Differences between two plants fruits: Amomum tsaoko and Amomum maximum, using the SPME-GC–MS and FT-NIR to classification

2.2.1 Solid-phase microextraction (SPME)

The SPME of information: Equipment: CTC Trinity Autosampler; Extraction: 50/30 μmDVB/CARonPDMS; Temperature: 50℃; Time: shock 15 min, extraction: 30 min, Concussion speed: 250 rpm; Parsing time: 5 min; GC cycle time: 50 min.

2.2.2 Operating conditions

Used DB-wax (30 m × 0.25 mm × 0.25 µm), the derivative substances were separated by constant flow helium of 1 mL/min, 0.5 g sample was added to the 20 mL headspace bottle and sealed. The temperature of the injection port is 260℃, the initial temperature is 40℃, the 5 min is maintained, the temperature rises to 220℃ at the rate of 5℃/min, and the temperature rises to 250℃ at the rate of 20℃/min, keeping 2.5 min. Interface temperature: 260 °C; Ion source temperature: 230 °C; Four-stage rod temperature: 150 °C; Ionization mode: EI+,70 ev; Scanning method: full scan; Mass range: 20 ∼ 400. The instrument GC–MS shown in Table 1. NIST2014 of spectral library. All metabolites were obtained from the following databases: Panomic’s own standard database (Suzhou, China) and the PubChem data (PubChem (nih.gov)).

2.2.3 Data optimization

Firstly, obtain the names of possible metabolites in each sample and their retention time, CAS number, related content, and other information through database annotation and integrate that annotation information of each sample to obtain the final table of possible metabolites for analysis. Secondly, to enable the data of different magnitudes to be compared, the standard internal normalization of the peak area is performed on the data. Internal standard: Benzene, 1,4-dichloro-, 100 μg/mL*10 μL, which is convert the relative content of each component.

2.4.1 Overall metabolite hierarchical clustering analysis

Conduct cluster analysis on the metabolite data to explore differences between the two species. Cluster heat maps showing the correlation of the data were drawn using R software. The heat map is the quality control of experimental data and the display of differential data. A single heat map is composed of a heat map body and heat map components. In the study, we adapt that agglomerate hierarchical clustering: Group each object into one class and merge these classes into larger and larger objects until the end. The data set is then scaled using the heatmap package in R (v3.3.2) and the relative quantitative values of the metabolites. Hierarchical clustering analysis is suitable for evaluating the similarity between samples. The distance calculation method, Euclidean clustering, calculates the pairwise distance between the respective data points of the two categories (Yuan et al., 2012).

2.4.2 Statistical analysis of differential metabolites

A Venn diagram is a graphical representation showing the overlapping areas of elements. It visually represents the similarity and overlap of different metabolite compositions in various comparison sets.

The box plots: A box plot is a statistical chart used to display the dispersion of a set of data, providing a relatively intuitive way to visualize the data distribution characteristics. Each data group can present its minimum, maximum, and average, and the range between the minimum and maximum values reflect the degree of variation in the data. Bar chart: The bar chart is a statistical report chart that uses the length of a rectangle as the variable for the graphical representation. It consists of a series of vertical bars of varying heights that indicate the distribution of the data, along with error lines indicating the standard error.

2.4.3 Multivariate statistical analysis

PCA generates new characteristic variables by linearly combining metabolite/spectrum variables with certain weights, categorizing each data group by the primary new variable (principal component). On top of that, it also removes poorly reproducible samples (outlier samples) and abnormal samples (samples outside the confidence interval, Hotelling T² ellipse). The reliability of the mathematical model calculated by PCA needs to be rigorously validated. An unreliable mathematical model fails to characterize the metabolomics data well and may seriously affect the acquisition of correct results or even mislead the analysis results. Therefore, the constructed model needs to be validated. The cross-validation of models mainly refers to parameters such as R²X, which is the interpretability degree of the model. Usually, R² higher than 0.5 is better.

The unsupervised analysis method must recognize intra-group errors, eliminate random errors unrelated to the study’s purpose, avoid excessive focus on details at the expense of overall patterns, and ultimately facilitate the discovery of inter-group differences and differential compounds. Therefore, this study sticks to this point, thus proposing using PLS-DA for analysis. PLS-DA commonly used pattern multivariate statistical analysis tool, which was a match for dealing with highly correlated variables. It combines the regression model with dimensionality reduction and utilizes specific discriminant thresholds to analyze the regression results. Compared with PLS-DA, OPLS can effectively reduce the model's complexity and enhance the model's explanatory power without reducing the model's predictive power, thus maximizing the variance of the formation. (Document of SIMCA-P 13.0) R2 is the goodness of fit, which is the percentage of the variation between the training set Y explained by the model and the partial least squares. as the degree of fit of the model to the data. Close to the value of 1 to the R² is a necessary condition for the existence of an excellent model. Q² is the prediction superiority as the percentage of variance between the cross-validated prediction training set X and the partial least squares. Q² indicates the degree to which the new data are predicted, and larger values, indicate that the model has good prediction performance.

