Morphotype Classication Criteria, Nomenclature of Shea Varieties [Vitellaria Paradoxa (C.F. Gaertn)], And Inuence of Sociocultural Factors on Perceived Shea Natural Variation Across Parklands in Benin

Background: Local knowledge and perception are crucial to undertake the domestication of useful species such as Vitellaria paradoxa that makes signicant contribution to rural household economy in Africa. This study aims to document shea morphotypes diversity based on folk knowledge especially the main criteria farmers used to distinguish shea trees and to examine the inuence of sociodemographic characteristics on that knowledge. Methods: 405 respondents were surveyed across shea parklands in Benin using semi-structured questionnaire. We used the relative citation frequency of criteria followed by Kruskal-Wallis test to evaluate the inuence of sociodemographic attributes on local knowledge of Shea morphotypes variation. Factorial Correspondence Analysis described the links between the different morphotypes and parklands, and Principal Components Analysis was used to characterize farmers perception on morphotypes’ abundance, fruits and butter yields. Results: Respondents identied 13 morphotypes based on the ve most cited criteria which are fruit size (55.5%), tree fertility (15.40%), bark colour (10.51%), timing of production (5.38%) and pulp taste (3.42%). The citation frequency of classication criteria varied signicantly depending on the age, the education level and the sociolinguistic group of the respondent. The Bembèrèkè zone shea parkland revealed higher diversity of morphotypes traits. The small fruit type (‘Yanki’) was reported to be widespread. It produces higher fruit and butter yields according to respondents. Conclusions: From our ndings, farmers perceived an important diversity of shea traits that are used to classify morphotypes with economic or sociocultural importance. The revelation of that natural variation in shea tree is a key step toward the development of shea improvement program that could focus on the morphotype Yanki reported to be potentially high in fruit production and butter yield.

incomes and food and nutrient availability. In tropical Africa, wild fruit tree species of high economic importance include Vitellaria paradoxa (C.F. Gaertn) referred to as Shea tree.
Vitellaria paradoxa (Sapotaceae), is a prominent multipurpose species occurring in a 500 km wide belt from Senegal to South Sudan. The tree is mostly valued for its oil-rich nuts [11][12][13]. Shea is a widely used traditional staple cooking oil. The average consumption of its oil is estimated at 21 g per person per day in rural areas [14]. In Benin, shea ranks among the top local wild edible trees that farmers protect or grow in their elds, and is mostly used for food processing and ceremonies [15]. Shea tree is also exported for use as a primary cocoa butter equivalent in the chocolate industry and a high-quality ingredient in cosmetics. The European Union's authorization to use shea up to 5% in chocolate products has resulted in a signi cant growth of international demand in the last twenty years [14]. The local sale of fresh shea fruits, kernels and butter also provides an alternative source of income to households in rural communities, particularly for women who are the main collectors and processors of shea nuts and also retailers of shea products. In Benin, shea products contribute up to 36 to 46% of the income of rural households in Atacora, Donga and Borgou regions [16]. Apart from its use for cooking, shea oil is also used in traditional medicine (nasal decongestant, treatment of wounds, child birth facilitation), in cosmetics (soap making, skin and hair moisturizers and skin lotion) and for rituals such as in traditional marriages as reported by Gwali et al. [17] in Uganda. Shea trees provide shade for farmers, herders and their animals and the ripe mesocarp (fruit pulp) is a key source of energy during the preparation of elds in the rainy season when grain supplies from the previous season are low [11,18,19]. The roots and bark also have numerous medicinal uses in the treatment of dysentery, suppurating wounds and other ailments [20]. Due to its importance, shea has been protected through participatory management involving a complex mix of traditions and customs [17].
Despite the importance of this species for thousands households in Benin, shea densities are declining and shea populations are ageing. This state of things is due to extended forest land clearing and the discontinuation of fallowing spell required for the natural regeneration of shea. Other causes include agricultural land expansion and mechanization for commercial crops, tree cutting and removal out of crop elds, rewood and charcoal production [9,21]. Moreover, shea resources degradation results in a loss of agricultural diversity and implicitly a potential loss of currency in the sector [22].
Studying the species' diversity and documenting traditional knowledge and local classi cation systems are prerequisite to support conservation action and crop improvement [23]. They are instrumental in understanding phenotypic variation and help guide the selection of ethnovarieties that address local priorities. Theories suggest that people's socio-cultural and demographic attributes such as gender, age, and ethnicity in uence their preference of a given species [24]. Moreover, people's preferences for a given species determine their conservation attitudes toward that species [25]. Tietiambou et al [6] also reported that attitudes toward the conservation of species vary according to socio-demographic pro les (gender, age, education, ethnicity) of local people, their geographical location and preferences based on use and market values.
In Benin, ethnobotanical studies have been conducted on some wild fruit tree species such as tamarind [2], African locust bean tree [26], baobab tree [27], marula or cider tree [28]. So far, very little has been done to document phenotypic diversity and local knowledge systems on shea tree. The study by Glèlè et al [29] analysed the structure of shea butter trees in parklands located in different climatic regions of Benin. Gnanglè [30] reported that shea is an economically, socially and culturally important tree for northern and central Benin. In addition. Agundez et al [31] have recently studied local preferences for shea nut and butter production in Northern Benin. To date, no study have been undertaken to assess local knowledge on shea morphotypes traits and how they can be used in shea cultivars development.
This study documented folk knowledge of shea tree morphotypes, including classi cation criteria, nomenclature, perceptions on abundance and yields in Benin; it also unravelled the underlying sociodemographic factors that in uence folk knowledge across the shea tree range in Benin. We hypothesized that (i) shea classi cation criteria by farmers are based on both fruit and tree traits and (ii) shea morphotypes knowledge is in uenced by socio-demographic factors such as age, gender, instruction level and sociolinguistic groups in Benin.

