Skip to main content
NCBI home page
Food Chemistry: X logoLink to Food Chemistry: X
. 2026 Jan 30;34:103612. doi: 10.1016/j.fochx.2026.103612

Key flavor compounds driven by microbiota in fermented cowpeas: Correlation analysis based on GC–MS/GC-IMS and amplicon sequencing

Zehui Li a, Longqun Fu b, Jiajia Song a, Huayi Suo a, Yu Zhang a,c,d,⁎
PMCID: PMC12906201  PMID: 41696635

Abstract

Fermented cowpeas, a traditional Chinese pickle, are valued for their characteristic taste and nutrition. This study examined flavor evolution and microbial dynamics during fermentation and storage via physicochemical analysis, volatile and nonvolatile compounds profiling, and microbial community assessment. 21 and 18 key discriminatory volatiles (VIP > 1) were identified via GC–MS and GC-IMS in cowpea samples, respectively. Multivariate analysis indicated that changes in flavor quality resulted from the combined effects of lactic acid, umami, and sweet amino acids, and 12 key aroma-active compounds (VIP > 1 and OAV > 1). Five core functional microorganisms (Debaryomyces, Companilactobacillus, Levilactobacillus, Chromohalobacter, and Microbacterium) were identified. Pearson correlation analysis revealed positive associations among these microorganisms and lactic acid, amino acids (Glu, Asp, Ala), and several key aroma-active compounds but negative associations with spermidine, tartaric acid, and Asn. These findings link microbial metabolism to flavor development in fermented cowpeas, revealing insights into fermentation-driven flavor modulation.

Keywords: Fermented cowpeas, GC–MS, GC-IMS, Key discriminatory volatiles, Microbial succession, Taste-active amino acids

Highlights

  • •

    Combining GC-IMS/GC–MS to track dynamic aroma changes in all cowpea fermentation stages.

  • •

    Ethyl octanoate, ethyl palmitate, 4-ethylphenol define fermented cowpea aroma.

  • •

    The core microorganisms were identified as Companilactobacillus, Debaryomyces, etc.

  • •

    Filling the research gap on aroma profile evolution across different fermentation periods.

1. Introduction

Paocai, a traditional Chinese fermented vegetable, is significant in East Asian cuisine owing to its deep cultural heritage and distinct flavor profile. Recent research on specific paocai varieties, such as pickled radish and chili peppers, has advanced rapidly. Studies using microbial community analysis and multiomics approaches, including metagenomics and metabolomics, have revealed the dominant role of core functional microbiota, particularly Lactobacillus, in fermentation processes (Yang et al., 2022). This work has established a theoretical foundation linking microbial dynamics to flavor development and laid the groundwork for industrial standardization. Fermented cowpeas are a traditional vegetable product favored in China for their unique nutrition and flavor (Chen et al., 2025). They are valued for their high levels of plant proteins, dietary fiber, polyphenols, and bioactive peptides (Jin et al., 2023). Lactic acid bacteria, especially Levilactobacillus and Lactiplantibacillus, is the core functional microbial group involved in the traditional fermentation process of cowpeas (Guo et al., 2021). Meanwhile, the Maillard reaction and proteolysis are the primary contributors to quality degradation in fermented cowpeas during storage (Chen et al., 2025). Current research remains largely limited to preliminary microbial community profiling and the observation of storage-induced changes. The key metabolic interactions between microbial communities and flavor-active compounds are still insufficiently understood. This critical gap in knowledge regarding flavor formation mechanisms significantly impedes efforts to improve product quality and market competitiveness and optimize processing techniques.

Flavor, the core quality attribute of fermented cowpeas, results from the synergistic effects of volatile and nonvolatile compounds. Analytical methods for flavor characterization are increasingly shifting toward multidimensional and integrative strategies. Gas chromatography–mass spectrometry (GC–MS) is effective for identifying volatile compound structures through spectral matching against reference libraries, but it is limited in its ability to directly detect aroma-active compounds that define sensory quality (Zhao et al., 2021). Gas chromatography–ion mobility spectrometry (GC-IMS) offers high sensitivity, superior separation capability, three-dimensional spectral visualization, and direct spectral comparison via specialized software, making it suitable for real-time flavor analysis. However, the limited GC-IMS reference database constrains its application (Chen et al., 2021). GC–MS plus GC-IMS offer complementary strengths, enabling more comprehensive characterization and accurate profiling of volatile flavor compounds.

Because the flavor of fermented cowpeas originates from microbial metabolism, understanding the contributions of microorganisms is essential. To address this, the current study investigates fermented cowpeas using physicochemical analyzes, volatile and nonvolatile metabolite profiling, and high-throughput sequencing to characterize microbial community dynamics throughout fermentation and storage. By integrating GC–MS and GC-IMS platforms, we constructed a detailed profile of volatile compounds, whereas Illumina MiSeq sequencing elucidated patterns of microbial succession. Correlation network analysis further revealed the relationships between microbial metabolic activity and the formation of flavor compounds. These findings enhance the mechanistic understanding of flavor development in fermented cowpeas and provide a theoretical framework for improving fermentation processes and product quality, which are essential for driving innovation and sustainability in the fermented vegetable industry.

2. Materials and methods

2.1. Sample preparation

Cowpea samples at different fermentation periods were kindly provided by Chongqing Fuling Zhacai Group Co., Ltd., Chongqing, China. Fresh cowpeas were cleaned, dried, and then added to a clean container, supplemented with 5 g/100 g of salt, followed by sealing of the container and allowing the cowpeas to ferment at 25 °C. After 5 days of fermentation, additional salt was supplemented to achieve a concentration of 8 g/100 g. Following 30 days of fermentation, the salt content was further increased to 16 g/100 g and the fermentation was continued for a year. The three periodic samples were labeled raw cowpeas (RC), fermented cowpeas (FC), and cowpeas stored for 1 year after fermentation (AC). The samples from each fermentation period were divided into two portions, one was stored at 4 °C for physicochemical indices and volatile component analyzes and the other was stored at −80 °C for microbial analyzes.

2.2. Determination of physicochemical parameters

The pH value and salinity of the samples were determined by using a pH meter (PHS-3E, Shanghai INESA Scientific Instrument CO., Ltd., China) and a salinity meter (Shanghai LICHEN-BX Instrument Technology CO., Ltd., China), respectively.

Titratable acidity measurements were made as per the methods of Meng et al. (2024), albeit with some modifications. Titratable acidity was determined by titrating to pH 8.2 ± 0.2 with 0.01 mol/L NaOH and expressed as grams of lactic acid per 100 g of the sample.

2.3. Determination of nonvolatile compounds: Organic acids, Free amino acids, Biogenic amines, and Nitrite

The extraction and measurements of organic acids and free amino acids followed the methods described by Yang et al. (2022), albeit with some modifications. To determine the organic acids, a paste was prepared using 10 g of the sample and an equal volume of deionized water. Then, 5 g of this paste was weighed and supplemented with 15 mL of an 80% ethanol solution. The resulting mixture was then incubated in a water bath at 75 °C for 0.5 h, cooled to room temperature, and transferred into a 25-mL volumetric flask with deionized water. The mixture was centrifuged at 9000 ×g for 30 min, and the resulting supernatant was filtered through a 0.22-μm membrane filter. The samples were analyzed by high-performance liquid chromatography (U3000, Thermo Fisher Scientific, Waltham, MA, USA) with UV detection. Chromatographic separation was performed on a C18 column (5 μm, 250 mm × 4.6 mm) with a mobile phase comprising a solution of potassium dihydrogen phosphate (pH 2.8) and methanol in a 97:3 ratio (v/v). The HPLC parameters were set as follows: injection volume of 10 μL, column temperature of 30 °C, flow rate of 0.8 mL/min, and a detection wavelength of 210 nm.

