Oxidative Degradation of Anthocyanins in Red Wine: Kinetic Characterization Under Accelerated Aging Conditions

Abstract

The oxidative degradation of anthocyanins in red wine was investigated under controlled conditions using hydroxyl radicals generated in the presence of Cu (II) as a catalyst. A full factorial experimental design with 23 replicates was used to evaluate the effects of hydrogen peroxide concentration, catalyst dosage, and reaction temperature on anthocyanin degradation over a fixed time. Statistical analysis (ANOVA and multiple regression) showed that all three variables and the main interactions significantly affected anthocyanin loss, with temperature identified as the most influential factor. The combined effects were described by a first-order polynomial model. The activation energies for degradation ranged from 56.62 kJ/mol (cyanidin-3-O-glucoside) to 40.58 kJ/mol (peonidin-3-O-glucoside acetate). Increasing the temperature from 30 °C to 40 °C accelerated the degradation kinetics, almost doubled the rate constants and shortened the half-life of the pigments. At 40 °C, the half-lives ranged from 62.3 min to 154.0 min, depending on the anthocyanin structure. These results contribute to a deeper understanding of the stability of anthocyanins in red wine under oxidative stress and provide insights into the chemical behavior of derived pigments. The results are of practical importance for both oenology and viticulture and support efforts to improve the color stability of wine and extend the shelf life of grape-based products.

1. Introduction

Anthocyanins (Greek: anthos (άνθος) = flower and kianeos (κυανός) = blue) belong to the group of flavonoids among the polyphenols, which are responsible for the red and blue color of grape berries and wine [1]. The dominant anthocyanins are derivatives of malvidin—malvidin-3-O-glucoside (Mvgl) and malvidin-3,5-O-diglucoside. Malvidin-3,5-O-diglucoside is characteristic of some hybrid grape varieties and young wines from Eurasian cultivated vines (Vitis vinifera L.). In red wine, the most important are glycosides of five anthocyanidins: delphinidin-3-O-glucoside (Dpgl), cyanidin-3-O-glucoside (Cygl), petunidin-3-O-glucoside (Ptgl), peonidin-3-O-glucoside (Pngl), and malvidin-3-O-glucoside, while pelargonidin-3-O-glucoside is absent or present only in trace amounts [2,3]. Anthocyanin compounds are responsible for certain organoleptic characteristics such as color, bitterness, and astringency of wine. Since the color of wine is a common quality criterion, the analysis of these pigments has attracted considerable attention [4,5].

Anthocyanins are water-soluble pigments from the phenol group that occur as secondary metabolic products in plants. They contribute to the coloring of flowers and fruits, help with pollination and seed dispersal, and play a role in plant development. Due to their broad color spectrum—from red to violet to blue—they are often used as natural dyes in various industries. Anthocyanins also have useful biological properties, including antioxidant and antimicrobial activity. However, their practical use is limited by their low stability, which is affected by factors such as pH, light, temperature, oxygen, enzymes, ascorbic acid, sulphites, and co-pigments [6].

Due to their unsaturated chemical structure, anthocyanins are very susceptible to oxidation through reactions with molecular oxygen. Oxygen is therefore a crucial factor influencing their stability, as its presence can accelerate the degradation of anthocyanins. In red wine, oxidative processes occur throughout production—during fermentation, aging in the barrel, and storage in the bottle—which influence both the chemical composition and the sensory properties of the final product. These reactions have a significant influence on the color of the wine [7,8] and its organoleptic properties [9].

Throughout the winemaking process, the wine is exposed to different amounts of oxygen, each of which influences its chemical evolution. Due to the complex interplay between the composition of the wine and its oxidative reactivity, it is still difficult to predict the exact effects of a given oxygen dose. Nevertheless, the results presented in this study provide practical insights that can help predict the behavior of anthocyanins and overall wine stability under defined oxidative conditions.

Oxygen is a key factor in wine production, especially in the post-fermentation stages. The highest oxygen exposure typically occurs during bottling, where the dissolved oxygen content can reach about 3.0–4.1 mg/L under controlled saturation conditions [10,11]. During the subsequent oxidative reactions, molecular oxygen (O₂) is gradually reduced to water via a series of steps, producing reactive intermediates such as hydrogen peroxide (H₂O₂) and hydroxyl radicals (HO•). These reactive oxygen species can initiate the oxidation of various wine components and thus influence the composition and stability of the wine [12,13,14].

Transition metals such as iron and copper are frequently present in wine and originate from various sources such as vineyard soils, pesticide treatments against Plasmopara viticola disease, winemaking equipment, and environmental influences. These metals can catalyse oxidative reactions by activating molecular oxygen and thus influence the stability and quality of wine [15]. Globally, iron concentrations in wine range from 0.061 to 50 mg/L, while copper concentrations range from undetectable to around 6.8 mg/L. Iron is primarily taken up by the grapevine from the soil and transferred to the wine during processing, while copper is often introduced using copper-based fungicides in viticulture [15].

The simultaneous presence of hydrogen peroxide and transition metals such as iron and copper in wine can lead to the formation of Fenton’s reagent, a highly reactive oxidizing system. While limited data are available on the combined effects of H₂O₂, Cu (II), and temperature on anthocyanin stability in red wine, this study fills this gap by investigating their effects on anthocyanin degradation in Serbian ‘Merlot’ cv. wine. The degradation kinetics are analyzed using a thermodynamic framework that incorporates the following main activation parameters: Gibbs free energy (ΔG*), enthalpy (ΔH*), entropy (ΔS*), and activation energy (Ea). These parameters enable a deeper understanding of the thermal stability and degradation behavior of anthocyanins in complex food matrices such as wine. Although the oxidative degradation of anthocyanins has previously been studied in systems involving Fe (II) and other transition metals [16,17], there is a lack of data specifically addressing Cu (II)-catalyzed degradation under realistic wine conditions, which represents the novelty of our approach.

