Introduction

Given the lack of effective control on the adverse effects of changing climate, adaptation strategies are increasingly searched for and developed to improve the resilience of local communities and contribute to food security and nutrition. In such a context, the traditional communities which are often vulnerable, and with low adaptation strategies, deserve more attention (UNFCCC 2018). Several efforts are being made to reduce the high dependence of their agricultural practices upon climate conditions. To this extent, introduction of important agroforestry species into traditional cropping systems is increasingly recommended as evidence exists on their contribution to locals’ subsistence and income generation (Schippmann et al. 2002). These initiatives are justified by the fact that with increasing food shortages caused by global climate change, it is expected that rural people will depend on wild edible fruit trees (WEFT) to meet their food and nutritional needs (Nebel et al. 2006; Bruschi et al. 2014; Berihun and Molla 2017).

However, the success of WEFT introduction in traditional agroforestry systems requires knowledge on the morphological variations within their populations and identification of desirable traits for genetic improvement. This is an important step towards domestication because, eventually, the morphological variations can reveal the best ideotypes suitable for different environments. So far, several studies have been carried out to reveal the morphological variation within WEFT across Africa (e.g. Leakey et al. 2000, 2005a; Abasse et al. 2011; Assogbadjo et al. 2011; Hounkpèvi et al. 2016; Padonou et al. 2017). Most of these studies have documented a continuous variation between and within studied populations for the investigated traits and have highlighted the potential to derive improved ideotypes from wild populations for the domestication purposes (Gouwakinnou et al. 2011). Although, domestication does not necessarily assure conservation and sustainable use of biodiversity, it has been shown that at least, while there is a substantial wild resource, domestication can increase intraspecific diversity (Leakey et al. 2004). In addition to providing important knowledge to sustainably manage useful plant species, studies of morphological variability can add insights into the adaptation capacity of the species in facing effects of unpredictable hazards like climate change (Hounkpèvi et al. 2016).

The monkey orange, Strychnos spinosa Lam. (Loganiaceae) is an indigenous WEFT to tropical and subtropical Africa, and widely spread from Senegal to South Africa (Sitrit et al. 2003; Isa et al. 2014). It naturally occurs in the savannah areas in tropical Africa and dwell in open woodlands including riverine fringes (Orwa et al. 2009). So far, researches undertaken on the species have underscored its importance for the local population welfare. It is among the topmost important edible tree species in the wild (Madzimure et al. 2013) which serves as an important food resource to poor farmers during periods of food shortage (Thiombiano et al. 2013; Bruschi et al. 2014). It is an important source of medicine for a wide range of diseases ( Arbonnier 2002; Isa et al. 2014; Plantz Africa 2017; World Agroforestry Centre 2018). It is also used to increase milk and crop production (Hoet et al. 2006, 2007; Bero et al. 2011; Salifou et al. 2017). In Benin, many parts of the species such as fruits, leaves, roots, seeds, etc. are used to fulfill these different functions (Avakoudjo et al. 2020). From the literature, knowledge on tree and fruit phenotypic diversity and differentiation across its distribution range in sub-Saharan Africa are rarely documented. This prevents from actions towards sustainable management of the species and exploitation of its potential in reaching food security with the local communities (Avakoudjo 2020).

Consistently with the ecological gradient hypothesis, climatic conditions have been shown to significantly determine natural variation of phenotypic traits at a relatively larger scale (climatic zones) in many WEFT (Fandohan et al. 2010; Hounkpèvi et al. 2016; Okello et al. 2018) and worth to consider in understanding differentiation in phenotypic variation. In this study, nine quantitative morphological descriptors were used to investigate the morphological variability and differentiation in S. spinosa populations in its natural range in two contrasting climatic zones where it is found in Benin: the drier Sudanian zone and the relatively more humid Sudano-Guinean zone. The ultimate goal was to identify elite individuals which can serve as basis for the successful domestication. Specifically, the study aimed to describe the relationship between dendrometric parameters and fruits descriptors and determine the extent of variation of the traits between and within populations of the two climatic zones. In particular, we tested the following hypotheses:

  1. 1.

