Region-specific aroma profiles of dark tea: Differences in key odorants are driven by solid-state fermentation with three Aspergillus cristatus strains

Abstract

To investigate the impact of geographically distinct strains of the same fungal species on the aroma of fermented tea, three Aspergillus cristatus strains were isolated from Fu brick teas produced in Shaanxi, Hunan, and Sichuan Provinces, China. These strains were subsequently used as inoculants to produce three distinct loose-leaf dark teas, designated SXFT, HNFT, and SCFT, respectively. Sensory analysis demonstrated that SXFT exhibited stronger flowery, sweet, woody, and minty aroma attributes than HNFT and SCFT. GC–MS analysis identified 89 volatile compounds and showed that total volatile levels increased significantly after fermentation, with SXFT showing the highest concentration (666.07 μg/L). Aroma recombination and omission assays revealed several distinct odorants: methyl salicylate, nonanal, linalool, and 6-methyl-5-hepten-2-one in the SXFT; linalool, β-ionone, and 6-methyl-5-hepten-2-one in the SCFT; and linalool, 6-methyl-5-hepten-2-one, and 1-octen-3-ol in the HNFT. These results provide a chemical basis for targeted starter selection in fermented tea production.

Highlights

• •Three distinct LDTs were prepared using A. cristatum from different regions in China.
• •Sensory attributes and key odorants were differentiated significantly among the LDTs.
• •LDT sample named as SXFT showed both high aroma complexity and high volatile abundance.

1. Introduction

Fu brick tea (FBT) is a dark tea traditionally produced in the Chinese provinces of Hunan, Shaanxi, and Sichuan (Wang et al., 2024). This tea is produced with a fermentation step, and during its characteristic “flowering” stage, Aspergillus cristatus is the dominant microorganism orchestrating aroma development of the FBTs (Wang et al., 2024). Previous studies have shown that inoculating with A. cristatus significantly enhances floral, minty, woody, and sweet aroma attributes by promoting the accumulation of alcohols, esters, and ketones, highlighting the crucial role of this fungal species in modulating the chemistry underlying the aroma of FBT (Shen et al., 2023).

Traditionally, dark tea is typically fermented under open or semi-controlled conditions, where the complexity and viability of natural microbial communities may lead to inconsistencies in aroma quality (Cheng et al., 2024; Du, Yang, Yang, & Yang, 2022). In recent years, producers aim to improve aroma quality and fermentation uniformity of various traditionally fermented foods and have increasingly adopted the use of specific strains for controlled fermentation (Xu et al., 2025; Zhang et al., 2024). Prior research has shown that loose-leaf dark teas (LDTs) fermented with A. cristatus exhibit increased levels of characteristic odorants such as p-methoxybenzaldehyde, coumarin, and methyl salicylate, leading to intensified fungal-flowery aroma notes of the tea (Shen et al., 2023). Solid-state fermentation of LDTs using A. cristatus similarly enhances the accumulation of volatile compounds and elevates sweet and flowery aroma attributes (Xiao et al., 2022). In recent years, several novel dark tea products have been developed and industrialized using a single A. cristatus strain as the starter culture (An et al., 2023; Shen et al., 2024; Shen et al., 2023). However, the differences in aroma formation and key odor-active compounds during dark tea fermentation by A. cristatus strains from different geographical origins have not yet been clarified.

Regional differences in the aroma profiles of FBTs produced across China have been reported. FBTs from Sichuan Province (inland southwest China) typically display vibrant grassy and floral notes, whereas those from Zhejiang Province (southeastern China) possess strong fruity and floral aromas (Xiao et al., 2022; Xiao, Zhong, Bai, Wu, & Gao, 2020; Zhao et al., 2024). Evidence from other fermented foods indicates that strains belonging to the same microbial species but originating from different geographical regions may exhibit distinct metabolic profiles that shape flavor formation. For example, yogurts fermented with four different Bacillus coagulans strains, isolated from Guangxi, Shandong, and Anhui Provinces, showed significant differences in sensory quality, with the Guangxi-derived strain yielding what was considered superior flavor and taste (Ma et al., 2021). Such strain-dependent variability underscores the importance of microbial biogeography in determining sensory outcomes in fermented foods.

In FBT, variations in raw materials and the dominant microbiota may contribute to regional differences in aroma, but emerging findings indicate that the geographical origin of the A. cristatus also influences fermentation performance and volatile formation (Li, Wang, et al., 2024). Recent work shows that region-specific A. cristatus strains can modulate microbial assembly and reshape the volatile profile during FBT fermentation (Xiao et al., 2024). Despite these insights, a systematic evaluation of LDT aroma quality resulting from controlled fermentation using A. cristatus strains from different geographical origins remains lacking.

In this work, three A. cristatus strains were isolated from commercial FBT samples produced in three traditional production regions in China, namely Shaanxi, Hunan, and Sichuan provinces. These strains were used to inoculate tea leaves to prepare three corresponding LDTs (SXFT, HNFT, and SCFT) through solid-state fermentation. We integrated Quantitative Descriptive Analysis (QDA), Solvent-Assisted Flavor Evaporation–Gas Chromatography–Mass Spectrometry (SAFE–GC–MS), Gas Chromatography–Mass Spectrometry/Olfactometry (GC–MS/O), Odor Activity Value (OAV) analysis, and aroma recombination and omission experiments to elucidate how three geographically isolated A. cristatus strains influence aroma formation. This layered analytical approach provided a comprehensive assessment of strain-specific contributions to key odorants and perceived aroma in loose-leaf fermented teas.

