Extraction and Identification of Volatile Organic Compounds Emitted by Fragrant Flowers of Three Tillandsia Species by HS-SPME/GC-MS

Numerous volatile organic compounds (VOCs) with a large chemical diversity are emitted by plant flowers. They play an important role in the ecology of plants, such as pollination, defense, adaptation to their environment, and communication with other organisms. The Tillandsia genus belongs to the Bromeliaceae family, and most of them are epiphytes. The aromatic profile of the Tillandsia genus is scarcely described. In this study, we use the headspace solid phase microextraction (HS-SPME) coupled with gas chromatography combined with mass spectrometry (GC-MS) method developed in our laboratory to explore the chemical diversity of the VOCs of fragrant flowers of three species of the genus Tillandsia. We were able to identify, for the first time, 66 volatile compounds (monoterpenes, sesquiterpenes, phenylpropanoids, and other compounds). We identified 30 compounds in T. xiphioides, 47 compounds in T. crocata, and 43 compounds in T. caliginosa. Only seven compounds are present in all the species studied. Comparison of the volatile compounds profiles by principal component analysis (PCA) between T. xiphoides, T. crocata, and T. caliginosa species showed a clear difference in the floral emissions of the studied species. Moreover, floral VOCs profiles allowed to differentiate two forms of T. xiphioides and of T. crocata.


Introduction
A wide variety of Volatile Organic Compounds (VOCs) are produced and emitted by plants, especially in organs such as flowers, leaves, fruits, and roots. More than  A principal component analysis (PCA) was performed to profile the variations of compounds of the different species T. xiphioides, T. crocata, and T. caliginosa. Having shown relatively similar profiles, a second PCA as well as a heatmap were performed to better visualize the differences in profiles of T. crocata and T. caliginosa. This study is the first to compare floral volatile profiles between fragrant Tillandsia species and provides additional information on the poorly described floral metabolic network of the genus Tillandsia.

Identification
CAR/PDMS fiber with optimal extraction conditions was used to study the volatile compounds emitted from the flowers of three Tillandsia species. Through the use of two HS-SPME/GC-MS methods, a total of 66 compounds were identified from the floral emissions of LP-xiphi, HP-xiphi, yellow crocata, orange crocata, and T. caliginosa (Table 2, Figure  1). Among these compounds, only seven-β-myrcene (6), limonene (8), eucalyptol (9), αterpineol (15), geranyl acetate (18), β-farnesene (23), and α-farnesene (24)-are present in all the species studied. The compounds were categorized into four different classes: monoterpenes, sesquiterpenes, phenylpropanoids, and other compounds including nitrogenous compounds, esters, and fatty acid derivatives ('others' family). Qualitative differences in the profile of flower volatiles of these species were observed.  A principal component analysis (PCA) was performed to profile the variations of compounds of the different species T. xiphioides, T. crocata, and T. caliginosa. Having shown relatively similar profiles, a second PCA as well as a heatmap were performed to better visualize the differences in profiles of T. crocata and T. caliginosa. This study is the first to compare floral volatile profiles between fragrant Tillandsia species and provides additional information on the poorly described floral metabolic network of the genus Tillandsia.

Identification
CAR/PDMS fiber with optimal extraction conditions was used to study the volatile compounds emitted from the flowers of three Tillandsia species. Through the use of two HS-SPME/GC-MS methods, a total of 66 compounds were identified from the floral emissions of LP-xiphi, HP-xiphi, yellow crocata, orange crocata, and T. caliginosa (Table 2, Figure  1). Among these compounds, only seven-β-myrcene (6), limonene (8), eucalyptol (9), αterpineol (15), geranyl acetate (18), β-farnesene (23), and α-farnesene (24)-are present in all the species studied. The compounds were categorized into four different classes: monoterpenes, sesquiterpenes, phenylpropanoids, and other compounds including nitrogenous compounds, esters, and fatty acid derivatives ('others' family). Qualitative differences in the profile of flower volatiles of these species were observed.  A principal component analysis (PCA) was performed to profile the variations of compounds of the different species T. xiphioides, T. crocata, and T. caliginosa. Having shown relatively similar profiles, a second PCA as well as a heatmap were performed to better visualize the differences in profiles of T. crocata and T. caliginosa. This study is the first to compare floral volatile profiles between fragrant Tillandsia species and provides additional information on the poorly described floral metabolic network of the genus Tillandsia.

