African forest systems contribute to household economy and conservation of biodiversity of forest resources (Shackleton et al., 2011; Yadav and Dugaya, 2013). However, many aspects of relevance for their conservation are still unknown, for instance tree phenotypic characteristics. The intraspecific variation of most plant species in natural populations could be a reflection of genetic variability and an adaptation to fluctuating environmental conditions (Padonou et al., 2014). In the case of within-species genetic variation, it is an important basis for selecting species for domestication or favored varieties suited for particular needs (Leakey et al., 2005; Ewédjè et al., 2012; Padonou et al., 2014). Knowing that the existing phenotypic variation is an essential component for selection and planning of sustainable management programs,   studies on phenotypic variability of useful plant species like Carapa procera DC is of high importance.

 

Carapa procera, also known as African crab wood, is a woody species widely distributed in Africa (Thiombiano et al., 2012). It is a multipurpose species that offers various products for diverse uses (Mulholland et al., 2000). Several studies have highlighted the importance of C. procera oil and also wood (Weber et al., 2010; Zhang et al., 2011; Djenontin et al., 2012). C. Procera has a potential for oil production in Burkina Faso (Ouédraogo et al., 2013). An expectation of increasing demand for this oil could lead to possibilities for production development by local communities. Presently, C. procera is threatened because of different perturbations (habitat loss due to agricultural extension, seasonal bush fires and over exploitation of seeds). This situation could contribute to reduce the genetic diversity of the species if no appropriate conservation measures are undertaken. Unfortunately, very little is known about the patterns of phenotypic and genetic variations within natural populations of the species in West Africa.

 

Phenotypic variation in fruit traits constitutes an important base for diversity assessment and development of improved and sustainable management programs as well as further domestication investigations (Anegbeh et al., 2005). Morphological studies carried out on the fruits of Pentadesma butyracea provided important information used in the selection of superior individuals for its domestication (Ewédjè et al., 2012). Recent studies on morphological variation in fruits and seeds of Afzelia africana and Jatropha curcas showed the existence of different morphotypes and made it possible to select best specimens for cultivation (Padonou et al., 2013, 2014). Such studies are all the more important for high value wild tree species like C. procera as they represent potentials for alleviating local communities’ poverty. The present study aims to (i) determine morphotypes of C. procera fruits (ii) assess phenotypic variation of fruits traits and (iii) test whether phenotypic traits of fruits are related to tree morphological traits.

Taxonomy, ecology and use of C. procera

 

The genus Carapa belongs to the Meliaceae family. According to Weber et al. (2010), three species of this genus are known in West Africa: C. microcarpa, C. procera and C. velutina. However, recent studies indicated that the other two species can be considered as synonyms of C. procera (Sanogo et al., 2013). C. procera occurs in the speciesthe gallery forests from Senegal to RD Congo and Angola (Thiombiano et al., 2012). In Burkina Faso, it grows in the south Sudanian zone, where the annual rainfall is around 1000 mm. C. procera is a tree with composite and alternate leaves. Inflorescences are spikelets bearing small white flowers with a flowering period from January to April. The fruit is a capsule containing several seeds reaching maturity between May and June. Fruits are mainly exploited for oil production, as they have very high oil content (61.5%) in the seeds and an  important  concentration  of oleic acid (59.1%) (Djenontin et al., 2012). This oil is used in cosmetic, parapharmaceutical and medical industries (Weber et al., 2010). It is also used in biological control against harmful insects (Onanga et al., 1997; Konan et al., 2003).

 

Study area

 

The study was carried out in the western Burkina Faso, mainly in two localities: Banfora (10°41’N - 4°56’W) and Orodara (11°05’N - 5°20’W) (Figure 1). Both localities possess a south Sudanian climate (Fontes and Guinko, 1995), with a mean annual rainfall up on 1000 mm and a rainy with 70 to 90 rainy days (May to October). The mean annual temperature is 26°C. The vegetation is dominated by tree savanna with patches of dry forests and gallery forests. Flora is characterized by high diversity of Sudanian species and some Guinean species (Schmidt et al., 2005; Gnoumou et al., 2011; Sambaré et al., 2011). Based on FAO (2008), the main soil types are vertisols, leptosols, luvisols and ferralsols and the relief is hilly. Two major ethnic groups constitute the local population, Gouin and Senoufo. Agriculture, livestock farming and non-timber forest product (NTFPs) exploitation are the main income generation activities among local communities.

