*2.4. Analytical Techniques*

The analysis of fermentative volatile or aromatic compounds after the AF was performed using the method described by Ortega et al. [21] with some modifications. The extraction was carried out by 4 mL of sample, 9 mL of (NH4)2SO4 saturated solution, 40 μL of internal standard solution (2-butanol, 4-methyl-2-pentanol, 4-hydroxy-4-methyl-2-pentanone, 2-octanol, and heptanoic acid, 40 mg of each of them in 100 mL of ethanol) and 300 μL of dichloromethane in tubes. The tubes were shaken for 1 h at 32 × g and then centrifuged at 3220× *g* for 10 min. Once the phases were separated, the dichloromethane phase was recovered. Two μL was injected onto a Hewlett-Packard (Palo Alto, California, CA, USA) 6890 series II gas chromatograph. Separation was carried out with a DB-Wax capillary column (60 m × 0.32 mm I.D. × 0.5 μm film thickness; J&W Scientific, Folsom, CA, USA).

## *2.5. Malolactic Fermentation*

After AF, the wines were drawn o ff the lees and transferred to 15 L containers that were inoculated with the commercial LAB *Uvaferm alpha* ® (Lallemand Bio S.L., Toronto, Canada) to carry out the MLF, at a temperature of 20 ◦C. The evolution of the fermentation was controlled by periodic determination of the malic acid content (g/L). After this, the wines were sulphited again and bottled. One month after MLF had ended, the wines were again analysed in terms of oenological and colour parameters, including the parameters described above for the AF (Section 2.2).

## *2.6. Statistical Treatment*

The statistical analysis of physicochemical data consisted of two multivariate analysis performed with discriminant analysis and classification by a hierarchical cluster. The analysis of the discriminate capacity of the oenological parameters was assessed for each replicate (n = 3) of must and wines after AF. The hierarchical cluster was built with the averages of every oenological and colour parameter assessed by triplicates (n = 3) for the oenological parameters of samples after FML. Both analyses were carried out by using the statistical package IBM SPSS Statistic 20.0 (Chicago, IL, USA). Raw data of replicates employed for statistical analysis could be consulted in the Spreadsheet S1.
