Characterization of the effect of different cooking methods on volatile compounds in fish cakes using a combination of GC–MS and GC-IMS

Graphical abstract

Highlights

• •Screening of key flavour substances according to OAV and ROAV.
• •Fifteen characteristic flavour compounds were screened according to VIP ＞1.
• •GC–MS and GC-IMS retained overall information on the volatile constituents of fish cakes from different cooking methods.

Abstract

In this study, gas chromatography-mass spectrometry (GC–MS) and gas chromatography-ion mobility spectrometry (GC-IMS) were used to analyze the volatiles of fish cakes obtained using five cooking methods, namely, steaming, baking, air frying, pan frying and deep frying. The odor activity value (OAV) and relative odor activity value (ROAV) were used to screen for the major aroma compounds. Orthogonal partial least squares discriminant analysis (OPLS-DA) and the variable influence on projection (VIP) were used to determine the characteristic flavor compounds in the fish cakes. A total of 72 volatile compounds were identified by GC–MS, and 41 volatile compounds were detected by GC-IMS. 3-ethyl-2,5-dimethylpyrazine and 2,5-dimethylpyrazine were not detected in either CK or SS. The OPLS-DA models for GC–MS and GC-IMS analyses were constructed based on VIP values, and 8 and 7 compounds, respectively, were screened as characteristic aroma compounds. The results of this study provide new insights into the dynamics of flavor formation in reheated fish cakes and provide a theoretical basis for the optimization of the production process of this food product.

1. Introduction

The accelerated pace of modern life has led to increasing demand for prepared dishes, as well as for increased variety of dishes. Aquatic dishes are aquatic products that can be made into finished dishes through simple cooking; these products include seafood and aquatic dishes, aquatic prepared food and deep-processed products, and other series of products. As an aquatic prepared food, fish cakes simplify the cooking steps of aquatic dishes and are highly convenient for use in daily life. To date, there have been few studies on the cooking methods of these materials. Therefore, flavor variations due to different cooking methods may be an interesting area of fishcake research.

Cooking is a heating process that results in varying degrees of changes in the composition, physicochemical indices and product flavor of food ingredients. However, cooking has a significant impact on the physicochemical properties and eating quality of meat, altering its nutritional and organoleptic qualities(Li et al., 2020, Roldan et al., 2015). Chen et al., 2021, Chen et al., 2021 compared tilapia meat prepared by microwaving, baking, steaming and boiling and found that the four preparation modes had significantly different effects on the meta flavor. Xiong et al.(2023) compared the effects of four cooking methods, namely, steaming, boiling, microwaving and baking, on the nutritional value, taste and eating quality of golden pomfret fish and found that steaming and microwaving were more conducive to nutritional preservation and improved human health. The flavor components of meat products are generally generated by the Maillard reaction, lipid degradation reaction and some vitamin degradation reactions(Zeng et al., 2020), and it is known that volatile compounds can significantly affect food flavor(J. Chen et al., 2021, Chen et al., 2021). Currently, volatile compounds in food are analyzed using both sensory and instrumental analysis, the latter of which can be applied to studies at the molecular level. GC–MS is an effective technique for the separation and identification of complex volatile compounds in qualitative and quantitative detection of volatile components in food and has the advantages of high resolution and high sensitivity(Mei and Chen, 2023, Song and Liu, 2018). GC-IMS, an emerging technique for food flavor analysis, has the advantages of high separation via gas chromatography and high sensitivity and low cost (Duan et al., 2023, Wang et al., 2020). It has been investigated in drug detection and environmental pollution monitoring(Wang et al., 2020, Yang et al., 2019). In recent years, the combined use of several instruments has become popular for both isolating and identifying a wider range of volatile compounds and providing a more comprehensive and diverse flavor profile of foods(Chen et al., 2021, Chen et al., 2021, Yang et al., 2022). A combination of GC-IMS and GC–MS has been reported to characterize the dynamics of volatiles during perch fermentation,The results showed that GC-IMS and GC–MS complement each other in validating the analytical results, thus differentiating the samples in a more comprehensive and effective manner(Nie et al., 2022). However, there are no reports on the combined use of these instruments for fish cakes.

The aim of this study was to apply GC–MS and GC-IMS to identify and quantify the volatile compounds in fish cakes cooked via different methods (steaming, baking, air frying, pan frying and deep frying). The major flavor compounds were screened based on their OAV and ROAV, OPLS-DA was used to construct a model to identify the characteristic flavor compounds among the major flavor compounds of fish cakes based on the VIP. The results of the present study provide reference information for thermal processing technology and industrial production of fish cakes. In addition, this approach can help promote the diversification of deep processing of aquatic products.

2. Materials and methods

2.1. Materials and reagents

Tilapia were sourced from the Walking Street Vegetable Market, Xiashan District, Zhanjiang city; yam, corn, carrot, salt and white pepper were sourced from the Walmart Supermarket, Zhanjiang city.

2.4.6-Trimethylpyridine (chromatographically pure, ≥99.9 % purity) was purchased from Macklin Biochemical Co., Ltd. (Shanghai, China). The mixed standard of n-alkanes (C₅–C₃₂) was purchased from Anpel Experimental Technology Co., Ltd. (Shanghai, China).

2.2. Sample preparation

2.2.1. Preparation of fish cakes

The fresh tilapia were slaughtered; rinsed after removing the head, scales and viscera; and deodorized with yeast patchouli compound solution. The deodorized fish meat was ground, and yam puree, glutinous rice starch, pork and other additives were added; then, the meat was seasoned. The seasoned meat was placed in a mold with a diameter of 7 cm and a thickness of 1 cm to form a cake. The cake was then set by a two-step shaping method, i.e., water was heated in a water bath at 50 ℃ for 30 min and then placed in a constant temperature box at 40 ℃ for 30 min.

2.2.2. Sample processing

The samples were divided into six groups. The samples without cooking treatment were used as controls (CK); for steam cooking, the samples were placed on a round ceramic plate that was placed in a multifunctional rice cooker (WFS-5017TM, Midea Group) and processed for 6 min (SS); the fish cakes were deep-fried in a pan with preheated soybean oil (180 °C) for 90 s (DF); the fish cakes were deep fried in a pan with preheated soybean oil (180 °C) for 5 min (PF); the fish cakes were placed in an air fryer (KZE5004, Midea Group) at 200 °C for 8 min (AF); and the fish cakes were placed in an oven (MG38-CB-AA, Midea Group) at 200 °C for 10 min (BS).

2.3. GC–MS analysis of volatile compounds

The fish cake volatiles were analyzed via solid-phase microextraction. The method of Zhang et al.(2020) was used with slight modifications. Briefly, the stranded sample (5.00 g) was accurately weighed into a 40 mL headspace bottle, a 2,4,6-trimethylpyridine standard solution (2.00 μL) was added, and the bottle was capped. The DVB/CAR/PDMS fibre was inserted into the headspace bottle, and the bottle was placed in a thermostatic water bath at 50 ℃ for 30 min to extract the sample. Immediately after extraction, the needle was transferred to the GC inlet port, and the sample was desorbed at 250 ℃ for 5 min while the instrument was activated for acquisition.

