Analysis of changes in flavor characteristics of Lipu taro pork at different stages of cooking based on GC-IMS, electronic nose, and electronic tongue Authors: Zhigui He, Yan Hu, Yuqing Zhang, Yingying Cui, Yingying Liu, Saeed A Asiri, Sam Al-Dalali Published: 2025 ABSTRACT AbstractThis study aims to examine the changes in volatile compounds of Lipu taro pork at various cooking stages, including raw pork meat (RP), pre-cooked meat (PM), roasted meat (RM), ultrasonically cured meat (UC), and steamed meat (SM). GC-IMS, fatty acid analysis, electronic tongue, and electronic nose were employed to identify flavor compounds at each processing stage, and the results were analyzed using multivariate statistical methods. Findings showed that the fatty acid content in PM was significantly higher than in pork from other stages, followed by the UC group, while SM had the lowest. GC-IMS detected a total of 60 compounds, including aldehydes, ketones, alcohols, and esters. Significant differences in flavor compound composition were observed across the various processing stages. Using the PLS-DA model, 21 compounds with VIP values above 1.0 were identified, with methyl-5-hepten-2-one (VIP > 1.5) being the most important for differentiation, followed by acetone, 2-pentanone, and ethyl hexanoate-D (VIP > 1.4).Keywords: Lipu taro pork, Flavor compounds, GC-IMS, E-nose, E-tongue, Fatty acids Introduction Steamed pork belly holds a prominent position as a traditional dish within the Chinese community (Bi et al., 2023; Quan et al., 2023). The prepared food sector in China is currently experiencing rapid growth at both the national and provincial levels. The Chinese pork belly is similar to other traditional meat dishes, which include vegetable components such as fermented soybean, dried mustard, and taro (Quan et al., 2023). The steamed pork belly, made from Lipu taro and pork, is delicate in texture, soft and fragrant when cooked, with a pleasing flavor and a fat that is neither greasy nor oily, as well as a reasonable combination of nutritional value, which people highly favor. The traditional method of making Lipu taro pork belly involves boiling it until tender, then frying it until the skin turns golden yellow, soaking it in cold water to soften it, slicing it, and marinating it in a complex spice sauce. After marinating, it is steamed with fried taro pieces till soft, sticky, and crispy. Taro, scientifically known as Colocasia esculenta (L.) Schott, is grown mainly in a region in southern China (Zeng et al., 2014). Starch is its principal chemical component; lipids, protein, trace minerals, and vitamins make up lesser quantities (Nip, 2023). Chinese pickled and steamed pork belly, mixed with taro slices, is a renowned delicacy in Chinese cuisine. It comprises taro slices, pork belly, and assorted additions, which are subsequently steamed. The resultant steamed dish provides a pleasing flavor and a gratifying sense of satiety (Quan et al., 2023). Meat is processed using various methods, including grilling, baking, microwave cooking, deep-frying, traditional oven cooking, boiling, and marination, which can enhance its taste, aroma, and other sensory characteristics (Akköse & Aktaş, 2014; Suleman et al., 2020). Traditional pickling involves slow, inconsistent salt penetration, which can lead to microbial contamination and make it unsuitable for large-scale industrial manufacturing. Ultrasonic technology accelerates marinade penetration into meat using heat, cavitation, and mechanical forces (Contreras-López et al., 2020). In most investigations, ultrasound-assisted pickling technology reduces pickling time, preserves meat color, texture, and tenderness, and enhances the product's chemical, physical, and sensory quality (Zhou, Hu, & Wang, 2022). Under heating conditions, substances such as fat, protein, and carbohydrates in meat undergo specific physicochemical reactions. Protein denaturation, fat oxidation, and decomposition generate flavor compounds, altering the sensory quality and structural morphology of meat (Bi et al., 2023). Additionally, cooking processes serve to kill pathogenic bacteria, improve taste, and accelerate the oxidation of fats (Cheng et al., 2023). The oxidation that occurs during cooking is vital for flavor development, and variations in oxidation levels can influence overall flavor. Research by Echegaray et al. (2020) found that reduced lipid oxidation in fried pork resulted in fewer flavor compounds than roasting. While animal fats are key to flavor formation, the breakdown of protein significantly affects meat's taste and flavor, in which, small percentage of aldehydes has been identified as originating from protein degradation (Huang et al., 2022). Steaming, as a thermal processing method, sterilizes food and promotes the development of volatile compounds (Jiang et al., 2022). Currently, commercially available Chinese steamed pork belly products vary in quality, have a short shelf life, and lack standardized production processes. Addressing these issues is essential for improving industrial efficiency and meeting growing consumer demand. Previous studies have examined the impact of taro slices on the chemical composition of Chinese pickled and steamed pork belly (Quan et al., 2023), and Bi et al. (2023) investigated the effect of steaming time on pork belly lipid profiles using lipidomics. The traditional technique for preparing Lipu taro pork belly involves boiling, roasting, marinating in a sauce mixture under ultrasonic treatment, and steaming. Previous research has not examined the variations in volatile flavor compounds of Lipu taro pork across different cooking stages. This study aims to track and measure the dynamic changes in volatile flavor compounds during the various cooking stages of Lipu taro pork using advanced instrumental analysis. The stages include raw pork meat (RP), pre-cooked meat (PM), roasted meat (RM), ultrasonically cured meat (UC), and steamed meat (SM). Analyzing flavor development at each stage provides a theoretical framework for regulating volatile flavor compounds in industrial production. It guides processing and quality control in the preparation of Lipu taro pork belly dishes. Materials and methods Materials and reagents Pork meat, Lipu taro, ginger, shallot, Sifangjing Fermented bean curd, Guilin sanhua wine, soy sauce, spiced essential oil, salt, sugar, soy sauce (food grade), etc., were obtained from Guangxi Sangu Cuisine Network Technology Co., Ltd. 14 % Boron trifluoride methanol solution was purchased from Sigma Aldrich, USA. Ketones: 2-butanone, 2-pentanone, 2-hexanone, 2-heptanone, 2-octanone, and 2-nonanone (all analytical grade) were bought from Aladdin Company, Shanghai, China. Acetic acid, sulfuric acid (high-purity), copper sulfate, potassium sulfate, sodium hydroxide, p-nitrophenol, sodium acetate, anhydrous sodium acetate, 37 % formaldehyde, acetylacetone, anhydrous ethanol, potassium hydroxide, and tartaric acid (analytical grade) were purchased from Guangzhou Chemical Reagent Factory, China. Sample preparation The raw pork belly (RP) was cut into 10 cm × 10 cm × 5 cm cubes along the muscle fibers, submerged in water, and boiled until fully cooked. The cooked meat was subsequently removed to obtain the pre-cooked meat (PM). The tissue absorbed the surface moisture from the PM skin. The cone was used to perforate the meat's skin at 2 mm intervals to a depth of 1 cm. Salt and dark soy sauce were incorporated, after which the meat sample was positioned on a baking tray and subjected to cooking in an electric oven preheated to the designated temperatures (230 °C on the top and 180 °C on the bottom) for 42 min, until the pork belly skin attained a reddish-brown color, resulting in roasted meat (RM). The RM was immersed in cold water for 10 min until the skin softened, after which the streaky pork was sliced into 8 cm × 6 cm × 0.7 cm pieces. The sauce was prepared by mixing fermented bean curd, Sanhua wine, spicy essential oil, salt, sugar, soy sauce, shredded ginger, and scallions. The streaked pork slices were meticulously combined with the sauce. The sauce and streaked pork slices were placed in a self-sealing bag, sealed, and subjected to ultrasonic curing (UC) for 20 min. Thereafter, the Lipu taro were peeled and sliced into 8 cm × 6 cm × 0.7 cm pieces, ensuring the dimensions were equivalent to or smaller than those of the pork belly. The taro slices were fried at 160 °C until they became light golden, then removed. Subsequently, the UC meat was mixed with taro slices, layering them until the bowl was filled, and then the sauce was poured over the mixture. The mixture was cooked for 60 min over medium heat until the taro and pork belly became soft, sticky, and crispy. The steamed Lipu taro belly meat was removed from the cage and garnished with minced green onions (Steamed meat (SM)). The whole procedure was illustrated in Fig. 1. Pork at various processing stages—RP, PM, RM, UC, and SM—was sampled for subsequent analysis. Schematic diagram of the production process of Lipu taro pork at different stages of cooking. RP, raw pork meat; PM, pre-cooked meat; RM, roasted meat; UC, ultrasonically cured meat; and SM, steamed meat. Fatty acid analysis The fatty acid composition of the Lipu taro pork samples at various cooking stages was analyzed using gas chromatography (Agilent 7890–5975 Gas Chromatography-Mass Spectrometer, Agilent Technologies (China) Co., Ltd.). The lipid fraction was extracted from 5 g of Lipu taro pork samples with a 1:1 (vol/vol) mixture of ether and petroleum, and the lipid extract was concentrated using a rotary evaporator. The extracted sample was then transferred to a 250 mL round-bottom flask, and the protocol outlined by Bazina and He (2018) was followed. The fatty acid methyl ester (FAMEs) was prepared using the boron trifluoride-methanol method. Subsequently, the FAMEs were qualitatively analyzed by comparing their retention times with those of the standards. The HP-5MS column (60 m × 0.25 mm, 0.25 μm) was used for separation. The inlet temperature was set at 280 °C, with an injection volume of 1.0 μL and a split ratio of 20:1. Helium was used as the carrier gas. The flow rate was set at a constant linear velocity of 1.5 mL/min. The heating program started at an initial column temperature of 120 °C and was maintained for 1 min. Subsequently, the temperature increased to 170 °C at 6 °C/min, with no hold time. Then, the temperature was raised to 215 °C at 2.5 °C/min for 12 min. Then, the temperature was increased to 230 °C at 4 °C/min for 10 min; subsequently, it was increased to 280 °C at 10 °C/min for 15 min. Ion source temperature was set to 200 °C. electron bombardment energy was 70 eV. The mass scanning range was m/z 40–550, and the solvent removal duration was 4.4 min. The FAMEs were quantified by the standard compounds. Electronic nose analysis The PEN3 electronic nose (PEN3 E-Nose, Airsense, Germany) consists of an array of 10 metal oxide sensors used to determine the odor profile of Lipu taro pork at various cooking stages; their descriptions are detailed in Table S1 (Zhou, Shi, et al., 2022). A 3 g portion of the pork sample was weighed, precisely, transferred to a 100 mL beaker, covered with plastic wrap, and allowed to stand for 30 min before testing. The direct headspace aspiration method was used with the CTC-PAL 3 Static Headspace Autosampler (CTC Analytics AG, Switzerland), in which the injection needle was inserted into a sealed sample cup containing the sample, and the sample was then measured with the electronic nose. Conditions for determination include a sampling time of 1 s per group, a self-cleaning duration of 80 s for the sensor, a zeroing time of 5 s for the sensor, and a sample preparation time of 5 s. The injection flow rate was 400 mL/min. The analysis sampling time was 80 s, and each sample group was tested in parallel three times. The intensities of 10 metal oxide sensors were measured and reported. Electronic tongue analysis The E-tongue of Lipu taro pork at various cooking stages was determined using the methodology of Miao et al. (2017), with necessary adjustments. Briefly, 10 g of pork samples were weighed, then 100 mL of distilled water was added for homogenization. The mixture was then centrifuged at 4000 rpm for 10 min, followed by filtration through gauze and filter paper to yield a clear solution for analysis. The sensors of the E-tongue (TS-5000Z electronic tongue, Japan INSENT company) were calibrated and diagnosed before the test. Afterward, the sensors were immersed in the extract solution for evaluation, and the sensor intensities were measured and reported. Headspace-gas chromatography-ion mobility spectrometry analysis The volatile flavor compounds of Lipu taro pork at various cooking stages were examined using the HS–GC–IMS (Flavor Spec Gas Phase Ion Mobility Spectrometer, G.S.A., Germany) according to the methodology outlined in prior literature (He et al., 2025). Briefly, 2 g of pork samples were precisely weighed and placed in a 20 mL headspace bottle. The samples were incubated at 60 °C for 15 min before injection, with each sample measured in triplicate. The incubation temperature is set to 60 °C, with a 15 min incubation and a splitless injection volume of 500 μL. Incubation