Plant aromas in wine: the case of cabernet sauvignon from typical regions in China

Abstract

22 Cabernet Sauvignon wines from 10 representative Chinese wine regions are collected, which wines contain plant aromas? GC × GC-QTOFMS/O analysis identified a total of 11 compounds with plant aroma characteristics, the average concentration of plant aroma-active compounds in the wine samples from the G2 (the second group region with cooler climate) (450 μg/L) was significantly higher than that in the G1 (the first group) (95 μg/L), and the average intensities of four out of the five plant aroma attributes in the wines of G2, namely green pepper, grass, mint, and eucalyptus, were higher than those in G1, plant aromas contributed to the sensory style of 95% of the wine samples analyzed. This systematic analysis of the distribution characteristics and intensity differences of plant aromas in typical Chinese wine regions provides theoretical support for exploring the regional features of Chinese wine, improving product quality and enhancing market competitiveness.

Graphical abstract

Highlights

• •Collecting wines from typical Chinese regions.
• •Identifying 11 compounds with plant aroma characteristics by GC × GC-QTOFMS/O analysis.
• •Describing the difference of Chinese wine regions based on the plant aromas.
• •The style of wines could be characterized qualitatively and quantitatively.

1. Introduction

Plant aromas in wine, as a distinctive flavor characteristic encompassing herbal notes such as green pepper, grass, and asparagus, holding a distinct place in wine olfactory perception that evoke associations with fresh or dried plant tissues (Zhao et al., 2019), has gradually become a focus of academic and wine industry attention in recent years due to its association with concepts like “natural” and “ecological”. Some consumers appreciate the fresh and subtle plant aromas, perceiving them as conveying a refreshing, natural sensation that better reflects the varietal characteristics of certain wines (Herrmann, 2016; Parr, Valentin, & Green, 2010). Conversely, other consumers associate these aromas with unripe berry (Chapman, Matthews, & Guinard, 2004).

The sources of such aromas primarily encompass two dimensions, varietal aromas, which are plant aromas with varietal character (such as green pepper, mint, eucalyptus) released from aroma precursors inherent to the grape berries themselves (Gregan & Jordan, 2016); and aromas from the winemaking process, which refer to plantaroma compounds (such as hay, tea, herbal notes) produced by relevant microbial metabolism during winemaking stages, particularly maceration and fermentation (Zhao et al., 2019).

As a beverage with distinct regional characteristics, wine's flavor profoundly influenced by the terroir conditions of the region (Zhang, Yang, Liu, Sun, & Yao, 2025). In some cool-climate regions, such as parts of the Loire Valley in France and New Zealand, Sauvignon Blanc wines exhibit pronounced plant aromas (Parr et al., 2010). Another study highlighted that Sauvignon Blanc wines from the Marlborough region of New Zealand are highly popular among consumers for their distinctive exotic character (Herrmann, 2016). Aromas reminiscent of boxwood, asparagus, and grass are fresh and recognizable, endowing the wines with intense and unique personality. These aromas provide a strong olfactory impact, allowing New Zealand Sauvignon Blanc to stand out. The plant aromas dominate the wine's aromatic profile and has become a signature flavor characteristic of Sauvignon Blanc from this region (Herrmann, 2016).

The study of the aroma profiles of Cabernet Sauvignon and Merlot dry red wines from five Chinese regions revealed that the characteristic aroma compounds in the Hexi Corridor region were primarily esters, with the highest content of C6 compounds. The aroma profile featured prominent plant aromas and red berry notes, potentially attributable to the cool climate and lowest average temperatures (Zhang, Qiao, & He, 2022). The Jieshi Mountain region exhibited higher hexanol content, possibly contributing to vegetal, grassy, and herbaceous notes, mainly due to lower photosynthetically active radiation (Zhang et al., 2022). Analysis of the impact of inter-parcel differences on wine aromas in the Eastern Foothills of Helan Mountain region showed that Marselan wine from parcel QTX1 contained the most abundant methoxypyrazines, exhibiting the highest intensities of green pepper and herbaceous notes. Compared to other parcels, parcel XX2 experienced lower temperatures and shorter sunlight duration, which may have led to higher pyrazine content in the corresponding wine (Dunlevy et al., 2013).

