**4. Discussion**

This study was focused on individual pilot plant vinifications of Tempranillo, Grenache and Graciano inoculated with non-*Saccharomyces* yeas<sup>t</sup> inocula for responding if every non-*Saccharomyces* yeas<sup>t</sup> would cause similar physicochemical and aromatic profiles in different grape varieties. The initial must of three grape varieties musts were separated according to parameters of APBV and acidity. Moreover, their indigenous yeas<sup>t</sup> communities were also different. These initial differences fitted with a standard winemaking of non-sterile grapes [22].

#### *4.1. Yeasts Establishment and Fermentation Kinetics*

Tempranillo grapes had low APBV and high malic acid content, which was initially positive for the establishment of yeasts and bacteria populations and consequently for the evolution of AF and MLF. The presence of *S. cerevisiae* in the grape surface and must is usually low and this was corroborated in this study [23]. The initial must had *S. cerevisiae* as residual yeas<sup>t</sup> and a high population of *H. uvarum*, *H. osmophila* and *T. delbrueckii* that were naturally present. The control sample inoculated only with *S. cerevisiae* had a large population of *M. pulcherrima* after three days, which might be due to an external contamination of the tanks with the *M. pulcherrima* inoculated vinification that coexisted in the experimental winery. Nonetheless, indigenous *T. delbrueckii* and *H. osmophila* were not detected and AF proceeded without problems; it was rapid and lasted only six days. In Tempranillo samples, early inoculated with non-*Saccharomyces* and then with *S. cerevisiae*, the establishment of the different yeas<sup>t</sup> species happened as it was expected, probably due to the preadaptation of the strains to the grape variety because they had been isolated from this same variety.

The microbial composition of the Grenache must was characterised by a large population of *H. uvarum*, with *S. cerevisiae* as a minority strain. A diverse indigenous population characterized the initial must. Furthermore, indigenous *M. pulcherrima* was found in Grenache grapes, although with low percentage. *S. cerevisiae* inoculated in the control sample was able to achieve total implantation in spite of the high APBV of the must, and of the ecological pressure that other initial yeas<sup>t</sup> species could have exerted. Indeed, the AF was not as rapid as it was in Tempranillo. The establishment of inoculated yeas<sup>t</sup> species in Grenache must sample were not so successful that the observed in Tempranillo samples, in effect, the diversity of indigenous and inoculated non-*Saccharomyces* stayed until the day 4 and after this, *S. cerevisiae* became the majority.

The Graciano must had a similar microbial composition to that observed for Tempranillo. *H. uvarum* was the most frequently detected species and *T. delbrueckii* was initially present in the must sample. Similarly, to what was observed in Grenache, the implantation of *S. cerevisiae* in the control sample was total in spite of the high acidity and low pH of the must although the AF kinetics was very slow and lasted thirteen days. Similar to the described in Grenache, the establishment of inoculated non-*Saccharomyces* species was even less successful in percentages of identification.

#### *4.2. Discriminant Analysis of Wines after AF*

#### 4.2.1. Statistical Analysis of Oenological Parameters

In order to know how the wine samples were separated depending only on the must inoculation strategy, the statistical analysis was performed without the ABV and the malic acid content that separated the must samples in the discriminant analysis.

The early inoculation of *S. cerevisiae*, *M. pulcherrima* and the mix of *L. thermotolerans* and *T. delbrueckii* did not provide enough changes in the oenological parameters of Graciano and Grenache wine samples, so that they appeared together regardless the inoculation strategy in the representation of the two main canonical discriminate functions. Only Tempranillo samples early inoculated with *L. thermotolerans* and *T. delbrueckii* was separated in the graph, from control wine samples and from wine samples early inoculated with *M. pulcherrima*. These Tempranillo wine samples early inoculated with the mix of *L. thermotolerans* and *T. delbrueckii* were characterized by a low pH and a high total acidity. As far as we know, this is the first time that the mixed inocula of *L*&*T* (30/70) has been tested in a pilot plant in three different grape varieties. Results showed that in the Tempranillo must, both yeasts achieved a total implantation maintaining a ratio of 13/87. Post AF, the inoculated wine had interesting increased

acidity parameters due to the capacity of *L. thermotolerans* to produce lactic acid [5], which could achieve balance in a grape variety generally characterised by high pH and low acidity.

