Modeling current and future distribution patterns of Uvaria chamae in Benin (West Africa): Challenges and opportunities for its sustainable management

Abstract

Uvaria chamae is a wild shrub species widely used as a source for traditional medicine, food and fuel in West Africa. The species is threatened by uncontrolled harvesting of its roots for pharmaceutical applications and by the extension of agricultural land. This study assessed the role of environmental variables for the current distribution and the potential impact of climate change on the future spatial distribution of U. chamae in Benin. We used data related to climate, soil, topography and land cover to model the distribution of the species. Occurrence data were combined with six least correlated bioclimatic variables derived from the WorldClim database, data on soil layers (texture and pH) and topography (slope) obtained from the FAO world database and land cover from the DIVA-GIS site. Random Forest (RF), Generalized Additive Models (GAM), Generalized Linear Models (GLM) and the Maximum Entropy (MaxEnt) algorithm were used to predict the current and future (2050–2070) distribution of the species. Two climate change scenarios (SSP245 and SSP585) were considered for the future predictions. The results showed that climate (i.e., water availability) and soil type are the key predictors of the distribution of the species. Based on future climate projections, RF, GLM and GAM models predict that the Guinean-Congolian and Sudano-Guinean zones of Benin will remain suitable for U. chamae, while it will decline in these zones according to the MaxEnt model. These results call for a timely management effort for the species in Benin through its introduction into agroforestry systems to ensure the continuity of its ecosystem services.

1. Introduction

Natural ecosystems and wild species of animals, plants or fungi are generally used by local communities to satisfy their basic needs [1]. Natural resources are therefore of paramount importance for improving food security and living conditions of local communities [2]. Wild fruit trees provide a significant number of non-timber forest products (NTFPs) which are key resources in the provision of rural livelihoods [3]. Human activities and changes in climatic conditions constitute major causes of degradation of natural resources [4,5]. Climate change effects that exacerbate risks to wild plant resources include increasing frequency and intensity of wildfires, storms, landslides, and floods [6]. Given the threats to plant natural resources due to increasing anthropogenic pressure and loss of natural habitats, knowledge of their ecological characteristics is important for their long-term strategic management [7,8]. This will ensure the availability of Wild Edible Plant (WEP) and provide local communities with adequate livelihoods [9].

Uvaria chamae P. Beauv. (Annonaceae) is a WEP providing-species which occurs in tropical West and Central Africa within tropical dry and moist forests [10,11]. The species occurs at higher densities in gallery forests of the Guinean and Sudanese savannas on rocky, and cooler soils (in plains and valleys) in Senegal, Cameroon and the Democratic Republic of Congo [12]. It has been reported that U. chamae is still in the wild and much more common in the Guineo-Congolian and Sudano-Guinean zone of Benin [13,14]. In recent ethnobotanical surveys, U. chamae has been identified as a key multipurpose fruit tree species that provides vital goods and services for rural communities [11,15]. Uvaria chamae is also reported to have great medicinal potential [16], cosmetics properties [17], and firewood [18]. It was revealed that the overexploitation of the species'roots is a major cause of its decline [19,20]. Despite increasing efforts to document the importance of the species for the locals' communities and the threats that lead to its decline, scientific knowledge on its geographical distribution and dynamics in different climatic zones in West African countries is limited. Identifying the current and future potential distribution areas and assessing the factors underlying the geographical distribution is an important task [21,22]. In order to predict the dynamics of the geographic distribution of species in the context of climate change, species range modeling (SRM) is applied [23]. The modeling approach facilitates the identification of suitable areas for domestication and conservation of species under future climates [24,25]. The modeling results provide a framework for the design of a concerted management plan to ensure the provision of ecosystem services by identifying favorable habitats for the species. This study aimed at: (i) examining the distribution patterns of U. chamae under current and future climate conditions in Benin and (ii) identifying current and future favorable habitats for the species countrywide. We expect that part of the distributional range in Benin will become less favorable for U. chamae under the influence of climatic variations.