Free Amino Acid Determination: 1 g of the sample was dried at 80 °C, ground into powder, and extracted with 0.1 mol/L hydrochloric acid. The mixture was sonicated for 40 min and then diluted to 25-mL volume with 0.1 mol/L hydrochloric acid in a volumetric flask. After filtration, 2 mL of the supernatant was derivatized with phenyl isothiocyanate and triethylamine. The derivatized lower solution was filtered through a 0.22-μm membrane filter. Free amino acids were quantified by HPLC (LC20A Shimadzu, Kyoto, Japan) with UV detection (equipped with a column measuring 5 μm, 250 mm × 4.6 mm). The gradient elution was performed with mobile phase A (50 mmol/L anhydrous sodium acetate solution, pH 6.5) and mobile phase B (methanol: acetonitrile: pure water = 1:3:1 v/v). The elution program was as follows: 0 min: 5% B, 7 min: 65% B, 39 min: 49% B, 40–50 min: 100% B, 51–60 min: 5% B. The HPLC parameters were set as follows: injection volume of 20 μL, column temperature of 35 °C, flow rate of 0.8 mL/min, and a detection wavelength of 254 nm.

The extraction and measurements of biogenic amines followed the methods described by Xiong, Xu, Liu, et al., 2024, albeit with some modifications. Briefly, the 10 g sample was mixed with 16 mL of 5% trichloroacetic acid solution and 2000 μL of internal standard solution (1 mg/mL 1,7-diaminoheptane). The mixture was then sonicated for 60 min and centrifuged at 2800 ×g for 8 min, and the supernatant was transferred into a 20-mL volumetric flask. The solution was diluted to the mark with 5% trichloroacetic acid. Then, 2 mL of this solution was derivatized with sodium chloride, sodium hydroxide solution, sodium bicarbonate solution, and dansyl chloride solution for 15 min. The reaction was terminated by adding 200 μL of the ammonia solution. The upper layer solution was filtered through a 0.22-μm membrane filter. Biogenic amines were analyzed using an HPLC (LC20A Shimadzu) with a C18 column (5 μm, 250 mm × 4.6 mm) and a UV detector. The gradient elution employed mobile phase A (pure water) and mobile phase B (acetonitrile). The elution program was as follows: 0 min: 55% B, 7 min: 65% B, 14–20 min: 20% B, 27 min: 90% B, 30–35 min: 100% B, 36–40 min: 55% B. The HPLC parameters were set as follows: injection volume of 20 μL, column temperature of 25 °C, flow rate of 1 mL/min, and a detection wavelength of 254 nm.

The nitrite content was measured by the naphthalene ethylenediamine hydrochloride method (Wei et al., 2023).

2.4. Determination of volatile flavor compounds

2.4.1. GC–MS analysis

The volatile compounds were extracted using SPME and separated and analyzed using GC–MS using the method described by Lian et al. (2024). The sample was chopped into particles of approximately 1-mm size. First, 3 g of the particles was added to a 15-mL headspace bottle, along with 5 μL of methyl heptanoate (internal standard, 17.58 g/L) (Shanghai Macklin Biochemical Co., Ltd., Shanghai, China). The bottle was then sealed and placed in a water bath at 75 °C for 45 min. Then, the solid-phase microextraction (SPME) fibers (Supelco Co., Ltd., Bellefonte, PA, USA) were inserted into the headspace of the vial for 20 min. At the end of the extraction period, the fiber was desorbed into the gas chromatography injection port for 5 min.

The determination was performed on GCMS-QP2010 (Shimadzu) equipped with a DB-5MS ultra inert (UI) column (30 m × 0.25 mm × 0.25 μm) under a pressure of 32.0 kPa and a flow rate of 1.0 mL/min. Briefly, the oven temperature was initially held at 35 °C for 4 min, then increased to 110 °C at a rate of 10 °C/min and held for 6 min, after which the temperature was raised to 150 °C at a rate of 5 °C/min and held for 2 min, finally heating to 230 °C at 7 °C/min and then held for 6 min. Helium was used as the carrier gas at a flow rate of 1.0 mL/min. The MS conditions were set as follows: ion source temperature of 230 °C, quadrupole temperature of 150 °C, scanning range of 50–550 mz-1, and ionization voltage of 70 eV. The spectrum peak was searched by the NIST14 database standard mass spectrometry library of the chemical workstation, and the unknown compounds were analyzed quantitatively. The relative area of methyl heptanoate served as an internal standard, and the volatile matter was quantitatively analyzed.

2.4.2. GC-IMS analysis

GC-IMS analysis was conducted with the flavorSpec® IMS instrument (G.A.S., Dortmund, Germany). A headspace sampling unit from CTC Analytics AG (Zwingen, Switzerland) was used for sample headspace extraction. The GC system had an MXT-WAX capillary column (15 m × 0.53 mm × 1 μm, Restek, USA). Then, 2 g of the cowpea samples were placed in a 20-mL headspace vial. The headspace injection conditions were as follows: in the splitless mode, incubation at 60 °C for 15 min with a shaking speed of 500 rpm, an injection needle temperature of 110 °C, and an injection volume of 500 μL. Nitrogen gas (99.999% purity) served as the carrier gas at programmed flow rates of 0–2 min: 2 mL/min, 2–10 min: 2–10 mL/min, 10–20 min: 10–100 mL/min, 20–30 min: 100 mL/min. Post-separation, the obtained analytes entered the IMS ionization chamber and were ionized at a constant temperature of 45 °C. High-purity nitrogen was used as IMS drift gas with a flow rate of 150 mL/min to ensure optimal performance.

Volatile compounds were qualitatively identified by cross-referencing RI values and the drift time was based on these established standards within the GC-IMS database. VOC data were collected and analyzed by using the VOCal analysis software. Qualitative analysis of VOCs was performed using the NIST and IMS databases in the software, and fingerprint differences were compared using the Reporter and Galerie Plot.

2.5. DNA extraction and sequencing of microorganisms in cowpeas

2.5.1. DNA extraction

The genomic DNA of cowpea samples was extracted by using the EZNA Soil DNA kit (Omega Bio-Tek, Norcross, GA, USA). The purity of the DNA was determined using 1% agarose gel, and the DNA concentration was determined by NanoDrop2000 (Thermo Fisher Scientific). For bacteria, the V5–V7 domains of 16S rRNA genes were amplified with primer 799F/1193R.11. For fungi, the internal transcribed spacer (ITS) regions were amplified with the primers ITS1F and ITS2R through a thermocycler polymerase chain reaction (PCR) system (GeneAmp 9700, Applied Biosystems, Foster City, CA, USA). The PCR products were recovered using 2% agarose gel electrophoresis. The purification of recovered products was achieved by the DNA Gel Extraction Kit (Axygen Biosciences, Union City, CA, USA), followed by validation using 2% agarose gel electrophoresis. The recovered products were quantified by Quantus™ Fluorometer (Promega, WI, USA). The amplicon library was generated using the TruSeqTM DNA Sample Prep Kit (Illumina, CA, USA). High throughput sequencing was performed using the Illumina Miseq PE300 platform (Beijing Novogene Bioinformation Science and Technology Co., Ltd., Beijing, China). For each measurement, three independent repetitions were performed.

2.5.2. Sequential data analysis

Illumina software was used to filter the sequencing readings to remove low-quality sequences. Flash software (version 1.2.11) was used for splicing, and the DADA2 plug-in in the QIIME2 process was used for noise reduction of optimized sequences after quality control splicing. Species taxonomic analysis of ASVs was performed using the naive Bayes classifiers in QIIME2, based on the SILVA (version 138) 16S rRNA gene database and UNITE (version 8) ITS databases. The PICRUSt2 software (version 2.2.0) was employed for functional prediction and analysis of 16S rRNA gene sequences, while Funguild was utilized for fungal functional prediction (Lian et al., 2024).

2.6. Statistical analysis

SPSS 27.0 software (SPSS Inc. Chicago, IL, USA) was used for one-way analysis of variance (ANOVA) and a minimum significant difference test. P < 0.05 indicated statistical significance. All measurements were expressed as the mean values ±standard deviations (SD). Simca 14.1 (Umetrics AB, Umea, Sweden) and Origin 2021 (Origin Lab Inc. Northampton, MA, USA) software was adopted. Each measurement was performed using three independent repetitions.

