Comparison of Chromatic and Spectrophotometric Properties of White and Red Wines Produced in Galicia (Northwest Spain) by Applying PCA

Wine is a complex matrix composed of numerous substances and color has an important influence on its quality and consumer acceptance. Color is affected by numerous factors such as pre-fermentation and fermentation operations, ageing, contact or addition of certain substances. In this study, different chromatic parameters were determined in 99 wines (58 red and 41 white) made from Galician (Northwest Spain) grape varieties. These parameters were obtained by using simple, rapid, and inexpensive spectrophotometric methodologies: color intensity, hue, total polyphenols content (Total Polyphenol Index TPI, Folin–Ciocalteu index, FCI), total anthocyans, total tannins, and color coordinates measured by the CIELab system. The influence of ageing in barrels (red wines) or using chips (white and red wines) on these parameters was also studied. A principal component analysis (PCA) was carried out to characterize the wines according to their chromatic characteristics. Application of PCA to the experimental data resulted in satisfactory classifications of studied white and red wines according to the cited enological practices.


Introduction
Galicia, a region located in Northwest Spain, is one of the Spanish regions with more wine "Denominations of Origin" (DO): 5 in total ("Monterrei", "Rías Baixas", "Ribeira Sacra", "Ribeiro" and "Valdeorras") (see Figure 1). However, unlike what happens with other Spanish DOs, the surface dedicated to grape crops is reduced. Approximately 10,900 ha were dedicated to vineyards in 2018 [1]. In the last decade, the wine production has been increased around 37% (mean value), although this percentage is not uniform for the 5 DOs (see Table 1).
Orography in this area has designed a viticulture based on "smallholding", where the viticultor proportion is higher compared to the total vineyard surface. There are numerous wineries: between 27 for Monterrei DO and 179 for Rías Baixas DO. Thus, relation liter/winery moves between 45,000 (Ribeira Sacra DO) and 166,000 for Rías Baixas DO (Table 1).
All the facts showed in Table 1 carry a lower input for laboratory analyses; therefore, the availability of sophisticated equipment to analyze grapes and wines is scarce. In general, the wineries have the basic instrumentation and big wineries have usually developed conjoint research projects with universities. Normally, university research groups provide chromatographic equipment, and wineries possess spectrophotometers or colorimeter apparatus that are ease of use as a routine analytical technique [2]. chromatographic equipment, and wineries possess spectrophotometers or colorimeter apparatus that are ease of use as a routine analytical technique [2].  Spectrophotometric methods have been extensively used in wine production to check maturity and quality parameters in grapes and wines as color, polyphenols, and their changes with different viticulture practices (grape variety, type of soil, climate, vineyard conducting systems...) and oenology treatments as yeast or enzymes addition, storage (in bottle) or aging processes (in wood barrels or with chips) [3]. These methods are adequate to be developed in small wineries that cannot afford the cost of sophisticated apparatus to test the wine quality. Moreover, these methods are simple, economic, and less time consuming than chromatographic methods (HPLC or GC), which also require previous expertise and more cost related to human resources, equipment, and other facilities.
Chromatic profile and phenolic composition of wines are increasingly used to characterize and typify them. Some authors reported that phenolics constitute a promising  Spectrophotometric methods have been extensively used in wine production to check maturity and quality parameters in grapes and wines as color, polyphenols, and their changes with different viticulture practices (grape variety, type of soil, climate, vineyard conducting systems. . . ) and oenology treatments as yeast or enzymes addition, storage (in bottle) or aging processes (in wood barrels or with chips) [3]. These methods are adequate to be developed in small wineries that cannot afford the cost of sophisticated apparatus to test the wine quality. Moreover, these methods are simple, economic, and less time consuming than chromatographic methods (HPLC or GC), which also require previous expertise and more cost related to human resources, equipment, and other facilities.
Chromatic profile and phenolic composition of wines are increasingly used to characterize and typify them. Some authors reported that phenolics constitute a promising class of compounds used to categorize wines [4]. Color is one of the main quality parameters in a wine and variations in wine types are largely due to the concentration and composition of wine phenols [5], anthocyans being the main contributors to a red wine color [6]. In fact, for red wines, the color is very relevant for their quality [7] and for consumer acceptance [8]. Moreover, the color influences sensory properties such as flavor, taste, and aroma [9][10][11][12]. Regarding white wines, there are significantly fewer studies related to the color and the phenolic composition, in comparison with red wines [13,14]. There are a large number of typical white wines in this Spanish geographical area which have not yet been studied extensively. In particular, there are very few works related to polyphenols in white wines from the Northwest Spain region [15,16], particularly in the case of wines obtained from autochthonous white grape varieties as Godello, Albariño, Loureira, or Treixadura.
The tendency of wine to improve, or at least change during aging, is one of its more fascinating properties [17]. Normally, the aging process is used in wine to stabilize it and to improve its quality. The aging processes modify sensory properties in wine as it is accompanied by the development of color, aroma, and flavor [18]. A traditional barrel can be effectively substituted by ageing with oak chips to improve color and woodaromas [19]. Enzyme addition is a known oenological practice that improves anthocyanins' extraction [20] and therefore color extraction. Yeast addition is another practice that can stabilize wine color [21]. Thus, all these practices were checked to measure various chromatic and color properties in order to classify wines.
Some works have studied spectrophotometric parameters used to characterize phenolic composition and chromatic properties in Galician young red wines [22,23], but there are very few papers reporting these parameters in Galician wines subjected to aging processes [24]. Regarding other Spanish wines, very few works have reported the effect of aging on color properties and phenolic composition of white wines [25][26][27] being more numerous the ones related to red wines [28,29].
Principal Component Analysis (PCA) is a statistical tool used to find correlations between wine properties and different treatments, and has been used effectively in some works, where color properties were analysed in wines [30,31] and allowed their classification [32].
The main objectives of this work were the determination of chromatic characteristic and total polyphenols in 99 wines (58 red and 41 white) produced in Galicia by measuring spectrophotometric parameters and compare these results in order to find differences and similarities in wine profiles by PCA. Some of the studied wines were obtained by different oenological processes (aging with wood barrels or chips, addition of enzymes and yeasts), and most of them were monovarietal wines. The establishment of chromatic relations between all the parameters considered: phenolic compounds (tannins, anthocyans) spectrophotometric measurements (color intensity, CI; Tint or Hue; CIELab coordinates; total polyphenol index, TPI; Folin-Ciocalteu index, FCI) will help the winery to focus on the main measurements to typify their wines correctly and will be a quality tool in order to consider a certain variety adequate for oenological treatment or aging processes.

