HS–SPME–GC–MS and Electronic Nose Reveal Differences in the Volatile Profiles of Hedychium Flowers

Floral fragrance is one of the most important characteristics of ornamental plants and plays a pivotal role in plant lifespan such as pollinator attraction, pest repelling, and protection against abiotic and biotic stresses. However, the precise determination of floral fragrance is limited. In the present study, the floral volatile compounds of six Hedychium accessions exhibiting from faint to highly fragrant were comparatively analyzed via gas chromatography–mass spectrometry (GC–MS) and Electronic nose (E-nose). A total of 42 volatile compounds were identified through GC–MS analysis, including monoterpenoids (18 compounds), sesquiterpenoids (12), benzenoids/phenylpropanoids (8), fatty acid derivatives (2), and others (2). In Hedychium coronarium ‘ZS’, H. forrestii ‘Gaoling’, H. ‘Jin’, H. ‘Caixia’, and H. ‘Zhaoxia’, monoterpenoids were abundant, while sesquiterpenoids were found in large quantities in H. coccineum ‘KMH’. Hierarchical clustering analysis (HCA) divided the 42 volatile compounds into four different groups (I, II, III, IV), and Spearman correlation analysis showed these compounds to have different degrees of correlation. The E-nose was able to group the different accessions in the principal component analysis (PCA) corresponding to scent intensity. Furthermore, the pattern-recognition findings confirmed that the E-nose data validated the GC–MS results. The partial least squares (PLS) analysis between floral volatile compounds and sensors suggested that specific sensors were highly sensitive to terpenoids. In short, the E-nose is proficient in discriminating Hedychium accessions of different volatile profiles in both quantitative and qualitative aspects, offering an accurate and rapid reference technique for future applications.


Introduction
Hedychium belongs to the family Zingiberaceae and is an economically important crop grown extensively for its ornamental and medicinal properties. The flowers of Hedychium are widely used in perfumed and cosmetic industries, while rhizome is used for medicinal purposes [1][2][3]. There are more than 80 Hedychium species that have been reported; however, new species are still emerging [4][5][6]. Hedychium species are diverse in color with numerous shapes. With respect to their floral scent, Hedychium species vary from scentless to rich in fragrance with high ornamental values [6][7][8]. The flowers of Hedychium are rich in an aroma that is mainly composed of a mixture of monoterpenoids, sesquiterpenoids, and benzenoids [9][10][11][12]. Recently, the main focus has been on the physio-biochemical properties of Hedychium; however, there are few reports on the qualitative and quantitative analysis of the floral volatile profiles of Hedychium.

Hierarchical Clustering Analysis (HCA) Based on GC-MS Data
To show the differences in the VOCs among Hedychium accessions, HCA analysis was performed (Figure 2a). The red color indicates a higher than average content value, while the blue color in the plot indicates a lower content value. The data showed that the contents of the VOCs (42 compounds) from six Hedychium accessions could be divided into four clusters (I, II, III, and IV) (Figure 2a). Cluster I include 24 compounds with high contents in the medium scented 'Gaoling'. Cluster II includes four compounds with high content in the faintly fragrant 'KMH' ((E)-β-farnesene) and the weakly fragrant 'Caixia' (limonene, dihydro-β-ionone, δ-cadinene). Cluster III includes eight compounds (linalool, α-amorphene, methyl benzoate, phenol, 2-methoxy-4-(1-propenyl), 1-butanol, 3-methyl, benzoate, benzyl benzoate, methyl jasmonate, and indole) with high content in the strongly fragrant 'ZS'. Likewise, cluster IV includes six compounds (β-pinene, caryophyllene, anisole, phenylethyl alcohol, benzyl nitrile, and eugenol) with high content in the weak scented 'Zhaoxia' and 'Jin'. Furthermore, Spearman correlation analysis showed that 42 VOCs had different degrees of correlation ( Figure 2b).