speed: 500 r/min. Temperature of the syringe: 85 °C. The column utilized was an MXT-5 capillary column with dimensions of 15 m × 0.53 mm and a film thickness of 1 μm. Column temperature set at 60 °C. Carrier gas: high-purity nitrogen (purity 99.999 %); Programmatic pressurization involved maintaining an initial flow rate of 2.0 mL/min for 2 min, followed by a linear increase to 10 mL/min over 8 min, and a further linear increase to 100 mL/min within 10 min. The total chromatographic run time was 20 min. Inlet temperature is 80 °C. Ionization source: tritium (3H) source; Migration tube length is 98 mm, with an electric field strength of 500 V/cm and a migration tube temperature of 45 °C. Drift gas consists of high-purity nitrogen with a purity of at least 99.999 %, and a flow rate of 150 mL/min is utilized in positive mode. The retention index (RI) was established to identify volatile compounds by injecting a mixture of n-ketones (C4–C9). The RI drift periods were subsequently compared to the GC–IMS library standards and the NIST database. Additionally, the area normalization method was employed to determine the relative quantity of volatile compounds using the subsequent equation: Data analysis The GC–IMS data was processed using the laboratory analytical viewer (LAV) computing software. This software included a GC–IMS library search and the reporter. The LAV software can autonomously produce top views of volatiles in different stages of Lipu taro pork cooking, represented in three dimensions (peak intensity, migration time, and retention time) and two dimensions (migration time and retention time). The shifts and variations in volatiles were more clearly illustrated in the two-dimensional overhead perspective. The three-dimensional graph clearly depicts the fluctuations in volatile compounds at different stages of Lipu taro pork cooking. The program includes a gallery plot plug-in that generates the gallery plot. The Gallery plug-in provided volatile fingerprinting, while the dynamic PCA plug-in facilitated data clustering. Data were initially processed in Microsoft Excel 2019, followed by statistical analysis in SPSS Statistics 25.0 (p < 0.05 was considered statistically significant). Multivariate analyses, including PCA and PLS—DA, were performed using Origin 2019 and SIMCA 14.1. Results and discussion Changes in fatty acids during the processing of Lipu taro belly meat Table 1 presents the composition and fatty acid content of Lipu taro belly meat at various cooking stages. A total of 15 fatty acids were identified across all meat samples, comprising 5 saturated fatty acids (SFA), 5 monounsaturated fatty acids (MUFAs), and 5 polyunsaturated fatty acids (PUFAs). The specific fatty acids detected include methyl myristate (C14:0), methyl cis-7-hexadecenoic acid (C16:1n9c), methyl palmitoleate (C16:1n7c), methyl palmitate (C16:0), methyl heptachyate (C17:0), methyl linoleate (C18:2n6c), methyl oleate (C18:1n9c), methyl oleate (C18:1n9t), methyl stearate (C18:0), methyl arachidonate (C20:4n6c), cis-8, methyl 11,14-eicosanenoic acid (C20:3n6c), methyl eicosadienoate (C20:2n6c), methyl eicosadienoate (C20:1n9c), methyl arachidonate (C20:0), and ethyl arachidonate (C22:4n8c). Table 1Composition and content of fatty acids during the processing of Lipu taro belly.Table 1Fatty acidsContent (g/100 g)RPPMRMUCSMC14:00.012 ± 0.00bc0.020 ± 0.00a0.010 ± 0.00c0.013 ± 0.00b0.011 ± 0.00cC16:1n9c0.002 ± 0.00b0.004 ± 0.00a0.002 ± 0.00b0.002 ± 0.00b0.002 ± 0.00bC16:1n7c0.020 ± 0.00c0.035 ± 0.00a0.022 ± 0.00bc0.026 ± 0.00b0.021 ± 0.00cC16:00.174 ± 0.01c0.325 ± 0.03a0.182 ± 0.02c0.236 ± 0.01b0.161 ± 0.02cC17:00.001 ± 0.00b0.002 ± 0.00a0.001 ± 0.00b0.001 ± 0.00b0.001 ± 0.00bC18:2n6c0.110 ± 0.01c0.223 ± 0.02a0.101 ± 0.01c0.177 ± 0.01b0.101 ± 0.01cC18:1n9c0.246 ± 0.02c0.469 ± 0.05a0.245 ± 0.02c0.317 ± 0.01b0.223 ± 0.03cC18:1n9t0.018 ± 0.00c0.037 ± 0.00a0.022 ± 0.00c0.030 ± 0.00b0.020 ± 0.00cC18:00.079 ± 0.01c0.161 ± 0.02a0.086 ± 0.01c0.115 ± 0.01b0.070 ± 0.01cC20:4n6c0.007 ± 0.00c0.012 ± 0.00a0.007 ± 0.00c0.014 ± 0.00b0.007 ± 0.00cC20:3n6c0.001 ± 0.00c0.003 ± 0.00a0.001 ± 0.00c0.002 ± 0.00b0.001 ± 0.00cC20:2n6c0.004 ± 0.00b0.007 ± 0.00a0.003 ± 0.00b0.004 ± 0.00b0.003 ± 0.00bC20:1n9c0.004 ± 0.00b0.007 ± 0.00a0.003 ± 0.00b0.004 ± 0.00b0.003 ± 0.00bC20:00.001 ± 0.00bc0.003 ± 0.00a0.001 ± 0.00c0.002 ± 0.00ab0.001 ± 0.00bcC22:4n8c0.002 ± 0.00c0.003 ± 0.00ab0.002 ± 0.00bc0.003 ± 0.00a0.001 ± 0.00cSFA0.267 ± 0.02c0.512 ± 0.05a0.281 ± 0.02c0.369 ± 0.01b0.244 ± 0.03cMUFA0.290 ± 0.02c0.552 ± 0.06a0.295 ± 0.03c0.380 ± 0.01b0.269 ± 0.03cPUFA0.123 ± 0.01c0.248 ± 0.03a0.114 ± 0.01c0.200 ± 0.01b0.113 ± 0.01cUFA0.414 ± 0.03c0.799 ± 0.08a0.409 ± 0.04c0.580 ± 0.02b0.382 ± 0.04cUFA: SFA1.551 ± 0.01a1.560 ± 0.03a1.458 ± 0.02b1.575 ± 0.02a1.562 ± 0.02aPUFA: SFA0.462 ± 0.00c0.483 ± 0.01b0.407 ± 0.01d0.543 ± 0.00a0.462 ± 0.01cNote: Different letters in the same row indicate significant differences (P <0.05). Raw pork meat (RP), pre-cooked meat (PM), roasted meat (RM), ultrasonically cured meat (UC), and steamed meat (SM). Composition and content of fatty acids during the processing of Lipu taro belly. Note: Different letters in the same row indicate significant differences (P <0.05). Raw pork meat (RP), pre-cooked meat (PM), roasted meat (RM), ultrasonically cured meat (UC), and steamed meat (SM). Generally, a greater presence of unsaturated fatty acids (UFAs) in a product correlates with an elevated nutritional value (Simoes et al., 2019). The composition of fatty acids in Lipu taro belly meat encompasses SFA and UFA, with the latter being present in significantly greater quantities than the former. Among the components, MUFA was the most significant, ranging from 0.269 % to 0.552 %, with oleic acid exhibiting the highest concentration. The presence of oleic acid in vegetable oils and animal fats significantly enhances the flavor profile of meat. Oleic acid serves as a precursor to various compounds, including 2-heptenal, 1-heptanol, 2-octenal, 2-nonenal, 1-octanol, 2-decenal, and 1-nonenol, during lipid oxidation reactions (Al-Dalali et al., 2022b). Subsequently, SFA constitutes 0.244 % to 0.512 %, with palmitic acid and stearic acid being the predominant components, each exceeding 0.05 g per 100 g. The primary constituent of PUFA is linoleic acid (LA), and our findings align with those documented by Da et al. (2021). The method of cooking and the duration therein profoundly influence the fatty acid composition in pork. The reduction of fatty acids can be attributed to oxidation and degradation processes that occur during cooking, with SFA concentrations typically declining as cooking time extends (Yingqiu