Abstract
Artemisia capillaris (Yinchen), a saline-alkali halophyte from China's Yellow River Delta, is made into functional herbal tea. Postharvest drying dominates its flavor and bioactive metabolites, yet temperature-gradient impacts on its sensory traits, two-dimensional volatiles and metabolome are unreported. This study characterized sample saline soil background and set four treatments: CK (fresh), T50 (50 °C), T100 (100 °C), T120 (120 °C), integrating QDA, GC × GC-TOFMS and UHPLC-QTOF-MS metabolomics. T50 brought bitter, grassy herbal scent; T100 sweet aroma; T120 roasted notes. Among 487 volatiles, low heat retained terpenoids while high heat generated Maillard heterocycles. 8821 differential metabolites were detected, with phenylpropanoid and flavonoid biosynthesis most disturbed. This work reveals temperature-driven flavor-metabolite remodeling mechanisms and guides targeted drying optimization for customized sweet/roasted Yinchen tea aromas.
Keywords: Artemisia capillaris, Coastal saline-alkali soil, Soil elemental characteristics, Thermal drying, Sensory quality, GC × GC-TOFMS flavoromics, Widely targeted metabolomics
Highlights
-
•
Uniform saline soil for A. capillaris; 50–120 °C heat shifts sweet/floral to roasted.
-
•
GCxGC-TOFMS: 487 volatiles, low heat retains terpenoids; high heat forms Maillard.
-
•
UHPLC-QTOF-MS: 8821 non-volatiles, altering phenylpropanoid/flavonoid pathways.
-
•
High heat reduces astringency and optimizes flavor via flavonoid deglycosylation.
-
•
Metabolite-sensory network reveals flavor, guiding targeted Yinchen tea processing.
1. Introduction
Coastal saline-alkali soil is a globally distributed degraded land ecosystem characterized by high salinity, high pH, and low nutrient availability, which limits conventional crop cultivation but enriches unique salt-tolerant halophyte resources with distinct secondary metabolic traits (Croppenstedt et al., 2018). The Yellow River Delta in Dongying, China, represents a typical coastal saline-alkali wetland that produces abundant wild halophytes with special medicinal and edible values (Xia et al., 2019). As a dominant halophyte in this region, Artemisia capillaris Thunb. has been traditionally used as herbal medicine and functional tea due to its excellent antioxidant, hepatoprotective, and lipid-regulating bioactivities (Wang, Liu, et al., 2025; Wang, Yang, et al., 2025). Different from ordinary farmland-grown A. capillaris, coastal saline-alkali stress acts as a critical abiotic driving variable that profoundly reshapes the adaptive metabolic network of local halophytic A. capillaris. To cope with high salinity, strong alkalinity, and nutrient deficiency stress, saline-grown A. capillaris specifically activates phenylpropanoid and flavonoid biosynthesis pathways, resulting in massive accumulation of stress-adaptive characteristic metabolites including flavonoids, phenolic acids, and terpenoids. Meanwhile, saline-alkali perturbation remodels primary metabolic homeostasis, causing unique differential accumulation of soluble sugars, amino acids, and organic acids. These stress-derived metabolic characteristics form a unique material foundation for flavor formation, endowing saline-alkali-grown A. capillaris with completely different metabolite precursors and flavor potential compared with normal soil-grown individuals. A. capillarisFurthermore, previous studies generally regarded the saline-alkali growth environment as a simple homogeneous background condition and ignored its functional role in shaping the unique metabolic characteristics of halophytic. In addition, most related studies only adopted conventional single-dimensional detection technology, lacking the combination of high-resolution two-dimensional volatile flavoromics and widely targeted non-volatile metabolomics to systematically resolve the coupling relationship between thermal processing temperature, metabolite remodeling and sensory quality differentiation. The lack of in-depth exploration of temperature-driven flavor formation mechanisms severely restricts the precise processing optimization and characteristic flavor industrial development of coastal halophyte tea resources.Saline stress significantly reshapes the secondary metabolism of halophytes, enabling saline-grown A. capillaris to accumulate richer terpenoids, flavonoids, and coumarins than ordinary field-grown samples (Shao et al., 2025; Wang et al., 2019).
Postharvest thermal drying is the most critical unit operation affecting the flavor formation and phytochemical retention of herbal tea (Yu et al., 2025). Processing temperature dominates enzymatic oxidation, thermal degradation, and Maillard browning reactions, thereby determining the final sensory quality and functional properties of herbal products (Karim & Mehdi, 2023). At present, wild Yinchen in the Dongying saline-alkali area is mostly processed by natural sun-drying or arbitrary high-temperature baking, resulting in unstable flavor quality and severe loss of active ingredients. Existing studies on Yinchen mainly focus on single-component extraction, pharmacological verification, or single-temperature processing evaluation, lacking systematic multi-dimensional comparison of sensory and metabolomic changes under continuous thermal gradients (Zhang et al., 2016). Notably, previous studies have ignored the interactive coupling relationship between inherent saline-alkali stress metabolic traits and postharvest thermal processing. Whether and how the unique stress-specific metabolites of saline-alkali-grown A. capillaris respond to gradient drying temperatures remains unclear, which restricts the precise processing optimization and characteristic flavor development of coastal halophyte tea resources.
Comprehensive two-dimensional gas chromatography coupled with time-of-flight mass spectrometry (GC × GC-TOFMS) possesses superior separation capability for complex volatile mixtures and can efficiently distinguish co-eluting trace aroma compounds in herbal matrices, which has become the mainstream technique for high-resolution flavoromics analysis (Drabińska & Jeleń, 2022; Izadmanesh et al., 2017). UHPLC-QTOF-MSwidely targeted metabolomics is a high-precision secondary metabolite quantification technology that integrates the advantages of untargeted global coverage and targeted accurate quantification, enabling systematic identification and quantification of flavor and bioactive secondary metabolites in herbal materials (Zhou et al., 2023). Combined sensory evaluation and dual-omics strategies have been widely applied to clarify the flavor formation mechanism of tea and herbal food materials (Du et al., 2019; Li et al., 2026). A. capillarisA. capillarisAgainst the above research gaps, this study innovatively takes saline-alkali habitat stress as a core functional variable, and adopts a GC × GC-TOFMS and UHPLC-QTOF-MS dual-metabolomics coupled quantitative sensory evaluation strategy to systematically reveal the gradient temperature-dependent flavor-metabolite remodeling mechanism of coastal saline-alkali-grown . This work clarifies the essential difference in processing response characteristics between saline-grown halophytic and ordinary farmland-grown materials, and provides new theoretical insights and targeted industrial processing parameters for the high-value utilization of coastal saline-alkali halophyte resources.
Rhizosphere soil elemental composition determines the basic growth stress background of field-grown halophytes and shapes their basal metabolic characteristics (Shao et al., 2025). To ensure the homogeneity of experimental materials, this study systematically determined the physicochemical properties and elemental background of Dongying coastal saline-alkali soil. All plant materials were collected from a unified saline habitat to eliminate environmental heterogeneity interference. Notably, soil elemental data were only used for habitat baseline characterization and were not included in subsequent quality correlation analysis, ensuring that drying temperature was the sole independent variable of this study .
Research Hypothesis: Different drying temperatures will produce distinct flavor profiles. Low-temperature drying (50 °C) will retain the natural fresh scent of fresh Artemisia capillaris with grassy and herbal notes, medium-temperature drying (100 °C) will generate a sweet aroma, and high-temperature drying (120 °C) will create a roasted aroma. This allows for the selection of appropriate drying processes according to different flavor requirements.
This study set four gradient thermal treatments (fresh, 50 °C, 100 °C, 120 °C) and adopted sensory QDA, GC × GC-TOFMS flavoromics, and UHPLC-QTOF-MS widely targeted metabolomics to: (1) clarify the sensory differentiation rule of Yinchen tea under gradient thermal processing; (2) screen temperature-responsive characteristic volatile markers; (3) reveal the non-volatile metabolic regulation mechanism of flavor and bioactivity changes. The findings aim to provide theoretical support for the standardized low-temperature processing and high-value industrial utilization of coastal saline-alkali halophyte tea resources.
2. Materials and methods
2.1.1. Plant material and saline-alkali soil background characterization
Wild Artemisia capillaris samples were harvested in April 2025 from typical severely saline-alkali coastal farmland in the Dongying Yellow River Delta (37°27′N, 118°33′E). Rhizosphere soil tightly adhered to plant roots was collected from a soil depth of 0–20 cm. The retrieved soil samples were fully homogenized, air-dried at ambient temperature, and sieved to prepare homogeneous specimens for the determination of soil elemental composition and physicochemical properties (Zheng et al., 2018). Notably, soil property detection in this study was solely implemented to confirm consistent saline-alkali growth conditions for all collected plant materials. No subsequent statistical correlation analysis was conducted between soil environmental indices and the quality characteristics of Artemisia samples.
2.1.2. Determination of soil physicochemical properties
Eight key soil physicochemical indices and enzyme activities were determined using national standard and classical analytical methods. Soil pH was measured via potentiometry with a soil-water suspension ratio of 2.5:1 (NY/T 1377–2007; Li, Chen, et al., 2025). Soil organic matter (SOM) content was quantified using the Walkley-Black potassium dichromate oxidation method (Pu et al., 2024). Alkaline hydrolyzable nitrogen (AHN) was determined by the alkaline diffusion method (De Rooij, 2004; Wu, Chang, et al., 2026). Soil available phosphorus (AP) was detected using the Olsen extraction method (Margarida et al., 2015; Wang, Zhang, et al., 2025a). Available potassium (AK) content was measured via ammonium acetate extraction combined with flame photometry (De Rooij, 2004; Zhang, Li, Zhang, & Zhang, 2024).
In terms of soil enzyme activity detection: soil urease activity was determined by sodium phenolate‑sodium hypochlorite colorimetry (Faizullah et al., 2020); soil sucrase activity was measured using the 3,5-dinitrosalicylic acid colorimetric method (Frankenberger & Johanson, 1983); soil catalase activity was detected via ultraviolet absorption spectrophotometry (Aebi, 1984). All samples were analyzed with three technical replicates, and final data were presented as mean ± standard deviation (SD) (Pu et al., 2024).
2.2. Thermal drying and processing treatments
In the present study, fresh Artemisia capillaris foliage was processed via four distinct thermal treatment protocols, with unified pre-treatment procedures applied toall experimental groups to eliminate procedural errors. The standardized pre-processing workflow included manual screening and harvesting of fresh foliage, mechanical cleaning via conveyor belt flushing, and rack-based withering to completely remove surface moisture of plant materials. The detailed treatment parameters for each group are illustrated in Fig. 1 and elaborated below. Notably, the presence or absence of high-temperature fixation was set as a hierarchical processing variable rather than a confounding factor, which was deliberately designed to simulate and distinguish traditional folk simple processing and standardized refined processing modes of Artemisia capillaris herbal tea.
Fig. 1.
Schematic flowchart illustrating four distinct processing protocols for Artemisia capillaris herb, with consistent preliminary operations across all experimental groups. All batches underwent identical unified preprocessing steps: manual picking of fresh A. capillaris foliage → conveyor belt washing → rack withering to eliminate surface moisture. A (CK, control): Only preprocessing steps, no subsequent thermal treatment; raw withered A. capillaris material was the final sample. B (T50): Post-withering constant-temperature drying at 50 °C for 9 h (simulating traditional natural shade-drying folk processing), without fixation and aroma-enhancing procedures. C (T100): Complete thermal workflow: high-temperature fixation high-temperature fixation (370 °C, 5 cycles of 3.5 min per cycle, total duration of 17.5 min) → oven drying at 100 °C → low-temperature aroma enhancement at 100 °C. D (T120): Identical picking, washing, withering, fixation (370 °C, 5 cycles of 3.5 min per cycle, total duration of 17.5 min) and primary drying (100 °C) as T100; the sole variable was elevated aroma-enhancing temperature set to 120 °C.All manufacturing procedures were performed under standardized constant laboratory environments, and three independent biological replicates were established for each treatment to ensure experimental reproducibility.
Group A (Control, CK): Samples only underwent the aforementioned pre-treatment procedures without any thermal processing, and the naturally withered raw materials were reserved as control specimens. Group B (T50): Pre-withered samples were subjected to constant-temperature drying at 50 °C for 9 h, which simulated the traditional natural shade-drying processing method commonly used in folk production without high-temperature fixation and aroma-enhancement treatments, until the sample moisture content was ≤5%. Group C (T100): A complete gradient thermal processing procedure was adopted, consisting of high-temperature fixation at 370 °C (5 cycles, 3.5 min for each cycle, total duration 17.5 min), constant-temperature oven drying at 100 °C, and low-temperature aroma enrichment at 100 °C, until the sample moisture content was ≤5% . Group D (T120): Samples received identical pre-processing, high-temperature fixation (370 °C, 5 cycles, 3.5 min for each cycle, total duration 17.5 min), and primary drying (100 °C) treatments as the T100 group; the only differential parameter was the elevated final aroma-enhancing temperature set at 120 °C, until the sample moisture content was ≤5%. All processing operations were performed in a strictly controlled constant laboratory environment. Three independent biological replicates were set for each treatment group to ensure the repeatability and reliability of experimental results.
Detailed hierarchical processing design principles:The four treatments in this study followed the actual production logic of Artemisia capillaris herbal tea with stepwise and targeted variable settings, without confounding experimental factors. CK and T50 groups simulated traditional folk household processing modes, which only included natural withering and low-temperature drying without fixation. This primitive processing method is widely applied in civilian homemade Yinchen tea, retaining the inherent grassy and bitter taste of raw materials. In contrast, T100 and T120 groups simulated industrial standardized refined processing modes, where high-temperature fixation serves as an indispensable core unit operation. Fixation can inactivate endogenous browning enzymes, alleviate raw bitter off-flavors, and reshape flavor precursor metabolism, which is a necessary procedure for high-quality aromatic Yinchen tea production. The stepwise variable differences among groups were defined as: CK vs T50 (with or without low-temperature drying), T50 vs T100 (with or without professional fixation and thermal refining procedures), and T100 vs T120 (sole difference in final aroma-enhancing temperature). The fixation step is a critical hierarchical variable distinguishing primitive folk processing from standardized refined processing, rather than an interfering confounding factor. It is worth emphasizing that the comparison between T50 and T100 involves multiple processing differences (with/without high-temperature fixation and thermal refining procedures), rather than a single temperature variable. In contrast, the T100 vs. T120 comparison strictly isolates the final aroma-enhancing temperature as the only independent variable, which can independently and accurately characterize the specific regulatory effect of gradient thermal drying temperature on sensory attributes and metabolic profiles of Artemisia capillaris herbal tea.