Previously, domestic research on the sensory mechanisms underlying the formation of typical aroma styles in Chinese wines has primarily focused on explaining fruity and floral aromas. For instance, the typical aroma characteristics of different wine styles, such as Hutai 8 rosé wine (fruity ester compounds) from the Weibei Dry Plateau region (Li, Wang, Yin, Ma, & Tao, 2022) and Vidal icewine (β-damascenone and ester compounds) from the Huanlong Lake region (Ma, Tang, Xu, & Li, 2017), have been systematically elucidated. The composition of hawthorn aroma (ethyl isovalerate) (Lyu, Tian, Ma, Xu, & Tang, 2025) and stone fruit aroma (monoterpenes) (Siebert et al., 2018) in wine have been comprehensively analyzed. Some studies have also addressed the overall flavor characteristics of wines from specific Chinese regions (Lan et al., 2022; Zhang et al., 2023). However, the specific contributions of plant aromas to wine styles across different regions have not been systematically analyzed. Systematic research on plant aromas in wines from typical Chinese regions remains relatively scarce. Among the numerous factors influencing wine flavor, plant aromas adds a unique charm due to its distinctive freshness and plays a significant role in shaping wine style (Zhang et al., 2022). China's wine industry has developed rapidly in recent years, various regions have nurtured wines with distinct regional characteristics (Tu et al., 2022). In-depth research into the presentation and stylistic contribution of plant aroma in wines from characteristic Chinese regions, and the establishment of a localized terroir aroma theory system based on this dimension, hold profound significance for balancing stylistic uniqueness with market acceptance, uncovering the regional characteristics of Chinese wines, enhancing product quality, and strengthening market competitiveness.

To identify plant aroma-active compounds, effective analytical tools are essential. Gas Chromatography-Olfactometry (GC-O) enables the qualitative characterization of volatile compounds' aroma profiles, simultaneously providing information on both chemical composition and odor characteristics of samples. It serves as an effective method for identifying aroma-active compounds (Li et al., 2025; Ni, Yan, & Tian, 2022).

This study took 22 Cabernet Sauvignon wines from 10 representative Chinese wine regions as samples, covering various climate conditions (i.e., cool, temperate continental, and plateau). Volatile compounds in the wines were quantitatively analyzed using headspace solid-phase microextraction (HS-SPME) combined with gas chromatography × gas chromatography quadrupole time-of-flight mass spectrometry (GC × GC-QTOFMS); the GC-O analysis employed the aroma intensity method, with trained assessors sniffing and recording aromas in real time, and calculating the mean intensity value, and a professional sensory panel was organized for sensory evaluation. The aim is to comprehensively reveal the distribution characteristics and intensity differences of plant aromas in wines from typical Chinese regions, clarify their contribution to wine styles, and explore their associations with regional terroir conditions. The research findings are expected to fill the research gap in this field, provide specific data support for the scientific identification and definition of the style typicality of the major wine regions in China, and help each producing region further tap into its advantages to create wine products with distinct regional characteristics.

2. Materials and methods

2.1. Wine sample

Taking typical Chinese wine regions as the core sampling areas, the regions span from west to east including Tibet, the southern and northern Foothills of Tianshan Mountains of Xinjiang, Southwestern High-Altitude, Ningxia, the Loess Plateau, Yanhuai Valley, Tianjin, Shandong, and the Ancient Course of the Yellow River, a total of 10 wine regions. Specific locations are shown in Fig. 1. Sample collection strictly adhered to the protected geographical indication areas. In 2022, recommended samples from each region were obtained through collaboration with wineries. A total of 22 valid Cabernet Sauvignon red wines samples were acquired, limited to single-variety and non-blended, from single-vineyard or single sub-regional wines, with alcohol contents ranging from 9.5% to 11.5% vol. Detailed information is presented in Table S1-Supporting Information. Sensory evaluation was completed immediately after bottle opening on three replicate for each wine.

Fig. 1.

Representative Wine Regions in China.

2.2. Extraction and quali-quantitave analysis of volatile compounds

Volatile compounds in wines were extracted using HS-SPME; specifically, 5 mL of wine sample, 1 g of NaCl, and 10 μL of 4-methyl-2-pentanol (324.4 μg/L, internal standard) were placed into a 20 mL sample vial, which was immediately sealed with a screw cap and positioned in the auto-sampler tray. Extraction was conducted at 45 °C for 30 min, followed by desorption at 250 °C for 3 min, prior to the analyses.