et al., 2016). Table 1 illustrates that the fatty acid content in the PM treatment exceeded that of pork at other cooking stages, with the UC group following, and the SM treatment exhibiting the lowest content. Fatty acids serve as a primary energy source for the body; however, an excessive consumption of SFAs may lead to elevated levels of specific indicators, including triacylglycerol, blood cholesterol, and LDL cholesterol. One of the key indicators linked to coronary heart disease is low-density lipoprotein cholesterol (Wereñska et al., 2021). Consequently, the World Health Organization (WHO) has been promoting a reduction in the consumption of SFAs in daily diets. The PUFA/SFA ratio is commonly used in nutrition to assess the nutritional quality of meat, with higher ratios indicating greater nutritional value, typically 0.45 or above (Mapiye et al., 2011). The PUFA/SFA ratio of pork in Lipu taro belly pork during processing exceeded 0.45, except for the RM group, which recorded a ratio of 0.407 (Table 1). This value is marginally below 0.45, suggesting that the nutritional value of pork in Lipu taro belly meat is elevated during processing. Furthermore, the fatty acids present in meat can indicate its nutritional value and influence the flavor profile of cooked meat products. During heating, the Maillard reaction and fatty acid oxidation occur rapidly, leading to the formation of various volatile compounds and potentially decreasing fatty acid levels (Han et al., 2021). According to Neff et al. (2014), UFAs can reduce cholesterol levels and combat atherosclerosis. Researchers have found that by chelating metal ions and scavenging hydroxyl radicals, Maillard reaction products can protect UFAs from autooxidation while processing (Huang et al., 2011). Compared to the other groups, the RM group had a somewhat lower UFA-to-SFA ratio, while the other groups showed less fluctuation. Another way to measure the post-processing flavor of a sample is to examine its SFA + MUFA content, as higher SFA + MUFA levels correlate with better tenderness, juiciness, and flavor in meat. In contrast, a higher PUFA content correlates with poorer tenderness, flavor, and juiciness (Camerona & Enser, 1991). Pork samples showed significantly higher overall amounts of SFA and MUFA than PUFA levels at various stages of cooking. In general, two factors affect the SFA + MUFA value: first, the evaporation of water and other moisture during the thermal process increases the SFA + MUFA value; second, oxidative cracking of SFA and MUFA at elevated temperatures decreases the SFA + MUFA value. These two factors are primarily related to the degree of processing (Camerona & Enser, 1991). Electronic nose analysis during the processing of Lipu taro belly meat E-noses can identify the odor of a sample, with minor variations in volatile compounds potentially leading to differences in sensor response (Chen, Chen, Xiao, et al., 2020). Data were collected via sample testing, with results from 118 to 120 s selected for PCA to illustrate differences among samples. Fig. 2a shows the change in the odor profile of pork during the processing of Lipu taro belly meat, with the corresponding sensor signal strength for each sample analyzed by the E-nose. The figure illustrates significant differences in signal strength among the W2W, W2S, W1W, W1S, and W5S sensors, with response values exceeding 1.0. This, in conjunction with the types of sensitive compounds detected by the sensors, suggests that the characteristic flavor of pork during the processing of Lipu taro belly meat may be attributed to inorganic sulfides, nitrogen oxides, methyl groups, alcohols, and aromatic compounds such as aldehydes and ketones. Conversely, no notable differences were observed in ammonia levels or certain aromatic compounds, hydrides, and long-chain alkanes. The response intensity of each sample was ranked as UC > RP > SM > PM > RM. Ultrasonic curing of pork exhibited lower response values on the W1C, W3C, W5C, W6S, and W3S sensors, whereas it achieved the highest values on the other sensors. This suggests that ultrasonic curing may facilitate the release of flavor compounds in pork. (a) Electronic nose radar map of pork; (b) PCA of the E-nose of pork during the processing of Lipu taro belly meat. Fig. 2b illustrates that the first PC accounts for 93.3 % and the second PC for 6.3 %, collectively representing 99.6 % of the total variance (exceeding 95 %), so demonstrating that PC1 and PC2 enclose the majority of the original information from the sample. The scatter plot shows a clear separation of samples across processing stages, indicating that the E-nose effectively distinguishes Lipu taro belly pork odors at these stages. The results indicated that when PC1 significantly exceeded PC2, an increase in the distance between samples along the abscissa corresponded to a greater disparity, with the distinction between the RP and UC treatments reaching its peak. The RP, PM, and RM groups were located on the left side, while the UC and SM groups were situated on the right side of the PCA, indicating substantial differences among these groups. Significant variances existed in the odors of RP, PM, and RM, although the odor characteristics of PM and RM were analogous. Electronic tongue analysis during the processing of Lipu taro belly meat The E-tongue mimics the human tongue to analyze, identify, and evaluate the tested sample, with the resulting data processed using multivariate statistical methods to swiftly convey the sample's overall quality information and facilitate its identification and classification. Fig. 3a presents the E-tongue radar chart for samples during the processing of Lipu taro belly meat. Based on the taste value of the reference solution, the sour taste and astringency aftertaste indices for each pork sample fall below the tasteless threshold, indicating an absence of sourness and astringency. Umami, saltiness, and sweetness exhibited elevated response values, serving as significant and prominent taste cues, followed by richness and bitterness. (a) Electronic tongue radar map of pork; (b) Bubble chart of umami, saltiness, and richness of pork; (c) Bitter and sweet bubble diagram of pork; (d) PCA of the electronic tongue of pork during the processing of Lipu taro belly meat. Figs. 3b and c illustrate that variations in processing technologies result in differing tastes of Lipu taro belly pork samples at each critical cooking stage. The RP treatment exhibits distinct sweetness, bitterness, and umami, with umami being the most pronounced, and its richness is similar to that of the SM group. Following PM treatment, the