Chemometric investigation of light-shade effects on essential oil yield and morphology of Moroccan Myrtus communis L.
Recorded values for the studied parameters
The recorded values for the six studied parameters are summarized in Table 1. The results have indicated that essential oil yield was ranged from 0.48 to 0.93 %
in the shaded site. While, in the exposed site, we have noted a small increase in
essential oil yield which varied between 0.88 and 1.06 %. Altitudes in the shaded
site were between 452 and 470 m while this parameter was between 467 and 485 m in
the exposed site. Otherwise, in shaded site morphological parameters were comprised
between the values 146–200 cm for the individual height, 3.25–4.5 cm for the leaves
length, 1.01–1.58 cm for the leaves width and 189–215 µm for the leaves thickness.
Concerning exposed site, morphological parameters were comprised between the values
67–130 cm for the individual height, 2.81–4.05 cm for the leaves length, 0.65–1.08 cm
for the leaves width and 375–412 µm for the leaves thickness.
Table 1. Essential oil yield, morphological traits, and altitude recorded for twenty studied
individuals
Principal component analysis
Explained variability
To decide about the number of retained components, we have adopted the Kaiser criterion
which says that during a standardized ACP, we must keep components whose eigenvalues
are higher than 1 (Jackson 1991]). The results (Table 2) have shown that the first two components satisfied this criterion and they can be
considered to explain data variability. Table 2 has also displayed the percentages of explained variability by each component and
the cumulated percentages. The first component which explains 72.22 % of the data
variability was mostly contributed by essential oil yield, altitude, height and leaves
thickness, while the second which explains 18.87 % was mostly contributed by leaves
length and leaves width. So, we were satisfied by retaining these two components explaining
91.09 % of the total data variability.
Table 2. Coordinates of parameters on each component, eigenvalues of components, explained
variability by each of them and cumulated percentages of explained variability
Parameters study
The Loading plot (Fig. 2) has revealed the existence of some correlations between studied parameters, in particular,
those between essential oil yield, leaves thickness, and altitude which were positively
correlated and the morphological parameters which were also positively correlated
between them and negatively correlated with the first three parameters. These results
mean that the more the altitude increases the shorter the individuals become and the
less high and wide leaves get. However, they also become richer in essential oils
and their leaves get thicker.
Fig. 2. Loading plot which shows that essential oil yield, leaves thickness, and altitude were positively
correlated between them and negatively with other morphological traits (plants height,
leaves length and width); PC1and PC2 are the first two principal components
Individuals study
The score plot (Fig. 3) has illustrated two groupings of individuals each one contains similar individuals
according to the studied parameters. Consequently, we have observed that the individuals
1–10 (shade plants) formed a group on the right of the first component (group A),
while the individuals 11–20 (sun plants) composed a group on the left (group B).
Fig. 3. Graph of individuals’ distribution (score plot) showing two groups of individuals classified according to the studied parameters.
PC1and PC2 are the first two principal components
Parameters and individuals bi-plot
By superposing the score and loading plots, we have got the bi-plot (Fig. 4). According to this graph, we notice that shade plants were higher, have longer and
wider leaves, but they were thinner and poorer in essential oil. Conversely, sun plants
were shorter, have less long and wide leaves, but they are thicker and richer in essential
oil.
Fig. 4. Bi-plot got by superposing score and loading plots showing the distribution of individuals
and parameters. PC1and PC2 are the first two principal components
Hierarchical cluster analysis
Aiming at a better visualization of the investigated population classification according
to altitude, essential oil yield and morphological parameters, a hierarchical cluster
analysis (HCA) was carried out (Fig. 5). As a confirmation of PCA score plots, the individuals were regrouped in two main
clusters. Cluster I represents shade plants (Individuals from 1 to 10 belonging to
group A in the PCA analysis) and Cluster II represents sun plants (Individuals from
11 to 20 belonging to group B in the PCA analysis).
Fig. 5. Hierarchical cluster analysis of Myrtus communis individual’s based on the studied parameters. The between-group method based on the
Euclidian distance was used; group A: shade plants, group B: sun plants
PCA and HCA results have pushed us to suppose that essential oil yield and morphological
parameters are affected by the difference in environmental conditions such as sun
exposure.
One-way ANOVA test
The results of Table 1 have shown a slight increase in essential oil yield observed in sun plants compared
to shaded ones. Also, for the three morphological parameters height, leaves length,
and leaves width, shade plants were characterized by higher values than those found
within sun ones. However, for the fourth morphological parameters (leaves thickness),
we observed the opposite. To confirm the effect of the sun exposure on the essential
oil yield and morphological parameters, a one-way ANOVA was conducted by considering
the sampling site as a variable (Table 3).
Table 3. Means and standard deviations with the F
calculated
obtained for the one-way ANOVA test carried out on the values of essential oil yield
and morphological parameters by considering the sampling site as a factor
All of the studied parameters have shown a significant change regarding the studied
factor (Table 3). This means that there is a statistically significant change in essential oil yield
and morphological parameters by passing from shaded to exposed sites. These results
confirm those obtained by PCA and HCA showing that the exposure to the sun had a direct
influence on studied parameters by increasing essential oil yield and leave thickness
and reducing plants height, leaves length and leaves width. Opposite results were
confirmed for shaded plants.
Modeling of essential oil yield in terms of morphological parameters
From the precedents results, we have detected several correlations between essential
oil yield and morphological parameters. To get more information about the essential
oil yield in individuals belonging to the studied location, we sought to model it
by a multiple model including morphological parameters as model coefficients. Before
proceeding to the modeling, the values of each parameter were centered. This step
allows the standardization of measurement units.
Variance analysis and model choice
The main effect of regression was statistically significant for the three tested models
(p value 0.05) (Table 4). Obviously, the calculation of F value (F
calculated
) has shown that it was higher than the theoretical F–value (F
theoretical
) for the same degree of freedom at 95 % confidence level. Moreover, the coefficient
of determination R
2
and the adjusted coefficient of determination R
2
a of the third model have the highest values compared with the obtained values for
the two first models (98.8 and 95.44 %, respectively). Therefore, we have chosen this
model containing main, interaction and quadratic terms in the essential oil prediction.
Table 4. F
calculated
, F
theoretical
, p values, coefficients of determination R
2
and adjusted coefficient of determination R
2
a related to the three tested models
The obtained R
2
and R
2
a have given a good agreement between the experimental and the predicted values of
the adapted model. These results were confirmed by those obtained in the graph of
correlation showing a linear curve for the observed values in terms of the predicted
ones (Fig. 6).
Fig. 6. Curve of observed values as a function of predicted ones obtained for the third tested
model containing all of the main terms, the interaction terms, and quadratic terms
Coefficients estimation and mathematical model
Coefficients estimations, statistical t student values, and the observed probability (p value) for each of the coefficients are summarized in Table 5. The results have shown that the factors b
0
, b
1
, b
13
, b
23
have a statistically significant effect (p value 0.05). As a result, these coefficients were included in the fitted model.
Table 5. Estimated regression coefficients for the model containing main, interaction and quadratic
terms
The mathematical model used for modeling the essential oil yield as a function of
morphological parameters is represented as shown in Eq. 1:
(1)
This model can be used to estimate the essential oil yield of the studied area and
can be a useful tool in the choice of individuals to operate.