All samples were analyzed using GC × GC-QOFMS, which consisted of an Agilent 8890–7250 gas chromatography tandem quadrupole time-of-flight mass spectrometer (Agilent, Santa Clara, CA, USA), a headspace solid-phase microextraction auto-sampler MPS (GERSTEL, Germany), and a solid-state thermal modulator (SSM1820, J&X Technologies, Shanghai, China). The comprehensive two-dimensional gas chromatography (GC × GC) system was configured as follows: the first-dimension column was a DB-WAX column (30 m × 0.25 mm × 0.25 μm, Agilent, Santa Clara, CA, USA), and the second-dimension column was a DB-17MS column (1.85 m × 0.180 mm × 0.18 μm, Agilent, Santa Clara, CA, USA). The initial oven temperature was set at 50 °C, increased at a rate of 3 °C/min to 230 °C, and held for 2 min. Splitless injection mode was employed with high-purity helium (99.999% purity) as the carrier gas at a constant flow rate of 1 mL/min. The ion source temperature was 230 °C, the EI ionization energy was 70 eV, the mass scan range was 45–500 m/z (full scan mode), the MS acquisition rate was 50 spectra/s, and the solvent delay time was 2 min. The solid-state modulator used an SV modulation column (C6-C40 hydrocarbon standard) with a modulation period of 4 s, a cold zone temperature held constant at −50 °C (Li et al., 2025; Zhang et al., 2024). Mass spectrometry data were processed using Canvas Panel software (version 2.5.710, J&X Technologies, Shanghai, China). The retention indices (RI) of volatile compounds were calculated using n-alkanes. Volatile compounds were identified based on mass spectra, matching against the NIST20 library (reverse match factor ≥ 700, difference between forward and reverse match factors <50), and RI (Dong et al., 2024). Quantification of volatile compounds was performed by standard calibration curve or using the calibration curves of compounds with similar chemical structures and carbon chain lenghts (Table S2-Surporting Information) (Xi et al., 2011).

2.3. GC-O analysis

According to Li et al. (2025), a Gerstel olfactory detection port (ODP4; Gerstel GmbH & Co. KG, Mülheim an der Ruhr, Germany) was interfaced downstream of the first DB-WAX column for olfactory evaluation. Odor assessment utilized the aroma intensity (AI) method and involved a panel of six trained assessors (3 males, 3 females; aged 20–30 years) experienced in GC-Olfactometry (GC-O). Prior to analysis, panelists underwent a minimum week training regimen to standardize aroma descriptor usage. During analysis, assessors documented both the perceived aroma descriptors and their corresponding intensities. Aroma intensity was quantitatively evaluated on a four-point scale (0 = none, 4 = very strong), with the final AI index reported as the mean score across the panel.

2.4. Sensory evaluation

The sensory panel comprised 12 trained assessors (6 male, 6 female), recruited from the student population at Northwest A&F University (aged 20–30 years). All panelists possessed relevant academic backgrounds and completed an intensive training program utilizing the 54-aroma Le Nez du Vin kit (France) experiencing 2 months (Tao et al., 2009). Panel performance was rigorously assessed using Panel Check software, evaluating two critical parameters: reproducibility (consistency across replicates) and discriminative power (ability to differentiate samples). This assessment was based on F-statistics and mean-square error (MSE) values, where higher F-values and lower MSE for a given attribute signify superior discriminative ability and repeatability, respectively (Cheng et al., 2025). Based on these metrics, 12 candidates demonstrated an average reproducibility index exceeding 0.25, coupled with low MSE values for the majority of aroma descriptors, confirming panel consistency (Cheng et al., 2025). Following this scientific and standardized training protocol, these 12 individuals were formally certified to form the sensory panel responsible for identifying characteristic wine aromas. Aroma descriptors were determined via quantitative descriptive analysis, adhering to the methodology outlined in the Chinese National Standard GB/T 16861–1997 (“Sensory Analysis: Identification and Selection of Descriptors for Building Sensory Profiles through Multivariate Analysis”) (Cheng et al., 2025; Lan et al., 2021). Prior to participation, informed consent was obtained from all volunteers. Sensory evaluations were conducted online using the Simplicity Sensory Analysis System (v2.0) software (China National Institute of Standardization) (Cheng et al., 2025; Lan et al., 2021). Within this system, assessors rated the intensity of each perceived aroma descriptor on a 0 to 5 scale, defined as: 0 (no sensation), 1 (weak), 2 (slightly weak), 3 (average), 4 (slightly strong), 5 (strong). The geometric mean intensity value (M) for each descriptor was subsequently calculated:

$$\mathrm{M}=\sqrt{F\times I}$$

The frequency (F) represents the percentage of times a descriptor was cited relative to the total possible citations, while the intensity percentage (I) reflects the actual intensity rating given relative to the maximum possible intensity for that descriptor (Cheng et al., 2025). This study received review and approval from the Ethics Committee of Northwest A&F University.

