Effect of Seasoning Addition on Volatile Composition and Sensory Properties of Stewed Pork

The study aimed to investigate the influence of seasoning formulations (SP1: water; SP2: water and salt; SP3: water, salt and spices; SP4: water, salt, spices and soy sauce; SP5: water, salt, spices, soy sauce, sugar; SP6: water, salt, spices, soy sauce, sugar and cooking wine) on the volatile profiles and sensory evaluation of stewed pork. Volatile compounds were extracted using solid phase microextraction (SPME), then analysed by gas chromatography-mass spectrometry/olfactometry (GC-MS/O) and two-dimensional gas chromatographic combined with time-of-fight mass spectrometry (GC × GC-TOFMS). The results revealed that the most abundant volatile compounds, especially aldehydes, were presented in the stewed pork using SP1 and SP2. This indicated that the stewed pork with water and salt could promote lipid oxidation and amino acid degradation. As revealed by principal component analysis (PCA), the stewed pork samples with SP3 were located on the opposite side of that with SP4, SP5, and SP6 in the first and third principal component (PC1-PC3), which indicated that the overall flavour formed by adding spices was significantly different from that of adding soy sauce, sugar, and cooking wine. Sensory evaluation showed that stronger spicy, caramel, and soy sauce odour were present in samples SP3, SP4, SP5, and SP6. This study has indicated that the addition of food seasoning had a positive effect on flavour profiles of stewed pork, particularly for salt and spices.


Introduction
According to the report of State Statistical Bureau of China, pork production has increased to 54.0 million tons from 2014 to 2019, accounting for more than 60% of the meat production. Stewed pork, a representative Chinese style meat product, is appreciated by consumers in most regions of China due to its simple processing technique [1] and distinct flavour [2]. Usually, the stewed pork is often produced by stewing the fresh pork in water with various condiments and spices for a long time [3]. Owing to the differences in the dietary habits of domestic consumers, the manufacturer would add seasonings to make different flavoured stewed meat products to meet the clients' needs. Some studies found that the seasonings created an enticing aroma during stewing and removed the undesirable odour in raw meat. Qin et al. [4] reported that a total of 37 volatile compounds were identified in stewed meat broths and the main volatile compounds, such as anethole, eucalptol, linalool, terpinen-4-ol, alpha-terpineol, and cedrol, may originate from star anise. More hexadecanal, octadecanal, and 9-octadecenal were present in soy sauce-stewed pork than in those water-boiled pork, and only 3,5-dimethyl-trans-1,2,4-trithiolane was detected in soy sauce-stewed pork [5]. It can be concluded that the addition of different (Zhengzhou, Henan Province, China). All pigs were reared under the same conditions, provided with the same feed and slaughtered following routine abattoir procedures (stunned, exsanguinated, scalded, dehaired and eviscerate). The hind leg muscles from the carcasses were cut into strips (6.0 cm × 4.0 cm × 10.0 cm) after removing visible fat and connective tissues. Then these muscles were packed in low-density polyethylene bags and stored at −20 • C. The sampling procedures were approved by the Animal Care and Use Committee of the Institute of Food Science and Technology, Chinese Academy of Agricultural Sciences, and performed in accordance with animal welfare and ethics. The 2-methyl-3-heptanone (99%) and n-alkanes (C 7 -C 30 ) of chromatographic grade were bought from Sigma-Aldrich (Shanghai, China).

