Mechanistic insights into aroma differences of Taiping Kuihong black tea processed by different withering methods based on sensory directed analysis, metabolomics, and key enzyme activity assays

Abstract

Withering is a critical process influencing the aroma formation of Taiping Kuihong Black Tea (TKBT). In this study, sensory directed analysis, targeted metabolomics, and key enzyme activity assays were employed to systematically compare the effects of hot-air withering (HW), mixed-mode withering (MW), and constant temperature and humidity withering (CTHW) on the aroma characteristics and their underlying formation mechanisms in TKBT. The results showed that HW exhibited stronger floral, fruity, sweet, and potato-like aroma, whereas CTHW was dominated by green aroma. Combined with gas chromatography–olfactometry, odor active value, aroma recombination, and aroma omission experiments, 9 key aroma-active compounds contributing to the characteristic aroma profile of the samples were further identified. Variations in these metabolites were mainly driven by heat-regulated enzyme activities and aroma precursors, leading to the redistribution of volatile compounds. This study provides a theoretical basis and technical support for improving the aroma quality of TKBT.

Graphical abstract

Highlights

• •Different withering methods significantly affect the aroma quality of TKBT
• •HW has bold floral, fruity, sweet, and potato-like aromas
• •SDA identified nine key aroma-active compounds in TKBT
• •HW temperature boosts enzyme activity and aroma precursor conversion

1. Introduction

Taiping Kuihong Black Tea (TKBT), produced in Huangshan City of Anhui Province, is refined using traditional black tea processing techniques (Yang et al., 2024). In recent years, driven by the growing market demand for high-quality black tea, the TKBT industry has shown a steadily increasing development trend. Statistics indicate that by 2025, its production had reached 246 tons, with an output value of USD 15.64 million, demonstrating considerable economic benefits. The core advantage behind this growth lies in the use of the ‘Shidacha’ cultivar. As a superior germplasm resource, ‘Shidacha’ is characterized by broad leaves, thick leaf texture, and an accompanying orchid-like aroma, and is widely used in the production of premium spring green tea (Jin et al., 2024; Zhou et al., 2022). However, during the summer and autumn seasons, large quantities of fresh ‘Shidacha’ leaves remain underutilized, resulting in waste and idle resources. Therefore, exploring efficient approaches to utilize ‘Shidacha’ leaves harvested in summer and autumn is not only essential for improving resource utilization but also crucial for ensuring the increasing production and stable quality of TKBT.

In this context, optimizing the withering stage of tea processing is considered a key breakthrough for achieving efficient utilization of tea resources (Li et al., 2025). Withering, as a key step in black tea manufacturing, can regulate the physiological changes of fresh leaves, activate multiple endogenous enzymes, and promote the transformation of aroma precursors, thereby laying the foundation for the formation of the characteristic flavor of black tea (Huang et al., 2022; Li et al., 2025). Previous studies have demonstrated that different withering methods exert significant effects on the aroma quality of Keemun black tea and Fujian white tea. Among these, aroma-active compounds such as (Z)-4-heptenal, linalool, geraniol, and (E)-β-ionone have been commonly identified as the key contributors to aroma differences under various withering conditions (Huang et al., 2022; Wu et al., 2022). These findings indicate that rational regulation of withering conditions plays a crucial role in shaping the aroma profile of tea. Therefore, systematic investigation and optimization of the withering process for TKBT would not only contribute to improving its aroma quality and market competitiveness, but also provide a feasible theoretical basis and technical pathway for the high-value utilization of summer and autumn fresh leaves of ‘Shidacha’.

Aroma is one of the core indicators for evaluating the quality of black tea, and the establishment of scientific and effective compound extraction and analytical methods is fundamental to the systematic investigation of aroma characteristics (Yang et al., 2025; Yin et al., 2023). Due to the complex composition of tea infusion systems and the pronounced matrix effects, a single extraction approach is often insufficient to comprehensively characterize volatile compounds (Yang et al., 2025; Yu et al., 2023). Although headspace solid-phase microextraction is widely used, its limited extraction phase volume makes it susceptible to competitive adsorption and matrix effects. In contrast, stir bar sorptive extraction (SBSE), with its larger sorptive phase volume, markedly enhances the enrichment efficiency and detection sensitivity of trace aroma compounds. Dynamic headspace (DHS) extraction, through continuous gas purging under non-equilibrium conditions, effectively captures highly volatile components and reduces quantitative bias caused by competitive adsorption. The combination of SBSE and DHS enables comprehensive extraction of volatiles with different physicochemical properties, providing a reliable technical approach for the systematic characterization of TKBT aroma. (Zhai, Zhang, Granvogl, Ho, & Wan, 2022; Zhang et al., 2026). Subsequently, one−/two-dimensional gas chromatography–olfactometry (GC–O) techniques are commonly applied to identify and compare sensorially active compounds in samples, in combination with chemometric analysis and sensory validation experiments, thereby enabling the quantitative evaluation and screening of key aroma compounds and their olfactory contributions (Xia et al., 2023; Zhang et al., 2026). By elucidating the relative impact of individual compounds on the overall aroma profile, this strategy provides a scientific basis for the selection of desirable aroma attributes and offers technical guidance for aroma recombination and optimization in food systems. In practice, this analytical strategy has demonstrated high accuracy and reproducibility in complex aroma systems such as tea, fruits, and baijiu, providing a reliable approach for effectively elucidating key aroma-active compounds in foods (Chen et al., 2022a; Xia et al., 2023; Zhang et al., 2026).

Although ‘Shidacha’ has been widely applied in green tea manufacturing, its processing adaptability and quality formation mechanisms when used as a raw material for black tea remain insufficiently investigated. Meanwhile, the fresh leaves of ‘Shidacha’ are characterized by long and thick stems and a high moisture content (approximately 82%), which makes natural withering difficult and necessitates the exploration of more suitable withering strategies. In this context, the present study employed ‘Shidacha’ as the raw material to systematically evaluate its feasibility for black tea production, with particular emphasis on the utilization efficiency of summer-harvested leaves and their impact on aroma quality. Specifically, sensory directed analysis (SDA), targeted metabolomics, and key enzyme activity assays were integrated to comprehensively compare the effects of three withering processes, namely hot-air withering (HW), mixed-mode withering (MW), and constant temperature and humidity withering (CTHW), on aroma formation in TKBT. The findings of this study are expected to provide a theoretical basis for the efficient utilization of ‘Shidacha’ and offer technical support and process optimization strategies for the high-quality development of TKBT.

