Carbon-Isotope Ratio (δ¹³C) and Phenolic-Compounds Analysis in Authenticity Studies of Wines from Dealu Mare and Cotnari Regions (Romania)

Abstract

In food quality, authenticity is one of the most important issues in the context of ensuring the safety and security of consumers, but it becomes even more important when wine is involved as this produce has become one of the most counterfeited foods in the world. A batch of 28 wines from Romanian grape varieties obtained in Dealu Mare and Cotnari regions was analysed from a physical–chemical point of view in order to assess the possibility of discriminating them according to geographical origin and variety. The samples were subjected to isotopic analysis, thus determining δ¹³C isotopic ratio using an elemental analyser EA, VarioMicroCube; while the targeted analysis of phenolic compounds was performed by UHPLC technique, using an Acquity UPLC H-Class. The basic physico-chemical analysis of the samples was carried out using FTIR spectroscopy (OenoFoss). Statistical analyses were performed using the TIBCO STATISTICAL SOFTWARE and the statistical test applied was the Tukey HSD test. Therefore, identified phenolic compounds such as hydroxytyrosol and coumaric acid are considered varietal markers. Tyrosol, dimers B1 and B2 and also catechin and epicatechin are indicators of geographical origin.

1. Introduction

In order to meet the needs of producers, consumers and authorities, and also because of the important value of the wine market, in recent years there has been a global interest in developing analytical methods that provide reliable information on issues related to the authenticity of wines. These methods are meant to diminish and prevent the falsification of wines and to guarantee their quality in the import–export market [1].

Factors such as grape variety, production area, year of harvest, type of soil, technology applied, and climatic conditions are all important in determining the chemical composition of wines [2].

Determining the isotopic ratios of elements such as H, C, N, O has proven, over time, to be a useful tool in determining the geographical origin of food [3,4], a tool that helps to confirm or to deny the information on the label or to verify the conformity of the products with legislation. Regarding wine, the determination of the ratios of stable isotopes (δ²H, δ¹⁸O or δ¹³C, constituents of water or ethanol extracted from the analysed wine) can provide information on the origin of raw materials [5]. The carbon isotope ratio in ethanol after a previous extraction from wine provides relevant information to validate the geographical origin of wines. Therefore, a correlation between the carbon isotope ratio and the phenolic compounds profile can provide an overview of wines of a certain variety or region. In addition, in recent years, untargeted metabolomic analysis (mass spectrometry (MS) hyphenated with separation techniques such as liquid chromatography (LC)) coupled with statistical analysis have allowed good discrimination between wines according to the area of origin, variety, harvest year, etc. [6,7].

The type and concentration of the phenolic compounds are very different depending on grape variety, climatic conditions in each area and the applied wine-making technology [8,9]. Generally, probably the major determinant factor for the variation in the polyphenolic content of different wines throughout the world is the amount of sunlight to which the grapes are exposed during cultivation, but also the average temperature and the water status [10,11].

In Romania, viticulture has an old tradition, over 2000 years old [12], even if at present many of the old, local grape varieties are no longer used in production or they occupy very small areas. Currently, Romania ranks 6th in Europe and 12th in the world in terms of wine production, with more than 180 thousand hectares of vines. In 2021, there was an increase in wine production by 37% compared to the previous year; the Romanian wine market being one of the most dynamic markets in Europe [13]. Two of the local grape varieties that occupy large surfaces and from which wines are highly appreciated by consumers are Fetească neagră and Fetească albă. Busuioacă de Bohotin has recently become a widely cultivated grape variety in Romania and appreciated for its aromatic rosé wines. It is also considered a Romanian variety due to the long time (centuries) it has been cultivated in Romania, although its origins are Greek [14].

The study’s main aim was to correlate, by applying statistical analysis, the chromatographic analysis of phenolic compounds with δ¹³C isotopic ratio in order to identify markers for the viticultural areas and functions towards discriminating between varietals.

2. Materials and Methods

The studied wines are in fact a set of 28 white, rosé and red commercial wines made in 2019, 2018 and 2017 from the 3 Romanian varieties mentioned above: Fetească albă, Busuioacă de Bohotin and Fetească neagră. These samples have their origins in two of the most important vineyards in Romania, namely: Cotnari (located in the northeast of the country, with 2037 ha of vines) and Dealu Mare (“homeland of red wine”, with 12,327 ha). Cotnari area is characterized by a hilly landscape with slopes of between 3 and 35%, while altitudes vary from 105 to 315 m above sea level. The climate is temperate, the summers are characterised by droughts while winters are cold due to a specific icy wind ”crivăț”, the absolute minimal temperatures can reach −29 °C [15]. In Dealu Mare, conditions are a little bit different. Slopes under vines vary between 8% and 30%. The average annual temperature varies between 10.8 °C and 11.2 °C. Winters are relatively short with cold bursts in January and the first half of February. The average temperature of the warmest month (July) is + 22.4 °C, with variations between + 20.7 °C and + 25.6 °C [16].

The grape varieties, the regions, the wine colours, the years of vintage and the physico-chemical characteristics are listed in

Table 1. Physico-chemical characteristics of the studied wines.

