Senegalia mellifera (Benth) Seigler & Ebinger., Dichrostachys cinerea (L.) Wight & Arn. and Terminalia sericea Burch. Ex DC., are three important bush encroacher species that contribute to the well-known ecological process named “thicketization” in Southern Africa. This issue has persisted for many years, impacting species distribution, plant communities, soil, and fauna dynamics. According to climate change projections, Southern Africa is expected to become drier and warmer in future scenarios, creating favourable conditions for proliferation of bush encroacher species. MaxEnt is a general-purpose machine learning method widely utilized in various ecological and biological scenarios to predict the potential suitable habitat of species. This is achieved by incorporating presence-only occurrence records and bioclimatic, and topographic variables. The analysis was performed in a Geographic Information System based on the current and future suitable areas for the respective Shared Socioeconomic Pathways (SSP) scenarios according to INM-CM5-0, UK-ESM1-0-LL and MPIES-M1-2-HR climate models. This was done to assess the potential effects of climate change on the distribution patterns of bush encroacher species. Model performance was assessed by the area under the curve (AUC) of the receiver operator characteristic (ROC), with 0.836, 0.822 and 0.738 as results to Terminalia sericea, Dichrostachys cinerea and Senegalia mellifera respectively. The current results show that Senegalia mellifera presents a habitat suitability of 56% (1,460,353 km²) of the total area, while Terminalia sericea and Dichrostachys cinerea have suitability over 37.9% (996,168 km²) and 43.9% (1,154,645 km²) of the area, respectively. These findings indicate that precipitation and temperature variables are the most important factors in explaining the spatial distribution of the bush encroacher species, predicting a future increase between 8–29.4%, 2.8–24%, and 3–24.2% for Senegalia mellifera, Terminalia sericea, and Dichrostachys cinerea respectively. Furthermore, each species has its own set of important variables and different ecological behaviour patterns. These results imply that an improved understanding of the response of woody species to a changing climate is important for managing bush encroachment in savanna ecosystems.

Bush encroaching, climate change, Dichrostachys cinerea, habitat suitability, MaxEnt, Senegalia mellifera, Terminalia sericea, thicketization

“Thicketization’’, “bush encroachment” or “woody plant encroachment” are different concepts used to explain the natural phenomenon characterized by an increase in density of woody plant bushes in the different savanna and grassland land uses (Archer et al. 2017; Schreiner-McGraw et al. 2020; Soubry and Guo 2022; Basant et al. 2023). The concept was first introduced by Bews (1917), describing an increase in the density of species of the genus Acacia sp., that refers to the one now classified as Vachellia sp. Or Senegalia sp. (O’Connor et al. 2014). This phenomenon has emerged as one of the most important problems of the rangelands. This impact will increase in the future, affecting the species distribution and altering the interactions between plant communities, soil, and fauna dynamics (Van Wilgen et al. 2022; Weber-Grullon et al. 2022).

Referring to the impact on rangelands and its effects on species distribution and plant-soil-fauna interactions, it aligns with the modern two-layer theory of savanna ecology (Kulmatiski and Beard 2013). According to this theory, savannas are characterized by a distinct vertical structure consisting of an upper layer dominated by trees and a lower layer dominated by grasses and herbaceous plants. If the herbaceous layer is over exploited, it loses its competitive edge and becomes unable to efficiently utilize water and nutrients, increasing the infiltration rate into the subsoil, favouring the growth of shrubs and trees, enabling them to assume the dominance (De Klerk 2004; Kulmatiski and Beard 2013). In this context, the encroachment of certain plant species or the disruption of natural disturbance regimes can lead to changes in the composition and structure of both the upper and lower layers of the savanna ecosystem (De Klerk 2004; Kulmatiski and Beard 2013). However, this theory is still inconclusive and recent literature suggests that it may be an oversimplification of the root system patterns among savanna plants overlooking the issue of plasticity in root development and response to environmental conditions (Kulmatiski and Beard 2013; Nakanyala and Hipondoka 2020).