2. Materials and methods

2.1. Materials and standard compounds

Green tea leaves were the raw material and were provided by Huangshan Seeking Tea Co. Ltd. (Anhui, China). The three FBTs were purchased from Jingwei Fu Tea Co., Ltd. (Shaanxi, China), Xiang Yi Fu Tea Co., Ltd. (Hunan, China), and Pingwu Xuebao Ding Tea Co., Ltd. (Sichuan, China).

Methyl salicylate (99.0%), phenethyl alcohol (99.5%), linalool (98.0%), geranylacetone (98.0%), benzyl alcohol (99.5%), phenylacetaldehyde (98.0%), coumarin (99.0%), acetophenone (99.5%), 4-ketoisophorone (95.0%), ethyl decanoate, n-alkane (C7–C25) were purchased from Sigma-Aldrich (St. Louis, USA). α-Terpineol (96.0%), 2-acetylpyrrole (99.0%), nonanal (96.0%) 4-methoxybenzaldehyde (98.0%), β-Ionone (97.0%), and 1-octen-3-ol (96.0%) were purchased from Shanghai Yuanye Bio-Technology Co., Ltd. (Shanghai, China).

2.2. Isolation of A. cristatus from different FBTs

Microorganisms were isolated from three FBTs exhibiting “golden flower” characteristics using the plate dilution method. Three A. cristatus isolates were screened, including HNAC, from FBT produced by Xiang Yi Fu Tea Co. Ltd., Hunan Province, China; SXAC, from FBT produced by Jingwei Fu Tea Co. Ltd., Shaanxi Province, China; and SCAC, from FBT produced by Pingwu Xuebao Ding Tea Co. Ltd., Sichuan Province, China. Each strain (HNAC, SXAC, SCAC) was preserved separately on Potato Dextrose Agar (PDA) medium at 4 °C for the next experiments. For each of the three A. cristatus strains, colony morphology on different media (PDA, Rose Bengal Agar, and high-osmotic-pressure PDA medium) and reproductive structures were analyzed by microscopy and scanning electron microscopy (SEM). A phylogenetic tree including all three isolated strains was generated using sequences of related Aspergillus species downloaded from NCBI. A schematic diagram of the experimental workflow is shown in Fig. S2.

2.3. Preparation of loose-leaf dark tea

Preparation of loose-leaf dark teas (LDTs) was carried out using methods described previously (Shen et al., 2023), with slight modifications. Briefly, the three A. cristatus strains (HNAC, SXAC, and SCAC) were individually activated on the PDA medium for 3–5 days. Golden mycelia growing on the PDA medium were gently scraped and suspended in sterile water to prepare spore suspensions, respectively. The raw material, sterilized green tea (105 °C, 10 min), was inoculated with one of the three spore suspensions (HNAC, SXAC, or SCAC) at a concentration of 1 × 10⁴ CFU/g of tea leaves. The inoculated samples were then incubated separately at 28 °C and 70% relative humidity for 10 days to obtain three LDT samples, designated as HNFT, SXFT, and SCFT, respectively. The sterilized green tea raw material without inoculation (GT) was used as the control sample.

2.4. Sensory evaluation

The training of the assessors was conducted following the method described by Shen et al. (2023). The sensory panel consisted of 10 trained evaluators (5 males and 5 females) aged between 22 and 27 years. Prior to the sensory evaluation, all panelists provided written informed consent. The study protocol was approved by the Ethics Committee of Anhui Agricultural University (Approval No. KJLL2025016). Four tea samples (GT, HNFT, SXFT, and SCFT) and ten aroma attribute descriptors (flowery, musty, stale, minty, woody, fruity, sweet, mushroom, green, and caramel) were submitted to the panelists. Subsequently, six aroma attribute descriptors were discussed and identified by the assessors: flowery, minty, woody, fruity, sweet, and green. Reference samples were dry mint leaves (minty), dry rose bud (flowery), fresh grass (green), sweet orange (fruit), sandalwood (woody), and honey (sweet) (Table S1). The intensities of the six aroma attributes across the four samples were evaluated by assessors using a 0 (not perceptible) to 9 (very strong) scale.

2.5. Analysis of volatile compounds by GC–MS/O

Volatile compounds were extracted from tea infusions using solvent-assisted flavor evaporation (SAFE), following a previously reported method, with minor modifications (Shen et al., 2023). Specifically, 5 g of tea was weighed into a conical flask and closed with a stopper until use. To make the tea infusion, 150 mL of boiling deionized water was added. After 5 min, the tea infusion was filtered using a 200-mesh filter and then cooled in an ice bath. Ethyl decanoate (internal standard, 5 μL, 500 ppm) was added to the tea infusion before distillation under a high vacuum at 45 °C using the SAFE technology.

Volatile compounds and odorants were identified using a gas chromatograph (7890B, Agilent, CA, USA) equipped with a mass spectrometer (5977B, Agilent, CA, USA) and an olfactory detection port (Gerstel, Mülheim an der Ruhr, Germany) (GC–MS/O). Separation was achieved on a DB-5MS capillary column (30 m × 0.25 mm i.d., 0.25 μm film thickness; J&W Scientific, Shanghai, China). The sample (2 μL) was subjected to the GC column via a non-split mode inlet. The gas chromatography oven temperature was programmed as follows: 40 °C (held for 5 min); 40–100 °C at a rate of 6 °C/min, 100–200 °C at a rate of 3 °C/min, 200–280 °C at a rate of 20 °C/min, with a final hold at 280 °C for 5 min. The effluent from the GC column was split post-column between the mass spectrometer and the olfactory detection port at a 1:1 ratio.