Current Number	Region	Grape Variety	Wine Colour	Year	Density (g/cm3)	Ethanol Concentration (% v/v)	pH	Total Acidity (g/L Tartaric Acid)	Sugars (g/L)
1	Cotnari	Fetească neagră	Red	2017	0.9934	13.26	3.63	5.57	5.90
2				2017	0.9920	13.66	3.72	4.97	4.80
3				2017	0.9937	13.08	3.64	5.54	4.90
4				2017	0.9938	13.37	3.57	5.79	6.10
5				2017	0.9931	12.16	3.73	5.02	5.10
6		Busuioacă de Bohotin	Rosé	2019	0.9897	13.63	3.15	6.61	3.70
7				2019	1.0012	12.79	3.35	6.38	29.50
8				2019	0.9937	12.77	3.36	6.14	11.40
9				2019	0.9935	12.91	3.37	6.14	11.20
10				2019	0.9934	12.98	3.30	6.28	12.10
11				2019	0.9903	13.05	3.26	6.64	3.90
12				2019	0.9935	13.18	3.34	5.91	12.10
13		Fetească albă	White	2018	0.9885	12.96	3.44	4.85	2.60
14				2018	1.0018	12.53	3.25	6.11	31.50
15				2018	0.9901	13.19	3.27	6.15	4.20
16				2018	0.9920	13.52	3.16	6.63	10.30
17	Dealu mare	Fetească neagră	Red	2017	0.9900	13.61	3.44	5.49	4.80
18				2017	0.9924	14.41	3.64	5.27	5.60
19				2017	0.9938	14.25	3.83	5.04	5.80
20				2017	0.9917	14.10	3.81	4.87	3.40
21		Busuioacă de Bohotin	Rosé	2019	0.9901	12.12	3.45	6.72	1.40
22				2019	1.0053	12.30	3.66	5.48	35.20
23				2019	1.0046	12.51	3.10	7.36	38.40
24				2019	0.9908	13.41	3.39	6.28	5.80
25		Fetească albă	White	2018	0.9905	11.49	3.41	5.49	3.40
26				2018	0.9900	13.61	3.44	5.49	4.80
27				2018	0.9947	13.00	3.47	5.79	12.30
28				2018	0.9885	12.96	3.44	4.85	2.60

.

The physico-chemical analyses were determined using FTIR spectroscopy (OenoFossTM type 4101, FOSS, Hilleroed, Denmark).

Targeted analysis of phenolic compounds was performed using UHPLC technique, using an Acquity UPLC H-Class (Waters, Milford, MA, USA) equipped with a quaternary pump and an autosampler that was coupled to a fluorimetric detector (FLD) and one with diode (DAD).

The column (Restek brand, model Report ARC-18, particle size 1.8 µm, length 150 mm and internal diameter of 2.1 mm) was thermostated at 35 °C, and the samples were kept at 12 °C. The wine was filtered before injection with PTFE filters with a porosity of 0.45 µm (Restek, Lisses, France). Composition of the mobile phase: an aqueous eluent (A) was prepared, containing 0.28% trifluoroacetic acid and 5% methanol (pH = 1.6). Solvent B was methanol 100%. The flow rate was constant at 0.30 mL min⁻¹. The optimized elution system consisted of a gradient obtained as presented in

Table 2. The HPLC gradient elution.

White and Rosé Wines			Red Wines
Time (min)	Eluent B (%)	C (%)	Time (min)	Eluent B (%)	C (%)
5	3	97	0–8	3–5	97–95
5–8	3–5	97–95	8–17	5	95
8–17	5	95	17–19	5–9	95–91
17–19	5–9	95–91	19–25	9	91
19–25	9	91	25–35	9–14.3	91–85.7
25–25	9–14.3	91–85.7	35–38	14.3	85.7
35–36	14.3–23	85.7–67	38–41	23–27	77–73
36–45	23–100	67–0	41–46	27–50	73–50
35–50	100	0	46–52	50–100	50–0
50–55	100–3	0–97	52–57	100	0
55–60	3	97	57–62	100–3	0–97
			62–68	3	97

. The diode array detector was set at 320 nm for caftaric acid, gentisic acid, and caffeic acid; at 305 nm for coumaric acid; at 280 nm for gallic acid; and at 260 nm for protocatechuic acid. The fluorescence detector was set at λex = 270 nm and λem = 322 nm for hydroxytyrosol, tyrosol, catechin, epicatechin, proanthocyanidin B1, and proanthocyanidin B2, while for trans-resveratrol it was set at λex = 330 nm and λem = 374 nm. Chemically pure standards of phenolic acids (gallic, protocatechuic, caftaric, caffeic, coumaric), stilbenes (trans-resveratrol), flavonols (catechin, epicatechin), hydroxytyrosol and tyrosol, procyanidin dimer B1 and procyanidin dimer B2 were purchased from Sigma-Aldrich (St. Louis, MO, USA). The purity of all phenolic standards was greater than 95%. The individual stock solutions (1000 ppm) were prepared in pure methanol and kept at −20 °C in the dark. The calibration standards were freshly prepared on the day of analysis by diluting the appropriate working solution with the initial medium phase solution. The concentration range was selected based on the sensitivity of UHPLC-PDA for each polyphenol [17].

δ¹³C measurements were carried out using an elemental analyser (EA, VarioMicroCube, Elementar, F-69623 Villeurbanne, France) which was coupled to isotopic ratio monitoring by mass spectrometry (IRMS, Isoprime/Elementar, F-69623 Villeurbanne, France). Internationally certified reference materials USGS40 (with δ¹³C org = −26.39‰) and caffeine IAEA-600 (δ¹³C org = −27.77‰) were used for calibration. As carrier gas, Helium (Air Liquide, 5.6) was used, while carbon dioxide (Air Liquide, 4.8) was used as the reference gas. A total of 10 μL of wine was added to tin cups, and these were then introduced into the oxidation tube (950 °C) under helium flow (200 mL min⁻¹) and oxygen flow (30 mL min⁻¹), the temperature of the reduction furnace being fixed at 550 °C. Tin cups (5 × 9 mm) and copper wire were supplied by Elementar (France). The gases obtained by combustion were dried and eluted in a specific column that physically retains CO₂ (60 °C) and then releases it with a higher temperature (110 °C). An open split system allowed adjustment of gas withdrawal to the IRMS; current trap was fixed at 200 μA. The total duration of one measurement cycle was 600 s. The masses measured by IRMS were m/z 44 and 45 corresponding to CO₂ without and with ¹³C, respectively. The isotope ratio is expressed as a relative deviation, δ¹³C in per mil (‰), from the international standard, Vienna-Pee Dee Belemnite (VPDB), according to δ¹³C (‰) = 1000 × [(Rs/Rst) − 1], where R corresponds to the carbon 13 isotope ratio of the sample(s) and the standard (st) [18].

