined in percentages of 30% and 70%, respectively, following the natural combination of the two species observed in other studies of non-*Saccharomyces* population in Rioja red wines [14,15]. All these yeasts were selected in the Rioja Qualified Designation of Origin (D.O. Ca. Rioja) from Spain, and

they are in the last stage of the selection program. Furthermore, they are stored in the Instituto de Ciencias de la Vid y del Vino (ICVV) collection. These yeas<sup>t</sup> were identified by Macrogen Inc. with the amplified region D1 of the 26S rRNA gene using the primers NL1GC and LS2 [16].

#### *2.3. Inoculation Procedure and Alcoholic Fermentation*

The must of each variety were put into nine 30 L tanks that were kept at 25 ◦C to carry out the AF (Figure S1). When the tanks were filled, potassium metabisulphite was added to the samples to achieve a total SO2 concentration of 50 mg/L.

After this, the 27 tanks were inoculated with the different yeasts following three different inoculation strategies. For each variety, three out of the nine tanks (n = 3) made up the control sample (C) and were inoculated with the commercial *S. cerevisiae* starter culture VRBTM following the producer's instructions, another three made up the sample early inoculated with *M. pulcherrima* (n = 3) (M) and the last three (n = 3) the sample early inoculated with a 30/70 mixture of *L. thermotolerans* and *T. delbrueckii* (*L*&*T*). The non-*Saccharomyces* yeasts had been pre-cultured in YPD liquid medium at 25 ◦C for 48 h with orbital shaking until the stationary phase. The concentration of cells/mL was counted with the Neubauer chamber. *M. pulcherrima* pure culture was inoculated in a concentration of 10<sup>6</sup> cells/mL counted while the mixed culture contained 3 × 10<sup>5</sup> cells/mL of *L. thermotolerans* and 7 × 10<sup>5</sup> cells/mL of *T. delbrueckii*. Three days later, all the 27 tanks were inoculated with the *S. cerevisiae* starter culture VRBTM at a concentration of 1 × 10<sup>6</sup> cells/mL.

The kinetics of AF was monitored by daily determination of the Brix degree and density decrease. Samples for implantation control were taken under aseptic conditions at three different moments. The first one was three days after harvest and initial inoculation with *Saccharomyces* and non-*Saccharomyces* yeasts (day 3). The second one was at the fourth day (day 4) when the 27 tanks had been inoculated with *S. cerevisiae* VRBTM. Eventually, the third control of implantation was performed one week after the first inoculation (day 7) (Figure S1). At these three moments, serial dilutions were carried out and the samples were microbiologically characterized as described above (Section 2.1). With the sequencing results, the percentage of each species composing each replicate was determined.

When the 27 wines had reached about 990 g/<sup>L</sup> density, they were pressed and fermented to dryness. The AF was complete when reducing sugars were lower than 2 g/L. Then, the wines were characterized by measuring the alcohol by volume (ABV), pH, total acidity, volatile acidity, colour intensity and hue according to official ECC methods [9]. Moreover, the malic and lactic acids, glycerol and acetaldehyde contents were determined by an enzymatic method carried out by an automated clinical chemistry analyser (Miura One) and tartaric acid by the Rebelein method [17]. Furthermore, total anthocyanins were measured by decolouring using SO2 [18] and total phenolics were determined as the total polyphenol index by spectrophotometric absorbance at 280 nm after dilution of samples. Ionized anthocyanins were determined according to Glories [19] and the polymerization index was calculated according to Ruiz [20].
