 Figure 3: Multivariate projection of the fatty acids in the plane of two greatest discriminants for opium samples collected
from licit opium growing divisions of India. 1. Barabanki, 2. Bareilly, 3. Bhilwara, 4. Chittorgarh, 5. Faizabad, 6. Garoth,
7. Jaora, 8. Jhalwar, 9. Kota, 10. Mandsaur, 11. Neemuch, 12. Pratapgarh, 13. Tilhar and 14. Ujjain.
| The opium samples can be classified into three groups based on the fatty acid quantitative data (Figure 3). The samples from
Mandsaur (14/18), Neemuch (8/12), Ujjain (4/7), Pratapgarh (4/6), Chittorgarh (5/15), Tilhar (1/8), Garoth (1/3), Jaora (1/3),
Kota (3/17), Jhalwar (3/9) and Bhilwara (4/8) are falling within one group (group I). The samples from Barabanki (4/6), Chittorgarh
(9/15), Faizabad (3/6), Bareilly (4/6), Tilhar (3/8), Mandsaur (2/18), Neemuch (2/12), Garoth (1/3), Jaora (2/3) and Jhalwar
(2/9) are falling within the second group (group II). Finally, the samples from Kota (14/17), Jhalwar (3/9), Bhilwara (3/8)
and Bareilly (1/6) constitute the third group (group III). The remaining samples (23/124) were found to be outside the classified
groups and as well outside the two-dimensional projection.
However, some of the samples (Bhilwara, Jhalwar and Kota) of group III were misclassified in group I. Similarly, some of the
samples (Tilhar, Chittorgarh) of group II were misclassified in group I and some of the samples (Mandsaur and Neemuch) of
group I were misclassified in group II. The misclassification of samples yielded no significant discrimination between the
divisions based on the quantitative fatty acid data. For example, opium samples from Garoth and Jhalwar were found in all
three groups making it difficult to discriminate one division from the other by means of the fatty acids as a biochemical
marker. The same was also observed with other divisions. Based on this discriminant analysis, the results have found to have
a 50% predictive value in relation to the source of the opium samples. A variable reduction process was adopted in order to pinpoint the fatty acid variables that can give a significant discrimination
between the licit opium-growing divisions. The mean and standard deviation of the fatty acid variables were calculated between
the groups as well as within each group. Variables that provide the largest difference in mean values, but having smallest
standard deviation within one group were selected. Variables that have the same or very small difference among the mean values
were ignored. The predictive value obtained after this process of variable reduction was found to be less than 50% compared
with the predictive value obtained before the variable reduction process. By the above study it is clearly evident that the
quantitative fatty acid data is not useful for source identification of Indian opium. Conclusion Qualitative analysis of fatty acids such as behenic, stearic and lignoceric acids in Indian opium are valuable biochemical
markers for the discrimination of licit opium growing divisions of India. Conversely, quantitative fatty acid data yielded
no significant results for the source identification. The methodology developed may find wide application for the determination
of fatty acids in opium originating from different licit opium growing countries.
|