    Individuals from the Sudanian zone have lower values for studied morphological traits compared to the ones in the Sudano-Guinean zone. We expected so, because drought triggers a wide variety of plant responses ranging from cellular metabolism to changes in growth rates and crop yields which often get lower in drier conditions (Shakeel et al. 2011);

  2. 2.

    As natural selection acts directly on phenotypes and important genetic variation within a population usually enables more phenotypic variation (Lumen Boundless Biology 2020), a higher morphological variation is expected for S. spinosa within populations rather than between populations. Indeed, the variation within a population allows some individuals to adapt to the changing environment by increasing their ability to survive and reproduce, which then ensures the survival of the allele in the population (Lumen Boundless Biology 2020). Furthermore, this expectation is also based on the reports of previous studies on WEFT in Benin along the same gradient (Fandohan et al. 2011; Gouwakinnou et al. 2011; Hounkpèvi et al. 2016).

Materials and methods

Study area

The study was carried out in the Sudanian and the Sudano-Guinean climatic zones which represent the distribution area of S. spinosa in Benin (Akoegninou et al. 2006; Fig. 1). The Sudanian zone (7° 30′–9° 45′ N) is characterised by a truly tropical climate with a unimodal rainfall pattern. The rainy season lasts for 7 months on average and covers the period from April to October with the maximum around August/September. The mean annual rainfall in is often less than 1000 mm. The relative humidity varies from 18 to 99% and temperature ranges from 24 to 31 °C. In the transition zone, which is the Sudano-Guinean zone (7° 30′–9° 45′ N), the climate is sub-humid with a tendency towards a unimodal rainfall regime. The annual rainfall varies between 900 and 1110 mm, whereas temperature and relative humidity ranges between 25 and 29 °C and 31 and 98%, respectively (Adomou et al. 2006; Assogbadjo et al. 2012; Gnanglè et al. 2012). Each of these climatic zones are further divided into phytodistricts which are meso-scale subdivision based on differences in natural vegetation, soil, and to some extent climatic conditions. So, the Sudano-Guinean zone is divided into three phytodistricts namely Zou, South-Borgou, and Bassila. The Sudanian zone also encompasses three phytodistricts including North-Borgou, Mékrou-Pendjari and Atacora chain (Adomou et al. 2006).

Fig. 1
figure 1

Map of Benin showing the location of the studied populations of S. spinosa across phytodistricts

Full size image

Sampling

Strychnos spinosa has already been subject to cultivation in some areas like Israel (Sitrit et al. 2003) and Australia. Though, no evidence on its cultivation exists for Benin, we focused only on natural populations in protected areas in order to depict natural variation of traits. This was done to limit bias on the studied traits assuming that individuals in the unprotected areas are more prone to human actions. Thus, individuals of the same phytodistrict were considered as a population (a sub-population level was considered when the population individuals’ come from different protected areas in the phytodistrict) and populations of the same climatic zone have been considered as provenance. Therefore, populations of Mékrou-Pendjari (Pendjari national park) and North-Borgou (W national park and Trois-rivières forest) were considered in the Sudanian zone (Supplementary Fig. 1) while populations of Bassila (Penessoulou forest), South-Borgou (Wari-Maro and Toui forests) and Zou (Agoua forest) were considered in the Sudano-Guinean zone (Supplementary Fig. 1), making a total of five populations at least 75 km apart. Tree inventory was done using linear transects of 3 km oriented North–South and East–West from a chosen random starting point based on the protected area map. Identification of trees of S. spinosa was made 50 m apart from both side of the transect and all individuals of the species encountered in the transect were considered and georeferenced. The first individual encountered in the transect was considered as the center of a 1000 m2 plot (50 m × 20 m) and from this first plot, other plots were installed at each 100 m till the end of the transect.