2.5. Statistical analysis

Single-factor analysis of variance (ANOVA) and t-tests were performed using SPSS version 21.0 (IBM, NY, USA). Processed data were visualized using GraphPad Prism 9.2.0. Principal component analysis (PCA), hierarchical cluster analysis (HCA), and heatmap construction were conducted with Origin 9.1 software (OriginLab, MA, USA). Orthogonal partial least squares discriminant analysis (OPLS-DA) was performed using SIMCA 14.1 (Umetrics, Umea, Sweden).

3. Results and discussion

3.1. Identification of plant aroma-active compounds by GC-O analysis

Volatile compounds, as the core elements affecting wine flavor, are the key carriers that highlight its typicality (Wu et al., 2025). To systematically and exhaustively capture the overall volatile composition of the samples, and to avoid overlooking potential unknown important components due to pre-defined targets, we performed metabolomic analysis on 22 Cabernet Sauvignon wines of this study samples using HS-SPME combined with GC × GC-QTOFMS. A total of 120 volatile compounds were identified, including 72 esters, 3 ketones, 17 alcohols, 10 aldehydes, 5 terpenes, 4 C13-norisoprenoid derivatives, 7 acids, and 2 phenols. Detailed information is presented in Table S3-Supporting Information and Table S4-Supporting Information. Combining GC × GC-QTOFMS/O analyses 11 compounds with plant aroma characteristic (Fig. 2a; Table 1; Table S10-Supporting Information), representing the core of the research. Of the 11 compounds identified, eight were found to exhibit distinct plant aromas via GC-O analysis, while the remaining three were drawn from previous research reports, (Z)-2-hexen-1-ol exhibits a grassy note (Calleja and Falqué, 2005; Cheng et al., 2025), 2-tridecanone displays herbal and hay characteristics (Jirovetz, Buchbauer, & Abraham, 2006), and theaspirane possesses a tea-like aroma (Calleja and Falqué, 2005). The absence of detectable plant aroma of these three 6compounds in this study may be explained by the principle that the contribution of a compound to the overall wine aroma is determined not solely by its concentration but also by its odor threshold (Li et al., 2025). Considering that the interaction and additive effects among compounds with plant aromas surpass the odor or even the threshold effects of individual compounds (Poitou et al., 2017), we therefore will discuss the 11 volatile compounds associated with plant aromas. The compounds contributing to the plant aroma in wine mainly include methoxypyrazine compounds (Sala et al., 2004), C6/C9 compounds, and other types of compounds (Poitou et al., 2017). The 11 compounds with plant aroma characteristic in this study are mainly categorized into C6/C9 compounds and other types of compounds (especially alchols, aldehydes, and ketones), whose chemical structures are shown in Fig. 2b. Specifically, these comprised 4 alcohols: 1-hexanol, (Z)-2-hexen-1-ol, 2-nonanol, and (Z)-3-Hexen-1-ol; 3 ketones: 2-tridecanone, geranylacetone, and valerone; 2 aldehydes: nonanal and (E)-2-nonenal; as well as methyl laurate and C13 norisoprenoid theaspirane.

Fig. 2.

Volatile compounds with plant aromas. (a) 3D Chromatogram obtained by GC × GC-QTOFMS for the detection of volatile compounds with plant aromas in 22 wines from different regions. (b) The chemical formula of volatile compounds with plant aromas.

Table 1.

Volatile compounds with plant aromas.

