Abstract

   Allium mongolicum is a wild vegetable with high nutritional value and
   is famous for its taste and aroma. This study used headspace
   solid-phase microextraction-mass spectrometry coupled with gas
   chromatography–mass spectrometry techniques to study the metabolic
   profile of A. mongolicum in different ecological environments. A total
   of 624 volatile organic compounds (VOCs) were identified. Ester
   compounds, heterocyclic compounds and terpenoids are the key
   metabolites that determine flavor differences. KEGG analysis showed
   that monoterpenoid biosynthesis, zein biosynthesis, α-linolenic acid
   metabolism and secondary metabolite biosynthesis were the most
   important metabolic pathways. Compared with Minqin A. mongolicum and
   Tengger A. mongolicum, Alxa A. mongolicum flavor substance notes
   sensory flavor has more green, fruity, sweet, floral, spicy, metallic,
   rose, almond, apple, grassy, tropical, citrus, fresh, herbal and other
   flavor combinations. Overall, this study reveals the main reason for
   the unique flavor of Alxa A. mongolicum through metabolomic evidence.

   Keywords: Allium mongolicum, Flavor substances, Metabolomics, Volatile
   organic compounds

Highlights

     * •
       Based on metabolomics analysis, a total of 624 volatile organic
       compounds were identified.
     * •
       The material composition of the unique flavor of the Alxa Allium
       mongolicum has been elucidated.

1. Introduction

   Allium is a group of common edible plants, such as Allium cepa, Allium
   fistulosum and Allium tuberosum Rottler ex Sprengle, which are of great
   economic value and cultural importance ([35]Şen, Göktürk, Hajiyev, &
   Uğuzdoğan, 2023; [36]Bi et al., 2024; [37]Wang et al., 2023). The
   volatile flavor components of allium are mainly composed of aldehydes,
   alcohols, ketones, acids, phenols, esters, furans and furanones, sulfur
   and nitrogen, of which sulfur compounds are the main source of the
   unique flavor of allium ([38]Hao et al., 2023; [39]Kim et al., 2022;
   [40]Nandakumar et al., 2018). Allium vegetables because of its leaf
   cells contain volatile sulfides, so it has a pungent smell of volatile
   oil, can remove the smell of odour and taste, oil and greasy, produce a
   special aroma, unique aroma and flavor make it a fresh vegetable,
   processed products and many processed food additives ([41]Kim et al.,
   2023; [42]Zhang, Jiang, Chen, Sun, & Cui, 2015). Sulfur compounds are
   characteristic aromatic constituents of allium plants, which have
   certain pharmacological functions, antibacterial and antifungal
   effects, can inhibit platelet aggregation, prevent and treat
   hyperlipidaemia and thrombosis, and significantly reduce blood
   cholesterol ([43]Richard et al., 2021).

   A. mongolicum is widely distributed and mainly grown in Qinghai, Gansu,
   Xinjiang and Inner Mongolia ([44]Muqier et al., 2017). A. mongolicum is
   a kind of green vegetable without pollution ([45]Zhang, Chen, & Zhang,
   2014). A. mongolicum contains relatively comprehensive nutrients,
   mineral nutrients, essential trace elements and amino acids are higher
   than ordinary vegetables, it is a good companion for cooking various
   nutritious soups and accompanying meals and wine, and fresh food has a
   unique flavor ([46]Bu et al., 2023; [47]Wang, Yang, Zhao, & Bao, 2013).
   Glutamic acid is most abundant in the leaves ([48]Bu et al., 2023;
   [49]Zhang et al., 2009; [50]Zhang & Ao, 2008). The main constituents of
   the extract are phenols, aldehydes, ketones, acids, alcohols,
   thioethers, non-heterocyclic sulfur-containing substances, alkanes,
   olefins and aromatic hydrocarbons ([51]Lu & Lu, 2002). In addition, A.
   mongolicum offer medicinal benefits, including the ability to regulate
   functional constipation and to prevent and treat obesity and
   hypertension ([52]Chen et al., 2020; [53]Wang et al., 2019). As one of
   the most promising functional foods of the future, demand for A.
   mongolicum is increasing, driving the development of A. mongolicum
   sauce, A. mongolicum noodles and other A. mongolicum processed foods.
   However, at the point of sale, Alxa A. mongolicum is preferred by
   consumers. Consumer preference is highly correlated with flavor
   intensity, but GC–MS analysis of flavor metabolite characteristics of
   Alxa A. mongolicum has not been reported.