2.3. Sensory quantitative descriptive analysis
Sensory evaluation was performed in strict accordance with the national standard GB/T 23776–2018 by a professional sensory panel consisting of seven evaluators (3 males and 4 females, aged 25–45 years) with more than five years of professional herbal tea sensory evaluation experience. Before formal evaluation, all panelists received two weeks of unified and standardized professional training, including the definition of nine sensory evaluation indices (aroma intensity, herbal note, grassy note, roasted note, sweet aroma, bitterness, astringency, mellow taste, and overall palatability), unified scoring standards, and standard tea infusion operation specifications. Standard samples with typical grassy, sweet, and roasted flavor characteristics were provided for repeated discrimination training to unify the sensory perception criteria of all panelists. The consistency of panel evaluation was verified by calculating the intra-class correlation coefficient (ICC), and only panel groups with ICC > 0.85 (excellent evaluation consistency) were allowed to carry out formal experiments. During the formal test, random blind sampling and parallel duplicate evaluation were adopted to eliminate individual subjective deviation and ensure the reliability of sensory data. For sensory testing, 3 g of each Artemisia sample was infused with 150 mL of boiling distilled water for 5 min. A 10-point scoring system (0 = undetectable, 10 = extremely prominent) was adopted to quantitatively evaluate nine sensory attributes covering aroma characteristics, taste performance and overall palatability. Each sample was evaluated in triplicate, and the average score of parallel tests was calculated as the final sensory score (Tian et al., 2026). A sensory radar chart was subsequently constructed to intuitively visualize the differences in sensory quality among different treatment groups (Li, Shao, et al., 2025) (Fig. 2).
Fig. 2.
Quantitative descriptive analysis results of CK,T50, T100 and T120. Sensory attribute data were analyzed using one-way ANOVA (α = 0.05); *P < 0.05 indicates a significant difference, and **P < 0.01 indicates an extremely significant difference. Error bars represent the standard error of the mean (SEM).
In terms of ethical compliance, this non-clinical sensory research involving healthy volunteer participants did not require institutional ethical review and approval, in accordance with local institutional research guidelines and relevant regulatory provisions. Prior to the formal sensory evaluation, all panelists were fully informed of the research objectives, experimental procedures, potential negligible risks, and their rights to withdraw from the experiment at any time without reason or adverse consequences. All participants signed written informed consent forms to confirm their voluntary participation and the authorized use of their sensory evaluation data. Throughout the entire research process, standardized protocols were strictly implemented to protect the personal rights, privacy and physical and mental health of all participants, fully complying with the basic ethical principles of human subject research (Chen et al., 2022).
2.4. GC × GC-TOFMS volatile metabolomics analysis
2.4.1. Instrumental detection conditions
An 8890 two-dimensional gas chromatography system coupled with time-of-flight mass spectrometry (GC × GC-TOFMS) was utilized to separate and detect volatile organic compounds (VOCs) adsorbed by CAR/PDMS extraction fibers. The one-dimensional separation column was an HP-5MS capillary column (30 m × 250 μm × 0.25 μm, Agilent Technologies, USA), and the two-dimensional separation column was a DB-17MS capillary column (2.89 m × 180 μm × 0.18 μm, Agilent Technologies, USA) (Chen et al., 2022; Tian et al., 2026). The temperatures of the injection port and transmission line were set to 250 °C and 280 °C, respectively. High-purity helium (99.99%) was used as the carrier gas, with flow rates of 1.0 mL/min and 1.5 mL/min for the first and second-dimensional columns, respectively. A SSM1800 solid-state modulator (J&X Technologies, Shanghai, China) equipped with HV-series modulation columns was installed between the two columns for targeted enrichment, aggregation and desorption of volatile components, with a modulation period of 4.0 s.The column temperature program was set as follows: initial temperature maintained at 40 °C for 1 min, increased to 180 °C at a heating rate of 4 °C/min, further raised to 250 °C at 20 °C/min and held for 1 min, and finally elevated to 350 °C at 20 °C/min with a 0.5 min holding time. The sample split ratio was set to 10:1 (Tian et al., 2026).
A 7250 TOFMS mass spectrometer (Agilent Scientific, CA, USA) was employed to acquire mass spectral information of volatile metabolites (Tian et al., 2026). The mass spectrometer operated in full-scan acquisition mode with a mass detection range of 45–450 amu, a scanning frequency of 50 spectra per second, and a filament current of 5 μA. Electron impact ionization (EI) was adopted with an ion source temperature of 200 °C and electron energy of 70 eV. The interface temperature of the mass spectrometry system was stabilized at 280 °C. A total of 487 volatile compounds were preliminarily identified by matching acquired mass spectra and retention indices with standard databases.
2.4.2. Qualitative identification and relative quantification of volatile active compounds
Original GC × GC-Q-TOFMS data were processed and analyzed using Canvas software (v1.0.0.25117) (Xue et al., 2022). For qualitative identification, a series of n-alkane standards (C7–C28) were used for retention index (RI) calibration. The mass spectrum of each chromatographic peak was matched with three standard libraries (mainlib, replib, nist_ri) in the NIST20 spectral database. Compound identification was strictly based on screening criteria: forward matching score > 750, reverse matching score > 800, and RI deviation <30. Ethyl decanoate (decanoic acid, ethyl ester) was selected as the internal standard with a final concentration of 10 μg/mL in the detection system, which was spiked into all sample extracts prior to instrumental analysis. Ethyl decanoate was chosen according to well-established aroma metabolomics protocols for herbal-tea matrices: it possesses stable chemical properties, is endogenously absent in Artemisia capillaris, and produces no overlapping chromatographic peaks that would interfere with identification and quantification of native volatiles. Moreover, its polarity, boiling point, extraction efficiency, and GC × GC-TOFMS ionization response are well-compatible with major aroma-active classes (terpenoids, esters, alcohols, aldehydes) present in herbal-tea samples. Given that the primary aim of this study was to profile differential volatile patterns and flavor-metabolite responses across thermal-processing treatments rather than to obtain absolute compound concentrations, this single-internal-standard relative-quantification strategy was adopted following mainstream tea-flavoromics practice. Uniform internal-standard calibration guaranteed consistent systematic bias and normalized quantitative baselines across CK, T50, T100, and T120 groups, offsetting deviations originating from sample pretreatment, instrumental drift and batch-to-batch variation. Although extraction and ionization responses differ among diverse volatile chemical classes, this approach reliably reflects relative accumulation differences of the 487 detected volatiles triggered by temperature gradients and satisfies the analytical requirements for screening processing-specific aroma markers and deciphering flavor-formation mechanisms.The relative content of each volatile component was calculated via the internal standard method, and the calculation formula was defined as: C = R × 10 μlL × 10 mg·L-1 M. In the formula, C represents the relative concentration of each volatile compound in Artemisia samples (μg/kg), R refers to the peak area ratio of the target compound to the internal standard, and M is the mass of the tested Artemisia sample (0.5 g). Each sample was detected in three technical replicates to ensure data accuracy (Chen et al., 2022).
2.4.3. Calculation of relative odor activity value
The relative odor activity value (ROAV) of each volatile compound was calculated to evaluate its contribution to the overall aroma of Artemisia samples. Referring to the sensory evaluation standard GB/T 23776–2018, a solid-liquid ratio of 1:50 (3 g sample per 150 mL water) was adopted to simulate the actual sensory infusion environment and calculate the effective concentration of volatile components in the infusion system. Combined with the reported odor threshold values of individual compounds in aqueous solution, the ROAV of each component was determined as the ratio of its actual content in samples to its minimum odor detection threshold (Tian et al., 2026). Notably, the ROAV calculation in this study only evaluates the theoretical aroma contribution potential of individual volatile compounds based on their relative concentration and aqueous odor threshold. The ROAV >1 screening criteria can identify theoretically dominant single aroma-active compounds, but it does not account for perceptual interactions including aroma synergistic effects and masking effects among diverse volatile components in the complex Yinchen tea infusion matrix.
2.5. UHPLC-QTOF-MS widely targeted metabolomics analysis
Secondary metabolite profiles of Artemisia samples were characterized based on a standardized widely targeted metabolomics platform (Zhou et al., 2023). Sample metabolites were extracted using a methanol-water mixed solution (7:3, v/v). The obtained supernatants were subjected to ultra-high performance liquid chromatography coupled with quadrupole time-of-flight mass spectrometry (UHPLC-QTOF-MS) detection under dual ESI positive/negative ionization modes. The widely targeted metabolomics strategy enabled high-precision qualitative identification and relative quantitative analysis of predefined secondary metabolites, covering flavonoids, coumarins, phenolic acids, terpenoids and other flavor-active and bioactive components in herbal materials. Metabolite annotation was performed by cross-comparison with self-built standard substance databases, HMDB and METLIN public databases. Differential metabolites were screened based on rigorous thresholds: variable importance in projection (VIP) > 3.5, p < 0.05, and fold change (FC) > 1.5 or FC < 0.67.
2.6. Statistical analysis
All experimental data in this study were expressed as mean ± standard deviation (SD) of three independent biological replicates. Independent sample t-tests were performed using SPSS 26.0 software to analyze significant differences between experimental groups, with P < 0.05 defined as the threshold for statistically significant differences. Principal component analysis (PCA) and orthogonal partial least squares discriminant analysis (OPLS-DA) were conducted via SIMCA 14.1 software to explore the overall differences in metabolite and aroma profiles among groups, and metabolites with VIP > 1 were identified as differential candidate compounds. All experimental graphs and charts were generated using Origin 2023, Cytoscape 3.9.1, and R software (ggplot2, vegan, igraph packages) (Pu et al., 2024).
3. Results
3.1. Basic physicochemical and elemental characteristics of saline-alkali habitat soil
The rhizosphere soil collected in this study was typical severe coastal saline-alkali soil, which was characterized by high salinity and alkalinity, low organic matter content, and insufficient available nutrients. The basic physicochemical properties and enzyme activities of the tested soil were quantitatively determined as follows: soil pH of 8.3440 ± 0.0073, soil organic matter (SOM) of 0.4993 ± 0.0032 g/kg, alkaline hydrolyzable nitrogen (AHN) of 138.4450 ± 0.7750 mg/kg, available phosphorus (AP) of 28.6400 ± 0.0727 mg/kg, and available potassium (AK) of 491.2041 ± 1.9587 mg/kg. In terms of soil enzyme activities, the urease, sucrase, and catalase activities were 54.1873 ± 0.3231 mg/d/g, 7.9750 ± 0.0350 mg/d/g, and 3.0349 ± 0.0172 μmol/d/g, respectively. The tested soil exhibited obvious sodium accumulation and severe saline-alkali stress, providing a consistent salt-alkali growth background for all wild Artemisia capillaris samples. The uniform soil habitat conditions eliminated heterogeneous environmental interference among different experimental groups, ensuring the reliability and accuracy of subsequent experimental results (Table 1).
Table 1.
Soil physicochemical properties.
| Soil organic matter(SOM)(g/kg) | Available phosphorus (mg/kg) | Available potassium (mg/kg) | pH | Urease (mg/d/g) | Sucrase (mg/d/g) | Catalase(μmol/d/g) | Alkaline hydrolyzable nitrogen (mg/kg) |
|---|---|---|---|---|---|---|---|
| 0.4993 ± 0.0032 | 28.6400 ± 0.0727 | 491.2041 ± 1.9587 | 8.3440 ± 0.0073 | 54.1873 ± 0.3231 | 7.9750 ± 0.0350 | 3.0349 ± 0.0172 | 138.4450 ± 0.7750 |
Note: Values are presented as mean ± standard deviation. Different lowercase letters in the same row indicate significant differences at P < 0.05. OM, organic matter,mean ± SD, n = 3.
3.2. Sensory quality differentiation under gradient thermal processing
The sensory descriptive profiles of Artemisia capillaris samples processed with four different thermal treatments (CK, T50, T100, and T120) are illustrated in the corresponding sensory radar chart. Obvious variations in sensory attributes were identified among different processing groups, mainly reflected in aromatic intensity, flavor richness, taste harmony, and undesirable grassy or bitter notes. Conventional shade-drying (simulated by T50) samples had a strong bitter taste, along with prominent fresh grassy and herbal aromas. High-temperature treatment at 100 °C produced a sweet aroma, and 120 °C treatment resulted in a roasted aroma.
In terms of aroma characteristics, the CK and T50 samples exhibited prominent fresh grassy and herbal aromas, with relatively weak fragrant layering and low aroma concentration. The low-temperature drying treatment (T50) slightly retained the natural fresh scent of fresh Artemisia capillaris but failed to form rich roasted and sweet aromatic substances. In comparison, the high-temperature fixation and aroma-enhanced treatments (T100 and T120) significantly improved the overall aromatic performance of samples. Both T100 and T120 groups showed markedly higher scores for roasted aroma and sweet aroma, accompanied by enhanced aromatic mellowness. Notably, the T120 sample presented the highest score of baked sweet aroma and the weakest green and grassy off-odors, indicating that appropriate high-temperature aroma enhancement effectively eliminated the raw green scent of fresh Artemisia capillaris and optimized the aromatic profile. The sweet aroma of T100 is mainly contributed by compounds such as trans-β-ionone, while the roasted aroma of T120 is mainly due to the formation of pyrazine derivatives and furan compounds through Maillard reactions.
Regarding taste attributes, the CK and T50 samples displayed stronger bitter and astringent taste and higher greenish flavor intensity, resulting in a rough and uncoordinated oral taste. The T100 treatment significantly reduced bitter and astringent attributes and improved taste mellow and smoothness. Further, the T120 sample obtained the highest scores in mellow taste, full-bodied flavor, and taste sweetness, with minimal undesirable bitter and greenish residues, presenting a well-balanced and harmonious taste quality. The reduction in bitterness and astringency in high-temperature treated samples is related to the degradation of certain bitter compounds and the formation of new taste-active substances.
The sensory radar plot intuitively indicates that high-temperature processing treatments (T100 and T120) occupied larger areas in desirable sensory indices including roasted aroma, sweet aroma, mellow taste, and full-bodied flavor, whereas the untreated and low-temperature dried samples (CK and T50) were dominated by undesirable greenish, grassy, and bitter attributes. These results demonstrate that fixation and gradient high-temperature aroma enhancement processing can effectively ameliorate the sensory defects of fresh Artemisia capillaris, eliminate raw green and bitter flavors, and significantly improve the overall sensory quality of processed Artemisia capillaris products. Consumers can choose the appropriate drying process according to their flavor preferences, whether they prefer the sweet aroma of 100 °C drying or the roasted aroma of 120 °C drying.
3.2.1. Qualitative and quantitative analysis of aroma compounds
In the PCA score plot (Fig. 3A), the first three principal components (PC1, PC2, and PC3) accounted for distinct proportions of the total variance, with a high cumulative contribution rate that sufficiently reflected the overall volatile metabolite information of Artemisia capillaris samples. The score plot revealed a clear separation among the four processing groups. The CK and T50 samples clustered closely together, while the T100 and T120 samples formed independent and well-separated clusters, indicating that low-temperature drying treatment exerted a limited effect on the volatile metabolic profile of Artemisia capillaris, whereas high-temperature fixation and aroma enhancement substantially altered its volatile characteristics. Additionally, biological replicates within each group exhibited tight aggregation, demonstrating excellent stability and reproducibility of the processing treatments and detection results.
