Chemical Characterization of the Essential Oil Compositions of Mentha spicata and M. longifolia ssp. cyprica from the Mediterranean Basin and Multivariate Statistical Analyses

This present study aims to characterize the essential oil compositions of the aerial parts of M. spicata L. and endemic M. longifolia ssp. cyprica (Heinr. Braun) Harley by using GC-FID and GC/MS analyses simultaneously. In addition, it aims to perform multivariate statistical analysis by comparing with the existing literature, emphasizing the literature published within the last two decades, conducted on both species growing within the Mediterranean Basin. The major essential oil components of M. spicata were determined as carvone (67.8%) and limonene (10.6%), while the major compounds of M. longifolia ssp. cyprica essential oil were pulegone (64.8%) and 1,8-cineole (10.0%). As a result of statistical analysis, three clades were determined for M. spicata: a carvone-rich chemotype, a carvone/trans-carveol chemotype, and a pulegone/menthone chemotype, with the present study result belonging to the carvone-rich chemotype. Carvone was a primary determinant of chemotype, along with menthone, pulegone, and trans-carveol. In M. longifolia, the primary determinants of chemotype were identified as pulegone and menthone, with three chemotype clades being pulegone-rich, combined menthone/pulegone, and combined menthone/pulegone with caryophyllene enrichment. The primary determinants of chemotype were menthone, pulegone, and caryophyllene. The present study result belongs to pulegone-rich chemotype.


Introduction
Lamiaceae, the sixth largest family among the angiosperms consisting of 236 genera with over 7000 species, is composed of conventionally used medicinal plants [1].Lamioideae and Nepetoideae are two of the most prevalent subfamilies among the total of 11 subfamilies of the Lamiaceae family [2].The genus Mentha L., belonging to the Nepetoideae subfamily, consists of 24 accepted species worldwide [3,4].Mentha spp., well known as "mint", is reported to have anti-inflammatory, sedative, antioxidant, antibacterial, and antifungal effects along with several traditional uses [5].One of the popular plants in this genus is M. spicata L., which is used worldwide for medicinal and culinary purposes [6].In Cyprus, the family Lamiaceae is represented by 32 genera, and the genus Mentha is represented by four species; M. aquatica L., M. pulegium L., M. spicata L., and M. longifolia (L.) Huds [7,8].
The essential oil components of M. spicata have been extensively reviewed [6,9,10].The essential oil composition studies have shown a variety of the major compounds in Molecules 2024, 29,1970 2 of 15 oils of M. spicata collected from the Mediterranean region [11][12][13][14][15][16][17][18][19][20].Previously, M. spicata ssp.spicata oils from Turkey have been reported as rich in menthone/isomenthone, trans-sabinene hydrate/carvone/terpinen-4-ol, and 1,8-cineole/linalool/carvone, respectively.It was also summarized in the same article that the chemotypes of M. spicata growing in the Mediterranean basin until that day were piperitone oxide-rich, piperitenone oxide-rich, carvone and/or dihydrocarvone-rich, dihydrocarveol-rich, linalool-rich, and pulegone/menthone/isomenthone-rich oils [11].There is only one report on the essential oil composition of M. spicata from Cyprus.The main compounds reported are carvone (71.3%) and limonene (12.5%) [12].However, there exists a general lack of comprehensive chemotaxonomic studies using the existing literature, with previous reviews reporting the essential oil compositions of their sampling volume, but no classifications were made between the samples apart from the original classifications made by the source literature utilized in each review study [6,9,10].
The essential oil compositions of M. longifolia were also previously reported, revealing that in the reported literature, there exists discrepancies and variations in the reported major components of essential oils.Among those reported as major compounds of M. longifolia essential oil are mainly pulegone, 1,8-cineole, menthone, menthol, carvone, limonene, piperitone, piperitenone oxide [16,[21][22][23][24][25][26][27].Chemotaxonomic research on this species is not forthcoming, with there existing only three studies with relatively large sample numbers in the previous literature, with none subjecting the samples to extensive analyses to solidify chemotaxonomic classification onto a statistical foundation [11,28,29].This is exacerbated by the fact that M. longifolia appears to exhibit heterogeneous essential oil composition, with significant numbers of chemotypes being identified in each previous study.However, these chemotype identifications rely on personal observational deductions from the raw data of the researchers, and, as such, they are highly subjective.An objective, statistical method is more reliable for high-fidelity, high-accuracy determination of chemotypes than any subjective measure.A recent review has concatenated and summarized these chemotypes, stressing the lack of phytochemical studies on the essential oil composition of M. longifolia [30].There is only one report on the essential oil composition of M. longifolia ssp.cyprica with pulegone (71.5%), 1,8-cineole (9.5%), menthone (5.0%), and limonene (3.4%) as major components [26].
There exists a large number of subspecies and variations in the essential oil composition among Mentha species in the Mediterranean Basin.Therefore, it is reasonable to think that there exist distinct chemotypes that can be identified with large-scale data processing, by employing statistical methods.This present study aims to characterize the essential oil composition of the aerial parts of M. spicata and M. longifolia ssp.cyprica, and to perform multivariate statistical analysis, principal component (PCA) and hierarchical cluster analyses (HCA), by comparing with the existing literature, emphasizing the literature published within the last two decades, conducted on both species growing within the Mediterranean Basin.Due to the endemic nature of M. longifolia ssp.cyprica, it was compared with M. longifolia.This study, therefore, uses statistical methods, for the first time as far as the authors know, to identify and establish chemotypes at a higher precision in the Mediterranean Basin.

