Savanna rangelands have experienced widespread degradation due to bush encroachment, raising significant concerns among conservationists and rural communities. In the context of climate change, these ecosystem shifts are likely to intensify, especially in South Africa's semi-arid regions. Understanding the impacts of climate variability and change on species distribution within these rangelands is crucial for mitigating further ecosystem disruption. Environmental factors, along with climatic variables, can accelerate the process of bush encroachment, threatening both biodiversity and land use. Early identification of areas vulnerable to invasion is key to developing effective and cost-efficient management strategies. This study aims to model the distribution of invasive species across protected and communal landscapes under long-term climate change projections. A Random Forest (RF) model produced the highest accuracy metrics for Area under the curve (AUC) = 0.99 and True Skill Statistic (TSS)=0.97, while a MaxEnt model recorded the second highest AUC (0.98) and TSS (0.97). The results show a clear difference between the current and future scenarios of the spatial distribution in all the models. Applying a species distribution model (SDM) using both MaxEnt and RF produced a higher degree of prediction accuracy because RF is susceptible to overfitting training data while MaxEnt can produce predictable and complex results. Moreover, the overall predictions using the ensemble model demonstrated an increase in areas suitable for encroachment under RCP 8.5 but a decrease in the bush encroachment rate under RCP 2.6. These findings underscore the critical need for proactive management strategies to mitigate bush encroachment, particularly under high-emission scenarios, ensuring the sustainability of semi-arid savanna rangelands in the face of climate change.