2. Material and methods

2.1. Study area

The present study was conducted in the Republic of Benin that covers three biogeographic zones (i.e. the Guinean-Congolian, Sudano-Guinean and Sudanian) where U. chamae occurs. It is a West African country located between 6 and 12°50′N and 1–3°40′E [26]. The Guinean-Congolian region of the country is marked by an average annual temperature of 25 °C–29 °C with a relative humidity of 69%–97% and a bimodal annual rainfall averaging 1200 mm. In the Sudanian region, rainfall is unimodal and sometimes less than 1000 mm per year, its relative humidity varies from 18% to 99% while its temperature varies from 24 °C to 31 °C. The Sudano-Guinean region is characterized by bimodal rainfall with a tendency to unimodality and an annual average rainfall that varies between 900 mm and 1110 mm, relative humidity varies from 31% to 98%, and annual temperature is 25 °C–29 °C [21]. In the Guinean-Congolese zone (southern Benin), the land cover types is mainly fallow land and agricultural fields, due to the high human population density [27]. Riparian forests, woodlands and dry forests characterize the Sudano-Guinean zone, while dry forests, riparian forests, tree, shrub and herb savannahs and woodlands, constitute the vegetation types of the Sudanian zone [27].

2.2. Study species

Uvaria chamae belongs to the Annonaceae family, it is locally called “finger root” or “bush banana”. The species is a shrub or a climbing woody liana reaching a height of 3.6–4.5 m [28]. The species is found in tropical and subtropical regions. It is widely distributed across western and central Africa, occurring in Senegal, Guinea, Mali, Sierra Leone, Liberia, Côte d'Ivoire, Ghana, Togo, Benin, Nigeria, Central African Republic, and the Democratic Republic of Congo [12]. It is also found in parts of South America, Australia and Asia [29]. Uvaria chamae is sometimes found near streams or in swampy areas of forests. It has low light requirements and can grow in a wide variety of soils and climatic conditions. The species can be found from 0 to 1500 m altitude with an average annual temperature between (22–35 °C) and rainfall between (1250–3000) [8,28]. In Benin, the species is found in fields, fallows, and savannahs [13]. Uvaria chamae is highly valued because of the diversity of uses of its roots, leaves, bark, fruits, seeds in traditional medicine [30], food through the consumption of its ripe fruits by humans and animals [31], firewood (e.g. the stems and branches of the species are used for cooking meals) [18]. For example, in folk medicine, extracts from the root, bark, and leaves are used to treat gastroenteritis, malarial fever, vomiting, diarrhea, sores, sore throat, and inflamed gums [16]. The species is used to treat fevers and has antibiotic properties [11]. The roots of the species have a widespread reputation [32]. The yellow coloration of the root of the species is widely used to dye fabrics which color is also useful in cosmetics [17]. Despite the importance of U. chamae to rural communities, overexploitation of its roots is a major cause of the decline of its populations [19].

2.3. Data collection

2.3.1. Species occurrence

In this study, we used 800 geographic coordinates (longitude and latitude) of U. chamae collected from the field between 2021 and 2022 (Fig. 1). Two (02) other data sources were explored to cover the entire distribution area of the species in West Africa. Thus, 1096 occurrences were collected from the Global Biodiversity Information Facility (GBIF, www.gbif.org) and 172 occurrences from the RAINBIO database [33] accessed in February 2022. The raw occurrences from GBIF and RAINBIO were cleaned by removing duplicates and individuals without geographic coordinates. In addition, records before the year 1985 were also removed to have update data. In whole, 2068 occurrences were used for the models (Fig. 2). Then, the Spatially Rarefy Occurrence tool from SDMtoolbox 2.0 was used in GIS to reduce spatial autocorrelation among occurrences by considering a distance of 1 km [34].

Fig. 1.

Geographical distribution of Uvaria chamae across West Africa and Benin.

Fig. 2.

A: Photo showing the occurrence record of U. chamae in northern Benin. B.C: Photos indicating the identification of unripe fruits on U. chamae in central Benin.