Fig. 3.
(A) Principal component analysis (PCA) plot; (B) Venn diagram of aroma compounds identified in CK,T50, T100 and T120. (C) Percentage stacked plot of different aroma compound categories in CK,T50, T100 and T120.
A total of 487 differential volatile compounds were identified in Artemisia capillaris samples with different thermal processing treatments via GC–MS detection (Fig. 3B and C), covering multiple chemical classes including alcohols, aldehydes, ketones, esters, heterocyclic compounds, and terpenoids, which jointly constituted the characteristic aroma profiles of processed Artemisia capillaris. The Venn diagram further visualized the overlapping and unique volatile components among the four groups. The CK and T50 groups shared the majority of volatile compounds with highly similar component compositions. In contrast, the T100 and T120 groups possessed abundant unique volatile substances, especially numerous roasted and sweet aroma-related esters and terpenoids generated by high-temperature thermal reactions. These results indicated that alcohols and aldehydes derived from fresh samples were the dominant volatile components in CK and T50 groups, while thermally induced esters and heterocyclic compounds dominated the volatile profiles of high-temperature-processed samples. The sweet aroma of T100 is mainly due to the high content of trans-β-ionone, which has a sweet floral aroma, while the roasted aroma of T120 is mainly contributed by pyrazine derivatives and furan compounds formed through Maillard reactions.
Collectively, these findings demonstrated that different thermal processing strategies significantly modulated the volatile metabolome of Artemisia capillaris. Low-temperature drying barely changed the original volatile characteristics of fresh samples, while high-temperature fixation and gradient aroma enhancement effectively promoted the generation of beneficial aroma compounds and reduced undesirable green and bitter-related volatiles, thereby fundamentally differentiating the volatile metabolic fingerprints and finally forming distinct sensory quality characteristics of processed Artemisia capillaris products. The formation of sweet aroma in T100 and roasted aroma in T120 is closely related to the specific volatile compounds generated under different temperature conditions, which provides a theoretical basis for the selection of appropriate processing techniques according to different flavor requirements.
3.2.2. Analysis of VIP >1 aroma compounds in CK,T50, T100 and T120
3.2.2.1. Multivariate statistical discrimination of volatile profiles
PCA (Fig. 4A) was first applied to visualize the overall separation of volatile profiles among Artemisia capillaris samples with different thermal processing treatments (CK, T50, T100, and T120). The score plot showed a clear separation among the four treatment groups, with all biological replicates clustering closely within their respective groups and exhibiting negligible intra-group variation. The first two principal components captured the majority of the cumulative variance, indicating that different thermal processing strategies induced fundamentally distinct volatile metabolic fingerprints in processed Artemisia capillaris samples.
Fig. 4.
Statistical analysis of VIP>1 volatile compounds and the schematic diagram of sample preparation; OPLS-DA analysis of potential VIP aroma compounds; Heat map,ven map of potential VIP aroma compounds and between quantitative descriptive analysis and aroma-active compounds (VIP > 1) in CK,T50, T100 and T120. (A) OPLS-DA result. (B) HCA result. R2X[1] and R2X[2] respectively represent the cumulative proportions of variance explained by the first and second principal components;(C),(E): Permutation test and VIP value in CK,T50, T100 and T120. (D),(F),(G): Heat map,Ven map of potential VIP aroma compounds,between quantitative descriptive analysis and aroma-active compounds (VIP > 1) in CK,T50, T100 and T120.
Consistent with the PCA results, HCA (Fig. 4B) revealed that samples from the same treatment formed independent and distinct clusters. Specifically, CK and T50 samples were grouped into one large branch, while T100 and T120 samples were classified into separate individual branches, with obvious inter-group spatial distances. This clustering pattern confirmed that low-temperature drying induced minor changes in volatile profiles, whereas high-temperature fixation and aroma enhancement generated highly group-specific volatile signatures, with excellent sample repeatability for each treatment.
To further identify discriminatory volatile aroma compounds (VACs) responding to thermal processing, OPLS-DA was performed on the volatile metabolite datasets. The OPLS-DA score plot (Fig. 4C) showed complete separation among all four processing groups, and the permutation test (Fig. 4E) validated the robustness and high predictive power of the established model, with no signs of overfitting. Based on the variable importance in projection (VIP) values, compounds with VIP > 1 were screened and defined as the key discriminatory VACs responsible for volatile profile differences among treatments.
A Venn diagram (Fig. 4D) was constructed to illustrate the overlap and specificity of these key discriminatory compounds across the four groups. A core set of common VACs was shared by all samples, representing the inherent conserved volatile components of fresh Artemisia capillaris. In addition, each processing group possessed unique discriminatory VACs: CK and T50 contained characteristic compounds related to fresh grassy and herbal notes, while T100 and T120 harbored abundant exclusive VACs associated with roasted and sweet aromas. This indicated that thermal processing altered both the composition and specificity of volatile metabolites, and high-temperature treatment significantly enriched the diversity of aroma-active compounds in Artemisia capillaris.
A hierarchical clustering heatmap (Fig. 4F) was generated to visualize the relative abundance of these VIP > 1 compounds across all samples. The heatmap clearly presented four distinct accumulation patterns of discriminatory VACs. A large number of fresh herbal and green-related compounds were highly accumulated in CK.
and T50 samples, while thermally induced esters, heterocyclic compounds and terpenoids were significantly enriched in T100 and T120 samples, especially in the T120 group. The sample clustering results were highly consistent with PCA and HCA classification, further confirming the reliability of treatment grouping and the stability of volatile metabolic characteristics in each group.
A correlation network (Fig. 4G) was constructed to explore the interactive relationships between these key VACs. The network revealed extensive positive and negative correlations among different volatile metabolites. The fresh and green-related compounds showed synergistic accumulation in CK and T50 groups, while the sweet and roasted aroma-related compounds exhibited positive correlation and accumulated massively in high-temperature processed groups. Meanwhile, significant antagonistic relationships were observed between green off-flavor compounds and thermally generated aromatic compounds, suggesting that metabolite interaction patterns jointly reshaped the overall aroma characteristics of processed Artemisia capillaris.
The key discriminatory VACs screened via OPLS-DA (VIP > 1) were further correlated with the sensory evaluation results. The differential accumulation patterns of volatile compounds shown in the heatmap and correlation network were highly consistent with the sensory quality variations among treatments. The green and bitter-related VACs enriched in CK and T50 corresponded to undesirable grassy taste and rough oral sensation, while the thermally induced characteristic VACs in T100 and T120 were closely associated with improved sweet aroma, roasted flavor and mellow taste. These results demonstrated that the group-specific accumulation of VIP > 1 volatile compounds driven by gradient thermal processing was the core metabolic basis for the significant differences in sensory quality of Artemisia capillaris products.
3.3. Analysis of key aroma compounds in CK,T50, T100 and T120
Key aroma compounds with relative odor activity value (ROAV) > 1 were identified and quantified in Artemisia capillaris samples with different thermal processing treatments (CK, T50, T100, and T120) (Table 2), and their contributions to differential sensory profiles were systematically analyzed. In total, 59 volatile compounds were confirmed with ROAV >1 across the four treatments, covering aldehydes, terpenes, alcohols, ketones, furans, and pyrazines. The distinct concentrations, odor thresholds, and accumulation patterns of these key aroma-active compounds fundamentally differentiated the aroma quality of Artemisia capillaris products processed by different thermal regimes.
Table 2.
Aroma compounds with an ROAV >1 in CK,T50, T100 and T120.
| NO | Aromasubstances | CAS | Odordescription | Threshold |
Concentration (mg/kg) |
ROAV |
||||||
|---|---|---|---|---|---|---|---|---|---|---|---|---|
| CK | T50 | T100 | T120 | CK | T50 | T100 | T120 | |||||
| 1 | Dimethyl sulfide | 75–18-3 | Cabbage, Organic, Sulfur, Wet Earth | 0.3 | 133.3801854 | 1376.559019 | 5157.371654 | 480.59071 | 444.6006178 | 4588.530062 | 17,191.23885 | 1601.969033 |
| 2 | β-Ocimene | 3338-55-4 | Floral | 6.7 | 1757.717373 | 0 | 0 | 0 | 262.3458766 | 0 | 0 | 0 |
| 3 | p-Cymene | 99–87-6 | Citrus, Fresh, Solvent | 5.01 | 1143.797688 | 0 | 42,340.20283 | 4366.673926 | 228.3029317 | 0 | 8451.138289 | 871.5916019 |
| 4 | β-Myrcene | 123–35-3 | Balsamic, Fruit, Geranium, Herb, Must | 1.2 | 195.0422972 | 0 | 16,254.93798 | 1457.566542 | 162.5352477 | 0 | 13,545.78165 | 1214.638785 |
| 5 | Benzeneacetaldehyde | 122–78-1 | Berry, Geranium, Honey, Nut, Pungent | 5.2 | 481.4095304 | 3720.279436 | 37,618.62753 | 3846.826991 | 92.57875585 | 715.438353 | 7234.351448 | 739.7744213 |
| 6 | Nonanoic acid | 112–05-0 | moldy, pungent | 1.5 | 135.9486777 | 6596.347933 | 4746.699001 | 494.2823265 | 90.63245177 | 4397.565289 | 3164.466001 | 329.521551 |
| 7 | Hexanal | 66–25-1 | green, grassy | 5 | 240.4534121 | 895.0695612 | 12,514.07832 | 846.5974337 | 48.09068242 | 179.0139122 | 2502.815664 | 169.3194867 |
| 8 | Eugenol | 97–53-0 | Burnt, Clove, Spice | 1.8 | 66.40744906 | 751.1383798 | 3710.114307 | 0 | 36.89302725 | 417.2990999 | 2061.174615 | 0 |
| 9 | γ-Terpinene | 99–85-4 | Bitter, Citrus | 65 | 1601.114191 | 0 | 27,152.33419 | 3469.395671 | 24.63252602 | 0 | 417.7282184 | 53.37531802 |
| 10 | Caryophyllene | 87–44-5 | Fried, Spice, Wood | 64 | 1040.291722 | 7854.300677 | 109,521.2222 | 25,243.22424 | 16.25455816 | 122.7234481 | 1711.269097 | 394.4253787 |
| 11 | 1-Octen-3-ol | 3391-86-4 | Cucumber, Earth, Fat, Floral, Mushroom | 1.5 | 24.08928034 | 5669.003753 | 7269.535677 | 600.3251264 | 16.05952023 | 3779.335835 | 4846.357118 | 400.216751 |
| 12 | Furan, 2-ethyl- | 3208-16-0 | coffee, caramel, and roasted nut sweetness | 2.3 | 23.77928833 | 283.3348834 | 3873.567464 | 332.5821489 | 10.33882101 | 123.1890797 | 1684.159767 | 144.6009343 |
| 13 | 1-Cyclohexene-1-carboxaldehyde, 2,6,6-trimethyl- | 432–25-7 | hay, tobacco, sweet woody notes, slight citrus fruit aroma | 3 | 25.44084542 | 17.72224749 | 16,624.11113 | 1140.049807 | 8.480281806 | 5.90741583 | 5541.370376 | 380.0166022 |
| 14 | cis-β-Farnesene | 0 | Boiled Vegetable, Floral, Wood | 87 | 673.4016379 | 905.762918 | 0 | 0 | 7.740248711 | 10.41106802 | 0 | 0 |
| 15 | Pentanal | 110–62-3 | green, fatty, moldy | 12 | 71.07659264 | 3828.591103 | 3533.694141 | 235.2195284 | 5.923049386 | 319.0492586 | 294.4745117 | 19.60162736 |
| 16 | Heptanal | 111–71-7 | Citrus, Fat, Green, Nut | 2.8 | 12.38748301 | 159.2115435 | 2908.933108 | 298.4163439 | 4.424101074 | 56.86126552 | 1038.904682 | 106.5772657 |
| 17 | Humulene | 6753-98-6 | balsamic vinegar-like | 160 | 706.8541639 | 5874.47374 | 45,602.43767 | 12,220.17579 | 4.417838525 | 36.71546088 | 285.0152354 | 76.3760987 |
| 18 | Octanal | 124–13-0 | Citrus, Fat, Green, Oil, Pungent | 3.4 | 14.68117179 | 181.3793439 | 5011.507713 | 694.2182925 | 4.317991704 | 53.34686586 | 1473.972857 | 204.1818507 |
| 19 | Furan, 2-pentyl- | 3777-69-3 | Butter, Floral, Fruit, Green Bean | 5.8 | 21.26854716 | 0 | 8022.684191 | 1060.555012 | 3.66699089 | 0 | 1383.221412 | 182.8543125 |
| 20 | 2-Octenal, (E)- | 2548-87-0 | Dandelion, Fat, Fruit, Grass, Green, Spice | 3 | 10.64812015 | 94.45186492 | 5116.015617 | 381.3146612 | 3.549373383 | 31.48395497 | 1705.338539 | 127.1048871 |
| 21 | 2,4-Heptadienal, (E,E)- | 4313-03-5 | Fat, Nut | 15.4 | 34.31269735 | 37.63251878 | 5086.079851 | 240.7284875 | 2.22809723 | 2.443670051 | 330.2649254 | 15.63171997 |
| 22 | Decanal | 112–31-2 | Floral, Fried, Orange Peel, Penetrating, Tallow | 9.3 | 20.09492924 | 145.8111145 | 3870.562108 | 446.1361141 | 2.160745079 | 15.67861446 | 416.189474 | 47.97162517 |
| 23 | Benzaldehyde | 100–52-7 | Bitter Almond, Burnt Sugar, Cherry, Malt, Roasted Pepper | 150 | 208.2681383 | 2297.064153 | 30,055.44085 | 3515.929496 | 1.388454255 | 15.31376102 | 200.3696057 | 23.43952997 |
| 24 | Germacrene D | 23,986–74-5 | Floral, soapy, green | 100 | 105.0117066 | 0 | 45,079.00269 | 2007.684994 | 1.050117066 | 0 | 450.7900269 | 20.07684994 |