Seasoning Formulations and Stewed Pork
The pork strips were boiled in water for 10 min at 100 • C to remove blood, and then stewed for 45 min at 98 ± 2 • C with different seasoning formulations, finally soaked for 60 min. The flow diagram of the stewed pork is shown in Figure S1. For stewed pork, the ratio of pork strip to seasoning formulation was 1:2 (w/v). In this study, there were six seasoning formulations to which the following treatments were randomly assigned (formulation 1:2 L/kg water, formulation 2:2 L/kg water + 60 g/kg salt, formulation 3:2 L/kg water + 60 g/kg salt + one spice bag, formulation 4:2 L/kg water + 60 g/kg salt + one spice bag + 20 g/kg soy sauce, formulation 5:2 L/kg water + 60 g/kg salt + one spice bag + 20 g/kg soy sauce + 30 g/kg sugar, formulation 6:2 L/kg water + 60 g/kg salt + one spice bag + 20 g/kg soy sauce + 30 g/kg sugar + 30 g/kg cooking wine). The water and seasoning were weighed based on each kilogram of pork. The stewed pork with formulations 1, 2, 3, 4, 5, and 6 were marked stewed pork one (SP 1 ), stewed pork two (SP 2 ), stewed pork three (SP 3 ), stewed pork four (SP 4 ), stewed pork five (SP 5 ) and stewed pork six (SP 6 ), respectively. It is worth noting that SP 3 was cooked with seasoning formulation that referred to aged brine according to the following procedure [20]. The first brine (fresh brine) was obtained by removing the spice bag and stewed pork, subsequently, supplementing water, salt and spice bag into fresh brine, the second brine was obtained by the same procedures. Aged brine was eventually produced based on this cyclic process until the eighth cycle. The processing methods of seasoning formulations of SP 4 , SP 5 , and SP 6 were the same as that of SP 3 . Only difference was the composition of seasoning formulation, which contained more soy sauce, sugar and cooking wine in sequence than SP 3 . SP 1 and SP 2 were cooked with water and salt-water. Fresh pork (FP) without stewing was used as the control group. All pork samples were collected the vacuum bags and stored at −20 • C until used.

Extraction of Volatile Compounds
The extraction of volatiles from the stewed pork was carried out using the manual solid-phase micro-extraction (SPME) equipped with a 50/30 µm divinylbenzene/carboxen/ polydimethylsiloxane (DVB/CAR/PDMS) fibre (Supelco, Inc., Bellefonte, PA, USA). Briefly, Foods 2021, 10, 83 4 of 30 5.0 g of the pork sample was weighed precisely and placed in a 40 mL headspace vial. Immediately after, 1 µL of 2-methyl-3-heptanone was added and sealed tightly with screw caps fitted with a Teflon/silicon septum. The vial was incubated in a thermostatic water bath at 60 • C for 20 min. The selected fibre was used to extract the volatile compounds in head space for 40 min at 60 • C. Upon completion, the fibre was inserted into the injection port (250 • C) of the GC instrument to desorb the analyses for 5 min.

GC-MS/O Analysis
The method was performed by the method of Han et al. [21] with minor modifications. The volatile compounds of stewed pork were analysed and identified by a GC-MS instrument (7890A-7000B, Agilent Technologise, Inc., Santa Clara, CA, USA) equipped with an olfactory detection port (Sniffer 9000; Brechbuhler, Schlieren, Switzerland). Capillary column DB-wax (30 m × 0.32 mm i.d., 0.25 µm film thickness; J & W Scientific, Inc., Folsom, CA, USA) was used with helium (purity of ≥99.999%) as the carrier gas at 1.2 mL/min flow rate. The front inlet temperature was 250 • C with a solvent delay of 4 min. The temperature program was as follows: Oven temperature was maintained at 40 • C for 3 min, ramped to 200 • C at a rate of 5 • C/min, then ramped to 240 • C at a rate of 10 • C/min with a 3 min final hold. The infector mode was splitless. The transfer line temperature and ion source temperature were kept 240 • C and 230 • C. Electro-impact mass spectra were generated at 70 eV, with m/z scan range from 50 to 400 amu. A sniffing port (Sniffer 9000) coupled to a GC-MS instrument was used for odour-active compound characterization. The effluent from the capillary column was split 1:1 (v/v) between the mass spectrometry detector and the olfactory detector port. The eight trained staff were utilized for the sniffing test on GC-O. The volatile compounds were initially identified by the National Institute of Standards and Technology (NIST) Mass Spectral Library (Version 2.0). Subsequently, the identified compounds were further confirmed based on a comparison of GC retention indices (RI) with authentic compounds. Qualitative analysis was also performed according to odour properties.