Results and Discussion
The essential oils of M. spicata and M. longifolia ssp.cyprica were isolated by hydrodistillation and analyzed for chemical characterization using simultaneous GC-FID and GC/MS.The yields of the essential oils were calculated on a dry weight basis as 4.0% and 3.0%, respectively.Overall, 32 and 22 identified compounds were detected, comprising the total of essential oils.The major compounds of M. spicata were determined as carvone (67.8%) and limonene (10.6%), while M. longifolia ssp.cyprica contained pulegone (64.8%), and 1,8-cineole (10.0%), respectively.Table 1 shows the detailed essential oil compositions of the aerial parts of M. spicata and the endemic M. longifolia ssp.cyprica from Northern Cyprus.The HCA results of the essential oils analyses from the present study and related studies of Mentha spicata from within the Mediterranean region are given in Figure 1.The results indicate the presence of three major clades (given as red squares A-C).A linaloolrich chemotype (ranging between 86.7% and 93.9%) from Greece was composed of the same population sampled at different times [17].Another outlier is also from Greece, with an unusual piperitone epoxide and piperitenone oxide (23.0% and 41.0% of total essential oil, respectively) dominant chemotype [34].The single sample comprising another outlier was a cis-carvone oxide (44.1%) chemotype, with dihydrocarvone also present at 8.9% [15].
The M. spicata essential oil sample obtained from the present study was placed in Clade A, indicating that it belongs to the carvone-rich chemotype dominated by this compound.The other study from Cyprus that also characterized M. spicata essential oil [17] was placed into the same clade, with low Euclidean distancing, indicating that they belong to the same chemotype and are of similar composition to the present study.
The three major clades (A, B, and C) were also subjected to ANOVA to determine statistically significant differences in essential oil content.Isolated samples were not considered for ANOVA due to limited sample sizes.The distinguishing feature of the three clades was found to be the carvone content, with clade A having the greatest concentration, followed by C, and the least carvone being present in B (p < 0.001).Therefore, carvone concentration in essential oil can be considered to be a major determinant in the distinction of these clades from each other.It was determined that 1,8-cineole and limonene contents were not significantly different among the three groups (p > 0.05) and, therefore, are not significant determinants of chemotype among these three clades.
Pulegone, cis-isopulegone, menthol, and menthone content of the essential oil were both determined to be a statistically significant determinant of chemotype, with A and C being similar to each other (p > 0.05), but B being significantly different from either (p < 0.001).All of these phytochemicals were present in higher concentrations in essential oils obtained from samples in B, compared to those in A or C.
A statistically significant increase was observed in the concentration of the essential oil of 2-hydroxy-3-(3-methyl-2-butenyl)-3-cyclopenten-1-one in C compared to A (p < 0.001) at an average concentration of 7.2% in C compared to an average of 0.1% in A, providing another differentiating characteristic between the two clades [14].However, the identification of this compound was determined to be irrelevant since it was not reported from any natural source in M. spicata in the previous literature.
In light of these statistical analyses, as explained above, it can be suggested that three definitive and four isolated samples can be associated with M. spicata.The isolated samples are putative chemotypes due to the presence of a single sample in each category; therefore, no definitive assertions can be made concerning their chemotypic uniqueness.However, much more definitive deductions can be made about clades A-C, which, according to the PCA, HCA, and ANOVA discussed herein, can be divided into three chemotypes: a carvone-rich chemotype with a simple majority of carvone associated with clade A, a carvone-poor chemotype that also features enrichment in menthol, menthone, pulegone, and cis-isopulegone, associated with clade B, and a carvone-rich chemotype that is not as rich in carvone as clade A but containing higher concentrations of trans-carveol (Table 2).The results indicate that among the M. spicata essential oil samples obtained from references, as well as the current study, there exists a significant variation in the essential oil composition, with certain possible clusterings (Figure 2).Among these, linalool, carvone, pulegone, and menthone appear to be significant components contributing to essential oil variation, with some others, such as limonene and 1,8-cineole, contributing to a lesser degree (Table 2).