Principal Component Analysis Based on GC-MS Data
To analyze the floral volatile compounds that play a crucial role in differentiating volatile profile, the 42 VOCs identified via HS-SPME-GC-MS were subjected to principal component analysis. The results showed that 65.28% of the total variability was accounted for by the first two principal components, and six accessions could be distinguished (Figure 2c). Moreover, terpenoids (monoterpenoids and sesquiterpenoids) contributed mainly

Principal Component Analysis Based on GC-MS Data
To analyze the floral volatile compounds that play a crucial role in differentiating volatile profile, the 42 VOCs identified via HS-SPME-GC-MS were subjected to principal component analysis. The results showed that 65.28% of the total variability was accounted for by the first two principal components, and six accessions could be distinguished (Figure 2c). Moreover, terpenoids (monoterpenoids and sesquiterpenoids) contributed mainly to PC1, while benzenoids/phenylpropanoids contributed to PC2 (Figure 2d).

Discrimination of the Different Taxa Using the E-Nose
In the current study, the floral volatile profiles of six Hedychium accessions were measured via E-nose, and the mean sensor values are listed in Table 2. The results showed that sensors W1W, W2W, and W5S were highly responsive when compared to other sensors. Moreover, the response values of the aforementioned sensors were highest for 'Gaoling', followed by 'Caixia', 'ZS', 'Zhaoxia', 'Jin', and 'KMH' ( Table 2). HCA analysis divided the ten sensors into three groups. Group I includes three sensors (W5C, W1C, and W3C) with high response values to 'Jin', 'Zhaoxia', and 'KMH', Group II includes two sensors (W6S and W3S) with higher responsive values to 'ZS', while Group III includes five sensors (W1S, W2S, W5S, W1W, and W2W) with high response value to 'Gaoling' (Figure 3a). Correlation analysis showed that sensor W1C was positively correlated with W3C and W5C, while W1C and W3C were negatively correlated with W1S, W2S, W1W, W2W, and W5S ( Figure 3b; Table 2). In addition, there was a significant positive correlation among W1S, W2S, W5S, W1W, and W2W (Figure 3b). Principal component analysis showed that the first two principal components explained 88.20% of the variance, and six accessions with different floral volatile profiles could be distinguished (Figure 3c). Sensors W1C, W3C, W1S, W2S, W5S, W1W, and W2W contributed greatly to PC1, while sensors W3S and W6S mainly explained the difference in PC2 (Figure 3d). Meanwhile, W5C contributed less toward both PC1 and PC2.

Correlation between GC-MS and E-Nose Sensors
To analyze the correlation between GC-MS and E-nose measurements, a PLS regression analysis was performed (Figure 4). The results revealed that the sensors showed different responses to different volatiles. Thirty-one volatile compounds and nine sensors are located in the right section of the plot, and they explained between 50 and 100% of the cross-validated variance, suggesting a significant positive correlation between these variables (p < 0.05) (Figure 4a). The heatmap of the PLS correlation coefficients showed that sensors W1S, W1W, W5S, W2W, W6S, and W3S were positively correlated with the VOCs, while sensors W3C and W1C had negative correlation coefficients (Figure 4b). Sensor W2W was positively correlated with 6 monoterpenoids and 2 sesquiterpenoids. Sensor W1W was positively correlated with 5 monoterpenoids and 2 sesquiterpenoids. The sensors W1S, W2S, and W5S were positively correlated with 3, 8, and 12 compounds, respectively ( Figure 4c). Meanwhile, sensors W3C and W1C were negatively correlated with 5 and

Correlation between GC-MS and E-Nose Sensors
To analyze the correlation between GC-MS and E-nose measurements, a PLS regression analysis was performed (Figure 4). The results revealed that the sensors showed different responses to different volatiles. Thirty-one volatile compounds and nine sensors are located in the right section of the plot, and they explained between 50 and 100% of the cross-validated variance, suggesting a significant positive correlation between these variables (p < 0.05) (Figure 4a). The heatmap of the PLS correlation coefficients showed that sensors W1S, W1W, W5S, W2W, W6S, and W3S were positively correlated with the VOCs, while sensors W3C and W1C had negative correlation coefficients (Figure 4b). Sensor W2W was positively correlated with 6 monoterpenoids and 2 sesquiterpenoids. Sensor W1W was positively correlated with 5 monoterpenoids and 2 sesquiterpenoids. The sensors W1S, W2S, and W5S were positively correlated with 3, 8, and 12 compounds, respectively (Figure 4c). Meanwhile, sensors W3C and W1C were negatively correlated with 5 and 13 compounds, respectively. Interestingly, sensor W6S was significant positively correlated with methyl benzoate.