| 25 | trans-β-Ionone | 79–77-6 | Floral, Violet | 0.021 | 0 | 530.2669445 | 14,820.22606 | 1082.818033 | 0 | 25,250.80688 | 705,725.0504 | 51,562.76349 |
| 26 | Nonanal | 124–19-6 | Fat, Floral, Green, Lemon | 2.8 | 0 | 2198.958124 | 25,510.0889 | 2923.165636 | 0 | 785.3421872 | 9110.746035 | 1043.987727 |
| 27 | 3-Octanone | 106–68-3 | Harbal | 21.4 | 0 | 4481.36423 | 615.9957384 | 0 | 0 | 209.4095435 | 28.78484759 | 0 |
| 28 | Styrene | 100–42-5 | Floral | 3.6 | 1.910355029 | 483.4625709 | 796.0585542 | 142.116804 | 0.530654175 | 134.2951586 | 221.1273762 | 39.47689 |
| 29 | 1,3-Cyclohexadiene-1-carboxaldehyde, 2,6,6-trimethyl- | 116–26-7 | Rich saffron sweet aroma, hay, tobacco, woody scent, mild medicinal powder aroma | 0.3 | 0 | 5.22311321 | 3115.368602 | 408.6337315 | 0 | 17.41037737 | 10,384.56201 | 1362.112438 |
| 30 | Methyleugenol | 93–15-2 | 0 | 68 | 0 | 728.2771456 | 0 | 0 | 0 | 10.70995802 | 0 | 0 |
| 31 | Phenylethyl Alcohol | 60–12-8 | Fruit, Honey, Lilac, Rose, Wine | 140 | 32.48481682 | 1477.401717 | 4823.486334 | 592.7909505 | 0.232034406 | 10.55286941 | 34.45347382 | 4.234221075 |
| 32 | 1-Butanol, 2-methyl-, (S)- | 1565-80-6 | malt, whisky, fermented wine aroma, light nutty flavor | 300 | 0 | 3147.362596 | 0 | 0 | 0 | 10.49120865 | 0 | 0 |
| 33 | endo-Borneol | 507–70-0 | Camphor, pine wood, cool herbal fragrance | 140 | 0 | 1464.912405 | 0 | 0 | 0 | 10.46366004 | 0 | 0 |
| 34 | 3-Hexen-1-ol | 544–12-7 | Grass, Green Fruit, Green Leaf, Herb, Unripe Banana | 70 | 46.03087667 | 666.9365317 | 1172.013799 | 102.8139236 | 0.657583952 | 9.527664739 | 16.74305428 | 1.468770337 |
| 35 | Caryophyllene oxide | 1139-30-6 | citrus-like, soapy | 400 | 48.61949546 | 3739.102116 | 8172.821872 | 2973.094111 | 0.121548739 | 9.347755289 | 20.43205468 | 7.432735277 |
| 36 | Pyridine | 110–86-1 | Pungent fishy smell, putrid odor, nauseating and irritating smell | 79 | 0.278483226 | 643.359596 | 58.87657528 | 20.07700776 | 0.003525104 | 8.143792354 | 0.745273105 | 0.254139339 |
| 2,4-Di-tert-butylphenol | 96–76-4 | phenolic-like, leather-like | 500 | 0 | 1799.346145 | 3368.634719 | 195.9476764 | 0 | 3.59869229 | 6.737269437 | 0.391895353 | |
| 37 | 1-Pentanol | 71–41-0 | Balsamic, Fruit, Green, Pungent, Yeast | 150.2 | 9.545639504 | 538.3509439 | 2538.772693 | 224.8612398 | 0.06355286 | 3.584227323 | 16.90261447 | 1.497078827 |
| 38 | Cyclohexanol | 108–93-0 | Schnabel K, Belitz H, v. Ranson C (1988) Z Lebensm Unters Forsch 187: 215–223, Untersuchungen zur Struktur-Aktivitäts-Beziehung bei Geruchsstoffen | 470 | 0 | 1121.530746 | 0 | 0 | 0 | 2.38623563 | 0 | 0 |
| 39 | 1-Octanol | 111–87-5 | Bitter Almond, Burnt Matches, Fat, Floral | 400 | 13.95553219 | 800.671815 | 1288.448114 | 126.925728 | 0.03488883 | 2.001679538 | 3.221120284 | 0.31731432 |
| 40 | 2-Hexene, 3,5,5-trimethyl- | 26,456–76-8 | faint waxy, hydrocarbon oil and gas | 1600 | 8.036525942 | 2858.625397 | 2594.32722 | 162.5335686 | 0.005022829 | 1.786640873 | 1.621454512 | 0.10158348 |
| 41 | 1-Hexanol | 111–27-3 | Banana, Flower, Grass, Herb | 590 | 73.71513811 | 661.8774662 | 2347.826128 | 268.4767496 | 0.124940912 | 1.121826214 | 3.979366318 | 0.455045338 |
| 42 | 4-Heptenal, (Z)- | 6728-31-0 | Dairy | 0.0087 | 0 | 0 | 26,481.08798 | 2989.914747 | 0 | 0 | 3,043,803.216 | 343,668.3618 |
| 43 | Butanal, 3-methyl- | 590–86-3 | malt, cocoa, nuts, fermented fruit aroma | 0.6 | 0 | 0 | 5922.626708 | 645.2086301 | 0 | 0 | 9871.044513 | 1075.347717 |
| 44 | 1,3,6-Octatriene, 3,7-dimethyl-, (Z)- | 3338-55-4 | Sweet floral aroma, green grass aroma, light citrus herbal aroma | 34 | 0 | 0 | 20,706.90043 | 0 | 0 | 0 | 609.0264834 | 0 |
| 45 | (E)-β-Farnesene | 18,794–84-8 | Spice, herb, fresh green, swee | 120 | 0 | 0 | 29,821.78889 | 5041.466401 | 0 | 0 | 248.5149074 | 42.01222001 |
| 46 | 5-Hepten-2-one, 6-methyl- | 110–93-0 | Citrus, Mushroom, Pepper, Rubber, Strawberry | 68 | 0 | 0 | 10,216.07707 | 1169.868628 | 0 | 0 | 150.2364275 | 17.20395041 |
| 47 | d-Limonene | 5989-27-5 | Citrus, Mint | 13 | 0 | 0 | 1526.434295 | 2211.213349 | 0 | 0 | 117.4180227 | 170.0933346 |
| 48 | γ-Muurolene | 30,021–74-0 | dry woody aroma, light herbal notes, slightly spicy fatty fragrance | 180 | 45.64908073 | 0 | 19,057.06777 | 636.0258477 | 0.253606004 | 0 | 105.8725987 | 3.533476931 |
| 49 | Bicyclo[3.1.1]heptane, 6,6-dimethyl-2-methylene-, (1S)- | 18,172–67-3 | Fresh turpentine, resin, cool pine spicy fragrance | 140 | 116.1020421 | 13.88638052 | 6007.575323 | 738.8801051 | 0.829300301 | 0.099188432 | 42.91125231 | 5.277715036 |
| 50 | 9-Oxabicyclo[6.1.0]nonane | 286–62-4 | odorless | 600 | 549.5544682 | 0 | 24,791.88645 | 4974.235096 | 0.915924114 | 0 | 41.31981074 | 8.290391826 |
| 51 | Benzene, 1-(1,5-dimethyl-4-hexenyl)-4-methyl- | 644–30-4 | mild woody aroma, light herbal scent, slightly sweet and spicy fragrance | 550 | 52.95487989 | 234.5228656 | 21,577.20293 | 1451.000988 | 0.0962816 | 0.42640521 | 39.23127806 | 2.638183615 |
| 52 | 3,5-Octadien-2-one, (E,E)- | 30,086–02-3 | Green | 100 | 11.72510682 | 4.3297613 | 3543.255272 | 423.3151186 | 0.117251068 | 0.043297613 | 35.43255272 | 4.233151186 |
| 53 | Cetene | 629–73-2 | The roasted tobacco leaves have a mild, smoky aroma. | 8000 | 284.2612129 | 5321.15033 | 34,274.5004 | 0 | 0.035532652 | 0.665143791 | 4.28431255 | 0 |
| 54 | Butanoic acid, 2-methyl- | 0 | fermented putridity, cheese, sweat | 5800 | 0 | 0 | 10,217.48855 | 0 | 0 | 0 | 1.761635958 | 0 |
| 55 | Benzyl alcohol | 100–51-6 | Boiled Cherries, Moss, Roasted Bread, Rose | 2546.21 | 17.24294688 | 607.3096662 | 3412.303627 | 388.0535625 | 0.006772005 | 0.238515152 | 1.340150116 | 0.152404382 |
| 56 | Hexanoic acid | 142–62-1 | rancid, body odor, cheesy greasy putrid acid smell | 3000 | 10.08438198 | 248.4315281 | 3324.208498 | 433.0920367 | 0.003361461 | 0.082810509 | 1.108069499 | 0.144364012 |
| 57 | Pyrazine, 3-ethyl-2,5-dimethyl- | 13,360–65-1 | Roasted peanuts, cocoa, caramel, potato nutty aroma | 1 | 0 | 0 | 0 | 922.1732908 | 0 | 0 | 0 | 922.1732908 |
| 58 | 3-Octen-2-one, (E)- | 1669-44-9 | mushroom, earth, hay, sweet herbal oil aroma | 50 | 0 | 0 | 0 | 796.9376452 | 0 | 0 | 0 | 15.9387529 |
| 59 | Pyrazine, 2-ethyl-5-methyl- | 13,925–05-8 | roasted peanuts, nuts, cocoa, burnt aroma | 100 | 0 | 0 | 0 | 1356.255823 | 0 | 0 | 0 | 13.56255823 |
Note: Odor description referred to the database (The Good Scents Company Information System, http://www.thegoodscentscompany.com/).
In the untreated control group (CK), the key aroma-active compounds were mainly composed of grassy and green-feeling volatiles. Representative green-related compounds including hexanal (48.09 mg/kg, ROAV >1), pentanal (5.92 mg/kg), and 3-hexen-1-ol (0.66 mg/kg) dominated the aroma profile of CK samples, contributing typical green, herbal, and fresh grass notes. Terpene substances such as β-ocimene (262.35 mg/kg) and γ-terpinene (24.63 mg/kg) were also uniquely detected in CK samples. These inherent volatile substances retained the original fresh plant odor of raw Artemisia capillaris, whereas almost no sweet, roasted, or floral thermal-derived aroma compounds were enriched in CK samples, resulting in a single and pure fresh green aroma characteristic.
For the low-temperature drying group (T50), the composition of key ROAV >1 compounds was generally consistent with CK, but significant concentration variations were observed. T50 samples exhibited dramatic accumulation of fatty and herbal volatile components, with nonanal reaching 785.34 mg/kg, benzeneacetaldehyde up to 715.44 mg/kg, and 1-octen-3-ol as high as 3779.34 mg/kg, all of which were significantly higher than those in CK, effectively enhancing the herbal and fatty aroma intensity. Nevertheless, T50 still lacked typical thermally induced aroma components such as pyrazines and high-content ionone substances, and green and grassy off-notes derived from residual hexanal (179.01 mg/kg) and pentanal (319.05 mg/kg) remained the dominant sensory characteristics, indicating that low-temperature simulated shade-drying could barely promote the formation of beneficial aroma compounds and failed to effectively eliminate the inherent green and bitter sensory defects of fresh Artemisia capillaris.
In contrast, the high-temperature processed groups (T100 and T120) exhibited completely different key aroma compound profiles with abundant thermally generated volatile substances. The T100 group possessed the highest concentrations of core aroma contributors among all treatments, including trans-β-ionone (705,725.05 mg/kg), 2,6,6-trimethyl-1,3-cyclohexene-1-carboxaldehyde (5541.37 mg/kg), caryophyllene (1711.27 mg/kg), and (Z)-4-heptenal (3,043,803.22 mg/kg). These high-abundance compounds provided rich sweet saffron, floral, woody, roasted, and creamy notes, which greatly enriched the aroma layering of Artemisia capillaris samples. Notably, two unique pyrazine compounds, 3-ethyl-2,5-dimethyl-pyrazine (922.17 mg/kg) and 2-ethyl-5-methyl-pyrazine (1356.26 mg/kg), were specifically and exclusively detected in the T120 group, contributing typical roasted nut and caramel sweet aromas. Meanwhile, T120 retained appropriate contents of floral and fruity volatiles while significantly reducing the concentration of green-related substances such as hexanal (169.32 mg/kg) and pentanal (19.60 mg/kg), forming a balanced aroma profile dominated by roasted sweet and floral notes.
Several typical green-related volatile compounds exhibited significantly higher ROAV values and concentrations in CK and T50 groups than in T100 and T120 groups. Specifically, hexanal and pentanal concentrations in T50 were 3.7-fold and 54-fold higher than those in T120, respectively, while (E)-2-octenal in T50 (31.48 mg/kg) was 2.5 times higher than that in T120. These green-feeling aldehydes were the main substances responsible for the raw green and bitter sensory defects of untreated and low-temperature dried samples. In comparison, thermal reaction-derived compounds showed sharply increased accumulation in high-temperature treatment groups: the content of trans-β-ionone in T100 was nearly 30 times higher than that in T50, and furan compounds and caryophyllene in T100 were 13.7-fold and 14-fold higher than those in CK, respectively, especially in T120. These results indicated that gradient high-temperature fixation and aroma enhancement triggered substantial thermal degradation, oxidation, and cyclization reactions, realizing the effective transformation from undesirable green-grass odor to pleasant roasted and sweet floral aroma.
Shared key volatiles such as nonanal and benzaldehyde provided a basic floral and green odor background for all Artemisia capillaris samples. However, the group-specific accumulation of ROAV >1 discriminatory compounds dominated the unique sensory characteristics of each treatment. The CK and T50 samples were defined by inherent plant-derived green and herbal volatiles with high contents of green aldehydes and grass alcohols, while T100 and T120 samples were characterized by thermally generated sweet, roasted, and floral aroma substances including ionones, pyrazines, and heterocyclic compounds. These results demonstrated that thermal processing modes and temperatures directly regulated the accumulation of key aroma-active compounds, and the differential ROAV distribution and concentration disparity of volatile metabolites were the essential material basis for the significant sensory quality differences among processed Artemisia capillaris samples, which was highly consistent with the aforementioned sensory evaluation and volatile metabolomic profiling results.he concentration differences of ROAV>1 volatiles directly determined the distinct aroma characteristics between CK,T50, T100 and T120, consistent with their sensory evaluation results.
3.4. Widely targeted metabolomics profiling of non-volatile metabolites
3.4.1. Comprehensive widely targeted metabolomics analysis of non-volatile metabolites across four experimental groups (CK, T50, T100, T120)
Widely targeted metabolomics was further performed to comprehensively characterize the non-volatile metabolic profiles and differential metabolite accumulation of Artemisia capillaris samples under four thermal processing treatments (CK, T50, T100, T120) (Fig.5). Systematic multivariate statistical analyses, including principal component analysis (PCA), hierarchical clustering analysis (HCA), metabolite composition analysis, differential metabolite screening, and Kyoto Encyclopedia of Genes and Genomes (KEGG) functional enrichment analysis, were conducted to elucidate the metabolic remodeling mechanism induced by gradient thermal processing. The results demonstrated that gradient temperature treatments dramatically reshaped the non-volatile metabolic landscape of A. capillaris, resulting in distinct accumulation patterns of primary and secondary metabolites, which fundamentally mediated the differential flavor and taste formation of processed samples. Low-temperature short-time treatment (T50) retained the inherent metabolic characteristics of raw samples, while medium-temperature (T100) and high-temperature (T120) processing induced drastic metabolic variations, facilitating the accumulation of flavor-active secondary metabolites and the reduction of undesirable taste-related primary metabolites.
Fig. 5.