Identification
CAR/PDMS fiber with optimal extraction conditions was used to study the volatile compounds emitted from the flowers of three Tillandsia species. Through the use of two HS-SPME/GC-MS methods, a total of 66 compounds were identified from the floral emissions of LP-xiphi, HP-xiphi, yellow crocata, orange crocata, and T. caliginosa (Table 2, Figure  1). Among these compounds, only seven-β-myrcene (6), limonene (8), eucalyptol (9), αterpineol (15), geranyl acetate (18), β-farnesene (23), and α-farnesene (24)-are present in all the species studied. The compounds were categorized into four different classes: monoterpenes, sesquiterpenes, phenylpropanoids, and other compounds including nitrogenous compounds, esters, and fatty acid derivatives ('others' family). Qualitative differences in the profile of flower volatiles of these species were observed.  A principal component analysis (PCA) was performed to profile the variations of compounds of the different species T. xiphioides, T. crocata, and T. caliginosa. Having shown relatively similar profiles, a second PCA as well as a heatmap were performed to better visualize the differences in profiles of T. crocata and T. caliginosa. This study is the first to compare floral volatile profiles between fragrant Tillandsia species and provides additional information on the poorly described floral metabolic network of the genus Tillandsia.

Identification
CAR/PDMS fiber with optimal extraction conditions was used to study the volatile compounds emitted from the flowers of three Tillandsia species. Through the use of two HS-SPME/GC-MS methods, a total of 66 compounds were identified from the floral emissions of LP-xiphi, HP-xiphi, yellow crocata, orange crocata, and T. caliginosa (Table 2, Figure  1). Among these compounds, only seven-β-myrcene (6), limonene (8), eucalyptol (9), αterpineol (15), geranyl acetate (18), β-farnesene (23), and α-farnesene (24)-are present in all the species studied. The compounds were categorized into four different classes: monoterpenes, sesquiterpenes, phenylpropanoids, and other compounds including nitrogenous compounds, esters, and fatty acid derivatives ('others' family). Qualitative differences in the profile of flower volatiles of these species were observed.  A principal component analysis (PCA) was performed to profile the variations of compounds of the different species T. xiphioides, T. crocata, and T. caliginosa. Having shown relatively similar profiles, a second PCA as well as a heatmap were performed to better visualize the differences in profiles of T. crocata and T. caliginosa. This study is the first to compare floral volatile profiles between fragrant Tillandsia species and provides additional information on the poorly described floral metabolic network of the genus Tillandsia.

Identification
CAR/PDMS fiber with optimal extraction conditions was used to study the volatile compounds emitted from the flowers of three Tillandsia species. Through the use of two HS-SPME/GC-MS methods, a total of 66 compounds were identified from the floral emissions of LP-xiphi, HP-xiphi, yellow crocata, orange crocata, and T. caliginosa (Table 2, Figure 1). Among these compounds, only seven-β-myrcene (6), limonene (8), eucalyptol (9), α-terpineol (15), geranyl acetate (18), β-farnesene (23), and α-farnesene (24)-are present in all the species studied. The compounds were categorized into four different classes: monoterpenes, sesquiterpenes, phenylpropanoids, and other compounds including nitrogenous compounds, esters, and fatty acid derivatives ('others' family). Qualitative differences in the profile of flower volatiles of these species were observed.   Figure S1 in the Supplementary Material. The identification numbers correspond to those shown in Table 2.
For T. caliginosa, 43 volatile compounds were identified for the first time in floral emissions. The volatile profile is quite similar to that of the species T. crocata; however, there are fewer sesquiterpenes, which are only five in T. caliginosa, and the compounds belonging to the 'others' family are more numerous in T. caliginosa than in T. crocata and in T. xiphioides. The first extraction method allowed us to isolate mostly limonene (8), eucalyptol (9), and phenylethyl acetate (39), while the second method allowed us to extract mostly hexadecyl ethanoate (66), 1-hexadecanol (61), phenylethyl acetate (39), and methyl nicotinate (54) from the flowers of T. caliginosa.