Volatile compounds with plant aromas
CAS Number	Volatile Compound	Category	RI	One-dimensional retention time (min)	Odor retention time (min)	Odor featurea	Odor featureb	Thresholdc (μg/L)
111–27-3	1-Hexanol	C6	1351	17.716	17.63–17.64	green, asparagus	green, mowing grass	1300
928–96–1	(Z)-3-Hexen-1-ol	C6	1361	18.183	18.71–18.72	green leaf, mushroom	green leaves	1000
928–94-9	(Z)-2-hexene-1-ol	C6	1413	20.18	—	—	green grass, herbs	400
628–99-9	2-Nonanol	C9	1517	24.316	24.25–24.38	green grass	green	70
124–19-6	Nonanal	C9	1392	19.313	19.08–19.09	hay	raw green	2.5
18,829–56-6	(E)-2-nonenal	C9	1533	24.913	24.55–24.57	green pepper	green, cucumber	0.6
108–83-8	Valerone	Other	1171	11.046	10.86–10.91	hay, herbs	green, mint	1800
593–08-8	2-Tridecanone	Other	1806	35.113	—	—	hay, herbs	—
3796-70-1	Geranylacetone	Other	1852	36.713	36.55–36.58	green grass	green, fresh	621
111–82-0	Dodecanoic acid, Methyl ester	Other	1801	34.913	35.31–35.33	hay, herbs	hay, mushroom	2
36,431–72-8	Theaspirane	Other	1542	25.247	—	—	green, herbs	—

Note: Odor feature^a represents the odor sniffed in GC-O; odor feature^b represents the odor acquired in the references, (Calleja and Falqué, 2005; Cheng et al., 2025; Jirovetz et al., 2006); and threshold^c represents the threshold acquired in the references, (Cheng et al., 2025; Van Gemert, 2011); “—” represents undetected.

3.2. The difference of compounds with plant aromas in wines from different regions

Based on the contents of these 11 volatile compounds, cluster analysis was performed on the 22 wines. The results are shown in Fig. 3a. The wines from 10 regions were divided into two major groups, the first group (G1) included wines from the Ningxia, Loess Plateau, Yanhuai Valley, Tianjin, and Ancient Course of the Yellow River wine regions, while the second group (G2) mainly referred to wines from the Tibet, Xinjiang, Southwest High-Altitude, and Shandong wine regions. This clustering result clearly demonstrated distinct volatile compound profiles among wines from different regions. More significantly, it indicates a strong correlation between regional characteristics and the chemical composition of the wines (Van Leeuwen, Roby, & de Rességuier, 2018). This correlation helps explain how specific terroir conditions influence the typical aroma profile of wine (Goulioti et al., 2025).

Fig. 3.

The difference of plant aroma-active compounds in wines from different regions. (a) Clustering heat map of producing areas based on 11 volatile substances of plant notes.

For abbreviations of the wine regions, please refer to Table S1-Supporting Information. (b) OPLS-DA between two groups of wines producing areas. G1 represents the first group of production areas, covering Wine Region of Ningxia, the Loess Plateau, the Yanhuai Valley, Tianjin and Ancient Course of the Yellow River; G2 represents the second group of production areas, including Wine Region of Tibet, Xinjiang, Southwest High-Altitude and Shandong. (c) Different metabolite screening. Pink indicated significant increase (P < 0.05), green indicated significant decrease (P < 0.05), and gray indicated no significant difference (P > 0.05), G2 comparing to G1. (For interpretation of the references to colour in this figure legend, the reader is referred to the web version of this article.)

An OPLS-DA was performed with the aim of screening for compounds with plant aromas characteristic that differed between the two groups wine regions. In this model, the wine regions served as independent variables and the contents of compounds as dependent variables. The results are shown in Fig. 3b and this set of models exhibited good predictability (Q² > 0.7) and strong fitting advantages (R²X > 0.9, R²Y > 0.9). In the OPLS-DA model, the two groups of wine regions were clearly separated (Fig. 3b), differential screening was performed on 11 volatile components based on the VIP values (VIP > 1) and t-test (P < 0.05) results to screen for the key differential metabolic compounds with plant aromas between the two groups of production regions, and the results are shown in Fig. 3c. The specific differential compounds are shown in Table S5-Supporting Information. Compared with the G1, the contents of 1-hexanol, 2-tridecanone, and methyl laurate in the G2 were significantly increased (P < 0.05), while the contents of 5 volatile compounds including (Z)-3-hexen-1-ol, (Z)-2-hexen-1-ol, geranylacetone, (E)-2-nonenal, and valerone were significantly decreased (P < 0.05). The two groups of regions exhibited marked differences in terroir (e.g., climate, soil, and topography) and geographical location. Within the G2, the Xinjiang region experienced significant diurnal temperature variations, known as cold desert climate (Lan et al., 2022), the Tibet and Southwest High-Altitude regions featured alpine or plateau climates with lower average temperatures (Xing, He, & Xiao, 2015), while the Shandong region, situated on the eastern coast, was distinctly influenced by a maritime climate (Li, Pan, & Jin, 2011). Collectively, these regions exhibited cooler climatic profiles compared to the first group.

