Metabolomic profiles from organisms in the Plantae and other kingdoms have been used to distinguish samples from different species and ecotypes, as well as aid in their classification (Dettmer et al., 2007; Allwood et al., 2011; Aliferis et al., 2013; Jing et al., 2015). In the case of plants, much attention has been given to one of the largest group of metabolites they produce, the terpenoids. In a study similar to ours which explored the chemodiversity of Selaginella in the context of evolution, Weng and Noel (2013) suggest that monoterpene and sesquiterpene synthases are derived from conserved diterpene synthases, based off of phylogenetic analyses of plant terpene synthases done by Bohlmann et al. (1998). Thus, it appears that monoterpenes and sesquiterpenes are frequently used as chemical markers for distinguishing between genera. For example, in the Lantana (Verbenaceae) genus, β-caryophyllene was the major compound suggested as a chemical marker (Sena Filho et al., 2010), while in Citrus, both cyclic (R-(+)-limonene and α-terpinene) and bicyclic monoterpenes (α-pinene and sabinene) were dominant in fruit essential oils and therefore proposed as markers for those species/varieties (Jing et al., 2015). Worldwide, the volatile metabolomics composition of Vitex species has been poorly investigated, likely due to their accessibility and phenotype, namely, their size and bad odor. From a total of 250 species, only about 35 are reported in the literature with in-depth chemical characterizations. Many of them are difficult to identify due to endemism, e.g., Vitex rufences A. Juss, Vitex capitata Vahl, Vitex gardneriana Schauer, which are found only in areas in the northeast of Brazil.

The main goal of this study was to develop a metabolomics approach for evaluating the volatile composition of Vitex species, using four species we collected (Vitex capitata Vahl, Vitex megapotamica (Spreng.) Moldenke, Vitex gardneriana Schauer and Vitex rufescens A. Juss) in addition to 13 Vitex species published in the literature. First, the chemical composition of the volatile fraction of essential oils from leaves of these species was characterized via GC-MS/FID. Subsequently, cluster analyses of the chemical composition of the 17 Vitex species were performed. Next, Inter-Simple Sequence Repeat (ISSR) was used to evaluate the genetic variation between our four Vitex species, followed by cluster analyses. Lastly, bioactivity of our four essential oil extracts was evaluated using toxicity assays against Aceria guerreronis Keifer (Acari: Eriophyidae), a major pest found in coconut plantations worldwide (Návia et al., 2013). From this study, we aimed to chemosystematically characterize the Vitex genus, in addition to expand our understanding of it in an evolutionary context.

GC-MS/FID analysis of essential oils collected from leaves of V. capitata, V. megapotamica, V. gardneriana and V. rufescens (yield of 0.13, 0.18, 0.33, and 0.45%, respectively) determined 46, 45, 41, and 37 constituents to be present, respectively (Table 2). Grouping for the species by Euclidian distance based on seven-member ring sesquiterpenes showed clades separating by the production of bicyclogermacrene, aromadendrane and 5,1-cycloaromadendrane sesquiterpenoids, e.g., aromadendrene, viridiflorene, ledol, viridiflorol, allo-aromadendrene, bicyclogermacrene, globulol, 6,9-guaiadiene, spathulenol and palustrol (Figure 1). We then analyzed the volatile composition of the essential oils from 13 previously published Vitex species, as well as the four Vitex species studied here, three of them endemic to Brazil (V. capitata, V. gardneriana and V. rufescens) via a clustering algorithm (Figure 2). The 17 species comparison showed strong similarities in the production of some monoterpenes, but varied by their production of larger terpenes, especially those with gem-dimethylcyclopropyl subunits on seven-member ring compounds.

The 14 ISSR primers generated a total of 90 fragments, 100% of which were polymorphic. The primer with the highest number of fragments was 857 (eleven), while the lowest were 845 and 888 (two). The values of PIC and MI are parameters that allow for estimation of the power of discrimination of the molecular marker by a primer; in our results, the primers 842 and 818 were the most informative (Table 1). We observed only fragments to V. capitata for primer 823. Genotypes were clustered by UPGMA using the Jaccard coefficient (JC), estimated from the binary data (Figure 3). The similarity mean was 0.21 JC (0.08 – 0.45 JC). We observed a clear separation of two groups, with V. gardneriana being the most isolated of the four species. In our study, ISSR fingerprints clearly distinguished all of the tested species, with cluster results similar to those performed for the chemical compounds (Figure 2).

In the biotoxicity assays, the essential oil extracted from V. gardneriana was highly toxic to A. guerreronis (LC₅₀ 0.8577 mg/mL; CI 0.67-1.15; x² = 3.17, df = 2, P = 0.20). In contrast, it was not possible to estimate the LC₅₀ from V. capitata, V. megapotamica and V. rufescens essential oils, as increasing concentrations (0.1, 0.5, 1, 1.5, 1.8, 2.0, and 2.3 mg/mL) of these oils did not kill adults of A. guerreronis over a period of 24 h. As a way of toxicity comparison, the acaricide abamectin, sprayed at its label rate, inflicted 100% mortality to A. guerreronis.

To our knowledge, this is the first report of the chemical composition and biological activity of the essential oil distilled from leaves of V. capitata, V. gardneriana and V. rufescens. Of the chemical constituents detected, 88.7, 94.5, 80.5, and 90.2% were sesquiterpenes, and 2.3, 0.3, 17.1, and 8.6% were monoterpenes for V. capitata, V. megapotamica, V. gardneriana and V. rufescens, respectively. Similar chemical compositions were found in the essential oil extracts from V. capitata and V. megapotamica, where α-copaene, (E)-caryophyllene, γ-elemene, α-humulene, gamma-muurolene, viridiflorene and γ-cadinene were the major components (Figure 2 and Table 2). Characterization of V. gardneriana revealed that the most abundant compounds were 6,9–guaiadiene (19.3%), caryophyllene oxide (18.6%), L-calamenene (13.9%) and α-cadinol (3.8%). In the case of V. rufescens, (E)-caryophyllene (21%), ledol (15.7%) and germacrene D (9.3%) were the main constituents detected.

