The study was carried out in Benin, which is located between 6°30′N and 12°30′N and 1°′E and 3°40′E and covers 114,763 km². The human population is about 12,909,041, of which 51.2% are women. This population’s annual growth rate is 3.5% [National Institute of Statistics and Economic Analysis (INSAE, 2016), 2016]. The annual rainfall ranges from 900 mm to 1,300 mm, forming three climatic zones (Guineo-Congolese, Sudano-Guinean, and Sudanian) split into 10 phytogeographical districts ( Figure 1 ). The phytodistricts correspond to the smallest floristic and climatic units encountered in each climatic zone (Adomou, 2005; Adomou et al., 2011). Each climatic zone has a human population comprising multiple ethnicities, and each of them holds several specifics habits. In rural areas, where farming is the primary activity, modern healthcare services are lacking and plant species are mostly used to treat diseases. Indigenous people use traditional medicine knowledge (TMK) to maintain their health systems. TMK has been crucial for centuries and is at present because many people in African rural communities have limited access to modern medicines, owing to their high cost and the increasing circulation of fake drugs in Africa (Klein, 2019; Vorrath and Voss, 2019). The knowledge of medicinal plant uses and practices seems greater among the older generations than the younger generations. Indeed, ethnomedicinal knowledge often increases with age, and the elders have more accumulated indigenous knowledge of medicinal plants than the younger generation (Abebe and Teferi, 2021).

A total of 590 informants ranging from 22 to 96 years old were interviewed, 11% of whom were women. The data collection focused on the sociodemographic characteristics of informants as well as ecological factors. The sociodemographic characteristics recorded comprised sex, ethnicity, age, level of education, marital status, household status, knowledge source, and primary activity ( Table 1 ).

Regarding the species’ ecological factors, species accessibility and ecology were considered. Species accessibility comprised the selling of its organs (availability on market: abundant, moderate, scarce), cost of its organs (free, moderate, expensive, very expensive), number of used parts (single or several), and reasons for the species’ disappearance [agriculture, medicinal overexploitation, technology, witchcraft (evil spirits)]. Ecological factors concerned ecological availability (abundant, moderate, scarce), habitat (forest, fallow, farm, home, or garden), conservation difficulty (easy or hard), and reproduction mode (slipping, seedling, rooting, or young plant).

To quantify the TMK held by each informant, an Ethnomedicinal Knowledge Index (EKI), adapted from Bennett and Prance’s (Bennett and Prance, 2000) relative importance index, was used and defined as follows:

where NBS = NBS_i/NBS_t and NP = NP_i/NP_t. NBS is the relative number of body systems. It was computed by dividing the number of body systems cited by informant i (NBS_i) by the total number of cited body systems by all informants (NBS_t). NP is the number of pharmacological properties. It was computed by dividing the number of pharmacological properties cited by a given informant (NP_i) by the total number of pharmacological properties cited by all informants (NP_t).

Descriptive statistics were used to describe TMK’s relationship with the sociodemographic and ecological characteristics of the informant’s area. To identify the socioeconomic and ecological factors that most influenced TMK, a conditional inference tree model was produced with “party” (Hothorn et al., 2006) and “strucchange” (Zeileis et al., 2002) packages on the matrix of sociodemographic and ecological factors ( Table 1 ) and EKI as a response variable. A first conditional inference tree model was generated with the sociodemographic factors sex, age, instruction level, marital status, household status, knowledge source, main activity, and ethnicity. The significance of influencing factors was also examined with a generalized linear model (GLM). A second conditional inference tree model was generated with species’ ecological factors. A final model was tested that included both sociodemographic and ecological factors. This allowed us to consider possible confounding effects that might not have been caught with the individual models (i.e., the sociodemographic factors model and ecological factors model).

The medicinal species considered in the study comprised 116 woody species belonging to 90 genera and 34 families. The predominant families were Leguminosae (28 species), Moraceae (11 species), Rubiaceae (nine species), and Combretaceae (nine species) ( Table 2 ). The Angiosperm Phylogeny Group IV (APG IV) classification of flowering plants was used to classify the species within families. Voucher specimens were also set for species identification with the help of experts from the National Herbarium, University of Abomey-Calavi (Ahoyo et al., 2021).

All statistical analyses were done with R statistical freeware (R Core Team, 2017).