2. Materials and methods

2.1. Chemicals and materials

All water used in the research was obtained from the distilled water of Wahaha Group (Hangzhou, China). Methionine and phenylalanine were purchased from Wako Pure Chemical Industries, Ltd. (Kanagawa, Japan). Hexanal, (Z)-4-heptenal, acetic acid, phenylmethyl ester, β-carotene, and neoxanthin were obtained from Shanghai Aladdin Biochemical Technology Co., Ltd. (Shanghai, China). (E)-β-damascenone was purchased from Shanghai Haohong Biomedical Technology Co., Ltd. (Shanghai, China). Geraniol, linalool, (Z)-3-hexen-1-ol, 1-octen-3-ol, linalool oxide (furanoid, mixture of isomers), linalool oxide (pyranoid, mixture of isomers), 2-phenylethanol, (Z)-jasmone, phenylacetaldehyde, methional, (E,E)-2,4-heptadienal, 6-methyl-5-hepten-2-one, limonene, (E)-2-hexenal, methyl salicylate, (E)-β-ionone, (E,E)-2,4-nonadienal, heptanal, (Z)-3-hexenyl hexanoate, neral, (E,Z)-2,6-nonadienal, benzaldehyde, β-myrcene, nerol, indole, dihydroactinidiolide, γ-nonalactone, (E,E)-2,4-decadienal, δ-octalactone, coumarin, jasmine lactone, ethyl decanoate, methyl linoleate, and methyl α-linolenate were purchased from Shanghai Macklin Biochemical Co., Ltd. (Shanghai, China). All reference standards had a purity of over 95%, except for (E,E)-2,4-heptadienal and (E,E)-2,4-nonadienal, which had a purity of 90%. Other reagents and materials are listed in Table S1.

2.2. Sample collection and preparation

Fresh leaves of TKBT, consisting of one bud with two to three leaves, were harvested on May 10, 2025, in Huangshan District, Huangshan City, Anhui Province, China. The fresh leaves were evenly divided into three groups and subjected to different withering treatments.

For hot-air withering (HW), fresh leaves were withered in a withering trough for 6 h at an air temperature of 35 °C, with turning every 2 h. During this process, the ambient indoor temperature ranged from 20.8 to 21.9 °C, and the relative humidity was maintained between 63% and 67%.

For mixed-mode withering (MW), fresh leaves were first spread on ventilated bamboo sieves under indoor conditions for 10 h and subsequently transferred to a withering trough for an additional 3 h at an air temperature of 35 °C. The indoor temperature and relative humidity during this process were maintained at 20.8–21.9 °C and 63%–67%, respectively.

For constant temperature and humidity withering (CTHW), fresh leaves were placed in a constant temperature and humidity chamber and withered for 10 h, with the temperature maintained at 26.2–26.8 °C and the relative humidity controlled at 41%–44%.

After withering, the moisture content of all three groups was adjusted to approximately 64% (±1%). The withered leaves were then subjected to identical rolling, fermentation, primary drying, and secondary drying conditions to produce the final tea samples (Fig. 1). All finished tea samples were stored at 4 °C until analysis. Samples collected during processing were taken every 2 h and immediately stored at −80 °C for subsequent analysis.

Fig. 1.

Flowchart of tea processing. Black teas processed by hot-air withering, mixed-mode withering, and constant temperature and humidity withering are labeled HW, MW, and CTHW, respectively.

2.3. Preparation of tea infusion, ethyl decanoate solution, and non-volatile matrix

Three grams of tea leaves were taken from each of HW, MW, and CTHW and placed in separate 250 mL Erlenmeyer flasks. Each flask was filled with 150 mL of boiling water, steeped for 5 min, and then the infusion was filtered into a clean Erlenmeyer flask. To minimize the loss of aroma compounds through volatilization, the filtrate was immediately cooled to room temperature in an ice bath. All samples were analyzed in triplicate.

A 15 μL aliquot of ethyl decanoate standard was added to a 10 mL volumetric flask and weighed, then brought to volume to 10 mL with anhydrous ethanol to prepare the stock solution. The stock solution was subsequently diluted with ultrapure water to prepare working solutions with final concentrations of 10 μg/mL and 6 μg/mL.

The SAFE technique efficiently separates volatile compounds from the non-volatile residue in tea infusions (Schlumpberger, Stübner, & Steinhaus, 2022). In this study, non-volatile residues were obtained from HW, MW, and CTHW using SAFE, collected, and dissolved in water. The resulting aqueous solutions for HW, MW, and CTHW were then repeatedly extracted by SAFE until the residues were essentially odorless, as confirmed by sensory panelists (Zhai et al., 2023). The odorless non-volatile residues were used as the matrix (tea-to-water ratio, 1:50) for subsequent sensory experiments.

2.4. Volatile extraction techniques

2.4.1. Dynamic headspace (DHS)

In a 20 mL headspace vial, 2 g of NaCl, 6 mL of tea infusion, and 6 μL of working solution (6 μg/mL) were added sequentially, and then placed in the autosampler tray for online analysis. The extraction and enrichment of volatile compounds were performed using the DHS dynamic headspace system (Gerstel, Germany). The samples were incubated at 40 °C with shaking at 500 rpm for 30 min. Then, the headspace components were purged and trapped onto Tenax® TA sorbent tubes (Gerstel, Germany) using a high-purity nitrogen flow at 50 mL/min (total 500 mL) at 35 °C. To remove moisture, the trapping tubes were dried at 35 °C with a flow rate of 100 mL/min for 25 min (total 2500 mL). The enriched components were transferred using the TDU2 thermal desorption unit (Gerstel, Germany), and in splitless mode, the desorption temperature rapidly increased from 40 °C to 260 °C (240 °C/s) and held for 5 min. The desorbed components were focused in the CIS4 cooled injection system (Gerstel, Germany), where liquid nitrogen was used to cool the temperature to −50 °C. After desorption, the temperature was raised at a rate of 10 °C/s to 250 °C and held for 3 min to ensure rapid introduction of volatile substances into the chromatographic column. All samples were analyzed in triplicate.

2.4.2. Stir bar sorptive extraction (SBSE)

Add NaCl (3 g), tea infusion (10 mL), a pre-cleaned stir bar, and working solution (6 μL; 10 μg/mL) sequentially to a 20 mL headspace vial. After sealing, the vial was placed on a magnetic stirrer and incubated at 40 °C with stirring at 1200 rpm for 60 min for extraction. After extraction, the stir bar was rinsed with ultrapure water and gently blotted dry with filter paper to avoid pigment residues and minimize potential damage to the chromatographic column (Zhang et al., 2026). The operating parameters of the TDU2 and CIS4 for volatile compound analysis were consistent with those described in Section 2.4.1. All samples were analyzed in triplicate.

2.5. Analysis of volatile compounds by gas chromatography–mass spectrometry (GC–MS)

Qualitative analysis of volatile compounds was performed using a GC–MS system (GC: 8890 B; MS: 5977 B) equipped with an HP-5MS capillary column (30 m × 0.25 mm × 0.25 μm). High-purity helium (>99.99%) was used as the carrier gas at a constant flow rate of 1.6 mL/min. Mass spectrometric detection was carried out using electron ionization (EI) at 70 eV in full-scan mode over an m/z range of 30–350. The ion source and quadrupole temperatures were set to 230 °C and 150 °C, respectively.