There were 9 wine samples in the G1, and the total concentration of compounds with plant aromas was 860 μg/L, while there were 13 wine samples in the G2 and the total concentration of compounds with plant aromas was 5846 μg/L. The average concentration of compounds with plant aromas in the wine samples of G2 (450 μg/L) was significantly higher than that of the G1 (95 μg/L). Multiple studies have demonstrated higher concentrations of volatile compounds with plant aromas in grape berries under cool-climate conditions compared to warmer regions (Allen, Lacey, Harris, & Brown, 1991; Belancic & Agosin, 2007; Lei et al., 2018; Mendez-Costabel et al., 2014). This phenomenon may be attributed to reduced activity of catabolic enzymes in biosynthetic pathways governing aroma compounds formation, potentially influenced by lower effective accumulated temperatures associated with cooler climates (Koundouras, 2018; Qin et al., 2017). Oliveira, Faria, and Sá (2006) also utilized the content of 1-hexanol to differentiate wines in their study. A similar phenomenon was observed in the study by Tian et al., (2023), where enhancing reflected light intensity and localized heating reduced C6 compound content, whereas diminishing reflected light intensity and localized cooling had the opposite effect. However, the decrease of C6 compounds ((Z)-3-hexen-1-ol and (Z)-2-hexen-1-ol) may suggest that the accumulation of aroma compounds in wine is a complex outcome, influenced by diverse environmental conditions within production regions, such as soil, sunlight, heat, water, slope aspect, and altitude (Lan et al., 2022). The contents of the other 3 volatile compounds, namely theaspirane, 2-nonanol, and nonanal, showed no significant changes.1-hexanol was identified as the most influential compound for differentiating the two regional groups, exhibiting the highest VIP value (3.0697) among the significantly differential volatiles (P < 0.05), thereby playing a key role in regional discrimination based on aroma profile.

3.3. The sensory contribution of plant aromas in wines from different regions

The sensory panel conducted sensory evaluations on 22 Cabernet Sauvignon wines from 10 regions, obtaining more than 40 sensory descriptors, of which 5 related to plant aromas, namely green pepper, green grass, hay, mint, and eucalyptus. The geometric mean (M) of the sensory descriptor was used to characterize the intensity of the aroma represented by the descriptor (Cheng et al., 2025) (Fig. 4). The wines from 10 regions were categorized into two groups (G1 and G2) based on 11 plant aroma compounds. The wines in G2 had stronger plant-like aromas than those in G1. Specifically, G2 wines had more intense scents of green pepper, grass, mint, and eucalyptus. Only the hay aroma was stronger in G1. This difference in aroma intensity aligns with the chemical analysis from Section 3.2, which found a higher concentration of these plant aroma compounds in G2 wines.

Fig. 4.

M-value of the plant aroma sensory descriptors. The M-value is geometric mean of the intensity of sensory descriptor.