3. Results and discussion

3.1. Analysis of physicochemical parameters

Table 1 summarizes changes in titratable acidity, pH, and salt content of cowpeas across fermentation stages. Fermented cowpeas exhibited markedly lower titratable acidity than cowpeas stored for 1 year after fermentation, whereas pH values exhibited the opposite trend. This inverse pattern likely reflects the metabolic activity of lactic acid bacteria, which dominate the early fermentation phase and convert carbohydrates into organic acids, reducing pH over time (Li, Lin, et al., 2025). Salt content also varied significantly between stages. Salt serves dual functions: it enhances flavor development during fermentation and inhibits microbial spoilage during long-term storage (Lee et al., 2022).

Table 1.

The dynamic changes in physicochemical parameters during cowpea fermentation.

Raw cowpeas Fermented cowpeas Cowpeas stored for 1 year after fermentation
Titratable acidity (g / kg) 0.37 ± 0.01c 8.43 ± 0.18b 9.36 ± 0.31a
pH 5.50 ± 0.01a 3.60 ± 0.01b 3.56 ± 0.01c
Salinity (g/100 g) 0.01 ± 0.01c 11.93 ± 0.15b 15.07 ± 0.06a
Open in a new tab

Data with different superscript letters in the same row represent the significant difference between groups (p < 0.05).

3.2. Analysis of nonvolatile compounds

3.2.1. Analysis of organic acids

Organic acids are crucial contributors to taste, particularly owing to their sourness, which enhances food palatability and substantially influences the overall flavor profile. Five organic acids (tartaric, oxalic, malic, lactic, and ascorbic acids) were identified in the raw cowpeas (Fig. 1A). Citric acid appeared in two samples after fermentation, whereas succinic acid appeared only in cowpeas stored for 1 year after fermentation. Succinic acid, mainly derived through the TCA cycle, is associated with the metabolic activity of fungi such as Debaryomyces and Pichia, indicating elevated fungal energy metabolism and activity during cowpea fermentation (Tang & Peng, 2024). Compared with fermented cowpeas, the samples stored for 1 year after fermentation exhibited stable levels of tartaric, malic, and ascorbic acids but significantly lower oxalic, lactic, and citric acid levels. Lactic acid predominates during early fermentation due to lactic acid bacterial glycolysis; its decline in the cowpeas stored for 1 year after fermentation samples likely reflects further metabolism by acetic acid bacteria and yeasts (Xia et al., 2022). These changes collectively influence sourness and contribute to the complexity and quality of the final product.

Fig. 1.

Fig. 1

Open in a new tab

The dynamic changes in the fermentation of nonvolatile compounds during the cowpea. (A) Organic acids, (B) Free amino acids, (C) Biogenic amines, (D) Nitrite. (RC represents raw cowpeas, FC represents the fermented cowpes, AC represents cowpeas stored for 1 year after fermentation).

3.2.2. Analysis of free amino acids

Free amino acids (FAAs) are pivotal precursors in flavor formation, and their profiles reflect the interplay between microbial metabolism and fermentation progression. These interactions drive the dynamic restructuring of flavor networks. Based on their taste attributes, FAAs are categorized as umami (Asp and Glu), sweet (Thr, Ser, Gly, Ala, Lys, and Pro), or bitter (Val, Met, Ile, Leu, Phe, His, Arg, and Tyr) (Zhang, Zhang, et al., 2023). In both fermented cowpea samples, sweet amino acids predominated, with overall abundance following sweetness > umami (Fig. 1B). This pattern may reflect the preferential metabolism of sweet amino acids by dominant lactic acid bacteria, which convert them into various flavor compounds. Compared with raw cowpeas, the total FAA content increased by 39.01% in fermented cowpeas and 56.96% in cowpeas stored for 1 year after fermentation. Bitter amino acids, especially Val and Met, decreased slightly after fermentation, although their total content increased during 1 year of storage. However, bitterness is often masked by the high salinity and acidity characteristic of paocai (Xiong, Xu, Du, et al., 2024). Conversely, sweet amino acids are major contributors to the characteristic sensory profile of fermented cowpeas. Among all FAAs, Ala was predominant in both fermented cowpea samples, followed by Gly, Asp, and Glu, aligning with findings in Laotan suancai, where Ala and Glu dominate (Xiong, Xu, Du, et al., 2024).

3.2.3. Analysis of biogenic amines and nitrite

The types and levels of biogenic amines in cowpeas varied significantly across fermentation stages (Fig. 1C). Four biogenic amines (putrescine, β-phenylethylamine, spermidine, and tyramine) were detected in raw cowpeas. The total biogenic amine content increased threefold in fermented cowpeas compared to that in raw cowpeas, primarily due to microbial biosynthesis by dominant microbial strains such as Companilactobacillus and Tetragenococcus (Xiong, Xu, Liu, et al., 2024). Notably, the total biogenic amine content in cowpeas stored for 1 year after fermentation decreased to levels comparable to those of the raw materials, likely because increased salinity and decreased pH inhibited the metabolic activity of biogenic amine-producing bacteria (Zhang, Zhang, et al., 2023). Despite these fluctuations, all samples remained below the safety threshold of 1000 mg/kg (Gao et al., 2023). During fermentation, nitrate in vegetables is microbially converted to nitrite. Accordingly, both fermented cowpea samples displayed significantly higher nitrite levels than raw cowpeas (Fig. 1D). Importantly, all nitrite levels remained within the safe limit of 20 mg/kg (Xing et al., 2023).

3.2.4. Analysis of GC–MS

GC–MS-based metabolomics identified 95 modulated volatile compounds in the cowpea fermentation system (Table S1). Multivariate statistical analyzes (Fig. 2A–D) revealed that fermentation induced compositional and quantitative changes in volatile metabolite profiles. The fermented cowpeas contained 65 distinct volatiles, whereas cowpeas stored for 1 year contained 52. However, the total volatile concentration increased markedly, reaching levels 1.26 and 1.49 times higher than in the fermented and raw cowpeas, respectively. This suggests that extended storage alters the volatile profile by enhancing overall flavor intensity and modulating compound diversity through complex biochemical pathways. Metabolically, ester and alcohol levels increased, whereas aldehyde and ketone levels decreased. Ester concentrations in cowpeas stored for 1 year after fermentation were 3.08 times higher than those in fermented cowpeas, indicating active microbial esterification. Microbial ester synthesis, primarily via enzymatic esterification, imparts fruity aroma notes and masks undesirable odors by enhancing aromatic dominance (Liu et al., 2024). In addition, organic acids, precursors in esterification, are crucial in shaping the flavor profile. The flavor of fermented cowpea arises from the synergistic effect of multiple volatile compounds. For example, Ethyl octanoate imparts fruity notes, 3-Octanol contributes an earthy aroma, Linalool imparts rose notes, and Phenylacetaldehyde imparts sweet and floral scent, thus forming a complex and coordinated flavor profile. These compound-specific trajectories reflect how chemical equilibria and directed microbial metabolism interact during fermentation to preserve flavor consistency and product identity (Qin, Xiong, Xiong, et al., 2024).

Fig. 2.

Fig. 2

Open in a new tab

Identification and analysis of volatiles in cowpea at different fermentation stages by GC–MS. (A) The pie chart of the categories of volatile compounds of the raw cowpeas (RC), (B) The pie chart of the categories of volatile compounds of the fermented cowpeas (FC), (C) The pie chart of the categories of volatile compounds of the cowpeas stored for a year after fermentation (AC), (D) Comparison of volatile components in cowpea at different fermentation stages.