For SBSE analysis, the oven temperature was initially set at 50 °C and held for 10 min, then increased to 100 °C at a rate of 6 °C/min, followed by a ramp to 180 °C at 3 °C/min, and finally raised to 260 °C at 20 °C/min and held for 5 min. The solvent delay was set to 8 min.

For DHS analysis, the oven temperature was initially set at 50 °C and held for 7 min, then increased to 90 °C at a rate of 10 °C/min, followed by a ramp to 180 °C at 3 °C/min, and finally raised to 280 °C at 15 °C/min and held for 5 min. The solvent delay was set to 5 min.

2.6. Gas chromatography–olfactometry (GC–O)

After volatile compounds were extracted using DHS and SBSE, samples were introduced into the GC system with the injector operated in solvent vent mode. The column effluent was evenly split at a ratio of 1:1 between the mass spectrometer and the ODP4 olfactory detection port. The transfer line temperature throughout the GC–O analysis was maintained at 280 °C to preserve the activity of volatile compounds. The olfactory detection port was supplied with nitrogen and water vapor at a flow rate of 50 mL/min to enhance sniffing performance and prevent nasal dryness (Zhao et al., 2025). During SBSE-GC-O and DHS-GC-O analysis, the GC oven temperature program and MS conditions were the same as those described in Section 2.5.

The Gerstel ODP recording software was used for data acquisition. When aroma-active compounds were perceived and identified by panelists, intensity ratings were immediately recorded by pressing a remote control button using a four-point scale (1, 2, 3, 4), accompanied by verbal descriptions of the odor characteristics. Aroma intensity, duration, and panelists' verbal responses were synchronously recorded and terminated upon release of the button (Gu, Jin, Schwarz, Rao, & Chen, 2022). The importance of aroma-active compounds was evaluated based on modified frequency (MF) values, with the general belief that a volatile compound with an MF > 0.5 contributes a significant aroma stimulus (Gu et al., 2022; Jiang et al., 2025). Notably, the MF metric was used to account for both perceived odor intensity and the consistency of its occurrence. This approach more effectively reduces bias arising from individual differences in olfactory sensitivity and helps identify characteristic components that show a high level of consensus among the panelists. Each sample was subjected to GC–O evaluation five times.

2.7. Qualitative and quantitative analysis of volatile compounds

This study employed four methods to ensure the accuracy of volatile compound identification: mass spectral library (MS) search, retention index (RI) comparison, odor characteristics (O) determination, and comparison with aroma standards (Std). The specific procedure is as follows: First, mass spectral information was extracted using AMDIS (NIST, USA) and compared with the NIST 20 database (with a match threshold of ≥80%). Next, the retention times of n-alkanes (C₇–C₄₀) were determined under the same chromatographic conditions, the RI of each compound was calculated, and compared with data from the NIST (https://webbook.nist.gov/chemistry/) website (with a difference controlled within 30). Finally, GC-O analysis combined with Std was used for final confirmation. The target compound's retention time (deviation ≤0.1 min), characteristic mass spectral fragments (match ≥80%), and sensory odor attributes were all matched with those of the standard compound under the same chromatographic conditions (Han et al., 2025; Liu et al., 2022; Zhang et al., 2026).

In this study, the absolute quantification of aroma compounds was achieved using the internal standard calibration curve method. Ethyl decanoate was selected as the internal standard, with the characteristic fragment ion m/z 88, and data collection was performed using selected ion monitoring (SIM) mode (Zhang et al., 2026). The response factors of each target aroma compound are detailed in Table 1. The process for constructing the standard curve is as follows: Aroma compounds and the internal standard (with concentrations ranging from 10 to 50 μg/L) were dissolved in pure water at five concentration gradient ratios of 5:1, 3:1, 1:1, 1:3, and 1:5. To eliminate the impact of sample preparation on quantification accuracy, the extraction and measurement procedures for the standard solutions were consistent with those described in 2.4, 2.5. Finally, the absolute quantification of each aroma-active compound was achieved by establishing a linear regression equation between the peak area response ratios of target compounds to the internal standard and their corresponding concentration ratios (Jiang et al., 2025).

Table 1.

The standard curve, modified frequency, and odor attributes of aroma compounds detected at the GC–O sniffing port.