Study area
The study villages are located in the shea parkland zones identi ed and described by Gnanglè [30] in Benin based on North-South and East-West rainfall gradient, socio-cultural groups, plant production period, and soil types [30,32]. We chose villages in close collaboration with resource persons of National Institute of Agricultural Research in Benin and shea industry (Fludor) staff. The choice of villages was proportional to the size and the dominant sociolinguistic groups of each of the shea parkland zone. In total, we surveyed ten villages in ve shea parklands zones including Bohicon, Savè, Parakou, Bembèrèkè and Kandi ( Figure 1).
Bohicon zone shea parklands is located between latitudes 7° and 8°N and includes the entire shea population stretching from Bohicon to Dassa-Zoumè. The average annual rainfall in the park is 1200 mm. Three types of soils can be distinguished, namely: the weakly desaturated ferralitic soils or ''terre de barre'' that stretches south of the park; the impoverished tropical ferruginous soils encountered in the center of the park and the hydromorphic soils north of the park. The average density of shea tree was 15 trees per hectare. The village Setto was selected in this zone for the study. The dominant sociolinguistic group was Fon.
The Shea park of the Savè region is located between latitudes 8° and 9°N and covers the shea population stretching from Glazoué to south of Tchaourou. The average annual rainfall is between 1100 mm to 1200 mm. Soils are basically leached or depleted tropical ferruginous soils. The Shea population of the Savè region has an average density of 26 trees per hectare. The village Toui was selected in this zone for the study. The dominant sociolinguistic group was Nagot The shea parklands of Parakou zone extends between latitudes 9° and 10°N and includes all the shea population located between Tchaourou and N'dali and between Parakou and Djougou. It is an area characterized by a Sudanian climate with an average annual rainfall ranging from 1000 to 1200 mm. In some places, moderately desaturated ferralitic soils are found. The average density of the park was 26 trees per hectare. The village Barei, and Sirarou were selected in this zone for the study. The dominant sociolinguistic group were Yoa and Bariba.
The Bembèrèkè zone shea parklands is located between latitudes 10° and 11°N and includes all the shea population extending from Bembérékè to Gogounou. The average annual rainfall in this park is 1100 mm. The average density of the park was 25 trees per hectare. Four villages namely Bensékou, Béroubouay, Soaodou, and Dipokor 2 were selected. The dominant sociolinguistic groups included Boko, Bariba and Ditamari. Three types of soils can be distinguished in this park: tropical ferruginous soils slightly leached, whether concreted or not on kaolinic material. In the depressions hydromorphic soils were found while under a vegetative cover we found ferralitic soils moderately unsaturated.
The shea parklands of Kandi zone is located beyond 11°N. It brings together the entire shea population ranging from Kandi to Malanville and from Kandi to Banikoara. This parkland receives an average rainfall of 800 mm. The average density of the park was 31 trees per hectare. The villages Birni-La a and Kokey were selected.
Dominant sociolinguistic groups in those villages included Dendi and Bariba people. Soils types encountered included poorly evolved soils, tropical ferruginous soils slightly leached and the hydromorphic soils.
The rural population of the ve zones host 1,637,434 inhabitants [33]. Livelihood activities carried out by the people of these sociolinguistic groups include agriculture, ranching, shing, hunting, processing of agricultural products, trade and craft [34]. Women mainly practice the processing of agricultural products individually or in groups with rudimentary equipment. The main processed products are Vitellaria paradoxa nuts (processed into butter), seeds of Parkia biglobosa (Jacq.) R.Br. ex G.Don (processed into a food condiment), the grain of Sorghum bicolor (L.) Moench (processed into an alcoholic beverage and used in some traditional ceremonies), groundnuts (Arachis hypogaea L.) processed into oil [34].