sweetness and richness of the pork intensified, whilst the bitterness and umami tastes diminished. The sweetness and bitterness of the RM group decrease while umami and richness are amplified, with saltiness approximating that of the RP group. Following UC treatment, the saltiness and richness of the pork attain their peak. The SM group in the final product exhibits the most pronounced bitterness, alongside the least saltiness and richness. Ultrasonic curing can markedly increase the salinity and richness of pork, while also enhancing its flavor, corroborating the findings of the prior investigation. Fig. 3d illustrates the PCA diagram, which reflects the overall relationship among the different cooking stages of Lipu taro belly meat. PC1 and PC2 represented 98.76 % of the total variation, exceeding 80 %, with PC1 and PC2 contributing 90.54 % and 8.22 %, respectively. This suggests that PC1 and PC2 accounted for the majority of the tasting information for pork (Salum et al., 2017), effectively illustrating variations in pork taste profiles at different stages. The pork samples at each processing stage were allocated to distinct regions without overlap, indicating substantial differences in overall taste across the stages. The proximity of samples on the PCA plot quantifies taste differences, with shorter distances indicating more similar tastes and vice versa (Moreno-Rojas et al., 2010). The RP, PM, RM, UC, and SM groups were positioned on the left and right sides of the PCA, indicating substantial differences among them. Ultrasonic curing significantly influenced the overall taste change in pork, and the findings aligned with the electronic nose measurements. Analysis of volatile flavor compounds during the processing of Lipu taro belly meat GC-IMS Spectrum analysis The GC-IMS system, characterized by high resolution and sensitivity, was used to assess the effect of different cooking stages on the volatile flavor components of Lipu taro belly pork. GC-IMS was used to analyze volatile compounds in pork at various cooking stages, with distinct volatiles identified in the comprehensive data presented in the two-dimensional map. Fig. 4(a) presents the three-dimensional GC-IMS spectra of volatile components in pork at various cooking stages during the processing of Lipu taro belly meat. The data showed distinct variations in volatile compounds across various pork samples. To facilitate a more natural comparison of changes in pork flavor compounds across various processing stages, the two-dimensional spectrum in Fig. 4(b) and the difference spectrum in Fig. 4(c) were derived by applying dimensionality reduction to the original spectrum. The regional points illustrate varying retention times (Y-axis), relative drift times (X-axis), and signal intensities (Z-axis), with a blue background representing lower concentrations and redder areas indicating higher concentrations of volatile substances (Christmann et al., 2022). From left to right: raw pork (RP), pre-cooked meat (PM), roasted meat (RM), ultrasonic-cured pork (UC), and steamed meat (SM). GC–IMS analysis of flavor compounds in pork samples during the processing of Lipu taro belly meat: (a) three-dimensional (3D) topographic, (b) two-dimensional topographic plots, (c) Difference spectrum. RP, raw pork meat; PM, pre-cooked meat; RM, roasted meat; UC, ultrasonically cured meat; and SM, steamed meat. Figs. 4(b) and 4(c) illustrate significant variations in the types and concentrations of volatile compounds across the samples. Using the spectrum of the RP group as a reference, the presence of a limited number of red spots indicates elevated levels of compounds at specific peak positions. Furthermore, the concentration of flavor compounds in pork at subsequent stages exceeds that of raw pork, suggesting that cooking and processing enhance the formation of flavor compounds. The UC and SM treatments exhibit elevated levels of volatile compounds, likely attributable to protein degradation during processing, thereby increasing flavor compounds. GC-IMS qualitative analysis The volatile compounds in pork during the preparation of Lipu taro belly meat were identified using the GC–IMS instrument, and Fig. 5 presents the qualitative spectra. The GC–IMS library was used for qualitative analysis and identification of volatile components, as shown in Table 2. In Fig. 6, the horizontal axis denotes drift time, whereas the vertical axis indicates retention time. In HS–GC–IMS, diverse volatile compounds produce unique product ions, such as M, D, and isomers, depending on their concentrations. The lifespans of these items are equivalent, though the drift time varies. The results demonstrate that Lipu taro belly pork samples at five cooking stages generated 60 volatile compounds, categorized into 6 esters, 16 aldehydes, 8 alcohols, 14 ketones, 2 acids, 6 terpenes, and 1 furan, in addition to 7 unidentified substances. Table 2 provides a comprehensive analysis of the volatile components, while Fig. 6a illustrates their relative quantities. Qualitative analysis of volatile compounds in pork during the processing of Lipu taro meat. RP, raw pork meat; PM, pre-cooked meat; RM, roasted meat; UC, ultrasonically cured meat; and SM, steamed meat. Table 2Determination of volatile flavor compounds of Lipu taro pork at different cooking points by GC-IMS.Table 2NumberCompound nameCASMolecular formulaMolecular weightRIRt[s]Dt[ms]Esters17Ethyl caproate-MC123660C8H16O2144.2999.1576.8111.3436518Ethyl caproate-DC123660C8H16O2144.2998.8576.3561.7950224Isoamyl acetateC123922C7H14O2130.2867.9347.2271.7506942Ethyl acetate-MC141786C4H8O288.1592.3138.061.0980843Ethyl acetate-DC141786C4H8O288.1591.5137.7721.3377753Ethyl lactateC97643C5H10O3118.1801.9274.5481.54029 Aldehydes1NonanalC124196C9H18O142.21100.9773.4351.4762(E)-2-OctenalC2548870C8H14O126.21062.2691.8991.333833OctenalC124130C8H16O128.21003.7584.4971.40416(E)-2-Heptenal-MC18829555C7H12O112.2952.6481.0311.255397(E)-2-Heptenal-DC18829555C7H12O112.2952.6481.0311.668829HeptanalC111717C7H14O114.2895.7383.8891.6786120BenzeneacetaldehydeC122781C8H8O120.21046.9662.031.2626930Hexanal-MC66251C6H12O100.2783.5257.1371.264331Hexanal-DC66251C6H12O100.2784.8258.3371.5600637Pentanal-MC110623C5H10O86.1688.4176.4061.18953392-Methylbutanal-MC96173C5H10O86.1654.9161.9731.16456403-Methylbutanal-MC590863C5H10O86.1638.2155.2131.17843413-Methylbutanal-DC590863C5H10O86.1635153.9341.4059482-Methylbutanal-DC96173C5H10O86.1655.6162.2611.4015549Pentanal-DC110623C5H10O86.1689.6177.2431.4246450(E)-2-HexenalC6728263C6H10O98.1844318.9121.18227 