Characterization of African Bush Mango trees with emphasis on the differences between sweet and bitter trees in the Dahomey Gap (West Africa)

 African bush mango trees (ABMTs) are economically the most important species within the family of Irvingiaceae. They are priority trees producing non-timber forest products (NTFPs) and widely distributed in the humid lowland forests of West and Central Africa. To boost their production and develop them towards a major crop for rural communities in Africa, a domestication program was initiated in the 2000s which is being coordinated by the World Agroforestry Centre. ABMTs belong to two taxa, one with sweet and one with bitter fruits which are morphologically difficult to distinguish. The fresh mesocarp of the sweet bush mangoes are consumed, while the seed of both bitter and sweet fruits are an important component of the African diet. The high oil content of this seed further increases their potential use.  Apart from the overlap of their morphological characters, the ecological and phenological distinction between sweet and bitter ABMTs is unclear due to: (i) the lack of comparative quantitative data and (ii) the lack of centralizing the existing country-level databases. Therefore, their taxonomic status is still not clear. Do they represent distinct species or varieties or are they mere forms within the same gene pool? It is also unclear whether the occurrence of ABMTs in traditional agroforestry systems in the Dahomey Gap, the dry savannah corridor between the Upper and Lower Guinean rain forest blocks, forms part of the natural distribution or not . Moreover, genetic studies addressing ABMTs diversity have been geographically restricted, and conclusions regarding the taxonomic status of sweet and bitter trees were not unanimous. This study was conducted in a perspective of developing suitable strategies for the conservation and use of ABMTs, mostly in the Dahomey Gap. First, differences in ethnobotanical knowledge of the major socio-cultural groups in the Dahomey Gap were linked to the agroforestry status of ABMTs. This was used to explain the characteristics of the spatial pattern of ABMTs abundance and the anthropogenic factors that govern this spatial structure as well as population survival in the Dahomey Gap. Second, occurrence data of wild and cultivated ABMTs were used in a species distribution modelling algorithm to calculate the niche space and potential distribution of bitter and sweet trees in Tropical Africa. The differences in the obtained distribution patterns were compared using ENM-Tools. Third, detailed monthly phenological data and morphological characteristics (qualitative as well as quantitative measurements on the leaves, bark, fruits, and seeds) were used to analyse the diversity of ABMTs and to discover differences between them in order to be able to identify bitter and sweet trees in the field. Finally, the molecular markers AFLP and cpSSR were used in order to map the genetic diversity of ABMTs and to discriminate sweet and bitter trees across Togo, Benin, Nigeria and Cameroon. The consumption of mesocarp and seed of bush mangoes is widely known throughout the Dahomey Gap. The level of knowledge within local communities of other types of uses (medical, social-cultural) is generally poor and decreases towards the western part of this region. This suggests that ABMTs (mostly the sweet trees) were introduced in this eco-region through the migration of human populations from the Lower Guinean forest block (Southeast Nigeria) to the West. In the Dahomey Gap, bitter trees are confined to the Volta forest region, a small-sized ecological area in south-western Togo. While low densities (< 462 trees per 25 ha) were recorded for wild bitter trees occurring in the Volta forest region, higher densities (up to 1020 trees per 25 ha) were found for sweet trees in human made agro-systems. This implies a clear difference in cultivation methods between bitter and sweet ABMTs. The intensive cultivation of ABMTs in the Dahomey Gap is influenced by farmland status, farmer’s socio-cultural group and type of ABMT. Small and exhausted farmlands are converted into sweet ABMT orchards indicating that their development is a small-scale process lead by individual farmers. Slash and burn agriculture and intensive collection of fruits for seed commercialization jeopardize bitter trees, while traditional fishing systems (using twigs), traditional mass selection strategy, and intensive land commercialization severely threaten sweet trees genetic resources. Using species distribution modelling, the potential distribution of wild sweet trees was predicted in the wetter zones of the Guinean-Congolian phytogeographical region, while that of bitter trees extended to drier zones in the Guineo-Congolia/Sudania and Lake Victoria regions. This difference is significant, supporting the idea that bitter and sweet trees belong to two different species. In the Dahomey Gap, bitter trees occur in the wild in the wettest ecological region of the Volta forest region which is a very small part of the Dahomey Gap. This region is ecologically particular among the ecosystems in which wild bitter trees generally occur. We also conclude that in the Dahomey Gap sweet trees occur only in cultivation. Within the Dahomey Gap, clear phenological differences exist between sweet and bitter ABMTs, mostly in their reproduction phases. Moreover, their reproductive success significantly depends on the type of ABMT, soil, climate and season and we conclude there is a low probability of hybridization between sweet and bitter trees in the area where they co-occur. The qualitative morphological characters, the type of bark, colour of the mature fruit exocarp and mesocarp, and fruit roughness, do not consistently discriminate bitter and sweet trees in the field. We strongly recommend broadening the geographic area of this study by increasing more bitter trees as well as the wild samples of both taxa to validate this conclusion. The bitter trees in the Volta forest region produce the heaviest seeds and this consistently distinguishes them from all the sweet trees sampled in the Dahomey Gap. However, a combination of quantitative morphological characters (from fruits, mesocarp, and seeds) failed to discriminate populations. On the other hand this indicates the presence of a high diversity and thus high potential for selection across all phytogeographical regions. However, domestication and climate appear to be playing a key role in the morphological differentiation of Dahomey Gap populations, and evidence of success in the traditional domestication and selection of sweet trees is proven.  Low genetic diversity was found for the bitter trees occurring in the Volta forest region in the Dahomey Gap due to the high fragmentation of the small-sized forest ecosystem in which they occur and the continuous reduction of the population size. The higher polymorphism and genetic diversity observed in the sweet tree population in Benin and Togo indicate the effect of domestication of material from different geographical origins as well as a frequent long distance transfer of genetic material. When used separately, the AFLP and cpSSR data failed to consistently discriminate geographical populations and bitter from sweet trees. But a combined dataset of both markers tends to differentiate such populations as well as tree types. Our results also provide evidence that the suitability of AFLPs and cpSSRs to assess genetic diversity patterns in Irvingia material needs to be thoroughly reassessed. Finally, although admitting that a broader study remains necessary, based on the presence of a consistent gap between both taxa regarding their reproductive periods, their different ecology and, of course, the consistent difference in taste of the fruit, we advise to treat the sweet and bitter ABMTs as two taxonomically different entities at species level: Irvingia gabonensis (Aubry-LeComte ex O’Rorke) Baill. and I. wombolu Vermoesen, respectively