#### 4.2.2. Statistical Analysis of Colour Parameters

Analysing statistically the colour parameters of the wine samples of the three grape varieties early inoculated with *S. cerevisiae*, *M. pulcherrima* and a mix of *L. thermotolerans* and *T. delbrueckii* provided similar results to the described for oenological parameters in the later Section 4.2.1. Again, Graciano and Grenache wine samples were separated only for being di fferent grape varieties, but not because of the three di fferent yeas<sup>t</sup> inoculation strategies. Moreover, the Tempranillo control wine samples and the samples early inoculated with *M. pulcherrima* reached high values of total polyphenol index, while samples early inoculated with a mix of *L. thermotolerans* and *T. delbrueckii* yeasts reached lower values what make them stay separated in the graph of the two main canonical functions extracted from the discriminant analysis. In one previous study, of this same mix of *L. thermotolerans* and *T. delbrueckii* was tested for oenological parameters and anthocyanins and stilbenes and similar results were described [14]. In general terms, the reduction of the total polyphenol index is not a good result for wine quality, but observing this e ffect only on Tempranillo that is a grape variety characterized for normal anthocyanins content, might not be so negative than if it happened in Grenache that has a low anthocyanins content [24].

#### 4.2.3. Statistical Analysis of Aromatic Profile

Results of the aromatic profile of the three varieties inoculated with di fferent strategies showed interesting results. For instance, wine samples were mainly separated in the graph of the two main discriminant functions by the content of propanol-1 compound that provide alcoholic and mature fruit notes. Any other aromatic compound was able to discriminate samples. Graciano wine control samples, with lower propanol-1 concentrations, were separated from samples that had been early inoculated with *M. pulcherrima* and with the mix of *L. thermotolerans* and *T. delbrueckii*, this would mean that early inoculation of non-*Saccharomyces* yeas<sup>t</sup> in Graciano must samples led to a more alcoholic profile than the samples inoculated only with *Saccharomyces*. Grenache wine samples were also separated by propanol content but in this case, the samples inoculated with *S. cerevisiae* had lower concentrations than the early inoculate with *M. pulcherrima* and these ones than the early inoculated with a mix of *L. thermotolerans* and *T. delbrueckii*. Just this same result was observed for Tempranillo wine samples. Giudici et al. [25] published that the higher alcohol n-propanol was directly related with the ability of some yeas<sup>t</sup> strains to metabolise methionine and threonine aminoacids and depended on their initial content in wine, what could explain why the same inoculation strategy led to di fferent concentration of propanol in wines depending on the variety. In any case, odour threshold for propanol was established by Peinado et al. [26] in 306 mg/<sup>L</sup> that was very high comparing concentrations obtained in the current study. This means that the di fferent concentration between wine samples observed in the current research would not probably led to a di fferentiation in sensory terms.

Furthermore, the three Tempranillo wine samples were slightly di fferenced by the hexyl acetate content. In this way, the control wine sample was the one with the highest content of hexyl acetate compared to the early inoculated with *M. pulcherrima* and with a mix of *L. thermotolerans* and *T. delbrueckii.* The hexyl acetate aromatic compound is related to apple, cherry, pear and floral aromas and the odour threshold is 1.5 mg/<sup>L</sup> [27]. Only Tempranillo samples inoculated with *S. cerevisiae* overcame this threshold so that it would be fruitier than the Tempranillo samples early inoculated non-*Saccharomyces*.

#### *4.3. Discriminant Analysis of Wines after MLF*

Aromatic composition of wines after MLF was not considered because this fermentation was seeded with one commercial strain of *O. oeni*, so that di fferences in aroma could probably be due to the e ffect of this strain but not to the di fferent inoculation strategies. Multivariate statistical analysis of oenological and colour parameters of samples of the three varieties showed clearly that early inoculation of Tempranillo and Grenache varieties with a mix of *L. thermotolerans* and *T. delbrueckii* caused separation of control wine samples while the early inoculation of Graciano with *M. pulcherrima* was the wine that was di fferenced of control Grenache wine samples.