Discussion
Hedychium is a perennial herb grown extensively as a garden plant as well as a cut flower in both tropical and subtropical regions. The flower blossoming results in the emission of abundant volatile compounds composed mainly of terpenes and benzenoids/phenylpropanoid substances [7,8,10,12,26]. Several accessions with high ornamental values are present in the market that range from scentless (H. coccineum) to highly fragrant (H. coronarium) [7,43]. The precise determination of volatile compounds and biosynthesis of floral volatile components will assist researchers in the Hedychium breeding program. In this study, Electronic nose technology was used for the first time to evaluate the volatile components of Hedychium, along with HS-SPME-GC-MS. Differences in the volatile profiles of six Hedychium accessions were evaluated, as volatile components are affected by their genetic contexts.
To measure the floral volatile contents and to analyze the differences among the floral volatile profiles of Hedychium accessions, the HS-SPME-GC-MS approach was used. The GC-MS data showed that the types and contents of VOCs in Hedychium accessions varied significantly (Table 1). Similarly, significant differences in volatile compounds were observed in Lilium and Anthurium [44,45]. A total of 42 volatile compounds were identified, and terpenoids were the main volatile contents in six Hedychium accessions, which is in agreement with previous studies [7,8,10]. Furthermore, linalool and (E)-β-ocimene constituted 62.34% of the total volatiles in 'ZS', which is in line with the previous findings that