It will pose a challenge to identify a singular factor as the exclusive reason for the disruptions linked to bush encroachment, particularly considering the spatial correlation of all influencing environmental variables (Kgosikoma and Mogotsi et al. 2013; Archer et al. 2017). The explanations of the main causal factors of this dynamic include changes in fire regimes, livestock grazing, loss of browsing herbivores, intensive harvesting of woody plants for fuel, pressure, nutrient availability and climate change (Kgosikoma and Mogotsi 2013; O’Connor et al. 2014; Turpie et al. 2019). The uncontrolled proliferation of woody plants in arid and semi-arid grasslands suggests the importance of broad-scale factors, such as climate change and increases in atmospheric CO₂. These southern Africa regions that are affected by the bush encroachment coincide with high-risk areas due to rising temperatures in the forthcoming decades (Beck et al. 2018; Engelbrecht 2019).

However, many studies have shown that the spread of bushes in semi-arid areas can also have substantially effects (Archer et al. 2017; Cai et al. 2020; Li et al. 2023b). Depending on the species, the soil under shrubs is often characterised by a higher organic substrate, seeds and nutrients than the soils without patches with shrub vegetation, following the ‘fertile island’ effect (Eldridge and Ding 2020). Furthermore, there is growing evidence that soil under shrubs and with controlled grazing was found to be more stable and had higher soil nutrient content, creating more favourable conditions for further vegetation development (Stafford et al. 2017 Cai et al. 2023). These favourable conditions for the development of biodiversity, raw materials, improved soil quality, etc., may coin the ecosystem service term (IPBES 2019).

The central issue in the woody plant encroachment is the role of the water and the landscape water balance. Several studies have shown that bush encroachment increases evapotranspiration and may lead to excessive water use contributing to more severe droughts (Grygoruk et al. 2014; IPCC 2021; Schick and Ibisch 2021). On the other hand, woody vegetation can improve ecosystem services with good management plans (Soliveres and Eldridge 2014; Stafford et al. 2017). According to the literature, soil carbon, microbial activity, nitrogen storage and above-ground net primary productivity were significantly higher under the trees/shrubs layer compared to herbaceous layer (Montané et al. 2010; Li et al. 2016; Turpie et al. 2019). Despite the above positive points shown above, the overgrowth of some shrub species such as Senegalia mellifera, which produces a very competitive lateral root system, displaces the herbaceous cover, decreasing biodiversity and drastically altering the habitat (O’Connor et al. 2014). This is the reason why it is so important to know the main areas of bush encroachment zones and where these species are distributed. This is important for good rangeland management that maintains a balance between herbaceous and shrub/shrub cover, avoiding the economic and ecosystem services losses (Briske et al. 2017; IPCC 2021).

Ecological Niche Modeling (ENM) has become a fundamental tool for understanding species’ ecological requirements and predicting their distribution patterns, particularly in light of current and future environmental changes (Hirzel and Lay 2008; Franklin 2013). This approach allows researchers to identify the environmental conditions suitable for species survival and forecast potential shifts due to climatic factors Various methodologies have been developed within ENM, each with its unique characteristics and specific applications, including generalized linear models (GLMs), generalized additive models (GAMs), genetic algorithm for rule set production (GARP), random forest models (RF), maximum entropy models (MaxEnt), and artificial neural network (ANN) models have been used to predict the distributions (Stockman et al. 2005; Elith et al. 2006; Chieverini et al. 2023; Munna et al. 2023a). Species distribution modelling has demonstrated its potency in conservation biogeography, serving as a valuable tool for predicting or extending knowledge into geographical areas where information is lacking (Franklin 2013). Nevertheless, an important point that these models do not consider how species interact in their environment in the presence of other species, i.e., biotic interactions (Araújo and Luoto 2007). It has been shown that interspecific interactions, such as competition for resources, predation and symbiosis, change in one way or another the possible distribution of species in their environment (Pearson and Dawson 2003; Abati et al. 2020). A further factor not considered in such models is the evolution and the genetic adaptation of species (Guisan and Zimmermann 2000). The potential for rapid genetic evolution is immense, occurring mainly in insects and plants, expanding their geographical limits of distribution by becoming more dispersive and/or adapting to incessant climate change (Maron 2004; Garnas 2018; Catullo et al. 2019).