2.3.2. Environmental data

Three (03) sources of variables were used in this study. The variables were related to bioclimatic, edaphic, and topographic parameters. The Nineteen (19) bioclimatic variables were downloaded from the WorldClim platform (www.worldclim.org) version 2.1. Soil layers (texture, and pH) and topographic (slope) variables were obtained from FAO world database (www.web archive.iiasa.ac.at/). The land cover variable was obtained from the DIVA-GIS site (https://www.diva-gis.org/datadown). These variables were clipped using a mask of the study area and resampled to 30 arcsecs (∼1 k) as it is the case for bioclimatic variables. To project the current distribution into the future, two Shared Socioeconomic Pathways (SSPs) of the newly updated future scenarios of CMPI6 were used: SSP245 and SSP585 for 2041–2060 time-horizon (averaged 2050) and (2061–2080) average (2070). In CMPI6, five scenarios were defined for WorldClim: SSP126, SSP245, SSP460, SSP370 and SSP585 [35]. The SSP245 and SSP585 were used in this study because they represent the worse and optimist cases of greenhouse [35] respectively. Thus, the sixth version of the Model for Interdisciplinary Research on Climate (MIROC6) [36] was used to project habitats in the future. The variables selection was conducted in three stages (03). The jackknife test (Supplementary S1) was first calculated using the SDMtune package [37] to identify the variable that most influenced the distribution of U. chamae. Then, Pearson correlation test (Supplementary S2) and variance inflation factor (VIF) (Supplementary S3) were performed using the usdm package [38]. For this purpose, one of each pair of highly correlated variables was removed considering a coefficient value of |0.7| and VIF < 8. In fact, the survival of U. chamae depends mainly on three (03) variables such as bioclimatic data, soil and land cover [39]. The species is mainly found on more or less rocky soils, but also on slightly cool soils in the lowlands and valleys and opened land [39]. Finally, six (06) ecologically sound variables [[40], [41], [42]] were selected to build the models for U. chamae.

2.3.2.1. Modeling approach

Four (04) machine learning modeling algorithms were used to assess the present and future distributions of U. chamae with the biomod2 package [43] in R4.2.0. These include the Generalized Additive Models (GAM) [44], Generalized Linear Models (GLM), Maximum Entropy (MaxEnt) [45] and Random Forest (RF) [46]. The models were run 10 times with 70% of the data for calibration whereas 30% of the data were used for validation. In addition, 1000 background points were automatically generated through the package Biomod2 [43]. To assess the model performance, cross-validation validation has been performed with the receiver operator curve (ROC) and the true skill statistic (TSS) [47]. Models are qualified as performing well when their average score was 1 [47].

2.3.2.2. Spatial analysis and mapping of U. chamae distribution

All the outputs files derived from the four algorithms were imported in geographic information system (GIS) and were adjusted on the Benin scale. Then, 30% of data were used to classify habitats into two (02) classes. Habitats below this value were assumed as unsuitable while all habitats above this value were considered suitable. 30% of the data served for model validation. The habitat dynamics was calculated using the following equation.

$$△=\frac{\left({FA}_i-{CA}_i\right)*100}{{TA}_i}$$

FA corresponds to an extent (e.g., high suitability of a class) in the target horizon under scenario i (here horizon = 2050 and 2070 and i = SSP245 or SSP585); CA corresponds to an extent (e.g., high suitability) in present-day condition; △ is the percentage of extent gained (△ > 0) or percentage of the extent lost (△ < 0) and stable when △ = 0.

3. Results

3.1. Contribution of environmental variables and validation of the models

3.1.1. Selected environmental variables

The results (see Supplementary S4) presented the six selected variables for the modeling. Isothermality (bio3) ranged from 67% to 72%. The results also indicated that the optimal mean temperature of the warmest quarter (bio10) for U. chamae was 25 °C. For the precipitation variables, the species does well under 250 mm with precipitation of the wettest month. However, under the driest quarter, the minimum precipitation (bio14) ranged from 5 to 12 mm. The optimum land cover was between 3 and 15 and the optimal value of the soil type characteristics was 1500–2000.

