1. Pinot noir wines

Two Vitis vinifera L. cv. Pinot noir wines (100 %) from the 2020 vintage, differing in their phenolic content (i.e., PN1 and PN2), were obtained from the Delegat Estate Winery, Blenheim, New Zealand. The Pinot noir grapes came from the same vineyard, from own-rooted Pinot noir vines Entav INRA 667 under the Smart-Dyson trellis system and were crushed and destemmed, while PN2 received the free-run juice that was drained from the skins to produce the commercial wines. Commercial Saccharomyces cerevisiae yeast (Lalvin RC212^TM, Lallemand Oenology, Blagnac, France) was inoculated for primary fermentation, in which both wines took 16 days to complete with temperature control at 26 °C. Both wines had also spontaneously completed malolactic fermentation (uncontrolled due to COVID-19 lockdown) before skins pressing. Basic parameters for PN1 and PN2 upon MOX treatments are presented in Table 1. Besides the variations in phenolic content, a slightly higher alcohol content and fructose level were also found in the PN2 wine (Table 1).

2. MOX Trial

The set-up for MOX and control treatments were the same as outlined by Yang et al. (2021a). The control treatment received no oxygenation and was sealed in the 2 L Fisherbrand^TM glass bottles. MOX treatments were carried out in 15 L stainless steel tanks and applied through a sealed-end diffuser tube, rolled up into a circular shape (r = 8.1 cm), with the fluorinated ethylene-propylene copolymer (FEP) membrane coating (TSFE14-0250-047-50, CHEMFLUOR^®, SaintGobain Performance Plastics, Courbevoie, France) to introduce the microlevels of oxygen (food grade, BOC Gas, Auckland, New Zealand) into the wines. The diffusion of oxygen (food grade, BOC Gas, New Zealand) in each MOX tank was controlled using 2-stage primary oxygen regulators (BOC-8000) and high precision line regulators (BOC BS1-000-1700-N). The dissolved oxygen (DO) levels were measured non-intrusively using the PSt3 oxygen sensor (Vinventions, CA, USA).

Each control and MOX treatment were carried out in triplicate. Two different oxygen doses: 0.50 ± 0.08 and 2.11 ± 0.3 mg/L/day of O₂ were applied to both PN1 and PN2 for 14 days with temperature control at 15 ± 1 ^°C. Treatments are labelled as follows: MOX applied at 0.50 and 2.11 mg/L/day on Pinot noir wines with the low phenolic content (T1 and T2) and the control (C1); and with the high phenolic content (T3 and T4) and the control (C2). The two oxygenation rates were determined in a preliminary study using a 12 % ethanol solution, where oxygenation was applied at a pressure of 200 kPa and 500 kPa, respectively, for 120 hours. Without an external oxygen supply, the oxygen ingress into the MOX tanks was 1.17 ± 0.46 mg/L/month. Prior to oxygenation, the wines were sterile filtered at 0.20 μm pore size (Sartopore 2 Midicap, Göttingen, Germany). After MOX, the wines were aged for 1 month at 15 ^°C and followed by 100 mg/L addition of potassium metabisulfite (Enartis Winy, Italy). The endpoint was determined 4 days after SO₂ addition.

3. Wine sampling

The sampling method was as described by Yang et al. (2021a). Colour parameters and polymeric pigment content were measured at the beginning of the experiment (time 0), day 5, day 14 (end of MOX), day 44 (after MOX and ageing for 1 month, i.e., before SO₂ addition) and day 48 (the endpoint). For phenolic composition, the samples were taken and analysed at time 0, day 44 and day 48. We had to skip the phenolic analyses on day 14 due to the unprecedented COVID-19 lockdown in New Zealand.

4. Conventional analyses

Wine pH and TA were measured using an automatic wine titrator (Hanna Instruments^®, Woonsocket, RI, USA). A photometric measurement was used to determine free and total SO₂, residual sugars (i.e., glucose and fructose), malic acid, acetaldehyde concentrations using enzymatic test kits from Megazyme (Bray, Ireland). For these analyses, a Thermo Spectronic Helios Gamma 9423 UV-vis spectrophotometer (Waltham, MA, USA) was used and absorbances were measured using visible cuvettes with a 2 mm path length (Mediray, Auckland, New Zealand). The alcohol content was measured using an Anton-Paar Alcolyzer (Graz, Austria).