Socio-demographic pro le of respondents
The characteristics of the study population varied according to the proportion of respondents' age, instruction level and sociolinguistic groups. Respondents age vary from 19 to 90 years old. There were mostly adults surveyed in all parklands and the majority of them were not educated. In Bembèrèkè, Bohicon and Savè zones, both genders were well represented while in Kandi and Parakou, more women than men were surveyed (Table  1).
In total, seven sociolinguistic groups were represented in the study. Six sociolinguistic groups were found in one municipality only. The Bariba sociolinguistic group was found in four survey villages (Sirarou, Soaodou, Beroubouay and Kokey). The Fon and Tchabè/Nagot (in Setto and Toui villages) belong respectively to the subgroups "gbe" and "ede", and both belong to the large linguistic group "Kwa". The other ve groups (Bariba, Boko, Ditamari, Yoa and Dendi) belong to the large group of languages known as "Gur" or "Voltaïque" comprising most of the languages of the northern part of the country [35].

Study species
Vitellaria paradoxa is the only Sapotaceae tree species living on dry soil in the Sudanian and Sudano-Guinean phytochoria [36]. Shea trees are allogamous. Flowers appear between 10-15 years. The tree attains full fruit production after 45-50 years. The pulp of the fruit, which is normally sweet, is an important source of nutrients for humans, other mammals, birds and bats. Each fruit generally carries one or rarely two nuts [23]. The shea tree is generally stocky, 10 to 15 m high, with short bole (3-4 m) which can reach rather large diameters (80 cm and more). It is easily recognizable by its gray or blackish bark with thick and square scales. Unlike young plants, older plants have a woody bark that is deeply cracked into more or less rectangular plates recalling the crocodile's skin [37]. The edge of the bark is reddish and exudes latex that is found in all parts of the plant. The fruit is an ovoid 4-5 cm long, yellowish green drupe. The nut contains an almond whose fat content (shea butter) is close to 50%.

