Flavors and fragrances samples were provided by China Tobacco Henan Industrial Co., Ltd. To simulate typical production process, we opened fully packaged samples, took portions, and then resealed and transported them to the laboratory. In the lab, the samples were divided into two groups: three were stored at −80°C as the initial storage group (Group A), and the remaining three were stored at room temperature (25°C, 60% relative humidity) for 14 days as the end of storage group (Group B). Additionally, samples were collected on days 0, 3, 7, 10, 14, and 17 for total viable count (TVC) and pH values measurement. Volatile compounds analysis was conducted on samples collected on days 0, 7, and 14.

TVC was measured according to the China National Food Safety Standard Method for Food Microbiological Examination (GB 4789.2-2022). Briefly, 25 mL of sample was mixed with 225 mL of sterile saline solution to prepare a 1:10 dilution. This mixture was then serially diluted in ten-fold increments. For each dilution, 1 mL was transferred to a sterile Petri dish, followed by the addition of 15 mL of plate count agar. After thorough mixing, the plates were incubated at 37°C for 48 h, and colony counts were expressed as CFU/mL.

For pH values measurements, a Mettler Toledo Seven2GO™ pH meter was used, following the manufacturer’s instructions. Measurements were taken at regular intervals to monitor pH changes throughout the storage period at room temperature.

Volatile compounds in flavors and fragrances were analyzed using gas chromatography–mass spectrometry (GC–MS). The 2 mL sample of flavors and fragrances was placed in a centrifuge tube, followed by the addition of 10 mL PBS. The mixture was vortexed for 5 min at 2,500 rpm. Next, 10 mL of HPLC-grade acetonitrile and 50 μL of internal standard (2,3-dichlorotoluene, 0.2096 g/L) were added, and the mixture was vortexed for 20 min at 2500 rpm. After freezing the sample at −20°C for 30 min, 1.5 g NaCl and 6 g MgSO₄ were added followed by quick shaking. Then, 5 mL of HPLC-grade dichloromethane was added, vortexed for 10 min at 2,500 rpm, and centrifuged for 3 min at 8,000 rpm. The supernatant was collected, concentrated, filtered, and subsequently analyzed using GC–MS.

The chromatographic conditions were as follows: an HP-5MS capillary column (60 m × 250 μm × 0.25 μm) was used. The inlet temperature was set to 240°C, with a 1.0 μL injection volume and helium as the carrier gas at a flow rate of 1.0 mL/min in splitless mode. The temperature program was as follows: hold at 50°C for 4 min, increase by 3°C/min to 70°C and hold for 5 min, then increase by 2°C/min to 100°C and hold for 17 min, followed by an increase to 120°C (hold for 10 min), and finally increase by 6°C/min to 280°C.

For mass spectrometry, electron ionization (EI) was used with an electron energy of 70 eV. The solvent delay was set to 10 min, and full-scan detection was performed with a mass range of m/z 35–500. Compound identification was achieved by comparing the retention times with known flavors and fragrances components and using the NIST20 spectral library. Quantification was performed using the internal standard method for compounds with available standards, while those without standards were quantified based on relative peak areas (compound peak area to internal standard peak area).

Microbial genomic DNA was extracted using the OMEGA Mag-Bind Soil DNA Kit (M5635-02) following the manufacturer’s instructions. The extracted DNA was stored at −20°C. DNA quality and quantity were measured using agarose gel electrophoresis and a Qubit™ 4 Fluorometer. Metagenome shotgun sequencing libraries were prepared using the Illumina TruSeq Nano DNA LT Library Preparation Kit with an insert size of 400 bp. Sequencing was conducted on the Illumina NovaSeq platform (PE150) at Personal Biotechnology Co., Ltd. (Shanghai, China).

Raw sequencing reads were processed to generate high-quality reads. First, sequencing adapters were removed using Cutadapt (v1.2.1), followed by trimming of low-quality reads using the sliding-window algorithm in fastp. Taxonomic classifications of reads were performed using Kraken2 against a RefSeq-derived database (including archaea, bacteria, viruses, fungi, protozoans, metazoans, and plants). Reads assigned to metazoans and plants were excluded from downstream analysis. Alternatively, Kaiju with greedy-5 mode was used against an NR-derived protein database including similar taxa.

Assembly of quality-filtered reads was conducted with Megahit (v1.1.2) using meta-large preset parameters. Contigs longer than 300 bp were clustered using mmseqs2 with an identity threshold of 95 and 90% coverage of the shorter contig. Taxonomic assignment was performed by aligning contigs against the NCBI-NT database using mmseqs2 with the taxonomy mode. Contigs assigned to plants or metazoans were excluded. Gene prediction was performed using MetaGeneMark, and coding sequences (CDS) were clustered using mmseqs2 with 90% identity and coverage.

To assess gene abundances, high-quality reads were mapped to predicted genes using Salmon in quasi-mapping mode. Abundance was normalized using CPM (copies per kilobase per million mapped reads). Functional annotation was performed using mmseqs2 against Kyoto Encyclopedia of Genes and Genomes (KEGG) database. Linear discriminant analysis effect size (LEfSe) was applied to detect differentially abundant taxa and functions across groups. Beta diversity analysis was conducted using Bray-Curtis distance metrics and visualized through principal coordinate analysis (PCoA).

