Effects of storage time on the quality and microbial community of ripe Pu-erh tea

Abstract

Long-term storage is beneficial to improving the quality of Ripe Pu-erh Tea (RPT). Using RPT with storage time of 0, 5, 10 and 15 years as materials, targeted (HPLC and GC–MS) and non-targeted (UHPLC-Orbitrap-MS and GC × GC-TOFMS) combined volatolomics and metabolomics coupled with electronic sensory evaluation were used to reveal the differences in chemical compositions and microbial community succession of RPT during storage. The results showed that catechin and caffeine were decreased during aging process, the volatile profile also evolved, with a decrease in aldehydes, alcohols, and an increase in hydrocarbons, contributing to the characteristic aged aroma. 1,2,3-trimethoxybenzene and (+)-cedrol were identified as the primary contributors to the “woody” character of RPTs with long storage years. Microbiological analysis identified Aspergillus and Bacillus as dominant genera, playing key roles in metabolite decomposition and synthesis, which influenced the flavor and aroma characteristics of aged RPTs. These findings provide insights into flavor development and quality evolution during long-term storage.

Highlights

• •Comprehensive analysis of Ripe Pu-erh Tea aged from 0 to 15 years using both targeted and untargeted metabolomics.
• •Integrating UHPLC-Orbitrap-MS, GC×GC-TOFMS, GC-MS, and electronic sensing to reveal chemical and sensory changes.
• •Demonstrated decrease in catechins and caffeine over time, contributing to reduced bitterness in aged tea.
• •Identification of key volatiles, especially 1,2,3-trimethoxybenzene and (+)-cedrol, driving the characteristic aged aroma.
• •Microbial analysis identified Aspergillus and Bacillus as key drivers of metabolite turnover and flavor development.

1. Introduction

Pu-erh tea is a type of tea with unique quality characteristics, which is made from Yunnan big leaf species within the protection scope of geographical indications, and is made by specific processing technology within the protection scope of geographical indications (Wang et al., 2022b). The processing of RPT includes the steps of sun-drying, pile fermentation, steaming and molding, and drying, etc. (Fig. 1A). Through the action of microorganisms to promote the fermentation and oxidation of the tea leaves, so that they are gradually transformed into aged tea with a unique aroma and flavor. (Lv et al., 2013; Huang et al., 2025). For a long time,Raw pu-erh tea and RPT have often been confused by consumers and researchers. RPT has always been said to get better with age, and with the increase of storage time, RPT will show a more mellow and complex flavor and taste (Shi et al., 2019). Due to the lack of relevant research, the physical changes of RPT over time are still unclear.

Fig. 1.

Information of tea samples and experimental methods used in this study. (A) Major processing procedures of Ripe Pu-erh tea (RPT) and RPT products aged for 0–15 years. (B) Schematic of the approach used in this study.

Current research related to the storage time of Pu-erh tea is mainly focused on raw tea, and studies related to ripe tea are still lacking, including targeted and untargeted analyses of non-volatile and volatile substances. The effect of storage time on the chemical composition and aroma of Raw Pu-erh tea has been studied in several papers. Wang et al. (Wang et al., 2022b) analyzed the chemical composition and biological properties of Raw Pu-erh tea. Long et al. (Long et al., 2020) used a combination of untargeted metabolomics and targeted metabolomics to reveal the chemical characteristics of Raw Pu-erh tea during pile-fermentation. Wang et al. (Wang et al., 2021) investigated the microbial community composition during storage of dark tea and evaluated the relationship between the microbial community and the main chemical components using Pearsons analysis. However, there is currently no research on the changes in microbial community composition and microbial-driven transformation of chemical components during the long-term aging of RPT, which is also a dark tea. In addition. There is still a lack of comprehensive electronic sensory analysis of RPTs. Some unscrupulous merchants took advantage of the fact that consumers lack awareness of Pu-erh tea, and sold many counterfeit and shoddy products as aged and high-quality Pu-erh tea, which seriously damaged the interests of consumers (Cao et al., 2018). Therefore, it is necessary to measure and analyze the chemical composition, aroma and microorganisms of RPTs with different storage times, and combine electronic sensory analysis to explore the effect of storage time on the quality of RPTs.

In this study, a global strategy was employed, integrating chemometrics with non-targeted (ultra-high performance liquid chromatography-Orbitrap-MS (UHPLC-Orbitrap-MS) and comprehensive two-dimensional gas chromatography-time-of-flight mass spectrometry (GC × GC-TOFMS)) and targeted (high-performance liquid chromatography and gas chromatography–mass spectrometry (GC–MS)) metabolomics technologies. This strategy was applied to reveal the differences in chemical compositions among typical ripe Pu-erh tea products and to determine the variations in the characteristic chemical compositions of ripe Pu-erh tea from different vintages. Additionally, we employed the electronic-tongue (E-tongue) and electronic-nose (E-nose) to further analyze the taste and aroma characteristics of all samples, exploring the relationship between taste, aroma values, and characteristic chemical components. (Fig. 1B) It is worth noting that the RPT samples studied in our research were made using fresh tea leaves of the same origin, batch, tenderness, and variety, processed using the same methods, and stored under uniform environmental conditions. This meticulous approach aims to minimize the effects of confounding variables (e.g., genetic factors, origin, harvesting time and standard, processing techniques, and storage environment) on RPT. Our results will provide new insights into the effects of the aging process on the chemical composition of RPT and contribute to the progressive understanding of the aging process of RPT.

2. Materials and methods

2.1. RPT samples

RPT samples were loose teas produced by Yunnan Longrun Tea Industry Co., Ltd. Lincang, Yunnan, China) in 2006, 2011, 2016 and 2021. They were named Y15, Y10, Y5 and Y0. All RPT samples were derived from the same tea tree variety (Camellia sinensis var. assamica) and processed using identical techniques. Post-production, the RPTs were stored in the Yunnan Longrun Tea Industry warehouse under non-controlled and natural conditions. The samples for this study were collected in September 2022 and subsequently stored in a laboratory freezer at −80 °C.

2.2. UHPLC-Q-Exactive orbitrap-MS analysis

UPLC–Orbitrap–MS/MS analysis was conducted using a Nexera X2 Shimadzu UHPLC system coupled with an AB Sciex TripleTOF 5600 mass spectrometer. Chromatographic separation utilized a Waters HSS T3 column, and the gradient program was applied as follows:0–5 min: 2 % B;5–7 min: 2–13 % B;7–21 min: 13–21 % B;21–23 min: 21 %–30 % B;23–28 min: 30–100 % B;28–30 min: 100 % B.The MS data were acquired using data-independent acquisition (DIA) mode. Parameters included electrospray ionization in positive and negative modes, a mass range of 100–1200 m/z for TOF-MS scan, and 50–1200 m/z for TOF-MS/MS scan. Metabolite identification used databases such as The Human Metabolome Database (HMDB), Lipid maps database, Metlin database, and a self-built database. After quality control, identified metabolites underwent multivariate statistical analysis.

2.3. Determination of major chemical components

Detection of total tea polyphenols, water extract, catechins, gallic acid, caffeine, and theobromine was performed using a high-performance liquid chromatography (HPLC) system (Agilent Technologies). All determinations were conducted with triplicate biological replicates.