Data collection
In the study area, we were granted verbal permission from traditional leaders before starting the surveys. Semistructured interviews were conducted using the respondents' preferred language which were Fon, Nagot/Tchabè, Bariba, Boko, Yoa, Ditamari or Dendi. To facilitate the communication between the respondents and the rst author (where necessary) she was accompanied by a well-trained local guide (who understood both the interviewee-spoken language and Fench) in each village to facilitate the questions/answers translation. A total of 39 to 45 farm household respondents were surveyed per village in ten villages totalling 405 respondents. Respondents were chosen through a random walk in all hamlets of the village. However, when the selected participant was not available or did not want to be interviewed, the random walk continued until another consenting respondent was found. The rst section of the questionnaire dealt with informants' age, gender, sociolinguistic group, and education level. The second part focused on the different morphotypes known to the respondent, the local names and their meaning and her /his perception on the morphotypes' relative abundance, fruit yield and butter yield after processing. Data are made available as additional les in Table S1 and Table S2.

Data analysis
We evaluated the relative frequency of sociodemographic attributes, especially age group, gender, sociolinguistic group, level of education. We used Fisher exact test to analyse contingency tables. We performed frequencies distributions by using the number of citation of each criterion by all respondents and the total number of citations of all criteria to estimate proportions of main morphotypes classi cation criteria cited by farmers and rank them from the most to the least important. Only variables exhibiting a relative frequency of citation of at least 1% (approximately 9 out of 405 respondents) were used in further analysis and interpretation. In addition, a detailed analysis of the diversity of shea morphotypes was carried out with BiodiversityR package [38]. The variation in local knowledge on shea morphotypes traits composition in zones was measured by computing Sorensen index using betapart package. in R. To characterize morphotype abundance and performance, we performed a Principal Component Analysis (PCA) using respondents perceptions of morphotype traits. Variables included frequency distribution, fruit and butter yields estimated using qualitative scales (e.g. low, medium, high).
To evaluate the in uence of socio-demographic factors on local knowledge of shea morphotype traits, a generalized linear mixed model (with Poisson error family) was performed using the number of morphotype traits known by the respondent as the response variable. Fixed factors included zones of occurrence, age group, gender, education level and sociolinguistic group; and village provenance was used as a random factor.
We performed a Kruskal-Wallis test on the relative frequency of citation of morphotypes traits per zone, age group, education level, gender and sociolinguistic groups, with the package Agricolae. A Factorial Correspondence Analysis was carried out to assess the most cited morphotypes traits across the parklands zones. All analyses and graphics were performed in R 3.5.1 software environment (R Core Team, 2018).

Identi cation and characterization of shea morphotypes
Criteria used by farmers for shea morphotypes identi cation and classi cation Overall, farmers identi ed 35 different traits based on sixteen primary criteria related to fruit, tree, and nut variants ( Table 2). Eleven traits were mentioned based on fruit characteristics (size, shape, fruiting period, and yield) and ve were identi ed primarily according to characteristics of the fruit pulp (e.g. taste and density).
Ten traits were identi ed based on tree characteristics (bark colour, height, scales size, heartwood colour and trunk shape) and another ve were identi ed based on shea nut traits (size, presence and number of seeds).
Farmers also differentiated shea trees based on leaf colour which could be light or dark green. The top ve classi cation criteria reported by farmers were fruit size (56.47%), tree fertility (15.67%), bark colour (10.70%), fruiting period (6.34%) and pulp taste (3.48%). Farmers reported 13 shea morphotypes traits across the phytogeographical zones in Benin. Some criteria were seldom used. Traits with a citation frequency lower than 1% were not further considered in the analyses.
Shea tree name in the sociolinguistic group of surveyed areas include Sombou (Bariba), Boulanga (Dendi), Tam (Yom), Wougo (Fon), Egui èmè (Tachabè), Kouli (Boko), and Mouta, Koacha, or Taou in Ditamari. Table 3 presents the local names of the most mentioned traits by farmers in the different zones. These names recorded across the shea parks of Benin were based on morphological and organoleptic traits. Shea tree traits according to local communities were mainly related to bark colour, fruiting period and fruit size; other traits were based on tree fertility and pulp taste. Farmers distinguished shea trees having white, black or red bark; they also differentiated early fruiting trees (that bear fruits before the beginning of the rainy season), trees that have normal fruiting time (they produce fruits during the rainy season) and the ones, which have late production (they fructify after the two previous ones). Some trees produce fruits with sweet or insipid pulp and others produce large, small or medium size fruits. Farmers also reported that there are male and female shea trees. Based on the fruiting ability of shea trees, local people distinguished two types of shea tree: the "female," fruit-producing trees, and the "male" trees, which never produce fruits. However, in the Ditamari sociolinguistic group, early maturing genotypes were referred to as ''Nda'' which means ''male'' ; this comparison was borrowed according to them from the rule of male gametes during fecundation. In the Ditamari community, the production of fruit in shea trees is comparable to the conception of a child; the male seed gives fruit very early (whereas the female keeps pregnancy for several months before giving birth. That is why they equate this trait with a male.

