Different drying methods reshape volatile aroma and bioactive non-volatile compounds in Camellia nitidissima flowers and infusions: UPLC-MS/MS and GC–MS insights

TL;DR

Different drying methods affect the aroma and health compounds in Camellia nitidissima flowers and tea infusions, with vacuum freeze-drying preserving more nutrients and sensory qualities.

Contribution

This study reveals that vacuum freeze-drying preserves more bioactive compounds and aroma in Camellia nitidissima flowers compared to hot-air drying.

Findings

Vacuum freeze-drying (VFD) preserves higher levels of flavonoids, polyphenols, and lipids compared to hot-air drying (HAD).

VFD flower infusions show greater diversity and higher concentrations of non-volatile metabolites like anthocyanins and isoflavones.

VFD flower infusions have higher relative odor activity values for 76 key odorants compared to other drying methods.

Abstract

Drying is a critical processing step in the production of Camellia nitidissima Chi (CN)-scented tea. We investigated the effects of hot-air drying (HAD) and vacuum freeze-drying (VFD) on non-volatile and volatile compounds in CN flowers and their tea infusions, utilizing mass spectrometry. Both methods reduced CN moisture to below 8% and significantly reduced volatile compound concentrations. VFD-treated flowers exhibited better morphology and higher levels of flavonoids, polyphenols, and lipids, thereby enhancing antioxidant capacity; while HAD promoted the upregulation of stress-related metabolites, such as phenolic acids. Moreover, VFD flower infusion showed greater diversity and significantly higher concentrations of non-volatile metabolites, especially lipids and flavonoids (anthocyanins, isoflavones). 78.3% of differential aroma compounds (VIP > 1) had higher relative odor activity than infusions from fresh and HAD flowers. These results may underscore VFD as the preferred method for retaining both the nutritional and sensory attributes of CN flowers and their tea infusions.

•Hot-air drying (HAD) alters more phenolic acid than Vacuum freeze-drying (VFD).•VFD flowers retain higher total flavonoids, phenolics, and antioxidant capacity than HAD.•Both methods significantly reduce volatile content, but VFD maintains the original profile.•VFD flower infusion has higher lipid, anthocyanin, and isoflavone content than HAD.•VFD flower infusion exhibits higher rOAV for 76 key odorants than other infusions.

Hot-air drying (HAD) alters more phenolic acid than Vacuum freeze-drying (VFD).

VFD flowers retain higher total flavonoids, phenolics, and antioxidant capacity than HAD.

Both methods significantly reduce volatile content, but VFD maintains the original profile.

VFD flower infusion has higher lipid, anthocyanin, and isoflavone content than HAD.

VFD flower infusion exhibits higher rOAV for 76 key odorants than other infusions.

Figures

Full text

Introduction

1

Edible flowers are important sources of bioactive compounds, with significant potential for human nutrition and health-promoting applications (Abbaszadeh et al., 2025; Wen et al., 2025). Historically integrated into diets across civilizations, they continue to attract scientific interest owing to their unique phytochemicals, including polyphenols, flavonoids, and terpenoids, which have demonstrated antioxidant, anti-inflammatory, and metabolic-regulatory effects (Rivas-García et al., 2021; Teixeira et al., 2023). In China, many edible flowers are processed into scented teas, which are distinctive tea substitutes that diversify food supply ((Li et al., 2019) Established varieties such as Jasminum sambac, Chrysanthemum morifolium, and Rosa rugose exemplify the successful utilization of floral resources in regional functional products (Gao et al., 2025; Hegde et al., 2022; Wang et al., 2024;Wang et al., 2024).

Camellia nitidissima Chi (CN), a Camellia species with rare golden petals known as the “Queen of Camellia,” has emerged as a candidate for scented tea production (Jiang et al., 2025). Recognized as both a medicinal and a food plant, CN is valued for its use in treating laryngitis, hypertension, and inflammation (Song et al., 2011). Modern studies have shown that CN flowers and their extracts are rich in health-promoting compounds, including flavonoids, polyphenols, and saponins (Lin et al., 2013; Tsoi et al., 2022). Notably, fresh CN flowers emit a pleasant fragrance and develop a harmonious floral tea aroma upon brewing, a key attribute for their development as a novel scented tea ingredient. Despite growing interest in CN flowers, current research predominantly focuses on CN leaves or essential oils, leaving critical gaps in understanding the factors that affect key quality parameters of CN flowers (e.g., flavor and bioactive compounds) (Yang et al., 2022).

Drying is an essential and common step in scented tea production, ensuring sufficient shelf life and preserving the desired qualities (Zhao et al., 2019). However, it inevitably alters flower aroma and the levels of bioactive compounds, directly impacting final product quality (Marchioni et al., 2021; Yang et al., 2024). Hot-air drying (HAD), the traditional and most common method, often employs high temperatures that can degrade heat-sensitive nutrients and alter color, although it may generate new flavors via reactions like Maillard browning (Xu et al., 2020). To mitigate these limitations, vacuum freeze-drying (VFD) offers an alternative. VFD minimizes oxidative damage through low-temperature processing under vacuum, making it particularly suited to preserving thermolabile, nutrient-dense materials such as flowers (Han et al., 2024). Each method has advantages and disadvantages, and no single method is ideal for drying edible flowers. The choice of method is influenced by available resources, expected consumer quality, and the final product's selling price (Zhao et al., 2019). However, the effects of two common CN drying methods (HAD and VFD) on the nutritional composition, flavor profile, and infusion quality of CN-scented tea remain unexplored. This knowledge gap hinders the optimization of CN processing and informed decision-making for producers.

Therefore, this study systematically investigates the effects of HAD and VFD on the nutritional value and flavor characteristics of CN flowers and their infusions. Using metabolomic techniques such as ultra-performance liquid chromatography tandem mass spectrometry (UPLC-MS/MS) and headspace solid-phase microextraction/gas chromatography–mass spectrometry (HS-SPME-GC–MS), we aimed to establish a comprehensive evaluation system. Our findings provide essential theoretical insights for refining CN scented tea processing technology, ultimately supporting the industrialization of this valuable botanical resource and maximizing its health-promoting potential.

Materials and methods

2

Sample

2.1

Camellia nitidissima Chi (CN) flowers were harvested from Nanning, Guangxi Zhuang Autonomous Region, China. Fresh flowers (FCN) were immediately transported to the laboratory on ice and stored at 4 °C to preserve metabolic integrity before processing. In addition to the samples used for testing, the remaining flowers were divided equally into six portions (n = 6): three for hot-air drying (n = 3) and three for vacuum freeze drying (n = 3). It should be noted that the drying process was carried out separately for each portion of the flowers. Specific drying methods are as follows:

Hot air drying (HAD): Fresh CN flower (FCN) samples were dried in a forced air circulation drying oven (CS101-2 EB, SDEI, China) at 40 °C for 4 h. The dried sample was labeled DCN.