3.2.5. Analysis of GC-IMS

GC-IMS is a nontargeted analytical method that identifies volatile compounds based on their retention and drift times (Feng et al., 2022). This analysis detected 65 volatile compounds in cowpeas across fermentation stages (Table S2), supported by three- and two-dimensional spectral visualizations (Figs. 3A and B). Using the raw cowpea spectrum as the reference, we observed stage-specific differences in volatile compound profiles. Most signal peaks emerged between 0 and 1200 s retention time and 1.0–2.0 ms drift time, whereas the cowpeas stored for 1 year after fermentation showed additional peaks between 1400 and 1600 s (Fig. 3C). Principal component analysis revealed clear differentiation across fermentation stages, with tight clustering of replicates and distinct group separation (Fig. 3D). Parallel samples were closely clustered, the separation between samples was high, and the differences between groups were obvious.

Fig. 3.

Fig. 3

Open in a new tab

Identification and analysis of volatiles in cowpeas in different fermentation stages by GC-IMS. (A) 3D chromatograms, (B) 2D chromatograms, (C) Comparison of chromatograms, (D) Principal component analysis (PCA), (E) Fingerprints of VOCs. (RC represents raw cowpeas, FC represents the fermented cowpeas, AC represents cowpeas stored for 1 year after fermentation).

To further explore stage-dependent changes, volatile compound fingerprints were generated. Each column represents a volatile compound, each row corresponds to a sample stage, and signal intensity (colour depth) indicates the compound concentration. Fingerprint maps of cowpeas were segmented into three distinct regions of A, B, and C, corresponding to raw cowpeas, fermented cowpeas, and cowpeas stored for 1 year after fermentation, respectively (Fig. 3E). Zone A was enriched in 1-Octen-3-one, (Z)-4-heptenal, 4-Heptanone, 1-Hexanal, and 3-Octanone, compounds associated with mushroom and vegetal aromas characteristic of raw cowpeas. Zone B was dominated by 2-Methyl propanal, Linalool, and Butanoic acid ethyl ester, marking the onset of fermentation. Zone C was enriched by esters, acids, and aldehydes, reflecting cumulative microbial metabolic outputs during extended fermentation. Notably, increased levels of some volatile compounds in cowpeas stored for 1 year after fermentation (such as esters including acetic acid butyl ester and ethyl propanoate) enriched fruity aroma notes and masked off-odors, improving overall product acceptability (Li, Chen, et al., 2025). Concurrently, Acetic, 1-Butanoic, and Propionic acids contributed to strong acetic acid notes, likely generated by the metabolic activities of microorganisms such as lactic acid bacteria. Overall, microbial metabolism drives the flavor transformation in fermented cowpeas from fresh vegetal tones to a complex fermented signature through bidirectional pathways of synthesis and degradation.

3.2.6. Comparative analysis of GC–MS and GC-IMS data

Orthogonal partial least squares discriminant analysis (OPLS-DA) was employed to identify differential volatile organic compounds (VOCs). This approach utilizes correlations within the dataset to visualize and quantify aroma differences among sample groups (Kang et al., 2022). For the GC–MS dataset, the resulting model demonstrated strong performance, with R2X, R2Y, and Q2 values of 0.918, 0.996, and 0.982, respectively, indicating excellent explanatory power and predictive ability (Fig. 4A). Cross-validation ANOVA and permutation tests further confirmed model stability and reliability, as both R2 (0.629) and Q2 (−0.653) were below the critical threshold of 1.0 (Fig. 4B). In OPLS-DA, the VIP score reflects the discriminatory power of each compound, with VIP > 1 considered significant. VIP values for the 95 volatile compounds detected by GC–MS are shown in Fig. 4C. Twenty-one volatiles with VIP > 1 were identified, including 9 esters, 7 alcohols, 3 aldehydes, 1 phenol, and 1 acid. Fig. 4D displays the biplot illustrating the relationship between compounds and samples. In this plot, compounds located closer to a given sample indicate higher relative content in that sample, signifying their significant contribution to sample classification and their role as key discriminatory volatiles (Huang et al., 2025). The three cowpea samples are distributed across different quadrants (Fig. 4D). The raw cowpeas are located in the fourth quadrant and are closely associated with compounds 38 (1-Nonanol), 57 (Benzaldehyde), 66 (4-Propyl-3-hexene), 6 (Isopropyl myristate), 60 (Tetradecanal), 16 (Leaf acetate), and 61 (Octadecanal). Among these, only Benzaldehyde, Isopropyl myristate, and Leaf acetate had VIP > 1, identifying them as key discriminatory volatiles in the raw cowpeas. The remaining compounds, although more abundant in the raw cowpeas, contributed minimally to group separation due to their lower VIP scores. The fermented cowpea sample is located in the second quadrant, with key discriminatory volatiles (VIP > 1) such as compounds 40 (Phenethyl alcohol) and 19 (Formic acid, hexylester). The 1-year-stored fermented cowpea sample lies in the third quadrant, characterized by key discriminatory volatiles such as compounds 9 (Ethyl caprylate), 1 (Octyl salicylate), 8 (Homosalate), 24 (Ethylene glycol monohexanoate), 35 (5-Ethenyltetrahydro-2-furanmethanol), and 48 ((−)-Alpha-terpineol).

Fig. 4.

Fig. 4

Open in a new tab

OPLS-DA score plot and 200-permutation test results for GC–MS and GC-IMS data. (A) score plot based on GC–MS data, (B) permutation test for the OPLS-DA of GC–MS data, (C) VIP scores calculated for volatile compounds on GC–MS data, the red portion represents the key differential compounds with VIP >1, (D) biplot of volatile compounds versus samples, (E) score plot based on GC-IMS data, (F) permutation test for the OPLS-DA of GC-IMS data, (G) VIP scores calculated for volatile compounds on GC-IMS data, the red portion represents the key differential compounds with VIP >1, (H) biplot of volatile compounds versus samples on GC-IMS data. (RC represents raw cowpeas, FC represents the fermented cowpeas, AC represents cowpeas stored for 1 year after fermentation. Each number in Fig. 4C and Fig. 4D corresponds to a substance, with names provided in Table S1. Each number in Fig. 4G and Fig. 4H corresponds to a substance, with names provided in Table S2.). (For interpretation of the references to colour in this figure legend, the reader is referred to the web version of this article.)

OPLS-DA of the GC-IMS dataset showed clear separation among all samples (R2X = 0.994 and Q2 = 0.999), indicating a well-fitted model (Fig. 4E). After 200 permutation tests (Fig. 4F), the Q2 regression line intersected the vertical axis below zero, confirming the absence of overfitting and validating model effectiveness. These results demonstrate that GC-IMS effectively distinguishes the three cowpea samples. VIP scores for the 65 volatile compounds detected by GC-IMS are shown in Fig. 4G. Eighteen compounds had VIP > 1, including 9 esters, 4 alcohols, 3 acids, 1 aldehyde, and 1 ketone. The compound–sample biplot (Fig. 4H) revealed that key discriminatory volatiles in the raw cowpeas included compounds 5 ((Z)-3-hexen-1-ol), 22 (Butanoic acid, 3-methylbutyl ester), 21 ((E)-ethyl-2-hexenoate), and 41 (3-Octanone). The main volatile in fermented cowpeas was compound 31 (Ethyl 2-butenoate). Key discriminatory volatiles in 1-year-stored fermented cowpeas included compounds 24 (Ethyl 2-hydroxypropanoate), 62 (Propanoic acid), 61 (1-Butanoic acid), 39 (Ethyl propanoate), and 38 (Acetic acid propyl ester).