Statistical analyses were performed using the TIBCO STATISTICAL SOFTWARE and the statistical test applied was the Tukey HSD test. Tukey’s honest significance test is a single-step multiple comparison procedure for evaluation of statistical significance between multiple parameters of the evaluation of global variance for the same characterization (region or grape variety). In addition, for the discrimination by the variety according to the phenolic content and δ¹³C ratio, the discriminant function analysis test was applied using the same software [19].

3. Results and Discussions

The study carried out the analyses of phenolic compounds and stable isotopes for the 28 wines from three consecutive years of harvest, from three local grape varieties (Fetească neagra, Busuioacă de Bohotin and Fetească albă) and with two different vineyards: Cotnari and Dealu mare. The results of δ¹³C analyses are summarized in

Table 3. Carbon isotope ratios (δ¹³C).

Current Number	Region	Grape Variety	δ13C VPDB (‰)	SD
1	Cotnari	Fetească neagră	−27.30	0.16
2			−26.89	0.02
3			−27.48	0.16
4			−27.36	0.16
5			−27.04	0.02
6		Busuioacă de Bohotin	−25.66	0.06
7			−26.13	0.09
8			−26.21	0.27
9			−26.35	0.00
10			−26.44	0.20
11			−26.34	0.17
12			−26.48	0.06
13		Fetească albă	−26.12	0.39
14			−26.37	0.11
15			−27.27	0.11
16			−26.63	0.25
17	Dealu Mare	Fetească neagră	−27.83	0.07
18			−28.12	0.00
19			−28.18	0.22
20			−27.92	0.05
21		Busuioacă de Bohotin	−27.13	0.25
22			−27.14	0.09
23			−26.98	0.04
24			−27.16	0.08
25		Fetească albă	−26.74	0.25
26			−26.28	0.05
27			−26.87	0.14
28			−27.20	0.09

, while the phenolic compounds concentrations are presented in Table 4.

It can be observed from Table 4 that the values of the parameter δ¹³C vary between 28.18 _VPDB (‰)—the minimum value reported in the case of a wine from Dealu Mare obtained from the Fetească neagră variety—and −25.66 _VPDB (‰)—the maximum value reported in the case of a wine from Busuioacă de Bohotin obtained in the Cotnari vineyard. All other values of this parameter fall within the reported range. A first observation is that the values in this range are generally specific to wines produced in Romania; there are data in the literature to confirm this observation [1,2,20].

Twelve phenolic compounds were targeted and determined for this study. The analyses were performed in duplicate. Table 4 summarizes the results obtained from the analysis of the 12 phenolic compounds grouped as follows: phenolic acids (gallic, protocatechuic, caftaric, caffeic, coumaric), stilbenes (trans-resveratrol), flavonols (catechin, epicatechin), hydroxytyrosol and tyrosol, procyanidin dimer B1 and procyanidin dimer B2. The values thus obtained and expressed in mg/L are comparable to those reported in the literature for Romanian wines [21,22].

From a simple interpretation of the averages in Table 4, variations in the content of phenolic compounds between the tested wines can be observed, depending on: geographical origin, grape variety used or type of wine. Thus, the levels of gallic acid accumulated by the wines obtained from the Fetească neagră variety are considerably higher in the case of the Cotnari vineyard, while the wines obtained from the same variety in Dealu Mare have lower concentrations of the same compound.

3.1. Statistical Tests

3.1.1. ANOVA Evaluation

Considering every categorization group, a normality projection of data was performed in order to verify the distribution of variable results (Figure 1).

Normality of distribution permitted evaluation after one-way ANOVA of the initial data for the variables included in the study considering the grouping of variables region and variety. This revealed that every parameter of characterization was significant at p < 0.05 in the wines that were proposed for analysis and characterization. Evaluation was performed taking Wilks Lamda, which indicated percent variance in the dependent variables not explained by the differences in the levels of categorization. The values of significance for the variables were of 0.00054 for Wilks Lambda (F = 38.75) for the variety as first grouping variable and 0.1072 (F = 7.69) for region as second grouping variable. Since Wilks values are close to zero and F values are positive, the variables contribute to the evaluation model.

3.1.2. Tukey HSD Test

From the statistical analysis performed (Tukey HSD test) whose results are presented in Table 5, it is confirmed that gallic acid can be considered, in the case of the analysed wines, as discriminating by area, p factor being p < 0.05.

Regarding protocatechuic acid, caftaric acid, caffeic acid, coumaric acid, trans-resveratrol and hydroxytyrosol, the differences between the averages of these compounds reported in Table 4 (for Cotnari and Dealu Mare wines) are smaller. Thus, it is difficult to suggest that these compounds may constitute area markers. This hypothesis is also confirmed by the statistical test applied; in their case the values of the p coefficient always exceeded the value of 0.05 (as can be seen in Table 5).