GC-O analysis with aroma intensity (AI) evaluation was used to identify aroma-active compounds in LDTs according to the method of Zhai, Zhang, Granvogl, Ho, and Wan (2022). The odor characteristics and aroma intensities of each sample were evaluated by three experienced assessors. Prior to evaluation, panelists received at least two weeks of training using real food samples and aqueous solutions of reference odorants with characteristic aromas. During GC–O analysis, an odor event was considered valid only when it was consistently perceived by at least two of the three panelists. The intensity of each odor was scored on a scale from 0 to 3, where “0” indicated no detectable odor and “3” represented a strong aroma intensity.

Using the mass spectrometer, volatile compounds were semi-quantified using ethyl decanoate as an internal standard. After GC-O analysis by the panelists, 20 aroma-active compounds with AI >1.5 were quantified by establishing standard curves of these aroma-active compounds using their authentic standards.

2.6. Odor activity value (OAV) calculation

The odor threshold (OT) values of the target compounds were obtained from previously published literature (Bao et al., 2025; Ma et al., 2025; Shen et al., 2023; Zhang et al., 2025). The OAV for each compound is calculated according to the following formula (1):

$$\mathrm{OAV}=\frac{\mathrm{C}}{\mathrm{OT}}$$ (1)

where C is the concentration of the volatile compound (μg/L) and OT is its corresponding odor threshold (μg/L).

2.7. Aroma recombination and omission experiment

The deodorized tea infusion matrix was prepared as follows: 3.0 g of fermented tea was infused in 150 mL of boiling water for 5 min, and the resulting tea infusion was obtained by suction filtration. The tea infusion was rotary evaporated in a water bath at 60 °C. The evaporation was repeated more than 5 times until the sample became odorless. Then, the deodorized tea matrix taken to 150 mL with deionized water. The main aroma-active compounds (AI >1.5, OAV > 1; 12 compounds) were added to the deodorized tea infusion matrix according to their corresponding concentrations in the original tea (SXFT, HNFT, and SCFT) infusion to prepare the reconstructed samples (re-SXFT, re-HNFT, and re-SCFT).

For aroma omission experiments, individual odorants were omitted one by one from the recombined model teas to generate different omission samples. Each comparison was assessed by 15 trained panelists, and each test was performed in triplicate. The three-point matching test was used to evaluate sensory differences between the omission samples and the recombination sample. A difference was considered significant when the number of correct identifications exceeded the critical value at p < 0.05 (Zhang et al., 2025).

2.8. Statistical analysis

Statistical analyses were performed using SPSS (Version 26.0, IBM Corp., New York, USA). Multivariate statistical analysis (Partial Least-Squares Discriminant Analysis) was carried out using SIMCA-P (Version 14.1, Umetrics, Umea, Sweden). Radar and bar charts were generated using OriginPro 2025b (version 10.1.5.132; OriginLab, USA). All experiments were performed and samples created in triplicate.

3. Results and discussion

3.1. Sensory characteristics of LDTs

Green tea (GT) was inoculated with the three new A. cristatus isolates (SX, HN and SC). After solid-state fermentation, the loose-leaf teas were labeled SXFT, HNFT, and SCFT. The GT and three fermented tea samples (dry tea leaves and tea infusions) were photographed and scored by the sensory panel (Fig. 1). A distinct “golden flower fungus” (A. cristatus) was observed, as a more golden and lustrous appearance relative to GT, on the surface of the dry LDT samples. The GT infusion was a yellow-green colour, while the LDT infusions were more orange-red (Fig. 1A).

Fig. 1.

Photographs of the dry tea leaves and tea infusions of GT, SXFT, HNFT and SCFT (A); Radar plots showing aroma attribute intensities (B) and taste attribute intensities (C) of the four teas. GT: unfermented green tea, which served as the raw material for fermentation; SXFT: loose-leaf dark tea fermented by A. cristatus isolated from Shaanxi Fu brick tea (China); HNFT: loose-leaf dark tea fermented by A. cristatus from Hunan Fu brick tea (China); SCFT: loose-leaf dark tea fermented by A. cristatus from Sichuan Fu brick tea (China). (For interpretation of the references to colour in this figure legend, the reader is referred to the web version of this article.)

The aroma attributes of the four samples were shown as radar plots (Fig. 1B). In comparison with GT, the three fermented LDTs demonstrated enhanced flowery, woody, fruity, sweet, and minty aroma attributes, while the green note significantly weakened. Among the LDTs, SXFT exhibited higher intensities in flowery, woody, sweet, and minty aroma attributes compared to HNFT and SCFT. In contrast, SCFT presented a more pronounced fruity aroma. Fermentation substantially improved the overall taste profile of SXFT, HNFT, and SCFT, reducing sourness, astringent, and bitterness while enhancing mellowness, umami, and aftertaste sweetness (Fig. 1C).

Fermentation of tea with A. cristatus has been reported to significantly reduce the grassy aroma while enhancing flowery, woody, and minty aromas (Shen et al., 2023). Similarly, fermentation with A. cristatus reduces fresh aroma and increases the stale and fungal-flowery notes when Shoumei white tea is used as the raw material (Zhu et al., 2024). Furthermore, tea infusions of FBT inoculated with A. cristatus from different sources show colors ranging from orange-red to reddish-brown and have different flavor profiles (Xiao et al., 2024). Consistent with these studies, our findings showed that LDTs fermented with A. cristatus isolates from different regions develop distinguishable aroma characteristics.

3.2. Volatile compounds analysis

To elucidate the differences in volatile compound (VC) profiles among LDT samples fermented with A. cristatus from different geographical regions, SAFE combined with GC–MS was employed to analyze the VCs in GT, HNFT, SXFT, and SCFT. A total of 89 VCs were identified across all samples, including 22 alcohols, 17 ketones, 7 aldehydes, 11 esters, 4 pyrroles, 10 acids, 7 heterocyclic compounds, and 11 hydrocarbons (Table S2).