Vacuum freeze drying (VFD): Fresh CN flowers were pre-frozen at −20 °C for 30 h. Under the conditions of 0.04 mbar and − 60 °C, and then vacuum freeze-dried (LGJ-18D, Sihuanqihang, China) for 24 h. This sample was referred to as FDCN.

Moisture determination

2.2

Triplicate samples (2.00 ± 0.01 g) were weighed (BSA124S, Sartorius, Germany) in pre-dried crucibles, placed in the oven at 103 ± 2 °C with lids ajar for 4 h, then covered and transferred to desiccators for 30 min before reweighing. This cycle was repeated at 1-h intervals until constant mass was achieved (defined as ≤0.005 g difference between consecutive weighings). Three biological replicates were established for each of the three CN flower types. Final moisture (wet basis) was calculated as:

$$[eqn]$$

Preparation of CN tea infusion of CN flower samples from various drying methods

2.3

To simulate the steps of “warming and moistening” or “washing” in daily tea drinking habits, following the procedure described by (Wu et al., 2023), FCN and CN-scented tea (DCN and FDCN) infusions were prepared using the green tea brewing method. Each sample (1 g) of FCN, DCN, and FDCN was rinsed with 50 mL of boiling water for 1 min (Methodology for sensory evaluation of tea). The water was decanted, and 50 mL of boiling water was added to extract the metabolites from the samples. After 6 min of extraction, all tea infusions were collected, filtered through filter paper, and cooled to room temperature to obtain CN tea infusions using different drying methods. Three samples of each tea infusion were prepared (n = 9). The resulting CN tea infusion samples were labeled FCNT, DCNT, and FDCNT, and were stored in an ultra-low-temperature refrigerator (−80 °C) prior to analysis. However, although discarding the rinse liquor may remove a fraction of water-soluble compounds, all groups were subjected to exactly the same rinse and subsequent main infusion procedure, so any loss introduced by the rinse step would be systematic and consistent across treatments, thus not biasing the relative differences attributed to drying methods.

Determinations of phytochemical composition

2.4

Sample preparation for total flavonoid and polyphenolic contents

2.4.1

The FCN (n = 3), DCN (n = 3), and FDCN (n = 3) samples were ground to a powder using liquid nitrogen, and the powder (0.2 g) was mixed with 1 mL of a 60% ethanol solution. The resulting mixture was subjected to ultrasonic extraction in a water bath at 60 °C for 30 min. The extract was then filtered through filter paper. The filtrate was cooled for 20 min, then centrifuged at 25 °C and 12,000 rpm for 10 min; the supernatant was collected for analysis.

Determination of total flavonoids

2.4.2

The total flavonoid content was determined according to the method described by (Zhang et al., 2012) with some modifications. For total flavonoid analysis, FCN (n = 3), DCN (n = 3), and FDCN (n = 3) samples (0.2 g) solutions were prepared according to the sample preparation steps described in Section 2.4.1. The supernatant (20 μL) was placed in a 96-well plate, and 6 μL of a 5% NaNO_2_ solution was added. The samples were mixed and incubated at room temperature for 6 min. Subsequently, 6 μL of a 10% Al(NO_3_)3 solution was added, and the mixture was allowed to stand for an additional 6 min. Then, 80 μL of a 4% NaOH solution and 88 μL of distilled water were added. The samples were mixed thoroughly and allowed to stand for 12 min. Finally, the absorbance of the samples was measured at 510 nm, with rutin used as the standard for the calibration curve.

Determination of total polyphenols

2.4.3

The total polyphenol content was determined according to the method described by (Chen et al., 2012) with some modifications. For total polyphenol analysis, FCN (n = 3), DCN (n = 3), and FDCN (n = 3) samples (0.2 g each) were prepared according to the sample-preparation steps described in Section 2.4.1. Subsequently, 1 mL of the supernatant was transferred into a 10 mL volumetric flask and diluted to 10 mL volume with distilled water for polyphenol analysis. Next, 10 μL of the supernatant was aspirated, 50 μL of FC color developer was added, and the mixture was mixed well. After allowing it to stand for 2 min at room temperature, 50 μL of a 12% Na2CO3 solution (mass fraction) and 90 μL of distilled water were added, mixed well, and left to stand for 10 min at room temperature. The absorbance was measured at 760 nm, with gallic acid used as the standard for the calibration curve.

Determination of antioxidant activities in vitro

2.5

Free radical scavenging activity: 2,2-Diphenyl-1-picrylhydrazyl,(DPPH) assay

2.5.1

The 2,2-diphenyl-1-picrylhydrazyl (DPPH) radical scavenging activities of FCN (n = 3), DCN (n = 3), and FDCN (n = 3) were determined according to the method described by Wang et al. (2023) with slight modifications. Approximately 0.2 g of the sample was weighed, and 1 mL 80% methanol was added. The mixture was then homogenized on ice and centrifuged at 10,000 ×g for 10 min at 4 °C. 10 μL of the sample solution was withdrawn, and 190 μL of the DPPH working solution was added. The reaction mixture was allowed to stand in the dark for 30 min. Subsequently, 200 μL of the sample was added to a 96-well plate, and the absorbance was measured at 515 nm using a microplate reader (SpectraMax190; Molecular Devices, USA). DPPH radical scavenging activity (%) = [(A_blank_-A_sample_) / A_blank]_ × 100.

Ferric reducing antioxidant power assay (FRAP)

2.5.2

The ferric reducing antioxidant power (FRAP) assay was performed to assess the antioxidant capacities of FCN (n = 3), DCN (n = 3), and FDCN (n = 3), following the method described by (Kavya et al., 2024) with suitable modifications. After the samples were ground to a powder using liquid nitrogen, 0.2 g of the sample was precisely weighed, and 1 mL of 80% ethanol was added. The mixture was vortexed, then oscillated to ensure thorough mixing, centrifuged at 10000 rpm for 10 min at room temperature (25 °C), and the supernatant was collected and kept on ice for further analysis. An aliquot (6 μL) was pipetted into 180 μL of FRAP working solution and 14 μL of distilled water, thoroughly mixed, and incubated for 10 min. Then, 200 μL was transferred into a 96-well plate, and the absorbance was measured at 593 nm. The FRAP value of each sample was calculated using FeSO_4_ as the standard for the calibration curve, with the FRAP antioxidant capacity (U/mL) determined by the equation = C × V_1_ × V_2_/V_sanple_/M × F (where C: FRAP concentration of the sample from the formula of the standard curve, μmol/mL; V_1_: total volume of the reaction, 0.204 mL; V_sample_: volume of the samples in the reaction mixture, 0.006 mL; V_2_: volume of the extract added, 1 mL; M: sample mass, g; F: -dilution factor).