Comparison of GC–MS and GC-IMS results showed eight VOCs detected by both techniques: Linalool, 1-Octen-3-ol, 3-Ethyl-4-methylpentan-1-ol, 3-Octanol, Ethyl octanoate, 3-Octanone, 2-Ethyl hexanol, and Benzaldehyde. The combined aroma profiles from both platforms indicated that raw cowpeas exhibited green, fruity, banana, apple, and grassy notes; fermented cowpeas exhibited floral and rose scents; and 1-year-stored fermented cowpeas exhibited fruity, pineapple, apple, brandy, and floral aromas. The combined use of GC–MS and GC-IMS offers complementary analytical strengths and yields consistent identification of key flavor compounds in fermented cowpeas. Both approaches detected and characterized essential aroma-active compounds (including esters, alcohols, and aldehydes), providing complementary evidence for defining the core flavor profile. Both methods effectively differentiated samples across fermentation stages. The quantitative GC–MS data and GC-IMS fingerprint profiles produced highly consistent clustering patterns in multivariate analysis, reinforcing the reliability of the observed fermentation dynamics. Moreover, GC–MS offered high-resolution identification of 95 volatiles, covering a wide range of long-chain compounds (C8–C21), including phenolic substances, across a broad volatility spectrum that includes high-boiling-point constituents. In contrast, GC-IMS enabled rapid, nontargeted detection and visual mapping of 65 volatiles, predominantly short-chain molecules (C2–C10) and facilitated clear differentiation among fermentation stages via fingerprinting and regional clustering (Zones A–C, Fig. 3E). This integrated approach enhances detection sensitivity, structural resolution, and temporal tracking of volatile evolution, offering multidimensional insight into the aromatic complexity of fermented cowpeas (Huang et al., 2025).

3.2.7. Analysis of odor activity value

The contribution of volatile compounds to the characteristic flavor of fermented cowpeas depends on their concentrations and odor thresholds, which are critical reference parameters. The odor activity value (OAV), defined as the ratio of a compound's concentration to its odor threshold, quantifies the impact of a single volatile component on the overall flavor profile. Compounds with OAV > 1 are considered key aroma contributors, with higher values indicating greater sensory relevance (Xiang et al., 2023). To improve screening accuracy, stricter criteria (OAV > 1 and VIP > 1) were used to identify key aroma-active compounds. Detailed information for volatiles with OAV >1 derived from GC–MS is listed in Table 2; all compounds included in the OAV analysis met the criterion VIP >1.

Table 2.

Volatile components of cowpea species OAV >1 in different fermentation periods.

No. Compounds CAS Odor
Thresholda (μg/kg)		 OAVb
 Odor descriptionc
RC FC AC
Alcohol
30 1-Octen-3-ol 3391-86-4 1 40,886.23 7779.33 17,685.08 mushroom, lavender, rose, hay
32 Linalool 78–70-6 6 561.99 2002.87 2160.25 citrus, rose, woody, blueberry
31 3-Octanol 589–98-0 78 169.22 366.94 472.30 earth, mushrooms, herb, melon, citrus, woody
40 Phenethyl alcohol 60–12-8 140 ND 46.15 ND floral, rose
Ester
9 Ethyl octanoate 106–32-1 5 ND 659.32 1593.52 fruity, pineapple, apple, brandy
16 Leaf acetate 3681-71-8 8 713.16 ND ND green, fruity, banana, apple, grassy
1 Octyl salicylate 118–60-5 40 19.77 40.46 268.85 mild, orchid, sweet, balsam
7 Ethyl palmitate 628–97-7 2000 ND 1.51 5.17 fruity, creamy
Aldehyde
51 Phenylacetaldehyde 122–78-1 4 788.36 372.85 1547.92 green, sweet, floral, hyacinth, honey, cocoa
58 2-Hexenal 505–57-7 30 268.19 457.53 ND sweet, almond, fruity, green, apple, plum, vegetable
57 Benzaldehyde 100–52-7 350 106.48 6.04 ND bitter almond, cherry, nutty
Phenol
81 4-Ethylphenol 123–07-9 21 ND 604.39 100.35 smoke
Open in a new tab

ND, not detected.

a: Odor threshold units of different volatile flavor substances are given in μg/kg, based on water as the solvent Threshold values of all flavor compounds are sourced from the book named: odor Thresholds: Compilations of Odor Threshold Values in Air, Water and Other Media (second enlarged and revised edition).

b: OAVs are calculated by dividing the average concentration by the odor threshold in water.

c: Odor descriptors are sourced from https://pubchem.ncbi.nlm.nih.gov/. and https://thegoodscentscompany.com/.

(RC represents raw cowpeas, FC represents the fermented cowpeas, AC represents cowpeas stored for 1 year after fermentation).

The key aroma-active compounds common to all three samples were 1-Octen-3-ol, Linalool, 3-Octanol, Octyl salicylate, and Phenylacetaldehyde. 1-Octene-3-ol, a signature C8 volatile of Lentinus edodes, imparts mushroom, lavender, and floral notes (Jung et al., 2021). Linalool, a monoterpene with a low odor threshold, exhibited higher intensity among alcohols and contributed a fresh aroma (Jeong et al., 2025). Phenylacetaldehyde adds floral and fruity notes and enhances aromatic complexity in fermented foods such as kimchi (Zhang, Li, et al., 2023). Together, these compounds contribute to the unique flavor of cowpeas, and some increase in abundance during fermentation. Ethyl octanoate, Ethyl palmitate, and 4-Ethylphenol were the main contributors to the unique aroma during fermentation. Ethyl octanoate, associated with pineapple, strawberry, and sweet notes, is typically produced during malolactic fermentation (Wang et al., 2025). Phenethyl alcohol also emerged as a specific marker compound; it imparts floral and rose aroma and is commonly found in fermented foods such as paocai and Baijiu (Li, Chen, et al., 2025). Overall, the unique flavor profile of fermented cowpeas is shaped by these differential volatiles. The mushroom flavor of 1-Octene-3-ol and 3-Octanol, the floral and fruity notes of Linalool and Ethyl palmitate, and the complex sensory contributions of compounds such as Ethyl octanoate define the sensory characteristics of the product throughout fermentation.

3.3. Analysis of changes in microbial community structure during cowpea fermentation

Microbial diversity during cowpea fermentation was investigated through high-throughput sequencing. The Chao and Shannon indices of bacterial diversity decreased initially and then increased, whereas the Simpson index displayed the opposite trend (Fig. 5A–C). For all samples, sequencing coverage exceeded 0.99, indicating that the sequencing depth was sufficient for accurately assessing microbial diversity (Fig. 5D). Similar trends were observed for the fungal community (Fig. S1), with coverage indices also exceeding 0.99. Moreover, the number of unique ASVs in raw cowpeas, fermented cowpeas, and cowpeas stored for 1 year was 91, 5, and 194, respectively (Fig. 5E). Only 59 ASVs were shared among the three samples, indicating significant shifts in the microbial community structure during fermentation. Principal coordinate analysis (PCoA) combined with ANOSIM (R = 0.9177, p = 0.001) confirmed these differences (Fig. 5F).

Fig. 5.

Fig. 5

Open in a new tab

The alpha diversity indices: (A) Chao 1, (B) Shannon index, (C) Simpson index, and (D) Coverage index (0.01 < P ≤ 0.05 is marked as *, 0.001 < P ≤ 0.01 is marked as * *, and P ≤ 0.001 is marked as * *). (E) The Venn diagram reveals the unique ASVs and shared ASVs across different fermentation stages. (F) The beta diversity analysis uses PCoA based on the Bray-Curtis algorithm. (G) The relative abundance of the bacterial microbial community composition in different fermentation stages at the genus level. (H) The relative abundance of the fungal microbial community composition in different fermentation stages at the genus level. The LEfSe analysis determined the biomarker across the varied fermentation rounds: cladogram (I) and bar plot (J). (K) The red portion represents the core microorganisms with VIP >1. (RC represents raw cowpeas, FC represents the fermented cowpeas, and AC represents cowpeas stored for 1 year after fermentation). (For interpretation of the references to colour in this figure legend, the reader is referred to the web version of this article.)