Comprehensive widely targeted metabolomics analysis of non-volatile metabolites across four experimental groups (CK, T50, T100, T120).(A) PCA score plot showing global metabolic separation of four groups; PC1 and PC2 explained 85.98% of total metabolic variation.(B) Venn diagram illustrating the shared and unique total non-volatile metabolites detected in all groups.(C) Hierarchical clustering heatmap of all differential metabolites across biological replicates, with red indicating high relative abundance and blue low abundance.(D) Stacked column chart displaying the relative composition and total abundance of major non-volatile metabolite classes in each group.(E) Venn diagram and bar plot summarizing overlapping and unique differential metabolites from pairwise comparisons (T50 vs CK, T100 vs CK, T120 vs CK), as well as the total number of differential metabolites per comparison.(F) Permutation testing and statistical significance of all detected metabolites; red dots represent significantly upregulated metabolites, blue dots significantly downregulated metabolites.(G) Abundance distribution curve of all detected non-volatile metabolites ranked by expression level.(H) Quantitative bar chart of orientin (a representative differential flavonoid metabolite) across groups, asterisks denote significant intergroup differences.(I) KEGG topology enrichment bubble plot of differential metabolites, highlighting key perturbed metabolic pathways based on enrichment significance and pathway impact value.(J) Functional clustering heatmap of differential metabolites annotated to corresponding KEGG metabolic pathways, revealing coordinated expression patterns of pathway-related metabolites. (For interpretation of the references to colour in this figure legend, the reader is referred to the web version of this article.)
The PCA score plot was applied to evaluate the global metabolic differences and sample reproducibility across all treatment groups (Fig. 5A). The first two principal components, PC1 and PC2, collectively explained 85.98% of the total metabolic variation, indicating that the PCA model could well reflect the overall metabolic differences of samples. All biological replicates within each group exhibited tight clustering with negligible deviation, verifying excellent repeatability and reliability of the metabolomic dataset. Obvious spatial separation was observed among the four groups: the CK and T50 groups clustered closely with highly overlapping metabolic fingerprints, whereas T100 and T120 groups were completely separated from the low-temperature and untreated groups. These results indicated that low-temperature treatment exerted a minor modulatory effect on the non-volatile metabolome of A. capillaris, while medium- and high-temperature thermal processing triggered profound and specific metabolic perturbations, which constituted the core metabolic basis for the differential flavor formation of processed A. capillaris.
The Venn diagram of total detected non-volatile metabolites was constructed to clarify the shared and group-specific metabolites across different thermal processing conditions (Fig. 5B). A certain number of core metabolites were commonly detected in all groups, which represented the constitutive basic metabolic components of A. capillaris and maintained the fundamental metabolic homeostasis of samples. In comparison, unique metabolites were specifically identified in each treatment group. The CK and T50 groups possessed abundant unique primary metabolites closely associated with herbal and grassy tastes as well as bitter astringency, while the T100 and T120 groups accumulated a large number of temperature-specific secondary metabolites. The distinct distribution of shared and exclusive metabolites further confirmed that thermal temperature gradient was the decisive factor driving the specific accumulation of non-volatile metabolites in A. capillaris.
Hierarchical clustering heatmap of all detected metabolites was generated to visualize the global accumulation patterns of non-volatile metabolites across individual samples (Fig. 5C). Consistent with the PCA results, all samples were clearly divided into two major clusters. The first cluster covered CK and T50 samples with highly homologous metabolite accumulation profiles, characterized by high abundance of primary metabolites. The second cluster consisted of T100 and T120 samples, which exhibited consistent and unique secondary metabolite enrichment patterns. The intra-group sample consistency and inter-group metabolic divergence fully validated the significant metabolic differentiation induced by medium and high-temperature processing, and further explained the sensory differences between low-temperature treated raw-like samples and high-temperature processed flavor-optimized samples.
Stacked column analysis of metabolite classes was performed to quantitatively compare the compositional variations of major non-volatile metabolite categories among different groups (Fig. 5D). The metabolic components of A. capillaris mainly included organic acids, amino acids, sugars, lipids, phenolic acids, flavonoids, and other secondary metabolites. The CK and T50 groups were dominated by primary metabolites such as organic acids and free amino acids, which were the main contributors to the bitter, astringent and green grassy off-flavors of raw A. capillaris. With the increase of processing temperature, the relative proportion of primary metabolites decreased significantly, while the proportion of flavonoids, phenolic acids and other functional secondary metabolites increased remarkably in T100 and T120 groups. The variation of metabolite category composition revealed that gradient thermal processing effectively optimized the metabolic component structure of A. capillaris, eliminating undesirable flavor substances and enriching flavor-preferential secondary metabolites.
To further explore the dynamic variation of differential metabolites during thermal processing, pairwise comparative analysis (T50 vs CK, T100 vs CK, T120 vs CK) was performed, and the quantity and overlap of differential metabolites were summarized via Venn diagram and statistical bar plot (Fig. 5E). The results showed that the number of differential metabolites increased gradually with the elevation of processing temperature, indicating that higher temperature induced more intense metabolic remodeling in A. capillaris. A small number of differential metabolites were shared in all comparison groups, which represented the universal metabolic response of A. capillaris to thermal stimulation. Meanwhile, each temperature group possessed a large number of unique differential metabolites, suggesting that different thermal processing regimes regulated specific metabolic pathways and induced exclusive metabolite accumulation, thereby forming distinct flavor characteristics.
Permutation testing and statistical significance of all detected metabolites(Fig. 5F). Metabolites with remarkable expression differences were intuitively screened, with red dots representing significantly upregulated metabolites and blue dots representing significantly downregulated metabolites. The results showed that low-temperature treatment (T50) caused only limited metabolite differential expression, with few significantly changed metabolites. In contrast, medium and high-temperature processing induced extensive metabolite differential regulation: a large number of secondary metabolites were significantly upregulated in T100 and T120 groups, whereas most taste-defective primary metabolites were markedly downregulated. These differentially expressed metabolites acted as key molecular markers for thermal processing-induced flavor optimization of A. capillaris.
The overall abundance distribution of all detected non-volatile metabolites was analyzed to characterize the metabolic expression profile of A. capillaris under thermal processing (Fig. 5G). The abundance distribution curve showed that most metabolites were stably expressed at low and medium levels, while a small number of core functional metabolites presented high relative abundance. Thermal processing did not alter the overall abundance distribution pattern of global metabolites, but specifically regulated the expression of key flavor-related metabolites. The targeted enrichment of high-value secondary metabolites and the reduction of undesirable primary metabolites in high-temperature groups further illustrated the targeted metabolic modification effect of gradient thermal processing on A. capillaris.
Orientin, a core flavor-active and functional flavonoid metabolite in A. capillaris, was selected as the representative differential metabolite for quantitative verification (Fig. 5H). The quantitative results showed that the relative abundance of orientin had no significant difference between CK and T50 groups, while it was significantly upregulated in T100 group and reached the highest level in T120 group. As a typical thermal-induced flavonoid derivative with antioxidant activity and flavor-modulating function, the massive accumulation of orientin in medium- and high-temperature groups effectively improved the taste quality of A. capillaris, reduced bitter astringency, and contributed to the formation of unique roasted and sweet aromas in processed samples.
KEGG topology enrichment bubble plot was used to analyze the key metabolic pathways perturbed by thermal processing (Fig. 5I). The enrichment results revealed that the differential metabolites induced by gradient thermal processing were mainly enriched in flavonoid biosynthesis, phenolic acid metabolic pathways, amino acid metabolism, carbohydrate metabolism and secondary metabolite biosynthesis pathways. Among them, secondary metabolite synthesis pathways exhibited the highest enrichment significance and pathway impact value in T100 and T120 groups, indicating that high-temperature processing mainly activated the biosynthesis of flavor and functional secondary metabolites. The rewiring of these core metabolic pathways fundamentally drove the metabolic quality improvement of thermally processed A. capillaris.
The functional clustering heatmap of pathway-annotated differential metabolites further clarified the coordinated expression patterns of pathway-related metabolites under different thermal treatments (Fig.5J). Metabolites involved in primary metabolic pathways (organic acid metabolism, amino acid catabolism) were highly expressed in CK and T50 groups, which corresponded to the raw grassy and bitter taste of unprocessed and low-temperature processed samples. In contrast, metabolites related to flavonoid and phenolic acid biosynthesis presented consistent upregulation trends in T100 and T120 groups, showing highly coordinated expression patterns. These pathway-specific metabolite accumulation characteristics confirmed that gradient thermal processing could precisely regulate the metabolic network of A. capillaris, thereby realizing the directional formation of different flavor qualities.
In summary, the widely targeted non-volatile metabolomics results systematically revealed the molecular mechanism of thermal processing regulating the flavor quality of A. capillaris. Low-temperature treatment maintained the original metabolic characteristics of raw samples with trivial quality improvement. Medium- and high-temperature gradient processing profoundly remodeled the non-volatile metabolic network, inhibited the accumulation of off-flavor-related primary metabolites, and significantly promoted the enrichment of flavor-active secondary metabolites via regulating core metabolic pathways. The specific metabolic variation patterns under different temperature conditions laid a solid material foundation for the formation of unique sweet aroma (T100) and roasted aroma (T120) of processed A. capillaris, and provided a theoretical basis for the precise selection and optimization of thermal processing technology for high-quality A. capillaris products.
3.4.2. Heat-processing remodels phenylpropanoid and flavonoid metabolic profilesacross four experimental groups (CK, T50, T100, T120)
Phenylpropanoid and flavonoid biosynthesis are central pathways generating bitter-astringent compounds and aroma-active substances in Artemisia capillaris. In total, 28 key metabolites covering these two pathways were quantified and statistically analyzed (Table 3). Thermal treatment induced widespread and temperature-dependent metabolic alterations, reshaping the pool of taste- and aroma-related metabolites.
Table 3.
Relative abundance of key metabolites involved in phenylpropanoid and flavonoid biosynthesis pathways under different heat-processing treatments.
| Metabolite name | KEGG_ID | CK | T50 | T100 | T120 | Sensory-related function |
|---|---|---|---|---|---|---|
| Phenylpropanoid biosynthesis | ||||||
| Coniferaldehyde | C02666 | 2.64e-05 ± 2.10e-06 | 2.24e-05 ± 1.67e-06 | 5.44e-05 ± 2.33e-06*** | 4.62e-05 ± 2.86e-06*** | Woody-herbal aroma |
| 5-Hydroxyferulic acid | C05619 | 1.05e-06 ± 2.26e-07 | 1.65e-07 ± 2.62e-07* | 8.12e-07 ± 7.56e-07 | 9.53e-07 ± 6.29e-07 | Astringent taste |
| Coniferin | C00761 | 4.83e-06 ± 3.98e-07 | 5.31e-07 ± 2.00e-07*** | 5.47e-07 ± 2.86e-07*** | 1.69e-06 ± 1.85e-07** | Bitter and astringent taste |
| Eugenol | C10453 | 3.18e-04 ± 1.91e-05 | 1.32e-04 ± 1.19e-05*** | 3.95e-04 ± 2.66e-05* | 3.80e-04 ± 3.14e-05 | Clove-like herbal aroma |
| Sinapoylcholine | C00933 | 1.06e-04 ± 1.10e-06 | 5.71e-06 ± 5.65e-08*** | 1.17e-05 ± 6.53e-07*** | 1.50e-05 ± 1.41e-06*** | Modulate bitter taste |
| p-Coumaryl alcohol | / | 9.76e-06 ± 1.13e-06 | 3.33e-06 ± 9.60e-07** | 5.20e-07 ± 3.43e-07** | 3.57e-07 ± 1.04e-07** | Bitter taste |
| 5-Hydroxyconiferaldehyde | C12204 | 3.34e-05 ± 1.70e-06 | 3.13e-05 ± 2.61e-06 | 4.67e-05 ± 7.06e-06 | 3.44e-05 ± 7.63e-07 | Astringent property |
| Sinapaldehyde | C05610 | 2.56e-05 ± 2.04e-06 | 2.56e-05 ± 8.05e-06 | 2.29e-05 ± 4.53e-06 | 2.43e-05 ± 3.80e-06 | Spicy-herbal aroma |
| Coniferol | C00590 | 8.73e-06 ± 1.48e-06 | 2.91e-05 ± 1.41e-06*** | 1.11e-05 ± 1.70e-06 | 9.37e-06 ± 4.21e-07 | Woody aroma |
| Sinapyl alcohol | C02325 | 1.34e-05 ± 8.07e-07 | 9.29e-06 ± 2.62e-07** | 1.46e-05 ± 5.72e-07 | 1.10e-05 ± 2.05e-06 | Herbal aroma |
| Phenylalanine | C00079 | 2.37e-04 ± 8.12e-06 | 2.71e-04 ± 2.67e-05 | 2.65e-04 ± 5.50e-06* | 2.91e-04 ± 1.08e-05** | Phenylpropanoid precursor; weak bitter taste |
| Tyrosine | C00082 | 1.21e-04 ± 6.96e-06 | 8.55e-05 ± 7.13e-06** | 8.34e-05 ± 3.97e-06** | 8.48e-05 ± 6.89e-06** | Phenylpropanoid precursor |
| Chlorogenic acid | C00852 | 1.10e-04 ± 2.31e-05 | 1.38e-05 ± 1.72e-05** | 1.83e-05 ± 2.66e-06* | 2.32e-05 ± 3.13e-07* | Astringent, bitter taste, herbal-grassy aroma |
| Flavonoid biosynthesis | ||||||
| Aromadendrin | C00974 | 3.91e-06 ± 3.67e-07 | 9.61e-06 ± 1.16e-06** | 8.86e-06 ± 1.43e-06** | 7.33e-06 ± 6.02e-07** | Astringent taste |
| Sakuranetin | C09833 | 9.44e-06 ± 1.06e-06 | 5.07e-07 ± 5.10e-08*** | 2.04e-07 ± 2.13e-07*** | 2.55e-07 ± 1.70e-07*** | Bitter taste |
| Galangin | C10044 | 1.43e-03 ± 1.87e-05 | 4.35e-03 ± 3.62e-04*** | 2.96e-03 ± 8.32e-05*** | 2.56e-03 ± 9.57e-05*** | Herbal-bitter note |
| Apigenin | C01477 | 5.60e-06 ± 8.72e-07 | 2.31e-05 ± 7.49e-08*** | 1.51e-05 ± 1.26e-06*** | 1.29e-05 ± 1.38e-06*** | Bitter, slight sweet |
| Luteolin | C01514 | 1.01e-05 ± 1.02e-06 | 7.76e-05 ± 5.30e-06*** | 1.66e-05 ± 9.13e-07** | 1.36e-05 ± 6.24e-07* | Bitter, herbal fragrance |
| Taxifolin | C01617 | 1.61e-07 ± 4.42e-08 | 9.25e-06 ± 1.48e-06*** | 2.16e-07 ± 3.44e-07 | 2.83e-07 ± 2.34e-07 | Bitter and astringent |
| Naringenin | / | 6.56e-06 ± 8.07e-07 | 8.24e-05 ± 4.97e-06*** | 3.16e-05 ± 1.68e-06*** | 3.40e-05 ± 1.28e-06*** | Bitter taste |
| Delphinidin | C05908 | 3.38e-05 ± 2.22e-06 | 2.40e-05 ± 2.41e-06* | 1.32e-05 ± 2.94e-06*** | 1.01e-05 ± 1.74e-06*** | Weak astringency |
| Prunin | C09099 | 2.74e-06 ± 3.56e-07 | 2.28e-06 ± 8.88e-07 | 1.15e-06 ± 7.10e-07* | 6.25e-07 ± 5.21e-07** | Bitter taste |
| Isosakuranetin | C05334 | 5.17e-06 ± 1.19e-06 | 1.89e-06 ± 5.28e-07* | 1.41e-06 ± 4.94e-07** | 1.23e-06 ± 1.74e-07** | Bitter taste |
| Naringin | C09789 | 9.07e-06 ± 1.15e-07 | 4.94e-06 ± 8.17e-07*** | 7.02e-06 ± 1.04e-06** | 6.23e-06 ± 4.55e-07*** | Strong bitter taste |
| Homoeriodictyol | C09756 | 4.50e-07 ± 5.31e-07 | 5.86e-06 ± 4.64e-07*** | 2.36e-06 ± 6.58e-07** | 1.87e-06 ± 6.96e-07* | Astringent taste |
| Phloretin | C00774 | 1.44e-05 ± 2.17e-06 | 2.22e-04 ± 2.22e-05*** | 2.59e-04 ± 1.02e-05*** | 2.45e-04 ± 1.81e-05*** | Bitter taste |
| Kaempferol | C05903 | 1.38e-04 ± 1.12e-05 | 5.01e-03 ± 1.76e-04*** | 3.99e-04 ± 1.06e-05*** | 2.91e-04 ± 1.47e-05*** | Astringent-bitter |
| Quercetin | C00389 | 1.49e-05 ± 8.37e-08 | 8.56e-06 ± 1.42e-06** | 5.64e-06 ± 1.13e-07*** | 4.84e-06 ± 4.42e-07*** | Astringent taste |
Note: CK, untreated control group; T50, T100, T120 represent thermal processing at 50 °C, 100 °C and 120 °C, respectively. Data are expressed as mean ± standard deviation (SD, n = 3). One-way ANOVA with Welch's t-test was used for pairwise comparison versus CK. *P < 0.05, **P < 0.01, ***P < 0.001; no symbol indicates no significant difference.