3.1.1 Multivariate statistical analysis

A total of 101 kinds of metabolites in D and F were detected, shown in Table S1. The differences between species are visualized from the PCA plot, and it is clear that D and F have been separated, but for F, its poor clustering effect is shown in Fig. 2 A. Data dimensionality is reduced for visual sample analysis. The metabolite PCA score shows that the two principal components have successfully explained 64.1 % of the information with an R²X (cum) value of 0.641 and a pre of 2. The biological repeat distribution of F is closer than that of D, indicating that the intraspecific variation of F is relatively tiny. Moreover, the distribution of D is also different from that of F, thus indicating the validity of the model, technical variability, and lower level of biological variation.

Close

Fig. 2

GC–MS of the PCA and PLS-DA model. (A1: PCA loading plot; A2: PCA Score Plot; B1: permutation test; B2: PLS-DA Score Plot; C: OPLS-DA.).

Export to PPT

HCA of the GC–MS data revealed that the six samples clustered into two major groups (Fig. S1), indicating differences in metabolites among the samples. The HCA diagram shows that D and F were strictly divided into two main components. This result is consistent with the PCA, where F acquired by growers could cluster together. At the same time, there was a clear difference between the D samples purchased through the market.

We can tell from the final results that D compounds were divided into 17 classes (the following only lists categories more remarkable than 3 %), including the terpenes (37.9 %), aldehydes (12.1 %), organic heterocyclic compounds (7.6 %), alcohols (6.1 %), aromatic (4.5 %), esters (4.5 %), ketones (4.5 %), peptide (3 %) being the predominant metabolite classes. On the contrary, F is divided into a total of 17 categories, of which, as the main metabolic categories, are terpenes (37.5 %), esters (12.5 %), alcohols (9.4 %), aldehydes (6.3 %), alkanes (7.8 %), aromatic (3.1 %), carboxylic acid (3.1 %), olefin (3.1 %). A more specific visualization of the classification information is shown in Fig. 3.

Close

Fig. 3

GC–MS of the D and F comparison of various metabolite. D: Amomum tsaoko Crevost et Lemarie;F: Amomum maximum Roxb.).

Export to PPT

Finally, a total of 101 metabolites were detected. The cluster analysis was used to determine the expression of metabolite relevance under different conditions or bases of metabolite visualization. A hierarchical clustering diagram of relative quantitative values of metabolites will be obtained, as shown in Fig. 4. The magnitude of relative content in the graph is shown by the difference in color, where the columns represent samples and the rows represent metabolites. The above results indicate that the composition of metabolite classes was approximately the same between the studied D and F species.

Close

Fig. 4

The clustering heat map of metabolites of two species.D1, D2, D3: Amomum tsaoko Crevost et Lemarie; F1, F2, F3: Amomum maximum Roxb.).

Export to PPT

Subsequently, using the HCA to reveals the magnitude of similarity between samples. Metabolite distribution and production are usually due to plant growth environment, growth stage, internal gene regulation, plant site differences, etc. According to the results obtained in this test, the amounts of D and F are also relatively close for terpenoids. D contains abundance of aldehydes more than F. For aromatic, this class of metabolites, as the aromatic odor of plants, has more D than F. The results of the comparison of the above two detected metabolite components are shown in the Venn diagram (Software: OriginPro 2022, ChemBioDraw Ultra 14.0, Figure S2). There are 29 metabolites in common; D and F have 37 and 35 different metabolites, respectively. Please refer to Figure S5-S7 and Table S5 for the specific chemical structures.

Histograms of two species show multiple metabolic species' relative content (Figure S3). Among the first five metabolites detected, the D fruit contained the following substances: Limonene, *-Terpineol, *-Muurolene, Copaene, 2-Decenal, (Z)-. F included Bicyclo[3.1.1]heptane, 6,6-dimethyl-2-methylene-, (1S)-, Naphthalene, decahydro-4a-methyl-1-methylene-7-(1-methylethenyl)-, [4aR-(4a*,7*,8a*)]-, 1,3,7-Octatriene, 3,7-dimethyl-, Humulene, (-)-*-Bourbonene exhibited a higher relative content. These substances were inconsistent in each of the replicate samples, indicating a large genetic diversity in the two species.

The fruit of D, its unique flavor is associated with the composition of aromatic, stimulating compounds, most of which are monoterpenes, of which TPS is the critical enzyme involved in the biosynthesis of D. The researcher attempts to elucidate who utilized the crude enzyme extract of A. tsaoko combined with geraniol would have stimulatory expression (Ping Li et al., 2022). The metabolites of aromatics include Estragole, p-Xylene, Toluene and Creosol.