Phytodistrict (nodes 1 and 3) and habitat of species conservation (node 5) were revealed to have the greatest effects on ethnomedicinal knowledge in Benin ( Figure 7 ). The ethnomedicinal knowledge associated with species that were generally conserved in fallow (2), the farm (3), and fallow and the home (7) seemed greater than that associated with species conserved in the forest (1), the home (4), fallow and the forest (5), the forest and home (6), the farm and home (8), the home and forest and fallow (10), and the farm and fallow and home (11).

In addition to these main factors, source of knowledge [e.g., friends (E = –0.457, p = 0.032), training (E = –0.841, p = 0.023)], membership position [e.g., communal (E = –0.503, p = 0.019)], and their interactions [e.g., friends: communal (E = 0.494, p = 0.039); and parents (father/mother): communal (E = 1.599, p = 0.021)] also played an important role in ethnomedicinal knowledge variation ( Table 5 ).

Sociodemographic attributes have an immense impact on the valuation of ecosystem services and govern indigenous ecological knowledge (Das et al., 2022). Hence, they determine the usefulness of a species and contribute to the tailoring of vegetation patches. In a study of children’s health in the south of Benin, Kouchadé et al. (2016) found that ethnomedicinal knowledge varied with ethnicity, source of knowledge, and phytodistrict. These findings were also found by other authors (Dassou et al., 2015; Kouchadé et al., 2016), who also cited instruction level as a determining factor of ethnomedicinal knowledge for traditional veterinary medicine in Benin. As in the present study, age and sex were found to be main factors accounting for variation in ethnomedicinal knowledge in Benin (Assogbadjo et al., 2011; Gouwakinnou et al., 2011; Ahoyo et al., 2018); women held more ethnomedicinal knowledge than men, as they often trade medicinal plants as their main activity. In addition, women are more engaged in primary healthcare within their families (Albuquerque et al., 2011). The ethnicities Gourmaché, Mahi, Fon, and Nago, were the most knowledgeable. This may be explained by the fact that most of these people are originally from the south: the most diversified area of Benin in terms of plant species. The Gourmaché are originally hunters, and thus seem closely familiar with traditional remedies.

The greatest ethnomedicinal knowledge was found to be held by people who gained it from their grandparents. Indeed, in traditional Beninese pharmacy, knowledge was orally passed down through the generations and, thus, owned by sociocultural groups (Upadhyay et al., 2011; Yoro, 2012). Vertical transmission remains the main diffusion method for ethnomedicinal knowledge. This tendency has been confirmed by previous studies, which argue that skills and medicinal knowledge are usually transferred from grandparents to their grandchildren (Shapi et al., 2009; Cheikhyoussef et al., 2011).

Instruction level was negatively correlated with ethnomedicinal knowledge, as also found by Ahoyo et al. (2018). People with higher levels of education seek office jobs and give up traditional knowledge. In contrast, those with less formal education remain dependent on forest resources and studies have shown that the majority of traditional healers are illiterate (Dassou et al., 2015; Ouachinou et al., 2019).

Membership position was also an important factor that contributed to ethnomedicinal knowledge. Local healers who were involved in national or at least departmental associations were found to be the most knowledgeable. In such associations, they have more access to modern medicine, which deepens their knowledge. Owing to the constant proximity of local healers and elders to plants, the local healers were expected to be more knowledgeable than those involved in national or departmental associations, who often travel for meetings, training with western (i.e., modern) medicine practitioners, etc. However, Vandebroek et al. (2004) found that, in South America, medicinal plant knowledge does not necessarily depend on the level of plant diversity, degree of modernization, or absence of western healthcare infrastructure in the healer’s environment. Therefore, the role of social factors in ethnomedicinal knowledge determination still varies with people’s history.

Regarding informants’ main activity, medicinal plant sellers and healers were the most knowledgeable. Their activities contribute widely to the nature of neighboring shrublands. They are also always in contact with plants and other people who use them widely. These people are often traders, healers, or patients, and thus knowledge is continually shared and renewed (Dassou et al., 2015). They derive their incomes from the plant, and thus were aware of saving this cultural patrimony.

The main factors which determine ethnomedicinal knowledge were phytodistrict, mode of reproduction, reasons for disappearance, species availability, habitat of species conservation, and conservation status.