The samples within M. longifolia were divided into three definitive clades and two putative clades, made of single members.The putative clades were disregarded for further statistical analyses.The definitive clades were named X, Y, and Z (Figure 3).Clade X was composed of a total of seven samples [21,22,25,27].Clade Y was composed of a total of four samples [28].Clade Z was composed of a total of three samples [24,26,27].
Clade X corresponds to a pulegone-rich clade with strong enrichment in pulegone (56.9% averaged), and low menthone content at approximately 6.3%.The current study sample is in clade X, with 64.8% pulegone and 7.6% menthone content.Clade Y corresponds to a menthone/pulegone chemotype that displays reduced pulegone (14.4% on average) but also displays enrichment in menthone (25.2% on average).Clade Z is characterized by a combined menthone/pulegone chemotype (12.6% and 18.3%, respectively) and slight caryophyllene enrichment (2.8% on average) (Table 3).The samples within M. longifolia were divided into three definitive clades and two putative clades, made of single members.The putative clades were disregarded for further statistical analyses.The definitive clades were named X, Y, and Z (Figure 3).Clade X was composed of a total of seven samples [21,22,25,27].Clade Y was composed of a total of four samples [28].Clade Z was composed of a total of three samples [24,26,27].[24], R3 [16], R4 [26], R5 [25], R6 [22], R7 [21], R8 [27].The samples within M. longifolia were divided into three definitive clades and two putative clades, made of single members.The putative clades were disregarded for further statistical analyses.The definitive clades were named X, Y, and Z (Figure 3).Clade X was composed of a total of seven samples [21,22,25,27].Clade Y was composed of a total of four samples [28].Clade Z was composed of a total of three samples [24,26,27].[24], R3 [16], R4 [26], R5 [25], R6 [22], R7 [21], R8 [27].[24], R3 [16], R4 [26], R5 [25], R6 [22], R7 [21], R8 [27].ANOVA revealed that menthone was a major determinant of chemotype, with clades X and Z on one hand and Y on the other, being significantly different in menthone concentrations of their essential oils (p < 0.05).X and Z did not have a significant difference in menthone concentration (p > 0.05).On the other hand, clade X was differentiated from Y and Z with a significant difference in pulegone concentration (p < 0.05), whereas Y and Z did not have a significant difference in pulegone content.Finally, there existed a statistically significant difference in the caryophyllene concentration in clade Z compared to clades X and Y (p < 0.05).Clades X and Y did not have significantly different caryophyllene content (p > 0.05).These findings were corroborated by the PCA conducted on M. longifolia, which indicated menthone, pulegone, and caryophyllene as significant contributors to variation among samples.Terpinen-4-ol and piperitone oxide could not be determined, due to inability to establish homogeneity of variances as a prerequisite for ANOVA (Table 3, Figure 4).
The present study results demonstrate the utility of multivariate statistical analysis of essential oil for the determination of chemotypic taxonomy.These novel methods, with their higher discretionary power, allow for more reliable classification of samples into existing chemotypes.Such applications are important when, for example, cultivating plants for medicinal or industrial purposes, so that the essential oil composition, and therefore the utility of plants from a medicinal or industrial standpoint, can be more definitively ascertained, increasing safety for the former, and yield, and efficiency for the latter.As an example, a recent study, also utilizing the principles of HCA and PCA to compare their samples, discovered previously undescribed chemotypes of Mentha sp.