Spectrophotometric Determinations
Regarding color intensity, for red wines (Table 2), the highest values were obtained for control wine and for aged wines in oak barrels or with oak chips (samples R1-R35). The commercial samples (R36-R58) gave the lowest values, especially the ones made with Brancellao and Merenzao grapes, due to their low content of anthocyanins. Color intensity values did not differ from other red grape varieties for control samples [33,34]. Color components (yellow, red, and blue) follow a similar tendency in aged wines for yellow and blue colorations, showing low values at 3 months barrel/oak contact, increasing after 5 and 7 months contact and reaching initial values after 9 or 12 months contact (see Supplementary Table S1). In these samples, the red component is high after 3 months and decreases after 5 and 7 months to be recovered at almost initial levels after 9 or 12 months of aging. The ratio between yellow and red colorations (A420 nm/A520 nm) corresponded to the tonality or hue, which gives an estimation of the color change toward the orange tones observed in wines during aging [2]. Tint or hue values in aged wines were higher in samples aged after 5 and 7 contact months and much lower in samples aged during 9 and 12 months, similar to what happens with other grape varieties [35].  In white wines (Table 3), color intensity varies between 0.3 and 1.5-higher values in comparison with other white grape varieties [36]. One study measures color intensity as absorbance at only 420 nm, without considering the contribution of red and blue colors [37]. Logically, the yellow component was contributing to a greater degree in CI, being the highest value in our study at 65.3% (sample W36). Similarly, in other work [38], control wine samples showed the lowest color intensity values, not being significant in our results. For aged white wines, the contact with oak chips has a little influence in yellow color (values between 46.9% and 52.1%, samples W10 and W12, respectively) (see Supplementary  Table S2).
Color measured by CIELab coordinates showed interesting results. Luminosity in red wines reached the highest values, oscillating between 95.5 and 97.9 in aged samples for three months, and these values were decreasing up to around 30.0 after 9 and 12 months of aging. Other authors have observed a similar effect, i.e., the wines darkened (lower L*) after aging, attributing this to their higher phenolic content [39]. Chroma data were presenting the lowest values for the red wines samples aged for 3 months.
For white wines, luminosity was quite similar in all the samples, varying from 94.5 (sample W18) to 103 (sample W33). Lower values for L coordinate were also found in other studies with white wines from other variety [40]. Regarding Chroma, in general, the samples aged with oak chips have lower values compared to commercial wines.
TPI and FCI are spectrophotometric parameters that provide winemakers with enough information about polyphenol concentration. In aged red wines, the higher values for TPI correspond to the samples with the longest contact with wood/oak-chips; this is provoked by the extraction of more polyphenols from wood. In white wines, IPT values are 10 times lower compared to red wines; contrary to red wines' samples, IPT values are lower in aged samples with oak-chips in comparison with commercial wines. It is well known that anthocyanins contribute to the red color of a wine and are present at a low concentration in white wines, being 50 times lower than in red wines, considering also flavonoids and catechins [41]. In our aged red wine samples, there is a global loss of anthocyans during aging, an effect also observed in other study [42], probably due to polymerization and reactions with other wine compounds.
Tannins are one of the critical classes of phenolic substances that undergo significant changes during winemaking. Total tannins in our red wines samples are higher in those samples aged for 9-12 months (10.1-14.5 g/L of cyanidin). Other study also found higher total tannins in aged wines for other grape varieties [43]. Tannins are present in low concentrations for white wines (lower than 0.15 g/L of cyanidin).