Bioclimatic factors in the study area are projected to change in unprecedented ways. Temperatures in Namibia, Botswana and western South Africa are expected to experience significant warming, ranging from 0.2–0.5 °C per decade, leading to an increase in the number of hot days (Beck et al. 2018; Engelbrecht 2019; Gornott 2023). Precipitation is projected to experience overall reductions in mean annual rainfall by mid-century, showing that the semi-arid and arid areas are likely to become drier than more humid areas. Furthermore, in spite of the projected decreases in rainfall, extreme rainfall events are going to increase over most of the central-east coast of South Africa in low-high mitigation scenarios. (Beck et al. 2018; Engelbrecht 2019; IPCC 2019; Gornott 2023).

The encroachment of both native and non-indigenous invasive woody species, exacerbated by the impacts of climate change, is causing significant environmental, social, and economic harm (De Klerk 2004; Archer et al. 2017). The SteamBioAfrica project aims to demonstrate the application of a superheated steam method to convert invasive woody biomass into environmentally friendly solid biofuel and water across Namibia, Botswana and South Africa. By promoting optimal shrub harvesting and land management techniques and establishing a market for the biomass produced, the initiative seeks to stimulate effective land restoration. This, in turn, will create promising opportunities for households and industries, benefiting both rural and urban environments.

The selection of three specific species—Terminalia sericea, Dichrostachys cinerea, and Senegalia mellifera—is based on their generalist behaviour, invasiveness, and biomass potential in semi-arid areas, aligning closely with the project’s objectives (Fig. 3) (Pedroso and Kaltschmitt 2012; Moyo et al. 2015; Ofentse et al. 2022; Neethling 2023). Terminalia sericea, known as “silver cluster-leaf,” thrives along mid-slope seep-lines, forming dense clusters and yielding substantial biomass (Moyo et al. 2015). Widely used as fuelwood and for medicinal purposes in southern Africa, it is a key focus of the initiative (Moyo et al. 2015). Dichrostachys cinerea, or “sickle bush,” possesses attributes suitable for clean energy generation and medicinal uses due to its rapid growth, high biomass density, and low burning emissions (Pedroso and Kaltschmitt 2012; Ofentse et al. 2022). Senegalia mellifera, commonly known as ‘black thorn’, is the most problematic species causing significant economic losses due to the bush thickening (Neethling 2023). With diverse uses as an energy resource, medicinal plant, and livestock feed, it is another crucial target for the project (Neethling 2023).

The study aims to elucidate the likely distribution of these three species, developing habitat suitability maps based on climate change scenarios proposed by the IPCC in 2021 following an accepted and well-known methodology (Fig. 1). This will increase knowledge about the behaviour patterns of these three invasive shrub species according to the proposed variables. Thus, soil management plans will be strengthened, adapted and improved, in addition to identify the main areas where the species can grow easily, providing additional knowledge for the exploitation of the resources. For instance, transforming the biomass of highly bush encroached areas into clean biofuel and condensate containing water and biochemicals with economic potential during the SteamBioAfrica project. Furthermore, we will use the model’s results to verify and compare the state of bush encroachment in southern Africa with existing literature, as well as its future outlook.

Figure 1. 

Schematic methodology representation.

A total of 57 distribution maps were acquired, with three generated under current conditions and 54 under conditions of climate change. The AUC values spanned from 0.738 to 0.836, denoting their classification performance as favourable (0.8 < AUC < 0.9). Specifically, Terminalia sericea with a mean AUC of 0.836, Senegalia mellifera with 0.738 and Dichrostachys cinerea with 0.822.

The outputs from the model were reclassified according to the common threshold established. The next tables show the area of the habitat suitability in km² and % for each climate model, climate scenario and in the different periods.

