• Tidak ada hasil yang ditemukan

That preparations derived from ethnomedicinal plants often show positive

pharmacological activities is widely acknowledged (Farnsworth

et al.,

1985; Fourie

et al.,

1992; Cox, 1994; Taylor

et al.,

2001). Considering the diversity of the flora

(approximately 24000 taxa in some 368 plant families)(Germishuizen and Meyer, 2003) and ethnomedicinal flora (3434 taxa from 1187 genera and 206 families)(Arnold

et al.,

2002) in southern Africa, the potential that the region holds in regard to efficacious medicinal plants is considerable. However, even with the recent advances in biodirected assays (Lewis, 2003), screening all regional taxa against ailments such as malaria is currently impractical. When considered in the light of the costs of bioprospecting, the necessity for techniques that prioritise taxa - particularly ethnomedicinal taxa

(Farnsworth, 1990) becomes clear.

The preferential use of ethnomedicinal data in drug discovery programmes has several potential spin-offs, such as providing short-term and long-term benefits to the

ethnomedicinal knowledge-holders, communities, host countries, and participating institutions (Soejarto

et al.,

2002a). Furthermore, natural products generally have a great potential for use in their native state, Le. with little or no structural modification, which can reduce the costs associated with re-engineering structures of complex compounds (Garrity and Hunter-Cevera, 1999). The methods described in this chapter therefore warrant attention from bioprospectors involved in similar projects.

One of the key reasons for incorporating regression analyses in the assessment of antimalarial data is the need to understand if plant selection by ethnomedicinal practitioners in the region is in any way related to current taxonomic constructs. If selection of ethnomedicinal plants is based purely on the placebo effect (Adler and Hammett, 1973) then the hypothesis that such selection is random (Moerman, 1979) would be true. The falsification of the hypothesis, as shown here, indicates a higher than predicted ethnomedicinal use of certain families; plants in such families are likely to be

more efficacious (Moerman and Estabrook, 2003), most likely due to the presence of certain secondary metabolites (Saxena

et al.,

2003). If so, then the recently validated status of plant families as indicators of evolutionary relationships adds strong support to the argument that closely related plants produce similar chemical compounds (Grayer

et

al., 1999; Moerman et al., 2003). While it is not the aim of this chapter to expound on the philosophy of plant selection by ethnomedicinal practitioners, the incorporation of such knowledge in the development of bioprospecting methods is seen as essential. This is particularly the case where ethnomedicinal preferences can be correlated with

phylogenetic perspectives.

3.4.1 Keyword associations

The keyword system of identifying plants with potentially efficacious anti-malarial extracts is analogous to the use of object attributes in data mining (Westphal and Blaxton, 1998). Keywords are attributes ascribed to certain objects (plant taxa) and as such can be used to identify those objects from the data source. Keywords considered synonymous and analogous with malaria and fever (Table 3.1) were incorporated into text searches on the obvious grounds that recorded ethnomedicinal literature reflects the knowledge of ethnomedicinal practitioners. Of presumed lesser importance is the

likelihood that plants may have been used in any number of ways, either singly or in combination with others. It is also unlikely that ethnomedicinal practitioners are familiar with the epidemiological basis of many diseases, including malaria (Randrianarivelojosia et al., 2003), and it is therefore unlikely that ethnomedicinal literature reflects bona fide links between the symptoms of the malaria (characterised by cyclic bouts of fever) and the Plasmodium pathogen. The assumption is that the occurrence of MAFEV keywords implies that the relative plants are efficacious in the treatment of MAFEV conditions.

It was unsurprising that 'fever' as a keyword showed a greater number of associated taxa (65.6%) relative to those taxa associated with 'malaria' (31.5%) due firstly to the greater number of keywords associated with fever conditions (Table 3.4). Secondly it is likely that fevers are a more ubiquitous phenomenon as opposed to malaria which is normally restricted to certain geographical regions.

3.4.2 Primary regression analyses

The selection of the taxonomic levels used here, was undertaken on the basis of certain published phylogenies (Bowe

et al.,

2000; Chaw

et al.,

2000; APG 11, 2003) though different classifications may have produced different results. The lack of detailed phylogenies prevented the use of phylogenetic comparative techniques, the disadvantages of which are discussed in Chapter 1.

Certain plant orders and families contained significantly greater numbers of MAFEV taxa (Figure 3.2) than would be expected if ethnomedicinal selection procedures were

random. While regression analyses may be considered unbiased, the application of these analyses do provide scope for subjective inputs. In this study, the selection of order and/or family taxonomic levels for analyses was considered subjective, in that other higher or lower taxonomic levels could have been used instead. It is also likely that results would have been different if the initial sampling was based on family level rather than at the level of order.

The use of least squares regression analyses for comparison of ethnomedicinally used anti-MAFEV higher taxa (Table 3.5) resulted in the falsification of the null hypothesis.

Ethnomedicinal practitioners may apparently select MAFEV plants at random but the experiential retention of taxa in that capacity is not random. The regression analysis of

families containing MAFEV taxa (Table 3.7) showed that greater residual values were associated with certain families than with others in the same order. This suggests a distinct plant selection bias towards certain taxonomic groups by ethnomedicinal practitioners. Certain 'hot' families clearly contributed more to their respective orders being designated as outliers. Repeated analyses of the same data set at order, family and genus levels would objectify the selection of the optimum taxonomic scale at which to run regressions. Of course none of these categories are equivalent due to different rates of divergence resulting from several suites of selective pressures. Families will be over-represented if the physiological and anatomical machinery required for the

manufacture of the particular suite of chemicals evolved as a synapomorphy for the ingroup (the family in question). Reasons for the dominance of certain families are therefore unclear, but the application of detailed phylogenies through additional

comparative methods may provide elucidation. This is because details of branch length and/or the evolution of characters on particular branches of a phylogeny should

necessarily be factored into the analyses in order to reduce artifactual signals (Felsenstein, 2003). Alternately, it could be that ethnomedicinal practitioners favour particular families due to phenotypic characteristics which they perceive as preferential.

Such practices may be particularly common where numerous closely related and similar- looking taxa are available for use in a given region. For example, the genus

Acacia

(Fabales) has 46 taxa (species, subspecies and variants) in the FSA region (SANB1, 2005) many of which are ethnomedicinally used (Watt and Breyer-Brandwijk, 1962;

Hutchings

et al.,

1996; Van Wyk

et al.,

1997). Uses include

inter alia

their inclusion in emetics and in enemas (Watt and Breyer-Brandwijk, 1962).

Comparison of selected taxa in 'hot' families (Table 3.8) from the regression analyses, with the antiplasmodial bioassay results revealed a majority with 1C50values :5 lOlJg/ml.

It can be concluded that regression analyses are a useful additional tool in bioprospecting.

3.4.3 Secondary regression analyses

The secondary regression analyses (Table 3.10) served to illustrate the potential for identification of additional higher taxa that could be prioritised for bioprospecting

purposes. The orders identified (Table 3.11) do however lie closer to the regression line, and as such may not yield candidates with particularly significant efficacies. The

repetition of regression analyses after removing outliers may be undertaken several times, but depending on the variability of the data, the data points will tend to move closer to the regression line. Where this is the case, the data will certainly be of reduced value. It is recommended that such analyses are not performed beyond the secondary stage as has been demonstrated.