Abstract
Theory in ethnobiology suggests that the selection of medicinal plants by local people in a given region is not random and evolutionary closely related species may have similar medicinal uses. Additionally, plants selection by local people is often driven by plant therapeutic efficacy, plant availability, plant versatility or local knowledge on medicinal plants. We tested the hypothesis of non-random selection of medicinal plants as well as the potential mechanisms explaining such non-random plants selection. We also tested for phylogenetic signal in medicinal plants. Our study was based in four villages across Benin, West Africa, where the local communities have deep knowledge about medicinal plants. We installed 91 plots around these four villages to establish the total list of plant species and their abundance. We then conducted ethnobotanical surveys in the same villages to identify medicinal plants used in the local pharmacopoeia. To test whether the selection of medicinal plants used in the region was non-random and whether plant selection was driven by plant therapeutic efficacy, plant availability, plant versatility or local knowledge, we used a generalized linear model. Furthermore, we used the D-statistic to test whether evolutionary closely related species are more commonly used as medicinal than other species. We found support for non-random medicinal plant selection. Such a non-random plant selection was driven by plant medicinal versatility. Plant availability and secondary compounds have no significant influence on plant selection. Local people's knowledge on medicinal plants was significantly affected by individuals' literacy but not by their gender, their age or the ethnic group they belong to. We found a weak phylogenetic signal in medicinal plant uses. Our study reveals that the most used families are not necessarily the ones that have more secondary compounds or that are the most available to the local people, but are the most versatile plants. The high level of medicinal flora used at the local scale, which contrasts with the country-level analysis found by previous studies, suggests new methodological guidance in testing the theory of non-random medicinal plants selection.