The Fig. 2 B reveal the total explanatory power is 63.8 %. Furthermore, it is worth mentioning that PLS-DA can extract the variation information between groups more efficiently and maximize the separation of assigned groups.

Finally, the metabolite with a VIP value greater than 1 under the OPLS-DA condition was selected as the difference value. OPLS-DA can decompose the X matrix information into two types of information related to Y and irrelevant. Then, after filtering out the information irrelevant to the classification, the relevant information mainly concentrated in the first prediction component, and the principal component 1 of this test was 39 %. Table 2 indicates the specific model parameters of GC–MS.

Conditions of differential metabolite screening: The differential metabolite biomarker was searched for by screening metabolites, and the conditions of relevant metabolite screening were p-value ≤ 0.05 + VIP ≥ 1 (Trygg and Wold, 2002, Haspel et al., 2014, Wang et al., 2014, Kieffer et al., 2016).

3.1.2 Discussion: The metabolites in the samples of analysis

The aromatic and pungent odor of D is associated with its rich terpene content. The metabolites of the two species that contain the most classes are terpenes. Terpenoids are a general term for a class of hydrocarbons with the molecular formula of isoprene integer ((C₅H₈)n), which are widely found in plants and marine organisms. These molecules can also udergo oxidation, rearrangement, and other chemical structural modifications to produce various alcohols, aldehydes, ketones, carboxylic acids, esters, and other terpenoid structures collective referred to as terpenes in a broad sense. Terpenoids (Boysen and Hearn, 2010) are the main natural products of many plant essential oils. They are used in large quantities in the food industry, cosmetics and even drugs. Such as limonene (C₁₀H₁₆), a monoterpene molecule with lemon aroma, or paclitaxel, a tricyclic diterpene with anticancer activity. Terpene synthase (TPS) has been found to be in terpenoid synthesis. The precursors were converted to monoterpenes, sesquiterpenes, and diterpenes, respectively (Wang et al., 2018a; Wang et al., 2018b). Terpenes are rightfully considered one of the largest classes of plant primary metabolites. Such metabolites can protect plants from environmental stress, surprisingly, have defenses against pathogens and herbivores (Wagner and Elmadfa, 2003, Huang et al., 2010, Goyal et al., 2012). The chiral identification of monoterpenes is crucial in the fields of spice chemistry and pharmacology (Boysen and Hearn, 2010). Monoterpenes and sesquiterpenes are the main components of volatile oils. Diterpenes are the main substances for forming resins; triterpenes are important substances for forming plant saponins and resins; tetraterpenes are mainly some fat-soluble pigments widely distributed in plants. Some of them have physiological activities, such as roundworm repellent, santonin with roundworm repellent effect, artemisinin with antimalarial effect, and andrographolide with antibacterial effect. Monoterpenes are synthesized in the plastids via the DOXP pathway; fatty acid derivatives are mainly produced by intermediates of the lipoxygenase-peroxidase pathway (Verdonk et al., 2003). Sesquiterpenes are often associated with mechanisms of organism protection. For instance, in marine green algae, in marine green algae, they are stored as polyacetates products and released as deterrent sesquiterpenes upon injury (Boysen and Hearn, 2010). The plant hormone abscisic acid is a sesquiterpene that induces fruit ripening and defoliation. Diterpenes can inhibit cell division and use in anticancer. Plant-derived diterpenes products have gibberellins and phytoalexins (defense) (Boysen and Hearn, 2010). This report shows a greater concentration on monoterpenes and sesquiterpenes for D and F. Sesquiterpenoids associated with spicy taste, Caryophyllene (Feng et al., 2017). Based on relevant evidence, isoprene, and β-caryophyllene have ecological functions as core components of plant signaling networks (Frank et al., 2021). Both have allelopathic chemical components that induce resistance to microbial pathogens in neighboring plants. This knowledge can be used to enhance plant resistance in various crop management programs, providing theoretical support.

Other reports that by GC–MS, were identified 73 compounds, including that monoterpenoids, sesquiterpenoids, diterpenoids, triterpenoids, diarylheptanoids, flavones, phenolic, etc. The main compositions were 1,8-cineole, ρ-propylbenzaldehyde, geraniol, geranial, α-terpineol, α-phellandrene (Yang et al., 2008) and fatty acid (Hang Liu et al., 2018).