In addition, according to the results of the statistical analyses highlighted in Table 5, the variables tyrosol, dimer B1, dimer B2, catechins and epicatechins have the p factor values p < 0.005, which again constitute statistically significant differences, translating into the possibility of discriminating the two areas analysed using these variables.

The same statistical test was applied in order to discriminate the analysed varieties according to the same variables: the concentrations of phenolic compounds (Table 6). Thus, it was observed that the recorded values of all the analysed phenolic compounds (except tyrosol) showed statistically significant differences between the varieties Fetească neagră and Fetească albă, while only some of them showed statistically significant differences between the other two combined pairs (Fetească albă/Busuioacă de Bohotin and Fetească neagră/Busuioacă de Bohotin). The values of protocatechuic acid also discriminated between the varieties Fetească neagră and Busuioacă de Bohotin, as well as those registered for caffeic acid. The concentrations of coumaric acid as well as hydroxytyrosol can act as discriminants for all three grape varieties. The variable contents of tyrozol recorded statistically significant differences between the varieties Fetească albă and Busuioacă de Bohotin, while the contents of dimers B1 and B2, as well as those of catechin and epicatechin, discriminated between the varieties Fetească neagră and Busuioacă de Bohotin.

A conclusion can be drawn from the obtained results: the isotopic analysis (δ¹³C) coupled with the statistical analysis makes possible to discriminate between grape varieties and areas. Thus, a clear differentiation could be observed between the wines obtained from the Busuioacă de Bohotin and Fetească neagră varieties and produced in the Cotnari vineyard, as shown in Table 7.

The same observation can be made in the case of wines obtained from the same varieties, but this time in the Dealu Mare vineyard. Within the Dealu Mare vineyard, the wines from the Fetească albă and Fetească neagră varieties could also be discriminated between with the help of this variable (δ¹³C). The values 0.007517 and 0.001653 of the p parameter in Table 7 show a clear discrimination between the two areas achieved using the δ¹³C isotopic ratio.

Table 4. Phenolic compound concentrations.

Sample		Gallic Acid	Protocatechuic Acid	Caftaric Acid	Caffeic Acid	Coumaric Acid	Trans-Resveratrol	Hydroxy-Tyrosol	Tyrosol	Procyanidin Dimer B1	Catechin	Procyanidin Dimer B2	Epicatechin
		mg/L	mg/L	mg/L	mg/L	mg/L	mg/L	mg/L	mg/L	mg/L	mg/L	mg/L	mg/L
Fetească neagră Cotnari (5)	Min	43.59	2.27	34.55	19.02	6.64	3.12	2.56	13.48	17.71	19.70	9.81	12.76
	Max	98.27	4.82	62.79	24.54	11.72	12.61	3.28	18.42	28.20	28.67	13.08	21.09
	Average	71.38	3.50	46.52	22.55	9.01	7.86	2.91	15.42	21.49	23.51	11.17	16.55
	SD	21.19	0.90	10.84	2.19	1.88	4.01	0.30	2.07	4.12	3.56	1.70	3.55
Busuioacă de Bohotin Cotnari (7)	Min	30.44	1.08	13.92	3.20	2.65	0.43	1.47	8.92	4.36	6.74	0.36	2.68
	Max	89.31	1.73	58.70	6.82	5.69	096	1.93	11.05	6.50	9.65	0.97	5.96
	Average	50.40	1.37	42.45	4.77	4.33	0.75	1.68	10.11	5.43	8.12	0.71	4.05
	SD	18.48	0.26	14.80	1.22	0.91	0.19	0.18	0.81	0.67	1.08	0.19	1.21
Fetească albă Cotnari (4)	Min	1.79	0.39	27.56	1.25	0.41	Nd	0.00	14.12	2.06	4.32	0.33	1.36
	Max	59.18	0.77	34.12	2.60	1.65	Nd	0.33	21.80	3.32	6.83	0.63	2.77
	Average	25.19	0.54	29.87	1.70	0.82	Nd	0.20	17.12	2.60	5.47	0.44	2.05
	SD	24.35	0.17	2.91	0.61	0.58	Nd	0.16	3.66	0.57	11.7	0.14	0.62
Fetească neagră Dealu Mare (4)	Min	33.68	2.44	43.26	12.15	5.51	3.77	2.29	24.63	36.77	33.75	20.26	27.80
	Max	70.99	3.66	88.62	17.30	8.85	12.07	3.31	32.15	40.69	40.81	24.46	37.32
	Average	49.56	3.16	63.99	14.44	7.36	6.78	2.72	28.43	38.92	37.06	22.71	32.39
	SD	17.27	0.57	1963	2.20	1.43	3.63	0.46	3.53	1.76	3.29	1.78	3.90
Busuioacă de Bohotin Dealu Mare (4)	Min	0.50	0.79	21.05	0.71	1.48	0.29	0.79	7.15	4.47	7.66	0.25	5.19
	Max	33.26	1.97	36.19	6.10	4.09	1.79	3.69	10.11	8.72	13.00	1.52	8.32
	Average	10.06	1.22	27.78	3.96	3.02	0.97	2.33	8.75	5.96	9.66	0.97	6.95
	SD	15.54	0.54	6.34	2.33	1.15	0.64	1.43	1.40	1.98	2.32	0.54	1.36
Fetească albă Dealu Mare (4)	Min	0.79	0.74	19.57	3.32	2.11	Nd	0.29	14.65	1.40	4.03	0.19	2.00
	Max	12.57	1.68	27.19	4.81	3.78	Nd	1.16	22.07	2.37	6.83	0.85	2.42
	Average	5.92	1.29	22.55	3.71	2.65	Nd	0.60	17.66	1.82	5.26	0.41	2.26
	SD	4.95	0.42	3.27	0.73	0.76	Nd	0.41	3.40	0.47	1.33	0.30	0.19

Table 4. Phenolic compound concentrations.

