For this study, the leaves and inflorescences from C. flexuosus var. albescens and C. nardus were harvested at 12 months in an experimental field. Leaves and inflorescences from C. flexuosus var. flexuosus were harvested at 18 months [3] .

A pilot extraction was carried out in a 250 L distiller. The plant material and distillation water, placed in the same reactor (0.5 m × 0.5 m × 1 m), were separated by a removable perforated steel partition and brought to a boil for 3 hours. The condenser was a pipe with an internal diameter of 2 cm, equipped with an expansion chamber, passing through a tank (0.5 m × 0.5 m × 1 m) in which the water flowed in the opposite direction to the distillate. The plant material was dried for 7 days and the following proportions were used: 15 kg of leaves per 100 L of distillation water and 10 kg of inflorescences per 90 L of distillation water. If m1 is the mass of the dried plant material and m2 is the mass of oil recovered, then the essential oil content is equal to (m2/m1) × 100.

GC analysis was performed on an Agilent GC 6890 instrument equipped with a split injector (280˚C), a flame ionization detector (FID) and a DB-5 column (20 m × 0.18 mm × 0.18 µm). The temperature was ramped from 50˚C (3.2 min) to 330˚C at 10˚C/min. Dihydrogen was used as carrier gas at a flow rate of 1 ml/min. The relative concentrations of the compounds were calculated thanks chemstation, an Agilent Instrument software, from the peak area obtained by gas chromatography without using correction factors.

GC/SM analysis was performed on an Agilent GC 7890/Agilent MS 5975 set-up operating in EI mode (70 eV), equipped with a DB-5 column (20 m × 0.18 mm × 0.18 µm). The injector temperature was 280˚C, and the carrier gas was helium at 0.9 mL/min. The temperature was ramped from 50˚C (3.2 min) to 330˚C at 8˚C/min and mass scan mode, m/z: 33 - 450. A mixture of n-alkanes (C8 - C30) was analyzed in the same conditions to calculate retention indices using the equation of [8] . Identification was carried out by comparing retention indices and mass spectra of studied samples with those in data banks [9] - [12] and our own (LEXVA https://www.lexva-analytique.com ).

Means, standard deviations and the usual graphs were obtained with Excel 8.0 software.

Table 1 gives the essential oil content and chemical composition of the studied oils which consist of oxygenated monoterpenes (81% - 88%), monoterpene hydrocarbons (2% - 7%), oxygenated sesquiterpenes (0.5% - 1%) and other compounds (1%).

Leaves gave, with a yield of 0.7% ± 0.1%, an essential oil with 30 identified constituents (content > 0.1%). The first 10 major constituents of the citronella chemotype studied were: geraniol (29.0% ± 3.1%), citronellal (25.6% ± 1.4%), citronellol (9.7% ± 0.9%), geranial (8.8% ± 0.4%), neral (7.1% ± 0.3%), geranyl acetate (2.7% ± 0.3)%), citronellyl acetate (2.3 ± 1.0), limonene (3.5% ± 0.6%), z-β-ocimene (2.2% ± 1.6%) β-caryophyllene (2.0% ± 0.2%), and made up 80% of the total oil. This list of major constituents is similar to that drawn up by [2] , after compiling data from the literature for the main representative of the citronella chemotype (from C. nardus).

SD: standard deviation; RIc: retention index calculated; RILit: retention index literature.

The major constituent profiles of the two parts of C. flexuosus var. albascens were very similar: like leaf essential oil, the inflorescence essential oil contained 6-methyl-heptan-5-en-2-one (1.4%), limonene (2.0%), linalool (1.0%), citronellal (8.3%), citronellol (6.7%), citral (neral: 8.7%. geranial: 13.4%), geraniol (33.0%), geranyl acetate (7.5%), and β-elemene (4.5%), but the individual levels of these constituents varied between leaves and inflorescences: (i) geraniol (29.0% leaves, 33.0% inflorescences); (ii) citronellal (25.6% leaves, 8.3% inflorescences; (iii) citral (15.9% leaves, 22.1% inflorescences and (iv) citronellol (9.7% leaves, 6.7% inflorescences). These results can be compared with those in Table 2 , which were obtained with the citral chemotype of Cymbopogon flexuosus acclimatized in Congo and were consistent with our previous results [3] .