Their aromatic and pungent properties are attributed to volatile and non-volatile, respectively (Starkenmann et al., 2007). It has been shown that there is a positive correlation between altitude and the yield of metabolites in D fruits, especially the aromatic components of the fruits, including 1,8-cineole, trans-citral, and the pungent compound, the decreasing trend at high altitudes of > 2000 m (Ping Li et al., 2022). An investigation was carried out to assess the presence of monoterpene hydrocarbons, oxygenated monoterpenes, sesquiterpene hydrocarbons, oxygenated sesquiterpenes, and other compounds in the fruits of D. Generally speaking, 1,8-Cineole is a kind of the oxygenated monoterpene, the main aroma component, with camphor, a cool and spicy flavor, as the most important constituent in oils (45.24 %) (Yang et al., 2010, Cui et al., 2017, Na Guo et al., 2017). Easily affected by light, temperature(Yi Lin et al., 2022), etc. However, the results of the current study did not find a clear explanation. This could be due to the prolonged storage time of the samples, the mixed storage of previous years by the vendors, the low content of growth and metabolism, or the differences in the methods of detection instruments. Then, there are no reports on the effects of storage periods of A. tsaoko on the quality of its processed products. Therefore, the storage period of A. tsaoko will become a hot research topic in the following period. Compounds will be decomposed and transformed under certain conditions. For example, limonene and linalool can be transformed into α-terpineol under certain conditions (Haleva-Toledo et al., 1999). D volatiles component has sedative, analgesic and hypnotic effects. The fruits of the 1,8 -Cineole make up 45.2 % (Sabulal and Baby, 2021). Studies have confirmed that the strong spice smell released by the crushed fruit is mainly related to 1,8-Cineole, α-phellandrene, α-pinene, β-pinene, and 2-isopropylbenzaldehyde (Zhangguo Lu et al., 2010). As a result, Pinene, as a typical terpene, is present in D and F. Caryophyllene has three isomers α, β, and γ. β-Caryophyllene is a bicyclic sesquiterpene also common in spices that is a ligand of the cannabinoid receptor 2 (CB2), which can be decrease pain for inflammatory responses and enhances wound healing (Koyama et al., 2019). Moreover, the above metabolites are only found in D. This exploratory study, for the first time, found Neburon, but it’s only found in D1, only skeptical about such compounds. Since it is impossible to test the sample grass fruit again directly, there is only one guess: It is likely that the growers used it to remove weeds, and the residues were absorbed from the soil during the growth and development of plants (JR. et al., 1981).

Aromatic compounds have excellent stability, which is related to their fully conjugated monocyclic compounds structure with a planar geometry. According to classical compounds, they must have a number of π-electrons equal to 4n + 2, where n = 1, 2, 3, etc. On the other hand, benzenoid compounds are collections of benzene rings with at least one joint side. One of the properties of these classical compounds is that the double bonds do not participate in addition reactions but perform electrophilic substitution reactions, thus preserving the stability of the π-electron system. Except in the case of C-C cleavage at high temperatures or by enzymatic action (Poater et al., 2022). Furthermore, this likely is one of the reasons why the aromatic flavor of the herb fruit is diminished after cooking at high temperatures. L-Cysteine (L-Cys), a non-essential amino acid, was detected in F at a low level of 1.6 %. L-Cys or its esters with aldehydes or ketones react with were obtained the reduced thiazole derivatives.

Two peptides, Lys-Cys and Ala-Gly, were present in this assay and were detected only in D. It has been reported that Lys-Cys is associated with the synthesis of hydrogels and bifunctional supervalent iodine reagents (Chowdhuri et al., 2020, Ceballos et al., 2021). It has been experimentally demonstrated that the mechanism of dipeptide inhibition was simulated using molecular dynamics (MD), and the results showed that the N-terminal end of the dipeptide is the key to the inhibition. Ala-Gly was appraised as an environmentally benign hydrate inhibitors for CH₄ hydrate (Woojin Go et al., 2022). Volatile metabolites are mainly derived from the chemical or enzymatic oxidation of unsaturated fatty acids and from further interactions with proteins, peptides,and free amino acids (Feng et al., 2017).

Inevitably, some pesticides appeared unexpectedly in this study. The source of these pesticides is not the plant's metabolism. During the growth process of grass and fruits, some weeds will appear around the plant, competing for nutrient resources and inhibiting photosynthetic agents for crop weeds. Therefore, growers will use herbicides to inhibit the growth of weeds around the plant and meet the needs of plant growth. And the choice of pesticides needs to comply with the relevant laws and regulations (Institute for the Control of Agrochemicals, 2018).

Esters, with more esters in F, compared to D. The proper level of esters will bring out the fruit flavor. Three were detected in D and eight were detected in F. Most of the naturally occurring fats and oils are fatty acid esters of glycerol. Low molecular weight esters are commonly used as fragrances and are found in essential oils and pheromones, and are more volatile (Sparkman et al., 2011). Esters are polar molecules but have lower boiling points than carboxylic acids and alcohols of similar molecular weight because there are no intermolecular hydrogen bonds between ester molecules. Esters are usually derived from inorganic or organic acids in which at least one hydroxyl group (OH) is replaced by an alkoxy group (-O-alkyl), most often from carboxylic acids and alcohols (Speight, 2011).