Therefore, the present study aims to characterize the anthocyanin composition of red wine from the ‘Merlot’ variety and to determine the oxidative degradation of individual anthocyanins through a controlled process of accelerated aging using hydroxyl radicals generated in the presence of Cu (II) as a catalyst. The research provides a better understanding of the stability of the anthocyanin compounds in ‘Merlot’ wine and the associated changes in oenological potential during the aging process under oxidative conditions.

2. Materials and Methods

2.1. Wine Sample Origin and Storage Conditions

The red wine used was a commercial Merlot (Vitis vinifera L.), produced using standard vinification procedures including alcoholic and malolactic fermentation. Sulfur dioxide (SO₂) was added at 50 mg/L. The wine was aged in stainless steel tanks for 12 months and bottled in dark glass bottles with screw caps under inert conditions.

Before the experiment, the wine was stored for 3 months in the dark at 15 ± 1 °C to minimize further oxidation. Although dissolved oxygen was not quantified, all degradation experiments were conducted in sealed Budarin-type reaction vessels to ensure that oxidative conditions were controlled. The wine matrix was characterized by pH 3.4, ethanol content 13.5%, and typical phenolic composition.

2.2. Chemicals and Reagents

All reagents were of analytical grade. Copper (II) chloride dihydrate (CuCl₂•2H₂O, ≥99%), hydrogen peroxide (30% w/w, p.a.), acetonitrile (≥99.9%, HPLC grade), formic acid (≥98%) and anthocyanin glucoside standards (purity ≥ 95%) were purchased from Merck (Darmstadt, Germany) and Sigma-Aldrich (St. Louis, MO, USA). The solutions were prepared using deionized water from a MicroMed high-purity water system (TKA Wasseraufbereitungssysteme GmbH, Niederelbert, Germany).

A Cu (II) stock solution (1.0 × 10⁻³ mol/L) was prepared by dissolving CuCl₂•2H₂O in deionized water. The hydrogen peroxide solution (0.979 mol/L) was freshly prepared from a commercial 30% stock solution immediately before use to ensure stability.

2.3. High-Performance Liquid Chromatography (HPLC-DAD) Analysis

Anthocyanin compounds were quantified by reversed-phase high-performance liquid chromatography (RP-HPLC) with diode array detection. The analysis was performed using an Agilent 1200 Series HPLC system (Agilent Technologies, Singapore) equipped with a UV–Vis diode array detector (DAD) for simultaneous monitoring of multiple wavelengths.

Sample injections of 5 µL were made onto an Eclipse XDB C18 column (4.6 × 150 mm, Agilent) maintained at 25 °C. Gradient elution was performed with two mobile phases: Solvent A (2% formic acid in water) and Solvent B (a mixture of 80% acetonitrile, 2% formic acid, and water). The elution profile was programmed as follows: 0–10 min, 0% B; 10–28 min, linear gradient from 0% to 25% B; 28–30 min, 25% B; 30–35 min, 25% to 50% B; 35–40 min, 50% to 80% B; and 40–45 min, re-equilibration back to 0% B.

Anthocyanins were identified by comparing both retention times (Dpgl- 22.04; Cygl- 22.60; Pngl- 25.32; Mvgl- 26.85 min.) and UV–Vis spectra with those of authentic standards. The quantification was based on an external calibration. The analytical parameters, such as calibration curves, coefficients of determination (R²), limits of detection (LOD), and limits of quantification (LOQ), are listed in

Table 1. Analytical parameters for anthocyanins used for HPLC-DAD analysis.

Compound	Calibration Curve	(R2)	Range (mg/L)	LOD 1 (mg/L)	LOQ 2 (mg/L)
Dpgl	$y=3.602x-0.10$	0.9996	0.50–20.00	0.03	0.10
Cygl	$y=5.738x+0.22$	0.9994	0.50–20.00	0.02	0.07
Pngl	$y=4.093x+0.43$	0.9998	0.50–20.00	0.03	0.10
Mvgl	$y=5.038x+0.36$	0.9998	2.00–50.00	0.12	0.39

¹ LOD—limit of detection; ² LOQ—limit of determination; Dpgl—delphinidin-3-O-glucoside; Cygl—cyanidin-3-O-glucoside; Pngl—peonidin-3-O-glucoside; Mvgl—malvidin-3-O-glucoside.

. The results are expressed in mg/L of red wine.

The concentrations of the individual anthocyanins were determined using calibration curves generated from authentic standards. For compounds for which no specific standard exists, quantification was performed using calibration data from structurally related analogues. For example, peonidin-3-O-glucoside was used as a reference for petunidin-3-O-glucoside. Similarly, delphinidin-3-O-glucoside, cyanidin-3-O-glucoside, peonidin-3-O-glucoside, and malvidin-3-O-glucoside were used to estimate the concentrations of their corresponding acetylated and p-coumaroylated derivatives. All measurements were performed in triplicate to ensure the reproducibility and reliability of results.

2.4. Reaction Conditions

A full factorial design (2³) with replication was used to investigate the influence of three variables on the degradation of anthocyanins in red wine. Two concentration levels of hydrogen peroxide (10 and 20 mmol/L), two levels of Cu (II) (0.01 and 0.02 mmol/L), and two reaction temperatures (30 °C and 40 °C) were investigated. Based on preliminary investigations, the factor levels were selected to ensure that the reaction time would allow for chromatographic measurements at five different time points. The three factors—H₂O₂ concentration (x₁), Cu (II) concentration (x₂) and reaction temperature (x₃)—and their respective maximum and minimum values are summarized in

Table 2. Experimental values and coded levels of the independent variables used for the Experimental design with the observed response values for Dpgl, Cygl, and Ptgl.