Data collection

At least 30 adult individuals (from the different populations) of the species were randomly selected in each climatic zone (provenance). A distance of at least 100 m was left between fruiting trees in order to avoid genetically close individuals. A total of 81 fruiting individuals of S. spinosa were assessed (30 and 51 respectively in the Sudano-Guinean and Sudanian zone). For each selected tree, the trunk diameter at breast height (Dbh at 1.3 m) and total height (htree, m) were recorded. Then a sample of 10 fruits was collected from each selected tree for further measurement following the protocol described by Leakey et al. (2000) and used by Fandohan et al. (2010) and Hounkpèvi et al. (2016). Seven morphological descriptors were measured on each fruit (Circ: fruit circumference; Thick: fruit thickness; DryM: fruits’ dry mass; SeedsN: seeds number; SeedsW: seeds weight; PulpM: pulp mass; and the Ratio: PulpM/DryM). These parameters were measured with an electronic digital calliper (0.01 mm resolution) and fruit weights were recorded using a 0.01 g-precision electronic balance. To improve accuracy, fruit’s circumference and thickness were measured at the first, the second and the third quarter of each fruit and the arithmetic means considered as the fruit’s circumference and thickness. Similar descriptors were used for other WEFTs (see Leakey et al. 2000, 2005a, b; El-Siddig et al. 2006; Fandohan et al. 2010).

Data analysis

Mean values and coefficient of variation were computed for dendrometric parameters per provenance and per population. The same statistics were calculated for each morphological parameter of fruits and seeds per population and provenance. The skewness was also calculated for each morphological trait to determine the symmetry of their distribution (Idohou et al. 2016).

For each trait, phenotypic classes were defined (Supplementary Table 1) following FAO (2018), Mengistu et al. (2015), and Mille and Louppe (2015). These classes were used to estimate the Shannon–Weaver diversity index (H′) which was used to characterize the phenotypic frequencies of the different traits. This index was computed using the following equation:

$$H = \mathop \sum \limits_{i = 1}^{n} P_{i} {\text{ln}}P_{i}$$

where n represent the number of phenotypic classes for a given trait and Pi the proportion of the total number of entries in the ith class. H was estimated for each trait, and per phytodistrict, and climatic zone. H was divided by Hmax = ln(n) in order to standardize H. The standardized phenotypic diversity index (H′) is in the range of 0–1. The diversity index was classified as high (H′ ≥ 0.60), intermediate (0.40 ≤ H′ ≤ 0.60), or low (0.10 ≤ H′ ≤ 0.40), as used by Mengistu et al. (2015).

One-way analysis of variance (ANOVA) followed by Student Newman Keul (SNK) test was performed to compare morphological traits among populations and between provenances. Linear mixed effect models were used to determine the relative importance of different sources of variability namely tree, phytodistrict, and climatic zone for each fruit trait. In this model, tree was considered nested in phytodistrict, and phytodistrict was considered nested in climatic zone. Canonical Discriminant Analysis (CDA) was performed to identify the most discriminating morphological traits and describe each population. This analysis also sought to highlight differences among populations with regards to the discriminant morphological characteristics. Hierarchical clustering analysis was performed to group trees in classes with homogenous characteristics based on both tree and fruit traits. The reliability of the cluster analysis was checked and validated through linear discriminant analysis. Morphological classes were then described using a Principal Component Analysis (PCA) and the results were graphically visualized in a biplot that displayed the groups and the association between individuals and morphological traits. Phylogenetic constellation plot was generated from the hierarchical clustering analysis to depict the genetic relationships among individual trees. The cluster analysis and phylogenetic constellation plots were performed in JMP pro 14.10 (JMP 2019). All remaining statistical analyses were performed using R 3.5.3 (R Core Team 2019).

Results

Morphological variability among S. spinosa individuals

Dendrometric traits of S. spinosa individuals varied significantly between climatic zones (provenances) and populations (phytodistricts) with high amplitude (CV > 25%) for total height and diameter, except for the diameter of North-Borgou population. However, the highest variation was observed in the Sudano-Guinean zone for both diameter and height. The diameter was more variable in South-Borgou and Zou populations whereas the highest variation in tree height was noticed in Zou and Mekrou-Pendjari populations (Table 1).

Table 1 Variability of trunk descriptors (diameter and height) of S. spinosa according to provenances and populations
Full size table

The coefficient of skewness varied between 0.2 and 1.72 for provenances (Table 1), indicating a right skewed distribution. The trend was almost the same at the population level except for Bassila where the coefficients were negative for the diameter and the height suggesting a left skewed distribution. The largest individuals were encountered in the Sudano-Guinean zone in Bassila population while the tallest ones were found in the Sudanian zone in the Mekrou-Pendjari population.