The metabolism of amino acids and the oxidation of lipids can decarboxylate to the corresponding aldehydes. Benzaldehyde has an almond and nutty odor and is present in both species. (Verma and Srivastav, 2020) Benzenoid and phenylpropanoid-related volatiles are derived from the common precursor L-phenylalanine (Verdonk et al., 2003). Benzaldehyde, benzyl alcohol, and benzoic acid are the most common tendencies(Van Moerkercke et al., 2009). A volatile substance, benzaldehyde, exists together in two fruits.

Ketones and phenolic compounds are considered to contribute to the antioxidant potential. Phenols were present in a small percentage of this assay, so it can be determined that phenols do not act as landmarks for D and F. However, phenolics' advantages (HVA) for treating diseases are undeniable. HVA contained in D is a major terminal metabolite of dopamine and one of the three catecholamines produced in the brain. Dopamine is broken down by the liver and excreted as HVA in the urine (Cynthia C. Chernecky PhD, 2013). Also, as a marker of metabolic stress, it is used as a reagent for detecting of oxidative enzymes (Szekely and Didaskalou, 2016). HVA is related to Aromatic L-amino acid decarboxylase deficiency (AADCD) (Manti et al., 2022), Celiac Disease (De Grandis et al., 2010), Breast cancer (Zniber et al., 2022) and Growth-Hormone Deficiency (Zielonka et al., 2015). The 4-Hydroxybenzoic acid detected in F is a phenolic derivative of benzoic acid, which can inhibit most Gram-positive and some Gram-negative bacteria. A study found that p-hydroxybenzoic acid has antiviral activity against SARS-CoV-2. Most promisingly, the study acted on Candida albicans by gamma irradiation, which led to the biotransformation of caffeic acid, resulting in a multiplication of its yield (Cho et al., 1998, Singab et al., 2022). Hence, based on the aforementioned results, it can be reaffirmed that D and F exhibit antibacterial effects.

The ketones detected this time contained: 2-Nonanone, 3,4,2,4,6-Pentahydroxychalcone and 2-Undecanone. 2-Undecanone is a volatile organic compound that inhibits DnaKJE-ClpB diketone-dependent refolding of heat-inactivated bacterial fluorophore enzymes. 2-Undecanone can inhibition of tumorigenesis (Melkina et al., 2017, Lou et al., 2019). Flavonoid metabolites in plants are often considered to have neuroprotective, antioxidant, and anticancer properties (Hostetler et al., 2017, Madunic et al., 2018). 3,4,2,4,6-Pentahydroxychalcone belong to the chalcone. Also, relevant chemical elements are classified as minor flavonoids, which play essential roles in plants. It is not a negligible part of their success in adapting to life as sedentary organisms. (Bentrad and Hamida-Ferhat, 2022). D contains ketone compounds, indicating that D is likely to have high medicinal value.

The above study proved that, in general, D contains more types of pharmacologically active substances than F and is relatively rich in them. To a certain extent, this study provides evidence that the pharmacological activity of D is more robust than that of F. It also proved that F has an acidic stimulating sensation regarding taste and flavor.

3.1.3 Discussion: Differential metabolite analysis

A total of 19 differential metabolites were screened, of which 12 were down-regulated and seven were up-regulated (D relative to F, the content of 12 metabolites was higher than that of F, and the content of 7 metabolites was lower than that of F. These differential metabolites are shown in the Table S2. Differential metabolite screening conditions: p-value ≤ 0.05, VIP ≥ 1.

In this study, as listed in Table S1, Butyrolactone, 2-Nonanone, Phenol, 3-methyl-, Benzaldehyde, 4-(1-methylethyl)-, *-Muurolene, these types of substances behave up-regulated in D. Bicyclo[3.1.1]heptan-3-ol, 6,6-dimethyl-2-methylene-, [1S-(1*,3*,5*)]- and Cyclohexene, 1-methyl-4-(1-methylethylidene)-, it is up-regulated relative to D. *-Myrcene, 1-Octanol, Naphthalene, decahydro-4a-methyl-1-methylene-7-(1-methylethenyl)-, [4aR-(4a*,7*,8a*)]-, (-)-*-Bourbonene, Butanoic acid, methyl ester, Tetradecane, Pentadecane, Bicyclo[2.2.1]heptan-2-ol, 1,7,7-trimethyl-, acetate, (1S-endo)-, Octanoic acid, methyl ester, (S,1Z,6Z)-8-Isopropyl-1-methyl-5-methylenecyclodeca-1,6-diene, it is obviously only detected in F. Tridecane and Bicyclo[3.1.1]heptane, 6,6-dimethyl-2-methylene-, (1S)-, it exists in the two, and shows a down-regulation to D. Tridecane is a differential metabolite, only exists in D3. All samples of F contain this substance. Tridecane is a component of the plant, playing a role in plant metabolites and volatile oils. The substance is downregulated.