GC–MS analysis was performed using a TQ8050NX gas chromatography mass spectrometer (Shimadzu, Kyoto, Japan) with an InertCap® Pure-WAX quartz capillary column (30 m × 0.25 mm, 0.25 μm); the carrier gas was He (99.999 % purity) at a flow rate of 1.0 mL/min; and the flow rate was 1.0 mL/min. The column was initially heated to 40 ℃ for 3 min, then to 100 ℃ at 4 ℃/min for 2 min, and then to 230 ℃ at 8 ℃/min for 5 min. The electron ionization energy was 70 eV, the interface temperature was 250 ℃, the ion source temperature was 230 ℃, the mass scanning range was 33–550 m/z, and the acquisition mode was Q3.

2.4. Characterization and quantification of the volatile compounds

Based on the total ion flow diagram of the volatile compounds, the first 200 peaks with the largest peak areas were integrated, and a similarity search was performed using the NIST05 and Wiley07 databases and combined with the RI of the compounds for characterization; the RI of the volatile compounds was calculated based on the peak times of the n-alkanes under the same analytical conditions for the same samples, and the content of the volatile compounds was determined by the internal standard method. The OAV is the ratio of the concentration of an aroma component in the aroma system (C_i) to its odor threshold (T) and was calculated as follows:

$$\mathit{RI}=100\times \left(\frac{t_x-t_n}{t_{n+1}-t_n}+n\right)$$ (1)

$$C_{\text{i}}=\frac{A_{\text{i}}}{A_{\text{s}}\times m_i}\times m_s$$ (2)

$$\mathit{OAV}=\frac{C_i}{T}$$ (3)

where (1) t_x, t_n and t_n+1 are the retention times of the volatile compounds to be measured; n-alkanes have n carbon atoms and n + 1 carbon atom n-alkanes, respectively, for which t_x < t_n < t_n+1; (2) C_i is the concentration of the volatile compounds (μg/kg); A_i and A_s are the peak areas of volatile Compounds i and that of the internal standard compounds, respectively; m_i is the mass of the samples (g); and m_s is the mass of the internal standard compounds (mg); and (3) T is the organoleptic threshold of the volatile compound.

2.5. GC-IMS analysis of volatile compounds

The volatile fingerprints of the samples were analyzed via GC-IMS (FlavorSpec®, G.A.S. Dortmund, Germany). The method of Zheng et al.(Zheng et al., 2022, Sun et al., 2022) was used with slight modifications. Briefly, the sample (2.00 g) was weighed into a headspace vial and incubated for 10 min at 60 °C at 500 rpm, after which the headspace sample (500 μL) was injected into a headspace autosampler at 80 °C. The carrier gas was programmed at the following flow rate: 2 mL/min for 0–2 min, increased to 100 mL/min from 2 min to 25 min and maintained for 30 min. The drift tube (5.3 cm) was operated at 45 °C and a constant temperature of 150 mL/min. The RI of the volatile compounds were calculated under the same analytical conditions using C₄–C₉ ortho-ketone (Sinopharm Chemical Reagent Beijing Co., Ltd., Beijing, China). The RI method was used for the determination of volatile compounds based on the data available in the IMS database.

2.6. Evaluation of flavorings

Based on the relative content, the method of Xu et al.(2023) was used with slight modifications. In this case, ROAV_stan = 100 for the most important component, and the remaining components were calculated according to Equation (4):

$$\mathit{ROAV}\approx \frac{C_i}{T}\times \frac{T_{\text{stan}}}{C_{\text{stan}}}\times 100$$ (4)

where C_i is the relative content of the compound (%); T_i is the aroma threshold of the compound in water (μg/kg); and C_stan and T_stan are the relative content and aroma threshold of the major components, respectively.

2.7. Statistical analysis

All of the samples were tested three times in parallel, and the results are expressed as the mean ± SD. SPSS 26 software (IBM, Armonk, NY, USA) was used to analyze the experimental data, and one-way ANOVA was used to test for significant differences (p < 0.05, significant difference). OPLS-DA was performed using the SIMCA-P 14.1 software (Umetrics, Umea, Sweden), and the Origin 2024 software (Origin Lab, Inc., Northampton, MA, USA) was used for graphing.

3. Results and discussion

3.1. GC–MS analysis

3.1.1. GC–MS analysis of volatile components produced by different cooking methods

To understand the effect of different cooking methods on the flavor compounds of fish cakes, GC–MS was used to analyze the volatile compounds in the samples after different cooking methods. As shown in Table 1, a total of 72 substances were identified by GC–MS.; these components were divided into 9 categories: 15 aldehydes, 11 alcohols, 7 ketones, 5 esters, 5 phenols, 13 hydrocarbons, 5 acids, 5 pyrazines and 9 others. The numbers of volatile components identified in the samples obtained using different reheating methods were as follows: 31 (CK), 46 (SS), 53 (BS), 54 (AF), 46 (DF) and 48 (PF). Compared to CK, the number of compound species increased significantly after the use of the different cooking methods, with the highest number of aldehydes in AF and DF, the highest number of hydrocarbons in SS and AF, and pyrazines most abundant in PF. The components with higher proportions were aldehydes, hydrocarbons, alcohols and phenols. In addition, the content of the volatile component classes varied between the cooking methods (Fig. 1). Aldehydes (57.25 ± 3.65 μg/kg) and alcohols (41.16 ± 6.00 μg/kg) were most abundant in the air-frying method; hydrocarbons (45.97 ± 6.00 μg/kg) were most abundant in the steaming method; and pyrazines did not appear in either the control or the steaming method. There were significant differences in the types and contents of volatile compounds in the samples obtained using different reheating methods, which may be due to the differences in the heat transfer mechanisms, heating times and temperatures used for the different cooking conditions(Sun et al., 2022, Zheng et al., 2022).

Table 1.

Information on volatile compounds identified by GC–MS.

