Influence of leaf tenderness on the aroma and taste characteristics of green tea from Camellia sinensis cv. Chuancha No. 2: integrated sensory and chemical profiling

TL;DR

This study shows how leaf tenderness affects the aroma and taste of green tea by analyzing chemical and sensory differences in samples with varying tenderness levels.

Contribution

The study identifies key aroma compounds and their biosynthetic pathways linked to leaf tenderness in green tea.

Findings

C1B (single bud) had a chestnut-like aroma and higher amino acids with lower polyphenols.

C1B1L (one-bud-one-leaf) had floral notes and the highest polyphenol content, with geraniol as a key contributor.

Five key volatiles were identified as major drivers of aroma variation across tenderness levels.

Abstract

Leaf tenderness influences green tea quality, yet its role in flavour formation remains unclear. This study analysed green teas from Camellia sinensis cv. ‘Chuancha No. 2’ with three tenderness levels—single bud (C1B), one-bud with one-leaf (C1B1L), and one-bud with two-leaves (C1B2L)—via sensory evaluation, chemical profiling, and GC–MS analysis. C1B showed the best appearance and a chestnut-like, refreshing aroma. C1B1L featured floral notes and a mellow, umami taste, while C1B2L was stronger and more full-bodied. C1B had higher free amino acids and lower polyphenols and caffeine; C1B1L had the highest polyphenol content. Among 97 volatiles, 11 had ROAV > 1. Five common aroma compounds, namely 1-octen-3-ol, geraniol, hexanal, benzeneacetaldehyde, and 2-pentylfuran, were identified as key contributors to the aroma profile. Geraniol, most abundant in C1B1L, shaped its floral aroma. These origins are supported by established pathways, including lipid-derived oxidation products, glycoside-released geraniol, and amino-acid-derived benzeneacetaldehyde. This study clarifies how tenderness drives flavour divergence and identifies compounds supporting sensory variation in green tea.

•Leaf tenderness reshaped the aroma and taste of green tea.•Bud tea featured a chestnut-like, clean aroma with higher amino acids and lower polyphenols.•Five common key volatiles were identified using ROAV and multivariate statistical analysis.•Geraniol, most abundant in one-bud-one-leaf tea, contributed to its prominent floral aroma.•Lipid, glycoside, and amino acid pathways explained aroma differentiation.

Leaf tenderness reshaped the aroma and taste of green tea.

Bud tea featured a chestnut-like, clean aroma with higher amino acids and lower polyphenols.

Five common key volatiles were identified using ROAV and multivariate statistical analysis.

Geraniol, most abundant in one-bud-one-leaf tea, contributed to its prominent floral aroma.

Lipid, glycoside, and amino acid pathways explained aroma differentiation.

Figures

Sensory evaluation of green tea samples with three levels of leaf tenderness: C1B (single bud), C1B1L (one bud with one leaf), and C1B2L (one bud with two leaves). (A) Visual comparison of dried tea leaves, tea infusions, and infused leaves (top to bottom). (B) Radar chart of traditional Chinese sensory evaluation scores across five attributes: dry tea appearance, liquor colour, aroma, taste, and infused leaf appearance. (C) Quantitative descriptive analysis (QDA) scores for eight aroma attributes: floral, tender, sweet, clean, chestnut-like, fresh, rich, and harmonious. (D) QDA scores for eig

Chemical composition of green tea samples with different leaf tenderness levels. (A) Moisture content. (B) Water extract content. (C) Caffeine content. (D) Free amino acid content. (E) Total polyphenol content. (F) Contents of four sweet-tasting amino acids (mg/g). (G) Contents of six umami-tasting amino acids (mg/g). (H) Contents of eight bitter-tasting amino acids (mg/g). (I) Total contents of sweet-, umami-, and bitter-type amino acids (mg/g). (J) Contents of non-ester-type catechins. (K) Contents of ester-type catechins. (L) Gallic acid content. (M) Total contents of non-ester-type and est

Multivariate analysis of volatile compounds in green tea samples with different leaf tenderness. (A–C) OPLS-DA score plots for pairwise comparisons: (A) C1B vs. C1B1L, (B) C1B vs. C1B2L, and (C) C1B1L vs. C1B2L. (D—F) Volcano plots showing significantly up-regulated (orange) and down-regulated (blue) volatile metabolites in: (D) C1B vs. C1B1L, (E) C1B vs. C1B2L, and (F) C1B1L vs. C1B2L. (G–I) Bar plots of the top 10 up-regulated and down-regulated volatile compounds based on log₂ fold change in: (G) C1B vs. C1B1L, (H) C1B vs. C1B2L, and (I) C1B1L vs. C1B2L. Only 9 down-regulated compounds were

Analysis of five key aroma-active volatiles. (A–E) Box plots showing the concentrations (μg/kg) of five key compounds across the three green tea samples. Different letters indicate significant differences at *p* < 0.05. (F) Aroma contribution mapping of the five volatiles using a Sankey diagram, grouped into three chemical categories: Alcohols (1-octen-3-ol, geraniol), Aldehydes (hexanal, benzeneacetaldehyde), and Heterocyclic compounds (2-pentylfuran). (For interpretation of the references to colour in this figure legend, the reader is referred to the web version of this article.)Fig. 5

Pearson correlation heatmap of key volatile compounds and sensory attributes. (Note: Red indicates positive correlation, blue indicates negative correlation; the colour intensity reflects the magnitude of the correlation coefficient). (For interpretation of the references to colour in this figure legend, the reader is referred to the web version of this article.)Fig. 6

Proposed biosynthetic pathways of five key aroma compounds in green teas with varying leaf tenderness. (A) Fatty acid–derived volatiles (FADVs); (B) Glycosidically bound volatiles (GVs); (C) Amino acid–derived volatiles (AADVs). Heatmaps represent relative abundance levels across samples, with colour gradients from blue (low) to red (high). LOX: lipoxygenase; ADH: alcohol dehydrogenase; GPP: geranyl diphosphate; PPO: geraniol synthase. (For interpretation of the references to colour in this figure legend, the reader is referred to the web version of this article.)Fig. 7

Full text

Introduction

1

Green tea, as one of the most widely consumed non-fermented teas globally, is highly valued for its distinctive sensory characteristics and health-promoting properties, primarily attributed to its abundant polyphenols and amino acids (Zhao, Li, Wang, & Song, 2022). The overall quality of green tea is strongly influenced by its flavour profile, encompassing both aroma and taste. Chinese green teas exhibit notable diversity in aroma types, ranging from chestnut-like (Xinyang Maojian, Henan) (Zhu et al., 2018), floral (Taiping Houkui, Anhui) (Jin et al., 2024), and bean-like (Longjing, Zhejiang) (Zhang et al., 2024; Zhang et al., 2024), to fresh and tender (Zhuyeqing, Sichuan) (Liu et al., 2025; Liu et al., 2025). Similarly, taste profiles vary from fresh and mellow (Taiping Houkui, Anhui) (Huang et al., 2024; Huang et al., 2024), sweet and mellow (Enshi Yulu, Hubei) (Guo et al., 2023), to strong and thick (Mengding Ganlu, Sichuan) (Zhao et al., 2024). Such diverse sensory attributes are closely associated with variations in chemical composition and manufacturing techniques.