   Metabolomics has become an effective tool to explore the biosynthesis
   pathway and molecular regulation mechanism of fruit and vegetable
   flavor chemicals ([54]Baky, Shamma, Xiao, & Farag, 2022; [55]Guo, Li, &
   Dou, 2024; [56]Nascimento et al., 2019). However, there have been few
   studies on the flavor compounds of A. mongolicum. Therefore, the
   comprehensive comparison of metabolites of Alxa A. mongolicum and A.
   mongolicum from different locations (Tengger A. mongolicum, Minqin A.
   mongolicum) will help plant breeders to cultivate and develop varieties
   with better flavor characteristics.

2. Materials and methods

2.1. Materials

   The fresh samples of A. mongolicm were collected from Alxa league,
   around the Tengger Desert, and Minqin, Gansu, in September 2023
   ([57]Fig. 1). Identified by Prof. ZhongRen Yang of Inner Mongolia
   Agricultural University (IMAU). A voucher specimen (No. 20230930–001)
   was deposited at the College of Horticultural and Plant Protection of
   IMAU.

Fig. 1.

   [58]Fig. 1
   [59]Open in a new tab

   A. mongolicum's different living conditions.

2.2. Sample preparation and GC–MS analysis

   The powder was processed according to the method of [60]Wang et al.
   (2023). Volatile organic compounds (VOCs) of A. mongolicum were
   determined by Wuhan Metware Biotechnology Co., LTD. (Wuhan, China,
   [61]http://www.metware.cn/) based on the Agilent 7890B—7000D platform
   at 250 °C in splitless mode (flow rate 1 ml /min). The detection time
   was 5 min. Each experiment was repeated 4 times. 1 g of the powder was
   transferred into a 20 ml headspace vial (Agilent, CA, USA) containing
   2 ml NaCl saturated solution to inhibit any enzyme reaction. Each vial
   was oscillated at 60 °C for 5 min, and a 120 μm DVB/CWR/PDMS extraction
   head was inserted into the sample headspace vial for 15 min. It was
   analyzed at 250 °C for 5 min, and then isolated and identified by
   GC–MS/MS (7890B gas chromatography and 7000D mass spectrometer
   (Agilent, CA, USA). Helium (99.999 %) was used as the carrier gas and
   the carrier gas flow rate was 1.0 ml/min. Maintain syringe temperature
   at 250 °C and detector temperature at 280 °C. The mass spectrometry
   (MS) was performed in electron shock (EI) ionization mode at 70 eV,
   with a scan range of m/z 20–550 amu at 1 s intervals. The quadrupole
   mass detector, ion source and transmission line temperatures are set at
   150 °C, 230 °C and 280 °C respectively. Volatile compounds were
   identified by mass spectrometry compared with NIST database and linear
   retention index ([62]Li et al., 2021).

2.3. Relative odour activity value (rOAV) analysis of metabolites

   rOAV analysis according to the method of [63]Huang et al. (2023). rOAV
   is a method of determining the major flavor compounds in foods based on
   the sensory threshold of the compounds and is used to clarify the
   contribution of each flavor compound to the overall flavor
   characteristics of the sample.

   In general, rOAV ≥1 indicates that the compound makes a direct
   contribution to the flavor of the sample.

2.4. Data analysis

   Unsupervised principle component analysis (PCA) was performed using the
   statistical function prcomp in R v3.5.0 ([64]www.r-project.org).
   Supervised multiple regression orthogonal partial least squares
   discriminant analysis (OPLS-DA) was performed using ropls v1.19.8 in R
   ([65]Chong & Xia, 2018). We used the threshold variable importance
   (VIP) value (VIP ≥ 1.0) in the OPLS-DA model to screen for these
   differential metabolites. To assess the significance of differences in
   metabolite abundance, a Student's t-test (mean two-tailed) was used
   (p = 0.05). Annotated metabolites were mapped to the Kyoto Encyclopedia
   of Genes and Genomes (KEGG) pathway database
   ([66]http://www.kegg.jp/kegg/pathway.html) to determine pathway
   associations. Pathway enrichment analysis was performed on the
   web-based Sever Metabolite Sets Enrichment Analysis (MSEA;
   [67]http://www.msea.ca). Bonferroni-corrected pathways with a P value
   ≤0.05 were considered significantly enriched.