Diversity of shea morphotypes traits in shea parks of Benin
Local knowledge on shea morphotypes traits diversity and Shannon index diversity varied from one zone to the another (Table 4). Based on the classi cation criteria used, different numbers of morphotypes traits were recorded across parks. Our results indicated that traits richness was greater in Bembèrèkè, Parakou, Kandi and Savè zones than in Bohicon shea parkland zone. There is a big difference in local knowledge between the northern zones and southern zone.
In order to understand how the richness of the morphotypes traits varied according to the number of zones considered, we realized the cumulative curve of morphotypes richness (Figure 3). Which revealed that the morphotypes richness increases and then becomes stable when all ve zones were explored.

Characterization of shea morphotypes' traits performance
Respondents reported three different level of shea morphotypes' abundance attribute such as 'widespread', 'few' and 'scarce'. Trees fruit yield and butter yield after processing could be according to respondents 'high', 'medium' or 'low'. Principal Component Analysis (PCA) performed on the perception scores of the distribution, the morphological and performance traits showed that the rst two dimensions accounted for 79.96% of the observed variation (Figure 4). The rst component is positively correlated with the variables medium and low yield (fruit yield), medium and low quantity of butter after processing (medium butter and low butter) and scarce and few distribution frequencies. The second component is negatively correlated with the variable null yield (no fruit production), and positively with the variables high yield, high butter and widespread.
According to the biplot of morphotypes and abundance, fruit yield and butter yield characteristics (Figure 4), it appears that, based on farmers' perception, shea morphotypes having the trait Yanki (small fruit) are characterized by high fruit yield, a high amount of butter after nuts processing and are widespread. On the other hand, shea morphotypes with the trait Bakanou (large fruit) and are rare (scarce) or very little spread (few). They may have medium and low fruit production and the nuts inside their fruits produce an average amount of butter after processing. The morphotypes characterized by the trait Akô (male tree) do not have ability to produce fruit (null yield). All other morphotypes traits do not reveal any particular characteristics. No speci c trend of these variables has been recorded for the other morphotypes traits.