GC × GC-TOFMS Analysis
The GC × GC-TOFMS system consists of an Agilent 7890 gas chromatography (Agilent Technologies, Palo Alto, CA, USA) equipped with cold-jet modulator and time-of-fight mass spectrometer (TOFMS; LECO Pegasus 4D). The first column was DB-WAX (30 m × 0.25 mm i.d. × 0.25 µm film thickness) and the second column was DB-17HT (2 m × 0.1 mm i.d. × 0.1 µm film thickness). GC × GC conditions: the temperature of the injection port was 240 • C; helium (99.999%) flow, 1.0 mL/min; splitless injection; 6.0 s of modulation period; the column temperature program for the 1st D column: initial temperature was 40 • C and held for 1 min, increased to 220 • C at 3.0 • C/min, to 230 • C at 10 • C/min and held for 5 min; the column temperature program of the 2nd D column: initial temperature at 45 • C (held for 1 min), increased to 225 • C at 3.0 • C/min (held for 2.5 min), to 230 • C at 10 • C/min (held for 5 min). TOF-MS conditions: The ion source and transfer line to the mass spectrometer were maintained at 220 • C and 290 • C, respectively. The ionization potential of MS was 70 eV, the detector voltage was 1620 V, the scan range was 33 to 450 m/z, and the mass spectra data acquisition rate was 50 spectra/s. The identification of volatile compounds was based on the NIST 2014 library (Hewlett-Packard Co., Avondale, PA, USA), the mass spectral match factor ≥ 800 and similarity ≥ 1000.

Quantification of Volatile Compounds
The volatile compounds of stewed pork were semi-quantitated by the method of calibration with an internal standard (IS). The concentration of the volatile constituent was measured by the calibration curves of the GC-peak area and the amount ratios for the target analyst relative to 2-methyl-3-heptanone. The quantitative data of the identified compounds were obtained without considering the calibration factors, that is, the calibration The odour activity value (OAV) of a compound was calculated as the ratio of its concentration in the stewed pork to its odour threshold in water. The equation is shown below: where C i is known as the concentration of the compound in the stewed pork and OT i is the odour threshold in water. Compounds with OAV ≥ 1 are considered to be the main contributors to the total flavour.

Sensory Evaluation
Sensory evaluation was carried out by 8 trained panellists (4 females and 4 males, aged 25-35 years). All assessors were recruited from Chinese Academy of Agriculture Sciences, Beijing and had at least one year of experience in the sensory descriptive analysis of stewed meat products. In order to be more familiar with the flavour characteristics of stewed pork, the 12 weeks of training sessions (2 times per week and 2 h per session) were conducted by assessors. The panellist have descripted and defined the flavour attributes, reference standard and intensities (Supplementary Table S1). Samples were coded with three-digit randomized numbers and presented to the assessors at room temperature. Panellists selected five flavour attributes to be evaluated, namely, fatty odour, meaty odour, caramel odour, soy sauce odour and spicy odour. Each attribute was scored on a 10 cm non-structured line with anchor points at each end (0 = not perceivable, 10 = strongly perceivable) [22]. The mean value of sensory attributes was shown in the radar chart.

Statistical Analysis
The contents of volatile compounds and odour activity values (OAVs) of the odouractive compounds in stewed pork were presented as the mean ± standard deviation (SD). Significant differences were determined by one-way analysis of variance (ANOVA) and Duncan's multiple range test at p < 0.05 of statistical product and service solutions (SPSS) software (v. 19.0, SPSS, Inc., Chicago, IL, USA). The PCA and PLS-DA were performed based on the mean OAVs using the software XLSTAT (2016) from Addinsoft (Barcelona, Spain). The data of odour-active compounds and sensory description in different stewed pork was also conducted with PLSR of the software XLSTAT (2016) from Addinsoft (Barcelona, Spain). The heat maps of the correlation data of PLS-DA and PLSR were conducted by R v3.2.2 (R Studio Team, 2012).