Habitat suitability for Terminalia sericea is mainly associated with annual precipitation (bio12; 55.2%) and annual mean temperature (bio1; 31.2%) (Table 2). It is mainly concentrated in the north-central, north-western, and east-central areas, with the species virtually absent from southern Namibia and north-western South Africa, and its habitat suitability restricted to the northern and eastern limits of the study area (Fig. 4). Terminalia sericea improves its habitat suitability compared to the current situation in all scenarios and time periods, with a maximum area gain of 24% in 2061–2080 for the UK-ESM1.0-LL climate model in the SSP585 scenario. The lowest area gains of 3% is in 2041–2080 for the MPI-ESM1.2-HR climate model for the SSP245 scenario (Table 7). Additionally, the categories of high suitability register their maxima in the SSP585 emission scenarios, with the UK-ESM1.0-LL climate model recording the highest value in the period 2061–2080, with up to 32.4% high suitability for the species. On the other hand, the INM-CM5-0 climate model shows lower values, reaching up to 23% for the same period and emission scenario, followed by the MPI-ESM1.2-HR with 25.5% under the same conditions (Table 4).

Prediction of habitat suitability for Senegalia mellifera was widespread and mainly associated with mean temperature of wettest quarter (bio8; 50.9%), annual mean temperature (bio1; 27.4%) and annual mean precipitation (bio12: 11.1%) (Table 2). The distribution of this species is the most habitat-adapted, occurring over much of central southern Africa, and is absent from the entire west coast of Namibia and the wetter areas of southern South Africa (Fig. 4). Senegalia mellifera improves its habitat suitability in all scenarios, with a maximum increase of 29.4% in 2061–2080 for UK-ESM1.0-LL climate model in the SSP585 scenario as happened with Terminalia sericea. The lowest gain is registered in the 2041–2060 period, with the INM-CM5-0 climate model, to the SSP245 scenario, with only an 8% of increase in distribution (Table 7).

Additionally, the categories of high suitability register their maxima in the SSP585 emission scenarios, with the UK-ESM1.0-LL climate model recording the highest value in the period 2061–2080, with up to 58.6% high suitability for the species. On the other hand, the INM-CM5-0 climate model shows lower values, reaching up to 34.9% for the same period and emission scenario, followed by the MPI-ESM1.2-HR with 45.7% under the same conditions (Table 5).

For Dichrostachys cinerea, the most related variables are precipitation in the driest month (bio13; 41.3%), mean temperature in the coldest quartile (bio11; 21.8%) and seasonality of temperatures (bio4; 18.9%) (Table 2). The distribution of Dichrostachys cinerea has the greatest increase in suitability with respect to the unsuitable areas in the current scenario. Distribution includes southwest of Botswana and northwest of South Africa, with high habitat suitability in the northern half of Namibia and Botswana and northeast of South Africa (Fig. 4). Dichrostachys cinerea improves its habitat suitability according to model variables with a maximum area gain of 24.2% during 2061–2080 period, for UK-ESM1.0-LL climate model in the SSP585 scenario as the other species. The lowest gain corresponds to the MPI-ESM1.2-HR climate model, in the SSP245 scenario during the 2041–2060 period, with only 3% of increase of suitable areas (Table 7).

Furthermore, high suitability categories reach their peaks in the SSP585 emission scenarios, with the UK-ESM1.0-LL climate model registering the highest value for the 2061–2080 period, showing up to 47.6% high suitability for the species. Conversely, the INM-CM5-0 climate model records lower values, with a maximum of 29.7% for the same period and emission scenario, followed by the MPI-ESM1.2-HR at 36.53% under the same conditions (Table 6).

Figure 4. 

Habitat suitability in the current scenario. This map illustrates variations in habitat suitability for the species, featuring distinct colors to delineate areas deemed unsuitable or suitable for the species. The color spectrum is utilized to convey different levels of suitability: areas in dark green represent the highest suitability, while light green signifies low suitability. Unsuitable areas for the species are depicted in grey.

Figure 5. 