The distinct flavour characteristics of green tea are largely determined by specific volatile and non-volatile metabolites. Volatile compounds, including alcohols, aldehydes, ketones, esters, and terpenoids, are key contributors to aroma (Yin et al., 2022; Yin et al., 2022). Non-volatile compounds such as amino acids, catechins, caffeine, and soluble sugars significantly influence taste attributes (Moreira et al., 2024). The formation of these compounds is a consequence of intrinsic biochemical processes in fresh tea leaves and subsequent metabolic transformations occurring during critical processing steps, including spreading, fixation, rolling, shaping, and drying (Zhang, Zhao, Lv, Qiu, & Tian, 2025). Therefore, clarifying the chemical pathways and factors governing tea flavour formation has become a fundamental research area, essential for sensory quality evaluation, geographic authentication, and processing standardisation.

Leaf tenderness, typically defined by the bud-to-leaf ratio (e.g., single bud, one bud with one leaf, or one bud with two leaves), is considered one of the most critical factors influencing green tea flavour quality. Younger leaves and buds generally contain elevated levels of free amino acids, glycoside-bound volatiles, and unsaturated lipids, while mature leaves have higher concentrations of structural polyphenols, cellulose, and fiber components (Xu et al., 2021). These compositional differences affect not only substrate availability for flavour-related enzymatic reactions but also the thermal stability of key metabolites during processing (Li et al., 2021). Consequently, teas made from raw materials of different tenderness exhibit notably distinct flavour profiles. For example, single-bud teas such as Zhuyeqing often present slender, attractive appearances with fresh and tender aroma and taste (Liu, Huang, et al., 2025; Liu, Ren, et al., 2025). In comparison, teas traditionally produced from one bud with one leaf (Longjing and Biluochun) generally offer balanced chestnut-like or floral aromas with mellow and umami-rich tastes, while teas composed of one bud with two leaves typically exhibit stronger and fuller-bodied infusions with lingering aftertaste. Despite the recognised importance of leaf tenderness for quality classification and consumer preferences, the precise chemical basis underlying tenderness-related differences in green tea flavour is not fully understood.

Recent research has substantially advanced the understanding of flavour compound formation pathways in green tea. Volatile formation mechanisms primarily involve lipid oxidation via the lipoxygenase (LOX) pathway, glycoside hydrolysis, and amino acid degradation, resulting in the generation of aroma-active compounds such as aldehydes, alcohols, ketones, esters, and terpenoids (Zhai, Zhang, Granvogl, Ho, & Wan, 2022). In contrast, taste-active metabolites—including free amino acids, catechins, caffeine, and soluble sugars—arise from biochemical processes intrinsic to fresh leaves or are subsequently modified through enzymatic and Maillard-type reactions during tea manufacture (Luo, Zhang, Ho, & Li, 2022). Advanced analytical approaches such as gas chromatography–mass spectrometry (GC–MS), high-performance liquid chromatography (HPLC), and liquid chromatography–tandem mass spectrometry (LC–MS/MS) have been widely employed for detailed chemical profiling of volatile and non-volatile compounds (Liu, Huang, et al., 2025; Liu, Ren, et al., 2025). Complementary sensory evaluation methods—including quantitative descriptive analysis (QDA) and relative odour activity value (ROAV) calculations—enable integration of chemical data with sensory perception, facilitating precise identification of aroma and taste contributors (Wang, Deng, Huang, & Ning, 2024).

To systematically investigate the chemical basis underlying tenderness-dependent flavour variations, this study utilised Camellia sinensis cv. Chuancha No. 2 (CC2), a cultivar specifically selected for its characteristically high theanine content, known to significantly enhance umami taste and overall sensory quality of green tea infusions. Three distinct tenderness levels—single bud (C1B), one bud with one leaf (C1B1L), and one bud with two leaves (C1B2L)—were processed into green teas under controlled manufacturing conditions. Sensory evaluation, including traditional assessment and QDA, was combined with instrumental analysis methods such as GC–MS-based volatile profiling, HPLC determination of taste compounds, and ROAV calculation to comprehensively characterise flavour-related metabolites. Multivariate statistical analyses and pathway interpretation were further applied to elucidate key aroma-active compounds and their corresponding formation mechanisms. This integrated analytical approach aims to clarify the chemical drivers behind tenderness-associated flavour differentiation, providing a theoretical foundation for raw material grading, cultivar selection, and targeted processing optimisation in green tea production.

Materials and methods

2

Chemicals and reagents

2.1

Standards for non-volatile compound analysis—including gallic acid, caffeine, and catechins (Epigallocatechin gallate (EGCG), Epicatechin gallate (ECG), Epigallocatechin (EGC), Epicatechin (EC), Catechin (C), Catechin gallate (CG), Gallocatechin gallate (GCG), Gallocatechin (GC))—were purchased from Sigma-Aldrich (USA), the National Institute of Standard Materials (China), or Chengdu Alfa Biotechnology Co., Ltd. (China). A mixed amino acid standard containing 17 amino acids and theanine was obtained from Shanghai Yuanye Bio-Technology Co., Ltd. (China). AccQ-Fluor reagents and Eluent A Concentrate were supplied by Waters Corporation (USA). For volatile compound analysis, GC-grade standards of geraniol, benzyl alcohol, 1-octen-3-ol, hexanal, 2-pentylfuran, and a series of n-alkanes (C8–C25) were obtained from Sigma-Aldrich, TCI, and Aladdin. Ethyl decanoate was used as the internal standard. SPME was performed using a DVB/CAR/PDMS fiber (50/30 μm, 1 cm) and manual holder from Supelco (USA). All chemical compounds used in this study were of HPLC grade or higher. Ultrapure water was prepared using a Milli-Q purification system (Millipore, USA).

Tea sample preparation

2.2

Fresh shoots of Camellia sinensis cv. ‘CC2’ were manually plucked in late March 2024 from a commercial tea garden at Yizhichun Tea Co., Ltd. (Leshan, Sichuan, China). According to tenderness, the raw materials were graded into three standards: C1B (single bud), C1B1L (one bud with one leaf), and C1B2L (one bud with two leaves). All samples were processed into green tea on-site using a uniform protocol comprising withering, fixation (pan-firing at 220–240 °C), rolling, and drying. The finished dry tea was sealed and stored at −20 °C until analysis.

Sensory evaluation

2.3

Five professionally trained panelists, each with prior experience in green tea grading, participated in the sensory analysis. The evaluation included both conventional scoring and quantitative descriptive analysis (QDA). For the conventional assessment, 3 g of tea was brewed in 150 mL of freshly boiled water (100 °C) for 5 min according to GB/T 23776-2018. Each sample was scored on a 10-point scale for five attributes: the appearance of dry leaves, liquor colour, aroma, taste, and the appearance of infused leaves. The final sensory score was calculated as a weighted sum, where aroma and dry leaf appearance each accounted for 25%, taste for 30%, liquor colour for 10%, and infused leaf appearance for the remaining 10%. Radar charts were generated to visualise the scoring results. Sensory evaluation followed China's National Standard GB/T 23776-2018. A panel of five certified and experienced tea tasters conducted the assessment after standardized training. All panelists met health and vision requirements, and hygienic procedures were strictly maintained throughout the evaluation to ensure reliability.