Alcohols51-Octen-3-olC3391864C8H16O128.2977.3530.5151.1589811Hexanol-MC111273C6H14O102.2866.8345.8681.3278512Hexanol-DC111273C6H14O102.2865343.6681.64018321-Pentanol-MC71410C5H12O88.1759.3233.5921.2561331-Pentanol-DC71410C5H12O88.1757.9232.3131.51316513-methyl-1-butanolC123513C5H12O88.1731.4209.1831.5029557EthanolC64175C2H6O46.1449.395.8291.14398602-MethylpropanolC78831C4H10O74.1615.1146.3261.37375 Ketones82-Heptanone-MC110430C7H14O114.2884.2367.9261.26107102-Heptanone-DC110430C7H14O114.2884.4368.2331.62907236-Methyl hept-5-en-2-oneC110930C8H14O126.2983.8544.3341.17718283-Octanone-MC106683C8H16O128.2984.5545.8671.31009293-Octanone-DC106683C8H16O128.2983.6543.951.72035353-Hydroxy-2-butanoneC513860C4H8O288.1710.2192.3021.331382-PentanoneC107879C5H10O86.1677.2171.4731.37168442-Butanone-MC78933C4H8O72.1562.6127.9761.06069452-Butanone-DC78933C4H8O72.1566.1129.1291.24761462,3-ButanedioneC431038C4H6O286.1552.8124.8071.1772447AcetoneC67641C3H6O58.1492.2106.9441.11897562-HexanoneC591786C6H12O100.2775.6249.1631.49544583-Pentanone-DC96220C5H10O86.1692178.911.35548593-Pentanone-MC96220C5H10O86.1692.2179.0711.11348 Acis154-Pentenoic acidC591800C5H8O2100.1896384.3281.4397819Benzoic acidC100527C7H6O106.1955.1485.7611.46903 Terpenoids13LinaloolC78706C10H18O154.31095.9762.4551.2185214EucalyptolC470826C10H18O154.31033.5636.9531.2986916LimoneneC138863C10H16136.21031632.3521.2182722α-pineneC80568C10H16136.2942.6462.271.2147425β-LeucineC28634891C10H16136.2968511.3551.2161926α-ThujeneC2867052C10H16136.2927.9436.1941.21474 Furan272-PentylfuranC3777693C9H14O138.2988.2553.921.25664 Others4–unidentified*0982.3541.1991.4171721–unidentified*0981538.291.4741934–unidentified*0726.7205.2741.4012736–unidentified*0718.7198.8791.0905952–unidentified*0743.2219.1581.5660554–unidentified*0701185.441.3944255–unidentified*0840.5314.9341.65606Note: M represents a monomer, D represents a dimer, and substances with the same CAS number are monomers; − “indicates that unidentified flavor compounds. RI: Relative retention index; Rt: Retention time; Dt: Relative migration time. The numbers in the table correspond to the identification numbers of volatile compounds identified in Fig. 6. Determination of volatile flavor compounds of Lipu taro pork at different cooking points by GC-IMS. Note: M represents a monomer, D represents a dimer, and substances with the same CAS number are monomers; − “indicates that unidentified flavor compounds. RI: Relative retention index; Rt: Retention time; Dt: Relative migration time. The numbers in the table correspond to the identification numbers of volatile compounds identified in Fig. 6. Relative contents of volatile components (a), GC-IMS fingerprint of volatiles in pork (b) during the processing of Lipu taro meat. The primary volatile flavor constituents of pork in the processing of Lipu taro belly meat are aldehydes, generated during heating, predominantly from the decomposition of fats and amino acids. This occurs primarily through the oxidative deamination and decarboxylation of amino acids via Strecker degradation during partial heat treatment (Al-Dalali et al., 2025). The high concentration of aldehydes and their low threshold render the aroma prominent, contributing to the meat's aromatic components (Chen, Chen, Qi, et al., 2020). Valeraldehyde emits a scent similar to grass and bananas; hexanal possesses a delicate fragrance of fresh grass; heptanal exudes a fruity aroma; and the odor of oil oxidation and frying characterizes nonanal. The oxidation by-products of oleic acid, linoleic acid, and arachidonic acid are hexanaldehyde and heptanal. Oleic acid is predominantly oxidized to produce nonanal (Al-Dalali et al., 2022a). The degradation of phenylalanine or linoleic acid yields benzaldehyde, which is readily discernible during heating and has a bitter almond odor. Phenylacetaldehyde is predominantly metabolized via the phenylalanine Strecker degradation pathway (Al-Dalali et al., 2022b). The extent of lipid oxidation, the generation of volatile compounds, and the transformation of specific aldehydes into other flavor constituents increased across various cooking stages, including pre-cooking, roasting, ultrasonic curing, and steaming, resulting in a gradual reduction in aldehyde content throughout the cooking process. The RP sample exhibited the greatest aldehyde level at 22.92 %, whereas the SM sample contained merely 4.44 %. Strecker breakdown and fat oxidation are the primary pathways for the production of alcohols that exhibit metallic and mushroom-like aromas. Most alcohols, unless used at very high concentrations or in unsaturated forms, do not significantly contribute to food flavor due to their high thresholds (Chung et al., 2002). Alcohols, including 1-octen-3-ol, 1-pentanol, and ethanol, are abundant in processed pork from Lipu taro belly meat and contribute significantly to the volatile flavor components. Of these, 1-octen-3-ol is the most abundant in the PM treatment and the least in the RP group, and it is made by oxidizing linoleic acid and has a flavor similar to cooked mushrooms, and contributes significantly to pork's flavor (Mottram et al., 1984). The PM treatment contains the highest concentration of 1-pentanol, which has a fermented bread flavor and gradually decreases with cooking and processing. The UC treatment yields the highest ethanol concentration because the ultrasonic effect effectively permeates the sauce, and ethanol is a characteristic volatile constituent of liquor. Due to the cavitation effect of ultrasonics, which accelerated the reduction reaction of aldehydes, amino acids, and sugars, as well as the oxidative degradation of fats, the relative alcohol content in the RP group was the lowest, in the UC treatment it was the highest, and in the SM treatment it was the lowest. This resulted in a decrease in the alcohol content in the SM treatment. Steaming can either oxidize alcohols to ketones and aldehydes or esterify them with acids to produce molecules having a ketone group. An excess of ethanol, a key component in ester formation, may result from the addition of Sanhua wine and other ingredients to ultrasonically cured pork. A distinctive flavor with a light sweetness and fruity aroma is imparted to pork by ketone compounds, which are generated when proteins react with sugars. These compounds can also be generated through the β-oxidation of fatty acids (Al-Dalali et al., 2022a). Several ketones, which are strongly associated with pork flavor, are influenced by factors such as cooking time, temperature, and mode (Li et al., 2022). Meat includes both saturated and unsaturated fats, as well as a high protein content. Amino acids in meat can be used to make branched-chain acids, and several short-chain ketones