 

 

Sampling design and data collection

 

Fruits were collected targeting at least ten fruit-bearing trees per encountered shape type (visual aspect of fruit surface, e.g. crested, goffering, smooth). Sampled trees were located at least100 m from each other. All growth stages, from young to adult were considered. For each tree, trunk diameter at breast height (dbh), height and crown diameter were measured and all the fruits were counted. Ten mature fruits (unopened) were randomly collected from each tree. The variable length (using a slide ruler with ± 0.02 mm precision), circumference at the middle (using a metric tape with ± 0.1 mm precision), weight (using a balance with ± 0.01 g precision), number of carpels and number of seeds, weight of seeds (using a balance with ± 0.01 g precision) were recorded for each fruit.

 

Fruit description included the apex, the grooves and the epicarp aspect. Apex was characterized as long, short or none, and grooves as deep, few or none. The epicarp aspect concerned the visual appearance of the fruits surface.

 

Data analysis

 

Hierarchical clustering (HC) using Ward’s method and Ecludian distance was performed on following variables: Length, width (provided by the conversion of the circumference which has been measured), number of carpels, number of seeds, weight of seeds, type of apex and grooves (Struyf et al., 1997). The HC allow identifying the different morphotypes. Principal component analysis (PCA) was used to establish relations between the fruit variables. One way ANOVA without transformation was performed to test differences in the phenotypic traits and also in fruit production between different morphotypes, at 5% threshold. Normality was tested by using the Shapiro Wilk test (Glèlè et al., 2006).

 

Additionally, morphological variation of fruits was determined within morphotypes through the size of fruits on the basis of ratio L/w where L and w are length and width, respectively. One way ANOVA was performed on L/w ratio to test for differences in morphological types between the individual trees. This analysis was followed by a post-hoc test.

 

Finally, HC and PCA were performed on tree variables (dbh, height and crown). The HC was used to determine the types of trees relative to the fruit morphotypes. The PCA was used to test the correlation between fruit characteristics and tree variables.

Phenotypic characteristics of the fruits of C. procera

 

The hierarchical clustering of fruit variables identified three morphotypes representing distinct groups, according to the epicarp aspect (Figure 2).

 

 

The morphotype 1, the morphotype 2 and the morphotype 3 included crested fruits, goffering fruits and smooth fruits, respectively. Crested fruits were characterized by the presence of projecting longitudinal crests on the epicarp, each crest corresponding to a carpel. Goffering fruits were characterized by the presence of nodosities on all the surface of the epicarp. Smooth fruits exhibited smooth epicarp surface without clear appearance of carpels contiguous lines (Figure 3).

 

 

 

Phenotypic variation between morphotypes

 

The fruits of C. procera generally consisted of an orbicular capsule with 5 carpels. One fruit had an average weight of 0.5 ± 0.2 kg and contained 5 to 20 seeds. The ANOVA performed on fruit traits showed significant differences between the three morphotypes. Significant differences were found among morphotypes for the number of seeds per fruit (Table 1). Morphotype 2 contained the highest number of seeds per fruit while morphotype 3 contained the lowest number. The width, length and weight of fruits were similar between morphotype 1 and morphotype 2, but differed significantly from morphotype 3. Morphotype 3 exhibited the lowest mean values for all the traits.

 

 

Phenotypic variation within morphotypes

 

The L/w ratio of fruits varied significantly (p = 0.001) among fruits bearing trees. Three groups of fruits that we named submorphotypes were distinguished through the significant difference in the fruits ratio L/w within morphotypes at 0.05 threshold. These submorphotypes included short fruits (SM1), long fruits (SM3) and intermediate fruits (SM2) (Figure 5).

 

 

Tree dendrometric traits

 

The hierarchical clustering of tree variables revealed three groups of trees (that were named «tree-type»). The first tree-type included trees bearing crested fruits. The second  included  trees  bearing  goffering  fruits  and  the

third tree-type included trees bearing smooth fruits. The PCA ordination of tree variables indicated that the first two axes were highly significant (P < 0.05) and explained 77.8% of the variation (Figure 6). The coefficient of correlation between the two ordination axes and the tree- type characteristics indicated that the first axis discriminated tree-types according to the dbh, the total height and the crown cover. A larger dbh was correlated with large height and large crown. As for the second axis, it discriminated the tree-types according to the total number of fruits produced.