QDA was applied to quantify specific sensory traits. Eight aroma attributes (floral, tender, sweet, clean, chestnut-like, fresh, rich, and harmonious) and eight taste attributes (fresh, sweet, brisk, aftertaste, strong, thick, bitter, and astringent) were evaluated. All attributes were rated on a 10-point intensity scale (0 = none, 10 = extremely strong). Samples were labelled with random codes and presented in randomised order to minimise bias. Each sample was evaluated three times, and the mean values were used for analysis.

Determination of main quality indicators and taste-related compounds

2.4

Representative non-volatile quality components of the tea samples—including moisture, water extractives, total polyphenols, caffeine, catechins, gallic acid, and free amino acids—were determined using standard methods with minor adaptations (Liu, Huang, et al., 2025; Liu, Ren, et al., 2025). Moisture and water extract contents were measured in accordance with GB/T 8304-2013 and GB/T 8305-2013. Total polyphenols were quantified using the Folin–Ciocalteu method and expressed as gallic acid equivalents (GAE). Free amino acids were analysed on a Waters HPLC system using AccQ-Fluor™ derivatization kits, following the manufacturer's instructions. For catechins and caffeine analysis, 0.2 g of finely ground tea was extracted twice with 70% methanol (5 mL per extraction) at 70 °C for 10 min. The pooled extracts were filtered through a 0.45 μm membrane and adjusted to 10 mL. Chromatographic separation was performed using an Agilent 1260 Infinity II HPLC system (Agilent Technologies, USA) with a binary solvent system: solvent A (9% acetonitrile, 2% acetic acid, and 0.2% EDTA-2Na) and solvent B (80% acetonitrile, 2% acetic acid, and 0.2% EDTA-2Na). The flow rate was maintained at 1.0 mL/min, the column temperature at 35 °C, and detection was carried out at 278 nm. All determinations were carried out in triplicate. Results were expressed either as percentages on a dry weight basis or in mg/g, depending on the compound class.

GC–MS analysis of volatile compounds

2.5

Volatile compounds were analysed using headspace solid-phase microextraction (HS-SPME) coupled with gas chromatography–mass spectrometry (GC–MS). Briefly, 0.05 g of powdered tea sample was weighed into a 20 mL headspace vial, and 1 μL of ethyl decanoate (1 ppm in methanol) was added as the internal standard. The vial was sealed and subjected to automated HS-SPME extraction using a DVB/CAR/PDMS fiber (50/30 μm, 1 cm; Supelco, USA). GC–MS analysis was performed on a Trace 1300 gas chromatograph coupled to an ISQ 3000 mass spectrometer (Thermo Fisher Scientific, USA). Separation was achieved on a TG-5SilMS capillary column (30 m × 0.25 mm i.d., 0.25 μm film thickness). The temperature program followed the method reported in a previous study (Cui et al., 2022). Mass spectra were recorded in EI mode (70 eV) with a scan range of m/z 35–450. All analyses were conducted in triplicate.

ROAV calculation

2.6

The relative odour activity value (ROAV) was used to estimate the contribution of individual volatile compounds to the overall aroma perception. The ROAV for each compound was calculated based on its concentration and corresponding odour threshold (OT) in water, using the following equation as described by Xiao et al. (2022): ROAV_i_ = 100 × (C_i/OTi)/(Cmax/OTmax). Ci_ and OTi represent the concentration and odour threshold of compound i, respectively. The compound with the highest odour activity was set as the reference (ROAV = 100). Volatiles with ROAV values greater than 1 were considered aroma-active contributors in the tea samples.

Statistical analysis

2.7

All experiments were performed in triplicate, and results were expressed as mean ± standard deviation (SD). One-way analysis of variance (ANOVA) with Tukey's post hoc test was used to assess significant differences among groups (p < 0.05), using SPSS Statistics 26.0 (IBM Corp., Armonk, NY, USA). Principal component analysis (PCA), orthogonal partial least squares discriminant analysis (OPLS-DA), and heatmaps were performed using MetaboAnalyst 6.0 (https://www.metaboanalyst.ca). Box plots and other visualisations were generated in OriginPro 2021 (OriginLab Corp., USA). ROAV-based river plots were created using TBtools.

Results and discussion

3

Sensory evaluation

3.1

The sensory profiles of the three green tea samples—C1B (single bud), C1B1L (one bud with one leaf), and C1B2L (one bud with two leaves)—were systematically evaluated using both traditional Chinese sensory evaluation and quantitative descriptive analysis (QDA), as presented in Fig. 1 and Table S1.Fig. 1. Sensory evaluation of green tea samples with three levels of leaf tenderness: C1B (single bud), C1B1L (one bud with one leaf), and C1B2L (one bud with two leaves). (A) Visual comparison of dried tea leaves, tea infusions, and infused leaves (top to bottom). (B) Radar chart of traditional Chinese sensory evaluation scores across five attributes: dry tea appearance, liquor colour, aroma, taste, and infused leaf appearance. (C) Quantitative descriptive analysis (QDA) scores for eight aroma attributes: floral, tender, sweet, clean, chestnut-like, fresh, rich, and harmonious. (D) QDA scores for eight taste attributes: fresh, sweet, brisk, aftertaste, strong, thick, bitter, and astringent. (For interpretation of the references to colour in this figure legend, the reader is referred to the web version of this article.)Fig. 1

As shown in Fig. 1A, distinct differences in morphology and infusion characteristics were observed across the three tenderness levels. C1B exhibited flat, straight, jade-green dry leaves with a lustrous surface and slightly exposed pekoe. The liquor appeared bright and tender green, while the infused leaves remained vibrant and uniform. With the inclusion of one or two leaves, both C1B1L and C1B2L showed increased leaf size and decreased uniformity. In particular, C1B2L displayed darker, curled dry leaves, deeper yellow-green liquor, and less vivid infused leaves, suggesting that increasing leaf maturity compromised the overall visual quality.

Traditional sensory scoring (Fig. 1B, Table S1) further highlighted these trends. C1B1L achieved the highest total score (9.4), outperforming C1B (9.3) and C1B2L (9.1). Notably, C1B1L excelled in aroma (9.8) and liquor colour (9.5), with a flavour described as “savory and mellow, accompanied by floral freshness.” C1B showed consistently good performance across all indices, particularly in appearance (9.1), aroma (9.5), and flavour (9.2). Its aroma was characterised as “clean and chestnut-like with refreshing notes,” while its taste was brisk and smooth. In contrast, C1B2L received slightly lower scores in appearance and taste, reflecting a shift toward a darker infusion and a stronger, more astringent profile dominated by roasted aroma.

To further dissect specific sensory attributes, QDA was employed and results are shown in Fig. 1C and D. For aroma (Figs. 1C), C1B was marked by high scores in tender (9.0), sweet (8.0), and rich (7.0) notes, reflecting its delicate and layered aromatic profile. C1B1L displayed the most distinctive floral note (9.0) among all samples, as well as strong freshness (9.0) and tenderness (7.0), indicating a floral-dominant style with youthful and vibrant characteristics. In contrast, C1B2L showed a significantly different profile, with a prominent chestnut-like aroma (8.0) and reduced tenderness and sweetness (6.0–7.0), suggesting a transformation toward a warmer, roasted aromatic structure as leaf maturity increased.