Free radical scavenging 2,2′-azino-bis(3-ethylbenzothiazoline-6-sulfonic acid) (ABTS) assay

2.5.3

The method described by Zhuang et al. (2010) was used to determine the 2,2′-azino-bis(3-ethylbenzothiazoline-6-sulfonic acid) radical scavenging assay (ABTS+) activity of the FCN (n = 3), DCN (n = 3), and FDCN (n = 3) samples, with slight modifications. A solution of 7.4 mmol/L ABTS+ was combined with a 2.6 mmol/L potassium persulfate solution in equal volumes and allowed to stand for 16 h in the dark as the ABTS+ radical stock solution. This stock solution was diluted 30–40-fold with anhydrous ethanol to stabilize A734 at 0.70 ± 0.02, yielding the working solution. For the ABTS assay, the sample solution was prepared in 80% ethanol as described in Section 2.5.2. To 0.02 mL of the sample extract solution, 2 mL of the ABTS+ working solution was added, and the mixture was allowed to stand in the dark for 6 min; the absorbance was then measured at 734 nm. The absorbance of a blank solution (A_blank_) with 0.02 mL of anhydrous ethanol +2 mL of ABTS+ working solution was measured at 734 nm, and the ABTS free radical scavenging was calculated as % = (A_blank−A_sample) × 100/ A_blank_, where A_blank_ is the absorbance of the blank, and A_sample_ is the absorbance of the sample.

Determinations of non-volatile compounds (nVOCs) by UPLC-MS/MS

2.6

Sample preparation and extraction

2.6.1

To investigate the non-volatile compounds (nVOCs) of CN (FCN, DCN, and FDCN), the samples (n = 3 each) were freeze-dried under vacuum (Scientz-100F; Scientz, China). The freeze-dried samples were ground (30 Hz, 1.5 min) into a powder using a grinder (MM 400; Retsch, Germany). The sample (100 mg) was accurately weighed, dissolved in 1.2 mL of pre-cooled 70% methanol, and thoroughly mixed. The mixture was vortexed for 30 s every 30 min, with 6 vortexing sessions per sample. After centrifugation (12,000 ×g for 3 min), the supernatant was collected and filtered through a microporous filter membrane (0.22 μm pore size). The samples were stored in injection bottles at −20 °C until UPLC-MS/MS analysis.

The nVOCs of tea infusions (FCNT, DCNT, and FDCNT) prepared using CN were also investigated. The tea infusion samples (2 mL) of FCNT, DCNT, and FDCNT were freeze-dried under vacuum (Scientz-100F, Scientz, China), and 1 mL of pre-cooled 70% methanol solution was added to each sample. The samples were vortexed in 70% methanol at room temperature for 15 min and then incubated at −20 °C for 30 min. The samples were centrifuged (12,000 ×g, 3 min) at 4 °C, the supernatant was pipetted and filtered through a microporous filter membrane (0.22 μm pore size) into a glass-lined tube in a brown injection bottle and stored at −20 °C until UPLC-MS/MS analysis.

UPLC conditions

2.6.2

UPLC conditions: SB-C18 column (1.8 μm, 2.1 mm × 100 mm, Foster City, CA); mobile phase A was ultrapure water with 0.1% formic acid), and mobile phase B was acetonitrile with 0.1% formic acid; elution gradient: 0.00 min, 95% A, 5% B; 0.00–9.00 min, 95% to 5% A, 5 to 95% B; 9.00–10.00 min, 5% A, 95% B; 10.00–11.10 min, 5% to 95% A, 95% to 5% B; 11.10–14 min, 95% A, 5% B, flow rate of 0.35 mL/min; column temperature was 40 °C; injection volume of 2 μL. The effluent was alternately connected to an electrospray ionization triple-quadrupole linear ion trap (QTRAP).

Mass spectrometry

2.6.3

Non-volatile metabolites were analyzed using a UPLC system coupled to an electrospray ionization (ESI) triple quadrupole mass spectrometer operated in scheduled multiple-reaction monitoring (MRM) mode. The ESI source parameters were: source temperature 550 °C; ion spray voltage 5500 V (positive) / −4500 V (negative); ion source gas I (GSI), gas II (GSII), and curtain gas (CUR) set to 50, 60, and 25 psi, respectively; the collision-activated dissociation (CAD) was set to high. QQQ scans were acquired as MRM experiments with collision gas (N₂) set to medium. For each metabolite, MRM transitions (Q1/Q3) and compound-specific declustering potential (DP) and collision energy (CE) were optimized and recorded (Table S1). Transitions were monitored in retention-time windows corresponding to chromatographic elution.

Determinations of volatile compounds (VOCs) by gas chromatography mass spectrometry (GC–MS/MS)

2.7

Sample preparation and extraction

2.7.1

The volatile organic compounds (VOCs) in the CN samples (FCN, DCN, and FDCN; n = 3 each) and tea infusion samples (FCNT, DCNT, and FDCNT; n = 3 each), subjected to various drying methods, were analyzed after freeze-drying the samples in a vacuum (Scientz-100F, Scientz, China) and grinding (30 Hz, 1.5 min) to a powder using a mill (MM400, Retsch, Germany). The powder (500 mg) was immediately transferred to a 20 mL-capacity headspace injection vial (Agilent, Palo Alto, CA, USA). Saturated NaCl solution (2 mL) and 10 μg/mL ethyl decanoate (20 μL) were added as internal standards. Sample extraction was performed using a fully automated headspace solid-phase microextraction (HS-SPME) method. During the SPME analysis, each vial was placed at 60 °C and shaken for 5 min. Subsequently, a 120 μm [DVB/CWR/PDMS, Stableflex (2 cm)]-coated fiber (Supelco, Inc., Bellefonte, PA, USA) extraction head was positioned 1 cm above the liquid surface and held in place for 15 min. Finally, the extraction head was inserted into the GC inlet, and the temperature was maintained at 250 °C for 5 min; thereafter, GC–MS analysis was performed. Three independent biological replicates were prepared for each treatment (Feng et al., 2019).

GC–MS conditions

2.7.2

An Agilent 8890-7000D (Agilent, Santa Clara, CA, USA) instrument was used to detect VOCs. GC analysis was performed using a DB-5MS capillary column (30 m × 0.25 mm; 0.25 μm, Agilent J & W Scientific, Folsom, CA, USA), and the carrier gas was high-purity helium (purity >99.999%) at a constant flow rate of 1.2 mL/min. The injector and detector temperatures were set to 250 °C and 280 °C, respectively. The GC oven was initially set to 40 °C for 3.5 min and then gradually increased at a rate of 10 °C per min to 100 °C, followed by a rate of 7 °C per min to 180 °C, and finally at a rate of 25 °C per min to reach the final temperature of 280 °C and held for 5 min.