No	Design Matrix			Dpgl		Cygl		Ptgl
	x1 (mmolL−1)	x2 (mmolL−1)	x3 (°C)	PARex 1	PARpre 2	PARex	PARpre	PARex	PARpre
1	10 (−1)	0.01 (−1)	30 (−1)	16.94 ± 0.30 a	16.87	26.81 ± 0.49	26.55	18.53 ± 0.18	18.42
2	20 (+1)	0.01 (−1)	30 (−1)	18.99 ± 0.22	19.06	31.15 ± 0.44	31.41	21.12 ± 0.17	21.23
3	10 (−1)	0.02 (+1)	30 (−1)	19.66 ± 0.45	19.73	32.97 ± 0.36	33.23	21.86 ± 0.24	21.96
4	20 (+1)	0.02 (+1)	30 (−1)	21.99 ± 0.52	21.92	38.35 ± 0.23	38.09	24.88 ± 0.32	24.77
5	10 (−1)	0.01 (−1)	40 (+1)	23.38 ± 0.48	23.26	43.96 ± 0.28	43.46	26.40 ± 0.26	26.25
6	20 (+1)	0.01 (−1)	40 (+1)	26.14 ± 0.36	26.26	51.04 ± 0.60	51.55	30.16 ± 0.35	30.31
7	10 (−1)	0.02 (+1)	40 (+1)	27.14 ± 0.33	27.26	53.64 ± 0.55	54.15	31.13 ± 0.26	31.28
8	20 (+1)	0.02 (+1)	40 (+1)	30.37 ± 0.62	30.25	62.75 ± 0.58	62.25	35.50 ± 0.42	35.35

x₁—H₂O₂ concentration; x₂—Cu (II) concentration; x₃—temperature; Dpgl—delphinidin-3-O-glucoside; Cygl—cyanidin-3-O-glucoside; Ptgl—petunidin-3-O-glucoside; ^a Mean values (n = 3) ± standard deviation; ¹ Experimental results; ² Predicted results.

,

Table 3. Experimental values and coded levels of the independent variables used for the Experimental design with the observed response values for Pngl, Mvgl, and Pngl-ac.

No	Design Matrix			Pngl		Mvgl		Pngl-ac
	x1 (mmolL−1)	x2 (mmolL−1)	x3 (°C)	PARex 1	PARpre 2	PARex	PARpre	PARex	PARpre
1	10 (−1)	0.01 (−1)	30 (−1)	16.47 ± 0.44 a	16.4	22.42 ± 0.59	21.15	16.16 ± 0.32	16.18
2	20 (+1)	0.01 (−1)	30 (−1)	18.12 ± 0.34	18.19	25.88 ± 0.39	26.1	17.45 ± 0.18	17.44
3	10 (−1)	0.02 (+1)	30 (−1)	18.97 ± 0.22	19.04	27.73 ± 0.53	27.8	18.68 ± 0.12	18.67
4	20 (+1)	0.02 (+1)	30 (−1)	20.89 ± 0.36	20.82	31.89 ± 0.29	31.7	19.93 ± 0.23	19.94
5	10 (−1)	0.02 (−1)	40 (+1)	21.10 ± 0.52	21.03	34.20 ± 0.35	33.9	20.36 ± 0.29	20.31
6	20 (+1)	0.01 (−1)	40 (+1)	23.21 ± 0.48	23.28	39.61 ± 0.43	39.9	21.98 ± 0.16	22.03
7	10 (−1)	0.02 (+1)	40 (+1)	24.33 ± 0.36	24.4	42.22 ± 0.50	42.5	23.54 ± 0.27	23.59
8	20 (+1)	0.02 (+1)	40 (+1)	26.72 ± 0.50	26.65	48.83 ± 0.38	48.5	25.35 ± 0.32	25.3

Pngl—peonidin-3-O-glucoside; Mvgl—malvidin-3-O-glucoside; Pngl—ac-peonidin-3-O-(6′-acetyl)-glucoside; ^a Mean values (n = 3) ± standard deviation; ¹ Experimental results; ² Predicted results.

and

Table 4. Experimental values and coded levels of the independent variables used for the Experimental design with the observed response values for Mvgl-ac, Pngl-p-coum, and Mvgl-p-coum.

No	Design Matrix			Mvgl-ac		Pngl-p-coum		Mvgl-p-coum
	x1 (mmolL−1)	x2 (mmolL−1)	x3 (°C)	PARex 1	PARpre 2	PARex	PARpre	PARex	PARpre
1	10 (−1)	0.01 (−1)	30 (−1)	24.41 ± 0.50 a	24.07	19.41 ± 0.28	19.05	22.58 ± 0.22	22.33
2	20 (+1)	0.01 (−1)	30 (−1)	28.40 ± 0.18	28.72	22.32 ± 0.32	22.67	26.31 ± 0.34	26.58
3	10 (−1)	0.02 (+1)	30 (−1)	30.48 ± 0.24	30.82	23.29 ± 0.20	23.87	28.90 ± 0.45	29.15
4	20 (+1)	0.02 (+1)	30 (−1)	35.82 ± 0.48	35.47	28.08 ± 0.26	28.26	33.67 ± 0.48	33.4
5	10 (−1)	0.01 (−1)	40 (+1)	36.80 ± 0.44	36.25	29.21 ± 0.42	28.8	35.22 ± 0.36	34.8
6	20 (+1)	0.01 (−1)	40 (+1)	43.23 ± 0.58	43.75	33.55 ± 0.35	33.18	41.04 ± 0.33	41.43
7	10 (−1)	0.02 (+1)	40 (+1)	45.37 ± 0.60	45.96	34.98 ± 0.30	35.16	45.08 ± 0.47	45.5
8	20 (+1)	0.02 (+1)	40 (+1)	54.00 ± 0.55	53.45	40.50 ± 0.52	40.32	52.52 ± 0.50	52.13

Mvgl-ac—malvidin-3-O-(6′-acetyl)-glucoside; Pngl-p-coum—Peonidin-3-O-(6′-p-coumaroyl)-glucosid); Mvgl-p-coum—Malvidin-3-O-(6′-p-coumaroyl)-glucoside; ^a Mean values (n = 3) ± standard deviation; ¹ Experimental results; ² Predicted results.

, where the coded values (−1) and (+1) represent the lower and upper levels, respectively.