No.	CAS	Compounds	Standard curves a	R2	Selected lions (m/z) b	MF c			Odor quality d	Identification basic e
						HW	MW	CTHW
1	928-96-1	(Z)-3-Hexen-1-ol	y = 0.4533x + 0.042	0.9984	67	0.74	0.67	0.84	Green	MS,RI,O,Std
2	3391-86-4	1-Octen-3-ol	y = 1.2472x + 0.1592	0.9952	57	0.40	0.40	0.40	Mushroom-like	MS,RI,O,Std
3	5989-33-3	(Z)-Linalool oxide (furanoid)	y = 0.51x + 0.1511	0.9945	59	0.53	0.40	0.55	Woody, floral	MS,RI,O,Std
4	34995-77-2	(E)-Linalool oxide (furanoid)	y = 0.4599x + 0.1252	0.993	59	0.87	0.81	0.92	Woody, floral	MS,RI,O,Std
5	78-70-6	Linalool	y = 0.4366x + 0.0552	0.9963	71	0.95	0.77	0.87	Citrus-like, floral	MS,RI,O,Std
6	60-12-8	2-Phenylethanol	y = 0.7116x + 0.1115	0.9961	91	0.71	0.50	0.59	Floral, honey-like	MS,RI,O,Std
7	14009-71-3	(Z)-linalool Oxide (pyranoid)	y = 0.173x + 0.0292	0.9951	68	0.71	0.59	0.74	Woody, floral	MS,RI,O,Std
8	106-25-2	Nerol	y = 0.2843x + 0.0258	0.9957	69	0.59	0.40	0.40	floral	MS,RI,O,Std
9	106-24-1	Geraniol	y = 0.2331x + 0.0191	0.9948	69	1.00	0.87	0.95	Floral, rose-like	MS,RI,O,Std
10	66-25-1	Hexanal	y = 0.3295x + 0.0832	0.993	44	0.77	0.71	0.87	Green, grassy	MS,RI,O,Std
11	6728-26-3	(E)-2-Hexenal	y = 0.22x + 0.0763	0.994	69	0.30	0.30	0.30	Green, grassy	MS,RI,O,Std
12	111-71-7	Heptanal	y = 0.3754x + 0.0141	0.9958	70	0.30	0.30	0.40	Fatty	MS,RI,O,Std
13	6728-31-0	(Z)-4-Heptenal	y = 0.2334x + 0.007	0.9933	68	0.30	0.30	0.30	Fishy	MS,RI,O,Std
14	100-52-7	Benzaldehyde	y = 0.6792x + 0.0132	0.9967	106	0.45	0.40	0.40	Bitter almond-like	MS,RI,O,Std
15	4313-03-5	(E,E)-2,4-Heptadienal	y = 0.7421x + 0.0643	0.9999	81	0.74	0.63	0.81	Fatty, green	MS,RI,O,Std
16	122-78-1	Phenylacetaldehyde	y = 0.4749x + 0.0607	0.9988	91	0.95	0.92	0.87	Honey-like	MS,RI,O,Std
17	557-48-2	(E,Z)-2,6-Nonadienal	y = 0.4311x + 0.0153	0.9943	70	0.71	0.55	0.63	Fruity	MS,RI,O,Std
18	5910-87-2	(E,E)-2,4-Nonadienal	y = 0.3044x + 0.0056	0.9975	81	0.39	0	0	Fatty, green	MS,RI,O,Std
19	106-26-3	Neral	y = 0.3079x + 0.0087	0.9947	69	0.35	0.30	0.30	Citrus-like,	MS,RI,O,Std
20	25152-84-5	(E,E)-2,4-Decadienal	y = 2.3886× - 0.0738	0.9992	81	0.24	0.20	0.20	Fatty, deep-fried-like	MS,RI,O,Std
21	123-35-3	β-Myrcene	y = 0.8391x + 0.1591	0.9932	93	0.71	0.50	0.50	Geranium-like	MS,RI,O,Std
22	138-86-3	Limonene	y = 0.1268x + 0.0549	0.9966	68	0.55	0.55	0.63	Citrus-like	MS,RI,O,Std
23	140-11-4	Acetic acid, phenylmethyl ester	y = 0.6975x + 0.0133	0.9973	108	0.35	0.20	0.30	Sweet, fruity	MS,RI,O,Std
24	119-36-8	Methyl salicylate	y = 0.8867x + 0.2255	0.9986	120	0.89	0.67	0.71	Mint-like, terpene-like	MS,RI,O,Std
25	698-76-0	δ-Octalactone	y = 0.2257x + 0.0092	0.995	99	0.84	0.74	0.67	Coconut-like	MS,RI,O,Std
26	104-61-0	γ-Nonalactone	y = 4.4974x + 0.0197	0.9987	85	0.50	0.40	0.30	Coconut-like	MS,RI,O,Std
27	31501-11-8	(Z)-3-Hexenyl hexanoate	y = 0.5087x + 0.0264	0.9937	82	0.24	0.20	0.24	Sweet, fruit	MS,RI,O,Std
28	25524-95-2	Jasmine lactone	y = 0.3674x − 0.0036	0.9997	99	0.81	0.71	0.63	Floral	MS,RI,O,Std
29	15356-74-8	Dihydroactinidolide	y = 0.4809x − 0.0275	0.9999	111	0.45	0.45	0.40	Woody, fruity	MS,RI,O,Std
30	110-93-0	6-Methyl-5-hepten-2-one	y = 0.5583x + 0.175	0.9968	43	0.74	0.63	0.30	Nutty	MS,RI,O,Std
31	23726-93-4	(E)-β-Damascenone	y = 0.5215x + 0.0152	0.9995	69	0.92	0.63	0.81	Baked apple-like	MS,RI,O,Std
32	488-10-8	(Z)-Jasmone	y = 0.1298x + 0.0253	0.9945	164	0.49	0.20	0.30	Floral	MS,RI,O,Std
33	79-77-6	(E)-β-Ionone	y = 0.4843x + 0.0092	0.9952	177	0.84	0.77	0.81	Floral, violet-like	MS,RI,O,Std
34	3268-49-3	Methional	y = 0.0228x − 0.0023	0.9986	104	0.95	0.87	0.77	Cooked potato-like	MS,RI,O,Std
35	120-72-9	Indole	y = 0.0663x − 0.001	0.9992	117	0.40	0.40	0.35	Mothball-like	MS,RI,O,Std
36	91-64-5	Coumarin	y = 0.3223x − 0.0324	0.9992	146	0.40	0.30	0.40	Woodruff-like	MS,RI,O,Std

Note: ^a Calibration curves were plotted based on the area response ratios of target compounds to internal standards and their concentration ratios. ^b Selected ions used for quantification ^cMF (%) = [F (%) × AI (%)]^1/2, where AI (%) is the average aroma intensity divided by the maximum scale (“4”), and F (%) is the frequency of detection of the same odorant in the panel. Black teas processed by hot-air withering, mixed-mode withering, and constant temperature and humidity withering are labeled HW, MW, and CTHW, respectively. ^d Odor quality of each odorant at the sniffing port. ^e Methods of identification: MS, identified by mass spectra; RI, by retention indices; O, by olfactometry; Std, by comparison with reference standards.

2.8. Sensory analysis

Twenty-five milliliters of tea infusion were transferred into a 50 mL amber sniffing bottle, which was maintained in a thermostatic water bath at 40 °C to prevent a decrease in sniffing intensity caused by temperature reduction. During the quantitative descriptive analysis (QDA), the tea infusion was replaced twice to ensure assessment accuracy. Aroma attribute evaluation was conducted by a panel of 12 assessors (6 males and 6 females). All panelists were selected based on the criterion that they had completed at least three months of systematic sensory training. This training was based on food and aroma reference standards and was intended to enhance the trainees' ability to recognize and discriminate odor signals. In this study, commercially available fruits and vegetables were selected as food references, whereas commonly used aroma standards were employed as odor references. These standards were typically first dissolved in ethanol or water and then serially diluted with water to 100 times their common odor threshold concentrations (Zhang et al., 2023). Aroma descriptors were first generated and then screened based on their frequency of occurrence. Six attributes were finally selected, including floral, fruity, sweet, potato-like, green, and woody aromas. The tea infusions were re-evaluated and scored using a 10-point intensity scale divided into five levels: weak (0–2), slightly weak (2–4), moderate (4–6), slightly strong (6–8), and strong (8–10). The results were presented as bar charts.

Aroma recombination experiments were performed using the odorless matrix as the experimental base. Eleven potential key aroma compounds with MF > 0.5 and odor activity values (OAV) ≥ 1 were first dissolved in ethanol and then individually added to 25 mL of the matrix according to their measured concentrations in the tea infusion to prepare the recombined aroma solution. The recombined sample and the original tea infusion were equilibrated simultaneously in a 40 °C water bath for aroma profile comparison, followed by evaluation using the QDA procedure described above.

Aroma omission experiments were also conducted using the odorless matrix as the base. Sensory differences between the complete recombination model and models lacking a single compound were evaluated using triangle tests. The minimum number of correct responses (X) required for statistical significance in the triangle test was calculated based on the number of panelists (N) and the significance coefficient (Z) using a binomial distribution, according to the following equation: X = N/3 + Z(2 N/9)^1/2. The significance coefficient Z was set to 1.64, 2.33, and 3.09 for significance levels of P = 0.05, 0.01, and 0.001, respectively (Zhao et al., 2024).

