Feature detection, retention time (RT) correction, and preliminary statistical analyses were carried out using XCMS online (XCMSOnline version 2.2.5; XCMS version 1.47.3; CAMERA version 1.26.0, the Scripps Center for Metabolomics, La Jolla, CA; https://xcmsonline.scripps.edu). Raw data conversion, GC/MS spectra processing, set parameters, etc., were described in Yu et al. (9). All chromatograms were simultaneously analyzed under the same conditions. From seven different conditions of the forced aging procedure (Figure 1), the multi-group job was selected, and the seven groups were classified.

Using the extracted features from XCMS-online software as the Y-Variable, a bivariate Pearson's correlation was performed. The marked significant Y-variables, which were the optimized candidate features based on a false discovery rate≤ 0.05, were selected as candidate features for network analysis and qualitative and quantitative analyses. To further simplify these analyses and delete the corresponding features of the IS and ethanol, the m/z and RT of all extracted features were compared in METLIN (https://metlin.scripps.edu) and the NIST 05 library on the GC/MS workstation (Agilent Technologies Inc., USA).

The identification of candidate features in network analysis directly referred to the existing our study (14): The aligned matrix (/.xlsx) from XCMS-Oline software (https://xcmsonline.scripps.edu) consists of all the exported metabolite features-putative identifications through METLIN standard database matching, which are defined as ions with unique m/z and RT values (Supplementary material). Finally, MS features were validated by (i) crosschecking their presence in the raw data, (ii) comparing their features with those present in the NIST 05 (matching degree %) and NIST 98 MS library (http://webbook.nist.gov), and (iii) comparing with the existing identifications (14–17, 46).

Finally, all identified volatile compounds, including unknowns, were directly quantified using HS-SPME-GC/MS.

Bivariate Pearson's correlation was carried out using IBM SPSS Statistics for Windows, version 19.0 (IBM Corp., Armonk, NY USA). Network analysis of the optimized candidate features was performed with Gephi version 0.9.2 (https://gephi.org/) using the Fruchterman–Reingold algorithm. All box plots and PCA sores plots were exported directly from XCMS-online software. Data trends were graphed with OriginPro 8.5.0 SR1 (The OriginLab Corporation, USA).

As shown in Figure 4, about 69% of all the candidate features in region A were identified as alcohols (e.g., 2- and 3-methylbutanol, 2-methylpropanol), followed by ethyl acetate, ethyl lactate, ethyl hexanoate, and acetic acid. Some common characteristics and trends were observed for samples from conditions (I) and (II). For example, the content of 2-methylpropanol, 3-methylbutanol, ethyl lactate, and acetic acid was significantly higher in sample treatments 3 and 5O₂ compared with 1O₂. Oxygen supplementation at week 5 (day 35, 1O₂) showed an increase in marker compounds, which then decreased in the latter stages of storage. In contrast, the four different temperatures of condition (I) had little effect on the amount of volatile aroma compounds. Hence, the results suggested that amounts of the volatile marker compounds in region A were dependent on dissolved oxygen and not readily oxidized.

Previously we found that amounts of 2- and 3-methylbutanol and 2-methypropanol showed significant decreasing trends during the aging of Huangjiu, and these compounds were representative aging markers, especially for the short-aged wine (14–16). Here, 3-methylbutanol and 2-methylpropanol showed some association with 3 and 5O₂, but the decreasing trend fluctuated greatly, especially in sample 5O₂. In a study on the aroma of fortified wine, a high initial oxygen concentration increased the concentration of aroma compounds during aging (21). Based on these findings, we speculated that the interval and frequency of oxygenation during aging could increase overall alcohol production. Since a higher oxygenation frequency also increased the concentration of other aroma compounds, we proposed that its introduction in the early stage of aging would be beneficial for the flavor of Huangjiu.

The organic acids in wine can be formed from the oxidation of alcohols and aldehydes and the hydrolysis of esters, while amounts of 2- and 3-methylbutanol, 2-methylpropanol, acetic acid, ethyl acetate, ethyl hexanoate, ethyl lactate mainly originated from the raw materials and fermentation process of Huangjiu. Consequently, the concentrations of these flavor compounds during fermentation can determine their individual reactivities (23). Since the hydrolysis of esters largely determined concentrations of the corresponding organic acids and alcohols in the aging process, it was reasonable to assume that the above three alcohols, acetic acid, and three ethyl esters occurred in the same environment of region A.

Most of the candidate features in region B were identified as phenethyl alcohol, with lesser amounts of γ-non-alactone, ethyl palmitate, 2,4-dimethylphenol, phenethyl acetate, and other unknowns (Figure 5). Contrary to region A, the interval and frequency of oxygenation (3 and 5O₂) decreased the concentration of these volatile aroma compounds.

Many phenolic compounds have antioxidant properties and are readily oxidized during aging. Correspondingly, the content of 2,4-dimethylphenol following 1O₂ supplementation was significantly higher than that with 3 and 5O₂, suggesting that a higher initial concentration of dissolved oxygen increased the oxidation of phenolic compounds during aging. The presence of both phenethyl acetate and phenethyl alcohol in the same region also indicated that the ester was mainly derived from the reaction of the parent alcohol with acetic acid.

The fatty acid ester ethyl palmitate is readily hydrolyzed to palmitic acid and ethanol in an acidic environment (47). Previously we reported that the relatively high concentration of ethyl palmitate in young Huangjiu decreased throughout the aging process, while the total acid content showed the opposite trend (14, 16).

Since the volatile compounds (including unknowns) in region B were readily oxidized, dissolved oxygen had a greater influence on the aging aroma of Huangjiu than temperature.