Compounds Name	RI	concentrations（μg/kg）
		CK	SS	BS	AF	DF	PF
Aldehydes
nonanal	1094.43	31.44 ± 6.4c	31.81 ± 3.09c	31.62 ± 2.96c	39.74 ± 0.62a	34.65 ± 5.02b	28.01 ± 0.39d
decanal	1129.35	3.86 ± 1.29a	2.39 ± 0.34b	3.19 ± 0.44ab	3.61 ± 0.27a	1.49 ± 0.19c	3.88 ± 0.92a
4-isopropylbenzaldehyde	1191.27	ND	2.18 ± 1.33a	1.06 ± 0.20bc	1.66 ± 0.42b	0.88 ± 0.13c	1.33 ± 0.06b
tetradecanal	1432.99	ND	1.74 ± 0.98a	1.12 ± 0.08ab	1.15 ± 0.24ab	0.62 ± 0.35b
piperonal	1469.29	ND	2.82 ± 1.22a	1.89 ± 0.61ab	1.62 ± 0.25b	1.13 ± 0.37c	1.61 ± 0.32b
octanal	1018.78	0.72 ± 0.02c	1.12 ± 0.15c	3.91 ± 1.17a	3.56 ± 0.93a	3.05 ± 0.32b	3.19 ± 0.26b
pentadecanal	1529.93	ND	ND	0.56 ± 0.17ab	0.39 ± 0.19b	ND	0.86 ± 0.10a
benzaldehyde	1129.80	ND	ND	2.73 ± 0.48a	1.99 ± 0.22ab	1.02 ± 0.37b	1.12 ± 0.01b
4-methoxybenzaldehyde	1410.89	1.35 ± 0.05b	2.14 ± 0.38a	ND	1.64 ± 0.18ab	1.36 ± 0.61b	1.72 ± 0.48ab
cinnamic aldehyde	1308.49	ND	0.83 ± 0.31a	ND	ND	ND	ND
(2E)-2-decenal	1168.16	0.38 ± 0.15c	0.71 ± 0.07b	0.75 ± 0.09b	1.17 ± 0.19a	0.94 ± 0.11ab	0.91 ± 0.15ab
(2E,4E)-2,4-decadienal	1001.98		0.73 ± 0.12b	0.79 ± 0.27b	0.72 ± 0.14b	0.87 ± 0.46a	0.83 ± 0.17a
4-ethylbenzaldehyde	1225.45	ND	ND	ND	ND	0.40 ± 0.02a	ND
1-hexadecanal	1600.20	ND	ND	ND	ND	0.40 ± 0.15a	ND
lauryl aldehyde	1253.42	ND	ND	ND	ND	ND	2.00 ± 0.57a
Total content		37.03 ± 7.89c	45.35 ± 7.84b	47.62 ± 6.47b	57.25 ± 3.65a	43.76 ± 7.78bc	42.27 ± 3.17bc
Alcohols
1-octen-3-ol	1135.68	ND	1.82 ± 0.29c	2.58 ± 0.41b	3.39 ± 0.51a	1.76 ± 0.23c	1.66 ± 0.13c
linalool	1144.04	7.52 ± 0.79b	14.53 ± 2.71ab	16.85 ± 1.42a	13.38 ± 0.62ab	7.77 ± 0.58b	12.37 ± 0.01ab
1-octanol	1220.68	ND	1.66 ± 0.44c	1.05 ± 0.86c	3.29 ± 0.52a	2.38 ± 0.05b	2.58 ± 0.08b
terpineol	1178.19	7.96 ± 2.03a	7.01 ± 0.85a	6.73 ± 1.07b	7.46 ± 1.95a	4.14 ± 0.67c	6.72 ± 1.78b
isospathulenol	1522.73	ND	1.16 ± 0.63a	ND	ND	ND	ND
geraniol	1205.75	ND	0.62 ± 0.34a	ND	ND	ND	ND
2-(4-methylphenyl)propan-2-ol	1146.94	ND	2.34 ± 0.95a	ND	0.65 ± 0.10b	ND	ND
2-(4-methylcyclohexa-2,4-dien-1-yl)propan-2-ol	1182.34	ND	0.55 ± 0.05d	0.30 ± 0.35d	2.72 ± 0.48b	1.56 ± 0.26c	6.30 ± 1.06a
(-)-terpinen-4-ol	1157.15	7.89 ± 1.43a	ND	6.16 ± 1.13ab	7.19 ± 1.68a	3.60 ± 0.62b	6.09 ± 0.80ab
2-(3-methylphenyl)propan-2-ol	1203.51	ND	ND	3.82 ± 0.76a	3.08 ± 0.14a	ND	ND
geranylgeraniol	2008.50	ND	ND	ND	ND	ND	1.73 ± 0.23a
Total content		23.37 ± 4.25d	29.69 ± 6.26c	37.49 ± 6.00b	41.16 ± 6.00a	21.21 ± 2.41d	37.45 ± 4.09b
Ketones
(-)-carvone	1183.18	ND	0.97 ± 0.01a	0.80 ± 0.17ab	0.99 ± 0.18a	ND	0.67 ± 0.09b
6,10-dimethyl-5,9-undecadien-2-one	1347.04	1.26 ± 1.31ab	1.04 ± 0.23ab	1.36 ± 0.49ab	2.06 ± 0.34a	2.12 ± 1.04a	0.96 ± 0.42b
4-methoxyphenylacetone	1241.06	2.47 ± 0.89a	1.20 ± 0.41b	0.78 ± 0.43c	ND	0.66 ± 0.19c	2.01 ± 0.66a
xanthoxylin	1286.29	2.57 ± 1.90a	2.07 ± 0.22a	1.60 ± 0.44b	1.79 ± 0.33ab	1.64 ± 0.57b	1.71 ± 0.39ab
6-methylhept-5-en-2-one	1047.36	0.70 ± 0.33c	2.20 ± 1.77a	1.69 ± 0.41ab	1.05 ± 0.51b	ND	ND
5-methyl-2-(1-methylethylidene)cyclohexanone	1165.56	ND	1.47 ± 0.90a	0.44 ± 0.09b	0.60 ± 0.32b	ND	ND
(-)-fenchone	1094.40	ND	ND	6.71 ± 1.14a	5.38 ± 3.02b	ND	4.55 ± 0.12c
Total content		7.00 ± 4.43bc	8.95 ± 3.54b	13.38 ± 3.17a	11.87 ± 4.70a	4.42 ± 1.80c	9.90 ± 1.68b
Esters
α-terpinyl acetate	1261.92	3.50 ± 1.26a	2.98 ± 0.33ab	3.92 ± 1.05a	3.95 ± 2.07a	2.25 ± 1.01c	2.42 ± 0.51c
diisobutyl phthalate	1625.97	0.69 ± 0.22b	1.57 ± 0.23a	0.73 ± 0.14b	1.34 ± 0.22ab	0.90 ± 0.66b	1.45 ± 0.36a
1,2-benzenedicarboxylic acid	1629.27	1.13 ± 0.57b	1.84 ± 0.08a	1.00 ± 0.15b	1.00 ± 0.28b	0.67 ± 0.33c	1.66 ± 0.51ab