The degradation of each anthocyanin was monitored over an oxidation period of 70 min oxidation period (adjusting the system from the start of the reaction to account for the retention time of each peak), and the response variable was defined as the percentage of anthocyanin reduction (PAR). The experimental matrix with the corresponding PAR values for each treatment combination is shown in Table 2, Table 3 and Table 4.

2.5. Anthocyanin Kinetic Degradation in the Presence of Hydroxyl Radicals

The kinetic degradation of red wine anthocyanins in the presence of hydroxyl radicals—generated by a Fenton-like system involving Cu (II) and H₂O₂—was studied at four different temperatures: 25, 30, 35, and 40 °C. The experiment was carried out in a Budarin vessel with four chambers. The Budarin vessel enabled separate thermostating of each component and precise timing of the reaction start. Cu (II) solution was added to one chamber, hydrogen peroxide to the second, red wine to the third, and a mixture of buffer solution (tartaric acid, pH 3.5) and deionized water (total volume: 2 mL) to the fourth. Prior to analysis, the wine sample was filtered through a 0.45 μm membrane filter. The vessel was thermostatted at the selected temperature (±0.1 °C) for 5 min. All reagents were freshly prepared and handled under subdued light to prevent photo-induced degradation. After equilibration, the contents of all four chambers were combined and stirred for 60 s. The resulting solution was immediately transferred to the chromatography cell, and the change in peak area was measured at 520 nm as a function of reaction time. All measurements were performed in triplicate to ensure analytical reliability and reproducibility.

2.6. Determination of Kinetic Parameters

In accordance with previous studies, anthocyanins follow first-order degradation kinetics (at constant H₂O₂ and Cu (II) concentration):

$$c=c_{o}exp\left(-kt\right)$$

(1)

where t is the time (min), k is the pseudo first order kinetic rate constant (min⁻¹), and $c_o$ and $c$ are the anthocyanin content (mg/L) at time zero and t, respectively.

The decimal reduction time (D-value), which is the time needed for a tenfold reduction in the initial concentration at a given temperature, is related to k-values according to Equation (2):

$$D={\displaystyle \frac{ln\left(10\right)}{k}}$$

(2)

The half-life (t_1/2) value of degradation is given by Equation (3):

$$t_{1/2}={\displaystyle \frac{ln\left(2\right)}{k}}$$

(3)

2.7. Thermodynamic Analysis

The temperature and degradation constant are related according to the Arrhenius equation:

$$k=A·e^{-E_a/RT}$$

(4)

where k is the rate constant of the degradation reaction, A denotes the pre-exponential (Arrhenius) factor, Ea is the apparent activation energy, R represents the universal gas constant, and T is the absolute temperature. By applying the natural logarithm to both sides of the Arrhenius equation,

$$lnk=lnA-{\displaystyle \frac{E_a}{RT}}$$

(5)

Equation (4) is obtained. When the natural logarithm of the degradation constant compared with the inverse of the absolute temperature is plotted according to Equation (4), the Ea value from the slope and the lnA value from the ordinate intercept are obtained. Thus, the thermodynamic parameters change in enthalpy (ΔH*), entropy (ΔS*), and free energy of activation (ΔG*) are obtained using the following equations:

$$∆H^{*}=E_a-RT$$

(6)

$$∆G^{*}=-RTln\left({\displaystyle \frac{k·h}{k_B·T}}\right)$$

(7)

where h (6.6262 1034 J s) is Planck’s constant and k_B (1.3806 1023 J K1) is Boltzmann’s constant. From Equations (6) and (7), it is possible to calculate the activation entropy ($∆S^{*}$):

$$∆S^{*}={\displaystyle \frac{∆H^{*}-∆G^{*}}{T}}$$

(8)

2.8. Statistical Analysis

The experimental data were statistically analyzed using Design-Expert^® software (Stat-Ease, Inc., Minneapolis, MN, USA). A full factorial experimental design (2³) with replication was used to evaluate the influence of three factors—hydrogen peroxide concentration, Cu (II) concentration, and reaction temperature—on the percentage of anthocyanin reduction (PAR). The significance of each factor and its interactions was evaluated using analysis of variance (ANOVA). In addition, first-order linear regression models were developed to describe the relationship between PAR and the experimental factors, including their interaction terms.

3. Results and Discussions

3.1. Wine Anthocyanin Characterization

The concentrations of the individual anthocyanins—Dpgl, Cygl, Ptgl, Pngl, Mvgl—and their respective acylated derivatives (esters of acetic and p-coumaric acid) in ‘Merlot’ cv. wine are listed in

Table 5. Anthocyanin composition in Serbian red wine ‘Merlot’, (mg/L).

Anthocyanin	mg/L	Anthocyanin	mg/L	Anthocyanin	mg/L
Dpgl	6.022 ± 0.102	Dpgl-ac	0.824 ± 0.023	Dpgl-p-coum	-
Cygl	2.356 ± 0.045	Cygl-ac	-	Cygl-p-coum	-
Ptgl	9.821 ± 0.110	Ptgl-ac	1.036 ± 0.018	Ptgl-p-coum	0.587 ± 0.017
Pngl	8.380 ± 0.092	Pngl-ac	1.380 ± 0.026	Pngl-p-coum	2.732 ± 0.032
Mvgl	50.252 ± 0.128	Mvgl-ac	8.681 ± 0.104	Mvgl-p-coum	7.721 ± 0.081
Σ	76.831		11.921		11.040

Dpgl-ac—Delphinidin-3-glucoside-acetate; Cygl-ac—Cyanidin/3/glucoside-acetate; Dpgl-ac—delphinidin-3-O-(6′-acetyl)-glucoside; Cygl-ac—cyanidin-3-O-(6′-acetyl)-glucoside; Ptgl-ac—petunidin-3-O-(6′-acetyl)-glucoside; Dpgl-p-coum—delphinidin-3-O-(6′-p-coumaroyl)-glucoside); Cygp-p-coum—cyaniding-3-O-(6′-p-coumaroyl)-glucoside); Ptgl-p-coum—petunidin-3-O-(6′-p-coumaroyl)-glucoside).