Chemometrics Analysis of Volatile Organic Compounds
In order to identify volatile compounds that contribute to differences or similarities between the species studied, data on the 66 identified volatiles were analyzed using principal component analyses (PCA). PCA is a multivariate analysis that allows the extraction of important information from a data table in which observations are described by several intercorrelated dependent variables and represents them as a set of new orthogonal variables called principal components. In Figure 2A, it is shown that the first principal component (PC1) contributes to 32.7%, and the second principal component (PC2) contributes to 17.29% of the total variance in the floral volatile peak area data. On this same figure, we can see three distinct groups; indeed, the flowers of the two forms of T. xiphioides are clustered together, the flowers of the two forms of T. crocata are clustered together, and the flowers of T. caliginosa form the third group. This shows a high degree of clustering of flowers belonging to the same species, thus suggesting a unique floral volatile profile for each species of the genus Tillandsia. PC1 allowed discrimination of LP-xiphi and HP-xiphi flowers that are clearly separated from yellow crocata, orange crocata, and caliginosa flowers. This separation is due to higher intensities of monoterpenoids and sesquiterpenoids in the floral emissions of both forms of T. xiphioides ( Figure 2B). PC2 showed a difference in floral volatiles of T. caliginosa compared to the two forms of T. crocata. This was due to  Figure S1 in the Supplementary Material. The identification numbers correspond to those shown in Table 2. For T. xiphioides, a total of 30 volatile compounds were identified in the floral emissions of both forms of this species. The volatile profile is largely dominated by terpenoids, which are 28 of the 30 compounds identified. Indeed, the first extraction method allowed the isolation of more monoterpenes, of which the major ones are limonene (8) and eucalyptol (9) for LP-xiphi and β-linalool (13) for HP-xiphi. With the use of higher values of temperature and extraction time, the second extraction method allowed the isolation of higher molecular weight compounds. The major ones are geranyl acetate (18) and nerolidol (27) for LP-xiphi and indole (55), nerolidol (27), and denderalisin (28) for HP-xiphi. Among the compounds identified in both forms of T. xiphioides, only indole (55) and methyl palmitate (63) are present in the HP-xiphi form and not in the LP-xiphi form.
For T. caliginosa, 43 volatile compounds were identified for the first time in floral emissions. The volatile profile is quite similar to that of the species T. crocata; however, there are fewer sesquiterpenes, which are only five in T. caliginosa, and the compounds belonging to the 'others' family are more numerous in T. caliginosa than in T. crocata and in T. xiphioides. The first extraction method allowed us to isolate mostly limonene (8), eucalyptol (9), and phenylethyl acetate (39), while the second method allowed us to extract mostly hexadecyl ethanoate (66), 1-hexadecanol (61), phenylethyl acetate (39), and methyl nicotinate (54) from the flowers of T. caliginosa.