Microbial succession contributes substantially to shaping the unique flavor profile of fermented foods, with different microbial taxa performing distinct functions at each stage. At the phylum level (Fig. S2), a shift from Pseudomonadota dominance in raw cowpeas to Bacillota dominance in both fermented cowpea samples was observed, with relative abundances of 90% and 60%, respectively. This transition is typically associated with decreasing pH and the development of anaerobic conditions (He et al., 2020). At the genus level (Fig. 5G), raw cowpeas exhibited a more diverse bacterial community, including Microbacterium, Pantoea, Pseudomonas, Staphylococcus, and Enterococcus. Following fermentation, Companilactobacillus and Levilactobacillus became dominant (up to 65% and 28%), although their relative abundance decreased to 36% and 17% after 1 year of storage. Lactiplantibacillus and Weissella were also relatively abundant (approximately 4% and 2%) in both fermented samples. The nomenclature of lactobacilli mentioned in this study follows the latest authoritative taxonomic revision (Zheng et al., 2020), which reclassified the former broad genus Lactobacillus into 25 genera based on whole-genome phylogenetic analysis. This updated classification system has been widely adopted by international repositories and regulatory bodies. Notably, bacterial diversity increased during long-term storage, with genera such as Salinivibrio, Chromohalobacter, Halomonas, and Brevundimonas increasing in abundance. These successional changes likely reflect environmental shifts (such as pH and salinity) that favor metabolically adaptable microorganisms (Rovira et al., 2023). Fungal communities also underwent significant changes. Although raw cowpeas contained a diverse fungal population, fermentation led to dominance by Ascomycota (99%) (Fig. S2), particularly Debaryomyces in both fermented samples. Minor populations of Pichia and Starmerella were also observed (Fig. 5H). These changes likely reflect nutrient depletion and metabolite accumulation that create selective conditions favoring specific fungi (El-Shahir et al., 2020).

LEfSe analysis identified microbial biomarkers associated with different fermentation stages (Fig. 5I and J). Microbacterium and Pantoea were enriched in raw cowpeas. In fermented cowpeas, Companilactobacillus was determined as a biomarker. Lactiplantibacillus, Chromohalobacter, and Halomonas were biomarkers in cowpeas stored for 1 year. These results suggest that microbial biomarkers correspond closely to the specific ecological niches formed during fermentation, similar to those observed in liquor fermentation (Yang et al., 2023).

Orthogonal projections to latent structures (OPLS) revealed core functional microorganisms based on three key criteria: (1) consistent presence throughout fermentation, (2) VIP > 1, and (3) significant correlation with volatile flavor compounds (Wang et al., 2020). Microbial genera fulfilling these criteria were considered principal contributors to flavor development. The OPLS model exhibited strong explanatory power and predictive accuracy (R2 = 0.902, Q2 = 0.737). VIP analysis quantified the contributions of specific microbial groups to flavor formation (Fig. 5K). Five genera (Debaryomyces, Companilactobacillus, Levilactobacillus, Chromohalobacter, and Microbacterium) were identified as dominant microbial drivers and key determinants of flavor profiles.

3.4. Correlation analysis between nonvolatile and volatile compounds and microbial community

The relationships among microbial communities, physicochemical properties, and nonvolatile and volatile compounds were analyzed through Pearson correlation analysis (Fig. 6). Companilactobacillus, Levilactobacillus, Lactiplantibacillus, and Debaryomyces exhibited strong positive correlations with key physicochemical parameters (total acid, salinity, and nitrite) and metabolites such as putrescine, cadaverine, histamine, lactic acid, malic acid, citric acid, GABA, and cystine. These results suggest that Lactic acid bacteria, the dominant fermentative microbe, lowers pH via organic acid production and generates flavor precursors through amino acid metabolism (Liu et al., 2021). Additionally, most microorganisms exhibited negative correlations with biogenic amines, consistent with the findings of Xiong, Xu, Liu et al. (2024)Xiong et al. (2024). At the amino acid metabolism level, Companilactobacillus, Levilactobacillus, Lactiplantibacillus, halophilic microorganisms (Chromohalobacter and Halomonas), Pichia, and Debaryomyces were positively correlated with umami (Glu and Asp) and sweet amino acids (Gly and Ala). These relationships suggest metabolic specialization: halophilic bacteria synthesize umami precursors through glutamate dehydrogenase pathways, and Debaryomyces accumulates sweet amino acids through glycolysis-coupled alanine transamination pathways (Wang et al., 2024).

Fig. 6.

Fig. 6

Open in a new tab

Correlation thermogram of nonvolatile compounds (A) and volatile compounds (B) with microbial communities.

Correlation analysis of 12 key aroma-active compounds (VIP > 1 and OAV > 1) revealed complex relationships between microbial taxa and aroma-active substances (Fig. 6B). Companilactobacillus, Levilactobacillus, Lactiplantibacillus, Chromohalobacter, Halomonas, Debaryomyces, and Pichia were positively correlated with most esters and aldehydes, indicating their involvement in flavor biosynthesis via esterification, redox reactions, and other metabolic pathways during fermentation ((Xiong, Xu, Du, et al., 2024)Xiong et al., 2024). Conversely, genera such as Pantoea, Pseudomonas, Enterococcus, Sphingomonas, and Rosenbergiella were positively correlated with Benzaldehyde, 1-Octen-3-ol, and Leaf acetate and negatively correlated with 3-Octanol, Linalool, and Ethyl palmitate. These organisms may produce specific alcohols and aldehydes through lipid oxidation and amino acid metabolism, contributing unique flavor characteristics to fermented cowpeas.

In summary, flavor formation in fermented cowpeas results from complex interactions between microbes and their metabolic activities. Companilactobacillus, Levilactobacillus, Lactiplantibacillus, Chromohalobacter, Microbacterium, and Debaryomyces were identified as key microbial drivers shaping the flavor profile. Earlier in this study, the OPLS-DA model clearly distinguished volatile compounds across fermentation stages. Key flavor compounds, including esters, alcohols, and aldehydes, were identified. These differences were mainly driven by the metabolic activity of core microorganisms. For example, Companilactobacillus, Levilactobacillus, and Lactiplantibacillus acidifie the fermentation system by producing lactic acid and creates an acidic environment that promotes ester formation. Lactic acid bacteria also directly produces several organic acids ((Li, Lin, et al., 2025)Li et al., 2025). Debaryomyces is a well-known producer of ester aromas in fermented foods and fruits. The flavor changes observed throughout fermentation reveal clear biochemical pathways ((Tang & Peng, 2024)Tang et al., 2024). A typical change during fermentation is the reduction of aldehydes with grassy notes and the increase of esters associated with fruity aromas. In this study, key compounds in raw cowpeas, such as Benzaldehyde and Leaf acetate, decreased after fermentation, whereas compounds such as Linalool, 3-Octanol, Ethyl octanoate, and Octyl salicylate increased. This shift is likely mediated by microbial enzymes such as aldehyde dehydrogenases and ester synthases, which convert precursors into more pleasant aroma compounds (Wu et al., 2025). Furthermore, the flavor profile continued to evolve during long-term storage, indicating that flavor compounds undergo further transformation and maturation.

4. Conclusions

This study systematically investigated the dynamic changes in physicochemical properties, volatile and nonvolatile compounds, and microbial communities during cowpea fermentation, revealing their interconnected metabolic networks. Fermentation caused significant metabolic restructuring, with GC-IMS and GC–MS identifying 65 and 95 volatile compounds, respectively. OPLS-DA confirmed distinct volatile profiles among raw, fermented, and 1-year-stored cowpeas and identified 21 volatiles from GC–MS and 18 from GC-IMS with VIP > 1 as key discriminators. Among these, 12 aroma-active compounds (VIP > 1 and OAV > 1) were established as flavor determinants, with Ethyl octanoate, Ethyl palmitate, and 4-Ethylphenol primarily contributing to the characteristic aroma of fermented cowpeas. Core microbial taxa, including Debaryomyces, Companilactobacillus, Levilactobacillus, Chromohalobacter, and Microbacterium, were identified as key metabolic drivers. They are correlated with organic acids, umami, and sweet amino acids, and key volatile substances (esters, alcohols, and aldehydes), and coordinate the production of organic acids, amino acid conversion, and volatile synthesis through enzyme activity. These interactions demonstrate that the flavor profile of fermented cowpeas is influenced by a microbial–metabolite network, in which microbial redox and esterification activities convert substrates into sour, umami, and fruity ester notes. The combined use of GC-IMS and GC–MS expanded the coverage of volatile compound detection, reducing the risk of missing compounds detected by only one technique, and enabled rapid differentiation of cowpeas at different fermentation stages. This integrated approach provides a solid foundation for comprehensive aroma analysis and accurate identification of fermented cowpeas. It also offers theoretical support for optimizing production processes and flavor evaluation in related fermented products. Future studies should elucidate the functional mechanisms of core microorganisms and explore the feasibility of guiding fermentation flavor and improving product consistency through inoculation with specific starter cultures. Such efforts would enhance the flavor quality and functionality of fermented cowpeas and support the development of standardized production processes.