Principal Component Analysis
Principal component analysis is the multidimensional technique most applied in sensory profiles, as it does not require and structure on samples (wines), and the number of variables has no limit [44]. This procedure extracts the dominant patterns in the data matrix in terms of a complementary set of scores and loading plots. PCA permits us to achieve a reduction of dimensionality, a data exploration finding relationships between objects, estimating the correlation structure of the variables and investigating how many components (a linear combination of original features) are necessary to explain the greater part of variance with a minimum loss of information. When PCA is performed on autoscaled matrix data, the principal component loadings are eigenvectors of the correlation matrix [45]. Therefore, it is a proper tool to typify wines according to their chromatic properties.
PCA seeks to establish and form the parameters analyzed, if the studied wines of our region differ or resemble each other. This requires finding the parameters specific enough to enable us to characterize our wines. In other words, the aim is to establish, on the basis of the parameters analyzed, whether the wines of our region are different or similar to each other. To do this, it is necessary to find sufficiently specific parameters that allow us to characterize our wines. This differentiation is much more difficult when the aim is to differentiate among wines of the same variety and grown in bordering areas where the climatological component has a dominant role [46].
PCA explains the pattern of correlations between a set of observed variables. In this study, the 12 analyzed variables were reduced to 8 and were used for 58 red wines, and to 5 variables for 41 white ones.

Principal Component Analysis in Red Wines
Variables used for red wines were: CI (color intensity), tint, C (chroma) and L (luminosity), TPI (Total Polyphenol Index), FCI (Folin-Ciocalteu Index), anthocyans, and tannins. The CIELab coordinates (a and b) were not considered as they are included in chroma calculation. Correlation between variables was adequate and the first discriminant functions obtained represented 80% of the total variability. Similarly to other studies for young red wines, chromatic parameters have a significant correlation with anthocyan pigments (around 0.5) [47].
A sample plot along first and second discriminant functions is showed in Figure 2. As it is observed in Figure 2, five wine groups were established, labeled as A, B, C, D, and E. Group A is formed by Mencía aged wines for 3 months in 4 tonnelleries (labels 2, 8, 12, 17 and 21) plus one control wine (label 1).
Regarding monovarietal wines, the ones obtained with Brancellao grapes are in group C (one sample is in group D, label 42) and the ones from Sousón grapes are inside group D (except one sample in C group, label 40). Wines made with Merenzao grapes are in group D (labels 36,45). These wines share properties of aged wines. This analysis suggests that both grape variety and aging conditions clearly modify chromatic properties in the studied red wines. This analysis also permits discarding samples not sharing general properties inside each group.
Commercial samples (labels 51-58) are elaborated with more than one grape variety and most of them are placed in group D. In this case, it is difficult to establish exhaustive conclusions due to less information related to oenological treatments. It is known that parameters as color and polyphenols can be modified with different viticulture practices as grape varieties and oenological treatments or aging processes [3,48].

