VOLATILE PROFILE OF TWO MONOVARIETAL WHITE WINES UNDER DIFFERENT ANTIOXIDANT ENVIRONMENTS DURING STORAGE IN BOTTLE

The volatile organic compounds (VOCs) formed during winemaking can be modulated by several additives, namely by the use of sulphur dioxide (SO 2 ) which has been well-accepted on winemaking as a preservative agent. However, some drawbacks associated with SO 2 wine application led to the need to reduce or replace its use. In this work, VOCs profile after storage in bottle under different antioxidant conditions of two Portuguese monovarietal wines (‘Arinto’ and ‘Síria’) was studied.


INTRODUCTION
Wine is a complex chemical mixture whose aroma is mostly originated by volatile organic compounds (VOCs), which depend on grape quality, winemaking and ageing processes. The matrix composition continues to change even during storage. This evolution can affect wine aroma properties presenting a significant impact on consumer acceptance of the product (Bindon et al., 2014;González-Barreiro et al., 2015;Belda et al., 2017;Piras et al., 2020;Echave et al., 2021). Oxidative reactions play an important role in aroma evolution during storage, in particular on white wines, which are more prone to the non-enzymatic oxidative phenomenon (Pati et al., 2014). In this process, a deterioration of white wines overall quality can occur by a controlled short-term oxidation promoting the development of a characteristic bottle bouquet (Cheynier et al., 1989;Kallithraka et al, 2009).
The oxygen present on the headspace of the bottle and dissolved in wine is reduced to hydrogen peroxide (H2O2) among other reactive oxygen species by metal redox mediation by iron and copper. The H2O2 formed reduce ketones, aldehydes and polyphenols to quinones, which results in loss of characteristic aroma (Waterhouse and Laurie, 2006;Oliveira et al., 2011). To prevent this aroma loss, several methods are used like the addition of antioxidant agents or in combination with a controlled maturation on lees. Sulphur dioxide (SO2) is a well-accepted preventing agent in the wine industry. It is used at different stages of the process due to its different functions, namely, the ability to act as an antioxidant and antiseptic. When added to musts or wine an acid-based equilibrium occurs between three forms, molecular SO2, sulphite ion (SO3 2-) and bisulphite ion (HSO3 -), known as the "free SO2". It can also react with unsaturated compounds producing the commonly known as "bound SO2". The HSO3reacts with H2O2 produced by oxygen metal reduction promoting the reduction of phenolic compounds. Also, SO2 reacts with quinones, reducing them back to the previous phenolic forms preventing wine aroma loss as already described (Roberts and McWeeny, 1972;Manzocco et al., 1998). However, an excessive exposer of the population to this preservative agent has led to an increase of reported allergic episodes (Vally et al., 2009;Zhang et al., 2014). As a response to the health concern, European Union established a requirement of a warning in the label indicating the presence of sulphites when present above 10 mg/kg or 10 mg/L (EU-28, 2016). The drawbacks associated to SO2 wine application either by the legal limits, health concerns, off-flavours that can be generated and formation of sulphuric acid, lead to the need to reduce or even replace the use of SO2 (Guerrero and Cantos-Villar, 2015;Echave et al., 2021). Maturation on lees has been used to increase the antioxidant protection of wine. Lees are a mixture of yeast cells, lactic acid bacteria and organic and inorganic precipitates. The contact of wine with this mass prior to bottling may promote the release of additional substances like polyphenols, mannoproteins or glutathione which, in a controlled manner, may help to prevent oxidation. Moreover, a controlled exposure has been indicated to generally improve wine sensory characteristics described as having more body (mouthfeel) and flavour (Fornairon-Bonnefond and Salmon, 2003;Comuzzo et al., 2015).
Previously, the impact on VOCs profile of different antioxidant conditions (different application doses of SO2 and its replacement by ascorbic acid) on monovarietal white wine fermentation was analysed (Almeida Santos et al., 2020). The influence of the presence or absence of bentonite was also studied. The same monovarietal musts under the conditions applied on the study lead to different wine VOCs profiles with a varietal dependency. The modification of these parameters on the fermentation step should hence be taken into consideration by winemakers. The varieties evaluated in the study were 'Arinto' and 'Síria', two Portuguese autochthonous white grape varieties. Both varieties integrate the list of mandatory varieties to produce wine products entitled as protected designation of origin (PDO) "Alentejo" (Portugal). 'Arinto' variety is typical from the north of Portugal but gaining importance in other wine-producing regions like Alentejo, south of Portugal. 'Síria' is a variety very common on Alentejo that as gathering increasingly interest by winemakers (Fernão-Pires, 2018;Almeida Santos et al., 2020).
This work is in line with the previous one (Almeida Santos et al., 2020). The present work is focused on the effect of different antioxidant conditions on VOCs, which was assessed by headspace-solid phase microextraction-gas chromatography/mass spectrometry (HS-SPME-GC/MS), comparing wine VOCs after three months on lees and three months in bottle under different antioxidant environments.