bornyl acetate		0.22 ± 0.10b	ND	0.38 ± 0.17b	1.07 ± 0.65a	ND	ND
linalyl acetate	1246.66	ND	ND	1.03 ± 0.43a	0.85 ± 0.26b	ND	ND
Total content		5.54 ± 2.15bc	6.39 ± 0.64b	7.06 ± 1.94ab	8.21 ± 3.48a	3.88 ± 2.00c	5.53 ± 1.38bc
Phenolics
butylated hydroxytoluene	1519.20	4.12 ± 0.25a	2.57 ± 0.53b	2.49 ± 0.64b	1.69 ± 0.38c	1.76 ± 0.37c	3.67 ± 1.65a
m-cresol	1440.16	1.66 ± 0.21a	1.49 ± 0.08ab	1.66 ± 0.36a	1.68 ± 0.57a	1.14 ± 0.27b	1.69 ± 0.67a
eugenol	1242.88	5.37 ± 0.57a	5.66 ± 0.71a	4.96 ± 1.09b	5.32 ± 0.47a	3.61 ± 0.94c	4.63 ± 0.72b
2,4-di-t-butylphenol	1445.72	2.13 ± 0.30a	2.94 ± 1.41a	ND	ND	ND	ND
methyleugenol	1235.34	ND	ND	0.84 ± 0.05b	1.18 ± 0.45a	ND	0.91 ± 0.09b
Total content		13.28 ± 1.33a	12.66 ± 2.73a	9.95 ± 2.14b	9.87 ± 1.87b	6.51 ± 1.58c	10.90 ± 3.13ab
Hydrocarbons
3-carene	1028.87	0.69 ± 1.03c	1.21 ± 0.98b	1.98 ± 0.37ab	0.83 ± 1.89c	0.32 ± 0.49c	4.13 ± 1.63a
d-limonene	1042.10	34.62 ± 1.36b	34.07 ± 2.30b	34.27 ± 3.31b	34.33 ± 1.16b	37.93 ± 3.76a	35.46 ± 3.39ab
2-methyl-1-phenylpropene	1107.53	ND	0.40 ± 0.09a	ND	0.38 ± 0.01a	ND	ND
alpha-caryophyllene	1509.04	2.81 ± 2.83b	2.71 ± 1.45a	2.81 ± 0.77b	3.09 ± 3.32a	2.91 ± 1.78b	3.43 ± 0.43a
caryophyllene oxide	1520.96	3.90 ± 2.45c	3.15 ± 2.12c	3.27 ± 0.21c	3.60 ± 1.41c	4.67 ± 0.25b	5.94 ± 0.41a
tetradecane	1400.02	ND	0.72 ± 0.36a	0.08 ± 0.25b	ND	ND	ND
myrcene	1033.93	ND	1.79 ± 0.17b	1.11 ± 1.46b	1.29 ± 0.05b	1.80 ± 0.77b	4.28 ± 0.80a
γ-terpinene	1052.21	ND	1.43 ± 0.33a	1.35 ± 0.32a	0.17 ± 0.02b	ND	ND
α-copaene	1497.48	0.62 ± 0.08ab	0.24 ± 0.19c	0.01 ± 0.94c	0.55 ± 0.39b	0.21 ± 1.13c	0.92 ± 0.50a
1,3-di-tert-butylbenzene	1400.30	ND	0.25 ± 0.01a	ND	ND	ND	ND
β-pinene	1017.96	ND	ND	0.92 ± 0.95a	0.74 ± 0.25b	0.56 ± 0.04b	ND
terpinolene	1066.00	ND	ND	0.80 ± 0.90a	ND	ND	ND
trans-anethole	1188.46	ND	ND	ND	ND	ND	5.58 ± 1.14a
Total content		42.64 ± 7.75b	45.97 ± 6.00a	42.33 ± 9.27b	44.78 ± 8.5ab	38.40 ± 8.22c	39.74 ± 8.30c
Acids
myristic acid	1457.20	3.77 ± 3.07d	4.89 ± 1.23c	4.59 ± 1.62c	4.72 ± 1.62c	16.56 ± 14.75a	9.36 ± 2.09b
pentadecanoic acid	1543.84	ND	5.52 ± 2.10c	5.96 ± 0.81c	5.90 ± 0.96c	25.44 ± 16.09a	16.65 ± 2.87b
nonanoic acid	1338.30	0.70 ± 0.32b	ND	0.45 ± 0.04b	0.58 ± 0.16b	0.78 ± 0.21b	2.29 ± 0.56a
dodecanoic acid	1309.08	ND	ND	0.49 ± 0.01a	0.50 ± 0.18a	0.49 ± 0.23a	ND
stearic acid	1819.52	14.66 ± 6.19a	ND	ND	3.54 ± 0.51d	8.80 ± 0.14b	5.02 ± 0.16c
Total content		19.13 ± 9.58c	10.41 ± 3.33d	11.49 ± 2.48d	15.24 ± 3.43c	52.07 ± 31.42a	33.32 ± 5.68b
Pyrazine
2,5-dimethylpyrazine	913.30	ND	ND	ND	ND	1.43 ± 1.21b	1.69 ± 1.03a
2,3,5-trimethylpyrazine	941.73	ND	ND	ND	ND	2.37 ± 0.22b	7.29 ± 5.54a
3-ethyl-2,5-dimethylpyrazine	1121.03	ND	ND	2.94 ± 0.31b	1.54 ± 0.46b	1.94 ± 0.41b	7.97 ± 2.37a
2,6-dimethylpyrazine	908.99	ND	ND	4.69 ± 0.50c	5.35 ± 0.50b	5.42 ± 0.50b	6.49 ± 0.50a
2,6-diethylpyrazine	1132.12	ND	ND	ND	ND	ND	1.99 ± 0.65a
Total content				7.63 ± 0.81c	6.89 ± 0.96c	11.13 ± 2.34b	25.43 ± 10.09a
Others
cis-anethol	1171.74	6.94 ± 1.75b	4.33 ± 1.89c	8.29 ± 2.32ab	7.09 ± 2.32b	9.86 ± 3.77a	7.00 ± 2.70b
safrole	1206.36	1.10 ± 0.50a	1.09 ± 0.24a	1.09 ± 0.11a	1.10 ± 0.46a	0.60 ± 0.09b	0.73 ± 0.19b
myristicin	1253.79	2.54 ± 0.10a	1.65 ± 0.15a	1.54 ± 0.58ab	1.67 ± 0.29a	1.17 ± 0.38bc	1.43 ± 0.35b
3-methylindole	1388.97	0.18 ± 0.17b	0.44 ± 0.23a	0.28 ± 0.05b	0.33 ± 0.08ab	ND	ND
Total content		10.76 ± 2.52ab	7.51 ± 2.51c	11.20 ± 3.06a	10.19 ± 3.15ab	11.63 ± 4.24a	9.16 ± 3.24b