Morphological variability of S. spinosa fruits

Measured parameters on fruits were significantly higher in the Sudano-Guinean zone than in Sudanian zone (Table 2; P < 0.001). Considering populations, Bassila had the highest value for fruit thickness, dry mass for fruits, seeds number and pulp mass; South-Borgou and Zou populations had the highest value for fruit circumference and seeds weight respectively (Table 2; P < 0.001).

Table 2 Comparison of fruit descriptors (Fruit circumference, thickness, dry mass, number of seeds, seeds weight and pulp mass) of S. spinosa according to provenances and populations
Full size table

The fruits descriptors were dispersed around mean values in all provenances and populations except for fruit circumference (Sudanian provenance and South-Borgou population) and fruit thickness (Sudanian provenance) which were relatively less dispersed (CV < 15%, Table 2). In most cases, the descriptors showed the highest amplitude of variation in the Sudano-Guinean provenance (Circ; Thick; DryM; SeedsW and PulpM), Zou population (Circ; DryM and PulpM) and Mekrou-Pendjari population (SeedsN; SeedsW and Ratio: PulpM /DryM).

Distribution of the descriptors was largely skewed (moderately or highly) across the provenances except for the circumference in both provenances, and thickness in the Sudanian provenance which were approximately symmetric (skewness = 0.19). The Skewness coefficient was close to zero (approximately normal distribution) for: circumference in all the populations except Zou; thickness in North-Borgou and Mekrou-Pendjari; dry mass, seed number and seed weight in the populations of Bassila, North-Borgou and South-Borgou. There was also a high variation in fruit characteristics within and between populations (Table 3). The within populations variation accounted for the greatest part (53–90%) of the total variation.

Table 3 Results of the variance components estimation procedure (in percentage) on S. spinosa fruit and seed traits
Full size table

Phenotypic diversity among S. spinosa individuals between populations and provenances

For all traits, there was a large natural variation as illustrated in Table 4. The phenotypic diversity estimation (H′) for individual traits ranged from 0.38 for the ratio (pulp mass over dry mass of the fruits) to 0.96 for the pulp mass. The overall mean values of the diversity for all the traits pool together was 0.81. All the traits showed high levels of polymorphism (H′ > 0.60) except the ratio which has the lowest level of diversity (H′ = 0.38).

Table 4 Descriptors used for estimating morphological trait diversity in S. spinosa and estimated phenotypic diversity index (H′) for each trait
Full size table

On the provenance basis, high diversity indices (H′ ≥ 0.60) pooled over descriptors were obtained for individuals from Sudano-Guinean zone (Bassila, H′ = 0.78; Zou, H′ = 0.75; and South-Borgou, H′ = 0.62). The individuals of the Sudanian zone showed intermediate diversity, 0.51 and 0.57 respectively for North-Borgou and Mekrou-Pendjari. The within population H′ depended on the indices of the descriptors. When H’ was considered for each trait, a monomorphism (H′ = 0) was found for ratio in South-Borgou and for diameter in North-Borgou. However, low divergence was recorded for ratio in Bassila (H′ = 0.09), for seed weight in North-Borgou (H′ = 0.35) and in Mekrou-Pendjari (H′ = 0.39), for fruits dry mass in North-Borgou (H′ = 0.2) and in Mekrou-Pendjari (H′ = 0.15).

Relationships between discriminant morphological descriptors

Results of the CDA showed that 93.82% of the information was represented on the first two canonical axes explaining respectively 69.2% and 24.62% of the total variability (Fig. 2; Table S2). Overall, the two axes were highly significant (P < 0.001) in discriminating the populations. The first axis was negatively correlated with the plant’s diameter, fruit circumference, dry mass, seeds number, seeds weight and pulp mass (correlation > 0.5). The second axis was positively correlated with plants height and diameter. All fruit traits were negatively correlated with plant height and positively correlated with plant diameter on the first axis.

Fig. 2
figure 2

Canonical discriminant analysis diagram of the 81 selected individuals showing association of the populations and the morphological traits of S. spinosa

Full size image

Overall, the population of Bassila had the highest values for all morphological characteristics except for the plant’s height (Fig. 2). The populations of South-Borgou and Zou showed the highest value for seed weight but medium values for the other parameters (Fig. 2). The populations of North-Borgou and Mekrou-Pendjari showed high value for height but low values for other parameters.