Butanoic acid, methyl ester is one of the aromas with an apple, banana, and pineapple fragrance in low concentration (Yun-tao Zhang et al., 2009). Tetradecane was also found in mango, which is a recognized fruit aroma substance. And owing to a kind of volatile compounds (Campos Alencar Oldoni et al., 2022). When the thrips felt conspecific extract by the Liothrips jatrophae, which include the tridecane, pentadecane, teradecyl isobutyraate and hexadecyl isobutyrate regards as the major ingredient. That will make thrips show threatened behavior, such as rapid sideways movement and others’ reactions (González-Orellana et al., 2021).

Regarding butyrolactone, there is a professional report that γ-butyrolactone and oxazolidinone-based ligands are pharmacologically necessary scaffolds in drug research (Bhandare et al., 2022). Relevant research indicates and proposes key markers for distinguishing the incorporation of quinoa flour into rice and wheat flour, particularly four volatile substances, including butyrolactone, and five other aromatic compounds (Yang et al., 2022). The relevant reached herein that the final heterocycles could be easily transformed γ-butyrolactone attribute to a straightforward photochemical method (Ye et al., 2022).

2-Nonanone is a colorless to light yellow liquid with fruit, flower, oil and herb-like aroma. It is reported that males of A. mysticus produce include two component mixture of (R)-3-hydroxy-2-hexanone and 2-nonanone. This pheromone can attract the aggregation of females and males of the same species (Molander et al., 2019).

Phenol,3-methyl-, is also a raw material for highly effective and low toxicity pesticides, such as speedwell and is used in the manufacturing of spices and resins. The presence of the substance is most likely due to the spraying of pesticides by growers during the growth of plants. It is obvious that the pesticide residue was found only in D and at relatively high levels. However, we were unable to determine whether the pesticide residue was enriched in the seed coat or the kernel. If a large amount of D is used, it is one of the breakthrough points that needs to be thoroughly investigated in subsequent studies to determine if it poses a potential health hazard to humans.

Benzaldehyde, 4-(1-methylethyl)- is a natural aldehyde that inhibits α-synuclein fibrillation and cytotoxicity, and has anticancer activity. This metabolite product, we also called Cuminaldehyde. It is direction a particular unit concentration of this substance can inhibit the biofilm formation of Pseudomonas aeruginosa and thus act against various acute and chronic infections caused by humans (Chatterjee et al., 2021).

*-Muurolene is one of a mixture of six compounds from the pigeon pea plants, which are attracted by steam distillate. It has the phenomenon of attracting female H. armigera moths present (Hartlieb and Rembold, 1996). It is known from the literature that this product is also present in the leaves of D (Yong Min et al., 2011). Our team found that there was a severe pest infestation in the area where the grass-fruit plants were planted. Whether this can provide some reference in the subsequent study needs to be studied in depth.

Cyclohexene, 1-methyl-4-(1-methylethylidene)- generated from 15 days of Sichuan sausages (Feng et al., 2017) and essential oils of three peppers(Renjie et al., 2009). Surprisingly, the PubChem database search showed it isomeric with Terpinolene (C₁₀H₁₆). However, this metabolite's specific chemical properties and physical qualities still need to be studied more, and it is detected in both D and F.

Myrcene revealed by this detected in the D, not only exists F. However, a researcher detected GC–MS for this chemical component (Yang et al., 2008). Myrcene exists as two isomers: the most natural terpenes were familiar presence occurring β-isomer, and there is less. In cyclization reactions leading to ρ-menthane derived from an isopropylidene group in the naturally isomers. For α-isome that is isopropenyl form, often easily prepared by humans. The chemical is a colorless oil with the characteristic odor of geraniums (Behr and Johnen, 2009). It can be used in perfume industry such as cologne and deodorant, which has a pleasant balsamic scent. Unparalleled importance of 1-octanol for the synthesis of 1-octene, a comonomer for polyethylene(Julis and Leitner, 2012). There is a colorless, transparent, oily liquid with a solid oily odor and citrus scent. Almost insoluble in water but miscible with alcohols, esters, chloroform, etc. Active ingredients from plant volatile compounds can serve as natural gaseous biocides. They regulate the degree of bacterial decay in agricultural products. A professional study revealed that 1-octanol exhibits remarkable antifungal potency, confirming that both the gas and liquid states of 1-octanol are associated with the growth and development of aflatoxin spores under varying exposure concentrations. This leads to irreversible damage to the plasma membrane, resulting in electrolyte leakage (Qin et al., 2022). As to previous research, odor signals could drive honeybee’s foraging preferences. (-)-β-bourbonene has reaction as far as bees are concerned. The compound showed a dramatic surge after infection (Antonio Cellini et al., 2018). An investigation with this result suggested the highly probable degradation product of germane-like precursors. A comprehensive experimental in-depth test confirmed that the extract containing bourbonene exhibited a strong anti-tick repellent effect (M. A. Birkett et al., 2008). Recently, a report was bound to investigate the anticancer potential of β-bourbonene on human prostate cancer PC-3 M cells. It was found by biological technology test that the concentration of this substance increased, and the protein expression of Bax was significantly increased in the drug treatment group, while the protein expression of Bcl-2 was decreased. Finally, it was concluded that β-bourbonene could inhibit the proliferation of prostate cancer PC-3 M cells while inducing apoptosis and G0/G1 phase arrest (Wang et al., 2018a; Wang et al., 2018b).

