Multifactorial discriminant analysis of leaf oil of C . odorata L . King and Robinson from Côte d ' Ivoire Esse

Laboratoire de Chimie Organique Biologique, UFR-SSMT, Université ́ Félix Houphouët-Boigny, BPV 34 Abidjan, Côte d’Ivoire. Université de Corse-CNRS, UMR 6134 SPE, Equipe Chimie ET Biomasse Route des Sanguinaires, 20000 Ajaccio, France. Laboratoire des Mathématiques et des Nouvelles Technologies de l’Information, Institut National Polytechnique Félix Houphouët Boigny BP 1093 Yamoussoukro, Côte d’Ivoire.


INTRODUCTION
The family Asteraceae is a very large cosmopolitan family.It is represented by 13 tribes, 84 genera and over 240 species (Adedeji and Jewoola, 2008;Walter, 1979).The family is highly advanced, easily recognized and with worldwide distribution.The members of the family are largely woody herbs or shrubs, a few are trees and climbing herbs (Adedeji andJewoola, 2008: Olorode, 1984).Many plants in the family Asteraceae are economically important as weeds, ornamentals, medicinals and green vegetables (Adedeji and Jewoola, 2008).Important weed species in this family include Chromolaena (C.) odorata Linn.
The biological activity of extracts of C. odorata has been shown by several studies in the world.Indeed, the extracts of fresh leaves of C. odorata have been used in the treatment of malaria in Ghana and Benin (Inya-Agha et al., 1987;Bedi et al., 2010).Aqueous extracts of this plant presented an anti-microbial activity against gonococcus in Guatemala (Caceres et al., 1995).
Extracts of C. odorata are used in folk medicine of Côte d'Ivoire as cataplasms to stop hemorrhages or as antiinflammatory drugs against pains (Bedi et al., 2001).
In continuation of our work on the characterization of C. odorata leaf oil growing wild in Côte d'Ivoire, due to the chemical polymorphism of this oil, four factors, notably, chemical composition, physicochemical constants, yields and geographical coordinates have been selected to improve the description of the leaf oil.Indeed, the aim of this study was to use, for the first time, the factorial discriminant analysis to observe homogeneity or to evidence an eventual chemical variability among the samples due to location effect of C. odorata from "Region du Belier" of Côte d'Ivoire.

Essential-oil isolation
Clevenger-type apparatus was used for 3 h.Essential-oil physicochemical constants and yields were estimated according to AFNOR (Association Française de NORmalisation, 1982).

GC-FID analysis
The oil samples were analyzed with a Perkin Elmer Clarus 500 apparatus equipped with FID and two fused-silica cap.columns (50 m × 0.22 mm i.d.film thickness 0.25 mm), an apolar BP-1 (polydimethylsiloxane) and a polar BP-20 (polyethylene glycol) column.
The oven temperature was programmed rising from 60 to 220° at 2°/min and then held isothermal at 220° for 20 min; injector temp., 250°; detector temp., 250°; carrier gas, He (0.8 ml/min); split ratio, 1/60.retention indices (RIs) were determined relative to the t R of a series of n-alkanes with linear interpolation using the software Target Compounds from Perkin Elmer.

C-NMR analysis
The 13 C-NMR spectra were recorded with a Bruker AVANCE 400 Fourier Transform spectrometer operating at 100.63 MHz and equipped with a 5-mm probe, in CDCl 3 , with all shifts referred to internal Me 4 Si.The 13 C-NMR spectra of the oil samples were recorded with the following parameters: Pulse width, 4 ms (flip angle 45°); acquisition time, 2.7 s for 128 K data table with a spectral width of 25000 Hz (250 ppm); CPD (Composite Pulse Decoupling) mode decoupling; digital resolution, 0.183 Hz/pt.The number of accumulated scans was 2000 to 3000 for each sample, depending on the available amount of oil (when available, 45 -50 mg of essential oil in 0.5 ml of CDCl 3 ).

Identification of components
The identification of the individual components was based on the comparison of the GC retention indices (RIs) for the polar and apolar columns, determined rel. to the Retention Time (t R ) of a series of n-alkanes with linear interpolation, with those of reference compounds.And for investigated samples, on the comparison of chemical shift values in the 13 C-NMR spectra of the essential oils with those of reference spectra compiled in a laboratory-built library, following a computerized method developed in our laboratories, using home-made software.In the investigated samples, individual components were identified by 13 C-NMR at contents as low as 0.3 to 0.4%.

Quantification of components
The relative contents of the oil constituents were expressed as percentage obtained by peak-area normalization without using correction factors.

Essential oil physicochemical constants
The leaves of C. odorata growing wild in "Region du Belier" (center) of Côte d'Ivoire were collected during three years in the rainy season, and the essential oils were isolated by hydrodistillation.
Observing Table 1, the oil yields calculated on the fresh weight basis were in the range of 0.05 to 0.35% (w/w).The oil optical rotation were in the range of -1.6°-(-0.6°).The oil refractive index was in the range of 1.499 to 1.509.The oil density was in the range of 0.725 to 0.995.The oil acid index were in the range of 0.24 to 1.8 (Table 1).
These essential oils are levorotatory, lighter than water and less acidic (AFNOR, 1982).These mean values are compared with those found in the literature on C. odorata, those of Benin, recorded in the Table 2, for all designated variables (Noudogbessi et al., 2006).The observation of this table shows similarities but also differences between the values from Côte d'Ivoire and those from Benin.The mean values of the yield, the refractive index and density are similar, and the two medium oils are levorotatory (Table 2).