Cluster analysis and group characterization

The hierarchical clustering analysis revealed three morphological classes of individuals of S. spinosa with a genetic similarity index > 80%. Individuals in different clusters were 88% similar while the different clusters were just 12% similar. The scatterplot from the linear discriminant analysis (Figure S2) showed that the first discriminant function (86.22%) well separated the individuals of class 2 from the ones of classes 1 and 3. The second discriminant function (13.78%) achieves a fairly good separation of classes 1 and 3, and classes 2 and 3, although it is not totally perfect. The first class comprised 29 individuals (35.8%), whereas the second and third classes consisted of 7 (8.6%) and 45 individuals (55.6%), respectively (Fig. 3). The individuals of class 1 were mostly (87.67%) of Sudanian provenance. The seven individuals of class 2 were all (100%) of Sudano-Guinean provenance. The individuals of class 3 comprised 12 individuals (26.67%) from North-Borgou population in the Sudanian zone and 17 individuals (37.78%) from the three populations of Sudano-Guinean zone.

Fig. 3
figure 3

Phylogenetic constellation plots displaying the relationships between 81 individuals of S. spinosa trees. Blue stands for individuals of class 1; green stands for individuals of class 2; red for individuals of class 3

Full size image

The first two principal components accounted for the total variability (100 %) among the individuals of Strychnos spinosa studied. The first principal component (Dim1) explained 86.52% of the overall variability and was positively correlated with all morphological descriptors (correlation > 0.9) except for the plant height (Table 5). Plant height was positively correlated with the second principal component (correlation = 0.970) which explained 13.48% of the overall variability. Fruit-related traits were discriminated by the first component and sorted the three morphological classes so that for these traits, class 2 < class 3 < class 1 (Fig. 4). In addition, the second component showed that class 1 has the highest value for plant height. This class is related to North-Borgou and Mekrou-Pendjari individuals while class 3 and class 2 were respectively linked to Bassila and South-Borgou, in one hand and Zou in the second hand.

Table 5 Principal components analysis (PCA) on morphological classes obtained from a hierarchical clustering: correlations between principal components (Dim1 and Dim2) and morphological traits
Full size table
Fig. 4
figure 4

Biplot showing the relationships between clusters and morphological traits of S. spinosa. Tree diameter; Height: tree height; Circ: fruit circumference; Thick: fruit thickness; DryM: fruits’ dry mass; SeedsN: seeds number; SeedsW: seeds weight; PulpM: pulp mass; Ratio: PulpM /DryM

Full size image

Correlation among fruits and trees traits

Overall, the correlations among traits were positive and statistically significant (Fig. 5). There was no significant correlation between tree height and fruits’ seed weight and other descriptors. The ratio of pulp mass over fruits’ dry mass was highly correlated with all the descriptors except for plant height, and seed weight. The tree diameter was highly correlated with the fruit mass, the pulp mass and the number of seeds. The dry mass of the fruit was correlated with all the descriptors except seed weight and tree height while the pulp mass was correlated with all the traits except tree height and seed weight. The fruit circumference was correlated with the fruit thickness, fruit mass and pulp mass. The fruit thickness and seed number were both correlated with the fruit’s pulp mass, dry mass and the ratio, and respectively correlated with fruit circumference and tree diameter.

Fig. 5
figure 5

Correlogram showing the pairwise correlation between tree and fruit descriptors. P value significance: ‘***’ significant at 0.001, ‘**’ significant at 0.01, ‘*’ significant at 0.05, ‘–’ significant at 0.1; DBH: Tree diameter; Height: tree height; Circ: fruit circumference; Thick: fruit thickness; DryM: fruits’ dry mass; SeedsN: seeds number; SeedsW: seeds weight; PulpM: pulp mass; Ratio: PulpM/DryM

Full size image

Identification of superior/desirable morphotypes of S. spinosa

Overall, individuals of class 2 showed good performances for the morphological traits (Table 6). This group was characterized by bigger individuals with the largest fruit, the highest amount of pulp and heavy seeds. The individuals of class 3, while being the smallest, scored well in terms of other parameters, which ranked them behind class 2. Individuals of class 1 were very tall, and had the lowest mean values for the different fruit descriptors. At population level, South-Borgou scored better for the assessed traits followed by the populations from Bassila and Zou (Table 7). At individual level, the elite individuals were found inside Bassila population. They showed morphological superiority in most of the traits assessed indicating a high potential for selection in the domestication process and can be considered for future research on the species’ improvement.