. The total anthocyanin content, calculated as the sum of the individual compounds, was 99.792 mg/L. Among the identified pigments, Mvgl was the most abundant, accounting for 50.25% of the total anthocyanin content, which is consistent with its predominance in most grape varieties and red wines. Acetylated derivatives accounted for between 0.82% and 8.70% of total anthocyanins, while p-coumaroylated anthocyanins ranged from 0.59% and 7.72%.

3.2. Effects of Kinetic Reactions

The influence of hydrogen peroxide concentration, Cu (II) concentration and reaction temperature on the percentage of anthocyanin reduction (PAR) in red wine was evaluated after a 70 min reaction time. The factors were selected based on preliminary screening results and practical limitations. For temperature, the upper limit was set at 40 °C, as higher temperatures led to too rapid degradation and HPLC analysis was not feasible. A 2³-factorial experimental design was created to determine the statistical significance of the selected variables. The experimental matrix comprised eight unique treatment combinations, each performed in duplicate to ensure reproducibility.

The results obtained show a significant change in the concentration of anthocyanin 3-O-glucosides during the first 70 min of the reaction. Ivanova-Petropulos et al. (2015) [16] also reported that the concentration of total and free anthocyanins—compounds responsible for the red-purple color of young red wines—decreases significantly during maturation and aging of 1–2 years, by up to 88%. Although a decrease in the concentration of free anthocyanins was observed during the degradation process induced by hydroxyl radicals, it is important to emphasize that remarkable differences in the degradation dynamics of these compounds were observed. In particular, significant differences in the stability of the individual free anthocyanins were observed depending on the conditions of oxidative degradation (Table 2, Table 3 and Table 4).

Among individual anthocyanins, Mvgl showed the highest initial concentration (50.252 mg/L), decreasing by nearly 49% under the most oxidative conditions. Its predominance and well-documented role in polymer pigment formation reinforce its central contribution to wine aging and color stability, and Mvgl exhibits high structural stability and a strong tendency to form polymeric pigments that preserve color and resist oxidative degradation during aging [17,18].

In contrast, the lowest concentrations were found for Cygl, which decreased from 2.356 mg/L before the degradation process to 0.877 mg/L after 70 min of reaction at 40 °C under H₂O₂ and Cu (II) concentrations of 20 mmol/L and 0.02 mmol/L, respectively. Cygl also showed the strongest reduction compared to its initial value, with a decrease in concentration of 62.75% after 70 min (Table 2, Table 3 and Table 4). Pngl, on the other hand, showed the highest stability with the lowest decrease in concentration—only 26.72%—under the same reaction conditions.

As for the acylated derivatives, Pngl acetate was present in low concentrations (Table 5). Interestingly, this compound also showed the highest stability to chemical changes during degradation by hydroxyl radicals. The enhanced stability of peonidin-based anthocyanins may be attributed to their methoxy substitution on the B-ring, which confers greater resistance to oxidative cleavage compared to hydroxylated analogs such as cyanidin [19]. In particular, during the first 70 min of the reaction, its concentration decreased by only 25.35%, from 1.380 mg/L to 0.350 mg/L (Table 2). In contrast, Mvgl acetate showed a much stronger reduction, with a decrease in concentration of almost 54% after 70 min of reaction time.

A similar pattern was observed for p-coumaroylated forms, with malvidin-based derivatives showing greater degradation than peonidin-based ones. The degradation of p-coumaroylated pigments during bottle aging is also supported by the study of Blanco-Vega et al. (2014) [20], where a reduction of nearly 50% was reported after several years of storage. The same authors reported that anthocyanin content reduction depends on grape varieties and the aging period.

3.2.1. Linear Regression Modeling

Table 2, Table 3 and Table 4 show the coded and actual values of the experimental factors as well as the corresponding reaction values, i.e., the percentage reduction in the individual anthocyanins: Dpgl, Cygl, Ptgl, Pngl, Mvgl, Pngl-ac, Mvgl-ac, Pngl-p-coum, and Mvgl-p-coum. A first-order linear regression model was applied to analyze the experimental data for percentage anthocyanin reduction (PAR). The general form of the regression equation used in this analysis is as follows:

$$y=b_o+b_1{x}_1+b_2{x}_2+b_3{x}_3+b_{12}{x}_1{x}_2+b_{13}{x}_1{x}_3+b_{23}{x}_2{x}_3+b_{123}{x}_1{x}_2{x}_3$$

(9)

In the regression model, the variables x₁, x₂, and x₃ represent the H₂O₂ concentration, Cu (II) concentration, and reaction temperature, respectively. Interaction terms such as x₁ x₂, x₁ x₃, and x₂ x₃ describe the combined effects of two independent variables, while x₁ x₂ x₃ considers the interaction between all three factors. The coefficient b₀ denotes the intercept, b_i represents the linear regression coefficients, and b_ij and b_ijk correspond to the coefficients for the two-factor and three-factor interactions, respectively.

The influence of these independent variables (x₁, x₂, x₃) on the percentage s degradation of the individual analyzed anthocyanins was assessed.

3.2.2. Analysis of Variance (ANOVA)

ANOVA results for the percentage reduction in anthocyanins are summarized in

Table 6. ANOVA of the first order polynomial model for the response variables (actual values).