The influence of leaf tenderness on taste was equally apparent (Figs. 1D). C1B demonstrated a mild and balanced taste, with moderate freshness (7.0), sweetness (8.0), and aftertaste (8.0). While it did not lead in any particular descriptor, it showed the lowest strength (7.0), indicating a clean and light mouthfeel with limited bitterness and astringency. C1B1L, in contrast, exhibited the most favorable taste attributes overall, with the highest scores in freshness (8.0), sweetness (8.5), harmony (9.0), and aftertaste (8.5). This profile suggests a well-integrated, mellow, and refreshing flavour. C1B2L displayed a notably robust taste, with elevated strength (9.0), bitterness (7.0), astringency (7.0), and thickness (8.0), but diminished sweetness (5.0) and harmony (8.0), reflecting a shift toward intensity and fullness at the expense of balance and delicacy.

Determination of major quality indicators and taste-related compounds

3.2

Taste is a key determinant of tea quality and consumer acceptance. To clarify the material basis underlying the differences in taste performance, we conducted qualitative and quantitative analyses of major quality indicators and taste-related components in the three green tea samples (C1B, C1B1L, and C1B2L). The evaluated parameters included moisture content, water extract content, total polyphenols, caffeine, eight major catechins, gallic acid, and 18 free amino acids (Fig. 2).Fig. 2. Chemical composition of green tea samples with different leaf tenderness levels. (A) Moisture content. (B) Water extract content. (C) Caffeine content. (D) Free amino acid content. (E) Total polyphenol content. (F) Contents of four sweet-tasting amino acids (mg/g). (G) Contents of six umami-tasting amino acids (mg/g). (H) Contents of eight bitter-tasting amino acids (mg/g). (I) Total contents of sweet-, umami-, and bitter-type amino acids (mg/g). (J) Contents of non-ester-type catechins. (K) Contents of ester-type catechins. (L) Gallic acid content. (M) Total contents of non-ester-type and ester-type catechins. Different letters indicate significant differences among samples at p < 0.05. (For interpretation of the references to colour in this figure legend, the reader is referred to the web version of this article.)Fig. 2

Significant differences were observed in moisture content, with C1B showing the highest level (5.12%), followed by C1B2L (4.18%) and C1B1L (4.13%). In contrast, the aqueous extract content showed no significant difference among the three samples, with values ranging between 40.20% and 41.23%. Caffeine content was significantly higher in C1B (4.44%) compared with C1B1L (3.99%) and C1B2L (4.07%). For total free amino acids, both C1B (5.03%) and C1B1L (5.13%) contained significantly higher levels than C1B2L (4.57%, p < 0.05). Regarding total polyphenols, C1B1L had the highest content (24.45%), which was significantly higher than that of C1B (21.20%) and C1B2L (19.70%).

Polyphenols are primarily responsible for bitterness and astringency, whereas free amino acids contribute to umami and freshness. The high amino acid content and relatively lower polyphenol concentration in C1B likely explain its clean, refreshing, and less astringent taste as perceived during sensory evaluation. In contrast, C1B1L exhibited the highest levels of both polyphenols and amino acids. Previous studies have shown that caffeine and polyphenols can interact through hydrogen bonding, π–π stacking, and CH–π interactions to form complexes that reduce bitterness and enhance the smoothness of tea infusions(Ishizu et al., 2014). This may account for the strong, fresh, and well-balanced taste profile observed in C1B1L. On the other hand, the reduced amino acid content in C1B2L could be responsible for its lower freshness scores, and in combination with moderate levels of caffeine and polyphenols, it produced a taste profile characterised by strength and mellowness, but with diminished brightness.

To further characterise the taste differences among the three green tea samples, the 18 free amino acids were categorised into three functional groups based on their flavour properties: sweet amino acids (Ser, Ala, Pro, Thr), umami amino acids (Asp, Glu, The, GABA, Asn, Gln), and bitter amino acids (Arg, His, Ile, Leu, Phe, Lys, Tyr, Val) (Yu et al., 2020; Zhao et al., 2024).

As shown in Fig. 2F, C1B contained the highest levels of the sweet amino acids Ser, Ala, and Thr, with significantly higher concentrations than those in C1B1L and C1B2L. Interestingly, Pro was only detected in C1B2L at a low level, while it was not detectable in C1B or C1B1L. For umami amino acids (Fig. 2G), C1B showed significantly higher levels of Asp, Glu, The, and Asn than the other two groups, indicating a stronger contribution to the umami and fresh taste. Of particular note is theanine, the most abundant free amino acid in tea (Lin et al., 2023). Its content in C1B, C1B1L, and C1B2L was 21.87 mg/g, 10.44 mg/g, and 14.33 mg/g, respectively, making it the dominant amino acid in all three samples. Theanine is a non-proteinaceous amino acid strongly associated with the umami and briskness of green tea infusions (Gong et al., 2020; Liu, Li, Huang, Liu, & Xiong, 2021), and its content typically correlates positively with leaf tenderness. This is consistent with our findings, where C1B (single bud) exhibited the highest theanine level. For bitter amino acids (Fig. 2H), C1B showed significantly higher levels of Val, Leu, Ile, His, Lys, and Tyr than the other samples. Ile was not detected in C1B1L, and His was exclusively detected in C1B. In contrast, Phe content was highest in C1B1L and significantly exceeded that in C1B and C1B2L. Arg content was most abundant in C1B2L and was significantly higher than in the other two samples. Collectively, individual amino acids within each flavour category exhibited notable variation among the three samples. When summed by taste type, the total contents of sweet-, umami-, and bitter-type amino acids all followed the same descending trend: C1B > C1B2L > C1B1L. The relative abundance of these key components interacts to shape the overall taste profile.

Catechins are the most abundant polyphenolic compounds in tea, accounting for approximately 70% of the total polyphenol content (Liang et al., 2010), and are key contributors to green tea taste. Based on their chemical structure, catechins can be classified into non-ester-type (C, EC, GC, EGC) and ester-type (CG, GCG, ECG, EGCG) catechins (Zhou et al., 2024), with the esterified forms contributing more strongly to bitterness and astringency (Wang et al., 2024; Wang et al., 2024; Wang et al., 2024; Wang et al., 2024; Wang, Deng, et al., 2024).