can be formed by reacting Strecker aldehydes. Several processes can alter the flavor of pork, including the Maillard reaction, the breakdown of amino acids, and the breakdown of unsaturated fatty acids, all of which produce ketones with low thresholds. Lipu taro belly pork is primarily composed of the following ketones: 3-hydroxy-2-butanone, acetone, 2-butanone, 2,3-butanedione, 2-heptanone, and 2-pentanone. These ketones have a distinct aroma characteristic of pork and cream, and due to their special characteristics, can enhance the aroma of meat. 3-Hydroxy-2-butanone is aromatic and tastes like cream and butter, two-heptanone is spicy and tastes like blue cheese, and 2-pentanone has a mild fermenting flavor of banana and fruit (Yanan et al., 2021). Throughout the thermal processing, the relative concentration of ketones decreased. The fingerprint and changes in the relative concentrations of volatile chemicals showed that the RP group had the highest ketone concentration at 57.46 %, followed by the RM treatment, and the UC treatment had the lowest ketone concentration at 17.14 %. Esters are generated from the esterification of free fatty acids and alcohols resulting from lipid oxidation in muscle tissue, which enhances the overall flavor of pork. The esters in pork emit aromas characteristic of fruits, spices, or flowers (Wang et al., 2018). Wang et al. (2018) showed that short-chain esters possess a fruity aroma, while long-chain esters exhibit a fatty aroma. Long-chain esters, shown by hexyl butyrate, have a somewhat oily aroma, whereas short-chain fatty acids and alcohol esters, such as ethyl acetate, exhibit nuanced aromas reminiscent of grapes and cherries (Carrapiso et al., 2015). Isoamyl acetate emits a banana-like scent and possesses a fruity-sweet flavor. Esters constitute a comparatively minor fraction of all volatiles. The ester concentration in the UC group was higher than in the other groups, reaching 37.46 %, with ethyl lactate being the most prominent at 21.24 %. The SM group had an ester content of 33.13 %, ranking second to the UC group. The RP group has a minimal ester content of 5.25 %. This may result from ongoing interactions between ethanol and acid during ultrasonic curing, which produce esters and thereby elevate ester concentration in pork. The pork at various cooking stages during the preparation of Lipu taro belly meat mostly contained two acids: 4-pentenic acid and benzoic acid, both present at low concentrations, except in the UC group, where the level was below 1 %. The origins of acid compounds are complex and may be derived from long-chain fatty acids. It possesses a high threshold and low content, so it contributes minimally to flavor development. Furan compounds are primarily generated through the Maillard reaction, which involves the interaction between amino acids and reducing sugars, as well as the pyrolysis of amino acids and thiamine, imparting a robust, meaty flavor frequently present in cooked products. Heterocyclic compounds, including furans and pyrazines, have been associated with roasted and high-pressure meat (Devine & Dikeman, 2014). Notably, 2-pentylfuran, an oxidation derivative of linoleic acid, was identified in the pork samples of Lipu taro betty meat, imparting a barbecue-like, ham-like aroma characterized by a low threshold and a desirable burnt aroma. Its concentration is predominantly observed in the PM and RM treatments. In addition, the terpene content in the RP treatment was 1.21 %, decreased to 0.78 % after PM treatment, increased continuously in subsequent processing stages, and reached its highest value at 5.75 % in the SM treatment, followed by the UC treatment, indicating that marinade and sauce seasoning accessories were essential sources of hydrocarbons. Fingerprint analysis of volatiles during the processing of Lipu taro belly meat To effectively compare the effects of stages of cooking on the volatile substance composition during the processing of Lipu taro belly pork, the program generates a fingerprint derived from the signal peaks of the volatile substances in each sample spectrum (Fig. 6b). Each row in the graph represents the overall signal peak. In contrast, each column illustrates the same volatile organic component among various samples. Each dot signifies a volatile component, with color denoting the concentration of the volatile compound. A higher content level results in a brighter color. Homologous chemicals comprise monomer and dimer structures, with variations in concentration-dependent signals and areas. The numerical value associated with the fingerprint denotes an unidentified compound. Fig. 6b illustrates that aldehydes, including 3-methylbutyraldehyde, 2-methylbutyraldehyde, and nonanal, were the predominant compounds in the RP group. Hexanal, valeraldehyde, heptanal, (E)-2-hexenal, (E)-2-heptenal, (E)-2-octenal, and other aldehydes were the predominant compounds identified in the PM group. Acetone, 2-butanone, 2-pentanone, 2-heptanone, and various other ketones were identified as the predominant compounds in the RM group. Ethyl caproate, ethyl 2-hydroxypropionate, 2-methylpropanol, 3-methylbutanol, phenylacetaldehyde, and additional compounds were the predominant substances identified in the UC group. Isoamyl acetate, methyl-5-hepten-2-one, limonene, α-pinene, and 1,8-cineole were identified as the predominant compounds in the SM group. The results of this study indicated that the volatile compounds present in pork at various cooking stages of Lipu taro belly meat primarily comprised esters, ketones, alcohols, and aldehydes. Principal component analysis (PCA) Fig. 7a shows the results of a PCA conducted using GC-IMS volatile component data of Lipu taro button pork at various cooking stages. From the picture, we can infer that PCA1 (55 %) and PCA2 (27.5 %) collectively account for 82.5 % of the total variance. The RP, PM, and RM groups were located on the right side of the PCA, while the UC and SM groups were positioned on the left side. This indicates significant differences between the groups and that the volatile components of pork undergo substantial changes as it progresses through various processing stages. Finally, volatile flavor components in pork can be successfully identified and distinguished at various processing stages using principal component analysis. PCA of volatile flavor substances of pork (a), PLS-DA (b), Replacement test model (c), VIP score chart (d), and cluster analysis of key flavor compounds in pork (e) during the processing of Lipu taro meat. PLS—DA model analysis PLS—DA is a supervised discriminant analysis method that reduces data