	SOV	x1	x2	x3	x1x2	x1x3	x2x3	x1x2x3	Error	Total
Dpgl	SS 1	13.442	23.495	108.41	0.0703	0.324	0.6441	0.0045	0.4298	146.82
	Df 2	1	1	1	1	1	1	1	8	14
	MS 3	13.442	23.495	108.41	0.0703	0.324	0.6441	0.0045	0.05373	146.82
	F	250.17	437.28	2017.7	1.3084	6.03	11.987	0.0837
	p	<0.00001 4	<0.00001	<0.00001	0.28575 1	0.03958	0.00854	0.77970 1
Cygl	SS	83.916	150.94	842.75	1.209	5.2326	8.0601	0.1128	7.2174	1099.4
	Df	1	1	1	1	1	1	1	8	15
	MS	83.916	150.94	842.75	1.209	5.2326	8.0601	0.1128	0.9022	73.296
	F	93.012	167.9	934.11	1.34	5.7998	8.9338	0.125
	p	<0.00001	<0.00001	<0.00001	0.28042 1	0.04262	0.01736	0.21465 1
Ptgl	SS	23.598	36.808	169.28	0.1352	0.7938	1.1101	0.004	1.0944	232.8223
	Df	1	1	1	1	1	1	1	8	15
	MS	23.598	36.808	169.28	0.1352	0.7938	1.1101	0.004	0.1368	15.5215
	F	172.5	269.06	1237.4	0.9883	5.8026	8.1147	0.0292
	p	<0.00001	<0.00001	<0.00001	0.34927 1	0.04258	0.02152	0.86856 1
Pngl	SS	9.4612	18.03	54.653	0.0378	0.1081	0.2701	0.00001	0.1464	82.7071
	Df	1	1	1	1	1	1	1	8	15
	MS	9.4612	18.03	54.653	0.0378	0.1081	0.2701	0.00001	0.0183	5.5138
	F	517	985.24	2986.5	2.0655	5.9071	14.759	0.0006
	p	<0.00001	<0.00001	<0.00001	0.18860 1	0.04117	0.00493	0.98105 1
Mvgl	SS	49.005	100.82	406.125	0.5	2.205	4.5	0.02	3.0947	566.269
	Df	1	1	1	1	1	1	1	8	15
	MS	49.005	100.82	406.125	0.5	2.205	4.5	0.02	0.3868	37.7513
	F	126.69	260.65	1049.9	1.2926	6.4762	11.633	0.0517
	p	<0.00001	<0.00001	<0.00001	0.288475 1	0.03445	0.09213	0.825834 1
Pngl-ac	SS	4.4551	16.6753	45.172	0.0028	0.099	0.3003	0.0066	0.1368	66.848
	Df	1	1	1	1	1	1	1	8	15
	MS	4.4551	16.6753	45.172	0.0028	0.099	0.3003	0.0066	0.0171	4.4565
	F	260.53	975.16	2641.6	0.1637	5.7894	13.561	0.3859
	p	<0.00001	<0.00001	<0.00001	0.69637	0.04276	0.03035	0.55174
Mvgl-ac	SS	73.811	135.3	454.51	1.5312	4.0612	4.3512	0.1012	5.4149	679.08
	DF	1	1	1	1	1	1	1	8	15
	MD	73.811	135.3	454.51	1.5312	4.0612	4.3512	0.1012	0.6768	45.272
	F	109.05	199.91	671.55	2.2624	6.0006	6.4291	0.1495
	p	<0.00001	<0.00001	<0.00001	0.17096	0.03996	0.03495	0.70909
Pngl-p-coum	SS	38.544	62.496	254.7	0.5832	1.1704	1.1858	0.0612	1.6424	360.38
	Df	1	1	1	1	1	1	1	8	15
	MS	38.544	62.496	254.7	0.5832	1.1704	1.1858	0.0612	0.2053	24.025
	F	187.74	304.41	1240.6	2.8407	5.7009	5.7759	0.2981
	p	<0.00001	<0.00001	<0.00001	0.13039	0.04401	0.04295	0.59996
Mvgl-p-coum	SS	59.187	153.3	486.72	0.8844	2.8322	7.5272	0.042	3.776	714.26
	Df	1	1	1	1	1	1	1	8	15
	MS	59.187	153.3	486.72	0.8844	0.8322	7.5272	0.042	0.472	47.617
	F	125.39	324.78	1031.1	1.8737	6.0004	15.9474	0.0889
	p	<0.00001	<0.00001	<0.00001	0.20825	0.03996	0.00398	0.77317

H₂O₂ concentration (x₁), Cu (II) concentration (x₂) and reaction temperature (x₃). SOV-Source of variation; ¹ Sum of squares; ² Degree of freedom; ³ Mean of square; ⁴ <0.00001 highly significant.

. The statistical relevance of the main factors and their two- and three-way interactions was assessed using the corresponding F and p values. A higher F-value indicates a more significant contribution to the model, with p-values below 0.05 indicating statistical significance.

To streamline the regression model, any factor or interaction term found to be statistically non-significant (p ≥ 0.05) was excluded. The resulting simplified regression equations are shown in

Table 7. Regression equations.

		R2	Radj2	CV(%)
Dpgl	$y=23.076+1.296x_1+1.714x_2+3.681x_3+0.201x_1{x}_3+0.284x_2{x}_3$	0.9970	0.9945	1.00
Cygl	$y=42.588+3.238x_1+4.344x_2+10.264x_3+0.808x_1{x}_3+1.004x_2{x}_3$	0.9934	0.9877	2.23
Ptgl	$y=26.197+1.717x_1+2.145x_2+4.600x_3+0.315x_1{x}_3+0.372x_2{x}_3$	0.9953	0.9911	1.41
Pngl	$y=21.226+1.009x_1+1.55001+2.614x_3+0.116x_1{x}_3+0.184x_2{x}_3$	0.9982	0.9966	0.64
Mvgl	$y=34.075+2.475x_1+3.550x_2+7.125x_3+0.525x_1{x}_3+0.750x_2{x}_3$	0.9945	0.9897	1.82
Pngl-ac	$y=20.431+0.746x_1+1.444x_2+2.376x_3+0.111x_1{x}_3+0.194x_2{x}_3$	0.9979	0.9962	0.64
Mvgl-ac	$y=37.312+3.037x_1+4.112x_2+7.537x_3+0.712x_1{x}_3+0.737x_2{x}_3$	0.9920	0.9850	2.20
Pngl-p-coum	$y=28.917+2.195x_1+2.795x_2+5.642x_3+0.382x_1{x}_3+0.385x_2{x}_3$	0.9954	0.9914	1.56
Mvgl-p-coum	$y=35.665+2.720x_1+4.377x_2+7.800x_3+0.595x_1{x}_3+0.970x_2{x}_3$	0.9947	0.9901	1.92

. The predicted values generated by the model are compared with the experimentally observed data in Table 2, Table 3 and Table 4 and show a strong agreement between the predicted and actual responses. This agreement indicates that the polynomial regression models are effective for predicting anthocyanin degradation behaviour under the oxidative conditions tested.