Habitat suitability area per scenario Senegalia mellifera. This map illustrates variations in habitat suitability for Senegalia mellifera, featuring distinct colors to delineate areas deemed unsuitable or suitable for the species. The color spectrum is utilized to convey different levels of suitability: areas in dark green represent the highest suitability, while light green signifies low suitability. Unsuitable areas for the species are depicted in grey. The map serves as a visual guide to understand how the suitability of habitat changes across different scenarios and climate models for Senegalia mellifera.

Figure 6. 

Habitat suitability area per scenario Terminalia sericea. This map illustrates variations in habitat suitability for Terminalia sericea featuring distinct colors to delineate areas deemed unsuitable or suitable for the species. The color spectrum is utilized to convey different levels of suitability: areas in dark green represent the highest suitability, while light green signifies low suitability. Unsuitable areas for the species are depicted in grey. The map serves as a visual guide to understand how the suitability of habitat changes across different scenarios and climate models for Terminalia sericea.

Figure 7. 

Habitat suitability area per scenario Dichrostachys cinerea. This map illustrates variations in habitat suitability for Dichrostachys cinerea, featuring distinct colors to delineate areas deemed unsuitable or suitable for the species. The color spectrum is utilized to convey different levels of suitability: areas in dark green represent the highest suitability, while light green signifies low suitability. Unsuitable areas for the species are depicted in grey. The map serves as a visual guide to understand how the suitability of habitat changes across different scenarios and climate models for Dichrostachys cinerea.

This study predicted habitat suitability for the species Senegalia mellifera, Dichrostachys cinerea, and Terminalia sericea in the southern African region. Distribution predictions were carried out under three different climate models (INM-CM5-0, UK-ESM1.0-LL and MPI-ESM1.2-HR), three emission scenarios (SSP245, SSP370 and SSP585) for the periods 2041–2060 and 2061–2080, highlighting changes relative from the current state. We used the MaxEnt methodology to predict areas at higher or lower risk of bush encroachment and to identify the key drivers influencing the distribution of these species. This approach provides valuable insights to support future tasks related to climate change and its underlying drivers that affect the distribution patterns of plant species. All models outperformed chance, achieving AUC values greater than 0.736 in all cases, with the Terminalia sericea model (mean AUC = 0.836) achieving the highest value.

In general terms, all three species show a pattern of increasing habitability in previously uninhabited areas according to climate scenarios and time periods. Habitability rises as the expected concentration of greenhouse gasses increases for each scenario. The SSP245 scenario (medium pathway of future greenhouse gas emissions) assumes that climate protection measures are being taken, while registering the lowest increase in habitat suitability among the climate change scenarios. Conversely, the SSP585 scenario (less optimistic), which involves drastic emissions increase, correlates with the highest expansion in habitat suitability for all three species.

The results confirm that variables related to rainfall and temperature are the most important for modeling the distribution of these bush encroacher species. Dichrostachys cinerea and Terminalia sericea showed similar changes in distribution between the current and climate change scenarios, with the species expanding their range limits mainly in South Africa and Botswana. Meanwhile, Terminalia sericea it does not have a big spread in Namibia, but it increases in its habitat suitability. However, Dichrostachys cinerea incresases the distribution along the north of the case study, incresing the habitat suitability in the future scenarios. In the other hand, Senegalia mellifera does not respond in the exact same way. The model results record an unsuitable area of 14.6% in the SSP585 scenario (2061–2080), which in comparison to the other species (Terminalia sericea: 38.1%; Dichrostachys cinerea: 31.9%) indicates that it will be the most generalist and widespread of the three species in scenarios with drastic emissions.

The results in the percentage change between the current scenario and future scenarios also support the idea that Senegalia mellifera will cover the most area, with a maximum increase in suitable distribution area of up to 29.4%, obtaining a maximum suitability of 85.4% of the total area. This is followed by Dichrostachys cinerea, with a 24.2% of maximum increase in habitat suitability, with a maximum suitability of 68.1% of the total area. Lastly, Terminalia sericea shows a 24% of maximum increase, with a suitability of 61.9% of the total area. All these results indicate that the UK-ESM1.0-LL climate model and the SSP585 emission scenario for the period 2061–2080 register the highest bush encroachment. The increase in habitat suitability for the species is highest in this climate model compared to INM-CM5-0 and MPI-ESM1.2-HR, which show lower average habitat suitability.

