Climatic niche of Dacryodes edulis ( G . Don ) H . J . Lam ( Burseraceae ) , a semi-domesticated fruit tree native to Central Africa

Dacryodes edulis is a fruit tree growing naturally in the rainforests of Central Africa. To provide data to scientists and forest managers for the habitat of D. edulis in the wild, the study of ecological niche and environmental adaptation was carried out. The geographical coordinates of 168 collections of wild individuals and 19 climatic parameters were treated under Geographical Information System. Potential spatial richness, favourable climatic parameters and elevation of the species were described. The favourable climatic conditions for natural growth of D. edulis were mainly revealed in the forests of Gabon and Equatorial Guinea. In Cameroon, these conditions were well marked in the regions of Kribi and Bipindi (South), in the forest of Yingui and Yabassi (Littoral), in the Dja Forest (East) and SouthWest towards the border with Nigeria. The mean of favourable annual precipitation is 2205.26 ± 505.5 mm per year. The optimal annual mean temperatures stretch from 23.5 to 25.5°C. The favourable elevations stretch from 0 to 800 m. These values are smaller than ones for domesticated individuals which represent fundamental climatic niche. Prospect in forest areas with the best favourable climatic conditions and elevation for natural growth of D. edulis should be encouraged to ensure in situ conservation of wild populations for maintaining genetic diversity.


INTRODUCTION
Dacryodes edulis (Burseraceae), commonly known as safou, plum tree or atanga, is an edible fruit (Figure 1) native to Central Africa.It grows naturally in evergreen forest.This species is characterized by two pseudostipuls, at the base of the petiole (Figure 2).This character differentiates it from other Dacryodes species.D. edulis is regarded as a multipurpose species.It produces edible fruits highly valued by local people and used in commercial transactions, even beyond the borders of Africa (Tabuna, 2000).In addition to human consumption and trade, the species is commonly grown in agroforestry systems used to shade the main species such as cocoa and coffee trees (Sonwa et al., 2002).Different parts of the plant are used in traditional medicine *Corresponding author.E-mail: gitodou@hotmail.com.
Abbreviations: CEFE, Centre d'Ecologie Fonctionnelle et Evolutive; IFORA, Iles Forestières Africaines.(Ajibesin, 2011).Culture of D. edulis in the Gulf of Guinea is secular (Chevalier, 1916).Because of domestication, its spatial distribution is extended inside Africa in twenty countries (Orwa et al., 2009) and also in tropical Asia in the Malay Peninsula (Aumeerudy and Pingo, 1989).Despite its importance in the live of local populations, D. edulis is among the species whose natural distribution and ecolo-gical niche are poorly understood.However, like most forest species of Congo Basin, wild D. edulis undergoes reduction in the number of individuals caused by agriculture, timber exploitation and global climate change.It appears very difficult to envisage a reliable in situ conservation of a species (for wild individuals) in order to maintain genetic diversity when the knowledge of the natural distribution and the ecological niche are poor.
The main objective of this work was to provide data to scientists and forest managers for the wild habitat of D. edulis.The specific objectives are: (1) to determine the geographical distribution and the spatial potential richness of D. edulis in its natural habitat, (2) to characterize its climatic niche (precipitation, temperature) and (3) to determine the favourable elevations for its natural growth.

Collection of geographical data
The geographical coordinates (latitude, longitude, and elevation) were obtained from three different sources.
1.The geographical coordinates of specimens conserved in six recognized Herbaria (Herbier National du Gabon (LBV); Herbier National du Cameroun (YA); Limbe Botanic Garden (SCA), Muséum national d'Histoire naturelle de Paris (P); Royal Botanic Gardens, Kew (K); Missouri Botanic Garden (MO) and Global Biodiversity Information Facility (GBIF) portal (www.gbif.org)were collected.Web sites including Google Earth, world-gazetteer.comand diva-gis.orgwere used to determine or complete geographical coordinates for samples with these data which are not indicated or are partially specified.The ecological notices indicated by collectors have been used to identify samples from wild individuals.2. Specimens collected in natural forests mentioned in the literature (Aubreville, 1962;Onana, 2008) with their geographical coordinates were selected.3. The geographical coordinates of specimens (leaves or cambium) collected in natural forests for DNA extraction (conserved in the laboratory of CEFE in Montpellier, France) during IFORA's prospection between 2006 and 2010 in Cameroon and in Gabon were considered.These leaf or cambium samples collected (with herbarium specimens in support) were identified using the identification key developed by Lam (1932), leaning on systematic reviews in the genera (Pierlot, 1996;Onana, 1998, Onana andCheek, 2003).
All duplicate samples or already mentioned in one of the sources have been eliminated to retain one.In total, the geographical data (latitude, longitude and elevation) of 168 collections were compiled.