Mean values with different lowercase letters in the same column are significantly different at p < 0.05. The data are presented as the mean ± SD (standard deviations).

ND: not detected.

Fig. 1.

Comparison of the percentages of volatile compounds in flavored fishcakes prepared using different methods. (CK: the control group; SS: steaming sample; BS: baking sample; AF: air frying; PF: pan frying; DF: deep frying).

3.1.2. OAV analysis

To further investigate the contributions of volatile components, OAV analysis was performed in conjunction with the sensory odor thresholds of each volatile component (Table 2). Compounds with 0.1 < OAV < 1 were considered to be odor-active substances, and compounds with OAV ≥ 1 were further defined as major odor-active substances(Yao et al., 2021). In this study, the compounds with an OAV ≥ 1 were nonanal, decanal, octanal, 3-ethyl-2,5-dimethylpyrazine, (2E,4E)-2,4-decadienal, 1-octen-3-ol, linalool, 1-octanol, eugenol, d-limonene, alpha-caryophyllene, caryophyllene oxide, myrcene, 2,5-dimethylpyrazine, and (2E)-2-decenal.

Table 2.

Thresholds and OAV of odor active compounds for different reheating methods.

Compounds Name	Threshold （μg/kg）a	Odorant Descriptionb	OAV
			CK	SS	BS	AF	DF	PF
nonanal	1.00	Fatty	31.44	31.81	31.62	39.74	28.01	34.65
decanal	0.10	Fatty、Sweet	38.60	23.90	31.90	36.10	38.80	14.90
octanal	0.70	Fatty	1.03	1.60	5.59	5.09	4.56	4.36
(2E)-2-decenal	0.30	Oils	1.27	2.37	2.50	3.90	3.03	3.13
(2E,4E)-2,4-decadienal	0.30	Fatty、Fruity	ND	2.43	2.63	2.40	2.77	2.90
1-octen-3-ol	1.00	Mushrooms	ND	1.82	2.58	3.39	1.66	1.76
linalool	6.00	Fruity	1.25	2.42	2.81	2.23	2.06	1.30
1-octanol	1.00	Citrusy、Oils	ND	1.66	1.05	3.29	2.58	2.38
eugenol	2.50	Clove	2.15	2.26	1.98	2.13	1.85	1.44
d-limonene	34.00	Floral、Citrusy	1.02	1.00	1.00	1.00	1.04	1.16
alpha-caryophyllene	2.40	Floral 、Citrusy	1.17	1.13	1.17	1.29	1.43	1.21
caryophyllene oxide	3.00	Citrusy、Clove	1.30	1.05	1.09	1.20	1.98	1.56
myrcene	1.10	Floral	ND	1.63	1.00	1.17	3.89	1.64
2,5-dimethylpyrazine	0.40	Nutty、Roasted	ND	ND	11.73	13.38	16.23	13.55
3-ethyl-2,5-dimethylpyrazine	0.40	Roasted、Meaty	ND	ND	7.35	3.85	19.93	4.85

ND: not detected.

^a,b Reference aroma description of published lecture from Sun et al., 2022, Zheng et al., 2022, Zheng et al., 2022, Sun et al., 2022, Deng et al., 2021.

Aldehydes, which are mainly generated by the oxidation of fats and amino acids during Strecker degradation reactions(Zheng et al., 2022, Sun et al., 2022), are important volatile compounds in foods and are thought to contribute significantly to the flavor of meat products due to their low threshold values(Shen et al., 2021). Nonanal and octanal have a fatty flavor, mainly due to the oxidation of oleic acid; both of these compounds generally have a high OAV and are important flavor compounds in fishcakes; additionally, (2E)-2-decenal and (2E,4E)-2,4-decadienal impart fatty, fruity, and fried potato flavor profiles. (2E,4E)-2,4-decadienal are important flavor compounds in surimi(An et al., 2020). Aldehydes are considered to be the main source of flavor in cooked aquatic products, and nonanal and octanal are important volatile compounds in turbot after the application of five cooking methods(Dong et al., 2018).

Alcohols, formed by the oxidation of polyunsaturated fatty acids, play a key role in the flavour of meat products. 1-octen-3-ol is produced by the oxidation of arachidonic acid and usually imparts a mushroom flavor(Duan et al., 2021), and linalool is derived from spices and has a woody and fruity aroma. 1-octanol has an aroma and sweetness and has been suggested to be an aroma activator in dry shrimp(Hu et al., 2021).

Phenolics are mainly derived from spices, and it has been shown that phenolics are the major constituents of some spice plants that are transferred to products during processing to enhance flavor(Cohen et al., 2019); hydrocarbons generally have a higher threshold and contribute less to the volatile flavor. In this study, eugenol was extracted from spices with a butyric flavor; the d-limonene and alpha-caryophyllene were extracted from patchouli extracts with a citrus and floral odor profile.

Pyrazines are important volatile compounds produced during thermal processing of aquatic products and are a volatile group resulting from the Maillard reaction. 2,5-dimethylpyrazine and 3-ethyl-2,5-dimethylpyrazine have meaty, nutty and toasted flavor profiles and are the key flavor compounds for flavored fishcakes. Similar to the results of Duppeti et al.(2022), pyrazines were not detected in CK or SS.

3.1.3. Multivariate statistical analysis

OPLS-DA is an analytical method for data visualization and quantification of the degree of variation between the samples according to the correlation between the data(Kang et al., 2022). The relative levels of substances with an OAV > 1 in Table 2 were selected as Y variables for the OPLS-DA model design. The explanatory power of the model was expressed as R²X and R²Y and the predictive ability of the model was expressed as Q², where the R² and Q² closer to 1.0 indicate a better fit of the model. As shown in Fig. 2A, R²Y = 0.978 and Q² = 0.933, which are both close to 1, indicating that the model has good explanatory and predictive power. As shown in Fig. 2A, the samples treated by different reheating methods were well separated. SS and CK were located in the first and fourth quadrants, respectively; BS and PF were located in the second quadrant; AF was located in the third quadrant; and DF was distributed in both one and four quadrants. In addition, the BS and PF samples were similar, indicating that they had similar flavor types.

Fig. 2.

(A) Score plot of OPLS-DA (R²Y = 0.978; Q² = 0.933); (B) Cross-substitution plot of 200 permutation tests (R² = 0.304; Q² = -1.030); (C) Distribution of VIP values (red represents the characteristic flavors with VIP > 1).

The reliability of OPLS-DA was tested by performing 200 cross-substitution tests on the model, and the results are shown in Fig. 2B. The horizontal coordinates in the graph are the retention of the samples, and the points at 1.0 are the R² and Q² of the original model. After validation, R² (0.304) and Q² (-1.030) are less than the retention value of 1.0, and the intercept of the Q² regression line of the model with the horizontal coordinate is negative, indicating that the model is free from overfitting and is stable and reliable.

The VIP is commonly used to analyze key variables in OPLS-DA models, and a VIP greater than 1 indicates a greater contribution. The VIP values obtained for each key component are shown in Fig. 2C. Among those with a VIP value greater than 1 were 3-ethyl-2,5-dimethylpyrazine, (2E)-2-decenal, decanal, n-octanal, nonanal, 1-octanol-3-ol, (2E,4E)-2,4-decadienal and 2,5-dimethylpyrazine, and the above eight substances were identified as characteristic odors by combining the OAV and odor descriptions of the key odor compounds.

3.2. GC-IMS analysis

3.2.1. Topography of flavored fishcakes cooked via different methods

GC-IMS is a nontargeted analytical method for the identification of volatile compounds using retention time and drift time(Feng et al., 2022) and has the characteristics of rapid detection, intuitive results and portability. The topography obtained from the GC-IMS analysis in this study is shown in Fig. 3A. The horizontal coordinates indicate the migration time, the vertical coordinates indicate the retention time, and the red vertical line at 0.5x the horizontal coordinate indicates the normalized reactive ion peak (RIP). The points on either side of the RIP peak represent the volatile components, and the colors represent the volatile component contents, with lower contents appearing in white and higher contents appearing in red. The types of volatile components in the samples obtained from the different cooking methods were similar, mainly with regard to the concentration differences.

Fig. 3.

Spectra obtained by GC-IMS of samples prepared using different methods; (A) topographic map; (B) comparison of differences (CK as reference).

To more clearly observe the differences between the different cooking methods, the CK spectrum was chosen as the reference, and differential spectra were obtained by comparison with the samples obtained using the other cooking methods (Fig. 3B). A concentration lower than the reference is shown in blue, that equal to the reference is shown in white and that higher than the reference is shown in red(Lourdes et al., 2014). It is observed from Fig. 3B that the volatile components of all of the samples completed gas-phase separation within 800 s, and significant differences among the volatile compositions of the samples obtained using different cooking methods were observed. Significantly more red dots are present for the BS, DF, PF and AF samples than for the SS and CK samples, indicating that the Maillard reaction became more intense and more volatile components were produced with higher temperature.