The VC contents in the GT, SXFT, HNFT and SCFT are shown in Fig. 2. In the four tea samples, the VC content was highest in the SXFT, at 666.07 μg/L, while the VC content in the GT control was only 372.09 μg/L (Fig. 2A). These findings suggest that fermentation with A. cristatus can increase the volatiles content in LDTs. Fermentation also shifted the dominant classes to alcohols, esters, and ketones, accounting for over 85% of the VCs compared to less than 50% in the GT (Fig. 2B). While the proportions between the volatile classes were similar among the three fermented LDTs, notable differences were detected in the contents of alcohols, esters, and ketones (Fig. 2B and C).

Fig. 2.

Contents of volatile compounds among the GT, SXFT, HNFT and SCFT. The content of total volatile compounds in each tea infusion (A); Percentage of total content of each volatile compound category (B); Content of different types of volatile compounds (C). GT: unfermented green tea, raw material; SXFT: loose-leaf dark tea fermented by A. cristatus from Shaanxi Fu brick tea (China); HNFT: loose-leaf dark tea fermented by A. cristatus from Hunan Fu brick tea (China); SCFT: loose-leaf dark tea fermented by A. cristatus from Sichuan Fu brick tea (China). (For interpretation of the references to colour in this figure legend, the reader is referred to the web version of this article.)

3.2.1. Alcohols

Alcohols accounted for the largest proportion in all four types of tea, but the levels of alcohols increased significantly in the fermented teas compared to the GT (Fig. 2C). The content of alcohol compounds was highest in SXFT (311.99 μg/L), followed by HNFT (267.90 μg/L) and SCFT (241.55 μg/L).

Alcohols are abundant in FBT and teas fermented with A. cristatus, contributing flowery, fruity, and woody aromas to the tea (Wang et al., 2024). In A. cristatus mycelium and spores, alcohol compounds, especially 1-octen-3-ol, are key VCs, with mushroom-like aroma. In addition, linalool and linalool oxide have been detected in FBT, where they serve as odorants contributing to flowery, sweet, and woody aromas (Zheng et al., 2025). During A. cristatus mycelial growth, the extracellular enzymes secreted by the fungus are considered the primary drivers for the accumulation of alcohols such as 1-octen-3-ol, linalool, linalool oxide I and linalool oxide II, and these alcohols contribute to the flowery aroma of FBT (Wang et al., 2024; Zhao et al., 2024).

3.2.2. Esters

Ester compounds were significantly elevated in LDT samples fermented with the three A. cristatus from different regions (Fig. 2C). Among the three LDTs, the content of esters was highest in SXFT (100.62 μg/L), followed by SCFT (90.57 μg/L) and HNFT (82.13 μg/L). Methyl salicylate was the most abundant ester compound in all three fermented samples, accounting for 73.14%, 69.21%, and 68.91% of the total ester content in SXFT, SCFT, and HNFT, respectively (Fig. S4).

Methyl salicylate, which imparts a minty aroma, is present at low levels in unfermented tea but accumulates to significantly higher levels in tea fermented with A. cristatus (Shen et al., 2023). The accumulation of methyl salicylate, a key odor-active compound, is likely associated with the activity of glycosyl hydrolases abundantly secreted by A. cristatus (Shen et al., 2023; Wang et al., 2024).

3.2.3. Ketones

Ketones were the third most abundant class of VCs in the tea infusions. Sixteen volatile ketones were identified across the three LDTs (Table S2). The content of ketone compounds in SXFT (52.60 μg/L) was notably higher than that in HNFT (34.87 μg/L) and SCFT (37.50 μg/L) (Fig. 2C).

Ketones, including ionone, geranylacetone, acetophenone, and 6-methyl-5-hepten-2-one, have been reported as characteristic flavor compounds in teas fermented with A. cristatus, contributing flowery, fruity and woody attributes (; Zheng et al., 2025). Among these compounds, β-ionone and geranylacetone have been reported to contribute significantly to the flowery and fruity aromas of FBTs with both stale-fungal and fresh-fungal aromas. In addition, 6-methyl-5-hepten-2-one is recognized as an important contributor to the fresh aroma of fresh-fungal and ripe-fungal aroma-type FBTs (Zheng et al., 2023).

3.3. Analysis of differential VCs

Partial Least-Squares Discriminant Analysis (PLS-DA) was employed to compare the differences in the VC profiles in the four tea infusions (Fig. 3). The LDTs were differentiated from the GT in the two-dimensional space (Fig. 3A). The SXFT was separated from HNFT and SCFT, while HNFT and SCFT were overlapping, indicating greater similarity between HNFT and SCFT. Samples were also grouped into three categories (GT, SXFT, and HNFT-SCFT) by hierarchical cluster analysis (Fig. 3B). To further explore the differences in VC profiles among the LDTs fermented by A. cristatus from different regions, PLS-DA was conducted for the volatile data from just the SXFT, HNFT, and SCFT samples (Fig. 3C). The permutation test confirmed the robustness and reliability of the PLS-DA model (Fig. 3D).

Fig. 3.

PLS-DA of the volatile compounds in GT, SXFT, HNFT and SCFT (A); Hierarchical cluster analysis of volatile compounds in the four tea samples (B); PLS-DA of volatile compounds of the three LDTs (C); Permutation test of the PLS-DA (D); Heatmap analysis of differential volatile compounds (E). GT: unfermented green tea, raw material; SXFT: loose-leaf dark tea fermented by A. cristatus from Shaanxi Fu brick tea (China); HNFT: loose-leaf dark tea fermented by A. cristatus from Hunan Fu brick tea (China); SCFT: loose-leaf dark tea fermented by A. cristatus from Sichuan Fu brick tea (China). (For interpretation of the references to colour in this figure legend, the reader is referred to the web version of this article.)