Mass spectra were acquired in the electron collision (El) ionization mode at an energy of 70 eV. The ion source temperature was 230 °C, the quadrupole temperature was 150 °C, and the interface temperature was 280 °C. The scanning method used was selected-ion detection mode (SIM) acquisition, and the selection of qualitative/quantitative ions was conducted in accordance with GB 23200.8–2016 (National Food Safety Standards).

Data processing and statistical analysis

2.8

All experiments and samples analyzed were performed in triplicate, and results are expressed as the mean ± standard error of the mean (SEM). Data were evaluated using one-way analysis of variance (ANOVA) to identify significant differences between samples. All statistical analyses were performed using SPSS statistical software (version 19.0; SPSS Inc., Chicago, IL, USA).

A combined detection platform based on UPLC-MS/MS and GC–MS–MS method (Chen et al., 2023; Zheng et al., 2015) was used for qualitative and quantitative analyses of metabolites based on fragmentation pattern, retention time, and mz using the standards from Wuhan Metware Biotechnology Co.'s self-built MWDBV2.0 database (Metware Biotechnology Co., Ltd., Wuhan, China) and public databases. Metabolite quantification was performed using MRM analysis with triple quadrupole mass spectrometry. After acquiring mass spectra of metabolites from various samples, the peak areas for each substance were integrated. Simultaneously, the mass spectral peaks of identical metabolites across different samples were integrated and corrected. Mass spectral data were processed using ANALVST 1.6.3, with the peak area of each chromatographic peak representing the relative concentration of the corresponding substance. Principal component analysis (PCA), hierarchical cluster analysis (HCA), and orthogonal partial least-squares discriminant analysis (OPLS-DA) were performed using the Maiwei Cloud. These variables were combined with variable importance in projection (VIP) values >1.0, fold changes (FC) >2, or FC ≤0.2 with P < 0.05 to identify differential metabolites. Significance analysis was performed using SPSS (version 26.0; P < 0.05). PCA and other plots were generated in R 4.3.1 (R Foundation) using the ggplot2 and pheatmap packages; bar graphs were generated in OriginPro 2023 (OriginLab, USA).

Results and discussion

3

Effects of different drying methods on CN

3.1

Drying fresh CN flowers (FCN) effectively removed moisture, extending their shelf life and maintaining their quality. In this study, HAD and VFD, which have been widely demonstrated as effective methods for food dehydration, were evaluated for their drying efficacy on CN flowers (Zheng et al., 2015). The results of this study show that the moisture content of samples treated with HAD (7.59 ± 0.09%) or VFD (7.23 ± 0.08%) was significantly lower than that of the sample FCN (86.59 ± 0.42%), representing a reduction of 91.2% and 91.6%, respectively (P < 0.05 in both cases; Fig. 1A). Although VFD is believed to remove water more thoroughly than HAD (Le et al., 2022), no statistically significant difference (P > 0.05) in final moisture content was observed between the two treatments under the specific experimental parameters employed. Notably, both methods achieved a moisture content below the 8% industry standard required for the preservation of floral products. (Topuz et al., 2023). Furthermore, the shorter drying duration of CN flowers compared with Bletilla striata may be attributed to the succulent CN petals (Lu et al., 2021). The high initial moisture content provides a strong internal driving force, facilitating rapid vapor diffusion through the parenchyma.Fig. 1Effects of different drying methods on Camellia nitidissima Chi (CN) flowers. (A) Moisture content of fresh CN flowers (FCN), hot air-dried CN flowers (DCN), and vacuum freeze-dried CN flowers (FDCN). Values denoted by different letters are significantly different at P < 0.05. Data are presented as the mean ± SEM. (B) Effect of different drying methods on the morphological characteristics of CN flowers. Bar = 10 mm.Fig. 1

Despite the comparable dehydration performance, the two methods yielded divergent outcomes regarding the samples' morphological characteristics (Fig. 1B). Whereas the flowers after both HAD and VFD had a golden yellow color, hot air-dried CN flowers (DCN) did not maintain the original morphological structure, whereas vacuum freeze-dried CN flowers (FDCN) effectively preserved a completely intact and flawless flower appearance(Zhang et al., 2024). Given that visual appearance is a crucial attribute that determines the commercial value of processed scented tea, FDCN may have greater commercial potential (Kelley et al., 2000).

Effect of different drying methods on the non-volatile components of CN flowers

3.2

UPLC-MS/MS identified a total of 1960 compounds in FCN, DCN, and FDCN samples, mainly including flavonoids (22%), phenolic acids (14%), and lipids (8%). While the dehydration process in both methods significantly expanded the detectable metabolite profile. Due primarily to the concentration effect from desiccating succulent CN flowers, 219 substances present in DCN and FDCN but absent in FCN were detected (Fig. 2A) (Uribe et al., 2024); and the divergent metabolic signatures between drying methods may result from modulated physiological metabolism or selective degradation of thermosensitive compounds (Table S2) (Li et al., 2024; Li et al., 2024).Fig. 2Effects of drying methods on non-volatile components of Camellia nitidissima Chi (CN) flowers. (A) Venn diagram showing metabolites detected in different CN flower groups (a) and the classification and counts of metabolites uniquely detected in DCN (b) and FDCN (c). (B) Total flavonoid (a) and total phenolic (b) contents of CN flower samples. (C) Antioxidant capacity of CN flower samples assessed by DPPH radical scavenging activity (a), ABTS+ radical scavenging capacity (b), and ferric reducing antioxidant power (FRAP) (c). FCN, fresh CN flowers; DCN, hot air-dried CN flowers; FDCN, vacuum freeze-dried CN flowers. Different letters indicate significant differences (P < 0.05). Data are presented as mean ± SEM.Fig. 2

Effects of HAD on non-volatile metabolites in CN flowers

3.2.1

HAD can alter floral metabolite profiles through concentration effects induced by dehydration or through the biological stress response accompanying elevated temperatures, thereby generating novel phenolic compounds. We applied the criteria of fold change (FC) ≥ 2 or ≤ 0.2 and VIP > 1 to systematically evaluate the impacts of HAD on the non-volatile metabolite profile of CN flowers. Compared with FCN, HAD resulted in 1108 metabolites being significantly upregulated and 41 downregulated (Table S3). Notably, 163 metabolites exhibited stronger enrichment (Log2FC > 3, P < 0.05; Table S4). Collectively, these quantitative shifts provide robust evidence that HAD processing enables effective metabolite concentration. Indicating that DCN has a rich nutritional composition. Additionally, our study found that DCN contained 45 unique metabolites (Fig. 2Aa, Table S5), which were predominantly amino acids and derivatives (9 compounds), flavonoids (6 compounds), and phenolic acids (9 compounds; Fig. 2Ab). These metabolite classes are frequently implicated in temperature-stress adaptation. Among these compounds, Viscumneoside III was markedly upregulated after HAD but was not comparably emphasized in FCN or FDCN (Wang et al., 2024; Wang et al., 2024). These indicate that the elevated temperature during HAD (40 °C) may function as a mild heat stimulus, triggering cellular stress responses and reshaping metabolomic (Thakur et al., 2019).