2.4. GC × GC-TOFMS analysis

A 1 g RPT sample was weighed and added to a 20 mL headspace flask, filled with 5 mL of boiling water, and heated in a solid-phase microextraction heating device. The volatile components absorbed on the SPME fiber were desorbed at the GC × GC-TOFMS injection port (250 °C). GC separation was performed on a DB-wax column and a DB-17ht column in the first and second dimensions, respectively. The modulation period was 4.0 s, with helium as the carrier gas. MS detection covered a range of m/z 33 to 550. The electron ionization energy was −70 eV.

2.5. HS-SPME/GC–MS analysis

1 g RPT powder was placed in a 20 mL headspace bottle, infused with 5 mL of boiling water, and sealed. A 65-μm carboxen/polydimethylsiloxane (CAR/PDMS) fiber was exposed to the sample headspace while continuously stirred for 60 min. Desorption and GC–MS analysis were conducted three times. The separation used a DB-WAX column, and the carrier gas was helium. The mass spectrometer conditions included an ionization mode of EI, ion source temperature of 230 °C, and a full scan mode with a mass range of 35–550 amu.

2.6. OAV calculation

The odor activity value (OAV) was calculated as the ratio of the concentration of a volatile compound to its odor threshold (OAV = Ci/OTi), where Ci represents the concentration of volatile compounds, and OTi stands for the odor threshold.

2.7. Sequencing processing

Fungal and bacterial community profiles of RPTs were analyzed by amplifying ITS1 and the v3-v4 region of the 16S rRNA gene, respectively. Genomic DNA was extracted using the CTAB method, and amplifications were performed with ITS1F/ITS2 and 515F/806R primers. Sequencing was conducted using the NovaSeq6000 system by Majorbio Co., Ltd.

2.8. Electronic sensory analysis

E-tongue Measurement:Taste evaluation utilized a TS-5000Z taste sensing system (Insent). Three grams of tea sample were infused with boiled deionized water, and the E-tongue analysis measured various taste characteristics in triplicate.

2.9. Statistical analysis

All data were presented as mean ± standard deviation and analyzed using Statistical Package for Social Sciences Version 26.0 (SPSS Inc.). Multivariate statistical analysis utilized SIMCA (version 14.1, Umetrics). One-way analysis of variance (ANOVA) and multiple comparisons were employed to test significant differences among groups (p < 0.05). Graphs were plotted using Origin (version 2021, OriginLab).

3. Results and discussion

3.1. Untargeted metabolomic analysis of non-volatile metabolites of RPTs

Metabolites undergo a series of transformations, resulting in distinctive chemical profiles within RPTs. A non-targeted metabolomics approach based on UHPLC-Q-Exactive Orbitrap-MS was used to investigate the non-volatile chemical profiles of aged RPTs. A total of 5978 metabolites were identified by analysis in both negative and positive ion modes. These metabolites were further categorized into 12 groups based on their chemical taxonomic superclass. These 12 groups included lipids and lipid-like molecules (654), organic acids and derivatives (517), phenylpropanoids and polyketides (358 e.g., flavans, flavones and tannins), organoheterocyclic compounds (423), organic oxygen compounds (379), benzenoids (243), nucleosides, nucleotides and analogues (78), organosulphur compounds (36), alkaloids and derivatives (20), organic nitrogen compounds (14), lignans, neolignans and related compounds (4) and others (1722). Based on the PCA and HCA results (Fig. 2A-B), it can be concluded that there is a significant difference between newly made RPTs and those stored for up to 15 years. We labeled them as Y0,Y5,Y10 and Y15, respectively. Using the orthogonal projection latent structure discriminant analysis model, a total of 118 different metabolites were identified between these years (PLS-DA, VIP > 1, P < 0.05).Other significant classes include amino acids, peptides and analogues with 9 (7.63 %), Carbohydrates and Carbohydrate Conjugates with 6 (5.08 %), Arylsulfates with 5 (4.24 %) and Flavonoids also with 5 (4.24 %).

Fig. 2.

Profiles of non-volatile compounds and among Ripe Pu-erh teas (RPTs) stored for different years. (A) Principal component analysis (PCA) score plot of Ripe Pu-erh teas (RPTs) stored for different years.In the PCA plot, the yellow ellipse represents the newly made RPT (Y0), the blue ellipse represents the RPT stored for 5 years (Y5), the red ellipse represents the RPT stored for 10 years (Y10), and the green ellipse represents the RPT stored for 15 years (Y15). (B) Hierarchical cluster analysis (HCA) plot of the metabolites of RPTs stored for different years. (C) Heatmaps of Z-score normalized relative contents of differentially accumulated nonvolatile compounds detected by UHPLC-Orbitrap-MS among RPTs stored for different years. (D—H) Contents of catechins, caffeine and water extract, gallic acid, theobromine and caffeine, respectively (mg/g). (For interpretation of the references to colour in this figure legend, the reader is referred to the web version of this article.)

Amino acids have been identified as a key component contributing to the health benefits and flavor of tea. In this study, a total of four amino acids and their derivatives were identified, and different storage periods RPTs had unique profiles of amino acids. In Y15, the average concentration of L-Theanine was higher compared to Y5 and Y10. In contrast, N-Phenylacetylaspartic Acid demonstrated lower average concentrations in Y0, but a significant increase during Y15, stabilising at significantly higher levels, indicating a continued upward trend over time. D-Phenylalanine exhibited relative stability across all years, with minimal fluctuations and no significant change. Furthermore, the analysis identified five types of flavonoids, including compounds such as (−)-Epigallocatechin 3,3′-di-gallate,Quercetin 3-O-alpha-D-arabinopyranoside, andLuteolin 7-glucuronide-4′-rham noside, Apigenin 7-glucoside-4′-p-coumarate, and 3,3′,5,5’-Tetrahydroxy-6,7-methyleneoxy-4′-methoxyflavone 3-glucuronide.The content of (−)-Epigallocatechin 3,3′-digallate increased significantly with the passage of time, while the content of other flavonoids also increased to some extent. Finally, the content of 3,3′,5,5′-tetrahydroxy-6,7-methyleneoxy-4′-methoxyflavone 3-glucuronide remained low throughout the duration of the experiment (Fig. 2C).

3.2. Targeted metabolomic analysis of non-volatile metabolites

This study determined the major flavor components of tea, including water extract, catechins, catechins, gallic acid, caffeine, and theanine. There was an overall declining trend in total catechin content with increasing vintage, where Epicatechin (EC) and Epigallocatechin gallate (EGCG) exhibited a similar trend to total catechins. Epigallocatechin (EGC) increased in Y10 but then decreased in Y15 (Fig. 2D), while gallic acid (GA) decreases significantly with storage time (Fig. 2F). These findings suggest that during microbial fermentation, gallic acid-bound catechins undergo hydrolysis to form non-gallic acid-bound catechins. The decrease in catechin content contributes to the reduction of bitterness in tea leaves, aligning with the common perception that RPTs acquires better quality with aging. Caffeine serves as the principal source of bitterness in tea infusions and is a determining factor in the formation of tea flavor. The caffeine content in RPTs exhibited a slight declining trend with increasing storage time, and the content of caffeine's precursor, theobromine, displayed a similar pattern to that of caffeine (Fig. 2G-H).