For phenylpropanoid pathway, upstream precursor tyrosine was significantly decreased across all heating groups, whereas phenylalanine accumulated significantly at T100 and T120, indicating substrate redistribution at the shikimate-derived aromatic amino-acid node. Glycosylated metabolites (coniferin, sinapoylcholine) were sharply depleted under heat treatment. The monolignol-related aldehyde coniferaldehyde significantly increased at higher temperature (T100, T120), while monolignol alcohol coniferol only accumulated markedly under mild heating (T50). As a hub metabolite linking phenylpropanoid and flavonoid branches, chlorogenic acid was significantly reduced in all thermal groups. Eugenol, a representative aroma compound, decreased sharply at T50 but recovered at T100. Several monolignol intermediates (5-hydroxyconiferaldehyde, sinapaldehyde) remained stable without statistical changes.
In flavonoid biosynthesis, two distinct response patterns were observed. A set of flavonoid aglycones (aromadendrin, apigenin, luteolin, naringenin, homoeriodictyol, phloretin, kaempferol) were significantly elevated after heat exposure, generally reaching maximum abundance in T50. By contrast, multiple flavonoid glycosides and related derivatives (sakuranetin, delphinidin, isosakuranetin, naringin, quercetin) were significantly reduced upon heating. Taxifolin only showed significant up-regulation in T50, with no statistical difference in T100 and T120 relative to CK. Prunin exhibited significant decline only in T100 and T120.
Collectively, heat-processing drove metabolic-flux redirection from glycosylated phenolics toward corresponding aglycone forms, which can be partially explained by heat-mediated deglycosylation. These metabolic shifts jointly determined the bitter-astringent intensity and herbal-woody aroma characteristics of thermally-processed A. capillaris. Mild heating (T50) produced the highest abundance of most flavonoid aglycones, while higher temperature (100 °C, 120 °C) partially suppressed aglycone accumulation but promoted the generation of aldehyde-type aroma intermediates.
3.5. Correlations between quantitative descriptive analysis, key aroma compounds and non-volatile metabolites
The correlation analysis between key aroma compounds and core non-volatile metabolites clarified the metabolic precursor mechanism underlying aroma differentiation induced by thermal processing (Fig. 6A). Primary metabolites (e.g., organic acids and soluble sugars) that were highly accumulated in untreated and low-temperature processed (CK, T50) samples exhibited significant positive correlations with grassy and green volatile components, which were responsible for the undesirable raw herbal odor and bitter taste of unprocessed A. capillaris. In contrast, thermally induced secondary metabolites dominated by flavonoids and phenolic acids (enriched in T100 and T120) showed strong positive correlations with high-value aroma compounds characterized by ROAV >1. These non-volatile secondary metabolites acted as critical precursors for the biosynthesis and transformation of sweet and roasted aroma volatiles, demonstrating that medium- and high-temperature processing facilitated the accumulation of flavor-preferential metabolites to reshape the aroma profile of A. capillaris. Additionally, significant negative correlations between flavor-defective primary metabolites and high-quality aroma-active volatiles further explained the elimination of raw off-flavors and formation of superior aromatic characteristics during thermal processing.
Fig. 6.
Correlations Between quantitative descriptive analysis, Key Aroma Compounds (ROAV >1) and non-volatile metabolites (VIP > 3.5) (Table S1) in CK,T50, T100 and T120. (A) Correlation analysis between Key Aroma Compounds (ROAV >1) and non-volatile metabolites(VIP > 3.5); (B) Correlation analysis between quantitative descriptive analysis, and non-volatile metabolites(VIP > 3.5). Significant correlations were expressed by *, ** and ***, which were *p < 0.05, **p < 0.01, and ***p < 0.001, respectively.
The pairwise correlation between quantitative sensory attributes and core non-volatile metabolites further verified the metabolic regulation of taste and aroma quality (Fig. 6B). The bitter, astringent, and green grassy sensory attributes of CK and T50 samples were positively correlated with the abundance of primary metabolites, confirming that the retention of inherent primary metabolic components accounted for the inferior sensory quality of low-temperature treated and raw samples. Notably, secondary metabolites significantly enriched in T100 and T120 groups were strongly positively associated with sweet flavor, mellow taste, and roasted aroma. Specifically, medium-temperature processing (T100) induced specific metabolite accumulation patterns that corresponded to the formation of distinct sweet aroma, while high-temperature treatment (T120) further optimized the secondary metabolic profile, ultimately generating intense and harmonious roasted aroma characteristics. The highly specific metabolite-sensory correlation patterns across temperature gradients solidified the notion that thermal processing remodeled the non-volatile metabolic landscape to directionally regulate the sensory quality of A. capillaris.
Collectively, the integrated correlation results established a complete regulatory network linking non-volatile metabolome, volatile aroma compounds, and sensory quality of thermally processed A. capillaris. Low-temperature drying failed to reverse the defective metabolic and sensory characteristics of raw materials due to insufficient metabolic modulation. By contrast, medium- and high-temperature thermal processing triggered coordinated metabolic optimization: it suppressed the accumulation of off-flavor-related primary metabolites, enriched functional secondary metabolites serving as high-quality aroma precursors, and ultimately eliminated raw grassy and bitter defects while promoting the formation of typical sweet and roasted aromas. These findings systematically elucidate the multi-level metabolic mechanism by which gradient thermal processing improves the comprehensive flavor quality of A. capillaris, providing solid theoretical support for the precise optimization of processing techniques to achieve targeted flavor modulation.
4. Discussion
4.1. Effects of thermal processing on sensory quality and aroma profiles of saline-grown Artemisia capillaris
Sensory evaluation clearly distinguished the comprehensive quality of Artemisia capillaris products under different thermal processing regimes. Fresh (CK) and low-temperature dried (T50) samples exhibited dominant grassy, herbal and bitter sensory attributes with rough oral texture, while medium-temperature dried (T100) samples presented a prominent sweet aroma, and high-temperature dried (T120) samples showed a distinct roasted aroma. Such sensory divergence was directly attributed to the distinct volatile composition mediated by thermal processing intensity (Wu, Dong, et al., 2026; Liu et al., 2024). These results indicate that consumers can choose the appropriate drying process according to their flavor preferences, whether they prefer the natural grassy and herbal notes of low-temperature drying, the sweet aroma of medium-temperature drying, or the roasted aroma of high-temperature drying. A. capillarisThis study provides clear and operable industrial processing guidelines for targeted production of herbal tea with differentiated flavor characteristics: For ecological functional medicinal tea products focusing on retaining the original medicinal and herbal flavor of wild halophytes, 50 °C low-temperature drying is recommended to preserve the unique stress-induced secondary metabolites and natural herbal scent of saline-grown; for daily popular herbal tea products oriented to mass consumer groups, 100 °C medium-temperature processing is the optimal industrial parameter, which effectively reduces bitter and astringent defects and forms harmonious sweet and mellow flavor; for high-end customized flavored tea products, 120 °C high-temperature aroma enhancement can be adopted to produce unique roasted aroma characteristics and improve product added value. This graded temperature processing strategy realizes precise docking of processing technology, flavor characteristics and market demand, and greatly improves the industrial applicability and market promotion value of coastal saline-alkali tea resources. Importantly, the distinct temperature-responsive flavor and metabolic remodeling revealed in this study are uniquely dependent on the saline-alkali-induced metabolic characteristics of A. capillaris, rather than universal processing effects of ordinary plant materials. The coastal saline-alkali habitat is not merely a geographical background, but a core functional variable shaping the unique metabolite precursor pool of experimental materials.
In this study, a total of 487 volatile compounds were identified across four treatment groups via GC × GC-TOFMS. Although all samples shared a core set of common volatile components, specific volatile compounds and differential metabolite accumulation existed among groups. Multivariate analysis confirmed obvious separation of aroma profiles between low-temperature (CK and T50) and high-temperature (T100 and T120) groups. Combined with VIP screening and ROAV analysis, differential volatile compounds effectively reflected the aroma discrepancy among processed Artemisia capillaris products. Further ROAV analysis indicated that high-temperature processed samples possessed higher contents and odor contribution of thermally generated aroma compounds, including trans-β-ionone (705,725.05 mg/kg, ROAV >1), caryophyllene, pyrazine derivatives, and furan compounds, which collectively provided rich floral, roasted and sweet notes. By comparison, CK and T50 accumulated higher hexanal, pentanal and (E)-2-octenal, which enhanced undesirable green and grassy attributes (Oussou et al., 2023). Terpenoids, aldehydes and heterocyclic compounds are core aroma-active substances determining herbal tea flavor (Guo et al., 2021; Shen et al., 2026). The differential distribution of these key odorants adequately explained the sensory differentiation, demonstrating that appropriate high-temperature fixation and aroma enhancement could effectively eliminate raw green and bitter defects of fresh saline-grown Artemisia capillaris and improve aromatic richness and sensory balance.
Specifically, the abundant terpenoids and phenolic acids accumulated under saline-alkali stress are well retained under low-temperature drying (T50), maintaining the unique grassy and herbal flavor fingerprint of saline-alkali halophytic A. capillaris. Moderate medium-temperature processing (T100) precisely modulates the stress-derived metabolic profile: it moderately degrades bitter phenolic substances induced by saline stress and retains stress-enriched soluble sugars, thus forming the exclusive sweet aroma characteristic of saline-grown A. capillaris tea. Under high-temperature aroma enhancement (T120), the unique primary metabolic substrates (differential sugars and amino acids) formed by saline-alkali stress undergo intense Maillard reaction and thermal caramelization, producing exclusive heterocyclic roasted aroma compounds that cannot be formed in normal soil-grown materials.
In summary, the differential flavor formation of thermally processed A. capillaris is the coupling result of saline-alkali habitat stress shaping unique metabolic precursors and postharvest thermal temperature regulating metabolite transformation. This study innovatively clarifies the scientific connotation of saline-alkali habitat as a decisive functional variable, rather than a simple geographical descriptor, providing targeted theoretical support for the characteristic flavor customization and high-value utilization of coastal saline-alkali medicinal and edible halophytes.
4.2. Temperature-driven volatile metabolic regulation and characteristic aroma marker screening
The PCA and HCA results demonstrated clear temperature-dependent clustering patterns of volatile profiles, with CK and T50 samples sharing highly similar metabolic fingerprints while T100 and T120 forming independent and well-separated clusters. This indicated that low-temperature drying exerted limited effects on the volatile metabolome, whereas high-temperature processing fundamentally altered volatile characteristics (Guo et al., 2021; Xie et al., 2024).
A total of 59 key aroma compounds with ROAV >1 were identified across treatments, exhibiting distinct temperature-responsive accumulation patterns. Green-related volatiles including hexanal, pentanal and 1-octen-3-ol were significantly enriched in CK and T50 groups, contributing typical fresh grass and herbal notes. These compounds are characteristic of intact plant tissues and are primarily derived from lipoxygenase-mediated oxidation of unsaturated fatty acids (Nishidono et al., 2019). In contrast, high-temperature treatments triggered substantial formation of thermally induced volatiles. Notably, two pyrazine compounds (3-ethyl-2,5-dimethyl-pyrazine and 2-ethyl-5-methyl-pyrazine) were exclusively detected in the T120 group, contributing roasted nut and caramel sweet aromas through Maillard reactions (Xue et al., 2026). The dramatic accumulation of trans-β-ionone in T100 (705,725.05 mg/kg) and T120 (51,562.76 mg/kg), compared with undetectable levels in CK, indicated that thermal degradation of carotenoids was strongly temperature-dependent (Pan et al., 2026).
The VIP > 1 screening identified key discriminatory volatile compounds that effectively reflected the aroma discrepancy among processed Artemisia capillaris products. Further ROAV analysis indicated that high-temperature processed samples possessed higher contents and odor contribution of thermally generated aroma compounds, including trans-β-ionone (705,725.05 mg/kg, ROAV >1), caryophyllene, pyrazine derivatives, and furan compounds, which collectively provided rich floral, roasted and sweet notes. By comparison, CK and T50 accumulated higher hexanal, pentanal and (E)-2-octenal, which enhanced undesirable green and grassy attributes (Oussou et al., 2023). Terpenoids, aldehydes and heterocyclic compounds are core aroma-active substances determining herbal tea flavor (Jie et al., 2026; Guo et al., 2021). The differential distribution of these key odorants adequately explained the sensory differentiation, demonstrating that appropriate high-temperature fixation and aroma enhancement could effectively eliminate raw green and bitter defects of fresh saline-grown Artemisia capillaris and improve aromatic richness and sensory balance.