Sample		Gallic Acid	Protocatechuic Acid	Caftaric Acid	Caffeic Acid	Coumaric Acid	Trans-Resveratrol	Hydroxy-Tyrosol	Tyrosol	Procyanidin Dimer B1	Catechin	Procyanidin Dimer B2	Epicatechin
		mg/L	mg/L	mg/L	mg/L	mg/L	mg/L	mg/L	mg/L	mg/L	mg/L	mg/L	mg/L
Fetească neagră Cotnari (5)	Min	43.59	2.27	34.55	19.02	6.64	3.12	2.56	13.48	17.71	19.70	9.81	12.76
	Max	98.27	4.82	62.79	24.54	11.72	12.61	3.28	18.42	28.20	28.67	13.08	21.09
	Average	71.38	3.50	46.52	22.55	9.01	7.86	2.91	15.42	21.49	23.51	11.17	16.55
	SD	21.19	0.90	10.84	2.19	1.88	4.01	0.30	2.07	4.12	3.56	1.70	3.55
Busuioacă de Bohotin Cotnari (7)	Min	30.44	1.08	13.92	3.20	2.65	0.43	1.47	8.92	4.36	6.74	0.36	2.68
	Max	89.31	1.73	58.70	6.82	5.69	096	1.93	11.05	6.50	9.65	0.97	5.96
	Average	50.40	1.37	42.45	4.77	4.33	0.75	1.68	10.11	5.43	8.12	0.71	4.05
	SD	18.48	0.26	14.80	1.22	0.91	0.19	0.18	0.81	0.67	1.08	0.19	1.21
Fetească albă Cotnari (4)	Min	1.79	0.39	27.56	1.25	0.41	Nd	0.00	14.12	2.06	4.32	0.33	1.36
	Max	59.18	0.77	34.12	2.60	1.65	Nd	0.33	21.80	3.32	6.83	0.63	2.77
	Average	25.19	0.54	29.87	1.70	0.82	Nd	0.20	17.12	2.60	5.47	0.44	2.05
	SD	24.35	0.17	2.91	0.61	0.58	Nd	0.16	3.66	0.57	11.7	0.14	0.62
Fetească neagră Dealu Mare (4)	Min	33.68	2.44	43.26	12.15	5.51	3.77	2.29	24.63	36.77	33.75	20.26	27.80
	Max	70.99	3.66	88.62	17.30	8.85	12.07	3.31	32.15	40.69	40.81	24.46	37.32
	Average	49.56	3.16	63.99	14.44	7.36	6.78	2.72	28.43	38.92	37.06	22.71	32.39
	SD	17.27	0.57	1963	2.20	1.43	3.63	0.46	3.53	1.76	3.29	1.78	3.90
Busuioacă de Bohotin Dealu Mare (4)	Min	0.50	0.79	21.05	0.71	1.48	0.29	0.79	7.15	4.47	7.66	0.25	5.19
	Max	33.26	1.97	36.19	6.10	4.09	1.79	3.69	10.11	8.72	13.00	1.52	8.32
	Average	10.06	1.22	27.78	3.96	3.02	0.97	2.33	8.75	5.96	9.66	0.97	6.95
	SD	15.54	0.54	6.34	2.33	1.15	0.64	1.43	1.40	1.98	2.32	0.54	1.36
Fetească albă Dealu Mare (4)	Min	0.79	0.74	19.57	3.32	2.11	Nd	0.29	14.65	1.40	4.03	0.19	2.00
	Max	12.57	1.68	27.19	4.81	3.78	Nd	1.16	22.07	2.37	6.83	0.85	2.42
	Average	5.92	1.29	22.55	3.71	2.65	Nd	0.60	17.66	1.82	5.26	0.41	2.26
	SD	4.95	0.42	3.27	0.73	0.76	Nd	0.41	3.40	0.47	1.33	0.30	0.19

Table 5. Tukey HSD test results.

Tukey HSD Test

No.

Region

Variable: Gallic acid, MS = 405.39, df = 24.000

Variable: Protocatechuic acid, MS = 0.51870, df = 24.000

Variable: Caftaric acid, MS = 222.41, df = 24.000

Variable: Caffeic acid, MS = 11.421, df = 24.000

Variable: coumaric acid, MS = 2.3497, df = 24.000

Variable: trans-resveratrol, MS = 718.87, df = 24.000

{1} 50.652

{2} 21.849

{1} 1.8283

{2} 1.8919

{1} 40.575

{2} 38.104

{1} 9.5612

{2} 7.3711

{1} 4.9151

{2} 4.3453

{1} 42.503

{2} 22.892

1

Cotnari

0.001133

0.819332

0.668389

0.102783

0.340172

0.067559

2

Dealu Mare

0.001133

0.819332

0.668389

0.102783

0.340172

0.067559

Tukey HSD Test

No.