Based on VIP > 1 and p < 0.05, a total of 16 differentially volatile compounds were identified in LTDs fermented by A. cristatus from the three regions (Fig. 3E). Several volatiles were more abundant in SXFT, including linalool, geranylacetone, linalool oxide I, linalool oxide IV, α-terpineol, methyl salicylate, and phenethyl alcohol. Previous studies have reported that methyl salicylate (imparting a minty note), linalool (flowery), and linalool oxide I (flowery) are typical volatile components in FBT, contributing to its characteristic fungus-flower aroma during fermentation (Li et al., 2020; Wang et al., 2024). Geranylacetone, known for its flowery and sweet aroma attributes, is high in Qing brick tea and Liupao tea (Ma et al., 2023). Moreover, methyl salicylate, linalool, linalool oxides, and α-terpineol have been associated with the fungal-flower aroma attribute in Liupao tea (Li et al., 2022).

In HNFT, linalool oxide II, benzyl alcohol, 1-octen-3-ol, and benzaldehyde were more abundant. Benzyl alcohol and benzaldehyde contribute a bitter almond-like aroma in FBT (Li, Wang, et al., 2024). 1-Octen-3-ol, which imparts a mushroom-like aroma, has been reported to occur in glycosylated form in soybeans and can be released via β-glucosidase-catalyzed hydrolysis of its glycosides (Matsui, Takemoto, Koeduka, & Ohnishi, 2018). In SCFT, several VCs, such as lavender lactone, 2-phenoxyethanol, and β-ionone, were present at relatively high levels. β-Ionone, characterized by a flowery aroma, serves as a key odor-active compound in premium-grade FBT and has been detected in FBT samples from seven different production regions across China (Li, Wang, et al., 2024).

3.4. Identification of aroma-active compounds using GC–MS/O and AI

Aroma-active compounds in the LDTs (SXFT, HNFT, and SCFT) were screened using GC–MS/O in combination with aroma intensity (AI) evaluation by human assessors, considering compounds with AI >1.5 as significant. A total of 27 odorants were detected and identified by GC–MS /O. Among them, the AI values of 20 compounds were greater than 1.5 (Table 1).

Table 1.

Odorants identified by GC–MS/O in LDTs fermented by different A. cristatus isolates.

Compounds	CAS	Odor description	Aroma intensity
			SXFT	HNFT	SCFT
Methyl salicylate	119-36-8	Minty	2.88 ± 0.10	2.50 ± 0.14	2.42 ± 0.21
Phenethyl alcohol	60-12-8	Flowery, honey-like	2.00 ± 0.20	2.13 ± 0.15	2.45 ± 0.28
α-Terpineol	98-55-5	Flowery, citrus, woody	2.22 ± 0.46	2.05 ± 0.15	2.15 ± 0.23
Nonanal	124-19-6	Citrus, soapy	2.20 ± 0.10	2.50 ± 0.25	2.75 ± 0.15
Linalool oxide II	34995-77-2	Sweet, flowery, creamy	1.80 ± 0.10	2.24 ± 0.12	1.76 ± 0.06
Linalool	78-70-6	Flowery, sweet, grape-like	3.00 ± 0.00	2.45 ± 0.20	2.63 ± 0.15
Geranylacetone	3796-70-1	Fresh, flowery, fruity	2.02 ± 0.10	1.55 ± 0.30	1.76 ± 0.11
1-Octen-3-ol	3391-86-4	Earthy, fungal	2.50 ± 0.10	2.12 ± 0.15	2.30 ± 0.10
Benzyl alcohol	100-51-6	Bitter almond-like, fruity	1.53 ± 0.22	2.13 ± 0.20	2.15 ± 0.05
4-Methoxybenzaldehyde	123-11-5	Woody	2.06 ± 0.24	1.85 ± 0.10	1.70 ± 0.35
Isophorone	78-59-1	Woody, sweet, fruity	1.60 ± 0.11	1.55 ± 0.12	1.53 ± 0.12
4-Ketoisophorone	1125-21-9	Woody, fruity, musty	1.86 ± 0.16	1.70 ± 0.10	1.50 ± 0.22
β-Ionone	79-77-6	Fruity, Woody, fruity	1.53 ± 0.11	1.53 ± 0.20	1.85 ± 0.18
Coumarin	91-64-5	Sweet, bitter almond-like	2.08 ± 0.14	1.80 ± 0.10	2.20 ± 0.13
1-ethyl-1H-pyrrole-2-carbaldehyde	2167-14-8	Leather-like	0.95 ± 0.35	1.21 ± 0.22	1.30 ± 0.25
2-Acetylpyrrole	1072-83-9	Nutty, musty, sweet	1.51 ± 0.33	1.75 ± 0.33	1.74 ± 0.35
Hexanoic acid	142-62-1	Sweat, pungent,	1.98 ± 0.18	1.52 ± 0.42	1.71 ± 0.23
Phenylacetaldehyde	122-78-1	Flowery, rose	2.15 ± 0.22	2.12 ± 0.24	2.25 ± 0.15
Acetophenone	98-86-2	Flowery	1.83 ± 0.16	2.05 ± 0.55	1.95 ± 0.25
6-Methyl-5-hepten-2-one	110-93-0	Fruity, sweety	2.07 ± 0.10	1.98 ± 0.30	1.95 ± 0.15

SXFT: LDT fermented by A. cristatus from Shaanxi Fu brick tea (Shaanxi, China); HNFT: LDT fermented by A. cristatus from Hunan Fu brick tea (Hunan, China); SCFT: LDT fermented by A. cristatus from Sichuan Fu brick tea (Sichuan, China).