4.3. Non-volatile metabolite under gradient thermal processing
Widely targeted metabolomics further revealed temperature-induced profound alterations in non-volatile metabolites. PCA and HCA showed consistent separation patterns with volatile results, confirming that high-temperature processing significantly remodeled the non-volatile metabolome while low-temperature drying maintained metabolic profiles close to fresh samples (Mohammed et al., 2024). The results demonstrated that gradient temperature treatments dramatically reshaped the non-volatile metabolic landscape of A. capillaris, resulting in distinct accumulation patterns of primary and secondary metabolites, which fundamentally mediated the differential flavor and taste formation of processed samples. Low-temperature short-time treatment (T50) retained the inherent metabolic characteristics of raw samples, while medium-temperature (T100) and high-temperature (T120) processing induced drastic metabolic variations, facilitating the accumulation of flavor-active secondary metabolites and the reduction of undesirable taste-related primary metabolites.
A large number of primary metabolites including organic acids, soluble sugars and amino acids were highly enriched in CK and T50 groups. These compounds are closely related to the fresh plant primary metabolism and contribute to the raw green taste and bitter-astringent sensory defects of untreated or insufficiently processed herbal materials (Liang et al., 2026). In contrast, massive beneficial secondary metabolites including total flavonoids, phenolic acids and terpenoid derivatives were significantly upregulated in T100 and T120 groups. The T120 group exhibited the highest accumulation of functional secondary metabolites and the lowest abundance of undesirable primary metabolites, showing the optimal metabolic modification effect. This temperature-dependent metabolic indicated that high-temperature processing effectively promoted the conversion from primary to secondary metabolism, enhancing the accumulation of bioactive and flavor-precursor compounds (Shen et al., 2026; Wu, Song, et al., 2026). The metabolic pathway analysis indicated that phenylpropanoid biosynthesis and flavonoid biosynthesis were the dominant pathways affected by thermal processing, which played crucial roles in the accumulation of flavor-active secondary metabolites.
The correlation analysis between differential non-volatile metabolites and sensory attributes demonstrated that thermally induced flavonoids and phenolic acids were positively correlated with roasted aroma, sweet flavor and mellow taste, while primary metabolites abundant in CK and T50 showed significant positive correlations with bitter, astringent and green off-flavors. This suggested that non-volatile secondary metabolites not only directly modulated taste properties but also served as critical aroma precursors, providing sufficient material basis for the generation of sweet, floral and roasted volatiles during high-temperature processing (Pan et al., 2026; Zhang, Zhang, Zhang, et al., 2024). The significant antagonistic relationships observed between flavor-improving secondary metabolites and defect-related primary metabolites indicated that high-temperature processing effectively reconstructed the metabolic correlation network and optimized overall flavor quality.
4.4. Thermal processing remodels phenylpropanoid and flavonoid metabolic networks to regulate taste and aroma quality
Phenylpropanoid and flavonoid biosynthesis are the core secondary metabolic pathways determining the flavor characteristics of herbal materials, which jointly produce bitter, astringent and aromatic compounds (Xie et al., 2026). The present study demonstrated that thermal processing induced profound and temperature-dependent metabolic reprogramming across the two pathways, forming a clear “precursor-intermediate-end product” metabolic flux shift, which fundamentally altered the sensory profile of the samples (Zhang, Zhang, Zhang, et al., 2024).
At the upstream precursor level, heat treatment significantly reduced tyrosine abundance while promoting the accumulation of phenylalanine, indicating a heat-induced substrate redistribution of aromatic amino acids in the shikimate pathway (Lynch et al., 2017). As the core precursor of phenylpropanoid synthesis, the increased phenylalanine provided sufficient carbon skeletons for downstream phenolic and flavonoid synthesis, which explained the massive accumulation of flavonoid aglycones in heat-treated groups (Pravallika et al., 2025). As a key junction metabolite connecting phenylpropanoid and flavonoid branches, chlorogenic acid was significantly depleted after heating, further confirming that thermal treatment redirected metabolic flux from intermediate phenolic acids toward end-stage flavonoid and monolignol products.
In the phenylpropanoid pathway, glycosylated bitter substances including coniferin and sinapoylcholine were almost depleted under heating, which greatly reduced the basic bitter and astringent taste of the sample. In contrast, high-temperature treatment (T100, T120) significantly increased the content of coniferaldehyde, a typical woody and herbal aromatic compound, while mild heating (T50) specifically promoted the accumulation of coniferol. This temperature-dependent differentiation indicated that low temperature favoured the synthesis of monolignol alcohols, whereas high temperature promoted the conversion of alcohols to aldehyde aroma precursors. Additionally, the significant decrease of multiple monolignol derivatives (p-coumaryl alcohol, sinapyl alcohol) in T50 further verified the metabolic flow transformation within the phenylpropanoid pathway under mild thermal stress.
Consistent with the phenylpropanoid response, flavonoid metabolites exhibited two completely opposite variation patterns after heat processing. Most flavonoid aglycones (kaempferol, naringenin, apigenin, luteolin, phloretin) were significantly up-regulated, while flavonoid glycosides (sakuranetin, naringin, quercetin, delphinidin) were universally down-regulated. This phenomenon strongly indicated the occurrence of heat-mediated deglycosylation: thermal treatment promoted the hydrolysis of bound glycosidic flavonoids into free aglycones (Wang, Zhang, et al., 2025b). Notably, T50 treatment induced the highest accumulation of flavonoid aglycones, while excessive high temperature (100–120 °C) partially decreased aglycone content, suggesting that mild heating optimizes flavonoid accumulation whereas ultra-high temperature may cause partial thermal degradation of phenolic compounds (Boonkorkaew et al., 2026).
The metabolic changes observed in this study perfectly corresponded to sensory quality variation. The degradation of glycosylated bitter substances (coniferin, naringin, sakuranetin) reduced the harsh astringency and bitter taste of the sample, while the enrichment of aroma-active aldehydes and flavonoid aglycones enhanced the layered herbal, woody and clove-like aroma. The unique metabolic characteristics of different temperature groups also explained the sensory differences among T50, T100 and T120 samples.
Nevertheless, this study only relied on untargeted metabolomic data to elaborate metabolic changes. The lack of transcriptomic evidence limited the interpretation of upstream gene regulatory mechanisms. Future multi-omics integration analysis will further clarify the key enzyme genes driving phenylpropanoid and flavonoid metabolic flow changes, so as to fully reveal the molecular mechanism of heat processing regulating herbal flavor quality.
4.5. Integrated metabolic regulatory network linking non-volatile metabolites, volatile compounds and sensory quality
Pearson correlation analysis was applied in the present work to establish associations between key aroma-active volatiles and core non-volatile metabolites. Nevertheless, correlation statistics merely demonstrate covariation patterns and cannot directly prove precursor-to-product transformation relationships. Combining our metabolomic dataset and well-established thermal processing theory for herbal tea materials, we further interpreted the potential formation mechanisms linking non-volatile substrates to aroma compounds under gradient heat treatments. Phenylpropanoid and flavonoid substances, as the predominant differential non-volatile metabolites, act as vital aroma precursors. Under low-temperature drying (T50), most polyphenol and flavonoid skeletons remain relatively stable, with limited degradation toward volatile derivatives, which preserves the original grassy and herbal notes of raw Artemisia capillaris. By contrast, medium- and high-temperature treatments (T100 and T120) initiate thermal degradation, decarboxylation and oxidation of phenolic acids and flavonoids, yielding abundant benzenoid and phenolic volatile constituents; these volatile products mitigate intense bitter-herbal perceptions and contribute to pleasant sweet-scented attributes (Xie et al., 2026; Zhou et al., 2017). Soluble sugars and amino acids, representing primary metabolic substrates, provide reactants for thermally driven aroma-generating reactions. High-temperature processing supplies sufficient thermal energy to facilitate Maillard reaction and caramelization, and the abundant heterocyclic volatiles observed in the T120 group originate largely from thermal reactions of differential amino acids and carbohydrates, conferring distinct roasted-like sensory characteristics to Yinchen tea. Accordingly, the temperature-dependent consumption of non-volatile precursors and concurrent build-up of volatile products reasonably account for inter-group sensory divergence. In this context, Pearson correlation served as a screening tool to mine candidate precursor-product pairs and pinpoint key regulatory nodes bridging metabolite variations and flavor phenotypes. Building upon these correlation outputs, the above thermal transformation and metabolic degradation mechanisms were integrated to interpret flavor formation beyond simple correlative description.
Integrated correlation analysis established a comprehensive regulatory network covering non-volatile metabolites, volatile aroma compounds and sensory quality of processed Artemisia capillaris. Primary metabolites enriched in CK and T50 exhibited significant positive correlations with green-feeling volatile compounds including hexanal, pentanal and (E)-2-octenal, jointly constituting the raw and grassy sensory defects. In contrast, secondary metabolites highly accumulated in T100 and T120 showed strong positive correlations with key high-quality aroma compounds with ROAV >1, such as trans-β-ionone, caryophyllene and pyrazine derivatives (Oussou et al., 2023).
These results revealed a hierarchical regulation mechanism: thermal processing directly altered non-volatile metabolite profiles, which in turn determined the availability of aroma precursors and flavor-active substances. The differential accumulation of non-volatile metabolites indirectly regulated the composition and abundance of volatile aroma substances, thereby differentiating the final sensory quality of processed Artemisia capillaris products. This multi-layered regulatory network explained why high-temperature processing simultaneously improved both taste (reduced bitterness and astringency) and aroma (enhanced roasted and sweet notes) characteristics (Wu, Chang, et al., 2026; Xie et al., 2024).
The significant antagonistic relationships observed between flavor-improving secondary metabolites and defect-related primary metabolites indicated that high-temperature processing effectively reconstructed the metabolic correlation network and optimized overall flavor quality. This comprehensive metabolic remodeling demonstrated that thermal processing temperature was the core driving factor for quality formation of processed Artemisia capillaris products, and gradient high-temperature fixation with aroma enhancement represented an effective processing strategy for saline-grown halophyte tea resources (Jie et al., 2026; Liang et al., 2026).
4.6. Implications and limitations of this study
This study systematically clarified the improvement effect of gradient thermal processing on the sensory quality of saline-grown Artemisia capillaris, and revealed the temperature-dependent regulatory mechanism of “non-volatile metabolite-volatile aroma formation-sensory quality differentiation”. High-temperature fixation and aroma enhancement effectively eliminated the raw green and bitter sensory defects of fresh saline-grown materials, promoted the accumulation of beneficial secondary metabolites and thermally generated aroma compounds, and significantly improved the overall quality of processed products. These findings provide theoretical support for standardized low-temperature processing and high-value industrial utilization of coastal saline-alkali halophyte tea resources (Liang et al., 2025; Miao et al., 2022).
Nevertheless, several limitations still exist. The present research only focused on a single harvest season (April) from one geographical location (Dongying Yellow River Delta). Seasonal variation, annual climate change and different saline habitats may affect the metabolic responses of Artemisia capillaris to thermal processing (Zhang, Zhang, Zhou, et al., 2024). In addition, further in-depth exploration of the specific thermal reaction pathways and kinetic modeling of key metabolite formation is required. Subsequent multi-season and multi-region verification experiments, combined with targeted pathway analysis and enzyme activity assays, will help to establish optimized processing parameters and promote the sustainable and high-quality development of halophyte functional food industry (Mohammed et al., 2024; Ye et al., 2026).
5. Conclusions
Gradient thermal drying treatments significantly differentiated the sensory quality and metabolic profiles of Artemisia capillaris tea cultivated in Dongying coastal saline-alkali soil. Low-temperature drying at 50 °C maintained the inherent grassy, herbal and bitter sensory attributes of fresh Yinchen samples with negligible metabolic modification. Medium-temperature processing at 100 °C effectively eliminated raw off-flavors and promoted the formation of distinct sweet aroma, while 120 °C high-temperature aroma enhancement generated unique roasted flavor characteristics via thermal reactions. A total of 487 volatile compounds and 8821 differential non-volatile metabolites were successfully identified across all treatments, with temperature gradient serving as the core driving factor for metabolic remodeling. Thermal processing primarily regulated phenylpropanoid and flavonoid biosynthesis pathways, induced flavonoid deglycosylation, reduced bitter and astringent metabolites, and enriched aroma-active secondary metabolites. This study systematically clarifies the processing response characteristics of saline-alkali grown halophytic Yinchen, reveals the intrinsic mechanism of temperature-driven flavor differentiation, and provides reliable theoretical guidance and technical support for precise processing parameter optimization and high-value industrial utilization of coastal saline-alkali halophyte tea resources.
CRediT authorship contribution statement
Lili Tian: Writing – original draft, Data curation, Conceptualization. Xinhui Hu: Investigation, Formal analysis, Data curation. Yuantao Yao: Validation, Software, Resources, Methodology. Ya Zhang: Visualization, Validation, Resources, Data curation. Lu Wang: Writing – review & editing, Writing – original draft, Funding acquisition.
Declaration of competing interest
The authors declare the following financial interests/personal relationships which may be considered as potential competing interests: Lu Wang reports financial support, article publishing charges, equipment, drugs, or supplies, statistical analysis, travel, and writing assistance were provided by Yellow River Delta Modern Agriculture Research Institute Shandong Academy of Agricultural Sciences. Lu Wang reports a relationship with Yellow River Delta Modern Agriculture Research Institute Shandong Academy of Agricultural Sciences that includes:. Lu Wang has patent pending to Lu Wang. no If there are other authors, they declare that they have no known competing financial interests or personal relationships that could have appeared to influence the work reported in this paper.
Acknowledgments
This research was supported by Key R&D Program of Shandong Province, China (2024SFGC0403).
Footnotes
Supplementary data to this article can be found online at https://doi.org/10.1016/j.fochx.2026.104528.
Appendix A. Supplementary data
Supplementary Table S1: Yinchen tea contains substances with a metabolic activity index (VIP) ≥ 3.5.
Data availability
Data will be made available on request.