Among the non-ester catechins (Fig. 2J), the content of catechin (C) was significantly higher in C1B (0.23%) and C1B1L (0.23%) compared to C1B2L (0.17%). Epicatechin (EC) content showed significant differences among all three groups, with values of 0.54% in C1B, 1.02% in C1B1L, and 0.83% in C1B2L. GC levels followed the order C1B2L (0.32%) > C1B1L (0.27%) > C1B (0.14%), with significant pairwise differences. EGC, the most abundant among non-ester catechins, was highest in C1B1L (4.43%), followed by C1B (4.02%) and C1B2L (3.46%), and all differences were statistically significant. For ester-type catechins (Fig. 2K), CG content was highest in C1B2L (0.24%), followed by C1B1L (0.18%) and C1B (0.11%), with C1B significantly lower than the other two. Interestingly, GCG content was highest in C1B (2.26%), markedly exceeding that in C1B2L (0.61%) and C1B1L (0.42%), which also differed significantly from each other. ECG levels decreased progressively with maturity: C1B (4.11%) > C1B1L (3.36%) > C1B2L (2.26%), with all pairwise differences being statistically significant. EGCG, the most abundant and biologically active catechin in green tea (Bandele & Osheroff, 2008), showed its highest concentration in C1B1L (12.57%), which was significantly higher than in C1B (7.53%) and C1B2L (10.26%). Gallic acid (GA), a key structural component of ester-type catechins, is also known to enhance taste complexity and contribute subtle umami properties (Zhang, Cao, Granato, Xu, & Ho, 2020). GA content was significantly higher in C1B compared to the other two samples (Fig. 2L), between which no significant difference was observed. The total content of non-ester-type catechins (Fig. 2M) followed the order: C1B1L (5.95%) > C1B (4.93%) > C1B2L (4.78%). A similar trend was observed for ester-type catechins, with C1B1L showing the highest level (16.53%), followed by C1B (14.02%) and C1B2L (13.38%). Notably, this pattern was consistent with the distribution of total polyphenols across the samples (Fig. 2E). Ester-type catechins, especially EGCG and ECG, are the main contributors to green tea bitterness and astringency, while non-ester catechins have negligible impact (Xu et al., 2018). C1B, characterised by lower ester catechins, higher amino acids, and elevated gallic acid, showed a clean and fresh taste with reduced bitterness. In contrast, C1B1L exhibited a higher polyphenol-to-amino acid ratio, contributing to greater bitterness and thickness. Despite its lower sweet and umami amino acid levels, its relatively high non-ester catechins may enhance aftertaste sweetness (Zhang et al., 2020). Soluble sugars, though not measured here, may also play a role (Li et al., 2022).

Identification and quantification of volatile compounds in green teas by GC–MS

3.3

To investigate the aroma profiles of green teas, a comprehensive volatile analysis was performed using gas chromatography–mass spectrometry (GC–MS). A total of 97 volatile compounds were identified across the three samples—C1B, C1B1L, and C1B2L—including 1 acid, 15 alcohols, 10 aldehydes, 2 esters, 3 heterocyclics, 17 alkanes, 31 alkenes, 10 ketones, 3 aromatics, and 5 others (Table S2). Fig. S1A–C illustrates the classification and proportional distribution of these volatiles in three green teas. In C1B (Fig. 3A), alkanes (26.88%), alcohols (21.18%), alkenes (19.17%), ketones (13.22%), and aldehydes (12.19%) were dominant. In C1B1L (Fig. 3B), alkenes (33.77%), alcohols (22.14%), and alkanes (21.89%) constituted the main volatile classes. For C1B2L, the primary categories included alkanes (26.50%), alkenes (25.88%), alcohols (15.86%), ketones (12.11%), and aldehydes (10.47%). Among these, hydrocarbons (alkanes and alkenes) and alcohols collectively accounted for 67.23%, 77.80%, and 68.24% of total volatiles in C1B, C1B1L, and C1B2L, respectively, indicating their central role in the aroma composition of green tea. Alkenes, in particular, are known to be crucial for the fresh and green-like aroma of green tea and may also act as precursors for other aromatic compounds.Fig. 3(A) Heatmap of 97 volatiles across the three leaf tenderness levels. Colour gradients from blue to red indicate increasing compound concentration. From the centre outward, sample groups are arranged as C1B, C1B2L, and C1B1L. Hierarchical clustering was performed on both samples and compounds. (B) PCA scores plot based on the relative abundance of volatile compounds. (C) PCA biplot, where each arrow represents an original variable (volatile compound concentration). The direction and length of the arrows indicate the strength and contribution of each compound to sample separation; longer arrows represent greater discriminatory power. (For interpretation of the references to colour in this figure legend, the reader is referred to the web version of this article.)Fig. 3

The most abundant compound in C1B was 2,3-octanedione (520.79 μg/kg), known for its sweet and creamy scent. This diketone has been reported as a significant contributor to the fresh aroma in Longjing green tea (Zhang, Mao, et al., 2024; Zhang, Zhang, et al., 2024). C1B1L was characterised by the highest level of δ-cadinene (300.63 μg/kg), a sesquiterpene with woody and herbal notes, and a known odour-active compound in Xinyang Maojian green tea with OAV > 1. C1B2L exhibited the highest content of (E)-4,8-dimethylnona-1,3,7-triene (411.07 μg/kg), a terpene compound previously detected in Yerba mate but not yet well-documented in green tea aroma studies (Yin, Kong, et al., 2022; Yin, Wang, et al., 2022). Its exact sensory contribution warrants further exploration (Machado et al., 2007).

To further explore the differences in volatile profiles among the three green tea samples, hierarchical clustering and principal component analysis (PCA) were performed based on the relative contents of 97 identified volatiles. As shown in the hierarchical clustering heatmap (Fig. 3A), samples clustered distinctly according to leaf tenderness. As shown in the hierarchical clustering heatmap (Fig. 3A), the three green tea samples were clearly grouped according to leaf tenderness. C1B formed a distinct cluster, indicating a unique volatile profile with generally higher overall contents. C1B1L also clustered independently, characterised by moderate to high levels of multiple volatiles. In contrast, C1B2L samples grouped together and were positioned closer to C1B, suggesting partial compositional similarity, yet with comparatively lower overall volatile content. The PCA scores plot (Fig. 3B) revealed clear separation among the three groups along the first two principal components, PC1 (45%) and PC2 (27%), which together explained 72% of the total variance. C1B was distributed primarily in the upper-left quadrant, indicating distinct volatile characteristics. C1B1L samples clustered in the upper-right quadrant, while C1B2L was mainly located in the lower-central region, closer to C1B, reflecting their relative chemical proximity. The corresponding biplot (Fig. 3C) illustrated the key volatiles contributing to group separation. C1B was strongly associated with α-phellandrene, β-maaliene, and acetoin, which loaded negatively on PC1 and positively on PC2. C1B1L was linked to geraniol and di-epi-1,10-cubenol, both with high positive loadings on PC1 and PC2. For C1B2L, (E)-4,8-dimethylnona-1,3,7-triene and γ-terpinene showed strong negative contributions to PC1, indicating their significant role in driving the volatile differences of the most mature leaves.

Differential volatile metabolite analysis

3.4

To further investigate aroma-related differences among green teas with varying leaf tenderness, orthogonal partial least squares discriminant analysis (OPLS-DA) was conducted based on the relative concentrations of the 97 identified volatile compounds. To validate model robustness and avoid overfitting, 200 permutation tests were performed for each pairwise comparison. As shown in Fig. 4A–C, clear separation was achieved among all sample groups, indicating substantial variation in volatile profiles across tenderness levels.Fig. 4. Multivariate analysis of volatile compounds in green tea samples with different leaf tenderness. (A–C) OPLS-DA score plots for pairwise comparisons: (A) C1B vs. C1B1L, (B) C1B vs. C1B2L, and (C) C1B1L vs. C1B2L. (D—F) Volcano plots showing significantly up-regulated (orange) and down-regulated (blue) volatile metabolites in: (D) C1B vs. C1B1L, (E) C1B vs. C1B2L, and (F) C1B1L vs. C1B2L. (G–I) Bar plots of the top 10 up-regulated and down-regulated volatile compounds based on log₂ fold change in: (G) C1B vs. C1B1L, (H) C1B vs. C1B2L, and (I) C1B1L vs. C1B2L. Only 9 down-regulated compounds were identified in (I), and are thus shown accordingly. (For interpretation of the references to colour in this figure legend, the reader is referred to the web version of this article.)Fig. 4

In the C1B vs. C1B1L model (Fig. 4A), the explained variation and predictive ability were high (R^2^X = 0.728, R^2^Y = 0.969, Q^2^ = 0.955), demonstrating strong model performance. Similarly, the C1B vs. C1B2L model (Fig. 4B) yielded R^2^X = 0.432, R^2^Y = 0.988, Q^2^ = 0.788, and the C1B1L vs. C1B2L model (Fig. 4C) showed R^2^X = 0.528, R^2^Y = 0.957, Q^2^ = 0.865. These results confirm that the OPLS-DA models were statistically sound and exhibited good predictive capacity. Differential metabolites were screened based on variable importance in projection (VIP > 1), fold change (FC > 2 or < 0.5), and p < 0.05 (Huang et al., 2022). Volcano plots (Fig. 4D–F) display the distribution of significant volatile differences in three pairwise comparisons: (D) C1B vs. C1B1L, (E) C1B vs. C1B2L, and (F) C1B1L vs. C1B2L. In total, 48 differential volatiles were identified between C1B and C1B1L (Table S3), 24 between C1B and C1B2L (Table S4), and 20 between C1B1L and C1B2L (Table S5).