Chemotaxonomy, or chemical variation in a group of compounds, is one of the tools used to define intra- and inter-specific variability in plants (Sena Filho et al., 2007, 2012; Duran-Peña et al., 2015). Our objective was to use the chemical profile in the four Vitex species we studied, plus 13 others from the literature, to develop a system for classification of plant specimens derived from this genus. The content of sesquiterpenes with seven and nine members in their ring system in the four Vitex species described in this study was remarkable. Recently, genes responsible for sesquiterpene cyclization were evaluated in fungi, which involved scaffolds generated by terpene synthases and cyclases (Quin et al., 2014). Our data indicates that sesquiterpene synthase is likely highly expressed in this genus and hypothesize that this enzyme plays an important role in its chemosystematics. Enzyme mechanisms of sesquiterpene biosynthesis are well known: 5-epi-aristolochene synthase (TEAS) found in Nicotiana tabacum L. (tobacco) was the first cloned and sequenced sesquiterpene synthase from Angiospermae species (Maillea et al., 2006), and could serve as a useful model for understanding how these compounds are built in the plant.

For both the chemical compound and genetic markers evaluated by clustering (Figures 1, 3), V. gardneriana was isolated from the other three species. This may be related to genotypic differences for this species, as it is found in a different biome (Sena Filho et al., 2008) which may influence the expression of genes used for synthesis of chemical compounds. The use of two data tools (chemical and genetic diversity) more thoroughly supports systematic characterization of the species studied (Rahalia et al., 2016), and validates our chemical marker proposal for the Vitex genus, namely, using the sesquiterpene skeleton with seven member rings as a biomarker candidate.

Interestingly, the chemical structure of the sesquiterpene viridiflorene found in V. megapotamica, V. capitata and V. rufescens possesses a very unique ring system, similar to 6,9-guaiadiene found in V. garnderiana (Figure 4). The enzymatic activity of recombinant Escherichia coli in TPS 31 from Solanum lycopenium led to the discovery of viridiflorine synthase (Bleeker et al., 2011). This compound possesses a similar structure to the africanone scaffold with a 2,6 ring closure (Quin et al., 2014); this type of ring closure could be considered as a more specific signature of the chemoprofile for species in the Vitex genus.

In the toxicity bioassays, the essential oil extracted from V. gardneriana was highly toxic to A. guerreronis, indicating its potential for use in controlling this major coconut pest. On the other hand, it was not possible to estimate the LC₅₀ from V. capitata, V. megapotamica and V. rufescens essential oils. This differential toxicity may be explained by the diversity and amount of sesquiterpenes found only in V. gardneriana, such as 6,9-guaiadine (19%), L-calamenene (13.9%), and caryophyllene oxide (18.6%). However, these compounds should be further evaluated concerning the toxicity to A. guerreronis using more modern toxicology bioassays. Further, predatory mites, especially those belonging to the family Phytoseiidae, are natural enemies of A. guerreronis worldwide (McMurtry and Croft, 1997; Návia et al., 2013). These beneficial mites are naturally found foraging on coconut plants and help to control A. guerreronis. Therefore, future research should also assess the acaricidal activity of V. gardneriana essential oil to predatory mites aiming at determining its selectivity toward non-target organisms, including pollinators and seed dispersers.

Biological activity of some of the prominent compounds we saw in the Vitex species studied here has been defined previously: 6,9-guaiadiene, prominent in V. gardneriana, was also found in essential oils extracted from Bracteosa ruilopezia (Standl.) Cuatrec and Baccharis salicifolia (Ruiz & Pav.) Pers. (Asteraceae)-both oils demonstrated antibacterial activity (Carrizo et al., 2009). It has also been suggested that caryophyllene oxide is toxic; this was evaluated using a direct spray test for females and eggs of the predatory mite Neoseiulus californicus (McGregor) (Momen et al., 2014). Further, α-cadinol and caryophyllene oxide were the main volatile compounds found in essential oil extracted from aerial parts of Teucrium polium L. – Lamiaceae. The essential oil showed fumigant and repellency effects against the cowpea weevil Callosobruchus maculatus F. (Coleoptera: Bruchidae) (Khani and Asghari, 2012; Khani and Heydarian, 2014). α-Cadinol also showed excellent inhibitory action against the termite Coptotermes formosanus Shiraki (Isoptera: Rhinotermitidae) (Cheng et al., 2014).

By combining the volatile profiles and ISSR molecular markers using bioinformatic tools, we were successful in clustering the Vitex species in order to identify markers for chemosystematic classification. In this work, V. gardneriana was isolated from the other species, expanding our understanding of the chemodiversity of this genus. In addition, the volatile components extracted from V. gardneriana showed potential for controlling A. guerreronis, as it was highly toxic to this coconut pest.

Conceived and designed experiments: JdSF, IB, AT, and HX. Conducted experiments: JdSF and IB. Analyzed data: JdSF, IB, AT, PN, DS, provided reagents: ACdS, AT, and JdSF. Wrote the manuscript: JdSF, AR, ASF, AT, DS, and JD.

The authors declare that the research was conducted in the absence of any commercial or financial relationships that could be construed as a potential conflict of interest.