3.2.2. Fingerprint analysis of volatile compounds produced by different cooking methods

To further observe the pattern of the changes in the volatiles, the migration time and retention indices of each substance were used to determine the volatile components in the GC-IMS database for comparison, and the results are shown in Table 3. 41 volatile compounds, namely 8 aldehydes, 8 alcohols, 9 ketones, 8 esters, 6 hydrocarbons, 1 pyrazine and 1 acid, were detected by GC-IMS. Compared with those in the CK group, the relative levels of aldehydes, alcohols and esters increased, and the level of ketones decreased after five different cooking treatments.

Table 3.

Information about volatile compounds identified by GC-IMS.

Compounds Name	Formula	RI	RT(sec)	DT(RIP relative)	relative content/%
					CK	SS	BS	AF	DF	PF
Aldehydes
benzaldehyde	C7H6O	961.4	335.427	1.15477	0.51 ± 0.02c	1.74 ± 0.17a	1.37 ± 0.04b	1.74 ± 0.07a	1.59 ± 0.03a	1.25 ± 0.15b
heptanal	C7H14O	899.5	279.496	1.34284	0.19 ± 0.04d	5.18 ± 0.11a	4.67 ± 0.09a	5.08 ± 0.36a	3.38 ± 0.54c	4.16 ± 0.21b
pentanal	C5H10O	694.1	166.226	1.42541	0.20 ± 0.03f	3.31 ± 0.67b	2.66 ± 0.19c	5.61 ± 1.04a	1.06 ± 0.21e	1.72 ± 0.49d
2-Methylpentanal	C6H12O	753.2	192.469	1.38877	1.84 ± 0.96a	0.45 ± 0.02b	0.52 ± 0.03b	0.36 ± 0.02c	0.31 ± 0.04c	0.40 ± 0.02b
nonanal	C9H18O	1109.9	549.562	1.48205	0.37 ± 0.04d	1.53 ± 0.01a	1.30 ± 0.10b	1.15 ± 0.04c	0.96 ± 0.23c	1.04 ± 0.17c
3-methylthiopropanal	C4H8OS	906.7	285.465	1.09276	0.22 ± 0.05c	0.53 ± 0.08b	0.57 ± 0.12b	1.06 ± 0.20a	1.26 ± 0.26a	0.66 ± 0.17b
3-methylbutanal	C5H10O	654.4	153.55	1.40238	7.27 ± 0.59b	5.26 ± 0.26c	5.73 ± 0.68c	7.66 ± 0.17b	8.13 ± 0.79b	9.50 ± 1.06a
2-methylpropanal	C4H8O	545.7	124.606	1.28771	0.25 ± 0.08c	0.44 ± 0.07c	0.38 ± 0.07c	1.99 ± 0.47b	5.70 ± 2.28a	1.32 ± 0.51b
Total content					10.86 ± 1.81d	18.44 ± 1.39c	17.19 ± 1.31c	24.65 ± 2.36a	22.38 ± 4.38b	20.05 ± 2.79b
Alcohols
5-methyl-2- furanmethanol	C6H8O2	957.8	331.972	1.26126	1.06 ± 0.26a	0.37 ± 0.12d	0.50 ± 0.05c	0.78 ± 0.10b	0.74 ± 0.22b	0.68 ± 0.19c
n-hexanol	C6H14O	869.6	258.238	1.33172	2.02 ± 0.35a	1.92 ± 0.14a	1.92 ± 0.21a	1.10 ± 0.11c	1.18 ± 0.17c	1.56 ± 0.05b
2,3-butanediol	C4H10O2	791.2	211.532	1.3701	0.45 ± 0.08d	2.07 ± 0.08b	2.61 ± 0.09a	1.43 ± 0.08c	2.80 ± 0.37a	2.45 ± 0.38ab
3-methyl-3-buten-1-ol	C5H10O	731.6	182.432	1.25249	0.46 ± 0.07e	2.17 ± 0.11b	2.53 ± 0.43a	0.88 ± 0.15d	1.48 ± 0.06c	1.79 ± 0.13c
2-butanol	C4H10O	635.6	148.102	1.32307	2.48 ± 0.65c	1.11 ± 0.03d	1.21 ± 0.19d	2.75 ± 0.53c	4.72 ± 0.22a	3.28 ± 0.32b
2-propanol	C3H8O	544.9	124.408	1.22823	1.57 ± 0.54c	2.69 ± 0.35b	1.91 ± 0.32bc	4.07 ± 0.45a	3.75 ± 0.79a	2.38 ± 0.41bc
pentan-1-ol	C5H12O	763	197.204	1.25904	5.91 ± 0.69d	7.66 ± 0.08b	8.76 ± 0.34a	6.14 ± 0.18c	4.57 ± 0.40e	6.91 ± 0.05c
methyl-4-penten-2-ol	C6H12O	757.4	194.442	1.23969	6.81 ± 1.21b	6.98 ± 0.21b	7.99 ± 0.24a	5.65 ± 0.24c	3.91 ± 0.33d	6.08 ± 0.36c
Total content					20.78 ± 3.85d	24.98 ± 1.12b	27.43 ± 1.86a	22.81 ± 1.84c	23.17 ± 2.58c	25.14 ± 1.89ab
Ketones
2-heptanone	C7H14O	889.9	271.904	1.26392	0.83 ± 0.08c	0.92 ± 0.06c	0.87 ± 0.06c	1.20 ± 0.25b	1.35 ± 0.19a	1.42 ± 0.18a
cyclohexen-2-one	C6H8O	899.8	279.749	1.40731	0.25 ± 0.02c	0.67 ± 0.12b	0.76 ± 0.03b	0.90 ± 0.06a	0.64 ± 0.09b	0.75 ± 0.05b
5-methylfuran-2(3H)- one	C5H6O2	872.3	260.01	1.36618	3.24 ± 0.68a	0.28 ± 0.02b	0.33 ± 0.01b	0.21 ± 0.02c	0.27 ± 0.02b	0.31 ± 0.04b
3-pentanone	C5H10O	695	166.631	1.35389	7.20 ± 0.91a	6.45 ± 0.71b	5.65 ± 0.77c	3.32 ± 0.52d	3.59 ± 0.66d	6.38 ± 1.33b
2,3-pentanedione	C5H8O2	691.9	165.364	1.3007	1.40 ± 0.38c	1.88 ± 0.18b	2.15 ± 0.11a	1.44 ± 0.05c	1.62 ± 0.10b	1.68 ± 0.34b
2,3-butanedione	C4H6O2	580.7	133.286	1.16444	4.56 ± 0.68a	1.66 ± 0.33d	2.04 ± 0.16c	1.94 ± 0.09c	2.31 ± 0.21b	2.45 ± 0.67b
2-Pentanone	C5H10O	688.7	164.061	1.11713	4.03 ± 0.53a	1.38 ± 0.02b	1.50 ± 0.17b	1.74 ± 0.16b	1.67 ± 0.30b	1.66 ± 0.29b
1-penten-3-one	C5H8O	680.6	161.496	1.07484	1.24 ± 0.13b	0.84 ± 0.11c	1.07 ± 0.17b	2.10 ± 0.56a	2.11 ± 0.30a	1.09 ± 0.38b
(E)-5-methyl-2-hepten-4-one	C8H14O	970.3	344.453	1.48847	2.78 ± 0.45a	0.17 ± 0.01b	0.15 ± 0.01b	0.18 ± 0.03b	0.17 ± 0.01b	0.15 ± 0.02b
Total content					25.53 ± 3.86a	14.24 ± 1.56b	14.54 ± 1.49b	13.04 ± 1.72b	13.74 ± 1.87b	15.9 ± 3.29b
Esters
ethyl (E)-2-hexenoate	C8H14O2	1050.6	448.224	1.31617	3.38 ± 0.49a	0.19 ± 0.04b	0.19 ± 0.01b	0.27 ± 0.03b	0.29 ± 0.11b	0.24 ± 0.09b
hexyl acetate	C8H16O2	1005.7	384.104	1.41393	0.38 ± 0.09d	1.69 ± 0.04b	1.57 ± 0.07b	1.85 ± 0.16a	1.03 ± 0.39c	1.28 ± 0.21c
2-methylbutylacetat	C7H14O2	871.9	259.757	1.29726	10.26 ± 0.99a	0.71 ± 0.07b	0.72 ± 0.07b	0.42 ± 0.07b	0.41 ± 0.06b	0.56 ± 0.07b
ethyl trans-2-butenoate	C6H10O2	848.2	244.572	1.18611	1.18 ± 0.39b	0.59 ± 0.06d	0.80 ± 0.08c	1.62 ± 0.02a	0.52 ± 0.14d	0.66 ± 0.15c
isoamyl formate	C6H12O2	792.3	212.139	1.27283	1.75 ± 0.64d	9.67 ± 0.42a	9.94 ± 0.19a	6.72 ± 0.18c	7.81 ± 0.60b	8.6 ± 0.73ab
ethyl butyrate	C6H12O2	792.7	212.341	1.56559	0.21 ± 0.03d	16.21 ± 0.99a	14.91 ± 1.16b	15.15 ± 0.27ab	12.49 ± 3.68c	14.2 ± 2.70b
methyl acetate	C3H6O2	520.4	118.687	1.19382	0.67 ± 0.12d	0.78 ± 0.16d	0.86 ± 0.19d	2.06 ± 0.83b	6.95 ± 3.07a	1.25 ± 0.34c
methyl propanoate	C4H8O2	622.9	144.531	1.1372	2.68 ± 0.83a	0.79 ± 0.08c	1.11 ± 0.15b	0.75 ± 0.02c	0.87 ± 0.10c	1.14 ± 0.16b
Total content					20.53 ± 3.58c	30.64 ± 1.87a	30.09 ± 1.93a	28.84 ± 1.58b	30.38 ± 8.15a	27.93 ± 4.45b
Hydrocarbons
limonene	C10H16	1028.3	415.1	1.22193	1.99 ± 0.28a	1.1 ± 0.09ab	0.98 ± 0.25b	0.45 ± 0.09d	0.72 ± 0.15c	0.65 ± 0.04d
β-pinene	C10H16	974.9	349.057	1.21846	9.91 ± 0.31a	1.57 ± 0.49ab	1.23 ± 0.26b	0.65 ± 0.17c	1.01 ± 0.09b	1.01 ± 0.13b
styrene	C8H8	899.7	279.66	1.43909	0.34 ± 0.04c	0.67 ± 0.03ab	0.75 ± 0.02a	0.61 ± 0.03b	0.73 ± 0.09a	0.72 ± 0.06a
2,4-dimethyl-1-heptene	C9H18	848.3	244.591	1.30258	0.32 ± 0.08b	0.22 ± 0.02c	0.36 ± 0.03ab	0.77 ± 0.05a	0.25 ± 0.09c	0.36 ± 0.07ab
1,2-dibromoethane	C2H4Br2	816.2	225.428	1.24709	7.59 ± 0.68a	6.08 ± 0.86ab	4.95 ± 0.82bc	5.75 ± 0.22b	3.54 ± 1.28d	4.61 ± 1.85c
1-(1,1-dimethylethoxy)-2-methylpropane	C8H18O	763	197.204	1.3343	0.48 ± 0.14c	0.76 ± 0.01b	0.97 ± 0.03a	0.55 ± 0.02c	0.82 ± 0.06ab	0.89 ± 0.04ab
Total content					20.63 ± 1.53a	10.40 ± 1.50ab	9.24 ± 1.41ab	8.78 ± 0.58b	7.07 ± 1.76c	8.24 ± 2.19b
Pyrazine
2,3-dimethyl pyrazine	C6H8N2	924.1	300.49	1.11928	0.33 ± 0.01c	0.18 ± 0.01d	0.22 ± 0.01d	0.55 ± 0.41c	2.06 ± 0.61a	1.52 ± 0.63b
acid
2-methylpropionic acid	C4H8O2	783.6	207.484	1.15636	1.34 ± 0.21a	1.13 ± 0.12b	1.27 ± 0.02ab	1.34 ± 0.06a	1.19 ± 0.11b	1.23 ± 0.14ab

