6 mm i.d.

5 μm particle size. The mobile phases consisted of A (water–acetonitrile–acetic acid, 67:32:1 v/v/v) and B (water–acetic acid, 99:1 v/v). The gradient elution conditions were as follows: 0 min (20% A + 80% B); 4 min (30% A + 70% B); 8 min (40% A + 60% B); 12 min (65% A + 35% B); 16 min (80% A + 20% B); 20 min (95% A + 5% B); 21.8 min (97% A + 3% B); 24 min (100% A) and 60 min (100% A). The flow rate was 0.8 mL min−1 and the injection volume was 20 μL ( Quirós, Lage-Yusty, & López-Hernández, 2009). Before analysis, wines were filtered with a 0.45 μm filter with a PTFE membrane (Millipore, São Paulo, Brazil). The programmable variable wavelength UV–Vis detector system allows detection at different wavelengths; so λ = 360 nm was set to rutin and kaempferol LY2109761 and λ = 373 nm was set to myricetin and quercetin. The fluorescence detector was set at λem = 392 nm and λex = 300 nm for trans-resveratrol. buy GDC-0973 Data were presented as mean ± pooled standard deviation. A bivariate linear correlation matrix of the data, displayed in Pearson’s correlation coefficient (r), was produced to measure the association between the response variables, and the significance (p-value) of such correlations was also provided. Retail price, antioxidant activity measured by ORAC and DPPH, and overall sensory perception of quality were used to classify

the set of red wines using hierarchical cluster analysis (HCA) ( Fig. 1). For this purpose, the values were autoscaled, and

sample similarities were calculated based on the Euclidean distance and the Ward hierarchical agglomerative method. To characterise the red wines in each of the four suggested clusters, Hartley’s or Levene’s test was applied to check for homogeneity of variances, and one-way ANOVA and Tukey’s HSD post hoc tests were then conducted to identify contrasts among clusters. For the variables that presented non-homogenous variances (p < 0.05), the equivalent to ANOVA non-parametric test was used. p-Values below 0.05 were considered significant. Statistica 9.0 software (Stat-Soft, Tulsa, OK, USA) was used for all statistical procedures. The results (Table 1) showed that the inhibition of DPPH ranged from 47.93% to 66.70%, while HSP90 the ORAC results varied from 13.87 mmol to 35.11 mmol TE/L. The redness of the wine varieties, measured by the a∗ coordinate, ranged from 39.17 to 52.60, while the colour intensity (C∗) ranged from 43.14 to 64.61. The phenolic compound contents varied within grape varieties and also within countries, as observed in Table 2: trans-resveratrol (1.56–4.30 mg/L), quercetin (5.18–21.81 mg/L), rutin (0.83–4.19), gallic acid (13.88–69.87 mg/L), caffeic acid (2.74–4.95 mg/L), epicatechin (19.75–44.53 mg/L), catechin (59.15–149.14 mg/L), myricetin (13.03–46.69 mg/L), ferulic acid (0.55–1.45 mg/L), p-coumaric acid (4.40–10.73 mg/L), vanillic acid (0.00–1.