Chemometrics Analysis of Volatile Organic Compounds
In order to identify volatile compounds that contribute to differences or similarities between the species studied, data on the 66 identified volatiles were analyzed using principal component analyses (PCA). PCA is a multivariate analysis that allows the extraction of important information from a data table in which observations are described by several intercorrelated dependent variables and represents them as a set of new orthogonal variables called principal components. In Figure 2A, it is shown that the first principal component (PC1) contributes to 32.7%, and the second principal component (PC2) contributes to 17.29% of the total variance in the floral volatile peak area data. On this same figure, we can see three distinct groups; indeed, the flowers of the two forms of T. xiphioides are clustered together, the flowers of the two forms of T. crocata are clustered together, and the flowers of T. caliginosa form the third group. This shows a high degree of clustering of flowers belonging to the same species, thus suggesting a unique floral volatile profile for each species of the genus Tillandsia. PC1 allowed discrimination of LP-xiphi and HP-xiphi flowers that are clearly separated from yellow crocata, orange crocata, and caliginosa flowers. This separation is due to higher intensities of monoterpenoids and sesquiterpenoids in the floral emissions of both forms of T. xiphioides ( Figure 2B). PC2 showed a difference in floral volatiles of T. caliginosa compared to the two forms of T. crocata. This was due to high phenylpropanoids intensities in both forms of T. crocata while T. caliginosa showed higher intensities of other compounds.  To better visualize the differences in volatile profiles between flowers of the two forms of T. crocata and those of T. caliginosa, a second PCA as well as a heatmap were performed (Figures 3 and 4). The first two components of the PCA explained 43.14% and 23.74% of the variation, explaining approximately 66% of the combined variance ( Figure 3A). We can see that the two forms of T. crocata have relatively different floral volatile profiles due to their higher intensities of sesquiterpenoids and phenylpropanoids, such as β-farnesene (23), farnesyl acetate (32), benzylacetate (37), and phenylethyl acetate (39), which have higher inten- To better visualize the differences in volatile profiles between flowers of the two forms of T. crocata and those of T. caliginosa, a second PCA as well as a heatmap were performed (Figures 3 and 4). The first two components of the PCA explained 43.14% and 23.74% of the variation, explaining approximately 66% of the combined variance ( Figure 3A). We can see that the two forms of T. crocata have relatively different floral volatile profiles due to their higher intensities of sesquiterpenoids and phenylpropanoids, such as β-farnesene (23), farnesyl acetate (32), benzylacetate (37), and phenylethyl acetate (39), which have higher intensities in floral emanations of yellow crocata compared to orange crocata ( Figure 3B). While compounds such as benzyl benzoate (48), eugenol (42), methyl eugenol (43), benzyl salicylate (50), and phenethyl benzoate (49) are more abundant in orange crocata flowers, T. caliginosa flowers are grouped on the right due to higher intensities of other compounds, such as 1-hexadecanol (61), hexadecyl ethanoate (66), methyl tetradecanoate (56), hexadecanal (59), and some phenylpropanoids, such as methyl benzoate (35) and methyl salicylate (38). To better visualize the differences in volatile profiles between flowers of the two forms of T. crocata and those of T. caliginosa, a second PCA as well as a heatmap were performed (Figures 3 and 4). The first two components of the PCA explained 43.14% and 23.74% of the variation, explaining approximately 66% of the combined variance ( Figure 3A). We can see that the two forms of T. crocata have relatively different floral volatile profiles due to their higher intensities of sesquiterpenoids and phenylpropanoids, such as β-farnesene (23), farnesyl acetate (32), benzylacetate (37), and phenylethyl acetate (39), which have higher intensities in floral emanations of yellow crocata compared to orange crocata ( Figure 3B). While compounds such as benzyl benzoate (48), eugenol (42), methyl eugenol (43), benzyl salicylate (50), and phenethyl benzoate (49) are more abundant in orange crocata flowers, T. caliginosa flowers are grouped on the right due to higher intensities of other compounds, such as 1-hexadecanol (61), hexadecyl ethanoate (66), methyl tetradecanoate (56), hexadecanal (59), and some phenylpropanoids, such as methyl benzoate (35) and methyl salicylate (38). This is confirmed by the heatmap (Figure 4); indeed, it is shown that the 57 volatile compounds detected in the floral emissions of T. caliginosa and in the two forms of T. crocata, are grouped in four large groups in the dendogram of the heatmap. The 14 compounds of group 1, consisting of methyl salicylate, methyl benzoate, methyl tetradecanoate, methyl palmitoleate, β-ocimene, hexyl acetate, methyl hexanoate, 11-hexadecenal, γ-terpinene, hexadecanal, β-fenchol, 11-hexadecenyl acetate, 1-hexadecanol, and hexadecyl ethanoate, are more abundant in the flowers of T. caliginosa. The 10 volatile molecules of Group 2, consisting of β-myrcene, limonene, α-pinene, eucalyptol, sabinene, β-phellandrene, β-pinene, zingerone, α-terpineol, and eugenol, are relatively abundant in the floral emissions of T. caliginosa and highly present in orange crocata flowers compared to yellow crocata. Group 3, consisting of 17 compounds, is formed mostly by sesquiterpenes and phenylpropanoids, which are abundant in yellow crocata flowers, while Group 4 is formed by 16 volatile compounds, which are abundant in orange crocata floral emissions.
ethanoate, are more abundant in the flowers of T. caliginosa. The 10 volatile molecules of Group 2, consisting of β-myrcene, limonene, α-pinene, eucalyptol, sabinene, β-phellandrene, β-pinene, zingerone, α-terpineol, and eugenol, are relatively abundant in the floral emissions of T. caliginosa and highly present in orange crocata flowers compared to yellow crocata. Group 3, consisting of 17 compounds, is formed mostly by sesquiterpenes and phenylpropanoids, which are abundant in yellow crocata flowers, while Group 4 is formed by 16 volatile compounds, which are abundant in orange crocata floral emissions.

