The Essential Oil Compositions of Three Teucrium Taxa Growing Wild in Sicily: HCA and PCA Analyses

The chemical composition and the qualitative and quantitative variability of the essential oils of three taxa belonging to the Teucrium genus were studied. The investigated taxa, that grow wild in Sicily, were Teucrium flavum L. (section Chamaedrys (Mill.) Scheb.), Teucrium montanum and Teucrium capitatum L. of section Polium (Mill.) Scheb. Essential oils were extracted by hydrodistillation and analyzed by GC-MS. In total, 74 compounds were identified. Sesquiterpene hydrocarbons were found to be the main group for T. flavum (48.3%). T. capitatum consisted essentially of monoterpene hydrocarbons (72.7%), with α-pinene (19.9%), β-pinene (27.6%) and sylvestrene (16.6%) as the most abundant compounds whereas ledene oxide (12.1%), epiglobulol (13.5%) and longifolenaldehyde (14.5%) were identified as the main constituents among the oxygenated sesquiterpenes (63.5%) of T. montanum. Furthermore, a complete literature review on the composition of the essential oils of all the other accessions of these Teucrium taxa, studied so far, was performed. Hierarchical Cluster Analysis (HCA) and Principal Component Analyses (PCA) were used in order to demonstrate geographical variations in the composition of the essential oils.


Introduction
According to The Plant List [1] more than nine hundred and fifty scientific plant names of species rank for the genus Teucrium are present. Of these, more than three hundred are accepted names, including species, subspecies, varieties, forms and hybrids. The southern, south-western and south-eastern parts of Europe are considered as the main center of differentiation of the genus although a significant number of these species grow also in Central Asia, south-western Asia, north-western Africa, southern North America, southwestern South America and Australia [2,3]. These perennial, bushy or herbaceous plants live commonly in sunny habitats [4] and on the basis of their calyx shape, inflorescence structure and pollen morphology they have been divided into ten sections (Teucropsis Benth., Teucrium Benth., Chamaedrys (Mill.) Schreb., Polium (Mill.) Schreb., Isotriodon Boiss., Pycnobotrys Benth., Scorodonia (Hill) Schreb., Stachyobotrys Benth., Scordium (Mill.) Benth.
lished [20]. Consequently, in the frame of our ongoing research on Sicilian plants [21][22][23][24], and in order to improve the knowledge on genus Teucrium, we decided to investigate the chemical compositions of the other three Sicilian taxa of Teucrium, which have never been analyzed: T. flavum, belonging to Section Chamaedrys, and T. montanum and T. capitatum, belonging to Section Polium. We also screened the reported literature in order to find data concerning the chemical composition of their essential oils and we performed Hierarchical Cluster Analysis (HCA) and Principal Component Analyses (PCA) in order to find a similarity among the Sicilian accessions and the other taxa belonging the same species studied so far.

Composition of the Essential Oils
Hydrodistillation of the aerial parts of T. flavum (O(f )) gave a pale-yellow oil. Overall, forty-one compounds were identified, representing 91.2% of the total compositions. The components are listed in Table 1 according to their retention indices on a DB-5 column and are classified on the basis of their chemical structures into five classes. This essential oil was rich in sesquiterpene hydrocarbons (48.3%). β-Bisabolene (26.8%) was, by far, the main components of this class as well as of the oil, followed by β-caryophyllene (6.6%), γ-cadinene (5.5%) and α-caryophyllene (3.1%). Oxygenated sesquiterpenes were present in a lower amount (11.0%) with caryophyllene oxide (3.1%) as the main compound of the class whereas oxygenated monoterpenes were practically absent. Monoterpene hydrocarbons accounted for 27.0%, limonene (12.7%) being the main product, followed by α-pinene and β-pinene (7.0% and 5.4%, respectively). Table 1. Composition (%) of the essential oils of T. flavum (O(f )), T. capitatum (V(c)) and T. montanum (F(m)) collected in Sicily.  Table 2 reports the main components of the essential oils, obtained by hydrodistillation, of the other populations of T. flavum, previously investigated, as well as of the other accessions of T. montanum and T. capitatum collected in different countries.   Portugal, Cantanhede