A further cluster analysis was performed on the 22 wines based on the intensities of 5 plant aromas, and the results are shown in Fig. 5a. There were 7 wines from the Xinjiang region, and 6 of them were clustered into one category, indicating that these 6 wines had high similarity in the intensity and types of plant aromas, mainly presenting green pepper, mint, and green grass aromas (Fig. 5a). Based on the M value (where a higher score indicates a more pronounced aroma) (Fig. 4), the green pepper aroma was the most prominent in the wine YN4–1 from the Southwest High-Altitude region, while the most obvious mint aroma was presented in the wine HB7–2 from the Yanhuai Valley region, and the green grass aroma was the most intense in the wine NX5–2 from the Ningxia region. Also, the eucalyptus aroma was relatively obvious in the wines SD9–2 from the Shandong region and XZ1–2 from the Tibet region, and the hay aroma of SD9–2 was also relatively prominent. Correlation analysis was performed between the 11 compounds with plant aromas and the 5 plant aromas, and the results are shown in Fig. 5b. The intensity of the green grass aroma showed a significant negative correlation (P < 0.05) with the contents of nonanal and (Z)-3-hexen-1-ol, while the intensity of the hay aroma was significantly positively correlated with (Z)-2-hexen-1-ol and (E)-2-nonenal. Indeed, for instance, the wine JS10–1 from the Ancient Course of the Yellow River region had relatively high contents of (Z)-2-hexen-1-ol and (E)-2-nonenal, and also had strong hay aroma; moreover, the wine SD9–2 had a relatively high content of (E)-2-nonenal, and its hay aroma was the most intense among all wines, the plant aroma of (E)-2-nonenal was also verified by Li et al. (2025) via GC-O. The mint aroma was significantly positively correlated (P < 0.05) with the content of (Z)-3-hexen-1-ol; the wine HB7–2 from the Yanhuai Valley region mainly described by mint aroma, the highest content of (Z)-3-hexen-1-ol (Fig. 3a and 4d). The fresh green aroma contributed by (Z)-3-hexen-1-ol to wine had also been mentioned previously (Martínez-Lüscher and Kurtural, 2023). The above analysis clarified the intensity differences of plant aroma attributes in wines from different regions and their associations with key active compounds. However, the overall sensory style of wine is not determined by a single aroma attribute but rather results from the synergistic effects of multiple aroma attributes. To investigate the role of plant aromas in shaping the overall aroma style of wines from different regions, it is necessary to further extract their characteristic aroma profiles from all aroma descriptors and analyze the role that plant aromas play within them.

Fig. 5.

The sensory contribution of plant aromas in wines from different regions. (a) Clustering heat map of producing areas based on 5 plant aromas. For abbreviations of the wine regions, please refer to Table S1-Supporting Information. (b) Heat map of correlation between 11 volatile substances and plant aromas.

3.4. The stylistic contributions of plant aromas in wines from different regions

Sensory characteristics represent specific, localized attributes, while the overall style emerges from the systematic integration of these attributes across perception (Cheng et al., 2025). Principal component analysis (PCA) was performed using the intensity values of the top 10 aroma descriptors (ranked by the M value for each wine) as original variables (Table S6-Supporting Information). As shown in Table S7-Supporting Information, the cumulative contribution rate of the first five principal components exceeded 85%, indicating that these five principal components essentially contained all the information included in the entire set of aroma descriptors. Eigenvectors represent loadings on a unit scale, based on the normalized eigenvectors corresponding to the eigenvalues, descriptors with higher weights were selected to identify the characteristic descriptors for each wine (Table S8-Supporting Information). Subsequently, cluster analysis was conducted on the intensity values of these characteristic aroma descriptors (Fig. 6) to further clarify the characteristic aroma profiles of wines from different regions (Cheng et al., 2025).

Fig. 6.

Systematic clustering of wine aroma descriptors. (a-v) wines from XZ1–1 to JS10–1, the specific corresponding order is consistent with Table S6-Supporting Information.

At an Euclidean distance of 7.5, the characteristic aromas of wine sample XZ1–1 (Fig. 6a) could be divided into 3 categories (Table S9-Supporting Information), combing these categories with the ranking of aroma descriptors by M value (Table S6-Supporting Information), which is dominated by spicy notes (cinnamon, clove, and licorice), supplemented by woody and plant aromas (oak and green pepper), along with black berry notes (mulberry) and toasty aromas (chocolate and smoke).