Principal Component Analysis in White Wines
PCA in white wines is usually applied to characterize its aromatic profile. As far as we know, no work related to PCA involved in wine chromatic properties (measured by spectrophotometry) has yet been published. There are some articles related to polyphenolic compounds measured by chromatography [49]. They are much fewer studies related to the color and the phenolic composition in white wines in comparison with red wines In particular, the works related to polyphenols in Galician white wines are still few, especially in the case of wines obtained from autochthonous white grape varieties as Albariño Treixadura, Loureira, and Godello [15,16,50].
PCA in white wines showed, when calculating the correlation matrix, that TPI and L exhibited a high correlation; therefore, they were suppressed from considered variables in the analysis. Finally, only five variables were considered (FCI, CI, tint, C, and L) in 41 wine samples. The obtained plot is showed in Figure 3, where four groups were established: F, G, H, and I. Regarding monovarietal wines, the ones obtained with Brancellao grapes are in group C (one sample is in group D, label 42) and the ones from Sousón grapes are inside group D (except one sample in C group, label 40). Wines made with Merenzao grapes are in group D (labels 36, 45). These wines share properties of aged wines. This analysis suggests that both grape variety and aging conditions clearly modify chromatic properties in the studied red wines. This analysis also permits discarding samples not sharing general properties inside each group.
Commercial samples (labels 51-58) are elaborated with more than one grape variety and most of them are placed in group D. In this case, it is difficult to establish exhaustive conclusions due to less information related to oenological treatments. It is known that parameters as color and polyphenols can be modified with different viticulture practices as grape varieties and oenological treatments or aging processes [3,48].

Principal Component Analysis in White Wines
PCA in white wines is usually applied to characterize its aromatic profile. As far as we know, no work related to PCA involved in wine chromatic properties (measured by spectrophotometry) has yet been published. There are some articles related to polyphenolic compounds measured by chromatography [49]. They are much fewer studies related to the color and the phenolic composition in white wines in comparison with red wines. In particular, the works related to polyphenols in Galician white wines are still few, especially in the case of wines obtained from autochthonous white grape varieties as Albariño, Treixadura, Loureira, and Godello [15,16,50].
PCA in white wines showed, when calculating the correlation matrix, that TPI and L exhibited a high correlation; therefore, they were suppressed from considered variables in the analysis. Finally, only five variables were considered (FCI, CI, tint, C, and L) in 41 wine samples. The obtained plot is showed in Figure 3, where four groups were established: F, G, H, and I.
Group F contains the white wines elaborated with Godello grapes and aged with chips along 2 months, and these wines are clearly different from the others. In group G, except two samples: 19 and 41 (made with Godello and with a mixture of Godello and Treixadura grapes, respectively). Samples of group H are commercial wines from different trademarks, obtained from Godello grapes. Finally, group I agglutinate mainly wines made with Albariño grapes.  Group F contains the white wines elaborated with Godello grapes and aged with chips along 2 months, and these wines are clearly different from the others. In group G, except two samples: 19 and 41 (made with Godello and with a mixture of Godello and Treixadura grapes, respectively). Samples of group H are commercial wines from different trademarks, obtained from Godello grapes. Finally, group I agglutinate mainly wines made with Albariño grapes.

Wine Samples
A total of 58 red wines (coded R1-R58) and 41 white wines (coded W1-W41) from Galicia (Northwest Spain) were analyzed (Tables 4 and 5). The wines were obtained from some native Vitis vinifera grape varieties collected in the five different Galician Denominations of Origin. Some samples were commercial wines obtained from supermarkets. Other samples were obtained from wineries at the industrial or semi-industrial scale from Mencía and Godello grapes, respectively, and subjected to oak-contact (with barrel or chips) and sampling at different times: 3, 5, 7, 9, and 12 months for Mencía wines (French oak barrels from 4 tonnelleries, and contact with French or American oak chips), and 7, 15, 30, and 60 days for Godello wines in contact with chips from 2 types of French oak and 1 of American oak). Wood-contact (during 5 months) R10 Wood-contact (during 7 months) R11 Wood-contact (during 12 months)

Wine Samples
A total of 58 red wines (coded R1-R58) and 41 white wines (coded W1-W41) from Galicia (Northwest Spain) were analyzed (Tables 4 and 5). The wines were obtained from some native Vitis vinifera grape varieties collected in the five different Galician Denominations of Origin. Some samples were commercial wines obtained from supermarkets. Other samples were obtained from wineries at the industrial or semi-industrial scale from Mencía and Godello grapes, respectively, and subjected to oak-contact (with barrel or chips) and sampling at different times: 3, 5, 7, 9, and 12 months for Mencía wines (French oak barrels from 4 tonnelleries, and contact with French or American oak chips), and 7, 15, 30, and 60 days for Godello wines in contact with chips from 2 types of French oak and 1 of American oak).