CRediT authorship contribution statement

Zehui Li: Writing – original draft, Visualization, Investigation, Formal analysis, Data curation. Longqun Fu: Resources, Methodology. Jiajia Song: Software, Methodology. Huayi Suo: Software, Methodology. Yu Zhang: Writing – review & editing, Validation, Methodology, Funding acquisition, Conceptualization.

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 study was supported by the Chongqing Technology Innovation and Application Development Project (grant number: CSTB2023TIAD-KPX0028) and the Southwest University Training Program of Innovation and Entrepreneurship for Undergraduates (grant numbers: X202410635183 and 202410635039).

Footnotes

Appendix A

Supplementary data to this article can be found online at https://doi.org/10.1016/j.fochx.2026.103612.

Appendix A. Supplementary data

Supplementary material 1

mmc1.docx (4.7MB, docx)

Supplementary material 2

mmc2.docx (92KB, docx)

Data availability

Data will be made available on request.

References

  1. Chen Y., Gong H., Ren H., Liu L., Zang S., Liu T., Zhao Q. A comprehensive study on the quality deterioration mechanism and main paths of cowpea pickles during storage. Food Chemistry. 2025;145445 doi: 10.1016/j.foodchem.2025.145445. [DOI] [PubMed] [Google Scholar]
  2. Chen Y., Li P., Liao L., Qin Y., Jiang L., Liu Y. Characteristic fingerprints and volatile flavor compound variations in Liuyang Douchi during fermentation via HS-GC-IMS and HS-SPME-GC-MS. Food Chemistry. 2021;361 doi: 10.1016/j.foodchem.2021.130055. [DOI] [PubMed] [Google Scholar]
  3. El-Shahir A.A., Abdel-Sater M.A., Yassein A.S. Lipolysis activity of aflatoxins and Ochratoxins producing fungal strains contaminating fresh cow meat. Biosciences Biotechnology Research Asia. 2020;17(1):163–172. doi: 10.13005/bbra/2821. [DOI] [Google Scholar]
  4. Feng H., Timira V., Zhao J., Lin H., Wang H., Li Z. Insight into the characterization of volatile compounds in smoke-flavored sea bass (Lateolabrax maculatus) during processing via HS-SPME-GC-MS and HS-GC-IMS. Foods. 2022;11(17):2614. doi: 10.3390/foods11172614. [DOI] [PMC free article] [PubMed] [Google Scholar]
  5. Gao X., Li C., He R., Zhang Y., Wang B., Zhang Z.H., Ho C.T. Research advances on biogenic amines in traditional fermented foods: Emphasis on formation mechanism, detection and control methods. Food Chemistry. 2023;405 doi: 10.1016/j.foodchem.2022.134911. [DOI] [Google Scholar]
  6. Guo Z., Wang Y., Xiang F., Hou Q., Zhang Z. Bacterial diversity in pickled cowpea (Vigna unguiculata [Linn.] Walp) as determined by Illumina MiSeq sequencing and culture-dependent methods. Current Microbiology. 2021;78(4):1286–1297. doi: 10.1007/s00284-021-02382-3. [DOI] [PubMed] [Google Scholar]
  7. He Z., Chen H., Wang X., Lin X., Ji C., Li S., Liang H. Effects of different temperatures on bacterial diversity and volatile flavor compounds during the fermentation of suancai, a traditional fermented vegetable food from northeastern China. LWT. 2020;118 doi: 10.1016/j.lwt.2019.108773. [DOI] [Google Scholar]
  8. Huang Y., Wang L., Li Y., Chen Y., Chen G., Jia L., Ge Y. Comparative analysis of key flavor compounds in various baijiu types using E-nose, HS-SPME-GC–MS/MS, and HS-GC-IMS technologies. Food Chemistry: X. 2025;102689 doi: 10.1016/j.fochx.2025.102689. [DOI] [PMC free article] [PubMed] [Google Scholar]
  9. Jeong J.Y., Kim M.J., Kim H.S., Lee J.H., Hwang I.M. Integrated VOC and non-VOC profiling for stage-specific quality assessment of kimchi fermentation using LC-MS/MS and GC-IMS. Food Chemistry. 2025;146275 doi: 10.1016/j.foodchem.2025.146275. [DOI] [PubMed] [Google Scholar]
  10. Jin Y., Qi Y., Fan M., Zhang J., Kong B., Shao B. Biotransformation of carbendazim in cowpea pickling process. Food Chemistry. 2023;415 doi: 10.1016/j.foodchem.2023.135766. [DOI] [PubMed] [Google Scholar]
  11. Jung M.Y., Baek S.H., Lim S.M., Chung I.M., Han J.G., Kim S.H. An unattended HS-SPME-GC–MS/MS combined with a novel sample preparation strategy for the reliable quantitation of C8 volatiles in mushrooms: A sample preparation strategy to fully control the volatile emission. Food Chemistry. 2021;347 doi: 10.1016/j.foodchem.2020.128998. [DOI] [PubMed] [Google Scholar]
  12. Kang C., Zhang Y., Zhang M., Qi J., Zhao W., Gu J., Li Y. Screening of specific quantitative peptides of beef by LC–MS/MS coupled with OPLS-DA. Food Chemistry. 2022;387 doi: 10.1016/j.foodchem.2022.132932. [DOI] [PubMed] [Google Scholar]
  13. Lee M.A., Choi Y.J., Kim Y.S., Chon S.Y., Chung Y.B., Park S.H., Seo H.Y. Effects of salt type on the metabolites and microbial community in kimchi fermentation. Heliyon. 2022;8(11) doi: 10.1016/j.heliyon.2022.e11360. [DOI] [PMC free article] [PubMed] [Google Scholar]
  14. Li H., Chen X., Liu J., Niu J., Zhu X., Zhu H., Sun B. Regulating fermentation based on dominant genus: Effects of phenethyl alcohol-producing Saccharomyces cerevisiae on microbial succession and flavor compounds in fermented grains. Food Bioscience. 2025;65 doi: 10.1016/j.fbio.2025.106079. [DOI] [Google Scholar]
  15. Li P., Lin Q., Song D., Luo Q., Wu L., Long Z., Chen Q. Microbial succession drives flavor development in low-salt fermented bamboo shoots: A correlation analysis of physicochemical dynamics and metabolic pathways. Food Chemistry: X. 2025;103283 doi: 10.1016/j.fochx.2025.103283. [DOI] [PMC free article] [PubMed] [Google Scholar]
  16. Lian Y., Song J., Mumby W., Suo H., Zhang Y. The correlation between flavor formation and microbial community dynamics during the fermentation of zha cai. Journal of the Science of Food and Agriculture. 2024;104(10):6233–6241. doi: 10.1002/jsfa.13447. [DOI] [PubMed] [Google Scholar]
  17. Liu D., Zhang C., Zhang J., Xin X., Liao X. Metagenomics reveals the formation mechanism of flavor metabolites during the spontaneous fermentation of potherb mustard (Brassica juncea var. multiceps) Food Research International. 2021;148 doi: 10.1016/j.foodres.2021.110622. [DOI] [PubMed] [Google Scholar]
  18. Liu M., Deng N., Li H., Hou X., Zhang B., Wang J. Characterization and comparison of flavors in fresh and aged fermented peppers: Impact of different varieties. Food Research International. 2024;182 doi: 10.1016/j.foodres.2024.114187. [DOI] [PubMed] [Google Scholar]
  19. Meng Y., Lian Y., Li J., Suo H., Song J., Wang M., Zhang Y. Quality characteristics and flavor analysis of five mulberry varieties. Foods. 2024;13(24):4088. doi: 10.3390/foods13244088. [DOI] [PMC free article] [PubMed] [Google Scholar]