Table 6 Average value of the morphological traits (mean ± SD) for each group
Full size table
Table 7 Average value of the morphological traits (mean ± SD) for each population per group
Full size table

Discussion

Variation of morphological descriptors of S. spinosa

This study showed existence of heterogeneity in S. spinosa populations. This is most likely the result of the species’ fitness in various habitats which could be useful to survive and thrive under climate change (Visser 2008; Hounkpèvi et al. 2016) which impacts are already visible in Benin (e.g. flooding, irregular distribution of rainfall, etc.) (see Gnanglè et al. 2011). Considerable morphological variation exists in the measured traits. Our results on variance components of fruit traits support the hypothesis of higher morphological variation within populations than between populations and in accordance with several previous studies on other tree/shrub species in sub-Saharan Africa (Abasse et al. 2011; Cuni Sanchez et al. 2011; Korbo et al. 2012) in general and particularly in Benin (Fandohan et al. 2011; Gouwakinnou et al. 2011; Hounkpèvi et al. 2016).

All fruit features contributing to the fruit weight such as number of seeds, seeds’ weight, and pulp weight varied significantly between climatic zones and among provenances within climatic zones. The parameters related to the weight had relatively higher variability (pulp mass, seed weight, fruit dry mass) than those related to the volume (circumference and thickness). The high variation in the fruit dry mass and pulp mass is similar to the findings of Mkonda et al. (2003) on Strychnos cocculoides Baker fruits from Zambia, and indicated the potential of S. spinosa to be improved from the wild. Therefore as proposed by Mkonda et al. (2003), genetic selection and methods inducing pathenocarpy might help to increase these traits of great interest. In a study on Tamarindus indica L., Fandohan et al. (2011) reported large variations in pulp/fruit ratio among climatic zones and among individuals of the Sudano-Guinean zone in Benin. It can thus be said from this result that fruits having a greater size and mass have a higher pulp/fruit ratio. This may indicate that for superior morphotypes, increase in pulp mass is higher than that of the remaining part of the fruits. Indeed, for tree genetic resources conservation, intraspecific genetic variation is needed to ensure the future adaptability of the species (Allendorf and Luikart 2008). It comes out that S. spinosa is suitable for domestication because selection for improvement would be possible (Zobel and Talbert 1984; Msukwa et al. 2016).

Correlation between fruit traits and tree size

Fruit traits except seed weight were highly correlated among each other, but were not correlated with tree characteristics (height and in some case tree diameter). Thus, tree height cannot be used as an indicator of fruit production. Msukwa et al. (2016) also reported that height is a weak predictor for fruit production hence, cannot be used in indirect selection. However, the tree diameter can be a good indicator for fruit and pulp mass and ratio of pulp/fruit mass. There was also a strong relationship between fruit mass and pulp mass showing that for future selection for pulp trait, the fruit mass can be a good indicator (Leakey et al. 2005a; Fandohan et al. 2011). For this latter, the variability of the relationship between fruit mass and pulp mass confirms the differences between morphotypes and may have been driven by both ecological and genetic variation. Futhermore, the weight and volume related parameters were highly correlated. This strong relationship can be exploited in the selection process and may eliminate the need for multiple selection criteria (Mkonda et al. 2003). Since correlated quantitative fruit characters are of a major interest in any improvement programme, the search for a desired character may cause simultaneous changes in the other characters (Maghembe et al. 1998).

