Seasonal and Regional Chemical Variability of the Wild Population of Lantana camara Leaf Essential Oil From Kenya

Studies examining the variability in wild plant metabolic expression propose that environmental factors significantly influence the essential oil (EO) quality and quantity in a plant. Lantana camara is a widely distributed invasive plant species worldwide. However, its immense metabolites can become a source of novel compounds to produce biopesticides in the agricultural industry. Although, the quality aspect has to be considered due to the environmental influence on the metabolites synthesised. Therefore, this research aimed to understand the influence of environmental factors and how it shapes the plant’s metabolite profile in multiple populations of L. camara. Leaf samples were collected from six different geographic regions of Kenya and the corresponding monthly climatic data and soil samples. GC-MS data from leaf EO were analysed with environmental variables (climate and soil data) using unimodally unconstrained and constrained ordination methods for untargeted metabolomics analysis. Partial Least Squares-Discriminant Analysis (PLS-DA) and Random Forests (RF) were used to confirm the variability further. Seasonal and regional variability was observed for secondary metabolites (SMs) in the leaf EO, which correlated to climatic factors and soil attributes. We highlight the season-al-geographic metabolism relationship for L. camara and the combined analytical approach to obtain data that contributes to understanding the influence of environmental factors on the synthesis and accumulation of SMs. This research will have all-embracing implications for maximising phytochemical uniformity.

The variation in secondary metabolites' production influenced by environmental conditions can characterise one species' plant populations. In this context, the metabolites may be used as a chemical marker to differentiate species found in specific geographical zones and seasons (De Souza, Ferri, Fiuza, Borges, & Paula, 2018;Khan et al., 2016;Pereira et al., 2019). Plants that produce essential oils vary considerably in their quality and quantity (composition and concentration of their constituents) due to their interaction with the natural environment.
The vast array of compound synthesis in the L. camara plant points out the adaptive significance for such a diversity of compounds. The variability for the essential oil (EO) composition of L. camara can be related to geographical distribution and pressures (Agil & Hosseinian, 2012;Benites et al., 2009;Javier, Ocampo, Ceballo, & Javier, 2017;Khan et al., 2016;Murugesan, Senthilkumar, Suresh Babu, & Rajasugunasekar, 2016;Pereira et al., 2019;Zoubiri & Baaliouamer, 2012). The metabolism and accumulation of secondary metabolites reflect the integrated influences of multiple ecological factors on the plant during their developmental and growth periods in addition to genetic factors (Liu et al., 2016). Some metabolites are synthesised only under specific environments, or their contents significantly increase under specific environments. Moreover, previous studies have demonstrated that medicinal plants growing in different regions and environments produce different SMs resulting in differences in their qualities (W. Liu, J. Liu, Yin, & Zhao, 2015). Therefore, studies examining the diversity in the production of SMs of L. camara will explain the influence of abiotic and biotic pressures on the EO quality found in a species.
Many studies investigating the influence of environmental factors on plant SM biosynthesis consider these factors' effects on individual compounds. However, individual compounds rarely happen in isolation (Gershenzon, Fontana, Burow, Wittstock, & Degenhardt, 2012). Instead, any compound's influence depends on conditions within the prevailing environment since a single factor cannot be extrapolated from a combination of environmental factors in plants growing in the natural environment (Berini et al., 2018). Thus, understanding how environmental factors will influence a plant's metabolic profile is vital for interpreting how these changes influence the abundance of individual compounds.
Considering the wide distribution and adaptive capacity of L. camara, and the lack of studies covering this subject, as well as the pesticidal properties of this plant, we proposed to carry out a comparative study with samples of EO from leaves of L. camara obtained from samples collected from six different climatic zones of Kenya and collected in different seasons. The approach involves comparing the data obtained by chemical profiles of these oils from six regions. To assess the influence of seasonal and geographical location on the production of SMs and determine the underlying factors responsible for the variations in SMs in L. camara, feasibly suggest the best harvesting seasons and regions for this wild species and promote its reasonable exploitation for specific compounds for biopesticide production.

Sampling Locations
Natural populations of L. camara plant, the first four leaves of the stem from the top ( Figure 1A

Gas Chromatography-Mass Spectrometry (GC-MS) Analysis
The oil samples were diluted in 100% n-hexane and transferred to the auto-sampler vials for GC-MS analysis. Agilent Technology (GC, Agilent 6890) (Agilent Technologies Inc., Santa Clara, CA, USA) gas chromatograph with a split detector and Mass Spectrometer Detector (MSD) coupled with an autosampler was used for this study. Each sample (1 μL) was injected into the GC-MS with an autosampler (Agilent Technologies). The injections of the volatile oil were conducted with a splitless injector at 220 °C. The compounds were then separated on a nonpolar capillary HP column at an average linear flow rate of 35 cm s -1 with helium as the carrier gas. The oven temperature was held at 35 °C for 5 minutes and then increased to 280 °C by 10 °C/min and held for 10 minutes. The collected volatile compounds were then identified by comparing their mass spectra and retention times with the National Institute of Standards and Technology (NIST) 2017 library of mass spectra.