Thereby, 19 differential metabolites were identified according to the screening conditions, as illustrated in Fig. 5. The ordinate refers to the relative quantitative value of the metabolite after normalization. Using the normalized data, the thickest black line in the middle of each box represents the average relative content of this substance in this group. This figure is based on the quantitative value of metabolites and whether the content of metabolites has significant differences between different groups. One asterisk indicates that P is less than 0.05, with a significant difference. The two asterisks are P less than 0.01, with higher significance. If there is no asterisk, there is no significant difference. Note: It should be noted that the 12th tested component is a pesticide, and it is not a result of the plant's metabolites. It can only be used for detecting pesticide residue of exogenous substances.

Close

Fig. 5

Box plot of 19 different metabolites of D vs F. (①*-Myrcene; ② 1-Octanol; ③Naphthalene, decahydro-4a-methyl-1-methylene-7-(1-methylethenyl)-, [4aR-(4a*,7*,8a*)]-; ④ (-)-*-Bourbonene; ⑤Bicyclo[3.1.1]heptan-3-ol, 6,6-dimethyl-2-methylene-, [1S-(1*,3*,5*)]-; ⑥ Butanoic acid, methyl ester; ⑦ Tetradecane; ⑧Pentadecane; ⑨Butyrolactone; ⑩ Bicyclo[2.2.1]heptan-2-ol, 1,7,7-trimethyl-, acetate, (1S-endo)-; ⑪ 2-Nonanone; ⑫ Phenol, 3-methyl-; ⑬ Tridecane; ⑭ Octanoic acid, methyl ester; ⑮ Bicyclo[3.1.1]heptane, 6,6-dimethyl-2-methylene-, (1S)-; ⑯ (S,1Z,6Z)-8-Isopropyl-1-methyl-5-methylenecyclodeca-1,6-diene; ⑰ Benzaldehyde, 4-(1-methylethyl)-; ⑱ *-Muurolene; ⑲ Cyclohexene, 1-methyl-4-(1-methylethylidene)-.).

Export to PPT

Close

Fig. 5

Box plot of 19 different metabolites of D vs F. (①*-Myrcene; ② 1-Octanol; ③Naphthalene, decahydro-4a-methyl-1-methylene-7-(1-methylethenyl)-, [4aR-(4a*,7*,8a*)]-; ④ (-)-*-Bourbonene; ⑤Bicyclo[3.1.1]heptan-3-ol, 6,6-dimethyl-2-methylene-, [1S-(1*,3*,5*)]-; ⑥ Butanoic acid, methyl ester; ⑦ Tetradecane; ⑧Pentadecane; ⑨Butyrolactone; ⑩ Bicyclo[2.2.1]heptan-2-ol, 1,7,7-trimethyl-, acetate, (1S-endo)-; ⑪ 2-Nonanone; ⑫ Phenol, 3-methyl-; ⑬ Tridecane; ⑭ Octanoic acid, methyl ester; ⑮ Bicyclo[3.1.1]heptane, 6,6-dimethyl-2-methylene-, (1S)-; ⑯ (S,1Z,6Z)-8-Isopropyl-1-methyl-5-methylenecyclodeca-1,6-diene; ⑰ Benzaldehyde, 4-(1-methylethyl)-; ⑱ *-Muurolene; ⑲ Cyclohexene, 1-methyl-4-(1-methylethylidene)-.).