As noted by Aklilu et al. (2021) [21], positive coefficients in the regression model indicate synergistic effects that increase the percentage of anthocyanin reduction, while negative coefficients reflect antagonistic interactions.

The analysis showed that all linear terms had strong statistical relevance, with F-values ranging from 93.0 to 2986.5 for factors x₁, x₂, and x₃, confirming their significant influence on anthocyanin reduction (PAR).

The Pareto charts shown in Figure 1 illustrate the relative influence of each factor on PAR. The length of each horizontal bar reflects the absolute value of the estimated effect, while the dashed line marks the threshold for statistical significance at a 95% confidence level.

Among the factors analyzed, reaction temperature (x₃) had the greatest influence on anthocyanin degradation, followed by Cu (II) concentration (x₂). In the two-factor interactions, the combinations of Cu (II) concentration and temperature (x₂ x₃) and H₂O₂ concentration and temperature (x₁ x₃) were found to significantly influence all reaction variables. In contrast, the three-way interaction (x₁ x₂ x₃) and the interaction between H₂O₂ and Cu (II) concentrations (x₁ x₂) did not show statistical significance (p > 0.05).

To further illustrate the interaction effects between the key variables, a composite response-surface visualization is provided in Figure 2. These 3D plots confirm that temperature and Cu (II) concentration jointly exert the most pronounced effect on anthocyanin degradation, whereas H₂O₂ concentration plays a secondary role within the tested range. The response surfaces are derived from the full factorial design and reflect trends observed for Mvgl, selected as the most abundant and representative anthocyanin in the studied wine matrix.

3.2.3. Model Fitting and Adequacy

To ensure that the developed regression models (Table 7) accurately reflect the behavior of the real system, the model fit was evaluated using regression analysis and ANOVA. The coefficient of determination (R²) was calculated together with the adjusted R² (R²adj) and the coefficient of variation (CV) to assess the appropriateness of the model (Table 6).

The R² value represents the proportion of variability in the response that can be explained by the model and not by random error. In general, a well-fitted model should have an R² of at least 80%. However, as R² can increase with the inclusion of additional predictors regardless of their significance, R²adj is a more reliable indicator of model quality. R²adj takes into account the number of predictors and should ideally be above 90% for a robust model.

The coefficient of variation (CV) reflects the relative dispersion of the data; values below 10% usually indicate acceptable precision and consistency of the experimental results [22,23].

In this study, R² values ranged from 99.20% to 99.82%, demonstrating excellent model fit. The adjusted R² values were above 98% for all responses, and the CV values remained below 10%, confirming the reliability, precision, and overall suitability of the fitted models.

3.3. Reaction Kinetics

The degradation kinetics of the individual anthocyanins were analyzed at four reaction temperatures—25, 30, 35, and 40 °C—over a period of up to 200 min (Figure 3). The kinetic parameters, including the rate constant (k), decimal reduction time (D-value), half-life (t_1/2), and activation energy (Ea), are summarized in Table 2, Table 3 and Table 4.

The concentration decay profiles of all analyzed anthocyanins over time were well described by a first-order kinetic model (Equation (3)). The model provided an excellent fit for all temperature conditions, with coefficients of determination (R²) above 0.96 for each compound, confirming a strong linear relationship between the logarithmic decrease in anthocyanin concentration and reaction time (Figure 3).

A similar degradation behavior was observed for Cygl and other anthocyanins in different fruit juice systems subjected to oxidative or thermal treatments. Although complete degradation is rarely reported, studies on juices such as blackberry, strawberry, sour cherry, and pomegranate generally describe first-order kinetics for the loss of individual or all anthocyanins. This degradation pattern is often influenced by the juice matrix, storage conditions, and processing parameters. These results support the application of first-order kinetic modeling in the description of anthocyanin stability during oxidation and thermal stress in complex food systems [24,25,26,27,28].

Reaction temperature had a pronounced effect on anthocyanin degradation, as illustrated in Figure 1. At 25 °C, the degradation rates were relatively low (k = 0.00197–0.00369 min⁻¹), while at 40 °C, a significant acceleration was observed (k = 0.00470–0.01113 min⁻¹), as shown in Table 2. Each anthocyanin exhibited a different degradation rate and sensitivity to temperature changes. Among the compounds analyzed, (Cygl) and (Mvgl-p-coum) were the most thermally sensitive with activation energies (Ea) of 56.62 kJ/mol and 55.33 kJ/mol, respectively. Interestingly, the activation energy obtained for Mvgl in this study is comparable to that previously reported in our work on Vranac red wine under similar oxidative conditions, where an Ea value of 57.70 kJ/mol was observed [29]. This consistency reinforces the thermal sensitivity of this anthocyanin across different grape varieties.

The activation energy reflects the minimum energy required for a reaction to take place and is usually derived using the Arrhenius equation. In this context, lower Ea values indicate reduced temperature sensitivity, which means that compounds with lower activation energies are more thermally stable during the degradation process [30].

The half-life values (t_1/2) of anthocyanins decreased with rising temperature, indicating thermal sensitivity across all compounds (Table 2). Peonidin-based forms generally exhibited higher stability compared to cyanidin and malvidin derivatives.

These results are consistent with previous studies on ‘Ruby’ and ‘Merlot’ grape juice, where elevated temperatures significantly accelerated the degradation of Mvgl, Dpgl, Ptgl, Pngl, and Cygl [31]. The observed differences in stability between the anthocyanins can be attributed to structural differences that influence their resistance to oxidation.

In addition,

Table 8. Thermodynamic parameters of degradion processes.