3.3. Global natural products social (GNPS) molecular networking analysis

In order to achieve the goal of overall metabolomics analysis, GNPS molecular networks were constructed for metabolites from RPTs with storage times of 0,5,10 and 15 years. In the current investigation, a total of 2008 nodes were observed within the global network (Fig. 3), indicating that MS/MS spectra were acquired for 2008 features during the UHPLC-Orbitrap-MS analysis. In molecular network, nodes are connected by edges to form clusters or groups, meaning that the MS/MS spectra of these chemical features share similarity above a certain threshold presented as both cosine scores and number of MS/MS fragments. Spectral features with analogous structures among RPTs stored for different years were connected by lines, resulting in the formation of metabolic clusters, mainly including Phenylpropanoids, Glycerolipids, Flavonoids. Furthermore, a novel catechin derivative (m / z: 394.889) was identified in RPT stored for over a decade, which is hypothesized to be a derivative of EGC due to its location in the same metabolic cluster as EGC. However, the chemical structure of this unknown spectrum remains to be elucidated.

Fig. 3.

Feature-based molecular network of metabolites among Ripe Pu-erh teas (RPTs) detected by the UHPLC-Orbitrap-MS/MS. The characteristic metabolites of RPTs stored for different years were clustered by ellipses with different colors. In the molecular network, the green ellipse represents the newly made RPT (Y0), the yellow ellipse represents the RPT stored for 5 years (Y5), the purple ellipse represents the RPT stored for 10 years (Y10), and the red ellipse represents the RPT stored for 15 years (Y15). (For interpretation of the references to colour in this figure legend, the reader is referred to the web version of this article.)

3.4. Untargeted metabolomic analysis of volatile metabolites of RPTs

Advanced GC × GC-TOFMS analysis of RPTs from four vintages was performed to investigate the combined changes in non-targeted volatile compounds (VOCs) among RPTs. This analysis identified a total of 507 VOCs among the RPTs. In order to deeply investigate the effect of storage time on the aroma of RPTs, we performed PCA and HCA (Fig. 4A-B). The results showed that there were significant differences among the RPTs of the four storage years. Based on the PLS-DA model (VIP > 1 and P < 0.05), 90 differential volatile metabolites were identified, encompassing an array of volatile compounds categorized into 6 classes, including 2 lipids, 15 ketones, 42 oxygenated compounds, 16 aldehydes, such as acetaldehyde, eight alcohols, including ethanol, and 8 hydrocarbons resembling benzene (Fig. 4C). Notably, among these compounds, only a small amount of compounds such as toluene and ethyl ether showed an increasing trend over time, while the remaining compounds decreased with the increase of storage time.

Fig. 4.

Profiles of volatile compounds and aroma characteristics among Ripe Pu-erh teas (RPTs) stored for different years. (A) Principal component analysis (PCA) score plot of Ripe Pu-erh teas (RPTs) stored for different years. In the PCA plot, the yellow ellipse represents the newly made RPT (Y0), the blue ellipse represents the RPT stored for 5 years (Y5), the red ellipse represents the RPT stored for 10 years (Y10), and the green ellipse represents the RPT stored for 15 years (Y15). (B) Hierarchical cluster analysis (HCA) plot of the metabolites of RPTs stored for different years. (C) Heatmaps of Z-score normalized relative contents of differentially accumulated untargeted detected by GC × GC-TOFMS among RPTs stored for different years. (D) Heatmaps of Z-score normalized relative contents of targeted volatile metabolites detected by GC–MS among RPTs stored for different years. (D) Content of main aroma compounds of RPTs stored for different years (μg/L). (F-K) Contents of 1,2,3-trimethoxybenzene,β-ionone,1,2,4-trimethoxybenzene,phenethyl alcohol,1-methylnaphthalene and (+)-Cedrol, respectively (μg/L), from RPTs of different years of storage. (For interpretation of the references to colour in this figure legend, the reader is referred to the web version of this article.)

3.5. Targeted metabolomic analysis of volatile metabolites of RPTs

3.5.1. Change of the volatile components (VOCs) of RPTs

To investigate the overall disparities in VOCs among distinct vintages of RPTs, Gas Chromatography-Mass Spectrometry (GC–MS) was employed for the detection of volatile compounds within these tea samples crafted from identical tea leaf source material. A total of 54 aroma compounds were detected in the four samples of different years (Table 1), including 10 alcohols, 14 aldehydes, 4 esters, 15 ketones, 3 hydrocarbons, 6 heteroxylates, and 2 phenols, a total of 7 major classes of compounds (Fig. 4D. In addition, the analysis of the changes in the content of aromatic compounds during the storage of Puerh ripe tea showed (Fig. 4E) that the total content of ketones, oxygen heterocycles, aldehydes, and alcohols showed an overall decreasing trend with the prolongation of storage time. It is worth noting that some of the compounds showed differential patterns of change: phenylethanol, phenylacetaldehyde, methyl palmitate, and camphor accumulated significantly with the extension of storage time, while the contents of nerolidol and (+)-cedrol continued to decrease. This volatility suggests that the storage process may selectively affect different types of aromatic substances through oxidation, degradation or transformation reactions, leading to the dynamic evolution of aroma characteristics.

Table 1.

Aroma substances of RPT with different storage years.