All sensory evaluation procedures complied with relevant regulations and were approved by the Science and Technology Ethics Committee of Anhui Agricultural University (approval no. KJLL2025032). A total of 12 trained panelists (6 males and 6 females, aged 22–28 years) were fully informed of the study objectives, procedures, and potential risks, participated voluntarily, and could withdraw at any time without penalty. Written informed consent was obtained from all participants prior to the sensory evaluation, including consent to participate and consent for the use of their anonymized data for research and publication. The rights and privacy of the participants were protected throughout the research; no personal identifying information was collected or disclosed.

2.9. Confirmation of key aroma-active compounds

The odor active value (OAV) of selected aroma compounds is the ratio of their quantified concentration to the corresponding odor threshold (Zhang et al., 2024). It is generally believed that components with an OAV ≥ 1 contribute significantly to the overall aroma profile of tea infusion (Yang et al., 2024). In this study, potential key aroma-active compounds were first screened based on the criteria of p < 0.05, MF > 0.5, and OAV ≥ 1. On this basis, further identification of key aroma-active compounds was carried out using aroma recombination and aroma omission experiments.

2.10. Determination of aroma precursors and enzyme activities in withered fresh leaves

This study performed the derivatization of linoleic acid (LA) and α-linolenic acid (ALA) following the method described by Wu et al. (2024). A 0.1 g sample of tea powder was mixed with 0.5 mL of extraction solvent (n-hexane/isopropanol, 3/2, v/v) and 0.25 mL of anhydrous Na₂SO₄ solution (67 mg/mL). The mixture was centrifuged at 12,000 rpm for 15 min at 4 °C. The extraction was repeated, and the supernatants were combined and evaporated to dryness under a nitrogen stream. The resulting residue was reacted with 3 mL of methylation reagent (methanol/toluene/H₂SO₄, 88/10/2, v/v/v) at 80 °C for 1 h. After cooling to room temperature, the mixture was extracted twice with heptane and brought to a final volume of 2 mL. After dehydration with 0.4 g anhydrous Na₂SO₄ and filtration through a 0.22 μm membrane, the samples were analyzed by GC–MS. The analysis was conducted using an Agilent 7890 A/5975C system, equipped with a DB-FFAP capillary column (30 m × 0.25 mm × 0.25 μm). High-purity helium was used as the carrier gas (1 mL/min), with an injector temperature of 230 °C and a 1 μL injection in splitless mode. The oven temperature program was as follows: initial temperature 50 °C, ramped at 40 °C/min to 200 °C, held for 1 min, then ramped at 3 °C/min to 230 °C and held for 10 min. MS parameters were set as follows: EI source energy 70 eV, scan range m/z 30–350; ion source and quadrupole temperatures were 230 °C and 150 °C, respectively, with a solvent delay of 7.0 min. Quantification of the components was based on external calibration curves (Table S2). All experiments were performed in triplicate.

Following the method described by Zhang et al. (2026). Geraniol β-primeveroside (GerPrim) and linalool β-primeveroside (LinPrim) in tea samples were analyzed using an LC–Orbitrap–MS system (UltiMate 3000, Dionex, Sunnyvale, CA, USA; Q-Exactive Focus, Thermo Fisher Scientific, Waltham, MA, USA). In the experiment, 100 mg of tea powder was ultrasonically extracted with 1 mL of pre-chilled 90% methanol, followed by centrifugation, and the two extracts were combined. To achieve relative quantification and purify the samples, 200 μL of internal standard (4-chloro-DL-phenylalanine, 1 mg/mL) and 50 mg of PVPP were added to the extract. After standing for 60 min, the mixture was centrifuged, and the supernatant was evaporated to dryness under a nitrogen stream. The resulting residue was reconstituted in 1.0 mL of ultrapure water, filtered, and the filtrate was diluted 40-fold for analysis. Chromatographic separation was performed using an ACQUITY BEH C₁₈ column (100 × 2.1 mm, 1.7 μm) at 40 °C. The mobile phase consisted of 0.05% formic acid in water (A) and acetonitrile (B). The gradient elution program was set as follows: 0–4 min (10%–15% B), 4–7 min (15%–25% B), 7–12 min (25%–40% B), 12–15 min (40%–55% B), 15–20 min (55%–95% B), and then returned to the initial conditions for rebalancing. The mass spectrometry was conducted in ESI negative ion mode with a spray voltage of 3800 V. The first and second resolution were set to 70,000 and 17,500, respectively. Full scans were performed using stepped HCD energy at 30, 40, and 60 eV (m/z 120–700). All samples were analyzed in triplicate.

Phenylalanine (Phe) and methionine (Met) contents were determined using a high-speed amino acid analyzer (L-8900, Hitachi, Tokyo, Japan) (Li et al., 2023, Zhang et al., 2024). Tea powder (100 mg) was mixed with 4% sulfosalicylic acid (4 mL) in a 10 mL centrifuge tube, ultrasonicated for 30 min, and then allowed to stand for 10 min. An aliquot of the supernatant (1.5 mL) was transferred to a 2 mL centrifuge tube and centrifuged at 12000 rpm for 30 min. The resulting supernatant (1 mL) was filtered through a 0.22 μm Millipore membrane and injected for analysis. The high-speed amino acid analyzer conditions were as follows: the mobile phase was lithium citrate buffer; detection wavelengths were 570/440 nm; the flow rates for the mobile phase and derivatization reagent were 0.35 and 0.3 mL/min, respectively; the column, post-column reaction, and autosampler temperatures were 38 °C, 130 °C, and 4 °C, respectively; and the injection volume was 20 μL. Phenylalanine and methionine contents were calculated based on the ratio of the sample peak area to the corresponding standard peak area. All samples were analyzed in triplicate.

Following the method described by Chen et al. (2022b). β-Carotene and (β-Car) neoxanthin (Neo) in tea samples were quantified using an LC–MS/MS system (QTRAP 6500+, SCIEX, USA) equipped with an APCI source. A 50 mg sample of tea powder was mixed with 0.5 mL of an extraction solvent (n-hexane/acetone/ethanol, 1:1:1, v/v/v) containing 0.01% BHT. After vortex extraction for 20 min and centrifugation at 4 °C, the extraction was repeated, and the supernatants were combined. The extracts were evaporated to dryness under a nitrogen stream and reconstituted in 0.1 mL of methanol/methyl tert-butyl ether (1:1, v/v), then filtered before analysis. Chromatographic separation was achieved using a YMC C₃₀ column (100 mm × 2 mm, 3 μm) at 28 °C, with a 2 μL injection volume. The mobile phase A consisted of methanol/acetonitrile (1:3, v/v, with 0.1% formic acid and 0.01% BHT), and mobile phase B was MTBE (with 0.01% BHT). The gradient program was as follows: 0–3 min, A/B = 100/0; 3–5 min, linear from 100/0 to 30/70; 5–9 min, linear from 30/70 to 5/95; 9–10 min, returned to 100/0; and 10–11 min, held at 100/0. Mass spectrometry was performed in APCI positive ion mode, with a spray voltage of +4500 V and an ionization temperature of 350 °C. The curtain gas, nebulizer gas, and auxiliary gas pressures were set at 25, 65, and 70 psi, respectively. The concentrations of the components were calculated based on external standard calibration curves (Table S2). All experiments were performed in triplicate.