3. Results

3.1. Overview of metabolic profiles

3.1.1. Volatile organic compounds (VOCs) in A. mongolicum.

   In order to detect the diversity of metabolites, HS-SPMEGC/MS
   technology was used to analyze volatile organic compounds in samples of
   A. mongolicum. A total of 624 metabolites were identified based on
   GC–MS detection platform and self-built database. Among them,
   terpenoids (102), ketones (44), hydrocarbons (66), alcohols (36), acids
   (15), esters (111), aldehydes (49), aromatics (38), heterocyclic
   compounds (106), sulfur-containing compounds (8), other classes (6),
   amines (20), halogenated hydrocarbons (4), nitrogen-containing
   compounds (6), ethers (2), phenols (11). Among them, esters contributed
   the most (17.79 %), followed by heterocyclic compounds (16.99 %) and
   terpenoids (16.35 %), which were the main components of volatile
   organic compounds of A. mongolicum in this study (Table S1). These
   results indicated that ester compounds, heterocyclic compounds and
   terpenoids were the most abundant metabolites in A. mongolicum.

3.1.2. Principal component analysis (PCA)

   To illustrate the presence of flavor substances in different habitats,
   we analyzed the metabolic profiles of 624 identified metabolites using
   PCA ([68]Fig. 2). The first principal component (PC1) explained 40.61 %
   of the metabolic differences among the three habitats, and the second
   principal component (PC2) explained 18.2 % of the metabolic differences
   among the three habitats. On PC1, group A was separated from group T
   and group M, which showed that the content of flavor substances in the
   samples of Alxa A. mongolicum, Tengger A. mongolicum and Minqin A.
   mongolicum etes was different. At the same time, the differential
   accumulation of volatile organic compounds is of great significance to
   elucidate the aroma formation mechanism of A. mongolicum. Therefore,
   these differentially expressed VOCs were analyzed in detail in this
   study.

Fig. 2.

   [69]Fig. 2
   [70]Open in a new tab

   PCA score of quality spectrum data of each group and quality control
   samples. The horizontal coordinate PC1 and the vertical coordinate PC2
   in the fgure indicate the scores of the frst and second ranked
   principal components, respectively. The scatter color indicates the
   experimental grouping of the sample. Each point in the figure
   represents a sample, and samples from the same group are represented by
   the same color. A = Alxa A. mongolicum, T = Tengger A. mongolicum,
   M = Minqin A. mongolicum.

3.1.3. Orthogonal partial least squares discriminant analysis (OPLS-DA) and
OPLS-DA model validation

   PCA analysis methods cannot ignore within-group errors and cannot
   exclude random errors. To further determine the differences in Volatile
   organic compounds (VOCs) under different growing conditions of A.
   mongolicum, we used supervised OPLS-DA analysis on different growing
   conditions of A. mongolicum ([71]Fig. 3a-c). As can be seen from the
   score plot, the samples from the different treatment groups were
   effectively differentiated, and unlike the PCA results, the degree of
   aggregation within each group was more aggregated than in the PCA
   method.

Fig. 3.

   [72]Fig. 3
   [73]Open in a new tab

   OPLS-DA scores. Scores of the OPLS–DA model with (a) A vs M, (b) T vs
   M, (c) A vs T. The permutation test chart for OPLS–DA model. The
   abscissa represents the model accuracy, and the ordinate is the
   frequency of the model classification effect. (d) A vs M, (e) T vs M,
   (f) A vs T. A = Alxa A. mongolicum, T = Tengger A. mongolicum,
   M = Minqin A. mongolicum.

   According to the model established by OPLS-DA, Q2 values were greater
   than >0.8([74]Fig. 3d-f), and there is no overfitting phenomenon,
   indicating that OPLS-DA has a high separation ability in this study.