In uence of occurence and sociocultural factors on local knowledge of shea morphotypes natural variation
The average number of traits cited by each respondent did not vary signi cantly (p> 0.05) regarding the parkland, age, gender, instruction level, and sociolinguistic group.
However, the relative frequency of citation of the 13 main local morphotypes varied very highly signi cantly regarding shea parklands zones (p< 0.001). Factorial correspondence analyses (FCA) plot ( Figure 5) shows traits mostly mentioned by farmers in each zone. Morphotypes that have red bark (Soarou), large and small fruit size (Bakanou and Yanki) were more common in Parakou and Kandi zones. Local morphotypes identi ed based on the pulp taste such as sweet and insipid (Dourobou and Yeniyando) were mostly mentioned in Kandi park than in others. Morphotypes traits relative to the timing of production, medium fruit size (Saganin) and bark colour speci cally morphotype having white and black bark, were common in Bembèrèkè shea parklands zone. On the other hand, the morphotypes traits Akô and Abô were mostly cited in the Savè and Bohicon zones.
With exception to gender (p = 0.218), all socio-demographic characteristics very signi cantly (p < 0.001) in uence the relative frequency of the 13 main traits of shea morphotypes ( Figure 6). Figure 6 shows the relative citation frequency (RCF) of traits according to age groups, gender, instruction level and sociolinguistic groups. All traits were known to the three age groups; however, local knowledge varied from an age group to another. The relative frequency of citation decreased from adult to old people passing by young for all traits except the trait "Dourobou" which was most reported by elders than young respondents.
General trend for the in uence of instruction level is the decrease of local knowledge on morphotypes traits from illiterate to literate respondents. Morphotypes citation frequency were higher for illiterate (25 < RCF < 85%) and respondents with primary school level (1 < RCF < 80%) than those of respondents who attained secondary and university (0 < RCF < 25%). Relative citation frequency of traits citation varied also from a sociolinguistic group to another. All sociolinguistic groups cited the female trait Abô and except Dendi, all sociolinguistic groups recognised the trait Akô which is quali ed as male by farmers. Bariba sociolinguistic group cited all traits mainly than those related to fruit size (Yanki, Saganin and Bakanou), bark colour (Kpika and Wonka) and pulp taste (Dourobou and Yeniyando) with RCF > 50%. The Fon sociolinguistic group mentioned only female and male traits. Shea morphotypes traits identi ed based on timing of the fruit production (Nina, Nda and Niendembou) were registered mostly within Ditamari sociolinguistic group (70 < RCF < 80%). The differentiation of morphotypes according to fruit size, pulp taste, bark colour criteria were almost common to Bariba, Dendi, Yom and Boko sociolinguistic groups located in the Northern part of the country.

Classi cation criteria of shea morphotypes in Benin
Farmers reported several classi cation criteria across shea parklands zones based on their importance. Fruit size, tree fertility (male and female), bark colour, timing of production and pulp taste were the top ve classi cation criteria most reported by farmers. Among those criteria, fruit size was the rst one mainly used to distinguish morphotypes. This result con rms that shea tree classi cation is mainly based on fruits variants as reported by Gwali et al. [39] in Uganda, Karambiri et al. [23] and Sandwidi et al. [37]in Burkina Faso. Ekué et al. [40] also reported for Blighia sapida K.D. Koenig that differentiation criteria included fruit size which was by far the most quoted criterion by farmers in Benin. Similarly, in a study on farmer classi cations of baobab tree (Adansonia digitata L.) in West Africa, they recognized 'male' and 'female' baobab trees [41]. These authors also reported that, local perceptions of baobab differentiation vary from one country to another. For instance, in Uganda and Burkina Faso, local shea classi cation was not based only on fruits but on nuts variants also. This nding revealed that variation can be observed in classi cation criteria depending on the geographical or sociolinguistic group. For example, local communities from Bohicon and Savè zones with Fon and Tchabè as dominant sociolinguistic groups have commonly distinguished male and female morphotypes. Moreover, the differentiation of morphotypes according to bark colour criterion among others, were almost common to Bariba, Boko, Ditamari and Yom sociolinguistic groups which belong to Bembèrèkè zone. As the result of the above, it seems that, contrary to what Sandwidi et al [37] have observed in Burkina Faso, certain sociolinguistic groups in speci c area in Benin were able to identify individual shea trees by merely looking at the bark colour of shea trees. This result con rms the rst hypothesis of this study concerning the using of tree characteristics to classify morphotypes. Assogbadjo et al. [42] reported that farmers can also distinguish different types of baobab trees using their own criteria based on the colour of the bark among others. Furthermore, utilization of the fruit variants as main description criterion could be linked to the fact that the description of the tree is related to the most important tree features for producers. Indeed, in the Bohicon zone, farmers distinguished trees on the basis of fertility and it is linked to the size and vigour of the tree, they also talked about the heartwood of the tree and ease of cutting or not. In fact, the wood only interests them. While in other sociolinguistic groups, the male morphotype is linked either to the capacity to produce the fruit, or to the time of production for others. Indeed, for these communities, the fruit is the most important part.