Related Data of Climatic Factors
The environmental data were divided into climate and soil data. The climate data consisted of Monthly average temperature (aT), maximum temperature (maxT) and minimum temperature (minT), average precipitation (P), sunshine duration (SD), and UV index for the collection month were collected from local meteorological stations for the six study sites and were pre-treated in MS Excel © and used for further analysis.

Data Processing and Analysis
The data for the study of seasonal and geographical variation in the composition and accumulation of SMs of the essential oil of L. camara (chemical profiles and environmental data) were divided into two sets of variables: chemical (secondary metabolites (SMs) and environmental (climate and soil data) variables. The data was obtained and used for further analysis after analysing the essential oils by GC-MS (chemical data).
The R software (version 3.6.3; R Core Team, 2020) and the RStudio graphical user interface (version 1.2.5033) were used to perform all the analyses. The total area of peak data was normally distributed (Shapiro-Wilk test: p > 0.05), and their variance was homogeneous (Barlet test: p > 0.05); therefore, we used an unpaired t-test to compare the amount SMs synthesised by L. camara between the rainy and dry seasons. For the same reason, we used the analysis of variance (ANOVA) followed by the Student-Neuman-Keuls (SNK) post hoc test to compare the amount of SM synthesised by L. camara across the different localities during a specific season using the R software package 'Agricolae' (de Mendiburu, 2020). We performed the one-way analysis of similarity (ANOSIM) using the Bray-Curtis dissimilarity matrix to compare the chemical profiles of the different compounds synthesised by L. camara between the seasons and across the different localities. Based on the similarity percentage (SIMPER) analysis, we identified the 10 most influential SMs contributing to L. camara EO's diversity between the seasons and across the different localities. To visualise this difference, we first used the non-metric multidimensional scaling (NMDS) plot, overlaid the physicochemical and environmental variables to the plot to see whether these parameters were related to SM profile diversities.
To better confirm the variation of L. camara EO SM composition between seasons and locations, we used two supervised machine learning algorithms, namely: Random Forest (RF) analysis (Breiman, 2001) and Partial Least Squares-Discriminant Analysis (PLS-DA) (Liland & Indahl, 2009). Helped by the R package called Random Forest (Liaw & Wiener, 2002), we ran the RF analysis using 10000 iterations (ntree) with 12 SM randomly selected at each split (mtry = √q, where q is the total number of SM (150)). Based on the function "importance ()", we generated the mean decrease in accuracy (MDA) for each compound selected. The compound with the highest MDA value was considered the most significant for L. camara EO's diversity between the seasons and regions. To visualise these differences, we generated the multidimensional scaling (MDS) ordination plot using the function "MDSplot ()" based on the proximity matrix from the RF analysis. Using the function "PLS-DA ()" embedded in the R package called mixOmics (Rohart, Gautier, Singh, & Kim-Anh, 2017), we performed the PLS-DA analysis. We visualised the difference using the function "plotIndiv ()". With this technic, we identified the most significant compounds for differentiating L. camara EO using the function "PLS-DA.VIP ()" found in RVAideMemoire R package (Maxime, 2020). All statistical results were considered significant when P < 0.05.

Seasonal Variation in Essential Oil Composition of L. camara
The GC-MS chromatograms from L. camara essential oil showed that the plant synthesised many Secondary Metabolites (SMs) for both rainy and dry seasons (Figure 2). The number of SMs synthesised by L. camara varied significantly between the seasons, except for Kiboko (p = 0.68; Figure 2C) and Njoro (p = 0.64; Figure  1F). The number of compounds was lower in Mtwapa ( Figure 2E) but higher in Embu (Figure 2A), Kandara ( Figure 1B) and Kakamega ( Figure 2D) during the rainy season. In comparison, there was a tremendous increase in the number of compounds observed in the plants sampled from Embu ( Figure 2A) and Mtwapa ( Figure 2E) and a reduction in Kandara ( Figure 2B) and Kakamega ( Figure 2D) during the dry season. Independently to the site, when we ran the analysis of similarity based on the Bray-Curtis distance matrix, we consistently found a significant difference of L. camara EO SM composition between rainy and dry season (ANOSIM: p < 0.0001, R = 0.465). This difference was depicted by the Non-metric multidimensional Scaling (NMDS) plot ( Figure 3A), with an excellent dissimilarity representation ( Figure 3B; Stress: 0.199). This plot distinguished L. camara EO obtained during the rainy season from those obtained during the dry season. The similarity percentage (SIMPER) analysis identified trans-cadina-1(6),4-diene, (E-)-Caryophyllene, and 1,8-cineole as the three significant compounds responsible for the distinction of L. camara essential oil between the rainy and dry season ( Figure  3C). jas.ccsenet.