3.1.2. Performance of the models

The performance of each algorithm is described in Table 1 and Supplementary S5. It appears that RF is identified as the best performing algorithm with the highest value for the validation parameters 0.96 and 0.99 for TSS and AUC respectively. The value of TSS and AUC for GLM were 0.91 and 0.99 respectively, which indicated the second-best performing algorithm. For GAM, the TSS value was 0.87, and the AUC value was 0.98. In opposite, MaxEnt had the lowest performance with a TSS value of 0.77 and an AUC value of 0.92.

Table 1.

Performance of the models based on evaluation metrics.

Algorithms	TSS	AUC
RF	0.96	0.99
GLM	0.91	0.99
GAM	0.87	0.98
MAXENT	0.77	0.92

RF: Random Forest; GLM: Generalized Linear Model; GAM: Generalized Additive Model; MAXENT: Maximum Entropy; TSS: True Skill Statistics; AUC: Received Operational Characteristics.

3.1.3. Contribution of the selected environmental variables to the models

Six (06) variables appeared to be determinant in modeling the distribution of U. chamae (Supplementary S6). These variables are: Isothermality (bio3), average temperature of the hottest quarter (bio10), precipitation of the wettest month (bio13), precipitation of the driest quarter (bio14), land use and soil type. Except for RF, bio14 was the most common important variable for U. chamae with 40% for GAM, 35% for GLM, and 33% for MaxEnt. Land cover was the least contributing variable for all algorithms except for GLM 16%. However, bio3 (65%) with GLM was the most important variable for all variables used. As for bio10, the highest contribution was that of GAM 28% but 22%, 21%, and 9% for RF, MaxEnt, and GLM respectively. Bio13 contributed the most to GLM 33%. It contributed 20%, 15%, and 5% for MaxEnt, RF, and GLM respectively. The soil type variable contributed 12%, 16%, 22%, and 19% for GLM, MaxEnt, RF, and GAM respectively.

3.2. Present-day and future suitable areas of U. chamae in Benin

3.2.1. Present-day distribution of U. chamae in Benin

The results of the analysis revealed that U. chamae is widely distributed across Benin for all algorithms used (Fig. 3). The suitable areas were from the coastal line to latitude 10°42′N encompassing the phytodistricts of the coast, Ouémé-Valley, Plateau, Pobè (Guinean-Congolian zone), Zou, Bassila, and South-Borgou (Sudano-Guinean). However, the RF predicted more suitable areas than the other three algorithms except in the western part of the Zou phytodistrict (Fig. 3d). In contrast, the MaxEnt method predicted less suitable areas for U. chamae compared to other algorithms. With MaxEnt, about a quarter of South-Borgou was suitable (Fig. 3c). The other currently suitable areas for U. chamae were concentrated in the South-Borgou and ranged from Bassila to the coast. However, some patches of unsuitable habitats were found in these areas. The GAM (Fig. 3a) and GLM (Fig. 3b) methods predicted similar suitable habitats. A large part of suitable areas was observed in the southern part of Mekrou-Pendjari, Atacora chain as well as in the phytodistrict of northern Borgou (Sudanian zone). Overall, northern Bassila, southeastern Borgou and western plateau phytodistricts were not suitable for the species in the present-day distribution.

Fig. 3.

Present-day distribution of the suitable habitats of U. chamae across Benin.

3.2.2. Future suitable areas under SSP245 of U. chamae in Benin

The projection of present-day distribution in the future indicated variable tendencies depending on the algorithms, scenarios, and horizons. Indeed, SSP245 scenario and GAM (Fig. 4a–e), and GLM (Fig. 4b–f) algorithms predicted a retraction of suitable areas in the southern part of Mekrou-Pendjari, Atacora chain as well as North Borgou phytodistricts whatever the time-horizon. However, an increase of suitable areas was found in north-Bassila, south-east of South Borgou, and the west of Plateau phytodistrict compared to the present-day distribution. Important retractions were predicted with MaxEnt under SSP245 by 2050 whereas an expansion was predicted by 2070. In fact, by 2050, along the western Bassila, Zou, Plateau, Ouemé Valley, and Coast will not be suitable for the species (Fig. 4c). In contrast, only the southeast of South Borgou was predicted to be unsuitable by 2070 compared to the initial distribution (Fig. 4g). As for RF method (Fig. 4d–h), a very important expansion was expected with SSP245 for the two time horizons. This technique indicated that not only were the initial suitable areas were conserved but also they could be expanded in the southern of Mekrou-Pendjari, Atacora chain as well as North Borgou phytodistrict whatever the horizon.