dimensionality to facilitate visualization, discriminant analysis, and prediction of complex datasets. Additionally, it assesses model performance indicators through cross-validation and permutation testing. A PLS—DA model was established using 60 qualitative compounds as independent variables and pork samples at various cooking stages as dependent variables. The discriminant effect of the model is illustrated in Fig. 7b and c. The values R2X = 0.963, R2Y = 0.986, and Q2 = 0.958 suggest that the model demonstrates high reliability in predicting and distinguishing the flavor of pork at various cooking stages. The permutation test was conducted, yielding R2 values of (0.0, 0.332) and Q2 values of (0.0, −0.583). Two groups of models were fitted for verification, with all R2 values positioned above the horizontal axis. The negative Q2 regression line suggests that the model is reliable and free from overfitting, demonstrating strong model correlation (Orley & Powers, 2013). To further characterize differences in volatile compounds in Lipu taro belly pork samples at various cooking stages, the PLS—DA model was analyzed, yielding a VIP value map. Using VIP > 1 as a reference, an increase in the VIP value indicates a more pronounced difference in the volatile compounds of Lipu taro belly pork samples at various cooking stages. A total of 21 marked volatile compounds were screened, as illustrated in Fig. 7d. This includes 8 aldehydes, 8 ketones, 2 esters, 1 alcohol, 1 acid, and 1 unidentified substance. The VIP value of methyl-5-hepten-2-one exceeded 1.5, serving as the primary differentiating factor. Acetone, 2-pentanone, and ethyl caproate followed, each with a VIP value greater than 1.4. Cluster analysis of volatile key compounds The volatile key compounds of pork at various cooking stages in the Lipu taro belly pork were subjected to cluster analysis, facilitating the practical identification of significant flavor components, as illustrated in Fig. 7e. The figure analysis indicates that region A contains the volatile components characterized by a considerable presence of pork, primarily octenal. Zone B comprises volatile components that are significantly present in the PM samples, including phenylacetaldehyde, 4-pentenoic acid, and ethyl caproate-D. Region C comprises volatile compounds, including methyl-5-hepten-2-one, 3-pentanone-M, and isoamyl acetate, the latter of which is a significant odor associated with the SM samples. Region D exhibits a characteristic volatile odor related to the UC samples, primarily influenced by compounds such as acetone, 2-pentanone, 2-butanone-D, 2-heptanone-D, 3-pentanone-D, 3-octanone-M, and 1-pentanol-D. Region E comprises volatile components with significant concentrations of the PM samples in Lipu taro belly pork, including octenal, hexanal-D, trans-2-heptenal-M, valeraldehyde-M, heptanal, trans-2-heptenal-D, valeraldehyde-D, and 1-pentanol-D, among others. The results align with the GC-IMS fingerprint and PCA findings of key compounds. Conclusion This study investigates the effect of various processing stages on the flavor profile of Lipu taro belly pork, utilizing an electronic nose, an electronic tongue, and fatty acid analysis in conjunction with gas chromatography-ion mobility spectrometry. The findings indicated that the fatty acid content in the PM group was significantly higher than that in pork at other cooking stages, with the UC group following, and the SM group exhibiting the lowest content. A total of 60 volatile compounds were identified using GC-IMS technology, comprising 6 esters, 16 aldehydes, 8 alcohols, 14 ketones, 2 acids, 6 terpenoids, 1 furan, and 7 unidentified compounds, with aldehydes, ketones, alcohols, and esters being the primary volatile compounds differentiating pork across various processing stages. Twenty-one compounds with VIP values exceeding 1.0 were analyzed using the PLS—DA model. These compounds include phenylacetaldehyde, 4-pentenoic acid, ethyl caproate-D, methyl-5-hepten-2-one, 3-pentanone-M, isoamyl acetate, acetone, 2-pentanone, 2-butanone-D, 2-heptanone-D, 3-pentanone-D, 3-octanone-M, 1-pentanol-D, octenal, hexanal-D, trans-2-heptenal-M, valeraldehyde-M, heptanal, trans-2-heptenal-D, and valeraldehyde-D, among others. The VIP value of methyl-5-hepten-2-one exceeded 1.5, making it the most significant differentiating factor. This was followed by acetone, 2-pentanone, and ethyl caproate, each with a VIP value greater than 1.4. The composition of flavor compounds in pork exhibited significant differences during the processing of Lipu taro belly meat. The content of these compounds at various stages was higher than that of the RP group, particularly in the case of aldehydes and ketones. This suggests that cooking and processing enhance the formation of flavor compounds. Subsequent research can integrate gas chromatography–mass spectrometry, gas chromatography-olfaction-mass spectrometry, metabolomics, and other technologies to establish a scientific foundation for understanding the flavor-formation mechanism of Lipu taro belly meat, thereby improving its flavor quality. CRediT authorship contribution statement Zhigui He: Resources, Project administration, Methodology, Investigation, Funding acquisition, Data curation, Conceptualization. Yan Hu: Methodology, Formal analysis, Data curation. Yuqing Zhang: Visualization, Software, Resources, Methodology, Formal analysis, Data curation. Yingying Cui: Visualization, Software, Resources, Methodology, Formal analysis, Data curation. Yingying Liu: Formal analysis, Data curation. Saeed A. Asiri: Visualization. Sam Al-Dalali: Writing – review & editing, Writing – original draft, Validation, Project administration, Investigation, Funding acquisition, Formal analysis. Ethical approval This article does not contain any studies with human participants or animals performed by any of the authors. 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. Acknowledgments This work was financially supported by the High-level Talent Research Funds of Guilin Tourism University (C00424002), the Guangxi Zhuang Autonomous Region's level financial science and technology plan project, Guangxi Key Research and Development Plan (AB19110023), and the Regional Innovation Cooperation Project of Sichuan Province (2024YFHZ0207). Footnotes Supplementary data to this article can be found online at https://doi.org/10.1016/j.fochx.2025.103320. Supplementary data Supplementary material Data availability Data will be made available on request. References Associated Data Supplementary Materials Supplementary material Data Availability Statement Data will be made available on request.