3.2. Differential metabolite analysis

3.2.1. Differential metabolite screening

   Based on VIP of OPLS-DA model, we found that there were significant
   differences in VOCs variance among different producing areas. Then, we
   further screened differential metabolites, with specific screening
   conditions as follows: fold change≥2 or fold change≤0.5, VIP ≥ 1, p
   value <0.05, a total of 249 differential metabolites were screened
   (Table S2), and these differential metabolites came from 15 types of
   substances, among which esters (n = 40) accounted for the largest
   proportion. Followed by terpenoids (n = 39), heterocyclic compounds
   (n = 38), hydrocarbons (n = 22), ketones (n = 20), aldehydes (n = 20),
   aromatics (n = 17), alcohols (n = 16), amines (n = 14),
   sulfur-containing compounds (n = 6), phenols (n = 6), acids (n = 5),
   others (n = 4), halogenated hydrocarbons (n = 1), and ethers (n = 1).
   Compared with Minqin A. mongolicum, 148 metabolites (80 up-regulated
   and 68 down-regulated) were found in Alxa A. mongolicum ([75]Fig. 4a),
   102 metabolites (7 up-regulated and 95 down-regulated) were found in
   Tengger A. mongolicum ([76]Fig. 4b). Compared with Tengger A.
   mongolicum, 194 metabolites (149 up-regulated and 45 down-regulated)
   were found in Alxa A. mongolicum ([77]Fig. 4c), indicating that more
   volatile substances accumulated in Alxa A. mongolicum than in Minqin A.
   mongolicum and Tengger A. mongolicum, and the different-expressed VOCs
   may be an important factor in the aroma formation of Alxa A.
   mongolicum. Detailed analysis showed that compared with Minqin A.
   mongolicum, the 80 VOCs significantly accumulated by Alxa A. mongolicum
   were mainly esters (15 kinds), followed by terpenoids (14 kinds),
   heterocyclic compounds (14 kinds) and aromatics (8 kinds). Compared
   with Tengger A. mongolicum, terpenoids (27 kinds) were the main VOCs
   accumulated in 149 VOCs, followed by esters (25 kinds), heterocyclic
   compounds (23 kinds) and hydrocarbons (13 kinds). From Venn diagram
   ([78]Fig. 4d), it can be seen that Alxa A. mongolicum and other local
   A. mongolicum share common metabolites and some unique metabolites.
   However, it is interesting to note that terpenoids and esters accounted
   for a small proportion of significantly down-regulated VOCs, while
   aldehydes accounted for a large proportion. The significantly
   upregulated terpenoids and esters are (1-Oxaspiro[4.5]dec-6-ene,
   2,6,10,10-tetramethyl-), (Bicyclo[3.1.0]hex-3-en-2-one,
   4-methyl-1-(1-methylethyl)-), (Propanoic acid, butyl ester),
   (Cis-cyclohexanone, 5-methyl-2-(1-methylethyl)-), (Diethyl malonate),
   (Propanoic acid, 2-methyl-, phenylmethyl ester), (Butanoic acid,
   2-methyl-, 2-methylbutyl ester), etc. The results indicated that
   terpenoids and esters might be the key substances for good flavor of
   Alxa A. mongolicum.

Fig. 4.

   [79]Fig. 4
   [80]Open in a new tab

   Volcano plot (a, b, c)and Venn diagram (d) of differential VOCs in A.
   mongolicum. (a) A vs M, (b) T vs M, (c) A vs T. A = Alxa A. mongolicum,
   T = Tengger A. mongolicum, M = Minqin A. mongolicum.

3.2.2. KEGG pathway analysis of differential metabolites

   We mapped these 249 metabolites to the KEGG database, and conducted
   pathway enrichment analysis of the metabolites significantly expressed
   in different habitats of the A. mongolicum, and identified the main
   pathways involved in these metabolites.

   Compared with Minqin A. mongolicum, the main pathways of flavor
   substances significantly enriched in Alxa A. mongolicum were
   monoterpenoid biosynthesis (3 metabolites), secondary metabolites
   biosynthesis (8 metabolites), zeatin biosynthesis (1 metabolite),
   α-linolenic acid metabolism (2 metabolites), metabolic pathways (8
   metabolites), etc. ([81]Fig. 5a).

Fig. 5.

   [82]Fig. 5
   [83]Open in a new tab

   KEGG enrichment map of differential VOCs of A. mongolicum under
   different growing conditions.

   (a) A vs M, (b) T vs M, (c) A vs T. A = Alxa A. mongolicum, T = Tengger
   A. mongolicum, M = Minqin A. mongolicum.

   Compared with Minqin A. mongolicum, the main pathways of flavor
   substances significantly enriched in Tengger A. mongolicum were
   monoterpenoid biosynthesis (1 metabolite) and biosynthesis of secondary
   metabolites (2 metabolites), etc. ([84]Fig. 5b).

   Compared with Tengger A. mongolicum, the main pathways of flavor
   substances significantly enriched in Alxa A. mongolicum were
   monoterpenoid biosynthesis (3 metabolites), biosynthesis of secondary
   metabolites (9 metabolites), zeatin biosynthesis (1 metabolite),
   α-linolenic acid metabolism (2 metabolites), metabolic pathways (8
   metabolites), etc. ([85]Fig. 5c). Obviously, the differential
   metabolites cis-3-Hexenol and cis-3-Hexenal of Alxa A. mongolicum were
   significantly enriched in α-linolenic acid metabolism level compared
   with Minqin A. mongolicum and Tengger A. mongolicum (Table S2).

