The study was carried out in Northern Benin, where the species widely grows. Northern Benin has four Departments (administrative subdivisions): Alibori, Donga, Borgou and Atacora (Figure 1). Each Department is divided into municipalities and each municipality has several villages. These Departments occupy a total area of 83723 km², or 74% of the total area of Benin (112622 km²). They belong to two climatic zones of Benin: the Sudanian climatic zone (between 9°45' - 12°25' N) and the Sudano-Guinean climatic zone (between 7°30' - 9°45' N). The rainfall in these two zones is unimodal. The Sudanian climatic zone is a woodland and savanna region with ferruginous soils. The mean annual temperature is 35°C. The mean annual rainfall varies between 900 mm - 1100 mm. The Sudano-Guinean climatic zone is a transitional zone between the sub-humid Guinean zone and the Sudanian zone. This zone is characterized by a vegetation mosaic of forest islands, gallery forests and savannas. The population of the study area was estimated at 2,144,743 inhabitants and the farm population was 1,637,434 inhabitants 31. The dominant ethnic groups in the study area are: Bariba, Fulani, Dendi, Mokolé, Berba, Waama, Lokpa, Yom and Nago. Other groups such as Otamari, Boko, Anii and Foodo are also present. Livelihood activities carried out by the people of these ethnic groups include agriculture, ranching, fishing, hunting, processing of agricultural products, trade and craft. The processing of agricultural products is mainly practiced by women individually or in groups with handicraft equipment. The main processed materials are Vitellaria paradoxa nuts (transformed into butter), seeds of Parkia biglobosa (transformed into a food condiment with a ready market), the seeds of Sorghum bicolor (transformed into an alcoholic beverage and also used in some traditional ceremonies), pods of Arachis hypogea (transformed into oil).

All municipalities of the study area and all villages of the selected municipalities were subjected into a hierarchical cluster analysis based on demographic data collected at the National Institute of Statistics and Economic Analyses of the Republic of Benin.

In each selected village, the sample size (n) used for the study was computed based on the following formula 32:

n = ⋃ 1 - α / 2 2 p ( 1 - p ) d 2 where,

n is the total number of surveyed people in the study;

⋃ 1 - α / 2 2 is the value of the normal random variable for a probability value of α = 0.05, ⋃ 1 - α / 2 = 1 . 96 ;

p is the estimated proportion of people in the village which was involved in the production, processing and/or marketing of P. biglobosa. During an exploration phase, it was noticed that most of the rural households used the P. biglobosa species as producers or processors or traders. In this study, p was taken as the ratio of agricultural population and total population of each selected village;

d is the expected error margin of any parameter to be computed from the survey, which is fixed at 0.1.

After estimating the sample size, local people were randomly selected. The target groups consisted of traders, traditional healers and farmers. Table 1 summarizes the sample size of people surveyed by Department, ethnic group, sex and age group. Semi-structured interviews surveys were conducted among various people from 13 ethnic groups in the study area, namely Berba, Waama, Otamari, Anii, Nago, Lokpa, Yom, Foodo, Dendi, Mokolé, Bariba, Boko and Fulani. The questionnaire was tailored for each category of target group. The questionnaire evaluated information concerning demographic characteristics (name, sex, age, ethnic group and main activity) and use forms of P. biglobosa.

For ethnobotanical data analysis, the following indices were used: (1) the interviewee diversity value (ID); (2) the interviewee equitability value (IE); (3) the consensus value for plant parts (CPP); (4) the use diversity value (UD); (5) the use equitability value (UE) and (6) the consensus value for the form of use (CMU). Table 2 provides a detailed explanation of each index. These parameters indicate how the species is used and how the knowledge of these uses is distributed among the interviewees (table 2). To compute the use diversity value (UD), specific uses of P. biglobosa were classified into different categories of uses.

The normality and homogeneity of the ID and IE indices calculated were audited and the non parametric Kruskal-Wallis test, using Minitab, was carried out to assess significant differences related to sex, age and ethnic group. The consensus value for plant parts (CPP) and the consensus value for the form of usage (CMU) calculated per ethnic group were subjected to Principal Component Analysis (PCA) using SAS 9.1 software to better describe the relationship between these use values and the ethnic groups.

Moreover, the retention degree of the medicinal recipes in each ethnic group was determined from the diversity index of Simpson (1/D); which is calculated from the Simpson index (D) as follows:

D = ∑ i = 1 S n i ( n i - 1 ) N ( N - 1 )

with S being the total number of interviewees per ethnic group; N, arithmetic sum of ni recipes and ni, the number of recipes per interviewee. The diversity index of Simpson (1/D) ranges from 1 to S. If 1/D tends to S, the recipes are shared by everyone in the group. When 1/D approached 1, the recipes were only retained by a small group of people.