Region

Variable: Hydroxytyrosol, MS = 0.38400, df = 24.000

Variable: Tyrosol, MS = 25.605, df = 24.000

Variable: Procyanidin dimer B1, MS = 51.513, df = 24.000

Variable: Catechin, MS = 35.256, df = 24.000

Variable: Procyanidin dimer B2, MS = 20.830, df = 24.000

Variable: Epicatechin, MS = 33.531, df = 24.000

{1} 1.6937

{2} 1.8820

{1} 13.520

{2} 18.277

{1} 9.7428

{2} 15.566

{1} 12.268

{2} 17.327

{1} 3.9133

{2} 8.0264

{1} 7.4569

{2} 13.864

1

Cotnari

0.433942

0.021524

0.044233

0.035432

0.026882

0.008065

2

Dealu Mare

0.433942

0.021524

0.044233

0.035432

0.026882

0.008065

MS—mean sum of squares, df—degrees of freedom, red values—p-value is statistically representative for discrimination; the numbers in parentheses correspond to the regions (in the matrix).

Table 5. Tukey HSD test results.

Tukey HSD Test

No.

Region

Variable: Gallic acid, MS = 405.39, df = 24.000

Variable: Protocatechuic acid, MS = 0.51870, df = 24.000

Variable: Caftaric acid, MS = 222.41, df = 24.000

Variable: Caffeic acid, MS = 11.421, df = 24.000

Variable: coumaric acid, MS = 2.3497, df = 24.000

Variable: trans-resveratrol, MS = 718.87, df = 24.000

{1} 50.652

{2} 21.849

{1} 1.8283

{2} 1.8919

{1} 40.575

{2} 38.104

{1} 9.5612

{2} 7.3711

{1} 4.9151

{2} 4.3453

{1} 42.503

{2} 22.892

1

Cotnari

0.001133

0.819332

0.668389

0.102783

0.340172

0.067559

2

Dealu Mare

0.001133

0.819332

0.668389

0.102783

0.340172

0.067559

Tukey HSD Test

No.

Region

Variable: Hydroxytyrosol, MS = 0.38400, df = 24.000

Variable: Tyrosol, MS = 25.605, df = 24.000

Variable: Procyanidin dimer B1, MS = 51.513, df = 24.000

Variable: Catechin, MS = 35.256, df = 24.000

Variable: Procyanidin dimer B2, MS = 20.830, df = 24.000

Variable: Epicatechin, MS = 33.531, df = 24.000

{1} 1.6937

{2} 1.8820

{1} 13.520

{2} 18.277

{1} 9.7428

{2} 15.566

{1} 12.268

{2} 17.327

{1} 3.9133

{2} 8.0264

{1} 7.4569

{2} 13.864

1

Cotnari

0.433942

0.021524

0.044233

0.035432

0.026882

0.008065

2

Dealu Mare

0.433942

0.021524

0.044233

0.035432

0.026882

0.008065

MS—mean sum of squares, df—degrees of freedom, red values—p-value is statistically representative for discrimination; the numbers in parentheses correspond to the regions (in the matrix).

Table 6. Tukey HSD test results.

Tukey HSD Test
No.	Variety	Variable: Gallic acid, MS = 405.39, df = 24.000			Variable: Protocatechuic acid, MS = 0.51870, df = 24.000			Variable: Caftaric acid, MS = 222.41, df = 24.000			Variable: Caffeic acid, MS = 11.421, df = 24.000			Variable: coumaric acid, MS = 2.3497, df = 24.000			Variable: trans-resveratrol, MS = 718.87, df = 24.000
		{1} 55.562	{2} 35.758	{3} 17.666	{1} 3.1021	{2} 1.3759	{3} 0.8286	{1} 50.957	{2} 36.978	{3} 27.159	{1} 17.660	{2} 4.2259	{3} 2.6210	{1} 7.8205	{2} 3.8622	{3} 1.4424	{1} 86.976	{2} 7.7264	{3} 0.0000
1	Fetească neagră		0.082862	0.002408		0.000162	0.000131	0.101911	0.101911	0.009557		0.000129	0.000129		0.000137	0.000129		0.000130	0.000130
2	Busuioacă de Bohotin	0.082862		0.172695	0.000162		0.277098			0.376352	0.000129			0.000137		0.008973	0.000130		0.823642
3	Fetească albă	0.002408	0.172695		0.000131	0.277098		0.009557	0.376352		0.000129	0.592940		0.000129	0.008973		0.000130	0.823642
Tukey HSD Test
No.	Variety	Variable: Hydroxy-tyrosol, MS = 0.38499, df = 24.000			Variable: Tyrosol, MS = 25.605, df = 24.000			Variable: Procyanidin dimer B1, MS = 51.513, df = 24.000			Variable: Catechin, MS = 35.256, df = 24.000			Variable: Procyanidin dimer B2, MS = 20.830, df = 24.000			Variable: Epicatechin, MS = 33.531, df = 24.000
		{1} 2.8841	{2} 1.6321	{3} 0.4128	{1} 19.797	{2} 10.356	{3} 17.680	{1} 26.774	{2} 5.4234	{3} 2.1837	{1} 27.471	{2} 8.4902	{3} 5.1588	{1} 14.695	{2} 0.8605	{3} 0.3586	{1} 21.897	{2} 4.7014	{3} 2.1414
1	Fetească neagră		0.000419	0.000129		0.000856	0.677037		0.000130	0.000129		0.000129	0.000129		0.000130	0.000131		0.000130	0.000130
2	Busuioacă de Bohotin	0.000419		0.001349	0.000856		0.016787	0.000130		0.624898	0.000129		0.487631	0.000130		0.971999	0.000130		0.637743
3	Fetească albă	0.000129	0.001349		0.677037	0.016787		0.000129	0.624898		0.000129	0.487631		0.000131	0.971999		0.000130	0.637743

MS—mean sum of squares, df—degrees of freedom, red values—p-value is statistically representative for discrimination; the numbers in parentheses correspond to the varieties (in the matrix).

Table 6. Tukey HSD test results.

