Modeling and predicting land use / land cover dynamics in Borana National Park, southern Ethiopia

Understanding LULC dynamics is essential for biodiversity conservation. This study quantified historical LULC changes from 1987 to 2023 and projected future trends (2023-2059) in Borana National Park (BNP) in southern Ethiopia. This area is increasingly threatened by long-term LULC dynamics that have not been comprehensively assessed. Multi-temporal Landsat imagery was classified using a Random Forest (RF) algorithm on the Google Earth Engine (GEE) platform. We compared the RF model using Spectral Bands only (SB) with a model integrating Spectral Bands, Spectral Indices, and Topographic variables (SBSITV). Future LULC dynamics were simulated using a CA–Markov model approach in TerrSet. Results reveal a dramatic regime shift from open rangeland to shrub-dominated and agricultural landscapes. Between 1987 and 2023, the park experienced a profound reduction in vital grazing and forest resources, with grassland, shrubby grassland, woodland, and forest by −24.5%, −51.9%, −63.9%, and −75.5% respectively. Conversely, shrubland expanded more than five-fold, increasing its landscape footprint from 7.5% to 50.7%, while agricultural land increased over 6-fold. Future projections indicate these trends will persist through 2059, with shrubland projected to expand by an additional +41.9% and agriculturalland by 55.8%, primarily at the expense of woodland (−69.4%), shrubby grassland (−54.5%), and grassland (29.9%). These changes charactized an extensive process of bush encroachment and rangeland fragmentation, signaling an ecological tipping point that threatens pastoral livelihoods and biodiversity. These findings underscore the urgent need for management interventions to control shrub encroachment and restore the ecological integrity of this fragile rangeland ecosystem.