 

 

The Pearson correlation of the variables indicated that the tree variables (dbh, height and crown) had a weak correlation with fruit morphotypes (r < 0.5). The results from ANOVA confirmed that the characteristics of fruit morphotypes were not correlated with the tree size, neither with dbh (p = 0.186), height (p = 0.299), nor crown (p = 0.616). Concerning the variation on fruit production, the results showed that the number of fruits per tree was not significantly different (p > 0.05) between the three tree types. Likewise, the results  indicated  that the fruit traits of submorphotypes were not correlated with the tree dbh, height and crown.

Fruits are the most important product from C. procera and are a good source of income for local communities. The selection of fruit base on their characteristics is therefore very important for the domestication of the species (Leakey et al., 2005). The overall fruit size is indeed an important and easily selected trait for productive cultivars (Atangana et al., 2002).

 

In the present study, considerable variations in fruit traits were observed among C. procera populations, which revealed three different fruit morphotypes. Some traits of the fruits are important assets for selection needs. They are the large size and weight of fruits and the high number of seeds. Similar diversity of fruits traits were recorded across the local distribution range of the species, which means variations cannot be attributed to a specific population or a provenance. Likewise, these variations cannot be attributed to environmental or climatic factors as reported for other species in previous studies (Anegbeh et al., 2005; Ewédjè et al., 2012; Padonou et al., 2014). The evidence is the environmental and climatic conditions are similar at our study area scale for all the trees populations. In this situation the important factors that may have affected the morphological traits of the fruits are the age and the genotype of trees (Assogbadjo et al., 2005, 2006). In the present study, the data were collected as well in the young trees as in adult ones, which indicated non-correlation with tree size. The PCA and ANOVA results showed that the variations in fruits were not related to the tree parameters. So, the final possible factor that could explain the phenotypic variations in the fruits of our study population of C. procera may be the genotype of trees. The phenotypic differences between fruits with genetic origin could result from adaptation of the species to diverse environmental conditions (Mathur et al., 1984). This possible genetic variation in C. procera fruits could be an important source for varietal selection. Within each morphotype, the significant variation in fruit size showed that the submorphotype 3, which provided biggest fruits, should be more desirable in terms of selection. Consequently, any improvement program must also take these valuable traits into account and select trees that combine large fruit size with high weight and number of seeds.

 

The tree-to-tree variation in fruit characteristics is consistent with results from other indigenous fruit trees, such as P. butyracea (Ewédjè et al., 2012). As in the case of this later riparian forest species of West Africa, the phenotypic assets coupled with the market opportunities for fruit products represent domestication opportunities for C. procera. Similar studies highlighted the variation in fruit and kernel traits of Irvingia gabonensis and Dacryodes edulis in West Africa and indicated that farmers have obtained a gain of 40 to 65% in fruit mass by their own procedures of genetic selection (Atangana et al., 2002; Anegbeh et al., 2005). In the case of C. procera, increasing fruit mass could lead to oil yield increasing, which is a positive element for oil production improvement. It is clear from the results of this study that there is opportunity to identify and promote individual trees with high valuable fruit characteristics.

This study provides interesting basic knowledge about the extent of variation in C. procera fruit traits. Trees of the species exhibited three morphotypes of fruits based on phenotypic traits. One morphotype raises higher number of seeds than the two other ones and includes a submorphotype with the biggest fruits being the best candidate to selection and domestication for oil production purposes. This study provides evidence that there is considerable intraspecific variation in fruit traits of importance to genetic selection. Based on these results, programs of participatory domestication can be developed for C. procera in Burkina Faso. Therefore, through the development of cultivars from the best trees, it should be possible to make substantial improvements to the quality of the marketable products. On a wider scale, it is expected that this approach may benefit not only on the livelihoods of people, but also on the sustainability of the species’ genetic resources. A tree selection based on fruit morphotypes could be a basis for conserving natural populations of C. procera and, thus, could be valued to develop management programs aiming at genetic resources conservation as well as plantations with highly productive varieties of C. procera. In terms of fundamental research, the actual results may raise research questions on the taxonomic status of C. procera in West Africa with regard to the morphological diversity of fruits.

The authors have not declared any conflict of interest.

This work was carried out within the frame work of the QualiTree Research project funded by Danida (10-002AU), to which the authors are grateful.