Tukey HSD Test
No.	Variety	Variable: Gallic acid, MS = 405.39, df = 24.000			Variable: Protocatechuic acid, MS = 0.51870, df = 24.000			Variable: Caftaric acid, MS = 222.41, df = 24.000			Variable: Caffeic acid, MS = 11.421, df = 24.000			Variable: coumaric acid, MS = 2.3497, df = 24.000			Variable: trans-resveratrol, MS = 718.87, df = 24.000
		{1} 55.562	{2} 35.758	{3} 17.666	{1} 3.1021	{2} 1.3759	{3} 0.8286	{1} 50.957	{2} 36.978	{3} 27.159	{1} 17.660	{2} 4.2259	{3} 2.6210	{1} 7.8205	{2} 3.8622	{3} 1.4424	{1} 86.976	{2} 7.7264	{3} 0.0000
1	Fetească neagră		0.082862	0.002408		0.000162	0.000131	0.101911	0.101911	0.009557		0.000129	0.000129		0.000137	0.000129		0.000130	0.000130
2	Busuioacă de Bohotin	0.082862		0.172695	0.000162		0.277098			0.376352	0.000129			0.000137		0.008973	0.000130		0.823642
3	Fetească albă	0.002408	0.172695		0.000131	0.277098		0.009557	0.376352		0.000129	0.592940		0.000129	0.008973		0.000130	0.823642
Tukey HSD Test
No.	Variety	Variable: Hydroxy-tyrosol, MS = 0.38499, df = 24.000			Variable: Tyrosol, MS = 25.605, df = 24.000			Variable: Procyanidin dimer B1, MS = 51.513, df = 24.000			Variable: Catechin, MS = 35.256, df = 24.000			Variable: Procyanidin dimer B2, MS = 20.830, df = 24.000			Variable: Epicatechin, MS = 33.531, df = 24.000
		{1} 2.8841	{2} 1.6321	{3} 0.4128	{1} 19.797	{2} 10.356	{3} 17.680	{1} 26.774	{2} 5.4234	{3} 2.1837	{1} 27.471	{2} 8.4902	{3} 5.1588	{1} 14.695	{2} 0.8605	{3} 0.3586	{1} 21.897	{2} 4.7014	{3} 2.1414
1	Fetească neagră		0.000419	0.000129		0.000856	0.677037		0.000130	0.000129		0.000129	0.000129		0.000130	0.000131		0.000130	0.000130
2	Busuioacă de Bohotin	0.000419		0.001349	0.000856		0.016787	0.000130		0.624898	0.000129		0.487631	0.000130		0.971999	0.000130		0.637743
3	Fetească albă	0.000129	0.001349		0.677037	0.016787		0.000129	0.624898		0.000129	0.487631		0.000131	0.971999		0.000130	0.637743

MS—mean sum of squares, df—degrees of freedom, red values—p-value is statistically representative for discrimination; the numbers in parentheses correspond to the varieties (in the matrix).

Table 7. Tukey HSD test results.

No.	Region	Variety	Tukey HSD Test; Variable δ13C; Approximate Probabilities for Post Hoc Tests; Error: Between MS = 0.9002, df = 22.00
			{1} −27.21	{2} −26.23	{3} −26.6	{4} −28.01	{5} −27.10	{6} −26.77
1	Cotnari	Fetească neagră		0.000292	0.057273	0.007517	0.993574	0.285852
2	Cotnari	Busuioacă de Bohotin	0.000292		0.394961	0.000144	0.001653	0.079276
3	Cotnari	Fetească alba	0.057273	0.394961		0.000153	0.207191	0.959551
4	Dealu Mare	Fetească neagră	0.007517	0.000144	0.000153		0.003671	0.000220
5	Dealu Mare	Busuioacă de Bohotin	0.993574	0.001653	0.207191	0.003671		0.636627
6	Dealu Mare	Fetească albă	0.285852	0.079276	0.959551	0.000220	0.636627

MS—mean sum of squares, df—degrees of freedom, red values—p-value is statistically representative for discrimination; the numbers in parentheses correspond to the mic between region and variety (in the matrix).

Table 7. Tukey HSD test results.

No.	Region	Variety	Tukey HSD Test; Variable δ13C; Approximate Probabilities for Post Hoc Tests; Error: Between MS = 0.9002, df = 22.00
			{1} −27.21	{2} −26.23	{3} −26.6	{4} −28.01	{5} −27.10	{6} −26.77
1	Cotnari	Fetească neagră		0.000292	0.057273	0.007517	0.993574	0.285852
2	Cotnari	Busuioacă de Bohotin	0.000292		0.394961	0.000144	0.001653	0.079276
3	Cotnari	Fetească alba	0.057273	0.394961		0.000153	0.207191	0.959551
4	Dealu Mare	Fetească neagră	0.007517	0.000144	0.000153		0.003671	0.000220
5	Dealu Mare	Busuioacă de Bohotin	0.993574	0.001653	0.207191	0.003671		0.636627
6	Dealu Mare	Fetească albă	0.285852	0.079276	0.959551	0.000220	0.636627

MS—mean sum of squares, df—degrees of freedom, red values—p-value is statistically representative for discrimination; the numbers in parentheses correspond to the mic between region and variety (in the matrix).

3.1.3. Statistical Tests—Discriminant Function Analysis (DFA)

The discriminant analysis made it possible to discriminate between varieties, the variables in this case being both the concentrations of the identified phenolic compounds and the results of the isotopic analysis, but not taking into account the origin of the wines because factors are considered as the average of the determined values in each region (Cotnari and Dealul Mare).