Export to PPT

3.2.1 Results: Spectroscopic analysis

Fig. 6 a showed the raw averaged NIR spectrum. NIR spectroscopy is a type of high-energy vibrational spectroscopy; Detailed chemical information is displayed in the scope of 750 to 2500 nm (13333 to 4000 cm⁻¹). At the molecular level, the replacement of light atoms (such as hydrogen) by heavy atoms can be sensed by NIR due to its impact on the strength of the remaining bonds and the combination of vibrational modes. The figure illustrates that the peak shapes and positions of the infrared spectra of the two species are identical, while the peak absorbance differs. This suggests that the chemical components of the two species are fundamentally similar, with the difference likely attributable to variations in the accumulation of chemical components. The spectral trends of the two are also comparable. In the original spectrum, the spectral absorbance of F is higher than that of D. Two spectra with similar trends can be clearly seen. It is speculated that this may be related to the selection of samples. The dried fruits from the market are most likely to be stored for a long time. But we can’t confirm the storage period, and the research on the storage period of D is also an extremely blank research field for now. FT-NIR can reflect the chemical information of hydrogen-containing organic components (N-H, C-H, O-H) in the samples. The result show that both species have multiple shared peaks at 7000–4000 cm⁻¹ and 8500–8000 cm⁻¹. The stretching vibration located between 8380 and 8230 cm⁻¹ is the C-H second overtone stretching models in –CH₃ (Liu et al., 2021a; Liu et al., 2021b). The first overtone of O-H stretching vibrations peaks could be seen near 6900–6800 cm⁻¹. While, the first overtone C-H stretching vibration belonging to –CH₂ could be revealed around 6000–5400 cm⁻¹ (Liu et al., 2021a; Liu et al., 2021b). One of the noticeable points is that around 4400 cm⁻¹, it shown that it is related to the combination band of O-H and C-O stretching modes in glucose, the Table 3 did some summary. Despite specific differences in peak shape, position, and intensity, direct classification and identification of samples from different species through the spectrum is challenging. As a result, the establishment of a classification model with the aid of chemometric methods becomes imperative.

Close

Fig. 6

The Raw of FT-NIR spectrum. D: is the grass fruit, represented by black lines; F: is the nine-foot cardamom, represented by red lines. (For interpretation of the references to color in this figure legend, the reader is referred to the web version of this article.)

Export to PPT

3.2.2 Discussion: Spectral discriminant model effect analysis

As shown in Fig. 7 A, the principal components can explain 96.1 % of the variables after fitting. To verify whether the traditional supervised algorithm PLS-DA can carry out the research exploration of the two species, the construction of a supervised model is therefore proposed. The results demonstrate a significant classification effect. To investigate the feasibility of employing the traditional supervised algorithm PLS-DA for the research exploration of the two species, we propose the construction of a supervised model.

Close

Fig. 7

The FT-IR of visualization model. A: PCA result; B: PLS-DA model; C: OPLS-DA model.

Export to PPT

As a classical linear classification method, it is well-suited for the analysis of intricate and convoluted data matrices. In order to enhance the evaluation of the model's effectiveness, 17 out of 25 samples were chosen as the training set, while the remaining 8 samples were allocated as the test set based on the Kennard-Stone algorithm (KS). And for validating the fitting extent of the final expression, 200-iteration was tested on the permutation test, the effect shown in Figure S4. It can be clearly seen that the value of R² is below 0.5, and the value of Q² is below 0. This parameter verifies that the model is robust and the fit is in place. The parameters, confusion matrix, and classification results are displayed in Table S4 in the Supplementary Material. It is had to describe that the confusion matrix included the total number of True Positive (TP), False Positive (FP), True Negative (TN), and False Negative (FN) of the train and test set of samples. The final confusion matrix resulted that all 17 test samples in the training set were correctly classified with a 100 % classification accuracy value, and the classification results in the test set were also excellent with all 8 samples correctly classified, which shows that the PLS-DA model is robust and accurate for the classification between two species. The parameter display result, the R² is 96.95 %, the Q² is 83.54 %, the result shown in Table 4.

The OPLS-DA is an extension of the PLS algorithm, which naturally simplifies the interpretability of the outcome components and allows additional assessment of intra- and inter-group differences. The unsupervised PCA only indicates the differential nature of the two categories of samples. The final distinction needs to be made with the help of the appropriate supervised methods of PLS. Based on the observation of the replacement test results for 200 iterations, the two PLS models did not show overfitting, which is likely to indicate the feasibility and robustness of the models. However, the effect of the models after increasing the sample size in this paper is also worth further testing in future studies. The OPLS-DA model has an R² of 96.59 % and a Q² of 86.58 %, with 100 % accuracy values for both the training and prediction sets, indicating the superiority of the model effect.

Finally, the classification effect of the PCA model has reached 96.1 %, the model effect of both PLS-DA and OPLS-DA has reached 100 %, and the prediction performance of the model of OPLS-DA is better than that of PLS-DA from the Q² value. In this study, the data pre-processing phase was omitted due to the high-performance parameters exhibited by the raw data.