[58].The present study results demonstrate the utility of multivariate statistical analysis of essential oil for the determination of chemotypic taxonomy.These novel methods, with their higher discretionary power, allow for more reliable classification of samples into existing chemotypes.Such applications are important when, for example, cultivating plants for medicinal or industrial purposes, so that the essential oil composition, and therefore the utility of plants from a medicinal or industrial standpoint, can be more definitively ascertained, increasing safety for the former, and yield, and efficiency for the latter.As an example, a recent study, also utilizing the principles of HCA and PCA to compare their samples, discovered previously undescribed chemotypes of Mentha sp.[58].
The statistical determination of chemotype Is also highly important in chemotaxonomy from a purely scientific perspective.In the present study, the present sample was chemotaxonomically classified purely on the basis of the HCA classification.While such a result could also be obtained from a subjective determination of the chemotype, there may always be confounding factors that a scientist may ignore or miss.For example, the M. longifolia analysis in this study indicated that the slight enrichment in caryophyllene is a determining factor in clade Z, despite the caryophyllene concentration not exceeding 3%.It, thus, demonstrates that different compounds can be strong determinants of chemotype even in small concentration differences, which may not necessarily be all that obvious to a subjective determination by a human.This chemotaxonomy determination based on statistical differences, regardless of the concentration scale in question, opens a new dimensionality to chemotype determination, and allows for the distinction between chemotypes that are superficially similar in the composition of the major substances, but may exhibit subtle differences in their composition, therefore comprising two different, albeit closely related, chemotypes.
These differences in composition are especially important in the medicinal and industrial cultivation of such plants, where the toxicological profile, which depends on not only the concentration, but also the specific toxicity of the compounds in question, chemotypic differences between a chemotype that does not contain any toxic compounds above threshold limits, and a chemotype that has a slightly higher concentration of a highly toxic compound, becomes crucial from a safety perspective, thus requiring the distinction of chemotypes based on chemical composition.The statistical determination of chemotype Is also highly important in chemotaxonomy from a purely scientific perspective.In the present study, the present sample was chemotaxonomically classified purely on the basis of the HCA classification.While such a result could also be obtained from a subjective determination of the chemotype, there may always be confounding factors that a scientist may ignore or miss.For example, the M. longifolia analysis in this study indicated that the slight enrichment in caryophyllene is a determining factor in clade Z, despite the caryophyllene concentration not exceeding 3%.It, thus, demonstrates that different compounds can be strong determinants of chemotype even in small concentration differences, which may not necessarily be all that obvious to a subjective determination by a human.This chemotaxonomy determination based on statistical differences, regardless of the concentration scale in question, opens a new dimensionality to chemotype determination, and allows for the distinction between chemotypes that are superficially similar in the composition of the major substances, but may exhibit subtle differences in their composition, therefore comprising two different, albeit closely related, chemotypes.
These differences in composition are especially important in the medicinal and industrial cultivation of such plants, where the toxicological profile, which depends on not only the concentration, but also the specific toxicity of the compounds in question, chemotypic differences between a chemotype that does not contain any toxic compounds above threshold limits, and a chemotype that has a slightly higher concentration of a highly toxic compound, becomes crucial from a safety perspective, thus requiring the distinction of chemotypes based on chemical composition.