Discussion
Hedychium is a perennial herb grown extensively as a garden plant as well as a cut flower in both tropical and subtropical regions. The flower blossoming results in the emission of abundant volatile compounds composed mainly of terpenes and benzenoids/phenylpropanoid substances [7,8,10,12,26]. Several accessions with high ornamental values are present in the market that range from scentless (H. coccineum) to highly fragrant (H. coronarium) [7,43]. The precise determination of volatile compounds and biosynthesis of floral volatile components will assist researchers in the Hedychium breeding program. In this study, Electronic nose technology was used for the first time to evaluate the volatile components of Hedychium, along with HS-SPME-GC-MS. Differences in the volatile profiles of six Hedychium accessions were evaluated, as volatile components are affected by their genetic contexts.
To measure the floral volatile contents and to analyze the differences among the floral volatile profiles of Hedychium accessions, the HS-SPME-GC-MS approach was used. The GC-MS data showed that the types and contents of VOCs in Hedychium accessions varied significantly (Table 1). Similarly, significant differences in volatile compounds were observed in Lilium and Anthurium [44,45]. A total of 42 volatile compounds were identified, and terpenoids were the main volatile contents in six Hedychium accessions, which is in agreement with previous studies [7,8,10]. Furthermore, linalool and (E)-β-ocimene constituted 62.34% of the total volatiles in 'ZS', which is in line with the previous findings that linalool and (E)-β-ocimene are the main floral volatiles of this species [43,46]. The HCA analysis data showed that 42 volatile compounds from Hedychium accessions could be categorized into four clusters, while Spearman analysis revealed that these VOCs had different degrees of correlation ( Figure 2). Furthermore, PCA data showed that floral volatile compounds were distributed among all accessions and that six accessions could be distinguished. These data demonstrated that monoterpenoids and sesquiterpenoids had a significant influence on the floral characteristics of Hedychium accessions. Previous studies also showed that the expression of genes related to the biosynthesis of terpenoids in Anthurium [45], Lilium 'Siberia' [29][30][31], and Freesia [42] had a certain correlation with the volatilization of terpenoids. In the future, the relationship between genes related to the biosynthesis pathway of terpenoids and their volatile compounds will be worth exploring in Hedychium.
The analysis of floral volatiles via E-nose requires little time, and the system is easy to operate, while sensory analysis via expert panel is costly [47,48]. Hence, the E-nose system was selected as a fast and qualitative technique to study the differences in the floral volatile profile of Hedychium. Previous studies have shown that the E-nose system can accurately detect the VOCs of various molecular structures regardless of the odor intensity [49]. In this study, there was no linear relationship between the data measured by E-nose and the intensity of the aroma, but there was consistency between the data measured by E-nose and the total amount of volatile compounds determined by GC-MS. This might be due to the influence of the properties of different compounds and the olfactory threshold. Many studies have shown that various volatile compounds have different degrees of correlation with different E-nose sensors [50,51]. Herein, the differences among the volatile profiles of Hedychium accessions were observed. Through PLS regression analysis, we observed that W1W, W2W, and W5S sensors with high response values had different degrees of correlation with monoterpenoids and sesquiterpenoids, suggesting that E-nose is an effective method to evaluate floral volatile profiles of Hedychium (Figure 4). Furthermore, sensor W6S was responsive to benzenoids/phenylpropanoids. In Actinidia, E-nose sensors S7, S10, S8, S6, S9, and S2 were highly responsive to terpene compounds [52]. Similar findings were observed in Prunus persica, where the E-nose accurately distinguished peach samples of two different flesh colors [53]. In Malus taxa, the nitrogen-containing compounds, terpenes, and sulfur-containing compounds were highly correlated with sensors W5C, W1W, and W2S [54]. These findings suggested that E-nose is an effective and rapid technique for studying the floral volatile profile of Hedychium flowers.
The evaluation of flower aroma is very complex, and the detection of floral fragrance is one of the key limitations of flower fragrance breeding programs. Previously, researchers have relied mainly on the GC-MS system to identify the floral components in ornamental plants; however, there have been few reports on the use of E-nose [55] and PTR-MS [56] to identify the floral volatile profiles of ornamental plants. Each method has its advantages and disadvantages; however, the combination of GC-MS with E-nose or PTR-MS can efficiently discriminate the aroma components [57]. Recently, a combination of E-nose and GC-MS, PTR-MS, and GC-MS joint analysis was used in Cucumis melo L. [58], Actinidia [52], Prunus [53], Vaccinium spp. [59], and several other plants to identify the VOCs. However, there are few reports on the combined analysis of multi-methods in flower fragrance. In this study, the floral characteristics of different Hedychium accessions were analyzed efficiently by combined analysis using GC-MS and E-nose. Herein, we provided a reference for the establishment of the rapid detection of Hedychium floral volatile profiles. In the future, the combination of different detection methods can be applied to precisely identify and measure the floral VOCs of ornamental plants.

Plant Materials
The scent intensity of floral volatile compounds from Hedychium accessions was assessed by 25 trained assessors, as described previously [60]. Based

Sample Preparation and HS-SPME-GC-MS Analysis
The samples for volatile determination were collected in the morning from September to October. The individual fresh flower (2 g) was placed in a 250 mL glass bottle supplemented with an internal standard. The flower was enclosed in a 250 milliliter (mL) glass bottle with the addition of 1.728 micrograms (µg) ethyl caprate, which served as an internal standard. After this, the bottle was immediately closed with aluminum foil. After 30 min, polydimethylsiloxane (PDMS, with 50/30 µm divinylbenzene/Carboxen, Supelco) fiber was injected into the bottle for 30 min to trap the volatiles, which was followed by insertion into a GC-MS (Agilent) or E-nose system for volatile analysis, as described previously [11,43]. The HS-SPME extraction was performed at 26 ± 2 • C. The GC-MS system with Agilent 7890A GC and Agilent 5975C MSD was provided with an Agilent DB-5MS capillary column (30 m × 0.25 mm), and helium gas was provided as a carrier. The flow of helium gas was kept constant at 1 mL/min. Initially, the GC injection port temperatures was maintained at 40 • C for 3 min, which was followed by an increase in temperature of 5 • C/min to 250 • C. The column effluent was ionized by electron ionization (EI) at an energy of 70 eV with a transfer temperature of 280 • C and a source temperature of 170 • C. The floral volatiles analysis was performed at the full-bloom stage. The chromatographic running time was 30 min. Three replicates were tested for each variety. The relative quantification of volatiles was calculated using Agilent ChemStation Data Analysis Application based on peak area ratio and the quantity of internal standard. Relative contents (%) = (area under peak/total peak area) × 100.