Compounds	Odor type	Relative contentsa
		Y0	Y5	Y10	Y15
1-Octen-3-ol	Woody	5.63 ± 0.71	6.28 ± 0.69	5.25 ± 0.74	4.72 ± 0.63
(+)-Cedrol	Woody	3.83 ± 0.74	19.05 ± 3.07	25.3 ± 3.29	57.26 ± 7.54
α-Terpineol	Floral	218.47 ± 41.92	73.62 ± 11.94	67.06 ± 12.7	51.57 ± 9.32
Phenethyl alcohol	Floral	1229.93 ± 237.26	317.31 ± 33.38	385.48 ± 60.97	432.86 ± 52.16
Nerolidol	Floral	12.87 ± 2.14	15.28 ± 1.81	15.25 ± 2.25	17.6 ± 2.31
Linalool	Floral	43.24 ± 8.1	19.9 ± 2.1	23.38 ± 3.05	14.28 ± 2.81
cis-3-Hexen-1-ol	Grassy	9.11 ± 1.41	4.72 ± 0.52	3.73 ± 0.46	5.04 ± 0.81
Linalool Oxide II	Floral	584.22 ± 81.35	390.7 ± 57.88	547.47 ± 103.99	412.91 ± 57.7
Linalool I oxide	Floral	1282.16 ± 249.56	691.94 ± 108.82	956.52 ± 185.88	561.2 ± 106.26
Phytol	Floral	13.61 ± 1.69	19.38 ± 2.11	25.6 ± 3.97	26 ± 3.23
Sapphire aldehyde	Herbal	7.59 ± 1.39	15.22 ± 1.76	14.41 ± 2.65	9.08 ± 1.03
β-Cyclocitraldehyde	Floral	2.35 ± 0.26	3.47 ± 0.66	4.14 ± 0.72	3.22 ± 0.59
Phenylacetaldehyde	Floral	271.12 ± 41.06	130.84 ± 14.96	167.36 ± 17.9	132.39 ± 21.72
(E)-2-Hexenal	Grassy	77.35 ± 9.65	132.06 ± 25.11	86.83 ± 14.09	95.86 ± 18.67
(E)-2-Nonenal	Fatty	6.21 ± 0.62	19.47 ± 3.5	13.37 ± 1.4	17.79 ± 2.91
(E,E)-2,4-Heptadienal	Grassy	79.41 ± 12.14	70.79 ± 12.52	42.39 ± 7.05	45.78 ± 5.31
(E,E)-2,4-Decadienal	Fatty	8.4 ± 1.05	9.25 ± 1.4	10.42 ± 1.87	10.92 ± 2.16
(E,E)-2,4-Nonadienal	Fatty	4.7 ± 0.73	7.82 ± 1.46	8.42 ± 1.34	11.2 ± 1.42
(E,E)-2,4-Octadienal	Fruity	2.13 ± 0.27	3.94 ± 0.4	4.29 ± 0.83	4.35 ± 0.52
Heptanal	Grassy	14.26 ± 2.21	32.51 ± 4.13	21.28 ± 2.6	18.64 ± 2.19
Decanal	Fruity	4.72 ± 0.87	7.09 ± 0.94	4.33 ± 0.51	4.85 ± 0.6
Citral	Lemon	2.98 ± 0.49	3.62 ± 0.41	3.04 ± 0.32	2.46 ± 0.34
Nonanal	Fatty	12.83 ± 1.63	20.63 ± 2.33	10.72 ± 1.78	11.78 ± 1.51
hexanal	Grassy	151.14 ± 20.12	380.94 ± 55.4	291.64 ± 46.99	262.46 ± 36.29
2-Heptanone	Fruity	0 ± 0	6.53 ± 0.78	7.84 ± 1.09	3.89 ± 0.51
2-Undecanone	Fruity	0.39 ± 0.05	0.78 ± 0.13	1.09 ± 0.11	1.12 ± 0.11
Tea flavoring ketone	Woody	13.56 ± 1.52	17.68 ± 3.52	20.99 ± 3.54	20.49 ± 2.35
6-Methyl-3,5-heptadien-2-one	Spicy	2.33 ± 0.29	13.41 ± 2.37	14.83 ± 2.7	13.96 ± 1.51
6-Methyl-5-hepten-2-one	Fruity	9.75 ± 1.51	14.92 ± 2.4	18.19 ± 2.15	10.34 ± 1.46
α-Violanone	Violet	3.7 ± 0.41	5.84 ± 0.71	5.44 ± 0.57	6.23 ± 0.97
β-Violanone	Violet	11.21 ± 2.1	12.9 ± 2.48	11.94 ± 1.98	11.37 ± 1.54
acetophenone	Floral	7.29 ± 1.12	13.35 ± 1.53	16.2 ± 3.06	14.74 ± 2.15
p-Methylacetophenone	Floral	1.29 ± 0.19	2.54 ± 0.28	3.09 ± 0.58	3.28 ± 0.41
Dihydro-beta-perillone	Woody	0.73 ± 0.13	0.71 ± 0.07	0.5 ± 0.08	0.74 ± 0.14
cis-Jasmone	Flowery	4.83 ± 0.83	2.23 ± 0.24	2.6 ± 0.45	2.47 ± 0.43
Geranylacetone	Floral	11.62 ± 2.24	20.91 ± 2.11	23.01 ± 2.83	22.87 ± 4.5
Isophorone	Woody	12.66 ± 1.85	11.62 ± 1.44	12.58 ± 2.1	10.14 ± 1.06
Camphor	Camphor	228.7 ± 44.36	38.45 ± 6.68	60.62 ± 6.81	46.32 ± 8.29
Phytone	Floral	13.72 ± 1.45	19.11 ± 3.77	27.08 ± 5.07	27.45 ± 3.85
Dihydroactinidiolide	Green	700.8 ± 114.69	1119.34 ± 133.83	1081.99 ± 133.68	1264.72 ± 209.08
Methyl salicylate	Minty	5.55 ± 0.92	5.7 ± 0.79	8.88 ± 0.96	6.29 ± 0.63
Benzyl acetate	Flowery	0.27 ± 0.03	1.55 ± 0.22	1.32 ± 0.24	1.08 ± 0.1
Methyl palmitate	Waxy	22.28 ± 4.04	7.9 ± 1.41	9.84 ± 1.94	4.56 ± 0.65
1,2,3-Trimethoxybenzene	Aged	1843.66 ± 248.11	1250.88 ± 160.52	1630.16 ± 198.2	1492.02 ± 173.81
1,2,4-Trimethoxybenzene	Aged	512.69 ± 101.96	466.04 ± 73.09	572.81 ± 79.32	409.76 ± 65
2,4,6-Trimethoxytoluene	Fruity	0.27 ± 0.05	5.02 ± 0.54	3.33 ± 0.55	8.59 ± 1.35
2-pentylfuran	Fruity	0 ± 0	0.73 ± 0.08	0 ± 0	0 ± 0
3,4,5-Trimethoxytoluene	Musty	36 ± 4.16	301.3 ± 37.55	214.03 ± 36.97	460.44 ± 87.27
3,4-dimethoxytoluene	Earthy	7.58 ± 1.35	25.23 ± 3.54	21.75 ± 2.29	33.34 ± 4.88
1-methylnaphthalene	Earthy	1.06 ± 0.13	1.19 ± 0.18	1.28 ± 0.13	1.48 ± 0.18
2-methylnaphthalene	Floral	1.7 ± 0.33	1.84 ± 0.25	2.17 ± 0.31	2.48 ± 0.36
β-stilbene	Woody	0.36 ± 0.06	0.63 ± 0.09	0.78 ± 0.07	2.14 ± 0.23
2,6-Di-tert-butyl-4-methylphenol	Toasty	0.51 ± 0.09	0.68 ± 0.11	0.49 ± 0.06	0.51 ± 0.05
4-s-butylphenol	Toasty	0.26 ± 0.03	3.96 ± 0.45	3.51 ± 0.43	5.33 ± 0.6

^aRelative content, the ratio of the peak area of the volatile compound to the peak area of the internal standard RPT.

3.5.2. Identification of key VOCs in RPTs

The key aroma components were screened out by calculating the OAV values, and a total of 29 compounds with OAV>1 were the key aroma components of RPT (Table 2), of which 23 compounds had OAV values greater than 1 in all the samples, among which 1,2,3-trimethoxybenzene had the largest OAVs and was the strongest contributor to the aroma (Fig. 4F), which might confer the characteristic woodsy, smoky, or aged aroma characteristics to the tea leaves (Yang et al., 2018). This was followed by phenethyl alcohol and 1,2,4-trimethoxybenzene (Fig. 4G-H), with the former possibly associated with floral or fruity tonalities and the latter contributing caramel or herbal notes (Yang et al., 2018). The oav values for (+)-Cedrol, β-ionone and 1-methylnaphthalene and were relatively low (Fig. 4I-K). Overall the contribution of these six compounds to the aroma of RPT was very significant.