Montenegro, River
Moraca 14.2 8.9 48.5 11.1 16.6 [47] Italy, Marche Italy, Tuscany Corsica Tunisia A comparison of our data with those reported in the literature (Table 2) shows some very interesting points. For our comparison and further statistical analysis, we reported all the papers related only to essential oils achieved by hydrodistillation and only obtained from aerial parts. With the exception of the populations of Corsica (France) [35], Tuscany (Italy) and Liguria (Italy) [48], all the other accessions indicate sesquiterpene hydrocarbons as the main class of the oils. Among these ones, the plants collected in Montenegro [27] showed a very similar profile with respect to (O(f )). In fact, in this oil, the main sesquiterpene components were β-bisabolene (35.0%) and β-caryophyllene (5.4%) and among the monoterpene hydrocarbons, as for (O(f )), the principal metabolites were α-pinene (17.5%), β-pinene (11.5%) and limonene (6.4%). A study on the different vegetative parts of a Sicilian population of T. flavum has been published some years ago [51] but the results were not inserted in Table 2 since the oils were obtained with a different method (microwave-assisted hydrodistillation). By the way, also in this case, the main component of all the investigated parts was β-bisabolene, although, the second more abundant compound, germacrene D, was totally absent in (O(f )).

PCA and HCA Analyses of the Essential Oil Composition of Teucrium Taxa
As stated before, the main compounds of the essential oils of Teucrium taxa, collected in different accessions, and their relative abundance are reported in Table 2. For the compilation of the Table 2, the following points were considered: (I) only compounds with abundance ≥3%, and (II) only essential oils EOs obtained by hydrodistillation were taken into consideration.
The analyses were carried out considering the classes' compounds with a significant contribution, according to the loading plot obtained by principal component analysis (PCA) for monoterpene hydrocarbons (MH), oxygenated monoterpenes (OM), sesquiterpene hydrocarbons (SH), oxygenated sesquiterpenes (OS) and other compounds (O).
For the T. montanum essential oils, as shown in the loading graph (Figure 1), all variables affected PC1 and PC2. In fact, PC1 (51.6%) was represented mainly by oxygenated sesquiterpene (OS) in the positive score, and in a minor contribution by MH, OM and O in negative scores; meanwhile, PC2 (27.7%) was represented mainly by a negative score of SH and positive scores of MH and OS.
HCA based on the Euclidean distance between groups indicated two species groups (A and B, Figure 2) identified by their essential oil chemotypes with a similarity <2.
A first Group A, in HCA analysis, the samples of T. montanum from Jabura (E(m)) and Mt. Orjen (C(m)), Serbia, whose oxygenated sesquiterpene compounds content (24.1-33.4%) differentiated them from the other oils were included. This cluster was characterized by the highest content of germacrene D (15.0%) and a relatively high content of β-caryophyllene (5.1-6.9%), δ-cadinene (3.6-4.5%) and γ-cadinene (3.6-4.1%). A definite Group B containing the samples collected in Croatia (A(m)) and Turkey (B(m)), was characterized by oils with a large percentage of monoterpene and sesquiterpene hydrocarbons; the difference between the two oils was mainly due to the O class (8.8-22.5%), which was the majority in Croatian EO.
The others two oils, Serbia Jadovnik (D(m)) and Sicily (F(m)), for their high contents of SH (74.2%) and OS (63.5%), respectively, were considered as two separate classes, without any similarity with the other clusters.
The PCA of T. flavum EOs (Figure 3), belonging to section Chamaedrys, presented a total variance of 90.2% of the original data. The most of samples showed a cluster formation in this model, affirming a similarity in the chemical composition of these essential oils. The PCA horizontal axis explained 55.3% of the total variance while the vertical axis a further 34.9%. HCA based on the Euclidean distance between groups indicated a solution with three clusters (A', B' and C'), with a dissimilarity <2 (Figure 4) which was mainly due to the variation along the major axis in PCA analysis. These clusters formed separate groups in the PCA biplot ( Figure 3).
HCA based on the Euclidean distance between groups indicated two species groups (A and B, Figure 2) identified by their essential oil chemotypes with a similarity < 2.
A first Group A, in HCA analysis, the samples of T. montanum from Jabura (E(m)) and Mt. Orjen (C(m)), Serbia, whose oxygenated sesquiterpene compounds content (24.1%-33.4%) differentiated them from the other oils were included. This cluster was characterized by the highest content of germacrene D (15.0%) and a relatively high content of β-caryophyllene (5.1-6.9%), δ-cadinene (3.6-4.5%) and γ-cadinene (3.6-4.1%). A definite Group B containing the samples collected in Croatia (A(m)) and Turkey (B(m)), was characterized by oils with a large percentage of monoterpene and sesquiterpene hydrocarbons; the difference between the two oils was mainly due to the O class (8.8-22.5%), which was the majority in Croatian EO.  Table 2.  Table 2.