In SXFT, linalool, methyl salicylate, and 1-octen-3-ol showed high AI values, contributing floral, minty, and mushroom aromas, respectively. HNFT was characterized by elevated AI values for nonanal, methyl salicylate, linalool, and linalool oxide II, which contributed to its fruity, minty, and floral aroma characteristics. In SCFT, nonanal exhibited the highest AI value, followed by linalool and phenethyl alcohol, imparting sweet and flowery aroma attributes.

Previous studies have also identified methyl salicylate and nonanal as the main odorants in LDT fermented by A. cristatus (Shen et al., 2023). Previous studies have indicated that methyl salicylate and nonanal contribute to the fresh mint-like aroma, while linalool and its oxides give floral sensory attributes in FBT (Xiao, Huang, Chen, Xiao, et al., 2022; Zhai et al., 2022). 1-Octen-3-ol exhibits a typical mushroom or soil aroma, which has been identified as a critical odorant contributing to both the complexity and intensity of the aroma profile in FBT (Wang et al., 2024; Zheng et al., 2023).

3.5. Quantitative analysis of the aroma-active compounds

The external standard method was used to quantify the concentrations of 20 aroma-active compounds (AI >1.5) identified by GC-O. Based on the OAVs calculated from these externally quantified concentrations, 12 aroma-active compounds with OAV > 1 were identified across the three LDT samples (Table 2), suggesting their potential contributions to the overall aroma profiles. Linalool exhibited the highest OAV (204.49) in the SXFT sample, followed by geranylacetone (111.99), β-ionone (105.75), 6-methyl-5-hepten-2-one (80.88), and nonanal (41.70). In the HNFT sample, linalool (161.18) had the highest OAV, followed by β-ionone (98.52), 6-methyl-5-hepten-2-one (90.96), geranylacetone (25.79), and nonanal (18.76). In contrast, β-ionone exhibited the highest OAV (130.69) in the SCFT sample, followed by 6-methyl-5-hepten-2-one (90.65), linalool (86.37), geranylacetone (20.32), and nonanal (13.71).

Table 2.

Odorants with odor active values greater than 1 (OAV > 1) in LDTs fermented by different A. cristatus isolates.

Compounds	Odor description	Content (μg/L)			OT (μg/L)	OAV
		SXFT	HNFT	SCFT		SXFT	HNFT	SCFT
Methyl salicylate	Minty	154.47 ± 7.11a	119.51 ± 7.88b	131.12 ± 4.68b	40	3.86	2.99	3.28
Phenethyl alcohol	Flowery, honey-like	14.08 ± 0.48a	12.15 ± 0.44b	13.72 ± 0.57a	1	14.08	12.15	13.72
α-Terpineol	Floral, citrus, woody	11.74 ± 0.06a	9.71 ± 0.48b	8.98 ± 0.74b	4.08	2.88	2.38	2.20
Nonanal	Citrus-like	13.34 ± 0.39a	6.00 ± 0.45b	6.50 ± 0.20b	0.32	41.70	18.76	20.32
Linalool	Floral, sweet	44.92 ± 4.06a	35.46 ± 0.63b	19.00 ± 0.52c	0.22	204.19	161.20	86.37
Geranylacetone	Fresh, floral, fruity	6.72 ± 0.02a	1.55 ± 0.11b	1.62 ± 0.37b	0.06	111.99	25.79	27.07
1-Octen-3-ol	Earthy, green, fungal	9.80 ± 1.47ab	11.55 ± 0.15a	9.48 ± 0.40b	1	9.80	11.55	9.48
Benzyl alcohol	Bitter almond-like, fruity	9.22 ± 0.98c	23.178 ± 1.18a	20.60 ± 0.88b	3	3.07	7.73	6.87
β-Ionone	Fruity, woody	11.00 ± 1.35b	10.25 ± 0.73b	13.59 ± 0.85a	0.104	105.75	98.52	130.70
Coumarin	Sweet, woody	15.27 ± 0.16ab	14.51 ± 1.01b	16.09 ± 0.49a	11	1.39	1.32	1.46
Phenylacetaldehyde	Floral, rose	4.38 ± 0.01a	3.60 ± 0.63ab	3.37 ± 0.40b	4	1.10	0.90	0.84
6-Methyl-5-hepten-2-one	Fruity, sweety	12.94 ± 1.90a	14.55 ± 1.48a	14.50 ± 0.89a	0.16	80.88	90.97	90.65

OT: Odor thresholds. Different letters indicate significant differences (p < 0.05) between samples.

Notably, these aroma-active compounds, which include alcohols, ketones, aldehydes, and esters, are consistent with previously identified key odorants in FBT from different sources (Li, Wang, et al., 2024). Linalool and geranylacetone show flowery and sweet aroma attributes, contributing significantly to the aroma of FBT with a stale-fungal aroma (Zheng et al., 2023). The ketone 6-methyl-5-hepten-2-one, with a sweet aroma, has been identified in Yunnan Pu-er tea and Guangxi Liupao tea (Li et al., 2022; Wang et al., 2025).