3.2.3. Analysis of sensory flavor characteristics of differential VOCs.

   A. mongolicum is widely consumed for its unique flavor and nutritional
   value. Flavor components can objectively reflect the flavor
   characteristics of different samples and are important indexes for
   evaluating flavor quality. Sensory analysis of A. mongolicum can be
   used to further understand their contribution to overall flavor. Make
   use of these databases of odour sensory information
   ([86]https://www.odour.org.uk; [87]https://www.flavornet.org;
   [88]https://www.femaflavor.org/flavor-librar y), The 249 differential
   metabolites obtained by screening were annotated to sensory flavor
   features, and the top 10 sensory flavors with the highest number of
   annotations were selected to draw the radar map ([89]Fig. 6).

Fig. 6.

   [90]Fig. 6

   [91]Fig. 6
   [92]Open in a new tab

   Analysis of sensory flavor characteristics of differential VOCs

   (a) A vs M, (b) T vs M, (c) A vs T. A = Alxa A. mongolicum, T = Tengger
   A. mongolicum, M = Minqin A. mongolicum.

   Compared with Minqin A. mongolicum, there are 140 sensory flavors
   annotated by flavor substances of Alxa A. mongolicum. The main flavors
   are green (25), fruity (22), sweet (21), floral (8), spicy (8),
   metallic (7), rose (7), almond (6), apple (6), grassy (6); Among them,
   spicy and sweet substances are listed in [93]Table 1 (A vs M). Compared
   with Tengger A. mongolicum, there are 156 sensory flavors annotated by
   flavor substances of Alxa A. mongolicum. The main flavors are green
   (34), fruity (25), sweet (25), tropical (11), citrus (9), floral (9),
   spicy (9), apple (8), fresh (8), herbal (8); Among them, spicy and
   sweet substances are listed in [94]Table 1 (A vs T). Compared with
   Minqin A. mongolicum, there are 98 kinds of sensory flavors annotated
   by flavor substances of Tengger A. mongolicum. The main flavors are
   fruity (15), green (14), sweet (11), tropical (7), herbal (6), roasted
   (6), apple (5), banana (5), citrus (5), fatty (5); Among them, spicy
   and sweet substances are listed in [95]Table 1 (T vs M). Therefore,
   more volatile substances were accumulated in Alxa A. mongolicum than in
   Minqin A. mongolicum and Tengger A. mongolicum, and the overlaps of
   differently expressed volatile organic compounds might explain the
   unique flavor formed by Alxa A. mongolicum.

Table 1.