  20. Qin X., Xiong T., Xiong S., Liu Z., Xie M., Guan Q. Metatranscriptomics unravel the formation mechanism of key flavors during the natural fermentation of suansun, a Chinese traditional fermented bamboo shoot. Food Bioscience. 2024;57 doi: 10.1016/j.fbio.2023.103436. [DOI] [Google Scholar]
  21. Rovira P., Brugnini G., Rodriguez J., Cabrera M.C., Saadoun A., de Souza G., Rufo C. Microbiological changes during long-storage of beef meat under different temperature and vacuum-packaging conditions. Foods. 2023;12(4):694. doi: 10.3390/foods12040694. [DOI] [PMC free article] [PubMed] [Google Scholar]
  22. Tang A., Peng B. Metatranscriptomics reveals microbial community function succession and characteristic flavor formation mechanisms during black rice wine fermentation. Food Chemistry. 2024;457 doi: 10.1016/j.foodchem.2024.140428. [DOI] [PubMed] [Google Scholar]
  23. Wang Q., Cui R., Liu X., Zheng X., Yao Y., Zhao G. Examining the impact of Tetragenococcus halophilus, Zygosaccharomyces rouxii, and Starmerella etchellsii on the quality of soy sauce: A comprehensive review of microbial population dynamics in fermentation. Critical Reviews in Food Science and Nutrition. 2024;64(29):10873–10884. doi: 10.1080/10408398.2023.2230285. [DOI] [PubMed] [Google Scholar]
  24. Wang Y., Li C., Zhao Y., Li L., Yang X., Wu Y., Yang D. Novel insight into the formation mechanism of volatile flavor in Chinese fish sauce (Yu-lu) based on molecular sensory and metagenomics analyses. Food Chemistry. 2020;323 doi: 10.1016/j.foodchem.2020.126839. [DOI] [PubMed] [Google Scholar]
  25. Wang Y., Li J., Zhu T., Li Y., Yang J., Ma T., Sun X. Unveiling sub-regional aroma differences in China's Ningxia region: Characterization of key aroma profiles and chemical foundations via sensomics and RATA. Food Chemistry. 2025;146557 doi: 10.1016/j.foodchem.2025.146557. [DOI] [PubMed] [Google Scholar]
  26. Wei W., Yang S., Yang F., Hu X., Wang Y., Guo W., Zhu L. Cold plasma controls nitrite hazards by modulating microbial communities in pickled radish. Foods. 2023;12(13):2550. doi: 10.3390/foods12132550. [DOI] [PMC free article] [PubMed] [Google Scholar]
  27. Wu Y., You Y., Wu L., Du M., Ibrahim A., Suo H., Zheng J. Integrated metagenomics and metatranscriptomics analyses reveal the impacts of different Lactiplantibacillus plantarum strains on microbial communities and metabolic profiles in pickled bamboo shoots. Food Chemistry. 2025;464 doi: 10.1016/j.foodchem.2024.141772. [DOI] [PubMed] [Google Scholar]
  28. Xia M., Zhang X., Xiao Y., Sheng Q., Tu L., Chen F., Wang M. Interaction of acetic acid bacteria and lactic acid bacteria in multispecies solid-state fermentation of traditional Chinese cereal vinegar. Frontiers in Microbiology. 2022;13 doi: 10.3389/fmicb.2022.964855. [DOI] [PMC free article] [PubMed] [Google Scholar]
  29. Xiang L., Zhu W., Jiang B., Chen J., Zhou L., Zhong F. Volatile compounds analysis and biodegradation strategy of beany flavor in pea protein. Food Chemistry. 2023;402 doi: 10.1016/j.foodchem.2022.134275. [DOI] [PubMed] [Google Scholar]
  30. Xing Y., Yi R., Yue T., Bi X., Wu L., Pan H., Che Z. Effect of dense phase carbon dioxide treatment on the flavor, texture, and quality changes in new-paocai. Food Research International. 2023;165 doi: 10.1016/j.foodres.2022.112431. [DOI] [PubMed] [Google Scholar]
  31. Xiong S., Xu X., Du T., Liu Q., Huang T., Ren H., Xie M. Organic acids drove the microbiota succession and consequently altered the flavor quality of Laotan Suancai across fermentation rounds: Insights from the microbiome and metabolome. Food Chemistry. 2024;450 doi: 10.1016/j.foodchem.2024.139335. [DOI] [PubMed] [Google Scholar]
  32. Xiong S., Xu X., Liu Q., Xu Y., Ren H., Xiong T., Xie M. Integrated metatranscriptomics and metabolomics revealed the metabolic pathways of biogenic amines during Laotan Suancai fermentation. Food Bioscience. 2024;57 doi: 10.1016/j.fbio.2023.103517. [DOI] [Google Scholar]
  33. Yang L., Xian C., Li P., Wang X., Song D., Zhao L., Zhang C. The spatio-temporal diversity and succession of microbial community and its environment driving factors during stacking fermentation of Maotai-flavor baijiu. Food Research International. 2023;169 doi: 10.1016/j.foodres.2023.112892. [DOI] [PubMed] [Google Scholar]
  34. Yang Y., Fan Y., Li T., Yang Y., Zeng F., Wang H., Zhang Y. Microbial composition and correlation between microbiota and quality-related physiochemical characteristics in Chongqing radish paocai. Food Chemistry. 2022;369 doi: 10.1016/j.foodchem.2021.130897. [DOI] [PubMed] [Google Scholar]
  35. Zhang C., Zhang J., Xin X., Niu H., Liao X., Liu D. Reduced formation of biogenic amines in low-salt Zhacai via fermentation under CO2-modified atmosphere. Food Research International. 2023;163 doi: 10.1016/j.foodres.2022.112256. [DOI] [PubMed] [Google Scholar]
  36. Zhang S., Li C., Wu J., Peng S., Wu W., Liao L. Properties investigations of rape stalks fermented by different salt concentration: Effect of volatile compounds and physicochemical indexes. Food Chemistry: X. 2023;18 doi: 10.1016/j.fochx.2023.100746. [DOI] [PMC free article] [PubMed] [Google Scholar]
  37. Zhao Y., Wei W., Tang L., Wang D., Wang Y., Wu Z., Zhang W. Characterization of aroma and bacteria profiles of Sichuan industrial paocai by HS-SPME-GC-O-MS and 16S rRNA amplicon sequencing. Food Research International. 2021;149 doi: 10.1016/j.foodres.2021.110667. [DOI] [PubMed] [Google Scholar]
  38. Zheng J., Wittouck S., Salvetti E., Franz C.M., Harris H.M., Mattarelli P., Lebeer S. A taxonomic note on the genus Lactobacillus: Description of 23 novel genera, emended description of the genus Lactobacillus Beijerinck 1901, and union of Lactobacillaceae and Leuconostocaceae. International Journal of Systematic and Evolutionary Microbiology. 2020;70(4):2782–2858. doi: 10.1099/ijsem.0.004107. [DOI] [PubMed] [Google Scholar]

Associated Data

This section collects any data citations, data availability statements, or supplementary materials included in this article.

Supplementary Materials

Supplementary material 1

mmc1.docx (4.7MB, docx)

Supplementary material 2

mmc2.docx (92KB, docx)

Data Availability Statement

Data will be made available on request.

Articles from Food Chemistry: X are provided here courtesy of Elsevier

RESOURCES