To identify volatiles most responsible for aroma differentiation among the three green tea samples, the top 10 up-regulated and down-regulated compounds (based on log₂ fold change) were selected for each pairwise comparison (Fig. 4G–I).

In both C1B vs C1B1L (Fig. 4G) and C1B vs C1B2L (Fig. 4H), compounds related to green and fresh notes were consistently up-regulated in C1B, including α-phellandrene (log_2_FC = 7.71/6.51), (Z)-2-heptenal (3.29/2.42), 2,3-octanedione (3.12/−), hexanal (2.07/2.19), 1-penten-3-ol (3.34/−), acetoin (1.90/1.88), and 1-octen-3-ol (2.21/2.01). These compounds are mainly derived from the lipoxygenase (LOX) pathway and monoterpene biosynthesis, which may be more active in tender buds due to higher enzymatic activity (Ho, Zheng, & Li, 2015; Huang, Li, et al., 2024; Huang, Tao, et al., 2024). Their enrichment contributes to the characteristic herbaceous and brisk aroma of C1B. Conversely, floral and woody volatiles were down-regulated in C1B and enriched in the more mature samples. In C1B1L (Fig. 4G), monoterpenes such as trans-β-ocimene (−2.23), geraniol (−2.10), and (−)-β-bourbonene (−3.27) as well as sesquiterpenes like nerolidol (−9.88), α-farnesene (−11.81), and γ-elemene (−2.75) were more abundant, suggesting a shift toward floral–woody elegance. Similarly, C1B2L (Fig. 4H) exhibited higher levels of cis-linalool-3,6-oxide (−1.93), cis-jasmone (−2.05), trans-linalool-3,6-oxide (−2.10), cis-linalool-3,7-oxide (−2.77), γ-elemene (−3.12), (−)-β-bourbonene (−3.68), and α-farnesene (−12.06), indicating enhanced linalool oxidation and sesquiterpene accumulation associated with a floral–woody character.

The comparison between C1B1L and C1B2L (Fig. 4I) revealed that C1B1L was enriched in benzyl alcohol (log_2_FC = 6.18) and geraniol (1.24), reflecting sweet-floral and rose-like aroma. Cis-3-hexenyl-α-methylbutyrate (1.31), a fatty acid–derived fruity ester, contributed green-fruity freshness. Concurrent up-regulation of nerolidol (5.83), α-bulnesene (2.20), caryophyllene (1.29), and epicubenol (1.35) added woody-spicy-resinous depth to the aroma profile. In contrast, C1B2L was enriched in lipid oxidation intermediates such as 1-penten-3-ol (−1.57) and pentanal (−2.00), as well as ketones like 4-methyl-3-penten-2-one (−1.36). The decreased levels of β-cyclocitral (−1.06) and 1-ethyl-1H-pyrrole-2-carboxaldehyde, 1-ethyl- (−1.45) suggested a reduction in compounds responsible for fresh/green notes and weakened roasted-fruity top notes.

Hexanal and 1-octen-3-ol are key volatile compounds generated from polyunsaturated fatty acids via the LOX-catalyzed pathway. As tea leaves mature, LOX activity tends to diminish, leading to a significant reduction in lipid oxidation products. Specifically, LOX exhibits higher catalytic efficiency in tender leaves, where it effectively converts substrates such as linoleic and α-linolenic acids into these characteristic aroma compounds. Conversely, the decline in LOX activity in mature leaves appears to be the primary limiting factor for their synthesis. This is corroborated by previous research on ‘Fuding Dabai’ green tea, which demonstrated that LOX pathway activity peaks in bud samples and systematically decreases with advancing leaf maturity (from one bud/one leaf to one bud/two leaves). We have incorporated these points into the revised manuscript to strengthen the discussion on formation mechanisms (Liu, Huang, Tang, Li, Li, Xie, et al., 2025). In addition to enzymatic activity, the concentration of fatty acid precursors in tea leaves also fluctuates with maturation. In mature leaves, the composition of fatty acids undergoes shifts—specifically, a reduction in polyunsaturated fatty acids may occur—thereby restricting the substrate availability for lipid oxidation. This is supported by empirical evidence showing that the content of linoleic acid, the essential substrate for the formation of hexanal and 1-octen-3-ol, significantly decreases as leaf tenderness diminishes, reaching its peak in the tea buds. These findings further explain why the levels of these lipid-derived volatiles are lower in more mature tea samples. (Chi-Tang, Xin, & Shiming, 2015) (Qingyang et al., 2024). As tea leaves mature, the concentration of fatty acids—particularly the polyunsaturated fatty acids serving as precursors for hexanal and 1-octen-3-ol—declines, thereby limiting the production of these volatile compounds. Consequently, the observed reduction in hexanal and 1-octen-3-ol levels may be attributed to the concerted action of diminished lipoxygenase (LOX) activity and an insufficient supply of substrate fatty acids.

The balance between these two reaction systems is heavily influenced by processing methods like pan-firing, steaming, and roasting. During high-heat treatments, both Maillard and lipid oxidation reactions are accelerated; yet, they affect the pool of amino acids and sugars differently while potentially enhancing lipid degradation. As a result, different processing techniques alter the relative ratio of lipid oxidation to Maillard products, leading to the divergence of aroma characteristics in the final tea products. (Chi-Tang et al., 2015)。.

ROAV analysis of aroma-active compounds

3.5

Relative odour activity value (ROAV) is a semi-quantitative index used to identify compounds contributing significantly to aroma perception, based on both compound concentration and odour threshold. A compound with ROAV > 1 is generally considered aroma-active (Chen et al., 2025).