Discussion
The identification of VOCs of three species of the genus Tillandsia was achieved through the use of the HS-SPME/GC-MS method. By using the CAR/PDMS fiber and two extraction methods with different values of extraction time and temperature developed in a previous study, the majority of the volatile compounds emitted by the flowers of Tillandsia species were extracted efficiently. Indeed, it was shown that the CAR/PDMS fiber allowed to extract the maximum of volatile compounds from T. xiphioides flowers compared to PDMS/DVB and DVB/CAR/PDMS fibers; moreover, considering the volatility of analytes affecting the optimal conditions of temperature and duration of extraction, it is necessary to use two extraction methods in order to trap the majority of volatiles emitted by Tillandsia flowers [11]. After the GC-MS analysis, the profiles of the volatiles emitted by the flowers of the different species studied were compared qualitatively. This is the first time that a comparison of the VOCs emitted by different fragrant flowers of the genus Tillandsia has been performed. The results showed a difference in volatile profiles between the flowers of the species T. xiphioides, T. crocata, and T. caliginosa. Indeed, the flowers of the two forms of T. xiphioides present a profile of volatile compounds much more enriched in monoterpenes and sesquiterpenes, compared to the flowers of T. crocata and T. caliginosa, which present relatively more similar profiles. This could be confirmed by a morphological difference but also by a difference in pollinator type between T. xiphioides species and those of T. crocata and T. caliginosa. Indeed, it is shown that T. caliginosa and T. crocata are morphologically closer-these species present the same types of foliage and flowers; moreover the old name given to T. caliginosa is T. crocata var. tristis [17]. This difference in the composition of floral volatiles could also be confirmed by the attraction of different pollinators [18] for T. crocata, whose floral odor attracts Euglossine bees [15] and T. xiphioides, which is considered sphingophilous [19]. Indeed, it is shown that plants pollinated by the same ecological guild show similar visual and olfactory signals [20].
Even though it is not the only characteristic that can play a role in the attraction of pollinators, the floral odor constituted by a complex mixture of volatile compounds belonging to different chemical families plays a very important role in the stimulation and attraction of pollinators. These floral volatile compounds can be classified in the family of terpenoids, phenylpropanoids/benzenoids, and fatty acid derivatives, according to their biosynthetic pathways. Found predominantly in the flowers of both forms of T. xiphioides, terpenoids are derived from two precursors with five carbons isopentenyl diphosphate (IPP) and its allylic isomer dimethylallyl diphosphate (DMAPP) and represent the largest class of floral volatile compounds. The latter are synthesized in plants from two independent pathways: the mevalonic acid pathway, where synthesis is carried out in the cytosol, giving rise to precursors of volatile sesquiterpenes (C 15 ); and the methyl-erythritol phosphate pathway, where synthesis is carried out in the plastids mainly responsible of the formation of monoterpenes (C 10 ) and volatile diterpenes (C 20 ) [1]. Regarding the odor of this group of compounds, monoterpenoid hydrocarbons generally have a spicy and/or resinous odor, oxygenated monoterpenes have a sweet or citrus odor, and sesquiterpenes have a green and floral odor; however, diterpenes are rarely present in floral fragrances due to their low volatility [7]. Although close in the composition of floral volatiles, the two forms of T. xiphioides present, nevertheless, a slight difference in the intensities of certain compounds, which can affect the floral odor and lead to a difference in odor for the two forms of this species [11].