Geographic distribution and estimation of the spatial richness
The geographical data of collections were entered into the software DIVA-GIS 4.1 (Hijmans et al., 2005).The natural geographical distribution was established from the geographical coordinates of each sample.The potential geographical distribution was modeled using nineteen climatic parameters extracted from BIOCLIM (Busby, 1991) and DOMAIN (Carpenter et al., 1993) available in DIVA-GIS.

Ecological niche and adaptability
The characteristics of ecological niche were studied based on the cover of the forest, climatic factors and elevation of collection sites.From these parameters, the map showing the ground cover in which the species could grow naturally was made following the spatial location of collection points.Graphs describing the ecological requirements from the data of the elevation and nineteen climatic parameters were made using the software STATISTICA version 6. Statistical processing was done by descriptive statistics and calculation of the correlation coefficient (r).Todou et al. 233

Natural geographic distribution of D. edulis
The natural distribution of D. edulis is focused in the center of endemism Lower Guinea.It stretches along the Atlantic in five countries (Congo, Gabon, Equatorial Guinea, Cameroon and Nigeria).The species occurs mainly throughout Gabon especially towards the Atlantic Coast, Equatorial Guinea and in the forests of South and south-West Cameroon.No individual with wild characters have been reported in the highlands of western Cameroon.Despite forest degradation in south of Nigeria, few individuals have been reported in close proximity to Cameroon.
Figure 3 shows the geographical distribution according to vegetative cover.It appears that D. edulis grows naturally in moist evergreen forest.In fact, D. edulis is not a naturally gregarious species like Dacryodes buettneri.Individuals were always observed isolated in forests.Identification is difficult because wild individuals are the upper canopy.In cases where several individuals were observed on a small area, archaeological remains and domesticated plants (mango tree, avocado tree or palm tree) demonstrating the existence of ancient habitations were recorded in places suggesting the old plantations.This is the case in Cameroon where several individuals were observed on the road of Kumba-Mamfe, beside the viaduct (Ndian Basin in Cameroon) cited by Kengue (2002a) as a natural wild population.During our prospection in Cameroon, eleven forest populations of wild D. edulis were observed and characterized; some have developed on old plantations (Table 1).These populations have been observed in oceanic regions directly subject to the monsoon, with prevailing winds from Southwest and in the regions of Centre, South and East characterized by equatorial climate or transition tropical climate, according to subdivisions based on regional climate stations homogeneous (Olivry, 2006).Individuals in the rain forests of centre, south and east Cameroon are sympatric with other species of the genera such as D. klaneana, D. macrophylla in undisturbed forests.

Potential natural geographical distribution of D. edulis
Based on nineteen climatic parameters, suitable forests for natural growth of D. edulis have been identified by the potential natural geographical distribution modeling (Figure 4).In Gabon, the ecological conditions favourable to natural growth were observed in the forests of Lopé and Waka Koulamoutou and in the forest of Oyem and Minkebe in the north.In Cameroon, these conditions were noted in the south (the forests of Kribi and Bipindi), in the atlantic cost (the forest of Yingui and Yabassi), in the congo basin (the forest of Dja) and in the south-west area (forests of Mamfe and the border with Nigeria).In  (Carpenter et al., 1993) available at DIVA-GIS.
Nigeria, wild individuals of D. edulis can be found in forests of southern part along the Atlantic Ocean.In Equatorial Guinea, natural forests of continental part have climatic conditions favourable for the natural growth of this species.
The potential natural location of D. edulis is much reduced comparatively to actual geographical distribution because of domestication.Research surveys in the areas shown are favourable to the natural growth of the species of interest in order to estimate the abundance of population.The knowledge of potential natural geographical distribution of D. edulis is essential to preserve wild populations for maintaining genetic diversity.A similar approach was used to plan conservation activities of six species of Passiflora from the Andes (Segura et al., 2003).Likewise, six new populations of Capsicum flexuosum in Paraguay were identified using the same approach (Jarvis et al., 2005).
The natural distribution of D. edulis is linked to theorical forest refuge of Maley (1987Maley ( , 1996)).Indeed, the dynamic character of tropical forests and climate change during the quaternary glaciations have caused changes in the expansion of forests.