Regional Variation of L. camara EO Secondary Metabolites (SMs) During Dry Season
We also found qualitative and quantitative variations in L. camara EO across the locations during the dry season. The total SMs synthesised by L. camara significantly varied across the different localities ( Figure 5A). L. camara plants sampled from Embu synthesised significantly more SMs than those sampled from Mtwapa, Njoro, Kandara, Kiboko, and Kakamega. According to their origin, there was a significant difference in L. camara EO SM composition as determined by the ANOSIM test and the NMDS plot ( Figure 5B). Based on the SIMPER analysis, we identified (E)-Caryophyllene, δ-3-Carene, and 1,8-Cineole as the most critical compounds contributing to L. camara differentiation in the dry season across the different ( Figure 5D). When we associated the physiochemical and the environmental parameters of each site to the NMDS plot ( Figure 5E), we found that the synthesis of SMs by L. camara varied from region to region and was primarily related to sunshine, rainfall, humidity, temperature, soil pH, nitrogen, and phosphorous ( Figure 5E).

Confirmation of L. camara SMs Profile Changes in the Function of Seasons and Locations Based on Supervised Machine Learning Algorithms
Results from Random Forest (RF) analysis and Partial Least Squares-Discriminant Analysis (PLS-DA) consistently confirmed the change of L. camara EO SMs in the function of seasons and localities. When we ran the multidimensional scaling analysis based on the RF analysis's proximity matrix, this technic differentiated L. camara oil SM based on their seasons ( Figure 5Ai) and location (Figure 6Aii, 6Aiii) of collections. Based on the mean decrease in accuracy (MDA), this analysis identified Camphor (with a classification accuracy of 83.33%) as the most influential SM for differentiating L. camara oil between rainy and dry season (Figure 6Bi). Similarly, this analysis respectively identified 14-hydroxy-(Z)-Caryophyllene (with a classification accuracy of 65%) and Eugenol (with a classification accuracy of 80%) as the most significant SM for distinguishing L. camara oil across the different localities during rainy ( Figure 6Bii) and dry (Figure 6Biii) seasons.
Also, the PLS-DA score plots separated L. camara EO in the function of seasons ( Figure 7Ai) and locations (Figure 7Aii, iii). As with the RF analysis, the separation of L. camara SM synthesised during the rainy and dry seasons identified Camphor as the most influential compound (Figure 7Bi). The nPLS-DA analysis identified 2-Cyclopenten-1-one, 3-methyl-2-(1,3-pentadienyl)-,(E, Z)-as the most significant SMs responsible for the differentiation of L. camara EO across the localities during rainy (Figure 7Bii) and dry (Figure 7Biii) Vol. 13, No. 11; Striking variation in the EO of L. camara leaf metabolic profile was observed from the samples collected from six different climatic zones in Kenya, dry and rainy seasons. The variability between the L. camara plant populations was compared using the ANOSIM test based on differences in the abundance and composition of multiple metabolites. Seasonal specificity is observed in this study, with variability in the number of SMs observed for both dry and rainy season in each region. Kakamega, Kiboko, and Kandara showed an increased metabolites number compared to the other regions in the rainy season. This increase in metabolic variation can be attributed to the plant's interaction with the prevailing condition at that particular season. These regions received adequate rainfall and maximum sunshine hours and UV index compared to the other regions where sunshine was reduced during the rainy seasons. The results agree with other researchers that the metabolite synthesis of essential oil-producing plants is strongly influenced by the slightest changes and how they interact with the natural environment without detrimental effects leading to newly formed compounds that could be advantageous during the prevailing conditions.
Ten dominant variable compounds contributed to shaping the seasonal variation of the L. camara plant populations. For the samples of L. camara evaluated in this work, we propose two chemotypes according to the season: a chemotype for the samples collected in the dry season (chemotype I) and another for the rainy season (chemotype II). Based on the NMDS test results and SIMPER analysis, the chemotype I is discriminated by the higher proportion of (E-)-caryophyllene, 1,8-cineole, spathulenol bicyclogermacrene, α-humulene, caryophyllene (II), and δ-3-carene in the leaf EO. The chemotype II is discriminated by the higher proportion of trans-cadina-1(6),4-diene, α-muurolene, and Iso-sylvestrene, as the dominant compounds. Overall, three significant compounds, trans-cadina-1(6),4-diene, (E-)-Caryophyllene, and 1,8-cineole, contributed significantly to the seasonal variability in the SM profile of the L. camara EO.