Analytical Methods
All spectrophotometric determinations were performed, in triplicate, diluting when necessary, using a UV-vis spectrophotometer (Hitachi U-2000) with 0.1 cm or 1 cm path length glass or quartz cell, and all absorbance values were corrected to 1 cm path length.
Total polyphenol index (TPI) was measured spectrophotometrically measuring the absorbance of the wine diluted with water (100-fold for the red wines and 10-fold for the white wines) at 280 nm [TPI = A 280 × Dilution factor].
For the Folin-Ciocalteu Index (FCI), the wine was diluted with water (5-or 10-fold) and added to the Folin-Ciocalteu reagent and Na 2 CO 3 , and then was measured by spectrophotometry at 760 nm [FCI = A 760 × Dilution factor × 20]. Table 4. Nomenclature and characteristics of red wine simples.

Wine
Code Treatment Variety Table 5. Nomenclature and characteristics of white wine samples.

Anthocyan Determination
Total anthocyans (TA) were analyzed according to Ribéreau-Gayon and Stonestreet's method [54]. The wine sample (1 mL) and ethanol (1 mL) were diluted with 20 mL of HCl (2%), and then divided into two tubes. Into one of the tubes, 10 mL of this mixture were mixed with 4 mL of distilled water (4 mL), and the other had 4 mL of sodium metabisulfite (at 15%, w/v) added. After 20 min of reaction, the absorbance of both tubes was measured at 520 nm (A 1 and A 2 , respectively). The TA content is calculated as follows: TA = (A 1 − A 2 ) × 875, and expressed in mg malvidin/L.

Tannin Determination
Total tannins (Tan) were analyzed spectrophotometrically at 550 nm in the wine diluted with water and hydrochloric acid and heated (A 550 ) vs. wine diluted in the same way but not heated (A' 550 ), following the method described by Zamora Marín [3]. The results are expressed as: Tan = (A 550 − A' 550 ) × 19.33.
Total tannins (as g/L cyanidin) were quantified following the Ribéreau-Gayon and Stonestreet methodology [55]. Stock solutions of cyanidin were prepared by dissolving the compound in methanol, stored at 4 • C in the darkness, and subjected to the same protocol.

Statistical Analysis
Statistical comparisons between both the red and white wines were made using the Student's t-test, and the least significant differences (LSD) were calculated (p < 0.05) to determine significant differences between wines. By using the SPSS software version 19 (SPSS Inc., Chicago, IL, USA), the mean averages of all data for each type of wine (red and white) were analyzed by Principal Component Analysis (PCA), which is a multivariate technique that analyzes a data table in which observations are described by several inter-correlated quantitative dependent variables. Its goal is to extract the important information from the table, to represent it as a set of new orthogonal variables called principal components, and to display the pattern of similarity of the observations and of the variables as points in maps [56].
Discriminant analysis is the most frequently used statistical technique to classify and differentiate wines [57]. This statistical tool was usually applied to differentiate and typify wine samples, using diverse variable types as sensory data [58,59], volatile compounds [57], spectrophotometry measurements [60] and chromatographic data [61][62][63]. Therefore, it is a proper tool to typify wines according to their chromatic properties.

Conclusions
Galician wineries could only perform methodologies involving inexpensive equipment as spectrophotometers to check the grape maturity and other quality characteristic in grapes, musts, and wines. Chromatic properties in wines (red and whites) are highly influenced by varietal grape and oenological treatments like age (with chips or barrels, this last case only for red wines). In red wines, the different chromatic properties in wines made with several grape varieties are remarkable. The red color, presenting lower values after 3-5 months of wood contact, is being stabilized after 12 months, and these results follow the same tendency in total anthocyans. In general, Luminosity and Chroma are changing about one third in red wines aged in oak barrels for 12 months, in comparison to ageing for 3 months, but not in wines aged with oak-chips. Ageing with chips in white wines is not so crucial to appreciate chromatic differences. The probable different enzymes and yeast used in commercial white wines do not have a great influence on the development of color, and the PCA grouping of these samples is more disperse. Total tannins and total anthocyans are not considered adequate parameters to classify samples by PCA in white wines.
Commercial samples are also more difficult to classify both for red and white wines. In many cases, the grape used is unknown, and this fact makes the comparison and grouping difficult. Briefly, PCA using few variables (less than 8 for red wines, anthocyans, tannins, FCI, CI, tint, TPI, chroma and L-, and 5 for white wines-FCI, CI, tint, chroma and L-), obtained by simple and inexpensive methods, is an efficient statistical tool allowing for classifying/typifying wines considering ageing and discarding the samples with defects/anomalies.