Analysis of Different Types of Volatile Components in Fresh and Stewed Pork
The kinds and content ratios of volatile components in all pork samples were shown in Table 1, the most abundant volatile compounds were aldehydes, followed by hydrocarbons, alcohols, heterocyclic and sulphur compounds, finally ketones, ethers, phenols, esters, and acids. These kinds of volatile compounds have been also found in cooked pork products [16,21]. In terms of the stewed pork with different seasoning formulations, the types and proportions of aldehydes, heterocyclic and sulphur-containing compounds increased significantly when the pork was stewed in water and salt (SP 1 and SP 2 ), while these compounds of pork samples treated with spices, soy sauce, sugar and cooking wine (SP 3 , SP 4 , SP 5 , and SP 6 ) decreased slightly. This result indicated that a large amount of flavour compounds was formed during the thermal processing of pork, however some of them were inhibited due to the addition of edible condiments. Additionally, it has been reported that the peak area and percentage composition of aldehydes increased in stewed chicken with the addition of star anise, whereas that of Maillard reaction products decreased [14]. The numbers of alcohols and ketones had gradually increased with the addition of seasoning in pork samples, and the content ratios of hydrocarbons in samples SP 3 , SP 4 , SP 5 , and SP 6 were much higher than those of samples from other treatment groups. These analyse concluded that the heat treatment and seasoning play the important role to the flavour of the cooked pork [23,24]. Furthermore, compared with compounds detected by GC-MS/O, more volatile compounds (e.g., aldehydes, alcohols, ketones, ethers, acids, heterocyclic, and sulphur-containing compounds) were identified by GC × GC-TOFMS, while some long-chain aldehydes and hydrocarbons were only detected using GC-MS/O (Supplementary Table S2). The above results have shown that the GC × GC-TOFMS combined with GC-MS/O could more comprehensively analyse the volatile profile in the fresh and stewed pork.
Ketones are often considered to have a great influence on the aroma of meat and meat products since they are presented in large amounts and exhibited specific aroma in food [32]. As shown in Table 2, 7 ketones such as 6-methylhept-5-en-2-one, 4-isopropyl-2cyclohexenone, piperitone, carvone, 4-phenylbutan-2-one, 4-methoxyphenylacetone and xanthoxylin were detected in the spice, soy sauce, sugar and cooking wine processed samples (SP 3 , SP 4 , SP 5 and SP 6 ), and they were not detected in fresh pork (FP), stewed pork with water or salt (SP 1 or SP 2 ). It indicated that more ketone compounds were formed due to the addition of food seasoning, which provided a richer fruity and nutty aroma [25] for the overall pork flavour. Pham et al. [33] reported that the methyl ketones were considered as the precursors to the fatty aromas related to cooked meat, which could be formed by the oxidative degradation of fatty acids [34]. In this study, the methyl ketones (e.g., heptan-2-one, octan-2-one, pentane-2,3-dione, octane-2,3-dione, 3-octen-2-one, (E,E)-3,5octadien-2-one, 6-methyl-3,5-heptadiene-2-one and acetophenone) were originated from the oxidation of lipids during heating.
There were few hydrocarbons detected in the fresh pork, however more hydrocarbons could be formed by added water, salt, soy sauce, sugar and cooking wine during the pork processing. All hydrocarbons could be divided into aromatic hydrocarbons and aliphatic hydrocarbons. Among them, 16 aromatic hydrocarbons had been identified in all stewed pork samples, and toluene, ethylbenzene, 1,3-dimethylbenzene, o-xylene and styrene were presented as common compounds. The production of toluene and ethylbenzene primarily come from amino acid degradation. This result was consistent with that reported by Olivares et al. [11]. It was also found that nine aliphatic hydrocarbons were detected in stewed pork samples (SP 1 , SP 2 , SP 3 , SP 4 , SP 5 and SP 6 ). Previous study has shown that aliphatic hydrocarbons had a limited influence on aroma perception due to their high threshold values [35] and raised mainly from lipid oxidation [27]. Additionally, phellandrene, (Z)-3,7dimethyl-1,3,6-octatriene, o-cymene, m-cymene, 1-methylindan, and longifolene could be formed from the added spices and soy sauce during the processing.               Ester compounds could be formed by the esterification of alcohols and carboxylic acids in the meat products [2]. The contribution of esters to the aroma of pork products depends on the length of their chain [27]. A total of 13 esters were identified in all stewed pork samples, where short-chain esters, such as ethyl acetate and ethenyl acetate, had fruity notes. While long-chain esters like isoamyl isobutyrate, hexyl butanoate, hexyl acetate and hexyl butanoate possessed a slight fatty odour [36]. In addition, when the salt was added, the relative content of esters was significantly increased (p < 0.05). This reason might be that the salt of meat products favoured the formation of ester compounds. For ether and phenol compounds, the anethole was detected in all pork samples and the methyleugenol, elemicin, myristicin, eugenol and trans-isoeugenol were found in SP 3 , SP 4 , SP 5 and SP 6 . Moreover, the ethers and phenols except for phenol were from the spices, and the acids (butanoic acid, pentanoic acid, and octanoic acid) come from the fresh pork.