Table 2.

OAV value of aroma substances in RPT with different storage years.

Compounds	Threshold of perceptiona	OAVb
		Y0	Y5	Y10	Y15
1-Octen-3-ol	1.5	3.1	3.5	4.2	3.8
(+)-Cedrol	0.5	114.5	50.6	38.1	7.7
Phenethyl alcohol	9	48.1	42.8	35.3	136.7
Nerolidol	10	1.8	1.5	1.5	1.3
Linalool	6	2.4	3.9	3.3	7.2
Oxidized linalool II	320	1.3	1.7	1.2	1.8
Linalool Oxide I	320	1.8	3.0	2.2	4.0
Phytol	0.64	40.6	40.0	30.3	21.3
Saffron aldehyde	3	3.0	4.8	5.1	2.5
β-Cyclocitraldehyde	3	1.1	1.4	1.2	0.8
Phenylacetaldehyde	40	3.3	4.2	3.3	6.8
(E)-2-Hexenal	17	5.6	5.1	7.8	4.6
(E)-2-Nonenal	0.4	44.5	33.4	48.7	15.5
(E,E)-2,4-Heptadienal	15.4	3.0	2.8	4.6	5.2
(E,E)-2,4-Decadienal	0.16	68.2	65.1	57.8	52.5
(E,E)-2,4-Nonadienal	0.16	70.0	52.6	48.9	29.4
Heptanal	3	6.2	7.1	10.8	4.8
Decanal	4.9	1.0	0.9	1.4	1.0
Citral	0.5	4.9	6.1	7.2	6.0
Nonanal	15	0.8	0.7	1.4	0.9
hexanal	4.5	58.3	64.8	84.7	33.6
α-Violanone	0.4	15.6	13.6	14.6	9.3
β-Purinone	0.1	113.7	119.4	129.0	112.1
Dihydrokiwifruit lactone	500	2.5	2.2	2.2	1.4
1,2,3-Trimethoxybenzene	0.75	1989.4	2173.5	1667.8	2458.2
1,2,4-Trimethoxybenzene	3.06	133.9	187.2	152.3	167.5
3,4,5-Trimethoxytoluene	100	4.6	2.1	3.0	0.4
3,4-dimethoxytoluene	5	6.7	4.3	5.0	1.5
1-methylnaphthalene	0.02	73.8	64.1	59.6	52.9

^aThe olfactory threshold of a compound in water, measured in mg/kg, obtained from literatures.

^bOav, the odorant activity value, OAV=Ci/OTi, Ci is the concentration of the odor component (μg/L), and OTi is the threshold value of that odor component (μg/L).

3.6. E-tongue and E-nose analysis of RPTs

To objectively assess the differences in taste among RPTs of different vintages, electronic tongue and electronic nose analyses were employed. The taste attributes of RPTs were visualized in radar plots in Fig. 5A, encompassing sourness, bitterness, astringency, aftertaste-B, aftertaste-A, umami, richness, saltiness, and sweetness. The 5 year RPTs had an overall balanced distribution of scores, with no particularly high or low scores for any of the attributes. RPTs stored for 10 years have high values for sourness and bitterness, while those stored for 15 years have high values for bitterness and Aftertaste - A, low values for sourness and moderate values for the other attributes. Electronic tongue analysis shows that the flavor attributes of RPTs show non-uniform dynamic evolution during the aging process: with the extension of aging time, bitterness and aftertaste B decrease overall due to the reduction of caffeine content, while other attributes show stage-by-stage fluctuations with the year. Newly produced RPTs are characterized by prominent aftertaste B and richness, low sourness, and moderate umami, saltiness, and sweetness; the attributes of 5 years old RPTs tend to be balanced, without obvious sensory extremes; by the 10-year aging stage, the sourness and bitterness values  increase again, which may be related to the oxidation of polyphenols or the accumulation of organic acids during storage; while the 15-year-old RPTs are characterized by the rebound of bitterness and aftertaste A values, the decline of sourness, and the maintenance of other attributes at a moderate level. Overall, the initial aging period (new tea to 5 years) is dominated by convergence and balance of flavor attributes, while long-term aging (more than 10 years) leads to non-linear fluctuations in sourness, bitterness, etc. due to the complex transformation of internal substances, revealing the stage-by-stage differences in sensory characteristics during the aging process of RPTs. We also performed an e-nose analysis and the results are shown in Fig. 5B, the response values of P40/1 exhibit no significant variations. In Y15, the response value of T70/2 is notably higher than that in Y0.Meanwhile the response value of LY2/G, LY2/gCT, PA/2 and P30/1 are lower in Y20. PA/2, LY2/gCT, and LY2/G sensors are sensitive to ethanol, ammonia, amine, propane, butane, carbon, and oxygen compounds. This result is consistent with the reduction in hydrocarbon content in the non-target VOCs analysis. Correlation analysis was performed to further elucidate the relationship between the E-nose sensors and E-tongue sensors. The LY2/G, LY2/gCT, PA/2, and P30/1 sensors exhibited more substantial positive correlations with astringency, sweetness, and more substantial negative correlations with sourness, bitterness, saltiness, and umami (Fig. 5C).

Fig. 5.

(A-B) Radar of sensory taste attributes profile during the storage process of RPTs. Y0,newly made RPT; Y5,RPT stored for 5 years; Y10,RPT stored for 10 years; Y15,RPT stored for 15 years. (C) The correlation analysis between Electronic nose and Electronic tongue values in RPT. (D) Relative abundances of the abundance>0.1 % fungi in the RPT. (E) Relative abundances of the abundance >0.1 % bacteria in the RPT. (F) Relationships of bacteria and fungi and differentially accumulated metabolites among RPTs by Pearson's correlation analysis (P < 0.05). Red lines represent positive correlation, and the blue lines indicate negative correlations. Flavonoid glycosides (FFGs), carbohydrates and carbohydrate conjugates (CCCs), amino acids and their derivatives (AADs). (F) Relationships of bacteria and fungi and differentially accumulated volatile metabolites among RPTs by Pearson's correlation analysis (P < 0.05). Red lines represent positive correlation, and the blue lines indicate negative correlations. Volatile compounds (VOCs). (For interpretation of the references to colour in this figure legend, the reader is referred to the web version of this article.)

3.7. Differences in microbial diversity among RPTs

Microbes play an integral role in the formation of pu-erh tea quality, involved in the decomposition and synthesis of metabolites in the leaves, as well as being important drivers of aroma formation in different types of pu-erh tea (Li et al., 2018; Zhu et al., 2020). To comprehensively explore the microbial dynamic evolution during the storage of RPTs, specific regions of the fungal ITS gene and bacterial 16 s rDNA were obtained and subsequently sequenced by high throughput sequencing.

The relative abundances of the top ten genera of fungi and bacteria were further schematized in Fig. 5C. The results indicated that the relative abundances of Aspergillus, Blastobotrys, Rasamsonia, Thermomyces, Candida,unclassified_f__Saccharomycetales_fam_Incertae_sedis, Trichosporon, Penicillium, Debaryomyces, and Apiotrichum were the top 10 fungal genera of RPTs. Throughout the storage process, the genus Aspergillus has been the most prevalent and was higher in newly made RPTs and not differing significantly in samples stored for 5, 10, and 15 years. Blastobotrys, on the other hand, was found to be progressively more abundant with increasing storage time. In addition, the relative abundances of the top 10 bacterial genera in RPTs were Bacillus,Pullulanibacillus,Paenibacillus,Kocuria,Scopulibacillus,Shimwellia, Staphylococcus, Brevibacterium,Ornithinibacillus, and unclassified_o__Bacillales (Fig. 5D). Throughout the storage process, the genus Bacillus has been the most prevalent and the overall trend of Bacillus content was progressively less with storage time, while the opposite trend was observed for Pullulanibacillus.

The correlation analysis based on the Pearson correlation coefficients between microorganisms (with an abundance >0.1 %) and differential metabolites (both non-volatile and volatile metabolites) was investigated. The fungi of Aspergillus exhibits a negative correlation with amino acid derivatives (AADs) and carbohydrate conjugates (CCCs), while showing a positive correlation with flavonoid glycosides (FGs). Paenibacillus demonstrates a significant positive correlation with differential non-volatile metabolites. Additionally. Bacteria of the genera Paenibacillus, Pullulanibacillus, and Bacillus are significantly correlated with differential non-volatile metabolites. Aspergillus, Blastobotrys, and Rasamsonia of fungi display a significant positive correlation with volatile metabolites, while other fungi show a negative correlation. In contrast, Scopulibacillus, Glutamicibacter, and Shimwellia of bacteria exhibit a negative correlation with volatile metabolites. Bacteria such as Paenibacillus, Pullulanibacillus, and Bacillus were positively correlated with non-volatile metabolites, suggesting that they may dominate the biosynthesis of such substances. Among the fungi, Aspergillus, Blastobotrys and Rasamsonia were positively associated with volatile metabolites, and may be involved in their release or transformation, while the other fungi and bacteria, such as Scopulibacillus, Glutamicibacter, etc., were negatively associated with volatile metabolites, or inhibited their production through metabolic competition or catabolism.

4. Discussion

The consensus within the scholarly community maintains that the process of aging contributes to the augmentation of sensory attributes and health-promoting properties in certain types of teas, including dark tea, white tea, and RPT (Lv et al., 2017). In recent years, there has been a surge in popularity of the aged tea market, with a concurrent increase in scholarly interest focused on investigating tea aging. Empirical evidence from studies indicates that the aging of teas has the potential to enhance taste and sensory qualities, introducing distinctive aged aromas to the tea. The aging process of RPT is recognized as a intricate phenomenon leading to discernible alterations in its quality and characteristics over the course of time.

In this investigation, a combination of untargeted and targeted metabolomics methodologies was employed to systematically examine the non-volatile chemical characteristics of ripe Pu-erh tea across various aging durations. The untargeted metabolomics approach applied in this study provided comprehensive insights into the chemical profiles of aged RPT, highlighting the complex transformations that metabolites undergo over time. Using UHPLC-Q-Exactive Orbitrap-MS, a total of 5978 metabolites were identified across both positive and negative ion modes, revealing a diverse spectrum of compounds. These metabolites were classified into 12 distinct chemical groups, with lipids and lipid-like molecules (654), organic acids and derivatives (517), and phenylpropanoids and polyketides (358) being among the most abundant. This classification underscores the broad biochemical complexity of RPTs and the numerous pathways involved in their metabolic evolution during aging (Zhou et al., 2020).Principal component analysis (PCA) clearly differentiated RPTs according to storage time, with RPTs stored for 0 years (Y0) and those stored for 15 years (Y15) showing significant differences.This segregation suggests that aging induces significant chemical changes that are detectable at the metabolite level, consistent with previous findings on the influence of microbial fermentation and oxidative processes on tea composition (Su et al., 2016). Orthogonal projections to latent structure discriminant analysis (OPLS-DA) revealed 118 differential metabolites between four different years of RPT, providing further evidence of the profound chemical shifts that occur with extended storage.Amino acids, known for their contribution to both the health benefits and flavor profile of tea (Zhu et al., 2016), were among the most notable groups in this study. The concentration of L-theanine, a major amino acid in tea associated with its umami taste and calming effects, exhibited distinct trends across the storage periods. In Y0, L-theanine levels were comparatively high, whereas in Y10, there was a notable decrease, possibly indicating the role of microbial processes in modulating amino acid concentrations during aging. N-Phenylacetylaspartic acid, another amino acid derivative, demonstrated lower levels in Y0, with a rise in Y15, further suggesting the influence of aging on the stability and transformation of specific metabolites. The dynamics of these amino acids are important for understanding the flavor evolution of RPTs, particularly the shift from a more bitter profile in younger teas to a smoother, umami-rich profile in older teas. Flavonoids, a class of compounds known for their antioxidant and potential health-promoting properties, also exhibited notable changes (Deng et al., 2021). Specifically, (−)-Epigallocatechin 3,3′-di-gallate, a flavonoid compound, increased significantly with aging, while others such as quercetin and luteolin remained relatively stable. The increase in (−)-Epigallocatechin 3,3′-di-gallate is consistent with its known role in the flavor profile of aged Pu-erh tea, which becomes more aromatic over time. Conversely, catechin showed a steady decrease, which may be associated with the ongoing microbial degradation processes during fermentation (Li et al., 2019). These changes reflect the complex interplay of biochemical processes that occur during the fermentation and storage of RPTs, leading to the gradual evolution of their chemical and sensory properties. The predominance of unclassified metabolites in the dataset highlights the need for further detailed investigations into the less understood components of aged RPTs. While the identified metabolites provide valuable insights into the chemical transformations during aging, the presence of numerous unclassified compounds suggests that there may be other key metabolites yet to be fully characterized. This gap underscores the potential for future studies to explore these unclassified compounds and their roles in the aging process of RPT.

Water extract is an important indicator of mellow and astringent tea flavor. We observed that the water extract content continued to decrease with the increase of storage years, but the content increased instead at 15 years, suggesting that the abundance of RPT diminishes during the first ten years of storage, but increases instead after ten years. With respect to catechins, the study documented a general decline in total catechin content with increasing vintage, particularly Epicatechin (EC), which exhibited a similar trend. This decrease may be attributed to microbial fermentation processes that occur during aging, where the gallic acid-bound catechins undergo hydrolysis, leading to the formation of non-gallic acid-bound catechins (Lv et al., 2015). This transformation likely contributes to the reduced bitterness and astringency commonly associated with RPTs. Epigallocatechin (EGC) increased significantly after ten years of storage, alongside the substantial reduction in gallic acid (GA), further supports the notion that microbial activity and enzymatic processes, including hydrolysis, play a crucial role in shaping the flavor profile of aged RPTs. The reduction in bitterness observed in older RPTs is consistent with previous findings that aging leads to a decrease in catechin content, which is often linked to reduced astringency (Liang et al., 2025). Caffeine, recognized for its bitter taste, also demonstrated a slight downward trend with increasing storage years. This decline in caffeine, concomitant with a parallel decrease in theobromine (its precursor), corresponds with the reduced bitterness perceived in aged RPTs. The decline in caffeine and its precursor over time may be attributable to microbial and chemical degradation processes during fermentation, resulting in a less bitter, more aromatic flavor profile in aged teas. This decline in bitterness is widely regarded as a pivotal factor contributing to the enhanced sensory appeal of ripe Pu-erh teas (Wang et al., 2022a).