Among the 97 identified volatiles, 39 compounds with reported odour thresholds were included in ROAV calculations (Table S6). As shown in Table 1, 11 compounds exceeded the ROAV threshold in at least one of the three green tea samples, suggesting their potential contribution to the overall aroma. These included aldehydes (e.g., hexanal, octanal), alcohols (e.g., 1-octen-3-ol, phenylethyl alcohol), monoterpenes (e.g., geraniol, linalool), furans, and aromatic compounds. Linalool and octanal showed universally high ROAVs across all samples. Linalool, used as the reference compound in ROAV normalization, maintained a value of 100, while octanal, a fat- and citrus-like aldehyde, consistently scored above 70. In addition, several volatiles exhibited clear sample-specific ROAV patterns. Hexanal and 1-octen-3-ol showed the highest ROAVs in C1B (12.01 ± 0.35 and 10.13 ± 2.03, respectively), but declined sharply in C1B1L and C1B2L, aligning with the green and fresh profile of tender buds. In contrast, benzeneacetaldehyde peaked in C1B (6.50 ± 0.68) and C1B2L (5.07 ± 5.61), contributing sweet-floral and nutty nuances. Geraniol was most aroma-active in C1B1L (13.80 ± 0.26), supporting its role in floral enhancement at moderate leaf maturity (Wang et al., 2020).Table 1. Relative odour activity values (ROAVs) of 11 volatile compounds with ROAV > 1 in at least one sample.Table 1. NameCASChemical classOdour descriptionsThreshold (μg/L)ROAVC1BC1B1LC1B2LHexanal66-25-1AldehydeGrass, tallow, fat0.007512.013 ± 0.3483.701 ± 0.4022.502 ± 1.908Heptanal111-71-7AldehydeFat, citrus, rancid0.0311.435 ± 0.1281.659 ± 0.11.169 ± 0.766Octanal124-13-0AldehydeFat, soap, lemon, green0.000184.625 ± 5.24571.246 ± 5.23472.762 ± 34.919Benzeneacetaldehyde122-78-1AldehydeHawthorne, honey, sweet0.0096.5 ± 0.6843.05 ± 0.175.069 ± 5.612Nonanal124-19-6AldehydeFat, citrus, green0.0151.823 ± 1.6341.323 ± 1.120.946 ± 0.805p-Cymene99-87-6Monoterpene hydrocarbonSolvent, gasoline, citrus0.01331.136 ± 0.0840.893 ± 0.0741.006 ± 0.61Linalool78-70-6Monoterpene alcoholFlower, lavender0.0015100 ± 0100 ± 0100 ± 0Geraniol106-24-1Monoterpene alcoholRose, geranium0.00752.745 ± 1.31513.796 ± 0.2594.76 ± 4.502Phenylethyl Alcohol60-12-8Aromatic alcoholCorn flakes, floral, fruit, honey, rose0.0454.931 ± 0.5856.01 ± 0.3953.716 ± 2.5061-Octen-3-ol3391-86-4AlcoholMushroom0.00210.125 ± 2.0312.727 ± 0.7251.99 ± 0.8182-pentylfuran3777-69-3Heterocyclic compoundsGreen bean, butter0.00484.059 ± 0.1941.481 ± 0.4231.052 ± 0.111Note: Odour descriptions were retrieved from The Good Scents Company database (http://www.thegoodscentscompany.com/search2.html). Odour threshold values were obtained from the Leffingwell & Associates database (http://www.leffingwell.com/) and related literature sources.

Key aroma compounds contributing to sample differentiation

3.6

To further narrow down the volatiles most relevant to aroma differentiation, the 11 ROAV-active compounds were intersected with differentially expressed metabolites (VIP > 1, p < 0.05, FC > 2 or < 0.5). This screening yielded five key aroma-active compounds: 1-octen-3-ol, geraniol, hexanal, benzeneacetaldehyde, and 2-pentylfuran. Their concentrations across the three green teas are shown in Fig. 5A–E.Fig. 5. Analysis of five key aroma-active volatiles. (A–E) Box plots showing the concentrations (μg/kg) of five key compounds across the three green tea samples. Different letters indicate significant differences at p < 0.05. (F) Aroma contribution mapping of the five volatiles using a Sankey diagram, grouped into three chemical categories: Alcohols (1-octen-3-ol, geraniol), Aldehydes (hexanal, benzeneacetaldehyde), and Heterocyclic compounds (2-pentylfuran). (For interpretation of the references to colour in this figure legend, the reader is referred to the web version of this article.)Fig. 5

In C1B (bud-only tea), 1-octen-3-ol (30.33 μg/kg, ROAV = 10.13), hexanal (154.73 μg/kg, ROAV = 12.01), and 2-pentylfuran (25.80 μg/kg, ROAV = 4.06) were significantly enriched. These compounds impart mushroom-like, grassy, fatty, and roasted notes. Although mushroom notes are commonly associated with aged or fermented teas (Wang et al., 2025), Wei et al. (Wei et al., 2022) found that 1-octen-3-ol addition enhanced sweetness and umami in green tea. Furthermore, benzeneacetaldehyde was also abundant in C1B (81.08 μg/kg, ROAV = 6.50), contributing honeyed and floral sweetness that complemented the clean, mellow, and slightly chestnut-like aroma.

In C1B1L, the dominant floral contributor was geraniol (114.22 μg/kg, ROAV = 13.80), a monoterpene alcohol associated with rose and geranium notes (Li et al., 2024). Moderate hexanal (31.47 μg/kg) added subtle green freshness. Benzeneacetaldehyde was lower in this sample (30.30 μg/kg), suggesting floral rather than sweet notes dominated the profile (Wang, Deng, et al., 2024; Wang, Gao, et al., 2024; Wang, Liu, et al., 2024; Wang, Wang, et al., 2024; Wang, Xie, et al., 2024). As a result, C1B1L exhibited a floral–fresh aroma balance.

In C1B2L, benzeneacetaldehyde emerged as the most aroma-active compound (58.49 μg/kg, ROAV = 5.07 ± 5.61), imparting sweet, hawthorn-like and corn-sweet notes. Its importance is further supported by reports identifying benzeneacetaldehyde as a key contributor to the chestnut aroma of Longjing tea (Wang et al., 2019). Although geraniol (48.47 μg/kg) and 1-octen-3-ol (18.71 μg/kg) were present at intermediate levels, the cumulative effect of these compounds, along with hexanal (26.80 μg/kg) and 2-pentylfuran (13.49 μg/kg), gave C1B2L a sweet, mellow, and lightly chestnut-like aroma characteristic of slightly mature leaves.

Aroma flow mapping (Fig. 5F) further illustrates how key volatiles with distinct descriptors—such as grassy (hexanal), floral (geraniol), and sweet-honeyed (benzeneacetaldehyde)—differentially contribute to the characteristic profiles of each sample, highlighting the shift from green freshness in C1B to floral and sweet richness in C1B1L and C1B2L.

The correlation heatmap combined with hierarchical clustering revealed distinct association patterns between sensory attributes and volatile compounds (Fig. 6). Sweetness, fragrance and duration, clean-refreshing quality, and tenderness clustered closely and showed strong positive correlations with each other, indicating that these attributes collectively contribute to the overall palatability of the samples. Florality and fruitiness formed another highly correlated cluster and were strongly associated with geraniol, suggesting that this terpene alcohol is a key contributor to the floral–fruity aroma profile.Fig. 6. Pearson correlation heatmap of key volatile compounds and sensory attributes. (Note: Red indicates positive correlation, blue indicates negative correlation; the colour intensity reflects the magnitude of the correlation coefficient). (For interpretation of the references to colour in this figure legend, the reader is referred to the web version of this article.)Fig. 6

Moreover, 1-octen-3-ol, hexanal, and 2-pentylfuran grouped together with significant positive correlations, implying similar variation trends and possibly related formation pathways. Benzeneacetaldehyde displayed moderate positive correlations with florality and fruitiness, indicating a potential synergistic role in aroma perception. Overall, these results demonstrate that floral–fruity volatiles and sweetness-related sensory attributes are the primary factors shaping the flavour characteristics of the samples.