5. Spectrophotometric analyses

Wine colour parameters including the absorbance of 420 nm and 520 nm, SO₂ resistant pigments, colour intensity, monomeric anthocyanins were determined using the methods described by Iland et al. (2013). Total anthocyanins (i.e., malvidin-3-glucoside equivalents mg/L), small and large polymeric pigments (i.e., SPP and LPP in absorbance units) were determined using the Harbertson-Adams assay (Harbertson et al., 2003), with the reaction volumes adapted from Heredia et al. (2006).

6. High-performance liquid chromatography analysis of monomeric phenolics

Monomeric phenolic compounds were determined using the high-performance liquid chromatography (HPLC) method as described by Yang et al. (2021a). In brief, approximately 1.5 mL of wine was filtered through a 0.45 µm membrane filters (MicroScience, Auckland, New Zealand), of which 20 µL was injected into a reversed-phased C18 column (Kinetex C18, 2.6 µm, 100 Å, 100 mm × 4.6 mm) on an Agilent Technologies (Santa Clara, CA, USA) 1200 series HPLC system. A G1315D diode array detector was used and chromatograms were obtained at 280 nm (for flavan-3-ols), 320 nm (for hydroxycinnamic acids), 360 nm (for flavonols), and 520 nm (for free anthocyanins).

For the quantifications, external standards were used: gallic acid, (−)-epicatechin, (+)-catechin, vanillic acid, syringic acid, caffeic acid, p-coumaric acid, ferulic acid, t-resveratrol, rutin and malvidin-3-glucoside. The measurements of quercetin, quercetin-3-O-glucuronide, quercetin-3-glucoside are expressed as rutin equivalents (mg/L). The concentrations of trans- and cis-coutaric acids are expressed as p-coumaric acid equivalents (mg/L). The content of trans-caftaric acid was calculated as caffeic acid equivalents (mg/L). The limit of detection (LOD) and limit of quantitation (LOQ) were calculated based on a signal-to-noise ratio of 3:1 and 10:1, respectively.

7. Tannin concentration and extraction method

Tannin concentration was measured using the revised version of methylcellulose tannin precipitation assay (MCP tannins, epicatechin equivalents (ECE) g/L) as detailed by Mercurio et al. (2007). The extraction of tannin for the phloroglucinolysis reaction was carried out as described by Yang et al. (2021b) based on the methods of Jeffery et al. (2008) and of Kassara and Kennedy (2011). After purification and drying, the tannin extracts were reconstituted to 10 g/L with pure methanol (HPLC grade), according to the MCP tannin concentration of wines. The methanolic extracts were then immediately analysed or kept in a –80 °C freezer until analysis. For each treatment, two extractions were undertaken for each vessel (i.e., 6 replications for each treatment).

8. High-performance liquid chromatography analyses of tannins

Tannin composition was analysed via the acid-catalysed depolymerisation reaction of tannins in the presence of excess phloroglucinol (Kennedy and Jones, 2001). For detection, the same HPLC system and the C18 column as for the analysis of monomeric phenolics were used, and the chromatograms were monitored at 280, 360, 520 and 620 nm, with 280 nm as the main detecting wavelength. Methods for the phloroglucinolysis reaction and the HPLC runtimes and mobile phases were as described in Yang et al. (2021b). The injection volume was 20 μL and the column was held at 30 °C. The solvents used were: solvent A, 0.1 % formic acid in Milli-Q water (v/v); and solvent B, 0.1 % formic acid in acetonitrile (v/v). The elution conditions were as follows: 0 min (95 % solvent A, 5 % solvent B); 3.08 min (95 % solvent A, 5 % solvent B); 9.23 min (80 % solvent A, 20 % solvent B); 16.92 min (60 % solvent A, 40 % solvent B); 18.64 min (10 % solvent A, 90 % solvent B); 21.54 min (10 % solvent A, 90 % solvent B); and 23 min (95 % solvent A, 5 % solvent B) with an additional 7 min post-run. The flow rate was 1.3 mL/min.

An external standard calibration curve was established using (−)-epicatechin, based on the peak area of the pure compound at 280 nm. The proportions of tannin subunits (%) were calculated based on their concentrations expressed as (−)-epicatechin equivalents (in moles). The mean degree of polymerisation (mDP) and the tannin mass conversion (yield %) were calculated according to Kennedy and Jones (2001). The mDP was calculated using the sum of all tannin subunits divided by the sum of all terminal units. The tannin yield (%) was determined by using the total mass of all tannin subunits (in grams), which excluded the phloroglucinol portion of the adducts, divided by the concentration of tannins (i.e., 10 g/L) used for the analysis. The degree of tannin galloylation (i.e., seed tannin indicator) was the sum of (−)-epicatechin-3-O-gallate subunits (i.e., both the extension and terminal units), and the degree of trihydroxylation (i.e., skin tannin indicator) was the (−)-epigallocatechin extension units (%).