The essential oils of the inflorescences and leaves of C. flexuosus var. flexuosus were very similar: same qualitative profile, with fewer components for inflorescences, and a narrow range of variation in the individual contents of these constituents ( Table 2 ). This major constituent profile is also consistent with those of C. flexuosus (citral chemotype), growing in India [7] , with: citral (geranial; 40.0% - 50.0%, neral; 30.0% - 35.0%), borneol (0.0% - 2.0%), citronellal (0.37% - 8.04%), citronellol (0.4% - 4.6%), citronellyl acetate (1.2% - 3.6%), geraniol (1.7% - 40.0%), geranyl acetate (2.0% - 5.1%), limonene (2.4% - 3.7%), methyl eugenol (20.0%), and myrcene (0.1% - 14.2%); however, the essential oil contents of these inflorescences were slightly lower.

This major Constituents profile is also consistent with those of C. flexouosus (citral chemotype) growing in India [7] with: citral (geranial; 40.0% - 50.0%, neral; 30.0% - 35.0%), borneol (0.0% - 2.0%), citronellal (0.37% - 8.04%), citronellol (0.4% - 4.6%), citronellyl acetate (1.2% - 3.6%), geraniol (1.7% - 40.0%) geramyl acetate (2.0% - 5.1%), limonene (2.4% - 3.7%) and methyl eugenol (20.0%) myrcene (0.1% - 14.2%).

Studied oils were characterized by a very high level of oxygenated monoterpenes (87% - 95%) and very low level of other constituents: monoterpene hydrocarbons (1% - 2%), oxygenated sesquiterpenes (0.2% - 1.1%) and other compounds (0.5% - 1.0%).

The leaf and inflorescence essential oils of Cymbopogon nardus L. Rendle

RI: retention index; NI: non-identified.

(reference of citronella chemotype) were examined in the same conditions as C. flexuosus (Nees ex Steud.) Wats, to assess the consistency of results with those of this reference ( Table 3 ). Studied oils contained oxygenated monoterpenes (65.5%

NI: non-identified; *superimposition of peaks with approx. 5% iso-isopulegol.

- 75.3%), oxygenated sesquiterpenes (18.5% - 22.0%), monoterpene hydrocarbons (1% - 2%) and other compounds (0.5% - 1.0%).

Table 3 lists 48 constituents of leaf or inflorescence of C. nardus essential oil (individual content > 0.1%). Sixteen components represent more than 80% of the oil composition: limonene, linalool, isopulegol, citronellal, isopulegol, citronellol, citral, geraniol, eugeniol, citronellyl acetate, geranyl acetate, β-elemene, germacrene-D, δ-cadinene, elemol, and α-cadinol. This major constituent profile, consistent with our previous results [5] , is also consistent with the following list compiled by [2] based on literature data: limonene, linalool, isopulegol, citronellol, citronellal, citral, geraniol, eugeniol, citronellyl acetate, geranyl acetate, and elemol.

According to the literature, “true” lemongrass (Indian C. flexuosus, source of citral) presents morphological and chemical variability. The citronella chemotype, which refers to C. nardus, was observed in Southeast Asia in the plantations of C. flexuosus mainly set up for the citral chemotype [2] [7] .

To highlight similarities and differences among these three essential oils, radar plots were used ( Table 4 , Figure 1 and Figure 2 ).

Close examination of representative radar plots of these oils built with 14 main constituents shows that: 1) Figure 1 and Figure 2 are totally different, with a very

high citral content (neral: 34.4%; geranial: 50.6%) for 2, 2) the two radar plots in Figure 1 are qualitatively similar for the major constituents, 3) C. nardus differs from C. flexuosus only by its citral content; C. flexuosus var. albascens contains

four times more citral (22.1%) than C. nardus (5.0%), which could be an marker for distinguishing between the two oils, 4) the “other” constituents, not considered for the radar plots are almost 10 times more abundant in C. nardus than in C. flexuosus, and 5) the difference between Figure 1 and Figure 2 illustrates the difference in composition between the two chemotypes of C. flexuosus (citronella and lemongrass).

The essential oils of the inflorescences of C. flexuosus var. albascens and C. nardus present the same chemotype according to their major constituent profiles; however, the citral content and the complexity of the complexity of the chemical profile are markers that could if necessary be used to distinguish between their two oils.