Fig. 4.

Future suitable areas under SSP245 of U. chamae across Benin: (a–d) future distribution under SSP245 by 2050: (e–h) future distribution under SSP245 by 2070.

3.2.3. Future suitable areas under SSP585 of U. chamae in Benin

The scenario SSP585 (Fig. 5), the algorithms GAM (Fig. 5a–e), and GLM (Fig. 5b–f) predicted similar suitable areas between the two horizons but also to those of the present days. Though, a retraction could occur in the southern of Mekrou-Pendjari, Atacora chain as well as North Borgou phytodistrict whereas an expansion could be expected in the known initial distributions. A similar prediction was also noted between the SSP245 and SSP585 for the two algorithms. The MaxEnt approach predicted an expansion of suitable habitats for the species by 2050 (Fig. 5c) whereas a retraction was expected by 2070 (Fig. 5g). Compared to SSP245 (2070), highly unsuitable areas for the species are found in the phytodistricts of Borgou Sud, Zou and Plateau with SSP585 in 2070. In addition, many unsuitable areas under SSP245 (2050) (especially the north part of the South Borgou) became suitable under SSP585 (2070). The RF method (Fig. 5d and h) predicted an expansion for U. chamae with the two horizons compared to the initial distribution. Between the two scenarios, the distribution of suitable areas was similar with an exception in Atacora chain where a retraction was found for SSP585 by 2070 (Fig. 5h). When overlapping the protected area (PAs) networks with the initial and future distribution of the species, the results indicated that only a limited portion of PAs appeared to be effective in the conservation of the species. However, according to MaxEnt the species could be threatened with decline for the SSP245 scenario by 2050.

Fig. 5.

Future suitable areas under SSP585 of U. chamae across Benin: (a–d) future distribution under SSP585 by 2050; (e–h) future distribution under SSP585 by 2070.

3.2.4. The dynamic of suitable habitats of U. chamae

Findings of this study showed the trend of habitat suitability. In fact, under SSP245 (2050), the tendencies of suitable areas were 1.5%, 1.94%, −29.12% and 17.84% respectively with GAM, GLM, MaxEnt, and RF methods (Fig. 6a). Likewise, by 2070 (Fig. 6b), an increase of (2.5%) was expected using GAM and GLM but it was 17.5% for RF. However, a slight decrease was expected with MaxEnt approach compared to the initial distribution with 2.48%. As for SSP585 by 2050, the tendencies showed a gain for U. chamae whatever the algorithms with 6.83%, 7.58%, 0.82%, and 20.95% respectively for GAM, GLM, MaxEnt, and RF (Fig. 6c). Compared to SSP585 (2050), a decrease was globally noted with SSP585 (2070) (Fig. 6d). Indeed, though the species could gain 20.07% and around 3.6% with respectively RF and (GAM, and GLM) a slight decrease 1.01% was expected with the MaxEnt method.

Fig. 6.

Percentage of habitat losses under the two scenarios using the four (04) algorithms: (a) SSP245 by 2050; (b) SSP245 by 2070; (c) SSP585 by 2050 and (d) SSP585 by 2070. The losses were computed by combining the present-day output with each future scenario. These tendencies express the spatial dynamic between an SSP and the present-day.