9. Microbial analysis

The effectiveness of the sterile filtration was evaluated during the experiment, where the samples were plated for microbial growth at time 0 (after filtration) and following the MOX treatments at days 5, 10, 44 and 48. Prior to plating, a 10 μL aliquot of the undiluted and stained sample with methylene blue was dispensed into a haemocytometer slide and examined under a microscope. For the control treatments, microbial analyses were determined at time 0, day 44 and day 48. The details of the method were as described in Yang et al. (2021a). In brief, samples were plated on Wallerstein Laboratory Nutrient (WLN) medium for yeast, and 50 % de Man, Rogosa, and Sharpe (MRS) medium (with the addition of 25 mg/L natamycin) for bacterial quantification. This was followed by a 7-day incubation at 28 ^°C, before determination.

10. Chemicals and reagents

All chemical reagents were of the highest available analytical grade quality and purchased from Merck (Castle Hill, NSW, Australia) unless specifically noted. Biological reagents and agar mediums for microbial analyses are purchased from Merck (Castle Hill, NSW, Australia). Total and frees sulfite assay kit (K-SULPH), D-fructose/D-glucose assay kit (K-FRUGL), L-malic acid assay kit (K-LMAL-116A), and acetaldehyde assay kit (K-ACHYD) were purchased from Megazyme (Bray, Ireland). Milli-Q water was obtained from a Thermo Scientific Barnstead Nanopure water purification system. Bovine serum albumin was purchased from pHScientific (Auckland, New Zealand).

11. Statistical Analysis

Statistical analyses were performed with R Studio version 1.2.5033 (R version 4.0.3, Inc., Boston, MA, USA) and graphs were produced using Excel 2019 (Microsoft, Redmond, WA, USA). Tukey’s honestly significant difference (HSD) test was used as a comparison test when the samples were significantly different after the analysis of variance (ANOVA), where raw p values < 0.05 were considered significant. For each analysis, quantitative data of each treatment are calculated as means of six values (i.e., 3 experimental replicates × 2 analytical replicates).

1. Dissolved oxygen, acetaldehyde concentration and basic wine composition

The dissolved oxygen (DO) concentrations detected in the control and MOX treatments during the experiment are shown in Figure 1. On the last day of MOX (day 14), T2 and T4 had approximately 4 times higher DO concentrations than T1 and T3, respectively, in accordance with the higher oxygenation rate. During ageing, from day 15 to 32, DO was not monitored in T1 and T2 due to the COVID19-lockdown. After 1 month of ageing and before SO₂ addition (day 44), the remaining DO concentrations in T1 (0.7 ± 0.5 mg/L), T2 (2.3 ± 0.4 mg/L), T3 (1.1 ± 0.3 mg/L) and T4 (1.4 ± 0.7 mg/L) indicated a lower DO consumption rate than DO accumulations, which might also be associated with the oxygen ingress through the MOX tanks. At the endpoint, i.e., 4 days after SO₂ addition, the DO concentrations in wines with MOX remained similar to that observed at day 44, showing the lack of reactivity of SO₂ directly with oxygen (Danilewicz, 2003).

Figure 1. The concentration of DO monitored in different treatments during this experiment. Note the missing data in C1, T1 and T2 was due to the COVID-19 lockdown.

Save View full size Expand inline Collapse inline

In previous MOX studies, a correlation between acetaldehyde production and the growth of Saccharomyces cerevisiae has been found (Sáenz-Navajas et al., 2018; Ji et al., 2020). It is known that Saccharomyces cerevisiae uses ethanol as a carbon source, when fermentable sugar becomes limited, which produces acetaldehyde (Galdieri et al., 2010; Pozo-Bayón and Moreno-Arribas, 2011). During the present study, no cells or viable growth of Saccharomyces cerevisiae were detected in any of the treatments, confirming the effectiveness of the sterile filtration. Free SO₂ was not detected upon the application of MOX (Table 1). Therefore, any impact on the wine can thus be ascribed to purely chemical oxidation phenomena.

The conventional analyses carried out at the endpoint are presented in Table 2. In both PN1 and PN2, the results did not show any clear impact of the MOX treatments. During MOX, a small but apparent increase of acetaldehyde (close to 2 folds) was seen with the higher oxygen dosage at day 5 (Table 2), from c. 4.7 to 8.4 mg/L in T2 and 6.3 to 10.4 mg/L in T4. This can be associated with the chemical oxidation of ethanol, due to H₂O₂ released from phenolic oxidation and catalysed by the redox cycling of Fe (II) and Fe (III) (Wildenradt and Singleton, 1974; Ribéreau-Gayon et al., 1983; Danilewicz, 2016).

Regarding the lower oxygen dosage, T3 had a similar acetaldehyde concentration as to C2 at the same time points, while a clear increase of T1 over C1 was observed by day 14. Cacho et al. (1995) found that acetaldehyde production from oxidation was affected by iron levels, in which the higher the iron concentration, the lower the increase in acetaldehyde content. This might be another influential factor between PN1 and PN2. Regardless, the total increase of acetaldehyde in these wines was very limited compared to previous MOX studies where the growth of Saccharomyces cerevisiae was also observed (Sáenz-Navajas et al., 2018; Ji et al., 2020; Yang et al., 2021a). In both Sáenz-Navajas et al. (2018), with oxygenation applied at 15 mg/L/month for 48 days, and Ji et al. (2020), with DO consumption at approximately 0.5 mg/L/day within 49 days, MOX treatments induced an increase of yeast growth to 10⁵ CFU/mL towards the end of MOX, which coincided with the increase of acetaldehyde (c. 30–35 mg/L) and the depletion of free SO₂ and DO. Yang et al. (2021a), applied MOX both before and after MLF (10.8 and 52.4 mg/L/month for 30 days) and increased acetaldehyde (from around 55 to 120 mg/L, and 35 to 70 mg/L, respectively) at the beginning of treatments, where free SO₂ was not readily detected and DO was still climbing. The authors indicated the influence of dormant yeast and yeast lees on acetaldehyde production (Yang et al., 2021a). From these previous studies, it also appeared that with MOX, the timing of acetaldehyde production could also be affected by the presence of free SO₂ that is not only known to bind acetaldehyde (Peterson and Waterhouse, 2016) but also can suppress yeast activity (Divol et al., 2012). In the present trial, the low acetaldehyde increase supports the idea that the larger increases in acetaldehyde observed during past MOX trials can be associated with residual yeast activity.

2. Colour parameters and polymeric pigment content

During the experiment, an important increase in the 420 nm (Figure 2a) and 520 nm (Figure 2b) absorbances and colour intensity (Figure 2c) was observed in all treatments. On day 44, MOX treatments applied to PN2 (T3 and T4) showed significantly higher increases on these colour parameters, which were not significantly affected by the different MOX dosages. In PN1, significant increases were only found in T2 with the higher oxygen dosage (p < 0.05, Table S1, supporting information), but smaller compared to T3 and T4. Similar observations were found in the absorbance of SO₂ resistant pigments at 520 nm in T2, T3 and T4 (Figure 2d).

Figure 2. Effect of MOX at 0.50 and 2.11 mg/L/day for 14 days on Pinot noir wines with a low phenolic content (T1 and T2) and a high phenolic content (T3 and T4) compared to the controls (C1 and C2, respectively) on the change of colour absorbance by spectrophotometric measures at (a) 420 nm, (b) 520 nm, (c) colour intensity (420 + 520 nm), and (d) SO2 resistant pigments at 520 nm (mean + standard error, n = 6).

Save View full size Expand inline Collapse inline

These occurred with an induced decline in anthocyanins, where the decrease of monomeric anthocyanins (Figure 3a) and total anthocyanins (Figure 3b) in T3 (10 % and 10 %, respectively) and T4 (17 % and 13 %, respectively) was considerably higher than in T1 (3 % and 6 %, respectively) and T2 (5 % and 7 %, respectively). The higher decrease of anthocyanins, but increase in colour parameters in Pinot noir wine due to MOX, has been reported in Durner et al. (2010) and Yang et al. (2021a) as well as in other red wine varieties (Anli and Cavuldak, 2012). Here, the results showed that MOX treatments had a much stronger impact on Pinot noir wines with higher phenolic content including more monomeric anthocyanins (i.e., PN2), which can be combined to form coloured polymeric pigments.