Compared with the box-plots derived from the untargeted GC/MS data (Figure 6), all the candidate features in region C were pre-confirmed as the same component (RT 29.47–29.48 min). The identity of the unknown could not be confirmed because of the limited mass accuracy of the single quadrupole mass spectrometer used. Inspection of the inset box plots of Figure 6 shows that the abundance of some unknown compounds (i.e., nodes 79, 105, and 130) approached zero at 3 and 5O₂ in condition (II). This implied that the molecular species were easily oxidized, and the aroma compounds in region C had similar characteristics to those of region B.

As shown in Figure 7, benzaldehyde was the most abundant volatile compound, followed by an unknown and furfural. Benzaldehyde can be formed by the oxygenation of its parent alcohol, which could account for the higher concentrations of the aldehyde found in samples 3 and 5O₂ compared with 1O₂. The highest concentration of benzaldehyde and the most significant change occurred in samples at 50°C and 5O₂, respectively, which agreed with the reported oxygen and temperature dependence of benzaldehyde (and sotolon) during Port wine aging (23). The identification of benzaldehyde may also be useful in unraveling the connection between the Maillard mechanism and oxidation during the aging of Huangjiu. Temperature also had a significant effect on the formation of furfural in region D during aging.

Maillard reactions are regarded as the major pathways of furan formation. Furans are responsible for the key “caramel-like” aroma in aged wine (17), and their formations are believed to occur via Maillard reactions and oxidation (21, 23). The aldehyde substituted furan, furfural, is a common flavor compound in wine, and its formation was shown to be significantly affected by temperature during the aging of a fortified wine (21). Here the concentration of furfural increased with increasing temperature (40 and 50°C) and dissolved oxygen (3 and 5O₂). Although the effect of condition (II) was much less than that of condition (I), furfural showed a strong association between time and dissolved oxygen (r values: 0.826^**, 0.802^**, and 0.819^** at 1, 3, and 5O₂, respectively). Hence, dissolved oxygen may also play an important role in the formation of furfural.

Temperature had little effect on the aging-aromas in regions A, B, and C, but the alcohols, acids, and esters in region A were highly dependent on oxygen. Although they were not easily oxidized, their concentrations increased with increasing dissolved oxygen. In contrast, the polyphenols, lactones, ketones, and unknowns in regions B and C were easily oxidized, and their concentrations were significantly affected by dissolved oxygen. The effects of both temperature and dissolved oxygen on the formation of the key marker compounds, furfural, and benzaldehyde, in region D, confirmed the importance of these parameters during the aging of Huangjiu (15, 17). The interval and frequency of oxygenation also increased their concentrations.

The aromatic compounds (2-phenyl-2-butenal, ethyl phenylacetate, acetophenone; Figures 8A–C) showed an initial increasing trend for most conditions during the forced aging process. These compounds, together with benzaldehyde, are characteristic markers of the long-aged product (10–15 years) (14–16) and are, therefore, indicators of high-quality Huangjiu. However, the concentration of these markers all decreased after 4 or 5 weeks for samples under condition (I), and this trend increased with increasing temperature (Figures 8A–C). In the presence of oxygen, the initial increasing trend persisted until weeks 8–9 (especially 3 and 5O₂). Compared with the traditional aging in pottery jars, the decrease in the three aromatic aging-markers after 4 or 5 weeks under the condition (I) represented the largest difference. This difference could be attributed to insufficient dissolved oxygen and poor oxygen permeability in the forced aging process.

In addition, the trends for each treatment condition showed that the effect of temperature on 2-phenyl-2-butenal and ethyl phenylacetate was greater than that of dissolved oxygen (Figure 8A), and so was ethyl phenylacetate, while dissolved oxygen had a greater effect on acetophenone, phenethyl alcohol, and phenethyl acetate (Figure 5).

The overall decreasing trends in the concentrations of isoamyl acetate showed a spiked increase for the condition (I) sample at week 4, the magnitude of which decreased with increasing temperatures (Figure 8D), and this agreed with a previous investigation of Huangjiu (15). Higher initial concentrations of isoamyl acetate, and a greater decreasing trend, were observed in samples subject to increased oxygenation (2 and 3O₂), and this mirrored the behavior of 3-methlybutaol and acetic acid in region A of network analysis (Figure 4).

The highest amounts of isovaleric acid (Figure 8E) were also found in samples with the highest oxygenation treatments. This was similar to the behavior of compounds in region A of network analysis, and its effect was greater than that of temperature. Hence, isoamyl acetate and isovaleric acid, together with the volatile compounds identified in region A of network analysis (Figure 4), constitute potential markers for the forced aging process of Huangjiu.

Lactones are cyclic organic esters of the corresponding carboxylic acids, and their formations in wine are reported to be sensitive to temperature (23). Here a slightly higher concentration of γ-butyrolactone (Figure 8F) existed at 50°C compared with the other temperature treatments. This followed the behavior of γ-non-olactone (Figure 5), but the trends were not clear. The concentration of γ-butyrolactone was highest at 1O₂ (Figure 8F), compared with 3 and 5O₂, indicating that the lactone was readily oxidized during the forced aging process. Hence dissolved oxygen had a greater influence on γ-butyrolactone than temperature. However, the regulation of dissolved oxygen on a commercial scale, and optimization of γ-butyrolactone and γ-non-olactone formation, may require development.

Although the key marker compounds ethyl butyrate and ethyl 3-methylbutyrate were expected, they were only detected in some samples at 50°C, and their concentrations were very low. Consequently, their absence/low concentration suggests that the forced aging procedure may require further development/validation.