Found mainly in the flowers of both forms of T. crocata, phenylpropanoids represent the second class of VOCs emitted by plants. The first molecule of the phenylpropanoid pathway is phenylalanine, which is derived from the shikimate biosynthetic pathway. The deamination of this molecule by phenylalanine ammonia-lyase (PAL) forms the phenylpropanoid skeleton producing trans-cinnamic acid. All phenylpropanoid compounds are derived from trans-cinnamic acid by a succession of hydroxylation, methylation, and reduction steps [21]. Among the phenylpropanoids identified in this study, benzylacetate and phenylethyl acetate were found mainly in the floral emissions of yellow crocata. Benzyl acetate is a compound with a pleasant odor and is commonly found in plants such as jasmine, hyacinth, gardenia, and azalea, and is thought to be formed from the oxidation of cinnamoyl CoA [7,8,22]. Phenylethyl acetate is widely used in the cosmetic, food, and pharmaceutical industries for its pleasant rose smell. It is present in rose essential oils and is the main aromatic volatile ester emitted by rose flowers. Traditionally, this compound is obtained by acetylation of phenylethyl alcohol; moreover, the acetylation of aromatic alcohols catalyzed by alcohol acetyltransferase leads to volatile acetate esters in plants [23,24]. Other phenylpropanoids, such as eugenol, methyl eugenol, benzyl benzoate, phenethyl benzoate, and benzyl salicylate, were very present in the floral emissions of orange crocata. In the phenylpropanoid biosynthetic pathway, eugenol and isoeugenol are obtained from coniferyl acetate through eugenol synthase and isoeugenol synthase. Subsequently, a methylation reaction of eugenol, catalyzed by an O-methyltransferase, allows us to obtain methyl eugenol [21]. Eugenol and methyl eugenol are major constituents of several essential oils, such as clove, basil, cinnamon, lavender, and jasmine essential oils. Eugenol is widely used in perfumery due to its powerful spicy odor, and methyl eugenol is widely used as a flavoring agent in food products [25][26][27]. In plants such as Clarkia breweri and Petunia hybrida, benzyl benzoate and phenethyl benzoate are obtained from benzyl alcohol and phenylethanol, respectively, through benzoyl-CoA: benzyl alcohol/phenylethanol benzoyltransferase [28,29]. In addition to its activity in the treatment of scabies and lice, benzyl benzoate is used as a fixative in perfumery [26]. Benzyl salicylate is the benzyl ester of salicylic acid, it is widely used in the cosmetics industry and is used in the composition of many perfumes [30]. Volatile aromatic esters generally contribute to a sweet fruity smell [31]. These results show that there is a slight difference in the floral volatile profiles of the two forms of T. crocata, especially in the intensities of some volatile compounds. This difference could be associated with the difference in floral colors of the two forms of this species. Indeed, it was shown that plants of the same species having a variation of floral color also presented a variation in the composition of floral volatiles. For example, a study on orchids showed that the color morphs of O. simia presented different volatile profiles; the white flowers emitted more benzenoid compounds and lipid products than the purple flowers [32]. However, there are studies that did not find a clear difference in the odor chemistry of color morphs of the same species [33,34]. Thus, even if the slight difference in volatile profiles of yellow and orange crocata flowers could be associated with the difference in flower colors, it could also be due to natural selection by pollinators or mutations in genetic drift or metabolic pathways.
In T. caliginosa, floral emissions showed, in addition to phenylpropanoids, higher intensities of compounds of the 'others' family than in the 'others' species. Indeed, compounds such as 1-hexadecanol, hexadecyl ethanoate, methyl tetradecanoate, and hexadecanal are very present in T. caliginosa. These compounds are derivatives of fatty acids, which constitute another large family of volatile compounds. They are produced following a series of reactions catalyzed by lipoxygenases, hydroperoxide lyases, isomerases, and dehydrogenases from polyunsaturated fatty acids and phospholipids [1]. They generally bring a waxy and oily note to the floral scent of the plants. It should also be noted that the flowers of both forms of T. crocata as well as T. caliginosa have shown a high intensity of methyl nicotinate. The latter is a vasodilator, which is used in topical preparations as a rubefacient in case of muscle pain [35].

