Characterizing and Mapping the Grassland Vegetation of the Colombian Orinoquia

This study presents a high-resolution mapping of grassland phytosociological alliances in the Colombian Orinoquia by integrating multi-source remote sensing data (Landsat-8 optical and Sentinel-1 SAR) with environmental variables within a Random Forest classification framework. Based on 292 rigorously classified vegetation plots, we modeled the distribution of 18 alliances across dominant geomorphological units: the alluvial plain (north) and the high plain (south-central). Results demonstrate that natural vegetation covers 73.74% of the region, with grasslands (41.55%) representing a more extensive formation than forests (32.19%). The alliances Paspalo pectinati–Axonopodion aurei (6.02%) and Axonopodo aurei–Trachypogonion spicati (4.37%) were identified as the most widespread. Ecological analysis revealed a pronounced dominance of C4 Poaceae species, particularly in alliances such as Sipaneo pratensis–Axonopodion purpusi (60% C4 diagnostic species), reflecting adaptations to seasonal moisture stress and high irradiance. Our methodology demonstrates that coupling phytosociological field data with multi-sensor remote sensing achieves high classification accuracy (79–87%), providing a robust tool for moving beyond descriptive vegetation mapping toward a comprehensive understanding of grassland distribution patterns at regional scales.