The environmental factors are the mechanism involved in influencing the accumulation and biosynthesis of L. camara SMs, and it is related to seasonal induction stress factors in the plant, such as changes in temperatures, humidity, rainfall and soil conditions (Guo et al., 2013;Pavarini et al., 2012;Ramakrishna & Ravishankar, 2011;Selmar & Kleinwächter, 2013;Yang et al., 2018). For example, plants collected during the dry season may have encountered high temperatures influencing an increase in (E-)-caryophyllene production, while low temperatures in the rainy season influenced an increase in trans-cadina-1(6),4-diene. This result agrees with De Almeida et al.
(2016), who reported an increase in (E-)-caryophyllene in the EO composition of Copaifera langsdorffii during the dry season as compared to the wet season. There is no official report in the literature regarding the influence of seasonal change on trans-cadina-1(6),4-diene accumulation. This variability can be attributed to the plants' relationship with the environmental conditions during plant growth and productivity at that particular season, changing the biosynthetic pathway of secondary metabolites toward metabolites' production. Therefore, different prevailing environmental pressures between the two seasons played a significant role in the diversity of the SMs of L. camara leaf EO. Our results corroborate with Nea et al. (2020), Pereira et al. (2019) and Dos Santos et al. (2019) reported fluctuation patterns of SMs produced by L. camara to correlate to seasonal changes. Collectively these results revealed that the seasonal variation predominantly influences the SM profile composition in wild plant populations. Therefore, this work provides new insight into understanding the response of L. camara plant populations SMs biosynthesis to the seasonal variation.
Our precedent results revealed seasonal variability in the synthesis of SM in the L. camara wild population plants.
We further aimed to determine whether this variability is regional and what environmental factors drive the variation. This study observed significant variability in L. camara EO's chemical profile across the different regions where samples were collected during the dry and rainy season as determined by the ANOSIM test and NMDS plot. Three significant compounds trans-cadina-1(6),4-diene, caryophyllene (II) and lavandulyl isovalerate were identified that contribute significantly to the regional variability in the rainy season, while (E-)-Caryophyllene, δ-3-carene and 1,8-cineole contributed to the differentiation during the dry season across the regions. These results are consistent with previous reports (Murugesan et al., 2016;Pereira et al., 2019;Sena Filho et al., 2012). Recently, Nea et al. (2020), and Pereira et al. (2019) confirmed high variability in SM composition of the EO of L. camara from Bregbo South of Côte d'Ivoire and samples collected from 21 municipalities representing three regions in Brazil respectively. Geographical location is a crucial factor that affects plant growth's prevailing conditions, having significant effects on secondary metabolic processes in a plant species (Liu et al., 2015)-Consequently, the cause of the relationship between the SM composition and contents in their growing locations.
The number of synthesised compounds variability across the regions in both seasons were more or less the same in the region of Njoro and Kiboko with a p-value of 0.64 and 0.68 respectively but differed significantly in Mtwapa, Embu, Kandara, and Kakamega with a p-value of < 0.0001, < 0.01, 0.017 and 0.002 respectively. Overall, Mtwapa jas.ccsenet.org Vol. 13, No. 11; showed high variability in the number of compounds synthesised as compared to the other regions. The number of compounds synthesised was very high during the dry season as compared to the rainy season. The significant difference in SM content is climatic factors of temperature, rainfall, UV index, and pH property. These changes in the environmental conditions may explain the differentiation of SMs in the Mtwapa region, located in the coastal areas and dominated with a more considerable climatic seasonal variability, particularly temperature, compared to Njoro and Kiboko that had a reasonably stable climatic condition. Temperature change is known to substantially affect SM synthesis since areas with more considerable climatic changes are faced with more variation and could lead to more significant variability in their SM profile (Allevato et al., 2019). Furthermore, Molina-Montenegro and Naya (Molina-Montenegro & Naya, 2012) argue that locations with slight seasonal variations and constant warm temperatures lead to a low environmental plasticity capacity. Subsequently, stability in the environment would reduce the plant's overall pressures, therefore reducing metabolite variation.