HS-SPME Conditions
In our previous study, the Carboxen/Polydimethylsiloxane (CAR/PDMS) 75 µm fiber was determined to be the most suitable for extracting the maximum amount of volatile compounds emitted by T. xiphioides flowers, so this fiber was used for the present study. The CAR/PDMS and the SPME device were ordered from Supelco (Bellefonte, PA, USA). Prior to use, the fiber was conditioned according to the manufacturer's recommendations.
Extractions were performed by placing a flower in a 20 mL amber vial sealed with a polytetrafluoroethylene (PTFE) septum-lined cap. For each Tillandsia flower, two successive extractions were performed: a first extraction method of 20 min at a temperature of 30 • C and a second extraction method of 65 min at 75 • C. These extraction methods provide a global view of the volatile compounds present in Tillandsia flowers and were retained after the optimization of the SPME method performed during our previous study. For both extraction methods, the equilibrium and desorption times are, respectively, 7.5 and 4 min.

Instrumentation and GC-MS Conditions
The analysis was performed on an Agilent 7890 B gas chromatograph coupled with an Agilent 5977 A mass spectrometer (Agilent Technologies, Santa Clara, CA, USA) with an MPS autosampler, a thermal desorption unit (TDU), and a cooled injection system (CIS) (Gerstel, Mülheim, Germany). MSD Chemstation version F.01.00 data acquisition software (Agilent Technologies, Santa Clara, CA, USA) was used to program the GC-MS. After extraction, desorption was performed thermally in the TDU at the recommended temperature of 300 • C for CAR/PDMS. After desorption, the injection was performed in split mode; the analytes were focused on the CIS at −10 • C for 2 min, then brought to 250 • C at a heating rate of 12 • C per second and maintained for 2.5 min. A DB-5MS capillary column (5% diphenyl cross-linked 95% dimethylpolysiloxane, 30 m × 0.25 mm × 0.25 µm) (Agilent Technologies, Santa Clara, CA, USA) was used and the separation conditions were as follows: initial column temperature of 40 • C for 2 min, then increased by 4 • C/min to 130 • C for 1 min, then increased by 7 • C/min to 230 • C, where it was maintained for 4 min. High purity Helium (99.999%) was used as the carrier gas at a flow rate of 1.2 mL per minute. The temperature of the transfer line was set at 250 • C, and the temperature of the ion source at 230 • C. The ions were generated by a 70 eV electron beam. The mass range was scanned from m/z 33 to 500 Da.

