Regression analyses of southern African ethnomedicinal plants: prioritising plant selections for bioprospecting and pharmacological screening

Using data sourced from the SANBI MedList database, the most comprehensive inventory of ethnomedicinal plants for the Flora of southern Africa region, taxa were grouped by order. A least squares regression analysis was used to test the null hypothesis that the use of these plants by traditional healers is strictly random. The above approach, following a North American study by Moerman (1979), sought to identify plant orders and subsequently families most highly selected for by traditional healers in southern Africa. The analyses identified seven principally ‘hot’ plant orders (Malpigiales, Fabales, Gentianales, Asteraceae, Solanales, Malvales and Sapindales) and ‘hot’ families therein from a total of 55 regional ethnomedicinal orders. Five ‘cold’ ethnomedicinal orders (Rosales, Proteales, Poales, Asparagales and Caryophyllales) were shown to be significantly less represented in the medicinal flora than predicted. These 12 outliers allowed for the falsification of the null hypothesis. Such regression analyses of local