All farmers and traditional healers surveyed in the study area used P. biglobosa. The values for total diversity value (ID) and total equitability value (IE) were generally high (> 0.50) and indicated that knowledge related to P. biglobosa was distributed homogeneously within the target groups (table 3). The additional file 1; 2; 3 and 4 give more detail on the quantitative measurements of knowledge about P. biglobosa in the different Department of the study area. Ethnic group, sex, age and locality (Department) were important factors relating to knowledge distribution. In the Borgou Department, the values of these indices were significantly different among ethnic groups (p = 0.001). These differences, within Borgou, were due to the fact that some interviewees had little knowledge on the species. This was the case in young Fulani women. In the Atacora Department, the ID and IE values for traditional healers differed significantly. These differences were due to the fact that knowledge of P. biglobosa were more diversified and more distributed within Waama aged men (ID = 0.10 and IE = 0.40) than within young Otamari men (ID = 0.008 and IE = 0.03). In the Donga Department, the ID and IE values calculated for the farmers were significantly different among ethnic groups. These differences were due to the fact that knowledge of P. biglobosa was more diversified and more distributed within Lokpa men (ID = 0.36 and IE = 0.52) than within young Nago and Foodo men (ID = 0.04 and IE = 0.06 for each ethnic group). In the study area, processing activities and sale of organs and products from P. biglobosa were mainly conducted by women.

Uses of P. biglobosa were grouped into different categories: medicinal (47% of the interviewees), food (25%), commercial (22%), handicrafts and domestic (3%), medico-magic (1%), veterinary (1%) and cultural (1%). In each Department, whatever the ethnic group, most interviewees mentioned medicinal uses (table 4).

The medicinal uses of the species were the most diversified. Particularly, the digestive system diseases (diarrhoea, dysentery, abdominal pain), diseases of the cardiovascular system, injuries and burns, infectious diseases (shingles, malaria, abscesses, yellow fever, scabies, measles, chicken-pox, edema, jaundice), pediatric pathologies, symptoms and syndromes (malaise, tiredness, headaches, hip pain, ache, rheumatism, excessive weight loss, elephantiasis, onset of paralysis) were reported by most interviewees.

The food uses (transformation of the seeds in condiment, consumption of pulp) and commercial uses (sale of seeds, condiment from seeds, pulp, sponge from roots, wood fire) were observed for all ethnic groups surveyed. These findings demonstrate the importance of P. biglobosa in eating habits as well as in the improvement of agricultural households' income. The seeds and pulp (the most marketed plant parts) are an important source of income for farmers. They sell the seeds to the women who invest in the processing of seeds in food condiment. This condiment is appreciated in the study area.

The P. biglobosa parts used were: fruits [shell (2%), pulp (22%) and seeds (36%)], bark (17%), leaves (9%), roots (3%), flowers (1%) and branches (10%). The fruits (seeds and pulp), bark, leaves and branches were the organs most used. The results of the Principal Component Analysis (PCA) performed on the overall consensus values on P. biglobosa parts regarding ethnic groups (table 5) showed that the first two axes accounted for 75.22% of the total variation. Therefore, only these axes were retained to describe the relationship between P. biglobosa parts and ethnic groups. Table 6 shows the correlation coefficient between the different parts and the two axes. The first axis shows a positive link between bark, leaves, roots, seeds and pulp. Axis 2 shows the positive link between flowers and branches. Moreover, Figure 2 shows the projection of different ethnic groups into the system axes 1 and 2. From these observations, it can be deduced that Lokpa, Waama and Bariba ethnic groups assigned a high consensus value for bark, leaves, roots, seeds and pulp. In contrast, Nago, Anii, Dendi, Otamari, Mokolé, Foodo, Yom, Berba, Boko and Fulani assigned low value to bark and leaves. Ethnobotanical consensus values for branches were high for Fulani, Waama and Boko but low for other ethnic groups.

The different forms of use of P. biglobosa in the study area were: decoction, maceration, filtration, powder, boiled pulp, sponge, vegetable brush (toothpick), condiment derivative of seeds/potash, direct consumption, pounding, small cushion (leaves of P. biglobosa rolled as a cushion), chewing, consumption after cooking on embers, soap and frankincense (table 7). The results of the Principal Component Analysis (PCA) performed on the overall consensus values per form of use by ethnic groups, showed that the first two axes explained 60.31% of the total variation. Therefore, only these axes were used to describe the relationship between use forms and ethnic groups. Figure 3 shows the projection of different ethnic groups into the system axes 1 and 2. The correlation coefficient between the different use forms and the first axis showed a positive link between decoction, maceration, filtration, condiment, powder, sponge, pounding and direct consumption. Axis 2 shows a positive link between vegetable brush (toothpick), pounding, powder, consumption after cooking on embers and soap. We therefore deduce that ethnobotanical consensus values for decoction, maceration, filtration, condiment, powder, sponge, pounding and direct consumption use forms were high for Bariba, Dendi, Fulani, Waama and Lokpa but low for other ethnic groups. Vegetable brush (toothpick), pounding, powder, consumption after cooking on embers, soap consensus values were high for Lokpa, Anii, Foodo and Yom but low for other ethnic groups.

The Simpson's diversity indices for different ethnic groups were significantly lower than S, i.e. the number of interviewees per ethnic group (table 8), meaning that each respondent knew practically at least one use of P. biglobosa. Each respondent had their way of using the species. Taking into account the Simpson's diversity indices and coefficients of variation, it can be concluded that the recipes are diversified and vary according to ethnic group.