Pretreatment and analysis of fresh tea leaf samples were performed according to Zhang et al. (2026). Freeze-dried, ground samples were extracted with PBS buffer (pH 7.4) at a buffer-to-sample ratio of 9:1 (v/w), followed by centrifugation at 8000 rpm for 30 min at 4 °C. The supernatant was collected for subsequent assays. Enzyme activities were measured using commercial assay kits according to the manufacturers' instructions (Li et al., 2023). After the microplate equilibrated to room temperature for 60 min, standard, blank, and sample wells were set up by adding 50 μL of standards at different concentrations, 50 μL of sample extract, and 50 μL of sample diluent, respectively. Then, 100 μL of HRP-labeled detection antibody was added to each well, and the plate was sealed and incubated at 37 °C for 60 min. The solution was discarded and the plate was washed five times (350 μL wash buffer per well, standing for 1 min each time). Substrates A and B (50 μL each) were added and incubated at 37 °C in the dark for 15 min, followed by addition of 50 μL stop solution. Absorbance (OD) was measured at 450 nm within 15 min. Activities of β-primeverosidase (β-pase), carotenoid cleavage dioxygenase (CCD), alcohol dehydrogenase (ADH), lipoxygenase (LOX), and hydroperoxide lyase (HPL) were calculated using standard curves (Table S2). All samples were analyzed in triplicate.

2.11. Statistical analysis

One-way analysis of variance (ANOVA) and Pearson correlation analysis were performed using SPSS software (v 21.0). Data visualization was carried out using Origin 2021, SIMCA (v 14.1), RStudio (v 2024), and an online platform (https://www.chiplot.online/).

3. Result and discussion

3.1. Overall aroma profiles of TKBT under three withering methods

After QDA evaluation of TKBT subjected to three withering treatments, six sensory descriptors were identified to characterize the overall aroma profiles (Fig. 2). Statistical analysis indicated that green aroma and potato-like aroma differed significantly among the treatments at p < 0.05, while floral aroma, fruity aroma, and sweet aroma exhibited highly significant differences at p < 0.01. Specifically, CTHW showed the highest intensity of green aroma and was clearly distinguished from the other treatments. In contrast, HW exhibited the highest intensities of sweet aroma, potato-like aroma, floral aroma, and fruity aroma. These results demonstrate that different withering methods can markedly reshape the sensory profiles of TKBT, likely by modulating the transformation of aroma precursors as well as the formation and retention of volatile compounds, thereby leading to distinct aroma styles.

Fig. 2.

Aroma profile analysis of tea infusion from different TKBT. Black teas processed by hot-air withering, mixed-mode withering, and constant temperature and humidity withering are labeled HW, MW, and CTHW, respectively. “*”, Significant (p ≤ 0.05); “**”, highly significant (p ≤ 0.01). Different letters indicate differences.

3.2. Screening of aroma-active compounds by GC–O

To screen aroma-active compounds that can be directly perceived by humans from the complex volatile matrix and to evaluate their odor characteristics and relative intensities for the identification of key aroma contributors, GC–O analysis was performed. A total of 36 aroma-active compounds were identified (Table S3). In parallel, quantitative determination of these 36 compounds using authentic standards provided a reliable data basis for subsequent comparisons among samples and contribution assessment (Table 1, Table S3). Differences were observed among samples subjected to different withering treatments in terms of both intensity distribution and odor-type composition of aroma-active compounds. However, overall aroma profiles were predominantly characterized by sweet aroma, floral aroma, and fruity aroma.

The MF value was used to evaluate the relative importance of individual compounds in the aroma profiles (Bueno et al., 2011; Gu et al., 2022; Xu et al., 2025). Based on the criterion of MF > 0.5, a total of 21 aroma-active compounds were selected. These compounds exhibited pronounced characteristic odor properties and were considered to play key roles in the construction of the overall aroma profiles, thereby warranting further investigation. From a chemical perspective, alcohols constituted the largest group, with eight aroma-active compounds identified, indicating that alcohols may serve as important structural components in the aroma formation of TKBT. Taking HW as an example (the same applies below), the alcohols ranked by decreasing MF values were geraniol (MF = 1.00), linalool (MF = 0.95), (E)-linalool oxide (furanoid) (MF = 0.87), (Z)-3-hexen-1-ol (MF = 0.74), 2-phenylethanol (MF = 0.71), (Z)-linalool oxide (pyranoid) (MF = 0.71), nerol (MF = 0.59), and (Z)-linalool oxide (furanoid) (MF = 0.53). Combined with their odor characteristics, these alcohols and their oxides mainly contributed to floral aroma and sweet aroma, accompanied by moderate woody aroma and green aroma (Zhai et al., 2022).

Aldehydes were also identified as important odorants in black tea (Yang et al., 2025; Yin et al., 2023). In this study, hexanal (MF = 0.77), (E,E)-2,4-heptadienal (MF = 0.74), and (E,Z)-2,6-nonadienal (MF = 0.71) were classified as fatty acid–derived aldehydes, which mainly impart fresh green and cucumber-like nuances and constitute important chemical contributors to green aroma (Chen et al., 2022b; Yang et al., 2024). In contrast, phenylacetaldehyde (MF = 0.95), primarily derived from amino acid degradation, exhibits a honey-like odor and is widely recognized as a key contributor to sweet aroma (Ho, Zheng, & Li, 2015).

Comparable numbers of ester and ketone aroma-active compounds were screened. For instance, lactones such as δ-octalactone and jasmine lactone are generally associated with hydroxylated fatty acid precursors derived from lipid oxidation and subsequent lactonization (cyclization) processes, contributing creamy, sweet, or jasmine-like pleasant aromas(Yu et al., 2022; Zeng et al., 2018). Among ketones, (E)-β-damascenone (MF = 0.92), (E)-β-ionone (MF = 0.84), and 6-methyl-5-hepten-2-one (MF = 0.74) are typical carotenoid-derived cleavage products. Their formation is closely related to oxidative degradation of carotenoids and they are widely regarded as important sources of floral aroma and fruity aroma (Ho et al., 2015). Overall, ester- and ketone-type compounds exhibited relatively higher MF values in HW, suggesting that the continuous heat input during hot-air withering may accelerate lipid oxidation and carotenoid cleavage reactions, thereby promoting the formation and expression of related volatiles and enhancing sweet aroma, floral aroma, and fruity aroma.

In addition, two alkene compounds and one heterocyclic compound were screened. Among them, β-myrcene (MF = 0.71) and limonene (MF = 0.55) are typically biosynthesized from isopentenyl pyrophosphate and dimethylallyl pyrophosphate, proceeding via the geranyl pyrophosphate intermediate under the catalysis of monoterpene synthases (Dudareva, Klempien, Muhlemann, & Kaplan, 2013; Qiao et al., 2022). These monoterpene compounds commonly exhibit fresh citrus-like aromas and contribute positively to the floral and fruity characteristics of the samples (Wang et al., 2022; Wu et al., 2024). Methional (MF = 0.95), a heterocyclic compound generated through Strecker degradation of amino acids, exhibits a cooked potato-like odor. Its high expression in HW likely contributed to the enhancement of potato-like aroma (Ho et al., 2015).