Pathways associated with stress-related secondary metabolism (e.g., phenylpropanoid biosynthesis and quercetin-related routes) were significantly enriched in comparisons involving DCN (Fig. S1), further demonstrating that even mild hot-air drying (40 °C) can induce a biosynthetic/stress response that reshapes the metabolite network rather than simply concentrating it (Ampofo et al., 2020; Vysochina & Voronkova, 2013). Simultaneously, after HAD, phenolic acid compounds showed greater variability (Fig. S2), which is consistent with their dual sensitivity under HAD: (i) heat-facilitated oxidative/enzymatic browning converting phenolics to quinones (Zhang et al., 2025) and (ii) stress-induced pathway activation increasing specific downstream metabolites (Fig. S3), while HAD induces broader stress-associated remodeling that may compromise net retention of key antioxidant pools. This pattern suggests that HAD may induce additional metabolic conversions, particularly within phenolic-associated pathways, beyond those observed under the FCN and FDCN conditions.

Effects of VFD on non-volatile metabolites in CN flowers

3.2.2

Compared to DCN, FDCN shows fewer changes in metabolite types, and enrichment analysis revealed no complex alterations in metabolic pathways, indicating that VFD primarily concentrates and stores substances. After VFD, 834 metabolites increased and 108 decreased, while 251 metabolites were strongly enriched (Log2FC > 3, P < 0.05; Table S6 and S7). This indicates that FDCN produces fewer overall shifts than DCN. Additionally, FDCN possessed 48 unique metabolites, lipids accounted for 54% of the unique metabolites in FDCN (Fig. 2Ac, Table S8), including 10 free fatty acids (e.g., 17-hydroxylinolenic acid), 7 lysophosphatidylethanolamines, and 5 lysophosphatidylcholines, suggesting that VFD may preferentially preserve labile lipid species by suppressing heat-driven oxidation, decomposition, and limiting enzymatic lipolysis under low temperature and vacuum (Xiao et al., 2024).

Meanwhile, the concentration and storage effects of VFD can more effectively enhance the nutritional properties of dried flowers (the highest total flavonoids and total polyphenols and the most antioxidant capacity than DCN and FCN) (Fig. 2B–C). This result aligns with previous studies, confirming the superiority of VFD in preserving the quality of edible flowers (Tan et al., 2021). The lower temperature and vacuum environment effectively prevented the degradation and transformation of flavonoids (114 compounds) and polyphenols (91 compounds) in FDCN (Li et al., 2024; Li et al., 2024). Here, we found that the composition of the top 15 relatively abundant flavonoid compounds was highly similar between FDCN and DCN samples (10 shared compounds), and 8 of these were present at significantly higher levels in FDCN than in DCN (Fig. S4). In conclusion, the use of VFD (FDCN) effectively preserves a broader range of metabolites, particularly lipids, flavonoids, and polyphenols, in comparison to other drying methods, especially in terms of maintaining the antioxidant capacity and overall quality of edible flowers.

Effect of different drying methods on the volatile components of CN flowers

3.3

Numerous teas have high commercial value owing to their distinctive flavors, making the detection and analysis of flavor compounds in different CN flower preparations essential for comprehensive product quality evaluation (Xiao, 2017). In total, 823 volatile compounds were identified in CN flowers (Table S9), and consistent with previously reported floral scent compositions, terpenoids (161), esters (159), and heterocyclic compounds (121) represented the three dominant chemical classes (Li et al., 2025). Consistent with observations in honeysuckle tea, no new volatile metabolites were detected in drying CN flowers (Peng et al., 2006), but drying drives aroma attenuation by reducing levels of terpenoids and esters in dried CN (Liu et al., 2024; Liu et al., 2024). However, although PCA indicated group differences (Fig. 3A), FCN and FDCN clustered more closely than either did with DCN, and the FCN vs. FDCN comparison yielded fewer differential metabolites (Fig. 3B).Fig. 3Effects of drying methods on volatile metabolites of Camellia nitidissima Chi (CN) flowers. (A) Principal component analysis (PCA) of volatile metabolite profiles in CN flowers. (B) Venn diagram showing the overlap of differential volatile metabolites identified in the DCN vs. FCN and FDCN vs. FCN comparisons. (C) Scatter plot of relative odor activity values (rOAVs) for volatiles detected in different CN flower preparations (FCN, DCN, and FDCN). (D) Odor-attribute annotation of downregulated differential volatile metabolites in the DCN vs. FCN (a) and FDCN vs. FCN (b) comparisons. FCN, fresh CN flowers; DCN, hot air-dried CN flowers; FDCN, vacuum freeze-dried CN flowers.Fig. 3

To further link chemical variation to sensory relevance, relative odor activity values (rOAV) were used to identify key odorants (rOAV >1): a total of 30 shared key odorant types were detected across FCN, DCN, and FDCN, with (E, Z)-2,6-nonadienal (cucumber, green) consistently exhibiting the highest rOAV in all samples due to its low perception threshold (Wang et al., 2025). However, the number of high-potency odorants (rOAV >100) differed among treatments (6 in FCN, 7 in DCN, and 5 in FDCN; Table S10), demonstrating that drying not only weakens overall aroma abundance but also reshapes the potency structure and dominant contributors of the odor-active profile (Fig. 3B). Overall, the compounds contributing most significantly to the odor were identical across all three samples, but compared to fresh flowers, both methods markedly diminished the composition of flavor compounds, with differing degrees and directions of reduction.

Effects of HAD on volatile metabolites in CN flowers

3.3.1

Overall, HAD (40 °C, no vacuum) resulted in a net depletion of the CN volatile pool, indicating that this mild drying process predominantly weakens aroma intensity (fruity and floral aromas) rather than generating new aroma complexity. This conclusion is directly supported by the quantitative pattern that 72 volatiles were significantly altered, with substantially more compounds decreased (57) than increased (15) (Tables S11 and S12; Fig. 3B). Under this condition, volatile losses are most plausibly explained by diffusion and evaporation during heating (Venskutonis, 1997).