Heterocyclic and sulphur-containing compounds are the important contributions to the formation of flavour in meat products [37]. As shown in Table 2, 17 heterocyclic compounds (furans, pyridines and pyrroles) and sulphur-containing compounds were detected in all pork samples. The 2-pentylfuran and safrole were derived from linoleic acid autoxidation [38] and spices (nutmeg, anise and ginger), respectively. The 2-pentylfuran is often used as an important indicator of the degree of oxidation of meat product. The contents of 2-pentylfuran in SP 1 and SP 2 were significantly higher (p < 0.001) than that in SP 3 , SP 4 , SP 5 , and SP 6 , indicating that the stewed pork with only water and salt had a greater effect on lipid oxidation. It has been reported that a large number of furans, pyridines, pyrroles and sulphur-containing compounds could be produced by Maillard reaction and amino acid degradation during cooking [15,23]. In our study, 3-(4-methyl-3-pentenyl)-furan, furfural, 2-furanmethanol, pyridine, 2-acetylpyrrole and dimethyl trisulfide displayed significantly higher levels (p < 0.01) in SP 5 and SP 6 than those in other groups, which indicated that the addition of sugar and cooking wine could promote the Maillard reaction. For sulphur-containing compounds, methanethiol and dimethyl disulphide from sulphurcontaining amino acid degradation were significantly lower (p < 0.001) in SP 1 than that in SP 2 , which indicated that salt-treated stewed pork was more conducive to the production of sulphur-containing compounds. This result was consistent with that reported by Liu et al. [39] who found that the levels of sulphur-containing compounds in Nanjing water-boiled salted duck were markedly higher than those in control samples. Regarding 2-acetylthiazole, 2-thiophenecarboxaldehyde and benzothiazole originated from Maillard reaction, and they contributed to roasted, caramel, and meaty notes [40] for the overall aroma of stewed pork.
The concentration of volatile compounds according to possible origins of the fresh and stewed pork are presented in Figure 1. It was found that the lipid oxidation, aged brine and amino acid degradation were the important origins of volatile compounds in all stewed pork attributed to their contribution to more aroma of the pork samples. For the lipid oxidation and amino acid degradation, their concentrations were highest in SP 1 and SP 2 , followed by SP 3 , SP 4 , SP 5 and SP 6 , finally FP. This indicated that heat-treated pork with water and salt would facilitate lipid oxidation and amino acid degradation to produce more volatiles, while there was an inhibitory effect on heat-treated pork with aged brine, especially for spices. Compared with volatile compounds from the aged brine in SP 3 , SP 4 , SP 5 and SP 6 , they were significantly higher (p < 0.05) than those in SP 1 and SP 2 , this may be due to the addition of food condiments (spices, soy sauce, sugar, and cooking wine) in stewed pork.

Odour-Active Compounds Analysis of the Fresh and Stewed Pork
To evaluate the contributions of volatile compounds to overall flavour of the fresh and stewed pork, the OAVs of these compounds were determined by dividing the concentration of the compound by its odour threshold in water. As can be seen from Table 3, a total of 29 odour-active compounds with OAVs greater than 1 were selected from 139 volatile compounds, including 14 aldehydes, four alcohols, three ketones, one hydrocarbon, one ester, two ethers, one phenol, thre furans, N-or S-containing compounds. Seven of them with relatively high OAVs were detected in all stewed pork samples: hexanal (OAV at 44.1-158.5), heptanal (OAV at 10.5-53.4), octanal (OAV at 72.2-329.9), nonanal (OAV at 110.3-1475.5), oct-1-en-3-ol (OAV at 13.9-68.5), 2-pentylfuran (OAV at 39.0-178.6) and methanethiol (OAV at 31.3-80.1). These compounds were known as the key odour-active compounds due to their significant contributions