3.3. Screening of potential key aroma-active compounds

GC–O analysis relies on human sniffing and is therefore susceptible to factors such as olfactory fatigue, individual variability, and column-related effects including high post-column temperatures and residual tailing. Consequently, sniffing intensity does not necessarily correspond to the actual contribution of a compound in the tea infusion (Qin et al., 2024). To improve the objectivity of compound screening, OAV was further introduced by integrating compound concentrations with their odor thresholds to quantify their potential aroma contributions (Wei et al., 2024; Zhang et al., 2023). Based on the combined criteria of MF > 0.5 and OAV ≥ 1, a total of 11 potential key aroma-active compounds were preliminarily identified, including (Z)-3-hexen-1-ol, (E)-linalool oxide (furanoid), linalool, geraniol, hexanal, (E,E)-2,4-heptadienal, phenylacetaldehyde, (E,Z)-2,6-nonadienal, (E)-β-damascenone, (E)-β-ionone, and methional (Table 2).

Table 2.

The odor thresholds and odor activity values of potential key aroma-active compounds in three withering samples.

Compounds	OTs (μg/L)	OAV
		HW	MW	CTHW
(E)-Linalool oxide (furanoid)	100	1.14	1.03	1.32
Linalool	0.58	71.24	45.95	62.66
Geraniol	1.1	312.91	208.58	248.49
(E)-β-Ionone	0.021	120.95	82.38	87.14
OAV of compounds with floral aroma		506.24	337.94	399.61
(E,Z)-2,6-Nonadienal	0.03	29.67	13.67	18.67
(E)-β-Damascenone	0.006	75.00	41.67	50.00
OAV of compounds with fruity aroma		104.67	55.34	68.67
(Z)-3-Hexen-1-ol	3.9	13.39	12.1	19.22
Hexanal	2.4	15.48	14.25	19.58
(E,E)-2,4-Heptadienal	0.032	60.63	45.63	74.06
OAV of compounds with green aroma		89.50	71.98	112.86
Phenylacetaldehyde	5.2	34.64	31.31	27.81
OAV of compounds with sweet aroma		34.64	31.31	27.81
Methional	0.43	14.79	10.30	8.53
OAV of compounds with potato aroma		14.79	10.30	8.53

Note: Odor thresholds (OTs) in water from: Leibniz-LSB@TUM Odorant Database (https://www.leibniz-lsb.de/en/databases) and (Zhai et al., 2022).

As shown in Table 2, the OAV distributions differed markedly among samples subjected to different withering treatments. For example, in HW, the OAVs of compounds associated with floral aroma and fruity aroma were 49.79% and 89.16% higher than those in MW, and 26.68% and 52.44% higher than those in CTHW, respectively. In contrast, the OAV of compounds related to green aroma in CTHW were 26.10% higher than those in HW and 56.80% higher than those in MW. Overall, the OAV-based results were consistent with the trends observed in QDA characterization, indicating good agreement between the quantitative evaluation of key aroma-active compounds and sensory perception.

3.4. Aroma recombination and omission experiments

Aroma recombination is a core approach for validating whether the combination of screened potential key aroma-active compounds can explain the overall aroma profile of a sample (Wei et al., 2024). In this study, the 11 screened potential key aroma-active compounds were added to an odorless non-volatile matrix according to their actual concentration ratios in the tea infusion to construct the recombination system. The recombined tea infusions were then subjected to QDA-based aroma profile characterization in comparison with the original tea infusions, and the results are shown in Fig. 3. Compared with the original tea infusions, the three recombined tea infusions generally exhibited more pronounced floral aroma, fruity aroma, and sweet aroma, whereas potato-like aroma and green aroma were slightly reduced. Nevertheless, significance analysis indicated no significant differences between the recombined and original tea infusions in the major aroma attributes, suggesting that the recombination system was able to effectively reproduce the core aroma characteristics of the original tea infusion.

Fig. 3.

Aroma recombination and aroma omission experiments of different TKBT samples. A: Radar chart of aroma recombination; B: Results of aroma omission experiments. Aroma omission experiments were conducted by 12 panelists, and X represents the number of panelists with correct answers. Black teas processed by hot-air withering, mixed-mode withering, and constant temperature and humidity withering are labeled HW, MW, and CTHW, respectively. “*, **, and ***” indicate significance at p ≤ 0.05, p ≤ 0.01, and p ≤ 0.001, respectively. Different letters indicate differences.

To further precisely identify the core aroma-active volatile compounds of TKBT, aroma omission experiments were subsequently performed based on the recombination system. In these experiments, individual compounds were removed one at a time from the recombined system, and the resulting sensory changes were compared to evaluate their contribution and necessity to specific aroma attributes. Ultimately, nine key aroma-active compounds were confirmed to constitute the core aroma components of TKBT (Fig. 3). Specifically, omission of linalool, geraniol, or (E)-β-ionone led to a consistent decrease in perceived floral aroma intensity. Removal of (Z)-3-hexen-1-ol, hexanal, or (E,E)-2,4-heptadienal resulted in a significant reduction in green aroma. In contrast, omission of phenylacetaldehyde, (E)-β-damascenone, or methional caused a noticeable decline in sweet aroma, fruity aroma, and potato-like aroma, indicating that these compounds serve as key driving factors for their corresponding aroma attributes.

3.5. Sources and differential analysis of key aroma-active compounds

To gain mechanistic insight into the differences observed in the nine key aroma-active compounds of TKBT produced under different withering methods, we shifted the analytical emphasis to biochemical events occurring during the fresh-leaf withering stage. Notably, we did not measure volatile aroma profiles of withering leaves at each time point; therefore, our discussion is not intended to establish a strict time-aligned causal link between enzyme activities and in-process aroma changes. Instead, because all post-withering processing steps were kept identical and the withering endpoint moisture content was strictly controlled, the differences in the final dry-tea aroma profiles are most plausibly explained by treatment-dependent variation in precursor conversion and associated enzymatic reactions accumulated during withering. Accordingly, we compared the levels of key aroma precursors at the end of withering and the temporal dynamics of key enzyme activities to infer how withering conditions may modulate aroma-formation pathways (Fig. 4, Table S4).

Fig. 4.

Analysis of the sources and differences of key aroma active compounds. Linalool β-primeveroside: LinPrim, Geraniol β-primeveroside: GerPrim, β-Primeverosidase: β-Pase, Linoleic acid: LA, α-Linolenic acid: ALA, Lipoxygenase: LOX, Hydroperoxide lyase: HPL, Alcohol dehydrogenase: ADH, β-Carotene: β-Car, Neoxanthin: Neo, Carotenoid cleavage dioxygenase: CCD. Methionine: Met, Phenylalanine: Phe.