References
- Aebi H. Catalase in vitro. Methods in Enzymology. 1984;105:121–126. doi: 10.1016/S0076-6879(84)05016-3. [DOI] [PubMed] [Google Scholar]
- Boonkorkaew P, Hayi-asan N, Thongtip A, et al. (2026). Preharvest temperature treatments influence physiological traits, secondary metabolites, and volatile compounds in holy basil under controlled environment conditions.Scientific Reports, 16: 63137. doi; 10.1038/s41598-026-63137-3 doi;10.1038/s41598-026-63137-3. [DOI] [PubMed]
- Chen Q.C., Zhu Y., Liu Y.F., et al. Black tea aroma formation during the fermentation period. Food Chemistry. 2022;374 doi: 10.1016/j.foodchem.2021.131640. [DOI] [PubMed] [Google Scholar]
- Croppenstedt A., Knowles M., Lowder K.S. Social protection and agriculture: Introduction to the special issue. Global Food Security. 2018;16:65–68. doi: 10.1016/j.gfs.2017.09.006. [DOI] [Google Scholar]
- De Rooij G.H. Methods of soil analysis. Part 4. Physical methods. Vadose Zone Journal. 2004;3:722–723. doi: 10.2136/vzj2004.0722. [DOI] [Google Scholar]
- Drabińska N., Jeleń H.H. Optimisation of headspace solid-phase microextraction with comprehensive two-dimensional gas chromatography–time of flight mass spectrometry (HS-SPME–GC × GC–ToFMS) for quantitative analysis of volatile compounds in vegetable oils using statistical experimental design. Journal of Food Composition and Analysis. 2022;110 doi: 10.1016/j.jfca.2022.104595. [DOI] [Google Scholar]
- Du L., Wang C., Zhang C., et al. Characterization of the volatile and sensory profile of instant Pu-erh tea using GC × GC-TOFMS and descriptive sensory analysis. Microchemical Journal. 2019;146:986–996. doi: 10.1016/j.microc.2019.02.036. [DOI] [Google Scholar]
- Faizullah M.M., Ramprakash T., Anjaiah T., Madhavi M. Variations in soil urease and dehydrogenase activities as determined by Diuron, Pyrithiobac sodium and Quizalofop ethyl applied to cotton cultivated in red and black soils. Current Journal of Applied Science and Technology. 2020;39:103–111. doi: 10.9734/cjast/2020/v39i1630742. [DOI] [Google Scholar]
- Frankenberger W.T., Johanson J.B. Method of measuring invertase activity in soils. Plant and Soil. 1983;74(3):301–311. doi: 10.1007/BF02181348. [DOI] [Google Scholar]
- Guo X., Ho C.T., Wan X., et al. Changes of volatile compounds and odor profiles in Wuyi rock tea during processing. Food Chemistry. 2021;341(P1) doi: 10.1016/j.foodchem.2020.128230. [DOI] [PubMed] [Google Scholar]
- Izadmanesh Y., Garreta-Lara E., Ghasemi J.B., et al. Chemometric analysis of comprehensive two dimensional gas chromatography–mass spectrometry metabolomics data. Journal of Chromatography A. 2017;1488:113–125. doi: 10.1016/j.chroma.2017.01.052. [DOI] [PubMed] [Google Scholar]
- Karim F.K., Mehdi M.M. Drying temperatures affect the qualitative–quantitative variation of aromatic profiling in aromatic herb ecotypes. Frontiers in Plant Science. 2023;14 doi: 10.3389/fpls.2023.1137840. [DOI] [PMC free article] [PubMed] [Google Scholar]
- Li G.Q., Chen Y.D., Li H.X., et al. Soil pH and organic carbon content governing the active iron in tea plantation and maize field. Land Degradation & Development. 2025;37:826–839. doi: 10.1002/ldr.70144. [DOI] [Google Scholar]
- Li W., Chen J., Zhang J., Peng Z. Multi-omics insights into the formation mechanism of aging quality in white tea (Bai Mudan) during storage: Integrating sensory evaluation, volatilomics, metabolomics, and metagenomics. Food Research International. 2026;224 doi: 10.1016/j.foodres.2025.117996. [DOI] [PubMed] [Google Scholar]
- Li Y.T., Shao S.B., Deng S.W., et al. Comparison of flavor characteristics and metabolite basis of oolong tea from six different tea plant cultivars under high- and low-altitude conditions. Plants. 2025;15 doi: 10.3390/plants15010023. [DOI] [PMC free article] [PubMed] [Google Scholar]
- Liang H.Z., Wang M.Y., Yang G., et al. Untargeted qualitative and targeted quantitative analysis of saponins reveal differential chemotypes of <sc>Gynostemma pentaphyllum</sc> and <sc>G. longipes</sc> from different geographical origins. Food Chemistry. 2025;468 doi: 10.1016/j.foodchem.2024.142412. [DOI] [PubMed] [Google Scholar]
- Liang Y., Yuan Y., Wang J., et al. Analysis of changes in taste characteristics of coffee at different primary processing methods using E-tongue, untargeted metabolomics and WGCNA. Foods. 2026;15(9) doi: 10.3390/foods15091475. [DOI] [PMC free article] [PubMed] [Google Scholar]
- Liu Z., Ma C., Liu L.X., et al. Comprehensive investigation on non-volatile and volatile flavor compounds in different varieties of rose tea by UPLC-Q-TOF-MS/MS-based metabolomics and GC-IMS, GC-MS. Journal of Food Composition and Analysis. 2024;136 doi: 10.1016/j.jfca.2024.106708. [DOI] [Google Scholar]
- Lynch JH, Orlova I, Zhao C, Guo L, Jaini R, Maeda H, Akhtar T, Cruz-Lebron J, Rhodes D, Morgan J, Pilot G, Pichersky E, Dudareva N. (2017). Multifaceted plant responses to circumvent Phe hyperaccumulation by downregulation of flux through the shikimate pathway and by vacuolar Phe sequestration.The Plant Journal, 92: 939–950. doi; 10.1111/tpj.13730 doi;10.1111/tpj.13730. [DOI] [PubMed]
- Margarida A., Claro A.M., Ferreira I.Q., Rodrigues M.Â. The effect of legume species grown as cover crops in olive orchards on soil phosphorus bioavailability. Journal of Plant Nutrition. 2015;38:2294–2311. [Google Scholar]
- Miao S., Yang W., Chen J., Wei X. Extraction methods, physiological activities and high value applications of tea residue and its active components: A review. Critical Reviews in Food Science and Nutrition. 2022;63(33):5511–5519. doi: 10.1080/10408398.2022.2099343. [DOI] [PubMed] [Google Scholar]
- Mohammed H.A., Sulaiman G.M., Khan R.A., et al. Factors affecting the accumulation and variation of volatile and non-volatile constituents in rosemary, Rosmarinus officinalis L. Journal of Applied Research on Medicinal and Aromatic Plants. 2024;42 doi: 10.1016/j.jarmap.2024.100571. [DOI] [Google Scholar]
- Nishidono Y., Chiyomatsu T., Sanuki K., et al. Analysis of seasonal variations of the volatile constituents in Artemisia princeps (Japanese Mugwort) leaves by metabolomic approach. Natural Product Communications. 2019;14(8) doi: 10.1177/1934578X19872600. [DOI] [Google Scholar]
- Oussou K.F., Guclu G., Sevindik O., et al. GC-MS-Olfactometric characterization of volatile and key odorants in Moringa (Moringa oleifera) and Kinkeliba (Combretum micranthum G. Don) herbal tea infusions prepared from cold and hot brewing. Separations. 2023;10(1) doi: 10.3390/separations10010010. [DOI] [Google Scholar]
- Pravallika K., Pradhan S., Prabha A., et al. Ultraviolet and pulsed light treatment of spices and herbs and their products: Microbial safety, enzyme inactivation, bioactive retention, and shelf-life extension. Comprehensive Reviews in Food Science and Food Safety. 2025;24(2) doi: 10.1111/1541-4337.70107. [DOI] [PubMed] [Google Scholar]
- Pu T.Y., Zhang N., Wang J.Q., et al. Organic management pattern improves microbial community diversity and alters microbial network structure in karst tea plantation. Heliyon. 2024;10 doi: 10.1016/j.heliyon.2024.e31528. [DOI] [PMC free article] [PubMed] [Google Scholar]
- Shao X., Yang C., Chen Y., et al. Sorghum-peanut intercropping under salt stress mediates rhizosphere microbial community shaping in sorghum by affecting soil sugar metabolism pathways. Frontiers in Microbiology. 2025;16 doi: 10.3389/fmicb.2025.1589415. [DOI] [PMC free article] [PubMed] [Google Scholar]
- Shen J., Li P., Li Z., et al. Widely targeted metabolomics analysis reveals effects of raw material parts, harvest season, and processing technology of Huangqin (Scutellaria baicalensis Georgi) tea. LWT - Food Science and Technology. 2026;240 doi: 10.1016/j.lwt.2026.119022. [DOI] [Google Scholar]
- Tian L., Yao Y.T., Zhang Y., et al. Seasonal and processing differences in nonvolatile compounds and volatile aroma of teas from Zhaoyuan, China: Insight into “spring green, summer black, autumn dark” quality pattern. Food Chemistry: Vol. X. 2026;35 doi: 10.1016/j.fochx.2026.103740. [DOI] [PMC free article] [PubMed] [Google Scholar]
- Wang D., Liu B., Li F., et al. Status and influential factors of soil nutrients and acidification in Chinese tea plantations: A meta-analysis. Soil. 2025;11:175–191. doi: 10.5194/soil-11-175-2025. [DOI] [Google Scholar]
- Wang K., Yang A., Li Y., et al. Artemisia argyi essential oil modulates lipid metabolism via linoleic acid and glycerophospholipid pathways in high-fat diet-induced obese mice. Frontiers in Nutrition. 2025;12 doi: 10.3389/fnut.2025.1650976. [DOI] [PMC free article] [PubMed] [Google Scholar]
- Wang Q., Pa B., Gong J., et al. Phenylpropanoids, flavonoids, and terpenoids from Artemisia ordosica Krasch. Magnetic Resonance in Chemistry. 2019;57:326–330. doi: 10.1002/mrc.4846. [DOI] [PubMed] [Google Scholar]
- Wang Y., Zhang Y., Gu M., et al. A green and efficient deep eutectic solvent for extracting total flavonoids from Artemisiae scopariae Herba (Yin Chen): Chemical characterization, antioxidant activity, and hepatoprotective effects, and mechanisms of action against liver injury. Food Research International. 2025;222(P2) doi: 10.1016/j.foodres.2025.117715. [DOI] [PubMed] [Google Scholar]
- Wu P., Dong T., Tian Z., et al. Effects of thermal processing (blanching and hot-air drying) on key aroma compounds, carotenoid content, total phenolic content, and lipoxygenase activity in carrot. Journal of Food Composition and Analysis. 2026;156 doi: 10.1016/j.jfca.2026.109344. [DOI] [Google Scholar]
- Wu X., Song X., Liu T., et al. Formation and transfer patterns of key aroma compounds in cigar tobacco based on SBSE-GC-O-MS, pyrolysis-GC-O-MS and flue gas analysis correlation studies. Frontiers in Chemistry. 2026;14 doi: 10.3389/fchem.2026.1755255. [DOI] [PMC free article] [PubMed] [Google Scholar]
- Wu Z.D., Chang Y.N., Yang X.D., Jiang F.Y. Effects of long-term nitrogen fertilization on soil respiration in acidic tea (Camellia sinensis L.) plantation soils. Horticulturae. 2026;12 doi: 10.3390/horticulturae12030372. [DOI] [Google Scholar]
- Xia J., Ren J., Zhang S., et al. Forest and grass composite patterns improve the soil quality in the coastal saline-alkali land of the Yellow River Delta, China. Geoderma. 2019;349:25–35. doi: 10.1016/j.geoderma.2019.04.032. [DOI] [Google Scholar]
- Xie J., Wang Q., Hu J., et al. Uncovering the effects of spreading under different light irradiation on the volatile and non-volatile metabolites of green tea by intelligent sensory technologies integrated with targeted and non-targeted metabolomics analyses. Food Chemistry. 2024;463(P4) doi: 10.1016/j.foodchem.2024.141482. [DOI] [PubMed] [Google Scholar]
- Xie S, Zhang T, Ou H, et al. (2026). Integrated multi-omics analysis reveals flavor characteristics and underlying formation mechanisms during withering of Huangjincha black tea.LWT - Food Science and Technology, 241: 119096. doi; 10.1016/j.lwt.2026.119096. doi;10.1016/j.lwt.2026.119096. [DOI]
- Xue J.J., Liu P.P., Yin J.F., et al. Dynamic changes in volatile compounds of shaken black tea during its manufacture by GC × GC–TOFMS and multivariate data analysis. Foods. 2022;11 doi: 10.3390/foods11091228. [DOI] [PMC free article] [PubMed] [Google Scholar]
- Ye Q., Wu M., Li H., et al. Metabolomics analysis reveals the effect of scenting process on flavor quality of jasmine tea. Food Bioscience. 2026;75 doi: 10.1016/j.fbio.2025.108089. [DOI] [Google Scholar]
- Yu X., Di H., Zhao Y., et al. Optimized processing and formulation for retaining the sensory and nutritional quality of mustard leaf tea. Food Chemistry: X. 2025;27 doi: 10.1016/j.fochx.2025.102515. [DOI] [PMC free article] [PubMed] [Google Scholar]
- Zhang D., Zhang F., Zhang X., Cao Z., Song X., Zhang T., Yang Z. Metabolomics revealed the characteristics of the unique flavor substances of AlxaAllium mongolicum. Food Chemistry: X. 2024;24 doi: 10.1016/j.fochx.2024.101979. [DOI] [PMC free article] [PubMed] [Google Scholar]
- Zhang H., Li C.L., Zhang L.X., Zhang Q.F. Effects of potassium on tea quality are associated with the accumulation of anthocyanins in tea plant. Horticulture, Environment, and Biotechnology. 2024;65:657–667. doi: 10.1007/s13580-024-00654-y. [DOI] [Google Scholar]
- Zhang M., Zhang L., Zhou C., et al. Metabolite profiling reveals the dynamic changes in non-volatiles and volatiles during the enzymatic-catalyzed processing of Aijiao oolong tea. Plants. 2024;13(9) doi: 10.3390/plants13091249. [DOI] [PMC free article] [PubMed] [Google Scholar]
- Zhang Y.X., Wang X., Wang S.F., et al. Traditional Chinese bitter flavor theory: Is there any relation with taste type II receptors? European Journal of Integrative Medicine. 2016;8(6):980–990. doi: 10.1016/j.eujim.2016.04.011. [DOI] [Google Scholar]
- Zheng H., Wang X., Chen L., et al. Enhanced growth of halophyte plants in biochar-amended coastal soil: Roles of nutrient availability and rhizosphere microbial modulation. Plant, Cell & Environment. 2018;41:517–532. doi: 10.1111/pce.12944. [DOI] [PubMed] [Google Scholar]
- Zhou J., Zhang X., Liu W., et al. Widely targeted metabolomics analysis of the main bioactive compounds of Ganpu tea processing through different drying methods. LWT - Food Science and Technology. 2023;189 doi: 10.1016/j.lwt.2023.115501. [DOI] [Google Scholar]
- Zhou L, Xu JD, Zhou SS, et al. (2017). Chemomics-based marker compounds mining and mimetic processing for exploring chemical mechanisms in traditional processing of herbal medicines, a continuous study on Rehmanniae Radix.Journal of Chromatography A, 1530: 232–240. doi; 10.1016/j.chroma.2017.11.036. doi;10.1016/j.chroma.2017.11.036. [DOI] [PubMed]
Further reading
- Wang Z., Ge C., Zhang J., et al. The metabolic mechanism of flavonoid glycosides and their contribution to the flavor evolution of white tea during prolonged withering. Food Chemistry. 2024;439 doi: 10.1016/j.foodchem.2023.138133. [DOI] [PubMed] [Google Scholar]
Associated Data
This section collects any data citations, data availability statements, or supplementary materials included in this article.
Supplementary Materials
Supplementary Table S1: Yinchen tea contains substances with a metabolic activity index (VIP) ≥ 3.5.
Data Availability Statement
Data will be made available on request.