Journal of Agricultural Science
The significant differences in climatic conditions and soil characteristics among the six regions and other explanations are the determinant factors for the variability in SM profile observed in L. camara EO of the same species growing in Kenya's diverse regions. It is observed that L. camara wild plants population responds differently to environmental variations, therefore variability in the production and accumulation of SMs. Ncube et al. (2012) explain that these variabilities are due to the physiological characteristics associated with genetic conditions, which arise probably with the prevailing climate in both seasons in the regions under study. Plant metabolism is influenced in many ways by those conditions. Factors such as temperature, humidity, rainfall, sunshine duration directly respond to these variations. These factors jointly influence the biosynthesis and accumulation of SMs and correlate with each other and do not act in isolation (Gobbo-Neto & Lopes, 2007).
Among the environmental conditions, soil characteristics represent a complicated biological system that strongly influences the plant's ability to produce SMs (Muscolo et al., 2019;Ramakrishna & Ravishankar, 2011 Yadegari, 2015). Correctly, soil characteristics play a crucial role in diverse soil conditions that causes significant differences in biosynthesis and SM accumulation in plants of the same species. The SM profile variation response of L. camara to the soils with diverse characteristics was in agreement with the findings of Ormeño and Fernandez (2012), and Muscolo et al. (2019), showing that soil, with its intrinsic characteristics, is directly responsible for plant metabolite production.
This study established a relationship between soil properties and metabolic profile to be regionally specific.
Extrapolating these results at the regional and seasonal scale suggests that any alteration of the soil properties leads to changes in SMs accumulation in the L. camara plant population, affecting its quality and quantity. The plant material analysed in this study was collected from plants growing under diverse natural conditions. Therefore, it was not easy to separate the effects of individual factors from the environment's multifactorial influence (Climatic and soil variables). Therefore, we conclude that different climatic factors (rainfall, temperature, UV, humidity) have different effects and intensities on the accumulation of SMs in the EO of L. camara. At the same time, different SMs are affected by different kinds of soil properties in the soil to a different extent, where the physicochemical properties including N, P, K, TOC, EC and pH in the soil all have a relatively significant strong effect on plant secondary metabolism. Thus the correlation differs significantly with SM composition from region to region.
Besides, further identifying the compounds through PLS-DA and RF confirmed SMs variability as a season and geographical location. The results indicated that SMs composition and contents were near related to the growing locations of L. camara, and environmental factors influenced the production of the SMs. The SMs synthesis and accumulation in medicinal plants are complex processes affected by many environmental factors comprising a multivariate system. The variability in chemical profile was expected because the soil factors were significantly different due to the different growing regions and the climatic factors changed with geographical conditions. Local adaptation depends on both genetic, soil properties and environmental factors; Thus, the high chemical variability reflects the need for the plant to adapt to such different environments (Ncube et al., 2012;Pereira et al., 2019;Sampaio & Da Costa, 2018;Sampaio et al., 2016).
This difference observed could lead to a more targeted analysis and understanding of biological pathways. Although determining the plasticity of a species is difficult without a conventional garden experiment. Our study used only ordination methods to associate climatic and soil variables with variation in chemical profiles of L. camara, and correlation does not imply a relationship. Many laboratory and greenhouse experiments are needed to confirm the effect of climatic and soil variables on the SM profile. Greenhouse experiments with the different species in the exact location could help detect whether chemical variation differences are due to environmental, soil, or genetic differences. The result would be beneficial for the production of EO with unique active ingredients. One could choose the location of plant growth for a particular compound, or one could add/omit nutrients to modify the EO quality and quantity. Therefore this exploratory analysis in wild populations of L. camara is advantageous and essential as it has reduced the environmental and soil variables and will allow for a more guided experimental analysis such as greenhouse experiments.

Conclusions
This study provides information on seasonal and regional variations of the quality and quantity of SMs of L. camara EO from the leaves. The results showed that environmental conditions in the drier season favour the production of more dominant compounds than in the rainy season. The optimum time to achieve the highest quantity of (E-) caryophyllene is during the drier period, while the trans-cadina-1(6), 4-diene is more abundant in rainy seasons. We can highlight the relationship between the environment and the metabolic profile of L. camara, and the variation in SMs is a direct response to fluctuations in conditions in the surrounding environment. Therefore, the knowledge gathered from this study on the influence of seasonal and regional variation on L. camara EO composition can help decide the best period to harvest the plant according to the desired compound for exploitation in the agricultural industry. The analysis of wild populations of L. camara has provided us with potential environmental variables that require more follow up with greenhouse experiments to determine their importance in SM biosynthesis.