to the integral flavour. Furthermore, it was found that the OAVs of hexanal, heptanal, octanal, nonanal, and decanal increased significantly (p < 0.01) in SP 1 and SP 2 . Statistical analysis showed that the total OAVs of odour-active compounds of SP1 and SP2 were significantly higher (p < 0.001) than those of FP and SP3, SP4, SP5 and SP6. Linear aldehydes like pentanal, hexanal, heptanal, octanal, nonanal and decanal have been reported to be generated from lipid oxidation [27]. Moreover, these aldehyde compounds could be detected in different processing methods and may contribute grassy, fatty and fruity notes to overall aroma of the pork samples. Unsaturated aldehydes such as (Z)-hept-2-enal, (E)-oct-2-enal, (E)-non-2-enal, (E)-dec-2-enal, (E,E)-2,4-nonadienal, undec-2-enal, and (E,E)-2,4-decadienal are degradation products of linoleate and linolenate hydroperoxides [41]. Among them, there was no significant difference in undecen-2-al in FP, SP1, and SP2 (p > 0.05), indicating that heating and salt treatment had no effect on the formation of undecen-2-al. On the other hand, the rest of the olefin aldehydes have relatively higher OAVs in SP1 and SP2. This showed that SP1 and SP2 could promote the increase of some unsaturated aldehydes. Benzeneacetaldehyde, with honey and sweet notes, is a well-known aroma component formed from Maillard reaction of phenylalanine [42] and the OAV in SP5 were significantly higher (p < 0.001) than that in other samples. , boiled pork with water; SP 2 , cooked pork with water and salt; SP 3 , stewed pork with water, salt and spices; SP 4 , stewed pork with water, salt, spices and soy sauce; SP 5 , stewed pork with water, salt, spices, soy sauce and sugar; SP 6 , stewed pork with water, salt, spices, soy sauce, sugar and cooking wine. LOP, Lipid oxidation products; AADP, Amino acid degradation products; MRP, Maillard reaction products; VFAB, Volatiles from aged brine; VFRM, Volatiles from raw meat; UO, Unknown origin.  Statistical analysis showed that the total OAVs of odour-active compounds of SP 1 and SP 2 were significantly higher (p < 0.001) than those of FP and SP 3 , SP 4 , SP 5 and SP 6 . Linear aldehydes like pentanal, hexanal, heptanal, octanal, nonanal and decanal have been reported to be generated from lipid oxidation [27]. Moreover, these aldehyde compounds could be detected in different processing methods and may contribute grassy, fatty and fruity notes to overall aroma of the pork samples. Unsaturated aldehydes such as (Z)-hept-2-enal, (E)-oct-2-enal, (E)-non-2-enal, (E)-dec-2-enal, (E,E)-2,4-nonadienal, undec-2-enal, and (E,E)-2,4-decadienal are degradation products of linoleate and linolenate hydroperoxides [41]. Among them, there was no significant difference in undecen-2-al in FP, SP1, and SP2 (p > 0.05), indicating that heating and salt treatment had no effect on the formation of undecen-2-al. On the other hand, the rest of the olefin aldehydes have relatively higher OAVs in SP 1 and SP 2 . This showed that SP 1 and SP 2 could promote the increase of some unsaturated aldehydes. Benzeneacetaldehyde, with honey and sweet notes, is a well-known aroma component formed from Maillard reaction of phenylalanine [42] and the OAV in SP 5 were significantly higher (p < 0.001) than that in other samples. 1,8-Cineole, anethole and estragole, with mint and aniseed flavour, were the most abundant in SP 3 . The OAVs of oct-1-en-3-ol, (E)-oct-2-en-1-ol, butane-2,3-dione, octane-2,3-dione and 2-pentylfuran were the highest in SP 1 and SP 2 , followed by SP 3 , SP 4, SP 5 and SP 6 , of which oct-1-en-3-ol and octane-2,3-dione was shown to be richer in SP 2 than that in SP 1 . Linalool and eugenol were found immediately when the spices were added to the cooked pork, which may be due to the flavour of the star anise itself. Dimethyl trisulfide, with fish and cabbage notes, was considered as the main sulphur-compound in SP 1 , SP 4 , SP 5 , and SP 6 .