Figure 3. Effect of MOX at 0.50 and 2.11 mg/L/day for 14 days on Pinot noir wines with a low phenolic content (T1 and T2) and a high phenolic content (T3 and T4) compared to the controls (C1 and C2, respectively) on the change of colour absorbance by spectrophotometric measures for (a) monomeric anthocyanins, by the Harbertson-Adams assay for (b) total anthocyanins (mg/L, malvidin-3-glucosie equivalents), (c) small polymeric pigments, and (d) large polymeric pigments (mean ± standard error, n = 6).

Save View full size Expand inline Collapse inline

After the final SO₂ addition, measured at day 48, the absorbance at 420 nm and 520 nm and colour intensity decreased in all treatments. However, a greater decline appeared in wines with MOX (8~26 %) compared to the control wines (5~15 %) and especially for T3 and T4. This could be due to the bleaching of anthocyanins derived from self-association and co-pigmentation reactions (Boulton, 2001) that were previously promoted by MOX. As a result, in both PN1 and PN2, no significant variation was found between the control and MOX treatments in the 420 nm and 520 nm absorbances and colour intensity in the end. Nevertheless, the content of SO₂ resistant pigments remained significantly higher in T2, T3 and T4, therefore, MOX still had a durable influence on the evolution of coloured pigments in these wines. In PN2, a further decrease in the monomeric (except for T4) and total anthocyanins (up to 7 % and 8 %, respectively) was also observed and the final concentrations were significantly lower in T3 and T4. However, these were not found in T1 and T2.

For the polymeric pigments measured using the Harbertson-Adams assay, SPP are a heterogeneous mixture of anthocyanin derived products that do not precipitated with protein and LPP from anthocyanins reacting with tannins that do (Harbertson et al., 2003). In PN1 wines, SPP decreased across all treatments (Figure 3c), while a higher increase of LPP (Figure 3d) was found in T1 and T2 (p < 0.05), showing the interactions between tannins and anthocyanins in the presence of oxygen (He et al., 2012) in these wines. At wine pH, this has been shown to generate coloured flavyliums, which would explain the increase in the 420 nm and 520 nm absorbances and colour intensity in T1 and T2. In PN2 wines, on the contrary, SPP remained almost unchanged in T4 and only slightly decreased in T3 after MOX, compared to a higher decline in C1. Meanwhile, LPP increased in PN2 wines over time, including the control (C2), but were significantly higher in T3 and T4. Therefore, to wines with the higher initial phenolic content, MOX had not only significantly promoted anthocyanin condensation reactions with tannins, but also maintained a higher concentration of polymeric pigments with anthocyanin colour. These variations between PN1 and PN2 reflected their differences in the anthocyanin content.

After SO₂ addition, a decline of SPP was seen in all wines (i.e., 15 % in C1, 13 % in T1, and 22 % in T2; 14 % in C2, 21 % in T3 and 16 % in T4), while the LPP content in wines remained similar to these before the SO₂ addition. The latter can be linked to the content of SO₂ resistant pigments in these wines. The decline in SPP would have contributed to the loss of wine colour at 420 nm and 520 nm absorbances and colour intensity observed after SO₂ addition. Cheynier et al. (2006) reported that the stability of pigments is not related to molecular weight, as some polymeric pigments are exactly like the anthocyanin precursors and are not resistant to sulfite bleaching. Moreover, the limited acetaldehyde production could lead to a lower content of acetaldehyde-mediated pigments in the wines. The importance of acetaldehyde on increasing colour stability and intensity has been well discussed, promoting pigment formation with the ethyl-bridged covalent bond that is more resistant to cleavage and SO₂ bleaching (Escribano-Bailón et al., 2001; Picariello et al., 2017; Han et al., 2019). In a recent study (Yang et al., 2021a), after SO₂ addition, a decrease of 520 nm absorbance and colour intensity was also observed in wines with MOX applied after MLF (approximately 6 % and 4 %, respectively), which, however, were much less than in the control treatments (approximately 15 % and 12.5 %, respectively). The acetaldehyde production in this previous study (Yang et al., 2021a) was at least 10 times higher than in the present study. Nevertheless, the content of SO₂ resistant pigments remained significantly higher in T2, T3 and T4. Therefore, MOX still had a durable influence on coloured pigments evolution. The trends seen in the present study could also suggest further modifications on the small polymeric anthocyanins, as well as reactions with other phenolic compounds released from SO₂ cleavage reactions to form colourless species, e.g., some of the anthocyanin-tannin adducts (Remy and Moutounet, 2000). Together, these could have also contributed to the decrease in wine colour observed in all wines after SO₂ addition.

3. Monomeric phenolic content

A total of 14 monomeric phenolics, including hydroxycinnamic and hydroxybenzoic acids, flavan-3-ols, flavonols and malvidin-3-glucoside, were identified in the two Pinot noir wines at the beginning of the experiment (except for ferulic acid in PN1) (Table 3). Their initial concentrations were higher in PN2, indicating a higher co-pigmentation capacity (Boulton, 2001). The monomeric phenolics were analysed again at day 44 and after the final SO₂ addition at day 48. With MOX treatments, significant declines were found in the concentration of malvidin-3-glucoside (i.e., the only monomeric anthocyanin detected), flavan-3-ols, and several flavonols.

In both PN1 and PN2, the decline in malvidin-3-glucoside was consistent with the results of spectrophotometric measures of monomeric and total anthocyanins. Between the control wines, the decrease in C2 was close to 2 times higher than in C1. Under the same MOX dosage, the overall decrease in PN1 was minimal (i.e., up to about 5 mg/L) compared to in PN2 (i.e., around 30 to 50 mg/L). After SO₂ addition, a further decrease of malvidin-3-glucoside was found, although this was not large (i.e., up to 12 mg/L). For the flavan-3-ols, the decrease in (+)-catechin induced by MOX was much greater in PN1, which was approximately 2 times more in T2 than T4 and only occurred in PN1 at the lower MOX dosage (T1). Meanwhile, a decrease in (-)-epicatechin by MOX was only found in PN2.

Among the flavonols, quercetin, and quercetin-3-glucoside, decreased during the experiment. Quercetin in the free form is an important cofactor for anthocyanin co-pigmentation (Boulton, 2001; He et al., 2012) and is also prone to oxidation (Zenkevich et al., 2007). In the control treatments, the decrease of quercetin from time 0 to the endpoint was very small, by less than 1 mg/L in C1 and around 2 mg/L in C2. With MOX, a higher decrease in quercetin was found in both PN1 and PN2. Under the same oxygen dosage, the decrease of quercetin in wines at the endpoint was somewhat similar, which was around 20 % and 19 % in T1 and T3, respectively, and higher at 29 % and 31 % with the higher oxygen dosage in T2 and T4, respectively. Thus, the influence of MOX on quercetin appeared to be dependent on the oxygen dosage and also in proportion to its initial content. Finally, similar losses of quercetin-3-glucoside were observed in PN2 with time, independent of the treatments, suggesting a loss due to the hydrolysis mechanism that was not affected by MOX. In PN1, quercetin-3-glucoside was not detected in the end.

4. Tannin composition

Results for the tannin profiles of the wines are presented in Table 4. The tannin concentration (i.e., tannin reactive with methylcellulose) increased in all wines during this experiment, but with a greater percentage increase in PN1 (30 % to 42 %) than in PN2 (22 % to 32 %). SO₂ addition did not show a clear impact on tannin concentration. In the end, the highest tannin concentration in both PN1 and PN2 was found with the highest oxygen dosage (T2 and T4). Oxidised tannins found in the modality are known to have greater intramolecular interactions, forming more condensed structures. However, this might have a negative impact on perceived astringency, which has been shown to increase with a higher tannin concentration (McRae and Kennedy, 2011). Despite the increase in tannin concentration, the mDP of tannins in both PN1 and PN2 remained very small (~2) and was not affected by MOX or the initial polyphenol content. This might be associated with the varietal characteristics of Pinot noir grapes that have a lower skin to seed tannin ratio (Sparrow et al., 2016), in which skin tannins are known to have a higher mDP (3 to 83), while a lower mDP (2 to 17) has been associated with seed tannins (McRae and Kennedy, 2011; Harrison, 2018). It needs to be also recognised that the measured mDP only applies to depolymerisable tannins, namely the molecules that can be broken up in the phloroglucinolysis reaction, which in the studied wines was no more than 60 % towards the end of the trial.

The yield (%) of total tannin polymers from the phloroglucinolysis reaction decreased in all wines, particularly with the T2 treatment of PN1. Together with the higher increase in tannin concentration, this showed a stronger impact of the higher MOX dosage on PN1, seen here by there being more tannin macromolecules that could not be easily depolymerised. Additionally, a loss in tannin yield (%) can also be associated with the formation of pyranoanthocyanins (McRae et al., 2015). The authors reported a lower tannin yield (%) when more pigmented tannins were present, due to a decline in the acid-labile bonds in the newly formed pigmented tannins. In theory, this would have affected PN2 with MOX more than PN1, due to the much larger increase in polymeric pigments. However, this was not the case in the present study. Here, the influence of pigment types on tannin composition and yield (%) determined by the phloroglucinolysis reaction needs to be considered, as Pinot noir wines are lacking in acetylated anthocyanins. This will require further investigation.

After SO₂ addition, a further decrease in tannin yield (5~7 %) was found in PN1, but not in PN2. It is well known that wine tannins can react with SO₂, forming sulfonated flavanols at the terminal positions (Ma et al., 2018). This group of tannin molecules are unfortunately not measured by the phloroglucinolysis method. Currently, tannin sulfonation has only been reported in wines after a long period of ageing (Ma et al., 2018, Arapitsas et al., 2018; Watrelot et al., 2020), while in model wine solutions they have been detected soon after SO₂ addition (Bonaldo et al., 2020). Future studies will be required to investigate the impact of tannin sulfonation on young red wines. Another possibility is that SO₂ addition may have resulted in bond breaking with some of the tannins producing smaller molecules (Ma et al., 2018), which could then undergo rearrangement reactions and even tannin polymerisation (McRae and Kennedy, 2011). For the PN2 wines, the influence of SO₂ on tannin composition may have been lower due to coloured polyphenols reacting with SO₂ in preference to the tannins, as seen by the larger decline in colour parameters with the PN2 wines described above. The variation in tannin yield (%) loss in response to SO₂ addition can therefore be associated with the differences in phenolic content between PN1 and PN2.

Regarding tannin composition, the degree of galloylation (%Galloyl) remained similar across all treatments, whereas trihydroxylation (%Tri-OH, (-)-epigallocatechin extension units) significantly declined in both PN1 and PN2 with MOX. In Yang et al. (2021b), an increase of %Galloyl was found in Pinot noir wines following MOX, although the wines had commercial tannin and oak products added at primary fermentation. The hydrolysis of gallate esters is known to occur during wine ageing but is not always directly affected by oxidation (Watrelot et al., 2020). In the present study, commercial tannin and oak products were not used, and the role of such additives on the evolution of Pinot noir tannin composition during MOX deserves further investigation.

A decrease in (-)-epigallocatechin extension units, associated with skin-derived tannins, was observed in both Pinot noir wines. Such a decrease in response to oxygen exposure has been reported previously in model wines (Lee, 2010; Carrascón et al., 2018), and in wines subject to oxygenation (McRae et al., 2015; Watrelot et al., 2020; Yang et al., 2021b), with linkage made to the impact of Saccharomyces cerevisiae as having a high affinity towards (-)-epigallocatechin units (Mazauric and Salmon, 2005). In Watrelot et al. (2020), this influence of oxygen on (-)-epigallocatechin extension unit was recorded in wines even after ageing for 3 years. This trends, seen here in the absence of yeast effects, could be an unwanted outcome, as (−)-epigallocatechin has been suggested as being responsible for the softer taste in skin tannins despite higher mDP values (Fernández et al., 2007; Ma et al., 2014), and for decreasing the perception of “coarse” sensations from tannins (Vidal et al., 2003). Also, (-)-epigallocatechin has been shown to have a weaker cellular response to human bitter taste receptors than the galloyl group of tannin subunits (Narukawa et al., 2011).

Besides the loss of (-)-epigallocatechin extension units, no other significant variation was found in PN2 wines between the treatments. In PN1 wines, before SO₂ addition, a decrease of (+)-catechin and (-)-epicatechin extension units was observed in T2, which led to a higher proportion of (+)-catechin, (-)-epicatechin and (-)-epicatechin-gallate terminal units in the depolymerised tannins. Tannin self-association and condensation reactions could have played a role in the changes seen in tannin composition seen with the phloroglucinolysis procedure. The final SO₂ addition affected the tannin subunits in PN1, where a decrease in the (-)-epicatechin terminal units and increase in (+)-catechin terminal units was observed. This is also of concern, as a larger proportion of catechin subunits could also enhance tannin binding with salivary proteins (McRae and Kennedy, 2011).