1. Ethnobotany, the science of human-plant interactions, has for long focused on documenting the traditional knowledge that humans have developed and accumulated over centuries towards plant uses. However, how such knowledge is constructed remains poorly understood and some of the methodological approaches developed for this purpose have been criticized. 

2. Here, we combine negative binomial models and phylogenetic comparative methods to test whether selection of medicinal plant species and organs by local people are non-random using data from Benin, a country in West Africa with roughly 3,000 plant species. 

3. We found evidence for taxonomic and phylogenetic non-random selection of medicinal plants: some taxa are preferentially used for medicine. Our analysis uncovers that plant organs are also non-randomly selected for medicine. Beyond plant taxonomy, similar plant organs of closely related species tended to be used for similar treatments, because these organs are likely to have similar secondary chemistry. Such non-random organs selection was more apparent for plant organs such as roots, bark and leaves which are predicted to be more vital for population fitness and species persistence. 

4. Collectively, our study suggests that the emerging non-random pattern of medicinal plant selection may be a consequence of heterogenous within-plant distribution of secondary chemistry across different organs proportionally to their importance to plant fitness as predicted by the optimal defence theory. 

We first digitalized the existing flora of Benin (Akoegninou et al., 2006) to retrieve species and family names and build the total plants list for Benin. We verified binomial name changes and family classification against APG IV using The World Flora Online (WFO, 2021). This flora includes 2802 angiosperms from 583 families. Second, to build the medicinal plants list for Benin, we obtained the most comprehensive compendium on Benin medicinal plants (Adjanohoun et al., 1989). This compendium includes 507 medicinal plant species from 83 families and was commissioned by the French Agence de Cooperation Culturelle et Technique and developed by the best-known botanists of West Africa at that time. From these two (total and medicinal) plant lists, we build a dataset that includes for each plant family occurring in Benin, the total number of species and number of medicinal species. For phylogenetic comparative methods, we reconstructed a time-calibrated phylogeny for all 507 species used as medicinal in Benin. To this end, we compiled a matrix of 488 DNA sequences for the matK region, a marker used as DNA barcode. These sequences were retrieved for 488 species from GenBank/EBI and BOLD (www.boldsystem.org, accessed June 2017).  For species for which DNA sequences are missing from these public repositories, we used DNA sequences from congeneric species that have their sequences on GenBank/EBI and BOLD.

Adjanohoun EJ, Adjakidje V, Ahyi MRA, Ake Assi L, Akoegninou A, d’Almeida J, Akpovo F, Bouke K, Chadare M, Cusset G, et al. 1989. Contribution aux études ethnobotaniques et floristiques en République Populaire du Bénin. Médecine traditionnelle et pharmacopée. Paris: ACCT.

Koegninou A, Van der Burg WJ, van der Maesen LJG. 2006. Flore analytique du Bénin. Backhuys Publishers, Leiden.

This repository includes 7 datasets and one R script that were used for all the analyses and illustrations in Gaoue et al. 2021 PPP

Dataset 1: Plant families from the flora of Benin with the number of total species per family and the number of medicinal plant species per family.

Dataset 2: Details of the DNA sequences used for phylogenetic reconstruction of the Benin medicinal flora. This includes all GenBank/EBI and BOLD accession numbers for gene sequences.

Dataset 3: Coding of plant organ selection for medicinal use. 1: an organ is selected for medicinal use; 0: the organ is not used for any medicinal use.

Dataset 4: Plant genera that form clusters of species that drive phylogenetic signal in organ plant selection

Dataset 5: Species classification into families that are over or under-utilized for medicinal purposes.

Dataset 6: Families classification into over or under-utilized categories for medicinal purposes.

Dataset 7: Phylogeny of medicinal plant species from the Benin flora.

Code 1: R Script for a corrected Moerman test of non-random medicinal plant selection and phylogenetic signal test for over/under-utilization of medicinal flora and non-random plant’s organs selection.