Importantly, the dominant decrease in volatiles implies not only weakened aroma intensity but also a directional reconfiguration of aroma quality. An orthogonal partial least squares discriminant analysis (OPLS-DA) model constructed using volatiles with rOAV >1 clearly separated DCN from FCN and exhibited strong predictive performance (Q^2^ = 0.818, R^2^Y = 0.997; Fig. S5), confirming that HAD results in reproducible restructuring of aroma-relevant composition. Consistent with the OPLS-DA separation, nine key odorants (VIP > 1) were identified as discriminative markers (Huang et al., 2025), indicating that group differentiation is driven by a limited set of aroma-active compounds rather than uniform shifts across all volatile classes. Among them, rose oxide isomers [(2S,4R)-4-methyl-2-(2-methylprop-1-en-1-yl)tetrahydro-2H-pyran and 2H-pyran, tetrahydro-4-methyl-2-(2-methyl-1-propenyl)-], which are associated with floral and fruity notes, peaked in FCN (Liu et al., 2022; Zhao et al., 2025), whereas several other discriminative markers exhibited higher abundance in DCN (Fig. 3 Da; Table S10; Fig. S6). Collectively, these results indicate that HAD (40 °C) not only causes a general loss of odorants but also reduces the relative contribution of floral- and fruity-associated odor-active compounds in CN, thereby weakening the characteristic floral–fruity aroma expression after hot-air drying (Wu et al., 2025).

Effects of VFD on volatile metabolites in CN flowers

3.3.2

FDCN shares a similar chemical composition to FCN, but exhibits a lower odor activity value (rOAV). After VFD, 111 volatiles showed differential abundance, including 104 decreased and only 7 increased (Fig. 3C; Tables S13 and S14). This highly unbalanced pattern demonstrates that VFD predominantly weakens volatile retention rather than promoting the formation or accumulation of new odorants. Such extensive losses are consistent with the physical–process characteristics of vacuum freeze drying, where the formation of a porous matrix, deep vacuum conditions, and sustained ice sublimation collectively facilitate volatile escape (Liao et al., 2025; Thamkaew et al., 2021; Wang et al., 2017; Zhao et al., 2023).

Beyond the overall decrease, VFD induced a directional reshaping of aroma quality by preferentially suppressing specific odorant categories. Based on database annotation, VFD significantly reduced the concentrations of multiple sweet-associated compounds. In addition, several compounds negatively associated with floral, herbal, and woody characteristics were markedly downregulated, including phenylacetic acid propyl ester, 2-Cyclohexen-1-ol, 2-methyl-5-(1-methylethenyl)-, acetate, and (Z)-6-Nonen-1-ol, acetate (Fig. 3Db). To quantitatively assess how these compositional changes translate into aroma relevance, an orthogonal partial least squares discriminant analysis (OPLS-DA) model was constructed using odor-active volatiles (rOAV >1). The model effectively discriminated aroma profiles across groups, and most key odorants exhibited lower rOAV values in VFD-treated samples than in other treatments (Table S10; Fig. S6). Collectively, these results indicate that VFD not only reduces total volatile abundance but also decreases the sensory contribution of sweet- and floral/herbal/woody-associated odorants, thereby broadly suppressing the characteristic aroma expression of CN after drying (Zhao et al., 2023).

Impact of drying methods on the metabolic profile of tea infusion

3.4

As a tea substitute, processed CN is brewed into a tea infusion for direct consumption. Consequently, analyzing only the dried flowers cannot fully reflect the quality attributes of the final product. Systematic comparison of the metabolic profile differences in tea infusions, the ultimate beverage product, derived from preparations dried by different methods, is essential. Although parameters such as water source and temperature significantly influence the composition of tea infusion metabolites, this study employed a standardized brewing protocol (FCNT: FCN tea infusion; DCNT: DDCN tea infusion; FDCNT: FDCN tea infusion) to minimize interference from brewing variables. This enabled precise elucidation of the drying method's impacts on the metabolite profile and flavor characteristics of the tea infusion (Cao et al., 2022).

To clarify the metabolic relationship between flowers and tea infusion, PCA was first performed on all metabolites based on UPLC-MS/MS and GC–MS data. Metabolic profiles of the tea infusions (FCNT, DCNT, FDCNT) and their corresponding raw flower materials (FCN or dried flowers DCN/FDCN) were significantly separated (Fig. 4A–C). This indicates that the brewing process induced systematic compositional transformations, a phenomenon well documented in tea-brewing studies (Pastoriza et al., 2017). Furthermore, OPLS-DA confirmed inter-group differences. All three models showed excellent predictive performance (Q^2^ > 0.99, R^2^Y = 1; Fig. 4D-F), statistically validating the robust separation between the metabolic signatures of the flowers and their tea infusions. More importantly, heatmap analysis indicated that the levels of most metabolites were significantly higher in the tea infusions than in the raw flower materials (Fig. S7). This was likely due to the enrichment of floral metabolites during brewing and the potential release of volatile compounds (Liu et al., 2024; Liu et al., 2024). Based on these findings, a subsequent comparison of the compositions of non-volatile and volatile compounds in tea infusions prepared from flowers dried by different methods is warranted to provide a more comprehensive understanding of their quality and value.Fig. 4Multivariate analysis of metabolite profiles in Camellia nitidissima Chi (CN) flowers and the corresponding tea infusions based on mixed data of UPLC–MS/MS (non-volatiles) and GC–MS (volatiles). (A–C) Principal component analysis (PCA) score plots showing the separation among flower samples (FCN, DCN, FDCN) and the corresponding tea infusions (FCNT, DCNT, FDCNT) using non-volatile and volatile datasets (as indicated in each panel). (D—F) Orthogonal partial least squares discriminant analysis (OPLS-DA) results for the same datasets, including score plots (a) and permutation tests (b) to validate the corresponding OPLS-DA models. FCN, fresh CN flowers; DCN, hot air-dried CN flowers; FDCN, vacuum freeze-dried CN flowers; FCNT, tea infusion of fresh CN flowers; DCNT, tea infusion of hot air-dried CN flowers; FDCNT, tea infusion of vacuum freeze-dried CN flowers.Fig. 4

Effect of different drying methods on the non-volatile components of CN flower infusions

3.5

Comparative analysis of tea infusions reveals that regardless of the drying method employed, the resulting infusions exhibit greater metabolic diversity. However, significant differences in the metabolic composition of infusions prepared from samples dried using different methods are apparent. Such variations may suggest distinct health benefits. PCA demonstrated good intra-group consistency and significant metabolic separation among the infusion groups (Fig. 5A), indicating that drying treatment is the key factor driving differences in the metabolic profiles of tea infusions. Based on UPLC-MS/MS analysis, a total of 1995 compounds were identified in FCNT, DCNT, and FDCNT (Table S15), while 1689 compounds were common to all three infusion types (Fig. 5B). Flavonoids (435, 22% of total metabolites), phenolic acids (278, 14%), amino acids and derivatives (179, 9%), and lipids (175, 9%) constituted the major classes of non-volatile metabolites shared by all three infusion types. What's more, the thermogram shows that the tea broth obtained from dried flowers is significantly richer in metabolites than that of fresh flowers (Fig. 5C).Fig. 5Effects of different drying methods on non-volatile metabolites in Camellia nitidissima Chi (CN) tea infusions. (A) Principal component analysis (PCA) of non-volatile metabolite profiles in different CN tea infusions. (B) Venn diagram showing non-volatile metabolites detected in three types of CN tea infusions. (C) Heatmap of non-volatile metabolite profiles across the three CN tea infusion groups. FCNT, tea infusion of fresh CN flowers; DCNT, tea infusion of hot air-dried CN flowers; FDCNT, tea infusion of vacuum freeze-dried CN flowers.Fig. 5