Mean values with different lowercase letters in the same column are significantly different at p < 0.05. The data are presented as the mean ± SD (standard deviations).

To respond directly to the differences in the volatile content in the samples cooked via different methods, a volatile fingerprint was constructed (Fig. 4). Each row represents a signal peak, and each column represents a compound. The content of volatiles in the samples varied with the brightness of the points, with lighter color corresponding to lower concentration. As shown in Fig. 4, the fingerprint can be divided into five regions, A, B, C, D and E. The volatile components in region A had the highest concentration in CK, and consisted mainly of 10 compounds, such as 5-methylfuran-2(3H)-one, 2-methylpentanal, β-pinene and methyl propionate. Among these, β-pinene and methyl propionate were obtained from a yeast-patchouli compound deodorant solution; both of these compounds have floral and fruity aromas(Rong et al., 2023). The B-zone volatiles had the highest concentration in DF; these volatiles consisted of methyl acetate, 2-methylpropanal, 2-propanol and 2,3-dimethylpyrazine. 2,3-dimethylpyrazine is a typical compound found in roasted meat and has a roasted flavor(Zhang et al., 2023). Zone C volatiles, consisting of ethyl trans-2-butenoate, 2,4-dimethyl-1-heptene and pentanal, had higher concentrations in the AF. The Zone D volatiles were generally more abundant for the five cooking modes and consisted of 21 substances, including 3-methylbutanal, nonanal and 2-methylbutylacetat. Zone E, which was common to both the CK treatment and the five cooking modes, consisted of n-hexanol, 2-pentanone and limonene.

Fig. 4.

Fingerprint profiles of samples prepared using different methods obtained via GC-IMS.

3.2.3. ROAV analysis

The ROAV method is widely used to analyze key aroma compounds, with higher ROAV corresponding to greater contribution of the volatile odor of the compound to the overall aroma(Yao et al., 2021). Compounds with 0.1 < ROAV < 1 were considered to have a modifying effect on the overall aroma, and compounds with ROAV ≥ 1 were considered to be key aroma compounds. As shown in Table 3, Table 3-methylbutanal was the component with the highest relative content and a low threshold (1.10 μg/kg). Therefore, 3-methylbutanal was selected as the main flavoring substance, which means that the concentration of ROAV_stan was 100. The ROAV of the remaining volatiles are shown in Table 4.