Figure 2. Dendrogram obtained by Hierarchical Cluster Analysis (HCA) based on the Euclidian distances between groups of A(m), B(m), C(m), D(m), E(m) and F(m)
Teucrium montanum essential oils. The plants codes are reported in Table 2.
The others two oils, Serbia Jadovnik (D(m)) and Sicily (F(m)), for their high contents of SH (74.2%) and OS (63.5%), respectively, were considered as two separate classes, without any similarity with the other clusters.
The PCA of T. capitatum revealed that the first principal component (PC1 and PC2) represented the 60% of the total information ( Figure 5). The general structure of the dendrogram ( Figure 6) generated by HCA indicated the existence of three main clusters of populations, based on their chemical composition and the Euclidean distance between groups (distance < 2). The graph, in fact, presented a first large cluster (Group A'') formed by essential oils harvested in Boussaada (A(c)) (Algeria), Bouira (B(c)) (Algeria), Bulgaria (C(c)), Athens (G(c)), (Fonte Coberta (N(c)), Rabaça (O(c)), Cantanhede (P(c); Q(c)), Serra D′Aire (R(c)), Portugal), Gennargentu (Sardinia) (T(c)) and Serbia (U(c)), which was characterized by a variable composition of monoterpenes and sesquiterpenes, both hydrocarbons and oxygenated ones.  Table 2. Different from this group, due to a variation in the positive scores of PC2 (34.9%), was the oil from Corsica (M(f )), essentially constituted by MH and characterized principally by α-pinene, β-pinene and limonene. Moreover, the M(f ) essential oil was out of the 95% confidence marked by the circle.
The second subgroup B', instead, including T. flavum collected in Sicily (O(f )) and in Tunisia (N(f )), showed oils containing relevant quantities of both hydrocarbon and oxygenated sesquiterpenes (SH, OS). A marked difference was observed for Iranian oil (D(f )), which was abundant in SH (82.2%).
The PCA of T. capitatum revealed that the first principal component (PC1 and PC2) represented the 60% of the total information ( Figure 5). The general structure of the dendrogram ( Figure 6) generated by HCA indicated the existence of three main clusters of populations, based on their chemical composition and the Euclidean distance between groups (distance < 2). The graph, in fact, presented a first large cluster (Group A") formed by essential oils harvested in Boussaada (A(c)) (Algeria), Bouira (B(c)) (Algeria), Bulgaria (C(c)), Athens (G(c)), (Fonte Coberta (N(c)), Rabaça (O(c)), Cantanhede (P(c); Q(c)), Serra D Aire (R(c)), Portugal), Gennargentu (Sardinia) (T(c)) and Serbia (U(c)), which was characterized by a variable composition of monoterpenes and sesquiterpenes, both hydrocarbons and oxygenated ones.   Table 2.
Lowering the level of dissimilarity (cut-off = 1, Figure 6), it is possible to divide this group into two subclusters (B''1 and B''2). Subcluster B''2 included two samples, (D(c)) and (E(c)), geographically close to each other and characterized by a high percentage of

Plant Material
Aerial parts of T. flavum (O(f)) were collected in June 2020, near Noto Antica (SR), (Sicily, Italy), at 380 m of altitude (36°57′25.37″ N and 15°02′18.76″ E). A voucher of the population analyzed was deposited in the herbarium of the University of Palermo (PAL 109709).
Aerial parts of T. montanum (F(m)) were sampled, in June 2020, in Contrada Quacella on the Madonie Mountains (Sicily, Italy) in an environment of mountain garrigue on Dolomite substrates, between 1450 and 1550 m of altitude in an area whose center had coordinates 37°50′49.78″ N; 14°1′22.59″ E. A voucher of the population analyzed was deposited in the herbarium of the University of Palermo (PAL 109708).
The aerial parts of T. capitatum (V(c)) were taken, in June 2020, in the chalky hilly center of Sicily, in an area in the countryside of Marianopoli (CL), Sicily, Italy with geographical coordinates 37°36′4.58″ N; 13°57′46.10″ E, about 700 m above sea level and the respective voucher was deposited in the herbarium of the University of Palermo (PAL 109711).