GNPS molecular networks serve as effective tools for the identification of novel natural products and the exploration of metabolite diversity in various samples, with widespread applications in tea research (Liu et al., 2012). Using GNPS molecular network analysis, it was observed that RPT stored for 15 years exhibited a reduction in the presence of certain metabolites compared to RPT stored for 0 year. This suggests a potential association between the decrease in these metabolites and microbial metabolic processes. The prolonged storage of RPTs led to complex biochemical transformations that extended beyond initial expectations. Additionally, a novel catechin derivative (m/z: 394.889) was detected in RPT aged for over a decade. This compound is hypothesized to be a derivative of epigallocatechin (EGC), as it was found within the same metabolic cluster as EGC. However, the precise chemical structure of this unidentified compound remains to be determined. Overall, the spectral differences between RPTs have not been fully characterized, and further structural analysis of these spectral variations will be crucial for understanding the material transformations occurring in ripe Pu-erh tea during storage. This will also aid in identifying compounds that are characteristic of the storage duration of RPT.

Aroma is among the most significant sensory characteristics that reflect the quality of tea. The findings of this study highlight the intricate and dynamic nature of volatile organic compounds (VOCs) in RPTs. Untargeted metabolomics analysis showed significant differences between RPTs of different storage years. The VOCs were classified into six primary groups: ketones, alcohols, aldehydes, hydrocarbons, esters, and oxygenated compounds, each contributing distinct characteristics to the aroma of RPTs. Among these, ketones, oxygen heterocycles, aldehydes, and hydrocarbons were identified as the key classes influencing the overall aroma of the tea, consistent with previous studies on the volatile profiles of fermented teas (Fan et al., 2021; Shi et al., 2019; Xu et al., 2020). A notable trend observed in this study is the decrease in the total content of aromatic compounds with prolonged storage, particularly in the categories of esters, aldehydes, and alcohols. While there was an overall increasing trend in hydrocarbon. This aligns with the general understanding that aging improves the aromatic quality of RPTs, as compounds responsible for fresh, grassy notes (e.g., aldehydes) gradually diminish while more complex, matured aromas emerge (Wang et al., 2022a). This shift reflects the transition from green, fresh notes to deeper, more mature and rich aromas characteristic of aged RPT (Tian et al., 2013).

Aged aroma is a predominant olfactory characteristic of RPT, encompassing a complex bouquet of fragrances such as aged, woody, herbal, sweet, camphoraceous, ginseng, lotus, and jujube (Li et al., 2018; Yang et al., 2018). Aged aroma is not solely a singular type of aroma but rather a composite of various aromatic profiles. The primary contributors to aged aroma are methoxybenzene compounds, while woodiness is primarily derived from compounds such as linalool, its oxides, α-terpineol, (+)-cedrol, α-ionone, and β-ionone.Through the identification of compounds with Odor Activity Values (OAVs) greater than 1 and their associated olfactory characteristics in this study, it is demonstrated that the major contributing compounds to the “woodiness” include (+)-cedrol, linalool, its oxides, α-terpineol, and β-ionone. Additionally, the primary contributing compounds to “aged aroma” are 1,2,3-trimethoxybenzene, 1,2,4-trimethoxybenzene, 3,4,5-trimethoxytoluene, and 3,4-dimethoxytoluene. Furthermore, 1-methylnaphthalene, with its medicinal and camphoraceous notes, along with benzyl alcohol, contributing herbaceous and saffron-like aromas, and β-citronellal, imparting sweet and herbaceous notes, can be classified as major contributing compounds to “herbal, sweet, and camphoraceous” aromas. Together, these compounds harmoniously contribute to the formation of the complex olfactory profile known as aged aroma. Alcoholic compounds normally associated with floral aromas increased with storage time and peaked at the RPT of 15 years of storage. Compounds such as (Z)-3-hexen-1-ol, α-terpineol, and phenylethyl alcohol, which contribute to the floral aroma of fresh RPTs, diminished over time, while other compounds like linalool, and geraniol increased with aging. The changes in alcohol content are indicative of the evolving floral and woody characteristics of RPTs, with cinnamyl alcohol, a compound associated with RPT's characteristic aroma, being more abundant in shorter-aged teas. This finding corroborates previous research suggesting that cinnamyl alcohol is a key marker for distinguishing between raw and ripened Pu-erh teas (Xue et al., 2020).Ketones also played a significant role in the aroma profile, with compounds like acetophenone and camphor contributing floral, fruity, and woody notes. Interestingly, the total content of ketones decreased with increasing vintage, although certain ketones, such as phytone, were more abundant in older samples. Phytone, known for its aged aroma, likely contributes to the maturation process, enhancing the depth and complexity of the tea's fragrance over time (Shi et al., 2019). The fluctuating levels of ketones across the different vintages suggest the involvement of both microbial and enzymatic processes during aging, which modify the volatile profile of the tea. Oxygenated compounds, particularly methoxybenzene derivatives, were consistently present at elevated levels in all vintages, with no significant trend of change. These compounds are known to result from the microbial degradation and methylation of tea catechins during fermentation, and their stable concentration suggests a continuous production during storage. Methoxybenzene compounds are particularly important for the characteristic aged and moldy aroma of RPTs, enhancing their depth and complexity (Yang et al., 2018). Hydrocarbons, which contribute to the woody and medicinal aromas of RPT, displayed a relatively lower overall content compared to other VOC classes. However, the content of β-caryophyllene, a key hydrocarbon, increased with aging, suggesting its role in the formation of the pure aged aroma typical of long-stored RPTs. This finding supports the hypothesis that hydrocarbons, particularly methyl naphthalene compounds, play a significant role in the distinct medicinal and woody notes of Pu-erh tea (Wang et al., 2022).