A maximum of three plant aromas appeared together in the characteristic profile of any single wine, such as XJ2–1 from the Southern Foothills of Tianshan Mountains in Xinjiang and SX6–1 from the Loess Plateau, even though five aromas were detectable overall (Fig. 6; Table S9-Supporting Information). There were eleven wines' characteristic descriptors containing two types of plant aromas, while eight wines contained only one type of vegetal aroma. Notably, only SD9–1 wine's characteristic aroma descriptors did not contain plant aromas (Table S9-Supporting Information). Climatic factors such as altitude, temperature, and humidity affect the volatile characteristics of Cabernet Sauvignon grapes (Perestrelo, Barros, Rocha, & Câmara, 2014). There are 10 wine samples from the Tibet, Xinjiang, and Southwest High-Altitude regions, among which 9 samples exhibit relatively distinct plant aromas, accompanied by fruity and spicy notes, showing overall similar aroma styles (Table S9-Supporting Information), this similarity has been reflected in Fig. 3a, in the previous classification of regions based on 11 plant aroma-active compounds (Section 3.2), the above three production regions were grouped into one category (Fig. 3a), indicating that they have certain similarities in the composition and content of compounds with plant aromas, such chemical similarities contribute to the consistency of their aroma-presenting styles (Luo, Zhang, & Loo, 2022). These three wine regions have shown cool climates with a plateau, alpine or desert climate. Previous studies have shown that under cool climates, the content of volatile compounds presenting plant aromas in grape berries is relatively higher than that under hot conditions (Allen et al., 1991; Belancic & Agosin, 2007; Mendez-Costabel et al., 2014; Lei et al., 2018), which may be due to the lower effective accumulated temperature associated with cool climates affecting the activity of decomposing enzymes in metabolic pathways related to the formation of aroma precursors (Koundouras, 2018; Qin et al., 2017), resulting in the incomplete decomposition of plant aroma-active substances. Additionally, the increased acidity possibly caused by low temperatures synergizes with plant aromas (Peirano-Bolelli, Heller-Fuenzalida, Cuneo, Peña-Neira, & Cáceres-Mella, 2022; Rojas et al., 2024), enhancing the perception of fresh plant aromas. Similarly, the two wines from the Yanhuai Valley region also have prominent plant aromas (Table S9-Supporting Information), which located at the southern foot of the Yanshan Mountains with undulating terrain, and vineyards are mostly situated on hillsides or at the foot of mountains, forming a unique cool microclimate. Such cool climate may lead to the accumulation and prominence of plant aromas in the wines (Koundouras, 2018; Qin et al., 2017). Plant aromas also contribute to varying degrees the aroma styles of wines from the other 5 wine regions, and moderate plant aromas can add complexity to wines.

4. Conclusions

A total of 120 volatile substances were detected in the 22 wines from 10 regions, among which GC × GC-QTOFMS/O analysis identified a total of 11 compounds with plant aroma characteristics. Based on the contents of these plant aroma-active volatile compounds, the 10 wine regions were divided into two groups. Collectively, G2 regions exhibited cooler climatic profiles compared to the G1.

There were 9 wine samples in the G1, with the total concentration of compounds with plant aroma being 860 μg/L, while the G2 comprised 13 wine samples, with the total concentration of which reaching 5846 μg/L. The average concentration of compounds with plant aromas in the wine samples from the G2 (450 μg/L) was significantly higher than that in the G1 (95 μg/L). Also, the average intensities of four out of the five plant aroma attributes in the wines of G2, namely green pepper, grass, mint, and eucalyptus, were higher than those in G1, only the hay aroma showed the opposite trend. And the intensity of the green grass aroma showed a significant negative correlation with the contents of nonanal and (Z)-3-hexen-1-ol, while the intensity of the hay aroma was significantly positively correlated with (Z)-2-hexen-1-ol and (E)-2-nonenal, the mint aroma was significantly positively correlated with the content of (Z)-3-hexen-1-ol.

Among the 22 wines, only the wine SD9–1 from the Shandong region does not have the contribution of plant aromas. The Tibet, Xinjiang, and Southwestern High-Altitude regions were grouped into one category based on 11 compounds with plant aromas of wines, and 90% of the wine samples from which exhibited relatively distinct plant aromas, accompanied by fruity and spicy notes, showing an overall similar style. The two wines from the Yanhuai Valley wine region, which also had a cool climate, also featured prominent plant aromas. Plant aromas also contributed to a certain extent to the aroma profiles of wines from the other 5 wine regions. Overall, plant aromas contributed to the sensory style of 95% of the wine samples analyzed. Furthermore, their contribution to the aroma profile was relatively higher in wines originating from cool-climate regions.

CRediT authorship contribution statement

Guo Cheng: Writing – original draft, Methodology, Data curation. Zhihao Deng: Visualization, Software, Investigation. Dingze Yin: Software, Project administration. Yue Wang: Software, Project administration. Jianing Li: Software, Project administration. Zhenglong Cheng: Resources, Project administration. Junxia Dou: Resources, Project administration. Zusong Liao: Resources, Project administration. Yulin Fang: Supervision, Project administration, Methodology, Funding acquisition. Xiangyu Sun: Resources, Project administration, Funding acquisition.

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.

Appendix A. Supplementary data

Supplementary material

Data availability

Data will be made available on request.