4. Discussion

4.1. Performance of the model

The present study aimed to project current distributions and model future favorable areas for the distribution and conservation of U. chamae at the scale of Benin. All models considered in this study have a good performance in predicting the potential distribution areas of U. chamae as their AUC is greater than 0.70 [48] and their TSS values were close to 1 [49]. Furthermore, it appeared that the RF model performed better in determining the distribution areas of the species followed the GLM, GAM and MaxEnt. These findings suggest that the mapping of suitable habitats for U. chamae in Benin can be done with confidence using RF, GLM, and GAM models and to a lesser extent with MaxEnt. Such trend support a recent study by Ref. [50] who have already demonstrated the performance of the RF, GLM, GAM and MaxEnt models in examining the potential impacts of climate change on Caesalpinia bonduc (L.) Roxb in Benin (West Africa). Similarly, Wang et al. [51] have been compared four distribution models (MaxEnt, GLM, BRT) to obtain similar results. In addition, it was found that the models considered predicted well the favorable areas for the species in the present and in the future climate conditions. This reflects the fact that the known presence of U. chamae in the field is similar to the mapped favorable areas of the species supporting recent findings [52,53].

4.2. Contribution of environmental variables to the models

The results showed that the distribution of U. chamae is influenced by a range of environmental factors. Globally, the distribution of the species is influenced by both climate and soil type. This corroborates the finding of Assogba et al. [21] who showed that soil factors and climatic variations contribute to the distribution of the species in West Africa. Similarly, Dimobe et al. [54] in Burkina Faso mentioned that rainfall and temperature are the main bioclimatic factors influencing the distribution of suitable habitats for plant biodiversity. According to Lompo et al. [55], the balanced influence of environmental factors in the distribution of plant biodiversity is in line with the concept that water availability and temperature variation are important factors in controlling the occurrence of plant species in subtropical and tropical African regions.

4.3. Potential effects of future climates on the distribution of U. chamae

U. chamae is a species with high medicinal potential for the local community [16]. Species Distribution Models (SDMs) are often used to elucidate natural resource management decisions [56]. Assessing the impact of climate change on the distribution of U. chamae has shown across models, scenarios, and time horizons a considerable shift in suitable habitats for the species. Indeed, the prediction of the current distribution of U. chamae showed that the species is widespread in two phytogeographic zones (Guinean-Congolian and Sudano-Guinean) and to a lesser extent in the third phytogeographic zone (Sudanian, especially in the Atacora chain and Mékrou Pendjari) in Benin much more according to the RF model followed by the GLM, and GAM. For the Maxent model, a large part of the Sudano-Guinean zone is currently not suitable for the species. While in the literature and according to our field observation, the species is not occurring in the Sudanian zone except the western part of the Atacora chain phytodistrict where a few rare individuals of the species can be found (Fig. 1, Akoègninou et al. [13]). Lompo et al. [55] found similar results in their study on “Ximenia Americana L. in Burkina Faso” which stated that the fundamental niche of a species is often larger than its realized niche. Because the species is in competition with other species by their presence and interaction, it occupies the realized niche, which is part of the fundamental niche [55]. A good prediction of future environmental conditions favorable to conservation of U. chamae is observed especially with the RF model, followed by GLM and GAM regardless of the scenarios. These result are consistent and compatible with the ecology of the species [13]. Variations in predictions under future climate conditions have been highlighted in previous studies [24,54]. With Maxent model, a significant decrease in the suitable areas for the species conservation rather was projected in the future. Our results are consistent with those of Lompo et al. [55] on Ximenia americana L., where climate change would lead to a decrease in the suitable area of X. americana by 2070 according to the MaxEnt model. This finding supports the hypothesis that climate change can alter the range of plant species [57]. It is therefore important to emphasize that the Guinean-Congolese and Sudano-Guinean zones are favorable for the conservation for the species, while efforts must be made for the conservation of the species in the Sudanian zone, because the species develops much more in the lower altitudes. For example, climatic and pedological factors condition the ecology of the species (Supplementary S4), while the Sudanian zone has climatic and soil conditions that do not meet the requirements of the species. In addition, the assessment of the conditions favorable to the distribution of the species in protected areas showed that only a limited number of protected areas offered good conservation potential for the species. However, many areas favorable to the species are located outside of protected areas (Fig. 3). This poses a threat to the species due to the expansion of human activities into areas outside of protected areas. In addition, the vulnerability of protected areas by climate variation in the future is reported by Parry et al. [58] and their fragmentation by Houehanou et al. [59]. It is therefore important to introduce the species into agroforestry systems to ensure its sustainable conservation.