Anthocyanins	T	k	D	t1/2	Ea	ΔH*	ΔS*	ΔG*
	°C	min−1	min	min	kJmol−1	kJmol−1	Jmol−1K−1	kJmol−1
Dpgl	25	0.00227	1014.10	305.28	43.73	41.28	−156.71	87.98
	30	0.00294	782.99	235.71		41.21	−157.42	88.91
	35	0.00373	617.16	185.79		41.17	−157.91	89.81
	40	0.00536	429.48	129.29		41.13	−157.30	90.37
Cygl	25	0.00369	623.85	187.81	56.62	54.18	−109.40	86.78
	30	0.00523	440.15	132.51		54.11	−110.08	87.46
	35	0.00730	315.34	94.93		54.06	−110.48	88.09
	40	0.01113	206.83	62.26		54.02	−110.04	88.46
Ptgl	25	0.00235	979.57	294.89	46.08	43.63	−148.54	87.90
	30	0.00320	719.37	216.56		43.56	−148.96	88.70
	35	0.00425	541.65	163.06		43.52	−149.21	90.18
	40	0.00576	399.65	120.31		43.48	−149.20	92.55
Pngl	25	0.00197	1168.53	351.77	43.61	41.16	−158.30	88.17
	30	0.00279	825.09	248.39		41.09	−158.26	89.04
	35	0.00350	657.71	198.00		41.05	−158.85	89.97
	40	0.00470	489.79	147.45		41.01	−158.79	90.71
Mvgl	25	0.00296	777.70	234.12	54.75	52.27	−117.62	87.32
	30	0.00423	544.21	163.83		52.23	−118.02	87.99
	35	0.00583	394.85	118.87		52.19	−118.25	88.61
	40	0.00863	266.74	80.30		52.15	−118.14	89.12
Pngl-ac	25	0.00201	1145.27	344.78	40.58	38.13	−168.29	88.28
	30	0.00272	840.15	252.92		38.06	−168.40	89.09
	35	0.00340	677.06	203.82		38.02	−168.92	90.05
	40	0.00450	511.55	154.00		37.98	−168.83	90.82
Mvgl-ac	25	0.00314	733.12	220.70	53.76	51.29	−120.43	87.18
	30	0.00455	505.93	152.31		51.25	−120.667	87.81
	35	0.00609	377.99	113.79		51.21	−120.26	88.43
	40	0.00908	253.52	76.32		51.16	−120.86	88.99
Pngl-p-coum	25	0.00265	868.68	261.51	46.21	43.76	−147.12	87.60
	30	0.00388	593.30	178.61		43.69	−146.94	88.21
	35	0.00531	433.52	130.51		43.64	−146.95	88.91
	40	0.00648	355.25	101.32		43.6	−147.37	89.73
Mvgl-p-coum	25	0.00319	721.63	217.24	55.33	52.88	−114.97	87.14
	30	0.0045	511.55	154.00		52.81	−115.61	87.84
	35	0.00647	355.79	107.11		52.76	−115.69	88.4
	40	0.00925	248.86	74.92		52.72	−115.73	88.95

shows the decimal reduction time (D value) for each condition. This parameter, which indicates the time required to reduce the anthocyanin concentration by one logarithmic cycle, decreased with increasing temperature. At 25 °C, D values ranged from 623.85 to 1168.53 min (Cygl to Pngl), while at 40 °C, they dropped to 206.34–511.55 min (Cygl to Pngl-ac), indicating faster degradation under thermal stress.

The evaluation of the thermodynamic parameters provides further insight into the kinetics of the thermal degradation of anthocyanins in the presence of a Fenton-like reagent. Table 4 summarizes the calculated values for Gibbs free energy of activation (ΔG*), enthalpy of activation (ΔH*), and entropy of activation (ΔS*) for all temperatures tested.

The values of ΔG*, which reflect the energy difference between the activated complex and the reactants [32], ranged from 86.78 to 92.55 kJ/mol. The positive ΔG* values indicate that the degradation of anthocyanins under these conditions is a non-spontaneous process.

The ΔH*, which represents the energy barrier associated with bond breaking and formation during the transition to the activated state (Vikram et al., 2005) [33], was relatively consistent under all conditions, ranging from 38.13 to 54.18 kJ/mol. The positive sign confirms that the degradation process is endothermic and requires energy input.

The activation entropy (ΔS*), which indicates the degree of molecular disorder in the transition state, was negative in all cases. These negative values indicate that the transition state is more ordered than the initial reactants, implying restricted molecular motion during the degradation process.

These thermodynamic trends are consistent with findings from other studies on anthocyanin degradation in non-wine matrices, such as fruit juices and plant extracts [34,35,36,37]. This similarity supports the broader relevance of the observed degradation behavior across different food systems.

4. Conclusions

The degradation of anthocyanins in Serbian ‘Merlot’ red wine under oxidative conditions followed first-order kinetics, with rate constants ranging from 0.00197 to 0.01113 min⁻¹, depending on temperature and anthocyanin structure. The activation energy (Ea) values varied between 40.58 and 56.62 kJ/mol, indicating moderate thermal sensitivity. Reaction temperature had the strongest influence on degradation rates, with Cygl being the most susceptible to oxidation. In contrast, Pngl and its derivatives demonstrated higher thermal and oxidative stability. Among the acylated anthocyanins, Mvgl-ac showed lower stability compared to Pngl-ac. A similar pattern was observed for their p-coumaroylated forms, with malvidin-based derivatives degrading more rapidly.

These findings highlight the role of anthocyanin structure in determining oxidative stability. From a practical standpoint, controlling oxygen exposure and metal ion content—particularly copper—during both vineyard management and winemaking is essential for preserving anthocyanin integrity and ensuring color stability in red wines. Reducing excessive copper residues—e.g., by managing pesticide inputs—and optimizing micro-oxygenation during fermentation and aging could help maintain anthocyanin composition and support the production of high-quality wines. These findings provide guidance for winemakers aiming to enhance shelf life, color stability, and overall market competitiveness.