E-Nose Analysis
A PEN3 portable E-nose (Airsense Company, Schwerin, Germany) was used in the assay. The structure includes a sampling apparatus, a sensor array unit, and patternrecognition software, as explained previously [54]. The sensor array consists of ten metal oxide semiconductor (MOS) sensors (W1C, W5S, W3C, W6S, W5C, W1S, W1W, W2S, W2W, W3S). The characteristics of the MOS sensors are provided in Supplementary Table S3. The gas in the headspace was thrust over the surface of the sensor for 5 min at a constant flow rate of 150 mL/min. The interval time was 10 s, while the values of the stabilized response sensors were 35-40 s. Three replicates were tested for each variety. For data processing, the stable value of each sensor was extracted.

Identification of VOCs
The floral volatile compounds were identified by comparing them with mass spectra from the NIST Mass Spectral Library (NIST 08), with existing works of literature, and with authentic standards. Mass spectra were obtained by automatic scanning at m/z 20 to 500 amu. The identification of compounds was perceived via comparing the mass spectra with NIST 08 at a match factor of ≥80 [42,51]. The data were processed using mass hunter qualitative analysis workflow software (Agilent Technologies Inc., Santa Clara, CA, USA). Linear retention indices (LRI) of the volatile compounds were measured via an alkane series standard (C7-C40) (Sigma, St. Louis, MO, USA) separated on the DB-5 MS capillary column under the same GC-MS analysis conditions.

Statistical Analysis
Statistical Package for the Social Sciences 19.0 (SPSS Inc., Chicago, IL, USA) was used for the statistical analysis. The differences among samples were calculated via analysis of variance (ANOVA). Data are presented as the mean ± SD (n = 3-5). R 4.0.5 internal statistical functions and the external packages "corrplot" and "mixOmics" were used for the multivariate statistical methods employed (Spearman correlation analysis, hierarchical clustering analysis (HCA), principal component analysis (PCA), and partial least squares (PLS)). Visualization of significant volatile organic compound correlations was performed via a PLS regression network using Cytoscape [61].

Conclusions
In the current study, we evaluated the floral volatile profile of six Hedychium accessions using HS-SPME-GC-MS and E-nose technology. The HS-SPME-GC-MS analysis revealed that (E)-β-ocimene, linalool, and methyl benzoate were abundant in highly scented H. coronarium 'ZS', while eucalyptol was the foremost compound in the medium fragrant H. forrestii 'Gaoling'. Furthermore, analysis of volatile compounds via E-nose technology showed that sensors W1W, W2W, W5S, and W6S played an important role in the recognition of volatile profiles of six Hedychium accessions. The correlation analysis between volatile compounds and sensors showed that specific sensors were more sensitive to terpenoids and benzenoids/phenylpropanoids. Our SPME-GC-MS and E-nose will facilitate the researcher in the Hedychium breeding program to develop new cultivars of high ornamental traits.
Supplementary Materials: The following are available online. Figure S1: Total ion chromatogram of the volatile compounds released from the flowers of six Hedychium accessions, Figure S2: A pictorial representation of the number of volatile compounds overlapping in different Hedychium accessions, Table S1: The list of floral volatile components in Hedychium accessions and their relative contents %, Table S2: The information of six Hedychium accessions, Table S3: Sensors used in this study and their main application in PEN3.