Effects of HAD on non-volatile metabolites in CN flowers infusion

3.5.1

Compared to FCNT, DCNT showed higher concentrations of certain trace substances, such as terpenes. Differential metabolite analysis revealed 1151 metabolites showing significant changes, including 1089 up-regulated and only 62 down-regulated compounds (Table S16). The upregulated metabolites were predominantly flavonoids (211), phenolic acids (156), and lipids (100). Notably, a stark contrast was observed in the analysis of unique metabolites: FCNT contained only two unique compounds (AICAR phosphate and 7β-hydroxydarutigenol, Table S17), whereas DCNT possessed 35 unique metabolites, including 7 nucleotides/derivatives, 6 flavonoids, and 5 phenolic acids (Table S18 and S19). These results confirm that HAD significantly enhances the metabolite abundance in CN flower infusions, establishing DCNT as metabolically richer than FCNT.

In addition to the concentration effects of drying, HAD-induced microstructural modifications (e.g., pore formation) in CN flowers facilitate the rapid and efficient dissolution of metabolites during brewing. This structural transformation results in higher metabolite concentrations in DCNT compared to FCNT (Yadav et al., 2017). Furthermore, among the top 30 differential metabolites ranked by VIP value (Fig. 6A), 29 were significantly upregulated in DCNT, with the majority being Amino acids and derivatives (8 types). The compounds with the highest VIP values included 5’-Demethylaquillochin, Homomangiferin, and 6,7-dimethoxy-2-[2-phenylethyl]chromone. The elevated levels of these bioactive substances contribute to the enhanced nutritional value and biological activity of the tea liquor prepared via HAD.Fig. 6Key differential non-volatile metabolites (VIP-ranked) in CN tea infusions. (A) Top 30 differential non-volatile metabolites with the highest VIP values in the DCNT vs. FCNT comparison. (B) Top 30 differential non-volatile metabolites with the highest VIP values in the FDCNT vs. FCNT comparison. FCNT, tea infusion of fresh CN flowers; DCNT, tea infusion of hot air-dried CN flowers; FDCNT, tea infusion of vacuum freeze-dried CN flowers.Fig. 6

Effects of VFD on non-volatile metabolites in CN flowers infusion

3.5.2

VFD preserved the highest levels of bioactive compounds (e.g., flavonoids and phenolic acids) and better lipid properties, thereby imparting FDCNT with a richer taste profile and deeper color and enhancing the potential health benefits of the infusion (Oluwole et al., 2022). Relative to FCNT, FDCNT exhibited 1179 differential metabolites, including 1122 up-regulated and only 57 down-regulated compounds. The up-regulated metabolites were predominantly flavonoids (226), phenolic acids (173), and lipids (122), with 59 compounds in FDCNT showing prominent upregulation (e.g., 4-hydroxybenzaldehyde, Log₂FC = 19.12; Table S20 and S21). FDCNT possessed 13 unique metabolites, including 3 lipids and 3 phenolic acids (Table S22). These findings indicate that infusions prepared from VFD-treated CN flowers contain more metabolite types and exhibit significantly higher concentrations of numerous metabolites than FCNT infusions.

Furthermore, among the top 30 differential metabolites ranked by VIP value (Fig. 6B), all 30 compounds were significantly up-regulated in FDCNT, predominantly lipids (15 types). The compounds with the highest VIP values included 2R-hydroxy-9Z,12Z,15Z-octadecatrienoic acid (a free fatty acid), 9-Oxo-10E,12Z-octadecadienoic acid (a free fatty acid), and Sibiricose A3. Effective retention of lipids and other bioactive substances may not only improve the infusion's mouthfeel (e.g., imparting greater body and smoothness) but also confer unique health benefits through specific biological activities (e.g., utilization of free fatty acids as energy substrates) (Li et al., 2023). Concurrently, the abundant phenolic acids and terpenoid compounds (particularly Monoterpenoids amd sesquiterpenoids) exhibit potent antitumor, anti-inflammatory, and antibacterial activities (Kumar & Goel, 2019; Yang et al., 2024; Yang et al., 2024).

Effect of different drying methods on the volatile components of CN flower infusions

3.6

Similar to analyses of flowers, the compound with the highest odor activity value was consistent across all three tea infusions, though the composition of the other compounds differed. Furthermore, while no specific compounds were detected in the comparison of tea flowers, they were present in the comparison of tea infusions. This indicates that the brewing process induces transformations in certain compounds, simultaneously shaping entirely distinct volatile profiles. This study identified 871 volatile compounds across FCNT, DCNT (HAD infusion), and FDCNT (VFD infusion), and no group-unique components were identified (Table S23). Major classes included 168 terpenes (19%), 148 esters (17%), and 172 heterocyclic compounds (16%). Moreover, the flavor differences between tea liquor groups brewed from different drying groups also exhibit significant distinctions from the flavor variations observed between tea leaf groups. PCA demonstrated good sample reproducibility and clear separation among the three infusion groups. Notably, although DCN and FDCN flowers showed partial overlap in the PCA results, DCNT and FDCNT infusions were completely separated (Fig. 7A). This suggests that drying method-induced differences in flowers (volatile composition, physical structure) significantly influence the release efficiency and composition of flavor compounds during subsequent brewing (Wang et al., 2024; Wang et al., 2024). Consequently, analyzing infusions is crucial for evaluating final product quality and understanding the practical impact of processing.Fig. 7Effects of drying methods on volatile metabolites in Camellia nitidissima Chi (CN) tea infusions. (A) Principal component analysis (PCA) of volatile metabolite profiles in CN tea infusions (FCNT, DCNT, and FDCNT). (B) Top 25 differential volatile metabolites identified in the DCNT vs. FCNT (a) and FDCNT vs. FCNT (b) comparisons. (C) Odor-attribute annotation of upregulated and downregulated differential volatile metabolites in the DCNT vs. FCNT (a, b) and FDCNT vs. FCNT (c, d) comparisons. (D) Scatter plot of relative odor activity values (rOAVs) for volatiles detected in different CN tea infusions. FCNT, tea infusion of fresh CN flowers; DCNT, tea infusion of hot air-dried CN flowers; FDCNT, tea infusion of vacuum freeze-dried CN flowers.Fig. 7

Effects of HAD on volatile metabolites in CN flowers infusion

3.6.1

Based on database annotation, we focused on down-regulated metabolites in DCNT and their potential flavor contributions, and found that HAD significantly reduced the concentrations of numerous volatile compounds, with 514 differential metabolites identified by comparing DCNT with FCNT (VIP ≥ 1, P < 0.05, |log2FC| ≥ 1). Among them, 403 were down-regulated, predominantly terpenes (92 compounds), esters (82 compounds), and heterocyclic compounds (55 compounds; Table S24 and S25). As these substances are key contributors to floral aroma, their widespread decrease implies severe flavor loss in DCNT (Li et al., 2025). Furthermore, β-eudesmol showed the greatest downregulation (log2FC = −8.59; Fig. 7Bb); this volatile component, widely distributed across various plants, further indicates the continuation of HAD-induced volatile losses in the infusion (Díaz-Maroto et al., 2002). Additionally, isomaltol is detected exclusively in DCNT but is absent in DCN, indicating that the hot-water infusion process triggers the transformation of certain compounds, thereby yielding a significantly more complex aroma profile (Okonkwo et al., 2024).