Essential Oil Extraction
A variable quantity of the aerial parts of T. flavum, T. montanum and T. capitatum (33-105 g) were subjected to hydrodistillation for 3 h second using Clevenger's apparatus [52]. The oils (yields 0.08% (v/w), 0.15% and 0.07% for (O(f)), (F(m)) and (V(c)), respectively), were dried with anhydrous sodium sulphate, filtered and stored in the freezer at −20 °C, until the time of analysis.

Chemical Analysis of Essential Oils
Analysis of essential oil was performed according to the procedure reported by Rigano et al. [22]. EO analysis was performed using an Agilent 7000C GC (Agilent Technologies, Inc., Santa Clara, CA, USA) system, fitted with a fused silica Agilent DB-5 capillary column ( Table 2. With a dissimilarity <1, group A" was divided in three subgroups (A"1, A"2 and A"3) in HCA analysis. The first subgroup A"1 is formed by (B(c)) and (T(c)) essential oils, and it is marked by a high content of sesquiterpene hydrocarbons (37.8-43.7%), a medium level of oxygenated sequiterpenes (10.0-13.5%) and a null contribution of others (0-0.1%).

Essential Oil Extraction
A variable quantity of the aerial parts of T. flavum, T. montanum and T. capitatum (33-105 g) were subjected to hydrodistillation for 3 h second using Clevenger's apparatus [52]. The oils (yields 0.08% (v/w), 0.15% and 0.07% for (O(f )), (F(m)) and (V(c)), respectively), were dried with anhydrous sodium sulphate, filtered and stored in the freezer at −20 • C, until the time of analysis.

Chemical Analysis of Essential Oils
Analysis of essential oil was performed according to the procedure reported by Rigano et al. [22]. EO analysis was performed using an Agilent 7000C GC (Agilent Technologies, Inc., Santa Clara, CA, USA) system, fitted with a fused silica Agilent DB-5 capillary column (30 m

Statistical Analysis
The essential-oil contents that exceeded 3.0% of the total oil composition in at least one species were considered as original variables and subjected, after normalization, to cluster analysis (CA) and to Principal component analysis (PCA). The statistical analyses were performed using PRIMER 6 (Massey University Eastbourne, Albany, New Zealand) with two principal components (PC) variables and the number of clusters were determined by using the rescaled distances in the dendrogram, using a cut-off point (Euclidean distance = 2) that allows the attainment of consistent clusters. The Principal Components Analysis (PCA) and the Hierarchical Cluster Analysis (HCA) were used to comprehend the similarity among the essential oils in relation to the contents of their chemical constituents. We tested two different cut-off similarity levels (cut-off level 1 and cut-off level 2), chosen on the basis of the mean distance between clusters measure and based on the similarities-differences between the samples belonging to the same cluster. Since the HCA analysis is a function of variables and observations, the highest correspondence between PCA and HCA resulted when we applied a cut-off of 2. A cut-off of 1 greatly increases the diversity between the analyzed theses and would lead to an incorrect clustering. The statistical analysis of the absence/presence was carried out using the cluster method of the PRIMER 6 software (Massey University Eastbourne, Albany, New Zealand) [53].

Conclusions
Chemical and statistical analyses (PCA and HCA) carried out on three Sicilian taxa (T. flavum, T. montanum and T. capitatum) belonging to the genus Teucrium, can provide chemotaxonomic information on the taxa investigated. All EOs were analyzed by GC-MS. T. flavum is mainly represented by sesquiterpene hydrocarbons (48.3%), the main class also found in most of all accessions already investigated. Sicilian T. capitatum essentially consisted of monoterpenes such as α-pinene (19.9%), β-pinene (27.6%) and sylvestrene (16.6%), finding a high similarity with the population collected in Bulgaria. T. montanum, on the other hand, showed a high abundance in both hydrocarbon and oxygenated sesquiterpenes (94.3%). The PCA analyses, and the subsequent analyses of the Clusters, based on the different chemical classes, represent a useful tool towards a complete taxonomic investigation, also leading to an understanding of the diversification of the genus Teucrium.