The independent contributions of each variable to the variety discrimination model were verified. It was found that all variables contribute differently to the discrimination between grape varieties. The hole model, based on the variables included, gave Wilks Lambda (WL) of 0.00054 for p < 0.05; thus, it was established that there are sufficient variables that influence the model and the discrimination between the functions can be performed. In this situationm only some variables have a greater influence for the determination of a group, namely, caffeic acid (WL-0.0016, p = 0.00052), coumaric acid (WL-0.000934, p = 0.027), trans-resveratrol (WL-0.00096, p = 0.022), tyrosol (WL-0.0014, p = 0.0013) and dimer B1 (WL-0.00097, p = 0.019).

To understand how these variables contribute to the identification of each group (respectively, the grape variety), in this sense, it was necessary to differentiate the functions on the basis of which the groups are individualized, with each function contributing to a greater or lesser extent to the discriminant analysis (

Table 8. Chi-square tests with successive roots removed.

Roots (Function)	Eigen Value	Canonic R	Wilks’ Lambda	Chi-Sqr	df	p-Value
1	151.4477	0.996715	0.000536	143.1067	26	0.000001
2	11.2449	0.958297	0.081667	47.5970	12	0.000004

).

Thus, two functions of discrimination were identified, called roots and which are statistically relevant because p < 0.05; therefore, there are two ways of interpreting how the variables (concentrations of phenolic compounds and δ¹³C ratios) contribute to the separation of groups (grape varieties).

As can be seen in

Table 9. Standardized coefficients for canonical variables.

Variable	Root 1	Root 2
Gallic acid	−0.7147	0.28852
Protocatechuic acid	0.3225	0.36014
Caftaric acid	1.0130	−0.25608
Caffeic acid	−2.6101	−0.15378
Coumaric acid	0.2592	−1.23542
Trans-resveratrol	−0.7378	1.36782
Hydroxytyrosol	0.8915	0.18484
Tyrosol	0.3968	2.06080
Procyanidin dimer B1	−0.3499	−6.83465
Catechin	0.2303	1.35097
Procyanidin dimer B2	−0.6267	5.42857
Epicatechin	−3.1831	−1.36891
δ13C mean	−0.3352	−0.32343
Eigenval	151.447	11.24487
Cum. Prop.	0.9309	1.00000

Red—variables that discriminates with statistical significance (p < 0.05) the first discriminant function (root 1). Green—variables that discriminates with statistical significance (p < 0.05) second discriminant function (root 2).

, the first canonical differentiation function (root 1) is more strongly influenced by variables such as caftaric acid, caffeic acid, and epicatechin; while root 2 is more strongly influenced by coumaric acid, trans-resveratrol, and procyanidin dimer B1, procyanidin dimer B2, hydroxytyrosol, tyrosol, catechin and epicatechin. A study of Montenegrin wines also allowed their discrimination with the help of the profile of polyphenolic compounds [23].

Based on the values of the root functions (as shown in Table 10) it could be shown that the first discrimination function (root 1—which is strongly influenced by the variables caffeic acid, caftaric acid and epicatechin) is the one that discriminates the Fetească neagră variety (and, implicitly, red wines) from the varieties Busuioacă de Bohotin and Fetească albă (meaning rosé and white wines). As can be seen in Figure 2, root 1 cannot make a strong discrimination between the varieties Busuioacă de Bohotin and Fetească albă.

On the other hand, the values of root 2 highlight a good differentiation between all three varieties. Thus, it can discriminate both red from white and rosé wines (respectively, the Fetească neagră variety of Busuioacă de Bohotin and Fetească albă varieties) but also between wines obtained from Busuioacă de Bohotin and Fetească albă varieties.

Table 10. Means of canonical variables.

Group	Root 1	Root 2
Fetească neagră	−16.8736	0.23597
Busuioacă de Bohotin	7.3757	−3.38780
Fetească albă	8.8411	4.39277

Table 10. Means of canonical variables.

Group	Root 1	Root 2
Fetească neagră	−16.8736	0.23597
Busuioacă de Bohotin	7.3757	−3.38780
Fetească albă	8.8411	4.39277

Figure 2. DFA plot of discriminating wine samples by grape variety according to the composition of the phenolic profile.

Figure 2. DFA plot of discriminating wine samples by grape variety according to the composition of the phenolic profile.

All the data used for the statistical analyses can be found in the Supplementary Materials.

In the literature were also reported situations when phenolic compounds were markers of the discrimination of varieties. An example is the study conducted on 52 wines of different origins (Romania, Bulgaria and Moldova) which demonstrated the possibility of using phenols, determined by liquid chromatography, to distinguish commercial table wines by variety [24].

The phenolic composition of wines is significantly influenced by genetic and environmental factors, but also an important contribution comes from technology. For this reason, the authentication of wines using phenolic imprinting can sometimes be quite a challenging task [25]. This is the reason why the identification and quantification of phenolic compounds cannot, by itself, answering authenticity issues. It can only be used as an additional verification method.

4. Conclusions

The Tukey Test allowed the discrimination of Romanian wines obtained from three different grape varieties in two separate regions. The discrimination by area and by grape variety was performed according to phenolic composition and δ¹³C.

The phenolic compounds that could be considered grape variety markers were: hydroxytyrosol and coumaric acid.

The phenolic compounds that could be considered area markers were: gallic acid, tyrosol, B1 and B2 dimers, catechin and epicatechin.

The results of δ¹³C coupled with statistical analysis made it possible to discriminate between different grape varieties and areas.

The characteristics of these wines were determined by their raw matter, geographical origin as well as technological processes applied specific to each region.