PCA and PLS-DA Analysis of Odour-Active Compounds
In order to clarify the differences in aroma profile of the fresh and stewed pork, a principal component analysis (PCA) was performed and showed in Figure 2a The first three principal components (PCs) accounted for 91.24% of the total variance and were enough to explain the maximum variation in all original data of the pork samples. As can be seen in Figure 2a, the PC1 and PC2 showed a clear-cut separation of the samples into three major groups. Among them, the sample dot of FP was located in the fourth quadrant, and sample dots representing SP 3 , SP 4 , SP 5 , and SP 6 were located in the second quadrant and could be recognized as one cluster, while sample dots of SP 1 and SP 2 located in the first and fourth quadrant were considered as a group because of their relatively closer distance. As shown in Figure 2b, the fresh and stewed pork samples were divided into four groups in PC1 and PC3, that is, group I: FP; group II: SP 1 and SP 2 ; group III: SP 3 ; group IV: SP 4 , SP 5, and SP 6 . The four group of sample points were located in the different quadrants indicating that the overall aroma of each group of samples were different. Moreover, SP 4 , SP 5 , and SP 6 samples was close and located in the third quadrant. It can be concluded that the overall flavour of SP 4, SP 5 , and SP 6 was similar each other. Similarly, so was SP 1 and SP 2 . It can be also found that the flavour of samples SP 4, SP 5 and SP 6 was significantly different from that of sample SP 3 in Figure 2a,b. This showed that the flavour compounds of stewed pork were also affected by soy sauce, sugar, and cooking wine.  Apart from PCA, the supervised PLS-DA was performed to evaluated the differences of volatile compounds of the stewed pork. As shown in Figure 3a, except for SP 3 , SP 4, SP 5 and SP 6 , only the separation was observed for FP, SP 1 , and SP 2 (R 2 X = 0.968, R 2 Y = 0.818 and Q 2 = 0.628). SP 3 , SP 4, SP 5 , and SP 6 were located on the negative side of axis 1, whereas FP, SP 1 , and SP 2 were founds the positive side of axis1, SP 1 and SP 2 were close each other. Obviously, the different stewed pork samples were separated into three group (SP 3 -SP 4 -SP 5 -SP 6 , SP 1 -SP 2 , FP). It could also be concluded that the overall flavour of SP 3 -SP 4 -SP 5 -SP 6 , SP 1 -SP 2 and FP were greatly different, and each group samples possessed the similar flavour profiles. The result is consistent with the PCA analysis (Figure 2a). In addition, to identify the most discriminative volatiles contributing to the observed the fresh and stewed pork samples, variable identification (VID) coefficients were calculated (Figure 3b). Volatiles with VID ≥ |0.80| discriminating FP, SP 1 , SP 2 and SP 3 were predominantly aldehyde compounds (pentanal, (Z)-hept-2-enal, nonanal and (E)-oct-2-enal) and related volatiles (linalool, octane-2,3-dione, ethyl acetate, anethole and estragole). Moreover, no compound with VID ≥ |0.80| was found discriminating SP 4 , SP 5 , and SP 6 . As shown in Figure 3a,b, it can be observed that pentanal, 1-hydroxypropan-2-one, D-limonene, ethyl acetate and methanethiol were not only on the opposite side of samples FP, but also strongly and negatively correlated with it (−0.83 ≤ r ≤ −0.69), while the remaining 24 odour-active compounds (−0.57 ≤ r ≤ 0.46) had a low correlation with it. Most odouractive compounds, such as hexanal, heptanal, octanal, (Z)-hept-2-enal, nonanal, (E)-oct-2-enal, decanal, (E)-non-2-enal, (E)-dec-2-enal, (E,E)-2,4-nonadienal, (E,E)-2,4-decedienal, oct-1-en-3-ol, (E)-oct-2-en-1-ol, octane-2,3-dione, and 2-pentylfuran, were close to SP 1 and SP 2 on the right side of t1, and the strong and positive correlation (0.60 ≤ r ≤ 0.97) were showed. Moreover, SP 2 induced an increase in the correlation coefficients of heptanal, octanal, (Z)-hept-2-enal, (E)-non-2-enal, oct-1-en-3-ol and octane-2,3-dione indicating that the addition of salt during the processing of stewed pork was beneficial to the formation of these compounds. For SP 3 , SP 4, SP 5 and SP 6 , benzeneacetaldehyde, 1,8-cineole, linalool, Dlimonene, anethole, estragole and eugenol had the high and positive correlation coefficients (0.63 ≤ r ≤ 0.99) on the left side of t1, only butane-2,3-dione had the high and negative correlation coefficient (r = −0.76). Because 1,8-cineole, linalool, anethole, and estragole had a higher correlation coefficient (0.79 ≤ r ≤ 0.99) with SP 3 and a lower correlation coefficient (−0.15 ≤ r ≤ 0.21) with SP 4, SP 5 , and SP 6 , this suggests that they may be potential flavour markers to distinguish SP 3 and SP 4, SP 5 and SP 6 .