Favourable annual precipitation for the natural growth of D. edulis
The favourable mean annual precipitation for the natural growth of D. edulis is 2205.26± 505.5 mm per year.The confidence interval of annual precipitation is not large [1500,3000] for the reason, that is, it is very close to the optimal annual precipitation ranging from 1600 to 2600 mm per year (Figure 5).This small extent is justified by the fact that these are the individuals that grow naturally without human intervention.They are growing in their original ecological niche.For the individuals collected in plantations, the confidence interval of precipitation is wide, ranging from 1400 to 4000 mm per year (Kengue, 2002b).The domesticated individuals can grow on the high rainfall areas on the slopes of Mount Cameroon to monsoon areas that have 4 months with less than 50 mm per month of rainfall (Verheij, 2002).The consequence is that the excess rainfall promotes vegetation against fruiting which is sparse and irregular.

Favourable mean annual temperature for natural growth of D. edulis
The confidence interval of mean annual temperature for the natural growth of D. edulis is 21.8, 25.8, while the optimal mean annual temperature range from 23.5 to 25.5°C (Figure 6).These values are compatible with the ones of Isseri (1998) who found that the mean annual temperature ranged from 23 to 25°C and are favourable for the best development and fruiting of D. edulis.The average mean annual temperature is 24.7 ± 1.4°C.The range of mean annual temperature favourable for natural growth of wild individuals of D. edulis is smaller than the one of individuals observed in plantations and they can grow in temperatures ranging from 19 to 26°C.However, low temperatures (<20°C) explain the very slow growth of individuals (5 to 7 years) and, individuals with high temperatures are characterized by poor fruit set (Kengue, 2002b).

Contributions of nineteen climatic parameters to natural growth of D. edulis
Considering only the significant factors (contribution of climatic parameters), two factors were retained for the principal component analysis (PCA) (Table 2).Both factors account for 68% of total variance.The first representing 39% of the variance showed a positive correlation with temperature.It distinguished individuals from those warm periods and cold periods.The second factor accounts for 29% of the variance.It showed a positive correlation with the number of individuals for precipitation of dry periods and negative corre- lations with the seasonality in temperature and annual temperature variation.Distribution of observations of wild D. edulis on orthogonal plans allowing for temperature and precipitation parameters showed significant correlations between these parameters (r ≥ 0.6 or r ≤ -0.6 and p < 0.05).Only one positive correlation was observed, between the precipitation seasonality and the temperature seasonality with r = 0.73 and p < 0.05.Five negative correlations were noted between the driest periods and annual temperature range and temperature seasonality.Precipitation of wettest month is negatively correlated with the mean monthly temperature range and precipitation seasonality with isothermality (Table 3).The negative relationships between precipitation and temperature suggest a substantial gradient in potential evapotranspiration that is in relation with distance of equator.

Favourable elevations for natural growth of D. edulis
The analysis of the relation between elevation and observation points of wild D. edulis showed that the natural growth of D. edulis is restricted to lowlands, from sea level to only 800 m of elevation (Figure 7).The mean of favourable elevation is 341 ± 200 m.No wild individual have been observed above 800 m elevation.Human planted D. edulis at various elevations up to 1800 m, in agro-ecosystems as the western highlands of Cameroon.According to Verheij (2002), plantations of D. edulis can successfully develop up to 1000 m.
According to Hutchinson (1957), there is difference between a fundamental niche (the abiotic conditions under which a population can persist in the absence of additional constraints) and a realized niche (where a population is competitive enough to occur naturally).Human successfully planted D. edulis even outside its native natural geographic range and under large interval of rainfall, under large interval of temperature and under large interval of elevation.To this successful degree that the plantations of D. edulis occur, the climatic range of domestic distribution represents attributes of the fundamental climatic niche.

Conclusion
Use of Geographical Information System (GIS) is an interesting approach to study the potential natural geographical distribution and ecological niche of plant species.This can help to plan in situ conservation activities

Figure 3 .
Figure 3. Distribution of wild individuals of D. edulis and existing vegetation.The blue points indicate actual distribution.The vegetation was extracted from DOMAIN(Carpenter et al., 1993) available at DIVA-GIS.

Figure 4 .
Figure 4. Potential natural geographical distribution of D. edulis.Potential geographic distribution estimated from simulations based on 19 climatic parameters extracted from BIOCLIM is represented in color.The red color indicates the area that connects the mos t favourable climatic parameters to the natural growth of D. edulis.The green color represents areas where the probability of finding individuals is low.The black points indicate the actual distribution.

Figure 5 .
Figure 5. Annual precipitation favourable for natural growth of D. edulis.

Figure 6 .
Figure 6.Favourable annual mean temperature for natural growth of D. edulis.

Table 1 .
Characteristics of forest with wild individuals of D. edulis in Cameroon.

Table 2 .
Mean of nineteen climatic favourable parameters for natural growth of D. edulis and values of two significant factors of PCA.

Table 3 .
Significant correlations between the precipitation and temperature parameters.