Plant Material
The collected plant materials were identified by K.H.C. Başer and D. Özkum Yavuz, according to the Flora of Cyprus [13].Aerial parts of the cultivated M. spicata (Flowering period: July-November), and natural M. longifolia ssp.cyprica (flowering period: June-November) were collected from Cengizköy-Lefka/Northern Cyprus on 1 July 2022 during the flowering stage.They were separately air-dried in the shade.Voucher specimens are kept at the Herbarium of the Near East University, Turkish Republic of Northern Cyprus (NEUN) with the voucher numbers NEUN 20001 and 20002.

Isolation of Essential Oils
One hundred grams of air-dried samples were separately distilled for 3 h using a Clevenger-type apparatus by hydrodistillation.The resulting essential oil was stored at 4 • C until further analyses.The oil yields were calculated as v/w on a dry weight basis.

Selection of Mentha spicata and M. longifolia Essential Oils for Multivariate Analyses
The samples for comparison were selected from previous research conducted on Mentha spicata and M. longifolia within the Mediterranean Basin.All samples were restricted to the Mediterranean Basin to prevent inevitable divergence in plant physiology due to divergent edaphic and environmental factors.Previous research was also limited to approximately the last 20 years so that differences in available technological tools and research methodologies could be minimized.Studies that indicated raw percentile values were taken into account, with those that gave mean ± SD also excluded from analysis.

GC-FID Analysis
The GC/MS analysis was carried out with an Agilent 5977B GC-MSD system (Santa Clara, CA, USA).Innowax FSC column (Agilent, 60 m × 0.25 mm, 0.25 mm film thickness) was used with helium as carrier gas (0.8 mL/min).GC oven temperature was kept at 60 • C for 10 min and programmed to 220 • C at a rate of 4 • C/min, and kept constant at 220 • C for 10 min and then programmed to 240 • C at a rate of 1 • C/min.Split ratio was adjusted at 40:1.The injector temperature was set at 250 • C. Mass spectra were recorded at 70 eV.Mass range was from m/z 35 to 450.FID results were used to report the characterized compounds' relative percentages (%) [59].

GC/MS Analysis
The analysis was carried out using an Agilent 7890B GC system.The integrated FID detector temperature was 300 • C. To obtain the same elution order with GC/MS, simultaneous auto-injection was performed on a duplicate of the same column applying the same operational conditions.Relative percentage amounts of the separated compounds were calculated from FID chromatograms.Identification of the essential oil components was carried out by comparison of their relative retention times (RRT) with those of authentic samples or by comparison of their linear retention index (LRI) to a series of n-alkanes.Computer matching against commercial (Wiley GC/MS Library, NIST Library) and inhouse "Başer Library of Essential Oil Constituents" built up by genuine compounds and components of known oils, as well as MS literature data, was used for the identification [59].

Statistical Analysis
All relevant data were imported to IBM SPSS Statistics v27.0 (International Business Machines (IBM) Corporation, Armonk, NY, USA).Principal component analysis (PCA) was performed using the correlation matrix method with sequential eigenvalues selected based on the introduction of "kinks" in the scree plot.The varimax rotation method was employed to improve the correlation between chemical constituents and principal components.Correlation matrices were employed to ascertain the effect of different constituents on chemotypes.Hierarchical cluster analysis (HCA) was performed using the squared Euclidean distance between-groups linkage method using agglomeration schedules.The dendrograms were produced from the HCA using this data.Only components that were deemed major by the authors of at least one of the references cited herein were included in the PCA and HCA.Previous studies from the Mediterranean Basin were included in the present study as references, with those outside of the Mediterranean Basin excluded from the study.The clades as determined by HCA were subjected to one-way analysis of variance (ANOVA) to confirm chemotypic differences, with Levene's test employed to test for homogeneity of variances, and Bonferroni's post hoc test was utilized to ascertain differences between identified clades.

Table 1 .
The essential oil compositions of the aerial parts of cultivated M. spicata and natural M. longifolia ssp.cyprica from Northern Cyprus.

Table 2 .
The rotated component (loadings) matrix of the essential oil compositions of Mentha spicata in the Mediterranean Basin.

Table 3 .
The rotated component (loadings) matrix of the essential oil compositions of Mentha longifolia in the Mediterranean Basin.