Descriptive Sensory Analysis
To describe the differences of odour profiles of seven pork samples, the flavour sensory evaluation was performed using the five representative descriptors, namely, "meaty", "spicy", "caramel", "soy sauce" and "fatty". As can be seen from Figure 4. Significant differences (p ≤ 0.01) were found among five odour attributes of the pork samples. The intensities of fatty notes in SP 1 and SP 2 were highest, followed by SP 3 , SP 4, SP 5 , SP 6 and FP. As described by the panellists from GC-MS/O, hexanal, heptanal, octanal, (Z)-hept-2-enal, nonanal, (E)-oct-2-enal, decanal, (E)-non-2-enal, (E)-dec-2-enal, (E,E)-2,4-nonadienal, (E,E)-2,4-decedienal and 2-pentylfuran might be closely related to the fatty odour. This result agreed with the result of Figure 3b. The meaty and caramel odour of SP 5 and SP 6 had the highest score in all samples, which could be mainly attributed to furans and N-containing compounds such as 3-(4-methyl-3-pentenyl)-furan, furfural, 2-furanmethanol, pyridine, and 2-acetylpryrrole. In addition, the strong spicy and soy sauce smell was presented in SP 3 and SP 5 , respectively. Heat map of the correlations between volatile compounds and the pork samples. FP, fresh pork; SP1, stewed pork with water; SP2: stewed pork with water and salt; SP3: stewed pork with water, salt and spices; SP4: stewed pork with water, salt, spices and soy sauce; SP5: stewed pork with water, salt, spices, soy sauce and sugar; SP6: stewed pork with water, salt, spices, soy sauce, sugar and cooking wine. The blue circle dots represent the fresh and stewed pork and the red triangle dots represent the odour-active compounds.

Relationship between Sensory Evaluation and Odour-Active Compounds
PLSR was employed to establish the relationship between the five sensory descriptors of the fresh and stewed pork and the odour-active compounds analysed by GC-MS/O and GC × GC-TOFMS, and the correlation coefficient between them was expressed in the heat map. As shown in Figure 5a, most of the X-matrix (contribution ratios of the odour-active compounds) and Y-matrix (intensities of the sensory attributes) are loaded around the circle (r 2 = 100%, r 2 = represent the degree of interpretation). The model quality (Q 2 = 0.846) ≥ 0.50 indicated that they were well explained by the PLSR model. The first two components explained 74.0% of X-matrix and 92.7% of Y-matrix. The dots corresponding to sample SP 1 and SP 2 had overlap in the second quadrant, and the samples points of SP 3 , SP 4 , SP 5 and SP 6 were close in the fourth quadrant, and as well as FP was found in the third quadrant. So, the fresh and stewed pork samples can be divided into three group and this result was consistent with previous PCA plots (Figure 2a). According to Figure 5a,b, it can be observed that SP 3 , SP 4 , SP 5 , and SP 6 were characterized by soy sauce, caramel and spicy odour because of their short distance, and the three aroma attributes aforementioned were positively correlated with pentanal, benzeneacetaldehyde, 1,8-cineole, 1-hydroxypropan-2-one, D-limonene, ethyl acetate, eugenol, methanethiol, and dimethyl trisulfide with a high correlation coefficient (0.60 ≤ r ≤ 0.92). On the contrary, the soy sauce, caramel and spicy notes located in the right side of the loading plot were strongly and negatively correlated with some aldehydes (heptanal, nonanal, (E)-oct-2-enal, (E)-non-2-enal, (E)-dec-2-enal, (E,E)-2,4-nonadienal, undec-2-enal and (E,E)-2,4-nonadienal) and unsaturated alcohols like (E)-oct-2-en-1-ol. SP 1 and SP 2 on the upper left side of loading plot were mainly descriptive fatty and meaty odour, which were in accordance with the descriptive sensory analysis. These two attributes were highly associated with hexanal, oct-1-en-3-ol and 2-pentylfuruan. Moreover, FP was located far from these flavour attributions and most volatiles, which indicated that there was not the unique flavour of the fresh pork.