Formation for key aroma compounds

3.7

The formation of tea aroma involves multiple biochemical transformations during processing, primarily including degradation of fatty acids, amino acids, and glycosidically bound precursors, as well as reactions involving carotenoid cleavage, Maillard reaction, and microbial transformation (Wang, Deng, et al., 2024; Wang, Gao, et al., 2024; Wang, Liu, et al., 2024; Wang, Wang, et al., 2024; Wang, Xie, et al., 2024; Zhang et al., 2023). Among these, the five key aroma compounds identified in this study—hexanal, 1-octen-3-ol, 2-pentylfuran, geraniol, and benzeneacetaldehyde—are mainly derived from three well-established aroma-generating pathways, as illustrated in Fig. 7: fatty acid–derived volatiles (FADVs), glycosidically bound volatiles (GVs), and amino acid–derived volatiles (AADVs).Fig. 7. Proposed biosynthetic pathways of five key aroma compounds in green teas with varying leaf tenderness. (A) Fatty acid–derived volatiles (FADVs); (B) Glycosidically bound volatiles (GVs); (C) Amino acid–derived volatiles (AADVs). Heatmaps represent relative abundance levels across samples, with colour gradients from blue (low) to red (high). LOX: lipoxygenase; ADH: alcohol dehydrogenase; GPP: geranyl diphosphate; PPO: geraniol synthase. (For interpretation of the references to colour in this figure legend, the reader is referred to the web version of this article.)Fig. 7

Three of the five compounds—hexanal, 1-octen-3-ol, and 2-pentylfuran—originate from the degradation of linoleic acid, representing typical FADVs. In the LOX (lipoxygenase) pathway, linoleic acid is enzymatically cleaved to form hexanal, a compound with characteristic grassy and fatty notes. It showed the highest ROAV and concentration in C1B, contributing to the tea's fresh-green aroma. 1-octen-3-ol, derived via sequential LOX and alcohol dehydrogenase (ADH) catalysis from 1-octen-3-one, imparts mushroom-like and savory nuances. It was also most abundant in C1B, supporting its role in adding depth and umami to the brisk aroma of bud-only tea. 2-pentylfuran, though also linoleic acid–derived, is thought to form through amino acid–catalyzed oxidation under moderate thermal conditions, such as those occurring during fixation and drying (Wang, Deng, et al., 2024; Wang, Gao, et al., 2024; Wang, Liu, et al., 2024; Wang, Wang, et al., 2024; Wang, Xie, et al., 2024). Its highest accumulation in C1B may result from the combined effects of tender bud structure and exposure to heat during processing, which together facilitate its generation. The roasted and sweet-bean characteristics of this compound likely reinforce the mellow base beneath C1B's fresh top notes.

Geraniol, a representative GV, is released from its glycosidically bound form under heat or enzymatic hydrolysis. It is biosynthesized from geranyl diphosphate (GPP) via geraniol synthase, and may also arise from interconversion with linalool or nerol (Zhang et al., 2023). In this study, geraniol was significantly higher in C1B1L, distinguishing it from the other four key compounds which all peaked in C1B. The elevated geraniol content in C1B1L is closely aligned with its sensory characteristics, particularly the prominent floral freshness, indicating that geraniol plays a central role in defining the one bud–one leaf sample's aromatic identity.

Benzeneacetaldehyde, the only AADV among the five, is produced through Strecker degradation of phenylalanine. This non-enzymatic reaction involves Schiff base formation followed by decarboxylation and hydrolysis, ultimately generating benzeneacetaldehyde (Delgado, Hidalgo, & Zamora, 2015). Although amino acid–derived volatiles are abundant in tea, this was the only compound in this class that passed our integrated screening criteria. Known for its sweet and honey-like notes, benzeneacetaldehyde accumulated in both C1B and C1B2L, and is a documented contributor to the chestnut-like aroma of Longjing tea (Wang et al., 2019; Wang, Deng, et al., 2024; Wang, Gao, et al., 2024; Wang, Liu, et al., 2024; Wang, Wang, et al., 2024; Wang, Xie, et al., 2024), reflecting its broad impact across different tenderness levels.

Conclusion

4

This study investigated the effect of leaf maturity on green tea flavour using a single cultivar (‘CC2’) processed into three plucking standards: C1B (single bud), C1B1L (one bud with one leaf), and C1B2L (one bud with two leaves). Sensory evaluation showed that C1B1L exhibited the most balanced and floral character, C1B was fresh and light, while C1B2L tended toward strength and thickness.

Taste-related compounds—including amino acids, caffeine, polyphenols, and catechins—varied significantly across samples. C1B had the highest levels of sweet and umami amino acids and lower ester-type catechins, contributing to its clean, brisk taste. C1B1L featured abundant polyphenols and amino acids, supporting its sweet–fresh profile, while C1B2L showed lower amino acid content and stronger astringency. Among 97 identified volatiles, five compounds—1-octen-3-ol, hexanal, 2-pentylfuran, geraniol, and benzeneacetaldehyde—were identified as key aroma contributors through ROAV and multivariate analysis. Except for geraniol, which peaked in C1B1L and shaped its floral aroma, the other four were most abundant in C1B and derived mainly from lipid and amino acid pathways.

These findings reveal the chemical basis of flavour variation—spanning both aroma and taste—across different leaf maturities. The dominant role of volatiles highlights how fatty acid, amino acid, and glycoside pathways drive aroma style transitions, offering a scientific foundation for raw material selection and aroma-oriented green tea production.

Ethics statements.

The sensory assessment of tea samples in this study was conducted by a panel of professionally trained evaluators. This research was performed in strict accordance with the ethical principles for human experimentation established by the World Medical Association's Declaration of Helsinki. While national legislation does not mandate formal ethical approval for sensory evaluation, and no specific institutional human ethics committee exists for such assessments, the authors implemented a rigorous internal protocol to ensure participant safety and welfare. Prior to the experiment, all panelists were fully briefed on the study's requirements and potential risks, and informed consent was obtained from every participant. The authors further guarantee the protection of participants' rights and privacy: participation was entirely voluntary, no coercion was used, and all individuals retained the right to withdraw at any stage without penalty. Furthermore, personal data remains strictly confidential and will not be disclosed without explicit authorization.

CRediT authorship contribution statement

Chen Liu: Writing – original draft, Resources, Funding acquisition, Conceptualization. Chenbo Wang: Writing – original draft, Validation. Mengyang Wei: Data curation. Meiyi Ning: Formal analysis. Zhiying Xu: Investigation. Shengxiang Chen: Methodology. Jilai Cui: Project administration. Chuankui Song: Software. Qian Tang: Writing – review & editing, Funding acquisition.

Funding

This work was supported by the Sichuan Province International Science and Technology Innovation Cooperation/Hong Kong, Macao and Taiwan Science and Technology Innovation Cooperation Project (Grant No. 2024YFHZ0284), the Independent Exploration Program of the Dual Support Plan for Discipline Construction at Sichuan Agricultural University (Grant No. 2024ZYTS019), and the Undergraduate Research Training Program of Sichuan Agricultural University (Grant No. 2025106X).

Declaration of competing interest

The authors declare that they have no known competing financial interests or personal relationships that could have appeared to influence the work reported in this paper.