2.3.2. Analytical Determinations

Residual sugar, organic acid, total and volatile acidity and pH were determined according to O fficial EU Methods (EC 2000). Total polysaccharides concentration was evaluated by HPLC [35], using a Varian instrument equipped with auto-sample injector (loop 20 μL) and coupled with refractive index detector. For the separation of total polysaccharides, a column Progel-TSK G-OLIGO PW (Supelco 808031) and a TSK-gel PW (Supelco 808034) precolumn were used with isocratic elution (NaCl 0.2 M; 0.8 mL/min; 40 ◦C). Before injection, the samples were filtered (1.2 μm) and purified on polyamide SC6 (Macherey-Nagel, Dylan, Germany). Polysaccharides quantification was performed by comparison with an external calibration curve of mannans from *S*. *cerevisiae* (M7504, Sigma-Aldrich, St. Louis, MO, USA), at concentrations ranging from 50 mg/<sup>L</sup> to 500 mg/L.

Total polyphenols in wines were determined by the Folin–Ciocalteu method according to Di Stefano [41] and expressed as catechin equivalents in mg of catechin/L, whereas total anthocyanins were determined by direct reading of the absorbance at 540 nm of wine in hydrochloric ethanol solution [42] and expressed in g malvidin/L [43].

Higher alcohols and acetaldehyde were determined by a GC method with flame ionization detector (FID) detection at 250 ◦C, on a Carlo Erba HRGC 5300 instruments equipped with a glass column (2 m; 2 mm ID) packaged with Carbopack C + 0.2% Carbowax 1500, 80–100 meshes (Supelco). The other chromatographic conditions were as follows: temperature gradient from 45 ◦C to 160 ◦C (3 ◦C/min), held to 160 ◦C for 20 min; inj. temperature 220 ◦C; carrier: Helium 2 mL/min; injection volume: 1 μL of distilled sample spiked with 3-methyl-2-butanol as internal standard. The acquisition and elaboration of the FID signal was carried out by means of Galaxy software (Varian Inc., Walnut Creek, CA, USA).

Minor volatile compounds were evaluated by capillary gas-liquid chromatography as previously reported [36]. In particular, the analyses were carried out on a Carlo Erba HRGC 5300 instrument, injecting an ether/hexane extracts (1/1, *v*/*v*) of the wine samples previously spiked with 3-octanol as internal standard. The chromatographic conditions were as follow: glass capillary column 0.25 μm Supelcowax 10 (60 m length, 0.32 mm internal diameter, 0.25 μm film thickness). One μL was injected in split-splitless mode (60 s splitless); carrier gas: helium at 2.5 mL/min flow rate; injection temperature: 220 ◦C; elution temperature gradient: from 50 ◦C (held 5 min) to 220 ◦C (3 ◦C/min); detection by flame ionization detector (FID) at 250 ◦C. The acquisition and integration of the FID signals were carried out using the Galaxy software (Varian Inc., Walnut Creek, CA, USA) and the content of each compound was evaluated in respect to an external standard curve. All analyses were carried out in double from each fermentation tank.

#### 2.3.3. Sensory Evaluation of Wines

The wines were left to mellow for about four months after bottling, before sensory evaluation. Wine tasting was performed by an 11 member trained and formed panel, in two sessions. Organoleptic evaluations were conducted by quantitative descriptive analysis, using a pre-defined protocol and descriptive terminology, previously developed by the tasting group. In particular, every sample was tasted twice by each taster, within completely randomized blocks, and the panelists were asked to express their judgment by quantification of each sensory descriptor (color intensity, floral, fruity, preserved fruit, spicy, candy, sulfur, chemical, earthy, mouthfeel volume, acidity, tannic intensity, astringency, dryness, bitterness) on the basis of a four point structured scale, from 0 to 3 for the olfactory descriptors (no presence of the perception to high intensity), or from 1 to 4 (low to high intensity) for color intensity and mouthfeel descriptors. The olfactive descriptors were chosen previously in a round table session of the panel basing on a free profile description of the same wines and the judges were subsequently trained on them.

For evaluation of gustative descriptors, the method used was that developed by ICV (Institute Cooperative du Vin, Lattes, France) for the Quantitative Descriptive Analysis of red wine [44,45].