To assess the sensory impact of these changes, differential metabolites were annotated for odor attributes, and metabolites with rOAV >1 were identified as key odorants (Table S26). Down-regulated metabolites in the DCNT vs. FCNT comparison were primarily enriched in sweet, green, and fruity attributes (Fig. 7Ca, b), likely attributable to the severe losses of terpenes and esters (Song et al., 2008). rOAV analysis identified 157 and 149 key odorants (rOAV >1) in FCNT and DCNT, with each possessing 3 unique key odorants (Fig. 7D). This pattern directly confirms that HAD broadly and profoundly induces the loss of aromatic compounds, leading to a significant reduction in the sensory contribution of key odorants in the infusion.

Effects of VFD on volatile metabolites in CN flowers infusion

3.6.2

Unlike the flavor characteristics of FDCN in the tea control group, FDCNT exhibits the most complex aroma profile among the tea infusion control groups. A comparison between FDCNT and FCNT identified 343 differential metabolites, including 153 up-regulated and 190 down-regulated metabolites. Of these, 99 exhibited log₂FC > 2 (Table S27 and S28). Notably, several compounds that were detected in both FDCN and FDCNT (e.g., 1-Octanol, 1-Undecanol, and Carvone) showed a higher abundance in FDCNT than in FCNT (Fig. 6Bb), indicating that VFD-treated materials may retain a larger pool of extractable aroma-related constituents that can be transferred into the infusion during hot-water brewing. Considering the complex effects of the brewing process on aroma release, further systematic studies are required to better understand the kinetics and drivers underlying these changes. Further analysis of FDCNT vs. FCNT showed increases in 25 fruity aroma compounds and decreases in 32 sweet aroma compounds (Fig. 7Cc, d). rOAV analysis revealed 171 key odorants (rOAV >1) in FDCNT (Fig. S8, Table S26), with 13 unique to this group—indicating that VFD-prepared tea liquor possesses the most abundant aroma composition.

Collectively, these results demonstrate that VFD not only alters volatile concentrations but also substantially reshapes the infusion flavor profile, a consequence of both drying-induced floral changes and brewing-triggered metabolite transformations (Li et al., 2024; Li et al., 2024; Wang et al., 2024; Wang et al., 2024). An orthogonal partial least squares discriminant analysis (OPLS-DA) model effectively discriminated infusion aroma profiles (rOAV >1) with excellent fitness (Q^2^ = 0.991, R^2^Y = 0.996, Fig. S9), identifying 97 key discriminative odorants (VIP > 1). Remarkably, 76 of these odorants exhibited the highest rOAV values in FDCNT (Fig. S10), indicating that FDCNT contains more high-potency odorant types, supporting its potential for a complex and unique flavor profile—an observation that stands in stark contrast to the results of floral analyses. Crucially, 3-Cyclohexene-1-methanethiol, .alpha., .alpha., 4-trimethyl-, which is characterized by sulfury, aromatic, grapefruit, naphthyl, resinous, and woody notes, was the highest rOAV odorant across all infusions (Fig. 7D). This compound is also recognized as an important odorant in grapefruit juice (Lin et al., 2002).

Conclusions

4

This study systematically compares the effects of hot-air drying (HAD) and vacuum freeze-drying (VFD) on the Camellia nitidissima Chi (CN) flowers and their corresponding tea infusions. For the flowers, both drying methods reduced moisture content to below 8%, with VFD preserving flower morphology significantly better. VFD-treated flowers exhibited higher concentrations of flavonoids, polyphenols, and lipids, while HAD induced the activation of stress-related metabolic pathways, upregulating a larger number of metabolites, particularly phenolic acids. Additionally, VFD-treated flowers had the highest levels of total flavonoids, total polyphenols, and total antioxidant capacity, making it the superior method for preserving these key bioactive compounds. Both drying methods resulted in significant losses of volatile floral compounds, with VFD leading to a more pronounced reduction in key flavor components, such as terpenes and esters.

For the tea infusions, VFD infusions showed a richer, more diverse metabolite profile, with higher levels of lipids and flavonoids (e.g., anthocyanins, isoflavones, flavonoid glycosides) than FCNT and DCNT infusions. The volatile profiles in FDCNT infusions were more complex, containing 171 key odorants (rOAV >1), surpassing those in FCNT (157) and DCNT (149). Among 97 differential aroma compounds, 76 exhibited peak rOAV values in FDCNT, contributing to its distinctive and complex aroma. In contrast, DCNT had advantages in certain nucleotide derivatives (e.g., cordycepin) and other metabolites but displayed a relatively simpler volatile profile.

In summary, the effects of HAD and VFD on CN flowers and their infusions differ, with each method offering distinct advantages. However, these results may underscore VFD as the preferred method for retaining both the nutritional and sensory attributes of CN flowers and their tea infusions, offering a more effective means of preserving bioactivity. Moreover, inconsistent results from comparisons of flowers and infusions prepared with different drying methods suggest that a comprehensive evaluation of both is crucial for selecting the best drying method for commercial use.

CRediT authorship contribution statement

Bin Yuan: Writing – original draft, Methodology, Investigation, Formal analysis, Data curation. Xiao-ming Tian: Writing – original draft, Methodology, Data curation. Xin-yu Yu: Writing – review & editing, Data curation. Lu Zhu: Investigation, Data curation. Mei-fei Zhu: Writing – review & editing, Formal analysis, Data curation. Yuan-yuan Lu: Writing – review & editing, Data curation. Guang-feng Xiang: Methodology, Data curation. Gao-fei Li: Investigation. Lan Zhou: Investigation, Data curation. Hao Lv: Writing – review & editing, Supervision, Methodology, Data curation. Fu-liang Hu: Writing – review & editing, Supervision, Methodology, Investigation, Funding acquisition.

Fundings

This work was supported by the Modern Agroindustry Technology Research System from the Ministry of Agriculture and Rural Affairs of China (grant number CARS-44), 10.13039/501100004735Natural Science Foundation of Hunan Province (grant number 2023JJ60116).

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.