3.6. Confirmation of key odorants using recombination and omission experiments

3.6.1. Reconstitution of aroma profiles

Aroma recombination experiments were used to validate the 12 key odor-active compounds (OAV > 1) to assess their contribution to the individual aroma profiles of the three LDTs. The 12 aroma-active compounds were added to deodorized tea infusions according to their respective concentrations in the three original tea infusions (Table S3). The sensory evaluation of the reconstituted samples (re-SXFT, re-HNFT, and re-SCFT) were compared to those of the original tea samples (Fig. 4). The aroma attribute intensities in the reconstituted samples showed a similar trend with the corresponding original LDTs. In the re-SXFT sample, the intensities of flowery, minty, fruity, and sweet aroma attributes closely resembled the original SXFT infusion, although the intensities of flowery and fruity attributes were slightly higher, while minty, woody, and sweet were slightly lower (Fig. 4A). The recombined model of HNFT (re-HNFT) also closely resembled the original HNFT, with a slightly higher intensity of the flowery attribute and lower intensities of the other attributes in re-HNFT (Fig. 4B). The aroma intensities in the reconstituted model of SCFT (re-SCFT) matched the SCFT, except the flowery and minty aroma attributes showed minor discrepancies (Fig. 4C).

Fig. 4.

Aroma profiles of LDTs and recombined tea models. SXFT and re-SXFT (A), HNFT and re-HNFT (B), SCFT and re-SCFT (C). SXFT: LDT fermented by A. cristatus from Shaanxi Fu brick tea (Shaanxi, China); HNFT: LDT fermented by A. cristatus from Hunan Fu brick tea (Hunan, China); SCFT: LDT fermented by A. cristatus from Sichuan Fu brick tea (Sichuan, China); re-SXFT: recombination aroma model of SXFT, re-HNFT: recombination aroma model of HNFT, re-SCFT: recombination aroma model of SCFT; *: p < 0.05.

The minor differences between the original teas and their recombinant models may be explained by interactions among aroma compounds. For instance, sweet-flowery compounds such as linalool may be released more rapidly in recombination models than in the original tea matrix, potentially masking woody notes (Zhao et al., 2024). Overall, the results indicate that the 12 key odorants identified in this work are the primary components that form the fundamental aroma profiles of each LDT.

3.6.2. Omission experiments of aroma compounds

Aroma omission experiments were then used to evaluate the contribution of 12 individual odorants to the overall aroma profiles of the tea infusions (Table 3). Compared to re-SXFT, the tea infusions lacking linalool, methyl salicylate, or nonanal showed significant differences (p < 0.001). Similarly, in HNFT, removing linalool, 1-octen-3-ol, or 6-methyl-5-hepten-2-one caused significant differences (p < 0.001). In SCFT, the absence of 6-methyl-5-hepten-2-one, β-ionone, linalool, or nonanal also caused significant changes (p < 0.001).

Table 3.

Triangle test results of omission experiments for each key aroma-active compound in LDTs.

Key aroma-active compoundsa	SXFT		HNFT		SCFT
	n(correct)b	Significancec	n(correct)	Significance	n(correct)	Significance
Methyl salicylate	12	***	10	**	8	*
Phenethyl alcohol	6	−	7	−	8	*
α-Terpineol	7	−	8	*	10	**
Nonanal	12	***	8	*	12	***
Linalool	15	***	13	***	12	***
Geranylacetone	10	**	6	−	7	−
1-Octen-3-ol	10	**	12	***	9	**
β-Ionone	8	*	8	*	13	***
Benzyl alcohol	5	−	8	*	7	−
Coumarin	8	*	9	**	10	**
Phenylacetaldehyde	4	−	6	−	7	−
6-Methyl-5-hepten-2-one	10	***	12	***	14	***

^aThrough GC-O and OAV analysis, 12 volatile compounds were selected which had important contribution to aroma characteristic of LDTs.

^bThe number of judges that correctly identified aromas among the 15 assessors obtained through the triangle test.

^cSignificance: *, significant (p ≤ 0.05); **, highly significant (p ≤ 0.01); ***, very highly significant (p ≤ 0.001); −, non-significant.

Additionally, the absence of geranylacetone, coumarin, or α-terpineol led to significant differences (p < 0.01) among all three reconstituted teas (re-SXFT, re-HNFT, and re-SCFT), whereas the removal of phenylacetaldehydes showed no significant impact (p > 0.05). Notably, the absence of linalool and 6-methyl-5-hepten-2-one showed highly significant differences (p < 0.001) in all models. These aroma-active compounds, contributing to flowery and sweet aromas, have been detected in FBTs from various regions and reported as biomarkers for distinguishing the aroma characteristics of FBTs from different geographic origins (Xiao, Huang, Chen, Xiao, et al., 2022). The importance of 1-octen-3-ol in the aroma profile of FBT has been validated through a similar method, and β-ionone has been shown to play a key role in its flowery aroma (Li, Wang, et al., 2024; Zhao et al., 2024).

In this study, the three LDTs fermented with A. cristatus from different regions exhibited different key odorant profiles. This variation could be due to strain metabolism, as demonstrated in studies of fermented orange peel by Aspergillus niger, where different strains led to varied volatile compositions (Wang, Bo, et al., 2025). Similarly, variations in the extracellular enzymes secreted by A. cristatus from different regions can significantly influence the formation of volatile compounds in FBT (Wang et al., 2024). In addition to the composition of volatile compounds, the overall tea aroma is influenced by interactions between volatile and non-volatile components, which together shape flavor perception (Song et al., 2025). Furthermore, olfactory perception is influenced by synergistic, masking, and additive effects among aroma compounds (Tian, Yu, Yu, & Chen, 2020).