The integration of electronic tongue and electronic nose analyses provides a comprehensive and objective evaluation of the taste and aroma profiles of RPT across different vintages. The findings reveal significant variations in sensory attributes such as bitterness, sourness, aftertaste, and richness, which evolve with storage duration. The observed trends align with the chemical transformations of volatile and non-volatile compounds during aging, highlighting the intricate interplay between taste-active components in shaping the overall flavor profile of RPTs. Bitterness and aftertaste-B (bitter aftertaste) demonstrated a general decline with aging, corresponding to the decreasing caffeine content observed in RPTs over time (Yang et al., 2021). This trend is particularly evident in newly produced RPTs, which exhibit higher scores for aftertaste-B and richness, alongside lower sourness, reflecting the initial robust bitterness and complexity of fresh tea. Conversely, the 5-year RPTs displayed a balanced sensory profile without extreme values for any attribute, suggesting that moderate aging might achieve a harmonious flavor composition. By contrast, RPTs stored for 10 years exhibited higher sourness and bitterness, whereas those stored for 15 years had pronounced bitterness and aftertaste-A (bitter lingering aftertaste), indicating the progressive development of distinct taste characteristics with extended storage.Electronic nose analysis further supports these findings by detecting changes in volatile compounds associated with sensory attributes. For example, sensors sensitive to ethanol, ammonia, amines, propane, butane, carbon, and oxygen compounds (e.g., PA/2, LY2/G, and LY2/gCT) showed lower response values in longer-aged RPTs (Y15) compared to shorter-aged samples (Y0). This result is consistent with the reduction in hydrocarbon content observed in untarget VOC analyses, which may contribute to the diminished bitterness and increased complexity in aged RPTs (Rong et al., 2023). Correlation analysis between electronic tongue and electronic nose data further elucidates the relationships between sensory attributes and chemical profiles. Notably, sensors such as LY2/G, LY2/gCT, PA/2, and P30/1 exhibited strong positive correlations with astringency and sweetness, and strong negative correlations with sourness, bitterness, and saltiness. These correlations suggest that ethanol, ammonia, amines, and hydrocarbons may play critical roles in modulating the flavor attributes of RPTs, influencing the balance between desirable and undesirable taste characteristics.

Microbial communities play a crucial role in shaping the quality and flavor profiles of RPT by driving the decomposition and synthesis of metabolites, as well as contributing to aroma formation (Li et al., 2019; Zhu et al., 2020). The dynamic evolution of microbial composition during storage profoundly influences the chemical transformations in RPTs, as evidenced by high-throughput sequencing of fungal ITS genes and bacterial 16S rDNA.The analysis revealed that Aspergillus was the dominant fungal genus throughout the storage process, underscoring its pivotal role in RPT quality formation. Aspergillus species are known for their involvement in the enzymatic breakdown of tea polyphenols and carbohydrates, facilitating the generation of aroma- and taste-active compounds (Zhu et al., 2020). Specifically, Aspergillus can produce enzymes such as β-glucosidases and esterases, which hydrolyze glycosylated precursors and esterify alcohols, leading to the formation of key volatile compounds like alcohols (e.g., linalool), esters (e.g., ethyl acetate), and aldehydes (e.g., benzaldehyde) that contribute to floral and fruity aromas in RPT during storage (Xu et al., 2020). Other fungal genera, such as Blastobotrys, Rasamsonia, and Thermomyces, were also prevalent, suggesting their contributions to the synthesis and degradation of metabolites. For instance, Blastobotrys may facilitate lipid oxidation pathways, promoting the production of volatile ketones and acids that enhance nutty and fermented notes (Li et al., 2019). Similarly, Bacillus emerged as the most prevalent bacterial genus, which aligns with its known capabilities to produce bioactive compounds that enhance tea flavor and aroma (Li et al., 2019). Bacillus species can secrete lipases and proteases that degrade lipids and proteins into volatile precursors, subsequently transforming them into compounds such as pyrazines and furans through Maillard reactions or microbial fermentation, which impart roasted and caramel-like aromas during the aging process (Xue et al., 2020; Zhu et al., 2020). Additional bacterial genera, including Pullulanibacillus, Paenibacillus, and Kocuria, also exhibited notable relative abundances, reflecting their roles in metabolic processes during storage. The correlation analysis provides deeper insights into the interactions between microbial taxa and tea metabolites. Notably, fungal genera Ascomycota and Mortierellomycota displayed significant correlations with differential metabolites. Ascomycota was negatively correlated with amino acid derivatives (AADs) and carbohydrates and carbohydrate conjugates (CCCs) but positively correlated with flavonoid glycosides (FGs), highlighting its influence on flavonoid metabolism, which contributes to tea's bitterness and astringency. Mortierellomycota showed strong positive correlations with non-volatile metabolites, suggesting its involvement in the synthesis of bioactive compounds (Xue et al., 2020). In terms of bacteria, genera such as Fibrobacterota, sva0485, and Nitrospinota exhibited significant positive correlations with non-volatile metabolites, indicating their roles in modifying the chemical matrix of RPTs. Conversely, bacterial genera Acidobacteriota, Bacteroidota, and Planctomycetota demonstrated negative correlations with volatile metabolites, suggesting their potential roles in modulating the aroma profile of RPTs by suppressing the accumulation of certain volatiles through competitive inhibition or degradation pathways, such as the breakdown of terpenoids into non-aromatic compounds (Zhu et al., 2020).

5. Conclusion

In conclusion, our study on the aging of RPTs has revealed significant insights into its chemical evolution and aroma development. The results revealed substantial shifts in non-volatile metabolites, particularly in amino acids, flavonoids, and catechins. Aging induced significant alterations in the metabolite profile, with compounds such as (−)-Epigallocatechin 3,3′-di-gallate showing an increase over time, contributing to the smoother, more umami-rich flavor of aged RPTs. Additionally, the reduction in catechins and caffeine, combined with microbial fermentation processes, played a key role in diminishing bitterness and astringency, thereby enhancing the overall flavor profile. The identification of 90 VOCs demonstrated that aging positively impacts the overall aroma of RPTs. Notable decreases were observed in aldehydes, alcohols, and heterooxygenated compounds, while hydrocarbons generally increased. Aging promoted a shift in aroma from fresh, green notes to deeper, more mature characteristics, notably the aged aroma, driven by compounds such as 1,2,3-trimethoxybenzene and (+)-cedrol. These findings reveal the complexity of aroma development in RPT. Sensory evaluation via electronic nose and electronic tongue analysis revealed that the taste and aroma profiles of RPTs evolve significantly with aging. Notably, bitterness, sourness, aftertaste, and richness exhibited clear variations. Bitterness and aftertaste decreased over time, correlating with the reduction in caffeine content. The 5-year RPTs displayed a balanced sensory profile, while those aged for 10 to 15 years showed more pronounced bitterness and sourness, reflecting the nuanced changes in flavor with extended aging. Microbial diversity analysis revealed that Aspergillus was the dominant fungal genus, playing a crucial role in the enzymatic breakdown of tea polyphenols and carbohydrates, thereby contributing to aroma and taste-active compounds. Bacillus, the most prevalent bacterial genus, was identified as a key producer of bioactive compounds that enhance flavor and aroma. Correlation analysis highlighted the dynamic interactions between microbial taxa and tea metabolites, with fungal genera such as Ascomycota and Mortierellomycota influencing the synthesis of metabolites related to bitterness, astringency, and bioactive compounds. These findings emphasize the complex role of microbial communities in the fermentation and aging processes that define the sensory characteristics of RPTs. In conclusion, this study provides valuable insights into the chemical, sensory, and microbial transformations that occur during the aging of RPT, underscoring the importance of integrating metabolomics, sensory science, and microbial.

CRediT authorship contribution statement

Liuyu Li: Writing – original draft, Investigation, Data curation, Conceptualization. Yiqiao Zhao: Investigation. Yilong Li: Investigation. Lisha Ran: Investigation. Jinhua Chen: Investigation. Kunbo Wang: Investigation. Zhonghua Liu: Supervision, Conceptualization. Shi Li: Investigation. Jianan Huang: Investigation. Mingzhi Zhu: Writing – review & editing, Project administration, 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.

Data availability

Data will be made available on request.