The figure that summarises changes in habitat suitability according to the proposed scenarios shows differences and similarities in the distribution patterns of the three species (Figs 5–7). For both Terminalia sericea and Dichrostachys cinerea, an increase in habitat suitability is observed in the northern part of the study area, which becomes more pronounced in the UK-ESM1.0-LL and INM-CM5-0 models, also indicating an increase in suitability across South Africa from east to west. The main difference between these two species lies in their potential distribution in Botswana, where Dichrostachys cinerea experiences a significant increase flowing from the north to the southeast, particularly in the scenarios of the UK-ESM1.0-LL model. Terminalia sericea, on the other hand, maintains low suitability, which seems to increase over time and with higher emission scenarios. In the case of Senegalia mellifera, Fig. 5 shows that habitat suitability is quite high, covering a large part of the study area, with high suitability increasing under higher emission scenarios. The UK-ESM1.0-LL and MPI-ESM1.2-HR models show the most significant changes in habitat suitability for this species, with a large portion of Botswana becoming highly suitable in both models.

According to the model outputs, these findings imply that Senegalia mellifera is the most potentially bush encroacher for the near future changes, with a larger distribution area and better habitat suitability in most of the scenarios.

This statement does not rule out or ignore the invasive potential of the remaining species due to the climate changes.

These results are similar to findings reported in other studies concerning various species. Based on studies involving growth rings and stem diameter (Shikangalah et al. 2021; Shikangalah et al. 2022), the relationships of Dichrostachys cinerea with wetter areas and the distinct adaptation strategy of Senegalia mellifera to drier regions have already been described. This specie is known for its tolerance to a wider range of environmental conditions, particularly in terms of water availability and temperature extremes (Guajardo et al. 2010; Shikangalah et al. 2021; Shikangalah et al. 2022). This flexibility allows it to expand into new areas more readily under changing climate conditions, leading to a higher predicted increase in suitable habitat compared to Terminalia sericea and Dichrostachys cinerea. These other two species may have more specialized ecological niches, with narrower ranges of tolerance for key climatic factors.

These findings support our model outputs and underscore the importance of the variables considered. In the current distribution scenarios, the model outputs generate distribution maps highly resemblant to the existing maps describing species presence boundaries (De Klerk 2004; Turpie et al. 2019; Marggraff and Venter 2020).

According to present and future Köppen-Geiger climate classification maps by Beck et al. (2018), southern Africa will experience an increase in arid and semi-arid climate areas, especially in South Africa, relating to our species distribution results. Changes in African savanna rangelands were already predicted years ago. Authors such as Midgley et al. (2005) and Tietjen et al. (2009) already expected a decrease in herbaceous vegetation cover and an increase in shrub cover in drylands in Africa caused by the CO₂ levels, rainfall, and temperature changes. According to the most recent studies, an annual increase in shrub covers of up to 0.27% yr-1 confirming global greening trends in Africa (Venter et al. 2018; Saha et al. 2015) a significant increase in vegetation optical depth (VOD) mainly in southern Africa, West and Central Africa was found (Wei et al. 2019) strengthening the idea of the continuous increase in bush encroaching, supporting our future findings in these species. All these studies conclude that the bioclimatic variables in our study are related to the increase in shrub cover throughout this region of Africa, in addition to other hardly measurable factors as far as overgrazing or fire suppression.

Certainly, socio-ecological processes and biotic interactions pose a significant challenge to quantification and modelling in scientific research. Numerous intricate interactions between species and their environments elude precise measurement, and even models exhibiting impeccable AUC values and statistical metrics cannot ensure absolute perfection. Furthermore, the broad geographical scope of our estimates introduces inherent uncertainty into the assessment of climatic and physical suitability.