Estimation and analysis of phenotypic diversity

Previous studies on native fruit tree species in Benin such as that of Assogbadjo et al. (2011) on Adansonia digitata L., Gouwakinnou et al. (2011) on Sclerocarya birrea (A. Rich) Hochst. subsp. birrea, Fandohan et al. (2011) on T. indica and Hounkpèvi et al. (2016) on Vitex doniana Sweet, revealed a high level of diversity, indicating the high genetic potential of these fruit trees for species improvement in domestication program. This study also found large natural variation in monkey orange population collected from different climatic zones. From the estimated diversity indices presented in Table 4, it can be said that all the traits presented diversity indices beyond 0.40, showing intermediate to large genetic diversity among populations for each assessed trait. It is noteworthy that the individuals in sudanian zone are less diversified (H′ range from 0.51 to 0.57) than those of the Sudano-Guinean zone (H′ = 0.62–078). From this observation, the populations of North-Borgou and Bassila can be respectively considered as the less and the most diversified. This situation can be explained by the fact that the climate in the sudanian is more stable than the one of the Sudano-Guinean zone (transition zone). Thus, species in the transition zone will be more subject to adapting continuously to the climate fluctuations; thus causing different phenotypic variations in them.

Implications for the species conservation and domestication

Domestication and improvement of wild edible fruit trees hold great promise in agroforestry development (Mkonda et al. 2003). For this, the large variation found in fruit parameters of the species indicates considerable opportunities for phenotypic selection and quantitative improvement of the traits (Leakey et al. 2005a; Hounkpèvi et al. 2016). The high correlation of the tree diameter with most of the fruit descriptors is interesting since the trait for vigorous vegetative growth (dbh) is good for selection in the domestication process (Shackleton et al. 2002). The assessed traits showed an important variability between and within climatic zones. This is essential for adaptive evolutionary change and selection for domestication purpose since it could offer some possibilities to identify and select new cultivars through an oriented selection scheme (Gouwakinnou et al. 2011). The higher variation within provenances also indicates that rapid improvement progress could be gained through clonal propagation of elite trees with superior phenotypic traits such as higher fruit pulp or big fruit size (Maghembe et al. 1998). The cluster analysis showed three groups of trees with different traits’ performances. It came out that “plus tree” can be found in Sudano-Guinean zone and especially in the phytodistrict of Bassila. This suggests that best individuals are found where the climate is more clement (warm and humid). According to Shakeel et al. (2011) drought induces changes in morphological, physiological and biochemical changes in plants. It progressively decreases CO2 assimilation rates due to reduced stomatal conductance, reduces leaf size, stems extension and root proliferation, disturbs plant water relations and reduces water-use efficiency. This can explain why the individuals of Sudanian provenance did not have a good performance for the assessed descriptors and that the elite individuals were found in the Sudano-Guinean zone. Additionally, the best individuals revealed by this study bear heavy seeds which mean that their seed have enough reserve to germinate and facilitate the survival of seedlings in their early age. This parameter can be considered during the selection process for ideotypes of Sudanian zone. As reported by Gouwakinnou et al. (2011), natural selection in dry areas should favor bigger seeds to allow quick germination and seedling establishment at the onset of rainy season. Altogether, the high variability in phenotypic traits observed in this study indicated significant potential for improvement through the development of cultivars from elite trees using horticultural techniques (Leakey et al. 2008). According to Leakey et al. (2005a), cultivar selection would increase uniformity in the product, increase productivity and provide an incentive for farmers to plant monkey orange trees in their farming systems and home garden. For the species introduction in agroforestry system, rapid progress can be done by selecting superior morphotypes and propagating them as clones (El-Siddig et al. 2006).

Conclusion

The present study assessed variability of the morphological traits of S. spinosa individuals across zones in Benin and provided useful information for selection of appropriate ideotypes for a sustainable management of the species populations. The findings revealed existence of significant variations among individuals from different climatic zones for the studied descriptors and showed that elite trees can be found in the Sudano-Guinean zone. Moreover, it showed that tree diameter can be used to predict fruit descriptors. Therefore, this study paves the way for future investigations leading to setting up appropriate strategies for the species’ domestication and its introduction in traditional agroforestry systems in Benin. Nonetheless, the study was limited by repetition of the data with the time (only one season of data collection). Thus, future investigations are necessary on a larger group of individuals well spread across the country and should include other descriptors in order to reveal all the phenotypic and genetic variation within the species. This will enable a proposition of a definitive management strategies of the species. Further studies should concentrate on the reproductive biology in order to assess the heritability of the interesting characters of the elite individuals (big tree diameter, large fruit, high amount of pulp, heavy seed) found in this study.