Identification
MassHunter Qualitative Analysis version B.06.00 (Agilent Technologies, Santa Clara, CA, USA) was used to integrate the majority peaks of the chromatograms, representing 95% of the peaks of the Total Ion Chromatogram (TIC) and being above the analytical noise. First, the volatile compounds were identified by comparing their mass spectra to those of compounds in the commercial databases, National Institut of Standard and Technologies (NIST) and Wiley7 (R > 800). Then, the retention indices of the compounds were calculated relative to the n-alkanes (C8-C20) and compared to those of compounds in the NIST online database (https://webbook.nist.gov/chemistry/cas-ser.html, accessed on 27 August 2021). Finally, the identifications were confirmed by injecting the available standards (mentioned above) into the GC-MS based on the comparison of mass spectra and retention times. For the standard solution, 0.2 µL of each standard was added in 2 mL of hexane, and the solvent was subsequently evaporated with nitrogen. The analysis was performed under the same conditions as for the flowers.

Statistical Analysis
For statistical analysis, peak area values of the total ion chromatograms were measured with MassHunter and transferred to Excel (Microsoft Excel 2013, version 15.0.5363.1000). Principal component analysis (PCA) was performed using SIMCA-P software (version 15.15, Umetrics, Umea, Sweden) and a heatmap with the R software (version 4.0.3, company Foundation for Statistical Computing, Vienna, Austria) based on an univariate scaling method. For this purpose, for each plant providing two flowers (the plants of orange crocata and the plants of T. caliginosa), an average of the areas of the peaks of the compounds was calculated in Excel. Odor characteristics were obtained from the "The Good Scents" company network database (www.thegoodscentscompany.com, accessed on 27 August 2021). Heatmap data was clustered (Ward's method was used to form hierarchical clustering) and visualized (using the pheatmap-package, version 1.0.12).

Conclusions
This study allowed the identification of 66 volatile compounds in three Tillandsia species with different colors and flower shapes. This identification was performed using the HS-SPME/GC-MS technique, where the use of two extraction methods, developed in a previous study, were necessary in order to efficiently extract the majority of floral volatiles. A comparison of the volatile compound profiles by PCA between T. xiphioides, T. crocata, and T. caliginosa species showed a difference in the floral emissions of the studied species. Indeed, the floral emissions of T. xiphioides were richer in terpenoids, those of T. crocata were enriched in phenylpropanoids, and those of T. caliginosa presented more compounds of the 'others' family, including fatty acid derivatives and nitrogenous compounds. In addition, a slight intraspecific difference was also observed in T. crocata, particularly variations in the intensities of some VOCs of yellow and orange crocata. These variations in floral volatile profiles between different Tillandsia species and between different forms of the same species could be associated with several phenomena, such as natural selection by pollinators, morphological and/or color differences, mutations in metabolic pathways, or genetic drift. The differences and variations observed in the VOCs profiles can explain the differences in the floral odor of these species and forms. This study provides additional information on the composition of floral VOCs of Tillandsia species and helps to better understand the diversity within the genus Tillandsia.
Supplementary Materials: The following are available online at https://www.mdpi.com/article/10 .3390/metabo11090594/s1, Figure S1: Chromatograms obtained using the first extraction method (low values of temperature and extraction time), Table S1: Complete table of  Funding: This work was funded by La Region Occitanie and Univ. Nimes. Tillandsia PROD plant nursery (28 chemin du Cailar F-30740 Le Cailar, France) provided all the plants that were necessary for the study (in-kind contribution).

Institutional Review Board Statement: Not applicable.
Informed Consent Statement: Not applicable.

Data Availability Statement:
The data presented in this study are available within the article and Supplementary Materials. Any other data are available upon request from the corresponding author.