Modeling the Impacts of Climate Change on the Distribution of the Invasive Cryptostegia grandiflora in Ethiopia, Using a Machine Learning Approach

Ethiopia, a biodiversity-rich country in the Horn of Africa, faces growing threats from invasive alien plant species, notably Cryptostegia grandiflora (rubber vine). This study assessed the current and projected future distribution of Cryptostegia grandiflora under climate change scenarios using ensemble species distribution models (MaxEnt, GLM, and Random Forest) and eight key bioclimatic variables. Model performance was high, with a mean AUC of 0.96 and a mean TSS of 0.88. The most influential predictors were mean diurnal temperature range, temperature seasonality, and precipitation seasonality. Under current climate conditions, 98% of Ethiopia (2018932.7 km²) is climatically unsuitable for the species, with suitable habitats concentrated in the central highlands and limited northern pockets. Future projections indicate substantial expansion of suitable habitat. By 2040, highly suitable areas are projected to increase by 162.0% under SSP2-4.5 and 131.2% under SSP5-8.5. By 2060, these areas are expected to expand further by 232.3% and 226.6%, respectively, relative to current climatic conditions. These projected shifts indicate an elevated invasion risk in central and southeastern Ethiopia, with significant ecological and socio-economic challenges, including suppression of native vegetation, reduced pasture productivity, and threats to pastoral livelihoods. Therefore, this study highlights the need for proactive monitoring, early containment, and climate-informed management strategies to mitigate future impacts of Cryptostegia grandiflora on biodiversity and ecosystem services.