Ethnomedicinal knowledge was found to be greater in the southern phytodistricts (Guineo-Congolese climate) than the northern ones (Sudanian climate). This decreasing trend moving northwards was also found by Kouchadé et al. (2016). Dassou et al. (2015) found the inverse trend in their study for traditional veterinary medicine. The inverse findings of Dassou et al. (2015) could be due to the study’s aim. The northern phytodistricts appear more knowledgeable on livestock breeding, owing to the importance of this activity in the region. Indeed, phytodistricts are ecosystems or habitats for particular animals or plant organisms. They each have homogeneous conditions regarding soils and climate, which determines the partitioning of different peoples in the country. Thus, the floristic diversity in an area depends strongly on the ecology of the zone and its inhabitants. The plants found in an ecological zone are considered to be owned by its people, and the zones are characterized by their relative ethnomedicinal knowledge (Dassou et al., 2015). Therefore, the variation of ethnomedicinal knowledge found among phytodistricts could be due to species diversity among phytodistricts. Moreover, the Guineo-Congolese climate of the south seems more favorable to biodiversity than the Sudanian climate in the north. Correspondingly, ethnomedicinal knowledge, which is often linked to available diversity, will be greater in the south than in the north.

The mode of reproduction and the reason for the disappearance of species also affect ethnomedicinal knowledge. The species that can be reproduced from roots and slipping/cutting were associated with the greatest ethnomedicinal knowledge. Those that needed to be reproduced in a nursery were associated with the weakest ethnomedicinal knowledge. This trend could be attributed to the ease of this reproduction mode. Indeed, healers were most able to remember and share knowledge about the easiest reproducible species. Moreover, the ethnomedicinal knowledge associated with species that were unavailable because of human pressure (i.e., carbonization, overexploitation for wood technology, extensive agriculture, and cutting to avoid habitation by evil spirits) seemed to be the greatest.

The species that were locally protected, and often less available, were found to be associated with the greatest ethnomedicinal knowledge. A similar tendency has been shown elsewhere (Sharafatmandrad and Khosravi Mashizi, 2020). The harvesting of species with high popularity levels, and thus associated with high knowledge levels, may lead to sustainability problems (Posthouwer et al., 2016). Therefore, it needs to reverse the decline in wild populations of these woody species, otherwise important ethnomedicinal knowledge associated with the threatened species will be lost. The effects of ecological availability and the used part (single or several) on ethnomedicinal knowledge have been proposed elsewhere (ecological apparency hypothesis) although no significant effects were found for them in this study.

The ecological availability of species has for several decades been found to be highest for the most numerous species that are the most used by people (Philips and Gentry, 1993; Nzuki, 2016). In this sense, the local therapeutic use of woody species may be associated with their wide visibility and availability throughout the year. The perennial and most available species were found to play a major part in medicinal plant trading in Benin (Kouchadé et al., 2016). However, this study seems to show that ethnomedicinal knowledge is greater when the species are scarce. Thus, the ethnomedicinal knowledge of woody species is not necessarily dependent on species availability. This change could be due to anthropogenic pressure exerted on the distribution area of the species, with its negative impacts on the sustainable conservation of forests (Sinsin et al., 2009; Djego et al., 2011). Moreover, in the case of some species, although availability has decreased over the years, people are willing to travel far to collect them because of their importance.

The ethnomedicinal knowledge associated with species that were conserved in fallow, the farm, and the home seemed greater than that associated with species conserved in the forest. Thus, people are more knowledgeable about species that grow or are available in their vicinity.

The regularly used parts of the species also determined the level of associated ethnomedicinal knowledge, although its effect was not significant. People who regularly used fewer species or single organs seemed more knowledgeable than those who used numerous species or several parts of the same species. The users of several parts for different therapeutic uses were generally the most numerous (Hseini, 2008; Mehdioui, 2008; Lahsissene, 2010). The wide use of some organs such as roots and barks could threaten the wild populations of species (Gaoué & Ticktin, 2007; Yaoitcha, 2016). Generally, the researched utility and the knowledge about them drive the sampled plant parts.

Furthermore, an initial approach based on the chemosensory perception of taste was developed to find a biological basis in the selection of medicinal plants in the human species. However, there was no correlation between the ability to identify tastes associated with plant chemicals and the identification of plants with pharmacological properties (da Silva et al., 2020).

An examination of the synergy between the effects of social, plant ecology, and accessibility factors on ethnomedicinal knowledge revealed phytodistrict and habitat of species conservation as the most important factors. These factors seem strongly correlated with ethnicity and the mode of reproduction. Indeed, ethnic groups are clustered around phytodistricts and, thus, a variation among phytodistricts also implies variation among ethnicities. In turn, the knowledge associated determines the pressure on the species.

Different environmental stressors have an effect on the reproductive potential of species (Lazarus et al., 2021). The presence of these stressors in a species conservation habitat could affect the species reproduction viability, depending on its mode of reproduction.