   The difference in the spiciness and sweetness of the A. mongolicum in
   different locations.
   A vs M A vs T T vs M
   Spicy substances (E,E)-2,4-Hexadienal)，
   (2-Heptanone)，
   (4-Hexen-3-one)，
   (1-Oxaspiro[4.5]dec-6-ene,2,6,10,10-tetramethyl-)，
   (2-Propen-1-ol, 3-phenyl-)，
   (3-Phenyl-2-propen-1-ol)，
   (2-Propenal, 2-methyl-3-phenyl-)，
   (p-Mentha-1,8-dien-7-ol) ((E,E)-2,4-Hexadienal)，
   (4-Hexen-3-one)，
   (2-Octanol)，
   (1-Oxaspiro[4.5]dec-6-ene, 2,6,10,10-tetramethyl-)，
   (2-Propen-1-ol, 3-phenyl-)，
   (3-Phenyl-2-propen-1-ol)，
   (2-Propenal, 2-methyl-3-phenyl-)，
   (p-Mentha-1,8-dien-7-ol)，
   (3-Octanol). 3-Octanol
   Sweet substances (1-Propanone, 1-(2-furanyl)-)，
   ((E,E)-2,4-Hexadienal)，
   (2-Butenal, 3-methyl-)，
   (2-Furanmethanol)，
   (2-Heptanone)，
   (2-Hexenal)，
   (3-Hexanone)，
   (4-Ethylbenzaldehyde)，
   (2-Acetylfuran)，
   (2-Methylnaphthalene)，
   (Toluene)，
   ((E)-2-Buten-1-one, 1-(2,6,6-trimethyl-1,3-cyclohexadien-1-yl)-)，
   (2-Propen-1-ol, 3-phenyl-)，
   (3-Phenyl-2-propen-1-ol)，
   (2-Propenal, 2-methyl-3-phenyl-)，(Acetophenone)，
   (BenzAldehyde, 3-methyl-)，
   (Butanoic acid, 2-methyl-, 2-methylpropyl ester)，
   (Diethyl malonate)，
   (Isobutyl isovalerate)，
   (Propanoic acid, 2-methyl-, phenylmethyl ester). (1,3-Dioxolane,
   4-methyl-2-(2-methylpropyl)-)，
   ((E,E)-2,4-Hexadienal)，
   (2-Butenal, 3-methyl-)，
   (2-Furanmethanol)，
   (2-Hexenal)，
   (4-Ethylbenzaldehyde)，
   (2-Methylnaphthalene)，
   (Toluene)，
   (Lilac Aldehyde C)，
   (Lilac Aldehyde D)，
   ((E)-2-Buten-1-one, 1-(2,6,6-trimethyl-1,3-cyclohexadien-1-yl)-)，
   (2-Propen-1-ol, 3-phenyl-),
   (3-Phenyl-2-propen-1-ol),
   (2-Propenal, 2-methyl-3-phenyl-),
   (Acetophenone),
   (BenzAldehyde, 3-methyl-),
   (Diethyl malonate),
   (Propanoic acid, 2-methyl-,
   phenylmethyl ester),
   (2H-Pyran-2-one, tetrahydro-),
   (Benzenemethanol, .alpha.,4-dimethyl-),
   (Butanoic acid, 2-methyl-, 2-methylbutyl ester),
   (Butanoic acid, 2-methyl-, 3-methylbutyl ester),
   (Butanoic acid, 3-methyl-, 3-methylbutyl ester),
   (cis-menthone),
   (Propanoic acid, butyl ester). (1-Propanone, 1-(2-furanyl)-),
   (3-Hexanone),
   (Butanoic acid, 2-methyl-, 2-methylpropyl ester),
   (Isobutyl isovalerate),
   (1,3-Dioxolane, 4-methyl-2-(2-methylpropyl)-),
   (cis-menthone),
   (Lilac Aldehyde C),
   (Lilac Aldehyde D),
   (Propanoic acid, butyl ester),
   ((Z)-1,3,6-Octatriene, 3,7-dimethyl-),
   ((E)-2-Hexen-1-ol, acetate)
   [96]Open in a new tab

4. Discussion

   As a characteristic vegetable with unique flavor and high nutritional
   value in desert, A. mongolicum is commonly used to please seasoning and
   assisted to cook delicious dish, providing health benefits to human
   body ([97]Wang et al., 2019). Alxa A. mongolicum had high levels of
   total amino acids and umami amino acids. Principal component analysis
   revealed a strong correlation between the sweet and savoury
   characteristics of A. mongolicum and the presence of sweet free amino
   acids, suggesting that A. mongolicum have a significant advantage in
   flavor development ([98]Hou, 2020). Esters, aldehydes and sulfur
   compounds were found to be the main volatile oil components in A.
   mongolicum ([99]Liu, Zhao, & Yang, 2007; [100]Wuren, 2011). Previous
   studies have shown that different extraction conditions and the
   selection of extraction head will affect the incompatibility of peak
   surface abundance and polarity of volatile components, which easily
   leads to the instability of component determination ([101]Li et al.,
   2019). Wang et al. detected a total of 24 compounds by headspace
   solid-phase microextraction combined with GC, and the volatile
   components were mainly 2-hexenal and thioether ([102]Wang, Yang, & Bao,
   2012). Wen et al. identified a total of 31 compounds by supercritical
   fluid extraction and Soxhlet extraction ([103]Wen, Liu, & Gao, 2018).
   Considering the instability and influence of traditional extraction
   methods on the detection results, some scholars turned to headspace
   solid phase microextraction (HS-SPME) combined with gas
   chromatography–mass spectrometry to study the volatile components of
   plants, and the detection results were relatively stable ([104]Hou,
   2020). In this study, a total of 624 volatile compounds were identified
   from three habitats by HS-SPMEGC/MS analysis (Table S1). Obviously,
   different extraction methods affect the types of volatiles. This may be
   due to the extremely complex composition of flavor substances, volatile
   substances are unstable and different experimental processes will
   greatly affect the experimental results ([105]Shi, 2008). The results
   of principal component analysis showed that metabolites of different
   species and varieties had different characteristics. The OPLS-DA model
   with high R^2Y and Q^2 values is established. This study provides
   metabolic evidence for understanding flavor differences in A.
   mongolicum.

   Flavor is one of the most relevant factors for consumers to evaluate
   quality, and their purchase preferences are related to both consumer