4.4. Sustainable management strategies of U. chamae

Management strategies for a species depend largely on the ecosystem in which it occurs, whether in-situ or ex-situ, and are influenced by the conservation status at the national level [50]. In Benin, U. chamae is a species that occurs in the wild and its roots are overexploited to satisfy the health needs of the population and its individuals are cut in agricultural fields for crop establishment [19]. However, Lawin et al. [60] reported that the species is threatened and field conservation is the measure adopted by the local population. It is therefore important to introduce the species into agroforestry systems and to encourage local communities to cultivate it in order to limit its possible extirpation as previously suggested for other declining agroforestry species in West Africa [21,55]. In addition, this study showcases that the bush banana is widely distributed in Benin in the Guineo-Congolian, Sudano-Guinean zones and to a lesser extent in the Sudanian zone. However, it is important to implement a sustainable management plan for the species in these biogeographic zones. Findings showed that both the Guineo-Congolian and Sudano-Guinean zones of Benin are identified as suitable zones for the conservation of U. chamae under current and future climates. These biogeographic zones will offer the species favorable climatic conditions for its growth and production. Thus, these areas can be used for the domestication and sustainable conservation of the species in Benin.

5. Conclusion

This study determined the current distribution and predicted future habitats suitable for the sustainable management of U. chamae in Benin. Nowadays, modeling algorithms are well adapted tools to provide useful information on species of interest for better biodiversity conservation. These algorithms provide details on the distribution areas of the species, which makes it possible to anticipate the variations that the species may experience in order to offer leads for possible conservation measures. It has been shown in this study that the distribution of U. chamae is globally determined by the climatic and soil conditions. Similarly, the current distribution map of the species showed that the bush banana is currently present in almost all of the Guinean-Congolian and Sudano-Guinean zones and, to a lesser extent, in the Sudanian zone. The modeling results showed that parts of the Guinean-Congolian and Sudano-Guinean zones are and will remain suitable for the conservation of the species for the RF, GAM and GLM models, while for Maxent, the suitable habitats for the species will decrease with the SSP245 and SSP585 scenarios. These results provide scientific support for planning and decision-making tools for the conservation of U. chamae in Benin. It should be noted that human pressures and climatic variations will have a significant impact on the distribution of U. chamae in Benin under the scenarios and models studied. Management and conservation measures must be implemented through sustainable development policies to strengthen endogenous knowledge and ensure the sustainable use of U. chamae in Benin. Although modeling the distribution of the species has allowed us to estimate its habitat suitability, it cannot predict the response of the species in terms of adaptation, dispersal, migration or extinction. Therefore, future studies of the species should take into account its phenotypic plasticity and local adaptation strategies, as conservation of a species involves adaptive measures.

Author contribution statement

Emilienne Houévo Daï: Conceived and designed the experiments; analyzed and interpreted the data; wrote the paper.

Juliano Sènanmi Hermann Houndonougbo; Amadé Ouédraogo: Analyzed and interpreted the data; wrote the paper.

Rodrigue Idohou: Performed the experiments; wrote the paper.

Romain Glèlè Kakaï: Contributed reagents, materials, analysis tools or data; wrote the paper.

Stefan Hotes: Analyzed and interpreted the data; wrote the paper.

Achille Ephrem Assogbadjo: Contributed reagents, materials, analysis tools or data; Analyzed and interpreted the data.

Funding statement

Mrs Emilienne Houévo DAÏ was supported by Organization for Women in Science for the Developing World [3240314465].

Data availability statement

Data will be made available on request.

Declaration of competing interests

The authors declare that they have no known competing financial interests or personal relationships that could have appeared to influence the work reported in this paper.

Appendix A. Supplementary data

The following is the Supplementary data to this article: