Study area

The study was conducted in Benin (6° 25′ N–12° 30′ and 0° 45′ E–4° E), West Africa (Fig. 1) [36]. Benin is characterized by three contrasting chorological regions [37], northwards: the Guinean region (humid climate), the Sudan-Guinean region (sub-humid climate), and the Sudanian region (semi-arid climate) regions [38]. The country’s native vegetation has suffered severe degradation as a result of various intense anthropogenic activities. The vegetation becomes dominated by woodland and savannah northwards. The resident population in 2013 was nearly 10 million inhabitants, unequally distributed across the territory [39]. The population is mainly young (more than 40% are < 15 years old) and slightly female-biased (51.2%) [39]. The local economy is agriculture-based [39]. More than half the population (53.9%) live with less than 1 US dollars per day [40]. The average size of farmland per farmer is 1.7 ha, and more than 1/3 had less than 1 ha [41]. Regardless of socio-cultural groups, women have no or limited access to land [41].

Study species

The African fan palm, B. aethiopum (Fig. 2), belongs to family Arecaceae, subfamily Coryphoideae [42]. It is widespread and common across sub-Saharan Africa where it is a well-known and conspicuous component of savannas. B. aethiopum has a large and straight stem to 25 m tall and may reach 80-cm diameter. The fruits are massive, ovoid, and orange at maturity (Fig. 2). The mesocarp is pulpy and fragrant with many longitudinal fibers [25]. In Benin, B. aethiopum is found in all three chorological regions but with great differences in regard with distribution and abundance [30]. Overall, B. aethiopum becomes common with higher abundance northwards, but in all three regions, there exist areas of local low and high abundance [30].

Sampling and data collection

In Benin, B. aethiopum has a wider distribution and larger abundance (trees ha⁻¹) in the semi-arid region (24.54 ± 1.51) than in the sub-humid region (16.24 ± 1.79), and than in the humid region (7.06 ± 0.56) [30]. However, these chorological region patterns hide strong local differences within each region, where it is possible to identify areas with low and high local abundance. Based on previous studies [30], we therefore identified villages with high (> 20 adult trees ha⁻¹) and low (up to 10 adult trees ha⁻¹) local abundance of B. aethiopum. In each chorological region, we randomly selected one village with high abundance and one with low abundance (Fig. 1). Abundance at chorological regions was assumed as an indicator of the regional apparency, while abundance at village level represented local apparency. Following [43], three age categories were considered (age < 30 years for young; 30 ≤ age < 60 years for adults; age ≥ 60 years for old persons). Gender (men and women) was also considered. Primarily, it was planned to select in each village 30 informants for each combination of gender and age category, making a total of 1080 informants. But because of insufficient informants in some villages, 992 informants participated to the work (Additional file 1). Data were collected using face-to-face individual interviews based on a questionnaire (Additional file 2). The goal of the study in terms of gathering knowledge on B. aethiopum for a PhD research (this study was part of the PhD research of the first author on the conservation biology of B. aethiopum in Benin) was first explained to village authorities and next to each of the informant as to have their consent for participation before starting the interviews. Only individuals that consented to participate in the study were considered. Interviews were conducted with assistance of a local translator when necessary. The questionnaire comprised three main sections. The first was related to informant’s socio-demographic information (age, gender, and ethnic affiliations). The second consisted of a free-listing of the use-reports of B. aethiopum, defined as each specific use mentioned by the informant per plant part in the sense of [44]. For example if an informant mentioned use of fruits to cure malaria and the use of leaves to also cure malaria separately, we considered them as two separate use-reports. Use-reports were arranged per plant part in six use-categories adapted from [45]: food, handcraft, construction, firewood, medicine, and ceremonies and rituals. The use-report does not distinguish between “knowledge” and “real use”, as various potential uses may be known but real use may be different [46]. To account for that, the third section of the questionnaire focused on the real use of the species by asking the informant to score the six use-categories based on the importance of their actual uses. The score varied from 3 (“high use”) to 0 (“not used”) with score 1 for “medium use” and score 1 for “low use” [7]. Informants were additionally asked to rank each plant part based on the importance of their actual uses. Each interview lasted between 2 h and 2 h 30 min.

Data analysis

Diversity of uses

The number of significant use-reports per plant part was higher for fruits (6 uses) and leaves (6 uses). Stem was involved in four uses while seeds, hypocotyles, and petioles were each involved in three significant use-reports. Roots had only two significant use-reports (Table 1).

Uses of plant parts of B. aethiopum (see Fig. 3) varied across regions. Fruits and leaves had multiple forms of use, some uses being reported by more than 50% of the informants. These two plant parts in addition to hypocotyl were involved in uses with the highest fidelity level, up to 86% in some regions. Seed uses were relatively not common, with fidelity level less than 15%. Use of root was mentioned only in the humid and sub-humid regions where it is used either to treat malaria or to strengthen children (Table 1). Processing methods were also not similar across regions (Table 1). For example, ripe fruits are toasted (Fig. 3b, c) only in the semi-arid region, while boiled (Fig. 3d) in the humid region before consumption (Table 1). In the semi-arid region, fruits are often boiled in association with cereals (either maize or millet or sorghum) (see Fig. 3e). The hypocotyles were consumed either boiled (Fig. 3g–i) (in all regions) or toasted (only in the semi-arid region) (Table 1). The use of fruits to discard shrews and snakes from homestead or chicken coop was essentially reported in the sub-humid region. There was no mention of this use in the semi-arid region. Many handcraft products (e.g., fan, mat, hat, sponge; see Fig. 3l–q) are made from leaves and petioles of B. aethiopum. It is worth noticing that other plant parts including bark and flowers of male and female trees were involved in medicinal and ceremonies and rituals uses that were not significant (FL < 5%), hence not presented here.

Based on the ranking of plant parts with respect to their importance for informants, fruit was ranked first and was followed by hypocotyl (Fig. 4) due to their food uses and commercial value. These two plant parts were the most sold on the local market either in rural or urban areas mainly by women and children. Fruits are collected from the wild. Hypocotyles are either collected from the wild or harvested from fruits sown on farm or at homestead (Fig. 3w–x). Apart from fruits and hypocotyles, the following plant parts were successively leaves, petioles, and stem (Fig. 4).

Use-value of B. aethiopum: effect of region, local abundance, gender, age category, and their interactions

Main effect of local and regional abundance, gender, and age category on overall and by use-category UV

There was a significant (p < 0.05) relationship between overall knowledge of B. aethiopum uses and age category, gender, local abundance and region, either as main effect or in a significant interaction term (Table 2). Informants in the drier regions have more knowledge on the species use. Informants in the humid region reported less uses (mean ± standard error; 3.21 ± 0.08) than in the sub-humid (5.32 ± 0.14) and semi-arid (4.07 ± 0.11) regions. The younger the interviewee is, the lesser he has knowledge on the species uses. Young informants reported less uses (3.80 ± 0.11) than adults and old informants who reported similar number of uses, 4.36 ± 0.11 and 4.24 ± 0.14, respectively. Men (4.31 ± 0.10) were more knowledgeable than women (3.98 ± 0.10). Informants from areas with high local abundance reported more uses (4.19 ± 0.09) than informants from areas with low local abundance (4.09 ± 0.10).

Predictors	Use-value			Importance index
	df	dev.	p value	df	dev.	p value
Region (R)	2	177.44	< 0.001	2	353.95	< 0.001
Local abundance (A)	1	0.01	0.915	1	2.22	0.137
Age category (G)	2	17.54	< 0.001	2	10.74	0.005
Gender (S)	1	6.36	0.012	1	4.57	0.045
R × G	–	–	–	4	9.20	0.036
A × G	2	8.94	0.011	–	–	–
G × S	–	–	–	2	5.41	0.049
Goodness of fit test (Error)	981	1053.66	0.753	976	2154.52	0.999
R-square (%)	36.96			34.80

df degree of freedom, dev. deviance, p value probability value computed from a Chi-square distribution

Knowledge on the uses of B. aethiopum varied greatly across use-categories (Fig. 5). Irrespective of the examined factors, knowledge was higher for food use (1.87 ± 0.04) followed successively by handcraft (0.72 ± 0.03), construction (0.60 ± 0.03), medicinal (0.41 ± 0.02), firewood (0.36 ± 0.02), and ceremonies and rituals uses (0.18 ± 0.01) (Fig. 5). Knowledge on food use was higher in the semi-arid region than in the sub-humid and humid regions that had similar knowledge (Fig. 5a). Knowledge on medicinal, construction, and handcraft uses were higher in the sub-humid region than in the other regions (Fig. 5a). Knowledge on food use was higher in areas with low local abundance while higher in areas with high local abundance for handcraft and ceremonies and rituals uses (Fig. 5b). Regarding gender, men always reported more knowledge than women irrespective of the use-category except for firewood use where the UV was higher for women (Fig. 5c). With respect to age category, knowledge was always lower for young than adults and old informants who had similar knowledge, except that for the food uses, the differences were relatively narrower than for the other use-categories (Fig. 5d).

Interacting effect of region and local abundance with socio-demographic attributes

Among socio-demographic attributes, only age category was involved in significant interactions with local abundance (Table 2). Young, adult, and old informants had similar knowledge in areas with high local abundance while greater differences were observed in areas with low local abundance, in particular between young and both adult and old informants who had similar knowledge (Fig. 6).

Cultural importance of B. aethiopum

Main effect of region, local abundance, gender, and age category on the overall and by use-category importance

There was a significant (p < 0.05) relationship between overall importance (IP) of B. aethiopum and region, local abundance, gender and age category; either as main effect or in interacting effect (Table 2). B. aethiopum was more important (higher IP) in the sub-humid region (mean ± standard error; 5.65 ± 0.08) than in the semi-arid (4.54 ± 0.08) and humid (3.51 ± 0.13) regions; slightly more important for men (4.52 ± 0.10) than for women (4.50 ± 0.09); more important in areas of high local abundance (4.55 ± 0.08) than in area of low local abundance (4.46 ± 0.10); and more important for adults (4.71 ± 0.10) and young (4.44 ± 0.12) than old informants (4.34 ± 0.12).

The IP varied greatly among use-categories and for each of the examined predictors (Fig. 7). Irrespective of the predictors, food use was the most important (2.45 ± 0.03) followed by construction (0.61 ± 0.03), handcraft (0.57 ± 0.03), and medicinal (0.56 ± 0.03) uses and then firewood (0.29 ± 0.02) and ceremonies and rituals uses (0.03 ± 0.01) (Fig. 7). Considering region, food use was more important in the semi-arid and sub-humid regions than in the humid region (Fig. 7a). Medicinal use was more important in the sub-humid region while the firewood use was roughly not important there (Fig. 7a). Considering local abundance, food and medicinal uses were more important in areas with high local abundance than in areas with low local abundance (Fig. 7b). Handcraft and construction uses were more important in areas with low local abundance than in areas with high local abundance (Fig. 7b). With respect to gender, food and firewood uses were more important for women than men while handcraft use was more important for men than women (Fig. 7c). Regarding age categories, food use was more important for young and adult than old informants. Contrary to food use, medicinal and construction uses were more important for adults and old informants than young (Fig. 7d).

Interacting effect of region and local abundance with socio-demographic attributes

There was no significant interaction involving local abundance and either socio-demographic attributes. However, significant effect was observed for the interaction of region and age category for the overall importance of B. aethiopum (p = 0.036; Table 2). B. aethiopum was more important for young and adults in the sub-humid and semi-arid regions while less important for young in the humid region (Fig. 8a). There was also significant interaction of gender and age categories (p = 0.047; Table 2). Accordingly, B. aethiopum was relatively more important for young women than young men, similarly important for adult men and women while relatively more important for old men than old women (Fig. 8b).

Interacting effect of age categories, gender, local abundance, and region as drivers of TK

Our data support the general trend that TK depends on age categories and gender [4, 54, 55]. Young informants often reported less uses than adults and old informants, hence congruent with the assumption that TK is a time-dependent process of learning [6]. Therefore, informants from older age category, having spent a longer time with their natural environment, would normally have more knowledge than informants from younger age category [4]. Although women are reputed to have more contacts with NTFPs than men [56], the fact that men often have more knowledge than women in our case study may be because women are often specialized in some uses of a species, most often food and to some extent medicinal uses while men uses expand to other uses (e.g., construction, ceremonies and rituals). This utilization pattern of resources is often dictated by differences in activities and roles of men and women within households [54] and sometimes to cultural taboos or prohibitions [7].

The EAH has multiple implications in conservation biology. For example, a positive relationship between species visibility and their use imply gradual elimination of the more apparent species by predatory collection through the constant pressure for domestic/commercial use [17]. The finding that TK was globally higher in areas of higher apparency (either local or regional) supports the ecological apparency hypothesis. Previous studies using abundance as a quantitative predictor of UV have come to similar conclusions [57, 58]. However, others found either no link or at best a weak relationship [15, 16]. Therefore, the EAH would not always explain pattern of uses of species, and other aspects such as socio-cultural, economic, and political aspects should be accounted for [14] and may explain the failure of our model to capture about 60% of the variation in TK. Indeed, human decision processes are complex and cannot be reduced to only ecological or economic considerations; other factors such as social and cultural are important in the processes involved in choosing and harvesting a plant [59].

Informants in drier regions reported more uses than informants in the wettest region which is congruent with the EAH prediction at the regional scale following the regional pattern of B. aethiopum across the country: B. aethiopum abundance and distribution is proportional to dryness, increasing northwards (from the wettest to the driest region) [30]. In Benin, previous studies have shown that diversity of wild edible plant species and plant in home gardens declines towards the semi-arid region [60–62], suggesting a likely higher intensity of use on a narrow number of species when the climate becomes drier which may in turn results in more knowledge on the resource at least as food use is concerned. At a local scale, the EAH also overall proved true: the higher the local abundance, the higher the knowledge informants have on the species. However, looking at patterns within each region (not statistically tested because of lack of replicates), this seems to not be the case in the semi-arid region where informants in the village of low local abundance reported up to two times more uses than informants in the village of high local abundance (low abundance: 5.50 ± 0.14 use-reports, high abundance: 2.89 ± 0.09 use-reports), compared to the sub-humid (low abundance: 3.70 ± 0.21 use-reports, high abundance: 6.49 ± 0.13 use-reports) and humid regions (low abundance: 3.24 ± 0.14 use-reports, high abundance: 3.19 ± 0.10 use-reports). As such, differences in TK due local apparency could be related to regional apparency. This pattern suggests two hypotheses. First, in the village of low local abundance in the semi-arid region, B. aethiopum was abundant in a recent past and likely has undergone rapid decline as reported by informants in that village (Additional file 3). This thus raises the importance of understanding the species past abundance in understanding current patterns of knowledge people have on them. Second, in the village of high abundance, the spread of uses of B. aethiopum is narrowed because other species fulfill the role of B. aethiopum observed in the village of low local abundance as suggested by the diversification hypothesis. The diversification hypothesis considers that the presence of multiple species in an environment amplifies the spectrum of alternatives [63] for subsistence, health, and livelihoods in general, thus reducing the pressure (use) on a single species, hence knowledge of its use. Apart from the regional apparency, one additional factor that may explain between regions differences for a given local abundance (e.g., considering villages of high abundance) is differences in ethnic affiliations [61, 64, 65]. Different ethnic groups have often different life style, beliefs, and perceptions of their environment that translate in different knowledge of resources of their environment [64] as previously reported for many other species in Benin [65–67].

This study also provides empirical evidence that pattern of knowledge distribution across age categories depends on local and regional apparency of the studied resources as we predicted. The most common (locally or regionally) a species is, the most likely knowledge on its utilization is similar across age categories. In contrast, the less common a species is, the relatively greater is the gap of knowledge among age categories, especially between young on one hand and adult and old people on the other hand. While knowledge acquisition as time-dependent process [6] is straightforward in explaining the often lower knowledge of younger informants as compared to older informants, local or regional apparency may provide interesting potential explanation of the magnitude of the gap of knowledge among age categories. Our proposition is that higher local or regional apparency seems to speed the process of knowledge acquisition resulting in similar knowledge among age categories. At the opposite side, lower local or regional apparency slow the process of knowledge acquisition and therefore result in greater discrepancies among age categories, in particular between younger and older informants. Additional studies on others species would be needed to clarify this proposition.

Cultural importance of B. aethiopum

Understanding the cultural importance of plant resource is crucial for an informed management [7]. Differences in form of uses across regions (see Table 1) are mostly linked to cultural differences due to different ethnic affiliations. For example, Gourmantché people from the semi-arid region toast the fruit before consumption. Such use was not mentioned in the other regions.

Most culturally important uses of B. aethiopum in study regions were successively food, handcraft, construction, and medicine. These use-categories are also the most known and important for palm species in Latin America [68, 69]. Irrespective of regions, local abundance, gender, and age categories, food use was the most culturally important, clearly indicating that B. aethiopum is primarily a food palm species in Benin, particularly in the sub-humid and semi-arid region. This corroborates Assogbadjo et al. [60] who reported B. aethiopum as a priority wild edible tree species in these two regions.

The finding that food uses was more important for women than men confirms the previous hypothesis of women specialization in the food use category but also stress on the relationship between patterns of plant uses and activities/roles in African households [54]: women are responsible for kitchen and most often are the sellers of food products in markets. This is further confirmed by the greater value of importance index for use-categories handcraft, construction and medicinal for men than women since culturally, men are often responsible for constructions and health care of the household members [70].

The fact that fruits and hypocotyles were the most important plant parts added to its greater commercial value are an additional evidence that B. aethiopum is a food palm species. Moreover, the greater commercial value of these two plant parts for women than men confirms B. aethiopum as a “women-palm” species (Fig. 9) as also reported in Brong Ahafo region in Ghana [35]. However, this great cultural and commercial value of fruits and hypocotyls if not well controlled may reduce regeneration potential in natural stands of B. aethiopum, threatening its population rejuvenation as reported for other species, e.g., Pentadesma butyracea [71].

Leaves were the second plant part with the highest number of significant use-report and the third most important plant part. They were often harvested from saplings and juveniles because adults are often taller (up to 20 m, see) and it is not easy to climb [30]. As reported in other studies [25, 72], stipe of B. aethiopum is a much appreciated material for construction and likely explain why it was so culturally important in the study regions. For example, roughly all houses in the village Loumbou-loumbou are made of B. aethiopum stipe (Salako, field observations; Fig. 3t). Gamba Begounou, an 83-year-old women, said “B. aethiopum is the only one good tree for house construction here. It is very resistant and can lives more than two hundred years”.

Medicinal use of B. aethiopum was also culturally important for surveyed informants and aligned with [35] who reported its medicinal use in 14 out of 28 African countries where it occurs and as the third most used palm species in traditional medicines in Africa. However, in spite of having four times more uses than food use-category (66 versus 16), only 5 were significant, suggesting a lack of consensus on most of the medicinal uses. This is likely due to the fact that the medicinal uses were mostly ethnic-specific. The common use of fruits against malaria also consensually reported in this study has been recently confirmed by pharmacological prospects [35].

Surprisingly, the use of sap commonly reported for B. aethiopum in other countries (e.g., Cote d’Ivoire, Senegal and Guinea; see [28, 32, 33]) was not mentioned in our study. In the humid region, this may be due to the fact that people reputed for palm wine extraction preferred Elaeis guineensis which is the ancestral source of palm wine in Benin [73]. In the sub-humid and semi-arid regions where E. guineensis is roughly absent, this may be due to (but not limited to) the fact that palm wine extraction is not their habit and that they do not have such knowledge. However, the use of fruits to discard shrews and snakes appears as a “new” reported use for B. aethiopum. This property may have potential for biological control of pests and hence required phytochemical screening prospection.