3.7. Possible terpene metabolic pathways

In this study, the contents of 11 terpenoid aroma compounds were significantly increased after fermentation (Table S2). Based on KEGG pathway analysis, a possible metabolic network for terpene biosynthesis in A. cristatus was inferred (Fig. 5). Geranyl pyrophosphate (GPP), synthesized via the mevalonate (MVA) pathway, is a key precursor for monoterpene biosynthesis (Li, Yao, & Zeng, 2024). Under the catalysis of terpene synthases, GPP goes through cyclization and rearrangement intermediates to form both cyclic and acyclic monoterpenes (Li, Yao, & Zeng, 2024). Noncyclic monoterpenoids such as linalool and β-myrcene are known to give flowery aroma to dark teas (Wang et al., 2025; Wen et al., 2023). During the fermentation of dark tea, linalool can be further oxidized by non-specific enzymes to form linalool oxides that enhance its flowery and woody aromas (Zheng et al., 2025). Cyclic terpenes, including α-terpineol, terpinen-4-ol, β-pinene, and 3-carene, have been identified as woody aroma compounds in dark tea (Chen et al., 2025). β-Phellandrene has been recognized as a key aroma compound of Fenghuang Dancong oolong tea, contributing a minty aroma (Qin et al., 2023). d-limonene, characterized by citrus-like and flowery aromas, has been reported as a key aroma-active compound in tea infusion (Liu et al., 2021).

Fig. 5.

Potential biosynthetic pathways of terpenoid-related compounds in LDTs fermented by A. cristatus. The squares represent the relative content of compounds in different tea samples fermented by the different isolates as well as the control green tea. (For interpretation of the references to colour in this figure legend, the reader is referred to the web version of this article.)

Previous studies indicate that various monoterpene alcohols with flowery aromas were generated through the monoterpene biosynthetic pathway in A. cristatus. Monoterpene alcohols such as linalool and α-terpineol enhance the aroma qualities of summer and autumn teas (Xiao et al., 2021). Moreover, fermentation with A. cristatus has been shown to elevate the levels of terpene aroma compounds (e.g., nerol, linalool, and geraniol) in white tea, resulting in a more pronounced flowery aroma profile (Xu et al., 2025). In recent years, multiple fungi, including Saccharomyces cerevisiae, Penicillium griseofulvum, and Aspergillus oryzae, have been reported to produce a variety of terpenoid metabolites via the MVA pathway (Li, Yao, & Zeng, 2024; Liang, Zhi, Fang, & Zhang, 2021; Ran et al., 2026). Nevertheless, the regulatory relationships among components of the A. cristatus MVA pathway, especially the expression of terpene synthase and enzymes that transform terpenes, remain to be further elucidated.

4. Conclusion

This study systematically elucidated the differences in aroma characteristics and key odor-active compounds in loose-leaf dark teas (LDTs) fermented by three Aspergillus cristatus strains found in commercial FBTs from different regions, namely Shaanxi, Hunan, and Sichuan provinces in China. The aroma profiles of LDTs fermented by different strains were clearly differentiated. Tea fermented by the Shaanxi isolate exhibited prominent flowery, sweet, woody, and minty attributes, from aromas provided mainly by linalool, methyl salicylate, nonanal, geranylacetone, 1-octen-3-ol, 6-methyl-5-hepten-2-one, β-ionone, and coumarin. The Sichuan isolate produced a tea with a pronounced fruity aroma, defined by nine key odorants: linalool, 1-octen-3-ol, 6-methyl-5-hepten-2-one, methyl salicylate, coumarin, α-terpineol, nonanal, β-ionone, and benzyl alcohol. Similarly, nine key aroma compounds were identified in the tea fermented by the Hunan isolate, namely β-ionone, nonanal, linalool, 6-methyl-5-hepten-2-one, α-terpineol, 1-octen-3-ol, coumarin, methyl salicylate, and phenethyl alcohol. Although several aroma-active compounds, such as linalool, methyl salicylate, 1-octen-3-ol and 6-methyl-5-hepten-2-one, were shared across all samples, their relative contributions differed markedly among fermented loose-leaf dark teas, leading to distinct sensory outcomes. These findings confirm that the geographical origin of A. cristatus strains significantly influences the aroma profile of fermented LDTs, primarily through variations in the composition and concentration of key odor-active compounds. This study provides new insights into the region-specific effects of microbial fermentation on tea aroma and offers a theoretical foundation for the targeted improvement and quality control of fermented tea products.

CRediT authorship contribution statement

Haoran Sun: Writing – original draft, Software, Methodology, Formal analysis, Data curation, Conceptualization. Zhongqi Zu: Methodology, Formal analysis, Data curation, Conceptualization. Jia Li: Visualization, Software, Formal analysis, Data curation, Conceptualization. Yaling Cao: Methodology, Formal analysis, Data curation. Rui Fang: Software, Methodology, Data curation. Shanshan Shen: Formal analysis, Data curation. Mengxue Chen: Validation, Software, Data curation. Yu Tang: Visualization, Supervision, Investigation. Zhengnan Jiang: Visualization, Supervision. Qi Chen: Visualization, Supervision, Investigation. Yu Wang: Supervision, Software, Investigation. Xueling Gao: Writing – review & editing, Validation, Resources, Project administration, Funding acquisition.

Ethical statement

The sensory evaluation of loose-leaf dark tea in this study has been approved by the Ethics Committee of Anhui Agricultural University (Approval No.: KJLL2025016). Participants gave informed consent via the statement “I am aware that my responses are confidential, and I agree to participate in this sensory evaluation” where an affirmative reply was required to enter the sensory evaluation. They were able to withdraw from the sensory evaluation at any time without giving a reason.

Funding

This work was supported by the National Key Research and Development Program of China (2022YFD2101103), the Science and Technology Innovation Project of Anhui Province (202423a11020009), the Special Fund for Anhui Agriculture Research System (AHCYJSTX-NCPJG), and the Anhui Province Innovation and Entrepreneurship Training Program for College Students (S202410364179).

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.