Essential oil composition
All the 71 samples were submitted to GC-FID analysis to determine retention indices (RIs) on two columns of different polarity.

Statistical analysis
The 71 oil compositions were submitted to statistical analysis, that is, factorial discriminant analysis (FDA; Figure 2).The essential oil samples were labeled according to the eight sites of harvest leaves: Toumodi, Dougba, Yamoussoukro, Zambakro, Toumbokro, Attiegbakro, Tiebissou and Tie-N'diekro (Figure 1).FDA of concentration data from 40 variables (4 physicochemical constants, yield, 31 chemical compounds and 4 geographic coordinates) showed differences of the constituent proportions according to the geographic origin.
Therefore, the mean content and the standard deviation of the major components were calculated for groups Toumodi, Dougba, Yamoussoukro, Zambakro, Toumbokro, Attiegbakro, Tiebissou and Tie-N'diekro.

Highlighting the discriminating power of the effect of harvest site leaves
The statistical treatment of data on Figure 2, has highlighted the discriminating power of the site effect; discriminating capacity, which existed statistically when samples were described by the chemical constituents (31 variables).When the chemical constituents have been associated with physicochemical constants and yield (36 variables), the discriminant power was visually reinforced.It has been total with the addition of the geographical coordinates (Figure 2).The use of geographic coordinates associated with chemical descriptors created the contraction of each group, to the point of reducing to    a point.This helped raise the apparent confusion that existed between these groups.The eight defined groups were so reconstituted.They are also different and each define a chemical variant: the chemical variability due to site effect.

Chemical composition of the different classes
Table 3 shows the mean chemical composition of each of the eight chemical variants.Their contents are given as mean±standard deviation.

DISCUSSION
C. odorata essential oil samples analyzed by gas chromatography (GC) and gas chromatography -mass spectrometry (GC/MS), published in the literature have shown the existence of a chemical variability.This first study using the factorial discriminant analysis, dealing with the combination data of 40 variables (4 physicochemical constants, yield, 31 chemical compounds and 4 geographic coordinates), led to observe the homogeneity or to evidence an eventual chemical variability among the samples due to location effect of C. odorata from "Region du Belier" of Côte d'Ivoire.
The additional use of geographical coordinates, as a descriptor of the essential oil, with the ability to reduce each of the eight groups at a single point is proof of the importance and supremacies knowledge of the origin of the oil sample.However, the mean values of physicochemical constants and yields comparing with those of Benin (Noudogbessi et al., 2006), according to optical rotation and acid index, essential oils from Benin seemed richer in volatile substances and more acidic than those from Côte d'Ivoire (AFNOR, 1982).The differences in the values of the physicochemical constants could be explained by the presence of isomers α-and β-pinenes as major constituents in the essential oil of Benin.
These results are broadly in agreement with those obtained by Smadja (1990) in a similar study on vetyver bourbon (Smadja et al., 1990).The leaf oil of C. odorata from Côte d'Ivoire exhibited a chemical variability with three composition patterns dominated either by geijerene,

Conclusion
Taking account the combination of chemical constituents, physicochemical constants, yield essential oils and geographical coordinates of the harvest sites of C. odorata leaves, the treatment of data by AFD showed clearly the discrimination from different sites in "Region du Belier" of Côte d'Ivoire.Each group became more homogeneous and better separated from other despite of certain similarity of major components.This has led to a chemical polymorphism according to eight leaves harvest sites: Toumodi, Dougba, Yamoussoukro, Zambakro, Toumbokro, Attiegbakro, Tiebissou and Tie-N'diekro.
The chemical variability found in our samples seemed to be linked to exogenous factors; however, the investigations taking account the geographic factors in the selected localities could confirm clearly the relationship between the samples and their correlation with the main constituents.

a)
Order of elution and contents determined on the apolar and polar column (BP-1 and BP-20); b )RI lar : Retention index determined on the apolar column (BP-1); c )RI lar : Retention index determined on the polar column (BP-20); d )Contents are given as mean±standard deviation.

Figure 2 .
Figure 2. Graphical representation of samples of C. odorata oil.Projection onto FDA discriminant axes F1 and F2 for the eight areas using 5, 31, 36 or 40 variables.Graphical highlighting of the discriminating power of the site effect by increasing multidimensional descriptors of essential oil samples.

Figure 3 .
Figure 3. Mean content of major component of the C. odorata leaf oil samples of the eight harvest sites.Error bar: standard-deviation

Table 1 .
Physicochemical constants and yield variability of the leaf essential oils from C. odorata.

Table 2 .
Physicochemical constants and yield variability of the leaf essential oils from C. odorata of Côte d'Ivoire and Benin.
*Data provided from literature.

Table 3 .
Chemical variability of the leaf essential oils isolated from C. odorata of Region du Belier center of Côte d'Ivoire.