Diversity of shea morphotypes
Overall, farmers identi ed 35 different shea morphotypes based on thirteen primary criteria relative to fruit, tree and nut variants. The number of morphotypes reported in literature varies from one country to another and from one area to another inside the same country. According to Karambiri et al. [23] who obtained a lower number of ethnovarieties in comparison of Gwali et al. [39] study, the larger number of ethnovarieties identi ed in a study could be linked to the size of the study area. This means that, larger the study area will be, larger will be the number of morphotypes identi ed. This study con rmed this assertion. Indeed, richness increases according to the number of citation and zones at rst and then becomes stable at the maximum reached for ve zones (Fig. 3). In other words, to assess the whole diversity, it is necessary to go through all shea distribution range. Indeed, Akohoué et al., [43] who recorded ve Kersting's groundnut (Macrotyloma geocarpum (Harms) Maréchal & Baudet) landraces in Benin and Togo while previous authors reported three traditional cultivars only, asserted that this is due to the fact that previous studies did not include all production areas and as result, a part of the existing genetic diversity was left out. More recently, Coulibaly et al. [44] collected six different phenotypic groups of Macrotyloma geocarpum in Burkina Faso and Ghana. It appears based on the analysis that, till the whole distribution range of a species has been explored, the entire cultivars or morphotypes diversity could not be catched. For this study, the diversity was greater in the Bembèrèkè zone and very low in the Bohicon. This nding could be linked to the scale of the zone and trees density. Bembèrèkè shea parkland zone is the largest with the highest average density per hectare (43 stems/ha) [29,30].

Characterization of shea morphotypes
Knowledge of the agromorphological diversity of the identi ed shea trees is essential to begin any program of improvement of the species [45]. Since the most treasured product of the shea tree is its butter extracted from the kernels[46] which also depends on trees production, there is a need to assess the potentialities of the available diversity in terms of abundance of morphotypes and their fruit production yield and butter yield associated. This study revealed that farmers' perception concerning the abundance and the fruit and butter production of the Yanki morphotype is unanimous while for the Bakanou morphotype it varies. On the one hand morphotypes Yanki (with small fruits) were reported to be widespread and characterized by high yield and high amount of butter. Moreover, morphotypes Bakanou (with large fruits) were reported to give an average amount of butter and were reported to be rare or very little spread. To date, no prior study on shea morphotypes abundance had been conducted. Since rural communities' knowledge on Tamarindus' ecological range was congruent with scienti c ndings [47], then, there is a crucial need to ll this gap to better understand factors that in uence this state of affairs which should meet farmers' preferences. In addition, perceptions were scattered among farmers about this morphotype' fruit production which can be high, medium and sometimes very low. Across the study areas, farmers associated fruit size with soil fertility and as such, the rarity of these morphotypes could be linked to the low level of soils fertility in their area. A major problem for cropping systems in the tropics is the reduction in soil productivity that accompanies most systems of continuous cultivation [48]. The past decades have shown a rapid decline of land productivity and soil fertility in particular [49]. Moreover, the observed inter-annual yield variability in fruit yield [50] can explain the absence of consistency among farmers about these variables. In fact, the main constraints encountered with nut production are the remarkable decrease of the production and its huge uctuation from year to year [51]. Authors have hypothesized many combined biotic and abiotic factors underlying the annual variation of shea trees' fruit production but this process still not fully understood.
In uence of ecology and sociolinguistic attributes on the knowledge of Shea morphotypes knowledge Analysis of determinants of shea morphotypes knowledge shows that none of the socio-demographic characteristics signi cantly in uence the number of shea morphotypes known to the respondent (Prob > 0.05). This result is in agreement with Karambiri et al. [23] who also reported a similarity across the villages they surveyed. There is therefore a minimum threshold of shea morphotypes knowledge reached by almost all respondents regardless of their origin, age category, sociolinguistic group and level of education. However, sociolinguistic groups strongly in uenced the diversity of morphotypes traits reported. For instance, the criterion timing of production were speci cally reported by Ditamari sociolinguistic groups than in others and could be linked to the degree of interest carried by each speci c group or area. For instance, Gwali et al. [39] have also found a signi cant in uence of ethnicity on ethno-variety nomenclature and claimed that the variation in folk knowledge among the various ethnic groups may be due to the intensity of utilization of the shea tree and its products. Moreover, morphotype knowledge increased following south and north gradient and this is re ected by morphotypes' diversity and speci c richness (number of different kind of morphotypes traits) mentionned by zone. Based on these results we can assert that shea morphotype knowledge was low in Bohicon zone, medium in Savè, and high in Parakou, Bembèrèkè and Kandi zones. In addition, certain morphotypes traits were speci c to a given sociolinguistic group as found by Fandohan et al., [34] for Synsepalum dulci cum (Schumach. & Thonn.) Daniell which belongs to the same family as Vitellaria paradoxa. The intercultural differences observed in the knowledge of the uses of S. dulci cum are related to the environment and the availability of the plant resource [28]. Surprisingly gender does not in uence the frequency of citation of the most reported traits. Meaning that both men and women have similar knowledge of the shea morphotypes. Currently, studies reported differences in men and women knowledge on plants.
According to Laleye et al. [52] in a case study related to knowledge of plants in traditional treatment of diabetes in Benin republic, the fact that men cited more species than women may be due to the close association of men's knowledge with the treatment of diabetes. The same level of knowledge of morphotypes between men and women reported in this study reveals that shea butter is certainly of different but important interest for both men and women. Indeed, the level of knowledge of the species by women is justi ed, as they are the main actors from the beginning to the end of the entire shea production chain. Women control shea production as 92 % of the collection and 98 % of the processing activities were performed solely by women and girls even if shea fruits are consumed by both men and women [53]. Men also have the power to make decisions about the conservation of shea trees since they are the owners of lands that shelters shea trees. Gender in uence could be perceived in processing of shea nuts where men were not involved. Along the same lines, Agúndez et al., [31] suggested that the development of shea resources management and conservation programs should include ethnic preferences and consider gender, to avoid reducing women's pro ts in the shea butter local market.