Microbial activity significantly influences the shelf life of flavors and fragrances, with total viable counts (TVC) serving as a direct indicator of microbial growth during storage. As shown in Figure 1A, the initial TVC was 2 log CFU/mL, indicating minimal contamination during production and packaging. For the first 3 days, TVC remained stable at 2–3 log CFU/mL, with no significant changes observed. However, after day 3, a rapid increase occurred, with TVC reaching 11 log CFU/mL by day 14. Beyond this point, microbial growth plateaued, likely due to the depletion of available organic nutrients, which limited further proliferation. According to the National Standard of China (GB 30616-2020), the acceptable TVC level is 5 log CFU/mL. Based on this standard, the sample remained within acceptable limits for up to 7 days.

In addition to TVC, monitoring pH values changes also provides valuable insights into spoilage progression. As shown in Figure 1B, the pH rose slightly from 3.69 to 3.74 early in storage, followed by a steady decline to 3.39 by day 14. This initial rise may be due to the production of alkaline compounds during early microbial metabolism. As storage continued, the accumulation of acidic by-products, particularly from acid-producing bacteria, became more pronounced, leading to a decline in pH as spoilage advanced. This trend is consistent with findings by Zhang et al. (2023), who observed a similar pH pattern in the spoilage of braised chicken, with an initial increase followed by a decline.

Aroma loss significantly impacts the quality of flavors and fragrances during storage (Obenland et al., 2011). In this study, gas chromatography–mass spectrometry (GC–MS) was employed to analyze dynamic changes in volatile compounds over time. Volatile compounds with initial concentrations of no less than 1 μg/mL were listed in Table 1. The internal standard, 2,3-dichlorotoluene, showed consistent concentrations across three measurements, confirming the instrument’s reproducibility.

A total of 3 aldehydes, 12 esters, 3 phenols, 3 acids, 4 alcohols, and 4 ketones were identified. Among the aldehydes, 5-hydroxymethylfurfural (5-HMF), vanillin, and its derivative ethyl vanillin are notable aroma contributors. However, these compounds were not detected in the later stages of storage. Vanillin can undergo biotransformation into vanillic acid and vanillyl alcohol, facilitated by microorganisms like Saccharomyces or Cystobasidium larynges (Rönnander et al., 2018). By day 7, vanillin had completely degraded, while vanillyl alcohol, initially undetectable, increased to 0.0102 mg/mL by day 14. This suggested that vanillyl alcohol might be the degradation product of vanillin. Both vanillin and ethyl vanillin are widely used flavoring agents with a characteristic rich vanilla scent, with ethyl vanillin providing a more intense and long-lasting aroma. Notably, in addition to their aromatic properties, these compounds have demonstrated antimicrobial activity (Fitzgerald et al., 2003). Vanillin, for instance, has been shown to inhibit the growth of bacteria and fungi such as Escherichia coli and Saccharomyces cerevisiae, and to exert cytotoxic effects on mammalian cells (Adeboye et al., 2014; Bezerra et al., 2016; Cheng et al., 2007). Therefore, the presence of vanillin in samples contributes not only to aroma but also to the preservation of flavors and fragrances. Its degradation diminishes both the aroma profile and the preservative potential.

Ester compounds play a pivotal role in contributing fruity, floral, or sweet aromas, making them essential for enhancing the complexity and longevity of fragrances. For example, phenethyl isobutyrate and phenethyl acetate are widely used in fine perfumes, shampoos, soaps, and non-cosmetic products such as detergents and household cleaners (McGinty et al., 2012a; McGinty et al., 2012b). Sotolon is another significant compound, recognized for contributing to the characteristic roasted chicory aroma (Wu and Cadwallader, 2019). Additionally, palmarosa essential oil contains several key aromatic components, with benzyl benzoate being one of the most significant (Kou et al., 2023). In the later stages of storage, a noticeable decline in the concentrations of several esters, including phenethyl isobutyrate, DL-malic acid diethyl ester, phenethyl acetate, ethyl glycolate, and benzyldimethylcarbinyl butyrate, was observed. Notably, the concentration of phenethyl isobutyrate decreased by 77.97%. Conversely, some esters remained relatively stable or even showed slight increases, such as benzyl benzoate, ethyl L(-)-lactate, sotolon, bis (2-ethylhexyl) phthalate, ethyl linoleate, dibutyl terephthalate, and 2-hydroxypropylacetate. This variation is likely due to complex biochemical reactions that occur during storage, leading to the production of these compounds.

Sorbic acid, along with its sodium, potassium, and calcium salts, is often used in food preservation for its antifungal and antibacterial properties, working by disrupting microbial metabolism (Brul and Coote, 1999; de Jesus et al., 2021). When combined with vanillin, potassium sorbate has been shown to have a synergistic preservative effect (Kimani et al., 2023; Matamoros-León et al., 1999). In this study, a decline in sorbic acid concentration was also observed in the later stages of storage, further reducing its preservative efficacy. Whether sorbic acid and vanillin exhibit a similar synergistic effect in this study remains to be confirmed through further experimentation. Additionally, with prolonged storage, fluoroacetone levels were detected to gradually increase, raising concerns about the safety and quality of the products.