Monoterpenes such as linalool and geraniol have attracted considerable attention because they impart pleasant floral notes to tea. In fresh leaves, these compounds predominantly occur in glycosidically bound forms, with β-primeverosides serving as the key precursors (Ho et al., 2015; Yang, Baldermann, & Watanabe, 2013). During withering, continuous moisture loss gradually loosens cellular structures and increases membrane permeability, weakening cellular compartmentalization. This, in turn, enhances the accessibility and reaction efficiency between hydrolytic enzymes (e.g., β-Pase) and their substrates, thereby promoting the hydrolysis of glycosidically bound precursors and the release of free monoterpene alcohols (Huang et al., 2022). Temperature is an important factor regulating this conversion. Compared with MW and CTHW, HW at 35 °C is more favorable for hydrolytic enzyme activity, which strengthens the glycoside-hydrolysis pathway and promotes the formation of floral aroma compounds (Wan, 2003). Our data support this interpretation: HW showed the highest β-Pase activity, along with significantly increased linalool and geraniol levels and concomitantly decreased levels of their direct precursors (LinPrim and GerPrim), indicating that HW accelerates the conversion and accumulation of free monoterpene alcohols by providing a favorable temperature.

(E,E)-2,4-Heptadienal, (Z)-3-hexen-1-ol, and hexanal are lipid-derived volatiles that typically impart characteristic green aroma (Ho et al., 2015; Zhang et al., 2023). Previous studies have shown that α-linolenic acid and linoleic acid serve as important precursors for these compounds and that their formation is regulated by enzymes such as LOX, HPL, and ADH (Chen et al., 2022b; Yang et al., 2024; Zhao et al., 2024). However, in this study, although the activities of LOX, HPL, and ADH were relatively higher under HW and the levels of fatty acid precursors such as LA and ALA were the lowest, the retained levels of these lipid-derived volatiles in the final product were not the highest. This suggests that their formation and volatilization losses likely occurred in parallel during withering, thereby reducing overall retention. In contrast, MW likely limited their formation during the early stage due to relatively low enzyme activities, and the short hot-air stage at the later phase was insufficient to compensate for this deficiency and may have caused additional losses, leading to the lowest contents. Under CTHW conditions, the milder temperature and weaker airflow likely reduced volatilization losses, favoring the retention of these low-molecular-weight volatiles and resulting in a more pronounced green aroma perception.

(E)-β-Ionone and (E)-β-damascenone are cleavage products derived from β-carotene and neoxanthin, respectively (Chen et al., 2025; Yang et al., 2024). In this study, CCD activity was highest under HW, and the declines in carotenoid precursors such as β-carotene and neoxanthin were the most pronounced. These results suggest that sustained heat input and relatively sufficient oxygen availability during HW enhanced carotenoid oxidative cleavage, thereby promoting the accumulation of (E)-β-ionone and (E)-β-damascenone and further intensifying the floral and fruity aroma expression of the samples (Ho et al., 2015).

Phenylacetaldehyde and methional are Strecker degradation products derived from phenylalanine and methionine, respectively, and typically impart sweet aroma and potato-like aroma (Ho et al., 2015). In the present study, the levels of these two compounds were higher under HW treatment, while the corresponding precursors, phenylalanine and methionine, were present at the lowest levels. This suggests that the elevated temperature provided by hot-air withering promoted Strecker degradation, thereby enhancing the formation and expression of these amino acid–derived aroma compounds.

4. Conclusion

This study systematically analyzed the effects of three processes, HW, MW, and CTHW, on the aroma formation of TKBT produced from summer and autumn ‘Shidacha’ raw materials, based on SDA, targeted metabolomics, and key enzyme activity assays. QDA results demonstrated that HW was characterized by more pronounced floral aroma, fruity aroma, sweet aroma, and potato-like aroma, whereas CTHW was dominated by a distinct green aroma, indicating that modulation of withering conditions can effectively reshape the aroma style of TKBT. Subsequently, GC–O analysis detected 36 aroma-active compounds, and integration of p < 0.05, MF > 0.5, OAV ≥ 1, along with aroma recombination and omission experiments, ultimately confirmed that linalool, geraniol, and (E)-β-ionone mainly contributed to floral aroma; (E)-β-damascenone was closely associated with fruity aroma; phenylacetaldehyde imparted sweet aroma; methional was responsible for potato-like aroma; and (Z)-3-hexen-1-ol, hexanal, and (E,E)-2,4-heptadienal were the primary contributors to green aroma. Further correlation analysis between aroma precursors and key enzyme activities revealed that the differential accumulation of these nine key aroma-active compounds fundamentally originated from variations in the activities of β-pase, CCD, LOX, HPL, and ADH, as well as differences in substrate metabolism. Notably, hot-air withering provided favorable temperature and dehydration conditions for these enzymatic reactions, promoting the hydrolysis of LinPrim and GerPrim, accelerating the cleavage of β-carotene and neoxanthin, and enhancing the transformation of precursors such as methionine, phenylalanine, linoleic acid, and α-linolenic acid. These processes collectively strengthened floral aroma, fruity aroma, sweet aroma, and potato-like aroma in HW. In contrast, the relatively mild temperature and humidity conditions of CTHW were more conducive to the retention of low-boiling green-related volatiles, resulting in a more prominent green aroma.

Overall, this study provides a theoretical basis and technological reference for producing high-aroma TKBT from summer and autumn ‘Shidacha’ raw materials, contributing to the high-value utilization of summer–autumn tea resources and enhanced industrial economic benefits. Moreover, the findings lay a foundation for further optimization of black tea aroma quality through the integration of additional processes, such as shaking, in future studies.

CRediT authorship contribution statement

Yuxuan Zhang: Writing – original draft, Software, Methodology, Formal analysis, Data curation, Conceptualization. Tianyuan Yang: Software, Methodology, Formal analysis, Data curation, Conceptualization. Jixin Zhang: Software, Formal analysis, Data curation. Tianzi Yu: Resources, Methodology, Investigation, Data curation. Luyao Yang: Resources, Methodology, Investigation, Data curation. Xuefei Peng: Visualization, Supervision, Investigation. Yanqun Jiang: Supervision, Investigation. Qianfeng Yang: Supervision, Investigation. Xiaoyuan Zhu: Project administration. Jingming Ning: Resources, Project administration, Funding acquisition. Shaode Hu: Resources, Project administration, Funding acquisition.

Ethics statement

The authors confirm that the appropriate protocols for protecting the rights and privacy of all participants were utilized during the execution of this research, including no coercion to participate, full disclosure of study requirements and risks, verbal consent of participants, no release of participant data without their knowledge, and the ability to withdraw from the study at any time.

Funding

This work was supported by the National Key Research and Development Program of China (2021YFD1601102) and supported by the earmarked fund for Agriculture Research System of China (CARS-19).

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.