Conclusions
Our study provides insight in the most reported traits of Vitellaria paradoxa in the ve shea parks zones of Benin. Number of morphotypes cited by a respondent does not vary signi cantly according to sociodemographic factors revealing that shea morphotypes knowledge is almost the same across the shea parklands. However, classi cation criteria of morphotype were diverse and strongly in uenced by age, instruction level, and sociolinguistic group. In fact, particularities speci c to certain zones and sociolinguistic groups concerning classi cation criteria have been underlined and emphasis the need to take into account sociocultural aspects to assess indigenous knowledge and by implication to identify indigenous preference of shea morphotypes. Local characterization of shea morphotypes should be taking into account for selection and breeding programs of the species. Availability of data and material The datasets supporting the conclusions of this article are included within the article (and its additional les).
Ethics approval and consent to participate A verbal agreement was obtained from traditional authorities of local communities and the informants prior to administering the questionnaire. A clear presentation of the study objectives made this easier.

Consent for publication
Prior and informed consent of local people's had been obtained for pictures publication.    Figure 1 Surveyed areas across shea parks in Benin and major sociolinguistic groups (in brown italics).

Figure 3
Cumulative curve of morphotypes richness in shea based on number of citation. Yanki= Fruit with small size; Bakanou= Fruit with big size; Saganin= Fruit with medium size; Akô=Male shea tree; Abô=Female shea tree; Kpika= White bark; Wonka= Dark bark; Soarou=Red bark; Nda:Precocious production; Nina=Normal production; Niendembou=Late production; Dourobou=Sweet; Yeni yando=Insipid; High Yield= high shea fruit production; Medium Yield= Average shea fruit production; Low yield= Low shea fruit production; Nul Yield=No fruit production; High butter = High amount of butter after nuts processing; Medium butter= Average amount of butter after nuts processing; Low butter= Low amount of butter after nuts processing.
Page 24/26  Relative citation frequency of the 13 main local morphotypes according to age, gender, instruction level, and sociolinguistic groups.

Supplementary Files
This is a list of supplementary les associated with this preprint. Click to download. TableS1Determinants.xlsx