Species distribution models have emerged as invaluable tools for implementing conservation strategies or conducting studies on the impacts of climate change, particularly in expansive regions characterized by notable physical and bioclimatic gradients. However, these models predominantly rely on interpolation methods and lack a robust foundation in ecological process theory. Thus, there remains substantial room for improvement, with the development and implementation of novel methodologies being imperative for enhancing our understanding of biotic interactions.

Additionally, it is crucial for biogeographers and ecologists to recognize that the pattern of species association in joint species distribution models is influenced not only by actual biotic interactions but also by shared habitat preferences, mutual migration history, phylogenetic background, and collective responses to absent environmental drivers (Dormann et al. 2018). Addressing the dearth of this type of data presents a formidable challenge, but it promises to greatly enhance species modeling and deepen our ecological understanding of them.

The objectives of this study were to investigate the potential impacts of climate change on the spatial distribution of these three bush encroachers, important woody species in southern African rangelands, and compare these findings with the existing literature. This study used INM-CM5-0, UK-ESM1.0-LL and MPI-ESM1.2-HR climate models, three emission scenarios (SSP245, SSP370 and SSP585) for the periods 2041–2060 and 2061–2080, combined with georeferenced species occurrence data. The results of the study indicate that the bioclimatic habitat of these species is expected to change significantly in response to climate change in the southern African savannas. Thus, supporting other studies that suggest the greening trends in southern Africa due to woody plant cover change. The main patterns of expansion are related to expected changes in temperature and precipitation over the century. According to the results shown above, Terminalia sericea, Dichrostachys cinerea and Senegalia mellifera improve their habitat suitability in almost all scenarios and time periods compared to the current one. Current results showed in Table 3 indicate that Senegalia mellifera exhibits the highest percentage of suitable area at 56%, though it has a smaller area of high suitability, at 3.1%. Closely following is Dichrostachys cinerea, with 43.9% suitable area and 5.2% highly suitable area. Terminalia sericea is not far behind, with 37.9% suitability and the highest percentage of highly suitable area at 5.6%.

Senegalia mellifera has the biggest increase in the area of high habitat suitability, with a 29.4% increase in the 2061–2080 time period for SSP585 scenario compared to the current scenario, making it the species with the best conditions for establishment throughout southern Africa in the future. Dichrostachys cinerea follows with a 24.2% and 24% for Terminalia sericea. In summary, Senegalia mellifera emerges as the most stable species covering the largest area across all scenarios. Conversely, Dichrostachys cinerea and Terminalia sericea exhibit a similar distribution pattern, with Senegalia mellifera being the most favoured species in future scenarios.

Finally, to contribute to the improvement of savanna rangelands management, we encourage developing new studies to enable innovative management and adaptation plans in line with ongoing climate change in the southern Africa regions. Finally, we recommend conducting validation studies at local scales, analyzing shrub regeneration and its growth in areas that were found to be highly suitable in future scenarios. This should be supported by localized studies measuring the differences between dense and sparse coverages in relation to the ecosystem. This would be very useful for future soil management plans and would support many issues addressed in the Africa 2063 agenda.

We want to thank Colin Lindeque and their team for graphical support of the Namibia study site and information related to bush encroached zones in the country. Also, thanks to Fernando Alonso-Martin, who read the manuscript a lot of times and gave statistics tips. SBA project , supported by the European Commission’s Horizon 2020 Programme. Grant agreement ID: 101036401.

Conceptualization, B.G.J.; methodology, B.G.J.; software, B.G.J.; validation, J.B.G., P.G.F and V.B.F.J; formal analysis, J.B.G., P.G.F and V.B.F.J.; investigation, B.G.J.; data curation, B.G.J.; writing—original draft preparation, J.B.G.; writing—review and editing, P.G.F., V.B.F.J., C.N.M.J; supervision, P.G.F., V.B.F.J and M.A.F.

The data used in this article are openly accessible through the following link on the Zenodo repository: [https://zenodo.org/records/13735837]. This ensures transparency and allows for further analysis or replication of the study by other researchers.