Wine samples
Wines studied in this work were obtained from 'Arinto' and Siria grapes harvested in September 2018 in Évora University vineyard. Grapes were destemmed, crushed, pressed and racked after a 24 h cold static sedimentation. The clarified must was divided into 10 L and 5 L vessels, and inoculated with a mixture of commercial Saccharomyces cerevisiae (mixture 1:1 of LEVULINE FB from Oenofrance and IOC 18-2007 from Lallemand OEnology). SO2, 0, 50 and 100 mg/L of SO2 ('Arinto' variety) and 0, 15, 30 and 45 mg/L of SO2 ('Síria' variety) using a commercial 6 % aqueous solution of sodium bisulfite (SAI, SOLFOX 6 Nº CE: 231-870-1) and ascorbic acid (100 mg/L) (MERCK CAS:50-81-7) were added. Bentonite (0.1 g/L) (aqueous solution of 10 % (w/v), MICROCOL ALPHA, LAFFORT), was added only to half of the samples. Samples were fermented at 16 °C in duplicates. The lower doses of SO2 used in 'Síria' musts are related to the presumably higher resistance to oxidation of this variety, according to local viticulture, but not yet studied in deep. After alcoholic fermentation, each wine was split into two sample series, and a second dose of 0 and 60 mg/L of SO2 for 'Arinto' wines and 30 e 60 mg/L of SO2 for 'Síria' wines was immediately added using the same commercial solution. Wines were kept in contact with lees for three months at 16 ºC, on amber 2.5 L glass carboys. After this period, wines were separated from lees, bottled and stored vertically for three months at 16 ºC on amber bottles of 75 mL. In total, 40 'Síria' wines and 30 'Arinto' wines were analysed. Figure 1 shows a summary of the experimental design. Wines were analyzed after alcoholic fermentation and after six months (three months sur lees and three months in bottle) Oenological parameters as total acidity (potentiometric titration, Method OIV-MA-AS313-01) and volatile acidity (distillation and titration, Method OIV-MA-AS313-02), alcoholic strength (distillation, Method OIV-MA-AS312-01B), pH (potentiometry, Method OIV-MA-AS313-15), reducing sugars (Method OIV-MAAS311-01A) and sulphur dioxide (iodometric titration, Method OIV-MA-AS323-04B) were measured according to OIV methods (OIV, 2018;Almeida-Santos et al., 2020). 'Arinto' and right scheme for 'Síria'. Same protocol was used without bentonite addition (5 L vessels were used for AA due to insufficient must for 10 L vessels).

HS-SPME sampling of wine volatiles
The HS-SPME sampling experiments were carried out as reported by Almeida Santos et al. (2020). Briefly, to 5.0 mL of sample in a 22 mL vial, 2 g of sodium chloride were added. The vial was sealed with a Teflon-lined rubber septum/magnetic screw cap and was equilibrated for 5 min at 30 °C, and then the sample headspace was extracted for 30 min at the same temperature with SPME fiber. The extraction occurred by exposing 1 cm of a DVB/Carb/PDMS fiber, 50/30 μm film thickness (df), supplied from Supelco, (Bellefonte, PA, USA). Prior to use, the fiber was conditioned following the manufacturer's recommendations. Fiber blanks were run periodically to ensure the absence of contaminants and/or carryover. Thermal desorption of the analytes was carried out by exposing the fiber in the GC injection port at 260 °C for 3 min in splitless mode, for the same time period. All samples were analysed in duplicate.

GC/MS analysis
The analyses were performed on a GC/MS system consisting of a Bruker GC 456 with a Bruker mass selective detector Scion TQ. An automatic sampler injector was used: CTC Analysis autosampler CombiPAL. Chromatographic separation was achieved on a SupelcoWaxTM 10 PLUS capillary column (60 m × 0.25 mm i.d., 1.0 μm, df), supplied by Supelco Analytical (Supelco, Bellefonte, PA). The oven temperature program began at 40 °C hold for 5 min, raised to 240 °C at 4 °C/min, holding for 5 min. Helium was used as carrier gas at constant flow of 1.7 mL/min at the electronic flow control (EFC 21). The MS transfer line and source temperatures were set at 260 °C. Data were acquired with MSWS 8.2 Bruker and analyzed with Bruker MS Data Review 8.0. Spectra were matched with NIST MS Search Program Version 2.3. The recording of retention times and characteristic mass fragments was obtained at an electron ionization (EI) of 70 eV. The mass spectra of the analytes were recorded at full scan, from 40 to 450 Da. The analyses of a commercial hydrocarbon mixture (C8 -C20) under the same chromatographic conditions was used to calculate the linear retention indices (LRIs) for each compound. The relative amounts of individual components are expressed as percent peak areas relative to the total peak area of the chromatogram (Relative Peak Area -RPA).

Statistical analysis
Principal components analysis (PCA) was performed using XLSTAT Version 2020.5.1 by Addinsoft, to reduce the number of variables and to detect a pattern in the relationship between the variables (% of peak area of each compound) and the wine samples.
One-way analysis of variance (ANOVA) using Tukey´s multiple comparisons test was performed to compare the means at the level of significance of p < 0.05, for total and free SO2 content in the wine samples. Two-way analysis of variance (ANOVA) using Tukey´s multiple comparisons test was performed to compare the means at the level of significance of p < 0.05, for VOCs, comparing wines immediately after fermentation and after 6 months (3 months on lees and 3 months on bottle). For ANOVA analysis three factors were considered, SO2 doses on fermentation (0, 50, 100 mg/L for 'Arinto' wines and 0, 15, 30, 45 mg/L for 'Síria' wines), SO2 doses after fermentation (0, 60 mg/L for 'Arinto' wines and 30 and 60 mg/L for 'Síria' wines) and the presence or absence of bentonite, using GraphPad Prism version 9.0.0 by GraphPad Software.

SO2 in wine samples
After alcoholic fermentation of 'Arinto' and 'Síria' wines, total and free SO2 content was analysed. A second addition of SO2 with two different doses was applied (0 and 60 mg/L of SO2 to 'Arinto' wines, 30 and 60 mg/L of SO2 to 'Síria' wines) and total and free SO2 content was analyzed again, using the same iodometric titration method, after six months (three months on lees and three months on bottle). On 'Síria' wines, 30 mg/L of SO2 instead of 0 mg/L of SO2 was applied since these wines were fermented with lower antioxidant doses and presented lower levels of total SO2. Table I for 'Arinto' wines and Table II for 'Síria' wines summarized the SO2 content for both wines with and without bentonite. The results show that both SO2 forms (total and free) decreased with time in both wine samples, considering the sum of the initial concentration of SO2 present on initial wines and the second addition of SO2. The decline of total SO2 may indicate its effectiveness as an antioxidant during the time, as already reported (Carrascón et al., 2018).
When performing ANOVA on the averages data for total and free SO2 on each wine, the observed differences do not have a response trend in relation to the applied conditions according to the three factors in the study (initial and second antioxidant conditions and presence or absence of bentonite on fermentation). Both total and free SO2 behaved as described before (Gambuti et al., 2020), showing a decreasing content over time, regardless of the added SO2 applied (Sacks et al., 2020;Stockley et al., 2021).

Volatile organic compounds identification
The HS-SPME-GC/MS method was applied to wine samples to characterize the VOCs profile. Each sample was analysed in duplicate (antioxidant dose combinations and presence or absence of bentonite on fermentation). Duplicates are indeed enough, since according to recently published works with comparable matrices (Almeida Santos et al., 2020;Ferreira et al., 2021;Saracino et al., 2021;Pereira et al., 2021;Ferreira et al., 2022) results showed that duplicates instead of triplicates do not jeopardize the associated error of the measurement and the resulted error for both cases was very similar. This fact allows to perform more analysis by time unit promoting the throughput of samples, an important issue in routine analysis, when needed. Table I Free and total SO2 content of initial 'Arinto' wines, with and without bentonite addition and six months after the alcoholic fermentation (AF) and the second SO2 addition

Table II
Free and total SO2 content of initial 'Síria' wines with and without bentonite addition, and 6 months after alcoholic fermentation (AF) and the second SO2 addition  100 mg/L of ascorbic acid 30 6.5(±0.7) 46.5(±2.1) ABa1 60 6.5(±0.7) 64.5(±0.7) Ab1 Different capital letters means that each calculated value differ significantly (p<0.05) according to initial antioxidant condition factor; different lowercase letters means that each calculated value differ significantly (p<0.05) according to second antioxidant addition factor; different number means that each calculated value differ significantly (p < 0.05) according to presence or absence of bentonite . VOCs were tentatively identified by matching mass spectra with spectra of reference compounds in NIST library. Linear retention indices (LRI) were also calculated using a commercial hydrocarbon mixture (C8 -C20) and compared with LRIs described in the literature (Bianchi et al., 2007;Oliveira et al., 2008;Mateus et al., 2010;Di Mattia et al., 2015;Almeida Santos et al., 2020;Pereira et al., 2021). Table III presents the information regarding VOCs tentatively identified by compound number.

Volatile organic compounds in wines
Only compounds that present a relative area above 0.0001% of the total chromatogram area were considered for statistical purposes. For both wines, the relative amount of each compound was calculated as the percentage ratio of the respective peak area in relation to the total peak area analyzed (RPA) of the chromatogram (Table IV for 'Arinto' wines and Table V for 'Síria' wines). 93 VOCs, which were present at least in one replicate sample, were tentatively identified. Additionally, 54 more compounds were also detected, but their identification was not possible to assign since the library match was below 850. This limit was established for identification assignments. comparing with literature data of the correspondent LRI data. The analyses of the VOCs profile for both wines ('Arinto' and 'Síria') allowed to observe that esters (44 compounds), alcohols (19 compounds), carboxylic acids (7 compounds), ethers (4 compounds), ketones (3 compounds) and aldehydes (4 compounds) were the most frequent chemical groups, together with 12 miscellaneous compounds, also confirmed by the literature (Cabrita et al., 2006;Almeida Santos et al., 2020;Petronilho et al., 2020). Regarding total chromatographic area, it was higher for wines fermented with bentonite analyzed after six months. For these wines it was possible to tentatively identify compounds accounting for 99 % of the total chromatographic area for both 'Arinto' and 'Síria' wines. Same result was obtained for wines fermented without bentonite analyzed after six months. The total area by chemical classes for each oxidant condition after six months (Table IV and Table V) was compared with the total area of wines after alcoholic fermentation, reported on a previous work (Almeida Santos et al., 2020). Esters and alcohols presented the highest variations in both wines. Figure  2 (a and b for 'Arinto' wines; c and d for 'Síria' wines, with and without bentonite respectively) illustrate the variations of VOCs in wines with six months using wines after alcoholic fermentation as reference.
Clearly, the effect of adding or not adding bentonite during fermentation, impacts the evolution of volatiles over time ( Figure 2). The evolution over six months led to an increasing of esters and a decrease of alcohols. The increase of esters amounts during ageing at temperatures above 5 ºC has been also reported by Garde-Cerdán et al. (2008). However, the total concentration of alcohols decreased, contrary of results reported for wines bottled above 5 ºC (Garde-Cerdán et al., 2008).
For VOCs, a two-way ANOVA using Tukey´s multiple comparisons test at the level of significance of p<0.05, was performed considering: initial and second antioxidant conditions and the presence or absence of bentonite on fermentation (Tables IV, V, VI and VII). A total of 11 compounds for 'Arinto' and 'Síria' wines presented at least one statistical difference: ethyl acetate (1), isoamyl acetate (9), isoamyl propionate (11), ethyl hexanoate (14), ethyl octanoate (23), ethyl decanoate (33) and 2-propanol (52) for both wine varieties. In general, compounds 1, 9, 11, 14 and 23 increased after wine maturation. Changes were even clearer between different antioxidant conditions on wines fermented with bentonite. Ethyl esters were already reported by other authors as being affected by bentonite (Vincenzi et al., 2015), indicating a decrease in their amount. 'Arinto' wines were more sensitive to the conditions applied. A decrease on ethyl esters of branched-chain fatty acids during maturation on lees was observed. This suggests an increase of long-chain alcohols and volatile fatty acids. The presence of lees and the autolysis on yeast cells also seems to contribute to the release of fatty acids that will produce VOCs such as esters, aldehydes, and ketones (Styger et al., 2011). Ethyl decanoate (33) and 2-propanol (52) on 'Arinto' wines fermented without bentonite increased, while in all the other wines decreased. Propyl acetate (3), 1-hexanol (58) and phenethyl alcohol (68) for 'Arinto' wines and ethyl 9-decanoate isomer (37) and heptanoic acid (78) for 'Síria' wines in general increased with time. This increase was less noticeable for the higher dose of SO2 applied. On the other hand, acetaldehyde ethyl amyl acetal (47) decrease over time on 'Síria' wines fermented without bentonite  (Janzantti and Monteiro, 2017;Wang et al., 2017;Kong et al., 2019;de-la-Fuente-Blanco et al., 2020;Almeida Santos et al., 2020;Pereira et al., 2021) (Janzantti and Monteiro, 2017;Wang et al., 2017;Kong et al., 2019;de-la-Fuente-Blanco et al., 2020;Almeida Santos et al., 2020;Pereira et al., 2021)

Table IV
Relative amount of each compound calculated as the percentage ratio of the respective peak area in relation to the total peak area analysed (RPA) of the chromatogram for 'Arinto' wines fermented without bentonite  .000 RPApercentage ratio of the respective peak area relative to the total peak area; n = 2 with SO2 and n = 1 with ascorbic acid. Different capital letters means that each value differ significantly (p<0.05) according to initial antioxidant condition factor; different lowercase letters means that each value differ significantly (p <0.05) according to second antioxidant addition factor; different number means that each value differ significantly (p <0.05) according to presence or absence on fermentation of bentonite. Relative amount of each compound calculated as the percentage ratio of the respective peak area in relation to the total peak area analysed (RPA) of the chromatogram for 'Arinto' wines fermented without bentonite  .000 RPApercentage ratio of the respective peak area relative to the total peak area; n = 2 with SO2 and n = 1 with ascorbic acid. Different capital letters means that each value differ significantly (p<0.05) according to initial antioxidant condition factor; different lowercase letters means that each value differ significantly (p <0.05) according to second antioxidant addition factor; different number means that each value differ significantly (p <0.05) according to presence or absence on fermentation of bentonite Table V Relative amount of each compound calculated as the percentage ratio of the respective peak area in relation to the total peak area analysed (RPA) of the chromatogram for 'Arinto' wines fermented with bentonite. Fermentation 0 mg/L SO2 50 mg/L SO2 100 mg/L SO2 100 mg/L ascorbic acid After fermentation Compound no. 0 mg/L SO2 60 mg/L SO2 0 mg/L SO2 60 mg/L SO2 0 mg/L SO2 60 mg/L SO2 0 mg/L SO2 60 mg/L SO2 1 6.835 Aa2 ± 0.001 6.260 Aa2 ± 0.002 6.020 Aa2 ± 0.008 5.270 Aa2 ± 0.004 3.362 Aa2 ± 0.019

Table V (continuation)
Relative amount of each compound calculated as the percentage ratio of the respective peak area in relation to the total peak area analysed (RPA) of the chromatogram for 'Arinto' wines fermented with bentonite

Table VI
Relative amount of each compound calculated as the percentage ratio of the respective peak area in relation to the total peak area analysed (RPA) of the chromatogram for 'Síria' wines fermented without bentonite RPApercentage ratio of the respective peak area relative to the total peak area; (n = 2). Different capital letters means that each value differ significantly (p<0.05) according to initial antioxidant condition factor; different lowercase letters means that each value differ significantly (p<0.05) according to second antioxidant addition factor; different number means that each value differ significantly (p<0.05) according to presence or absence on fermentation of bentonite Relative amount of each compound calculated as the percentage ratio of the respective peak area in relation to the total peak area analysed (RPA) of the chromatogram for 'Síria' wines fermented without bentonite. RPApercentage ratio of the respective peak area relative to the total peak area; (n = 2). . Different capital letters means that each value differ significantly (p<0.05) according to initial antioxidant condition factor; different lowercase letters means that each value differ significantly (p<0.05) according to second antioxidant addition factor; different number means that each value differ significantly (p<0.05) according to presence or absence on fermentation of bentonite.

Table VII
Relative amount of each compound calculated as the percentage ratio of the respective peak area in relation to the total peak area analysed (RPA) of the chromatogram for 'Síria' wines fermented with bentonite. RPApercentage ratio of the respective peak area relative to the total peak area; (n = 2). Different capital letters means that each value differ significantly (p<0.05) according to initial antioxidant condition factor; different lowercase letters means that each value differ significantly (p<0.05) according to second antioxidant addition factor; different number means that each value differ significantly (p<0.05) according to presence or absence on fermentation of bentonite.

Table VII (continuation)
Relative amount of each compound calculated as the percentage ratio of the respective peak area in relation to the total peak area analysed (RPA) of the chromatogram for 'Síria' wines fermented with bentonite. RPApercentage ratio of the respective peak area relative to the total peak area, (n=2). Different capital letters means that each value differ significantly (p<0.05) according to initial antioxidant condition factor; different lowercase letters means that each value differ significantly (p<0.05) according to second antioxidant addition factor; different number means that each value differ significantly (p<0.05) according to presence or absence on fermentation of bentonite

Principal component analysis of volatile organic compounds after storage in bottle
Principal Component Analysis (PCA) was conducted to study the evolution of wines subject to different oxidant conditions (from fermentation to bottle) and allow to observe how the presence of bentonite could impact on VOCs profile of the 'Arinto' and 'Síria' wines. This statistical procedure is very useful for chromatographic aroma analysis, also demonstrated elsewhere (Gomes Da Silva and Chaves Das Neves, 1997;Gomes Da Silva and Chaves Das Neves, 1999;Mateus et al., 2010;Almeida Santos et al., 2020;Pereira et al., 2021) in order to understand underlying data structures. Indeed, PCA reduces the number of variables that explain the system variance allowing to detect a pattern in the relationship between the variables and the wines. For the PCAs illustrated in Figure 3 ('Arinto' wines) and Figure 4 ('Síria' wines), the relative area of the tentatively identified VOCs (Table III) were used. These relative areas were normalized against the initial wines (after alcoholic fermentation), fermented with or without bentonite under different antioxidant combinations ('Arinto' wines fermented with 0, 50 and 100 mg/L of SO2 and 100 mg/L of ascorbic acid and 'Síria' with 0, 15, 30 and 45 mg/L of SO2 and 100 mg/L ascorbic acid). As described before (Almeida Santos et al., 2020), when data are normalized considering as reference the initial wines without treatments, there are more evident differences among samples. Hence, this methodology was applied. For maturation, 'Arinto' wines were kept with additional 0 and 60 mg/L of SO2, and 'Síria' wine with 30 and 60 mg/L of SO2. For 'Arinto' and 'Síria' wines, 74% and 55 % of the variance was explained by the first and second principal components (PC1 and PC2), respectively. Figure 3 shows the PCA for 'Arinto' wines. PC1 explains 63.58% of the variance and clearly separates the samples regarding the presence of bentonite. The PC2 explains 10.55% of the variance and separates the samples by the combination of the antioxidant conditions. On the positive side of PC1 is possible to observe only wines fermented without bentonite (except for G sample), and in the negative side of PC1 the wines fermented in the presence of bentonite. Indeed, wine samples that fermented without bentonite and with ascorbic acid (G and H samples), are the ones that presented a more distinct VOCs profile evolution. Differentiation is due to the contribution of isoamyl propionate (11), ethyl 2octenoate isomer (30) and unknown 12 (105). Regarding the other samples fermented without bentonite, a higher dispersion is observed, comparing with the wines fermented with bentonite. Propyl acetate (3), 2,4,5-trimethyl-1,3-dioxolane (46), 2undecanone (50), decanal (74) and unknown 35 (128) seem to explain this behaviour.
Besides the VOCs profile of the studied wines after fermentation presented good separation after PCA analysis, as described before (Almeida Santos et al., 2020), an effect is now observed regarding the different conditions applied. Indeed, the evolution of wines fermented with bentonite lead to less differentiation consider the combination of the antioxidant conditions. Among this group of samples, the wines that fermented with 100 mg/L of SO2 and without a second antioxidant addition presented a VOCs profile slightly different. Diethyl succinate (36), an ester associated with the ageing of wine (Comuzzo et al., 2020) and the unknown compound 6 (99) are responsible for that differentiation. Also, wines fermented with ascorbic acid with and without a second antioxidant addition are separated from the group by the unknown compound 32 (125).  wines. The initial wines, after fermentation were used as reference. Black squareswines fermented with bentonite; Black trianglewines fermented without bentonite; Dark blue dotsesters; Dark green dotsethers; Yellow dotsketones; Purple dotsalcohols; red dotsaldehydes; Grey dotscarboxylic acids; Light bluemiscellaneous; Orangeunknowns.
explain the discrimination of wines which fermented and maturate under lower antioxidant conditions. Ethyl butyrate (5), hexyl acetate (16) and butanoic acid (78) contribute to the distribution of wines under higher antioxidant conditions.

CONCLUSIONS
In this work, 'Arinto' and 'Síria' wines were produced under different antioxidant conditions, in the presence or absence of bentonite, and were maturated in different antioxidant conditions. The study of maturated wines allowed studying their impact on the respective VOCs profiles. Regarding free and total SO2 it was possible to conclude that replacing SO2 by ascorbic acid, does not change the SO2 decrease observed. The same can be seen for samples treated with and without bentonite for both varieties. Regarding the VOCs profiles, 11 relevant compounds reflecting the conditions impact were identified by ANOVA analysis, such as the esters ethyl acetate (1), isoamyl acetate (9), isoamyl propionate (11), ethyl hexanoate (14), ethyl octanoate (23), ethyl decanoate (33) and alcohol 2propanol (52) in both wines. Isoamyl butyrate (13), ethyl 9-decanoate isomer (37), heptanoic acid (78) and acetaldehyde ethyl amyl acetal (47) only affected 'Síria' wines, and propyl acetate (3), isoamyl alcohol (55), 1-hexanol (58) and phenethyl alcohol (68) only impacted on 'Arinto' wines. Through the PCA analysis, using wines after fermentation as reference, it was possible to observe that VOCs profile evolution of wines was mainly influenced by the fermentation conditions and less by the postfermentation antioxidant conditions. Regarding the second antioxidant conditions, 'Síria' wines were more sensitive than 'Arinto' wines. 'Arinto' wines fermented without bentonite and ascorbic acid lead to completely different VOCs profiles.