Table 4.

Thresholds and ROAV of odorants for different cooking methods.

Compounds Name	Threshold （μg/kg）a	Odorant Descriptionb	ROAV
			CK	SS	BS	AF	DF	PF
heptanal	3.00	Nutty、Fruity	0.96	36.10	29.89	24.31	15.21	16.06
pentanal	12.00	Grassy	0.25	5.77	4.26	6.71	1.19	1.66
2-methylpentanal	1.60	Fatty	17.40	5.88	6.24	3.23	2.61	2.90
nonanal	1.00	Fatty	5.60	31.99	24.96	16.51	12.96	12.04
3-methylbutanal	1.10	Meaty、Fruity	100.00	100.00	100.00	100.00	100.00	100.00
2-methylpropanal	1.50	Fatty	2.52	6.13	4.86	19.05	51.3	10.19
n-Hexanol	5.60	Floral	5.46	7.17	6.58	2.82	2.84	3.23
pentan-1-ol	150.00	Sweet、Floral	0.60	1.07	1.12	0.59	0.41	0.53
2-heptanone	140.00	Meaty	0.09	0.14	0.12	0.12	0.13	0.12
3-Pentanone	40.00	Floral	2.72	3.37	2.71	1.19	1.21	1.85
2,3-pentanedione	29.00	Floral	0.73	1.36	1.42	0.71	0.75	0.67
2,3-butanedione	10.00	Floral	0.73	3.47	3.92	2.79	3.12	2.84
2-pentanone	10.00	Cheesiness	6.90	2.89	2.88	2.50	2.25	1.92
1-penten-3-one	23.00	Spicy	6.10	0.76	0.89	1.31	1.24	0.55
hexyl acetate	2.00	Fruity、Sweet	2.87	17.67	15.07	13.28	6.95	7.41
2-methylbutylacetat	5.00	Fruity、Sweet	31.05	2.97	2.76	1.21	1.11	1.30
ethyl butyrate	73.00	Fruity、Sweet	0.04	4.64	3.92	2.98	2.31	2.25
methyl acetate	2.00	Fruity、Sweet	5.07	8.15	8.26	14.79	46.91	7.24
methyl propanoate	100.00	Fruity、Sweet	0.41	0.17	0.21	0.11	0.12	0.13
limonene	4.00	Floral、citrusy	7.53	5.75	4.70	1.62	2.43	1.88
β-pinene	6.00	Floral、citrusy	24.99	5.47	3.94	1.56	2.27	1.95

^a,b Reference aroma description of published lecture from Sun et al., 2022, Zheng et al., 2022, Zheng et al., 2022, Sun et al., 2022, Deng et al., 2021.

Aldehydes have a low odor threshold and strong aroma and are an important component of meat products(Elmore et al., 2005). For example, heptanal has a fruity and nutty flavor, and 3-methylbutanal has a fruity and cheesy flavor. Studies have shown that heptanal and 3-methylbutanal are widely present in aquatic products such as chub, tuna and sturgeon(Li et al., 2022, Wang et al., 2022). Ketones are mainly derived from fat oxidation and usually have floral, creamy and fruity aroma characteristics. 2-heptanone is thought to impart a meaty flavor(Bassam et al., 2022), while 2-pentanone has a creamy and cheesy taste(Xie et al., 2022). Esters are formed by the combination of acids and alcohols from the thermal degradation of lipids, and short carbon chain acids and alcohols form esters with a low threshold. For example, hexyl acetate and ethyl butyrate have fruity and sweet flavors, respectively(Deng et al., 2021)^, and these esters are derived mainly from fruit and vegetable products.

3.2.4. Multivariate statistical analysis

The relative contents of substances with ROAV > 1 in Table 4 were selected as Y-variables for the OPLS-DA model design. As shown in Fig. 5A, R²Y = 0.969 and Q² = 0.920 are both close to 1, indicating that the model has good explanatory and predictive ability. As shown in Fig. 5A, the samples treated with different cooking methods were well separated. DF and PF were in the first quadrant, CK was in the third quadrant, BS and SS were in the fourth quadrant, and AF was distributed in one or four quadrants.

Fig. 5.

(A) Score plot of OPLS-DA (R²Y = 0.969; Q² = 0.920); (B) Cross-substitution plot of 200 permutation tests (R² = 0.182; Q² = -0.797); (C) Distribution of VIP values (red represents the characteristic flavors with VIP > 1）.

The results obtained by performing 200 cross replacement tests on the model are shown in Fig. 5B. Validation showed that R² (0.182) and Q² (-0.797) were both less than the retention value of 1.0, and the intercept between the regression line of Model Q² and the abscissa was negative, indicating that the model was not overfitted and was stable and reliable. The results for the VIP values of the key components are shown in Fig. 5C. Among those with VIP values greater than 1 are 3-methylbutanal, 2,3-pentanedione, 3-pentanone, pentan-1-ol, pentanal, 2-methylpropanal and 1-penten-3-one, which were identified as characteristic odors in combination with ROAV and as the odor description of the key odor actives.

4. Conclusion

In summary, volatile organic compounds (VOCs) in tilapia cakes prepared by different cooking methods (steaming, baking, air frying, pan frying and deep frying) were comprehensively analyzed using GC–MS and GC-IMS. A total of 72 volatiles were identified by GC–MS, and 41 volatiles were detected by GC-IMS. An OPLS-DA model for GC–MS analysis of flavored fishcakes cooked with different methods was constructed based on the VIP values, and eight characteristic flavoring substances were screened, namely, 3-ethyl-2,5-dimethylpyrazine, decanal, (2E)-2-decenal, octanal, nonanal, 1-octanol-3-ol, (2E,4E)-2,4-decadienal and 2,5-dimethylpyrazine. The OPLS-DA model for GC-IMS analysis revealed seven characteristic aroma compounds, namely, 3-methylbutanal, 2,3-pentanedione, 3-pentanone, pentan-1-ol, pentanal, 2-methylpropanal and 1-penten-3-one. The combination of GC–MS and GC-IMS maximized the retention of overall information regarding the volatile components of flavored fishcakes obtained using different cooking methods and allowed rapid differentiation of different cooking methods using multivariate statistical analysis. Future research will be focus on consumer sensory preferences and taste profiles.The results of this study may provide new insights into the dynamics of flavor formation in culinary fishcakes and Provide theoretical and technical references for the industrialised production of fishcake processing, as well as a better after-sales experience for the consumer's home cooking style.

CRediT authorship contribution statement

Xuebo Yang: Writing – original draft, Investigation, Data curation. Qiuhan Chen: Formal analysis, Conceptualization. Shouchun Liu: Writing – review & editing, Methodology, Investigation. Pengzhi Hong: Project administration, Funding acquisition. Chunxia Zhou: Validation, Software. Saiyi Zhong: Validation, Software.

Declaration of competing interest

The authors declare that they have no known competing financial interests or personal relationships that could have appeared to influence the work reported in this paper.

Data availability

Data will be made available on request.