3.2.3. Graciano

The species identified in Graciano samples are shown in Figure 2B. The initial Graciano must (day 0) had 75% *H. uvarum*, 12% *S. cerevisiae*, and 13% of *T. delbrueckii* and *H. osmophila*. Control samples were composed of 100% *S. cerevisiae* at each sampling checked. The samples inoculated with *M. pulcherrima* (M) at the third day had 17% *M. pulcherrima* and 83% *S. cerevisiae.* One day later (day 4), *S. cerevisiae* was 93% and one week later (day 7) all the samples were composed of *S. cerevisiae*. The samples inoculated with *L. thermotolerans* and *T. delbrueckii* (*L*&*T*) three days later (day 3) were made up of 7% *H. uvarum*, 43% *S. cerevisiae*, 43% *T. delbrueckii* and 6% *L. thermotolerans*. One day later (day 4), samples had 87% *S. cerevisiae* and one week later (day 7), they were composed solely of *S. cerevisiae*.

The Graciano AF kinetics (Figure 3C) of control samples and samples early inoculated with non-*Saccharomyces* yeasts were similar regardless of the inoculated yeasts used and took thirteen days to complete.

#### *3.3. Characterisation of Wines*

The statistical CDA of the oenological parameters of the samples of Tempranillo, Grenache and Graciano wines after AF, are shown in Figure 4. The variability between the samples (n = 3) was explained by four possible canonical functions (F) with statistical significance. F1 explained over 78.9% of variability and F2 15.2%, both explaining the 94.1% of the variance. Four out of the five assessed parameters contributed to the separation along F1, but the pH was the most influencer. F2 was mainly loaded by the total acidity, F3 and F4 (not included in the graph) were loaded by the volatile acidity. The three samples of Graciano wines stayed close in the negative part of axis F1 and Grenache and Tempranillo wines were separated by the axis 2. The three Grenache wines were clustered together. The sample of Tempranillo early inoculated with *L. thermotolerans* and *T. delbrueckii* (LT) was separated from the other two types of Tempranillo wines (control –C- and inoculated with *M. pulcherrima –M-*).

Results of statistical CDA of the colour parameters of the samples of Tempranillo, Grenache and Graciano after AF, are shown in Figure 5. The variability between the samples (n = 3) was explained by four possible canonical functions (F) with statistical significance. F1 explained over 89.2% of variability and F2 8.3%, both explaining the 98.5% of the variance. The six colour parameters analysed contributed to the separation along F1 and F2, but the most important one was the total polyphenol index. F3 and F4 (not included in the graph) were loaded by the hue and the colour intensity, respectively. The three samples of Graciano wines stayed together in the positive part of axis F1 and Grenache and Tempranillo wines were in negative part of axis F1 and separated by the axis 2. The three Grenache wines were clustered together in the positive part of axis F2. The samples of Tempranillo were placed in the negative part of both axis and the samples early inoculated with *L. thermotolerans* and *T. delbrueckii* (LT)were separated from the other two types of Tempranillo wines (control –C- and inoculated with *M. pulcherrima –M-*).


**Figure 4.** (**A**) Canonical discriminant graph of oenological parameters in control samples (C), samples early inoculated with *M. pulcherrima* (M) and samples early inoculated with a mix of *L. thermotolerans* & *T. delbrueckii* (LT), after alcoholic fermentation (2.) of Tempranillo (Temp), Grenache (Gre) and Graciano (Gra) (**B**) Standardized canonical coefficients of the four main discriminant variables in functions (F1 and F2) for oenological parameters.

**Figure 5.** *Cont*.


**Figure 5.** (**A**) Canonical discriminant graph of colour parameters in control samples (C), samples early inoculated with *M. pulcherrima* (M) and samples early inoculated with a mix of *L. thermotolerans* & *T. delbrueckii* (LT), after alcoholic fermentation (2.) of Tempranillo (Temp), Grenache (Gre) and Graciano (Gra). (**B**) Standardized canonical coefficients of the four main discriminant variables in functions (F1 and F2) for colour parameters.

The statistical CDA of the aromatic compounds of samples of Tempranillo, Grenache and Graciano after AF, are shown in Figure 6. The variability between the samples (n = 3) was explained by four possible canonical functions (F) with statistical significance. F1 explained over 99.1% of variability and F2 0.4%, explaining the 99.5% of the variance. 12 alcohols and six esters out of the 34 aromatic compounds measured, contributed to the separation along F1 that was mainly loaded by propanol-1 compound and F2 by the hexyl acetate contents. The F3 by 2-phenylacetate and F4 by ethyl-3-hidroxibutyrate although not included in the graph. The three samples of Graciano wines stayed together in the negative part of axis F1, being separated the control sample (C) of the other two samples. Grenache and Tempranillo wines were separated by the axis 1 but in the positive part. In this case, wines were separated, being the control samples of Tempranillo and Grenache very close while the samples of both varieties but early inoculated with *L. thermotolerans* and *T. delbrueckii* (LT) were quite distant.

The MLF of each wine was completed without problems (data no shown). Six months after completion of MLF and bottling, the wines were analytically analysed in the colour and oenological parameters described for samples after AF and the hierarchical cluster built with the average data is shown in Figure 7.

**Figure 6.** *Cont*.


**Figure 6.** (**A**) Canonical discriminant graph of aromatic compounds in control samples (C), samples early inoculated with *M. pulcherrima* (M) and samples early inoculated with a mix of *L. thermotolerans* & *T. delbrueckii* (LT), after alcoholic fermentation (2.) of Tempranillo (Temp), Grenache (Gre) and Graciano (Gra) (**B**) Standardized canonical coefficients of the four main discriminant variables in functions (F1 and F2) for aromatic compounds.

**Figure 7.** Hierarchical clusters assessed with average oenological and colour parameters of control samples (C), samples early inoculated with *M. pulcherrima* (M) and samples early inoculated with a mix of *L. thermotolerans* & *T. delbrueckii* (LT), after the malolactic fermentation (3.) of (**A**) Tempranillo (Temp); (**B**) Grenache (Gre); and (**C**) Graciano (Gra).

Tempranillo and Graciano grapes inoculated early with *L. thermotolerans* and *T. delbrueckii* (LT) were separated from the control samples and from the samples inoculated early with *M. pulcherrima* (M) that were clustered together. In contrast, in Grenache wines after MLF the samples inoculated early with *M. pulcherrima* (M) were separated from the other two samples that stayed clustered together.
