Unravelling drivers of burned area change in southern Ethiopia: climate, vegetation, and dryland development
Аннотация
African savannas are the most extensively burned regions globally; however, burned area (BA) in these regions has declined in recent decades. Agricultural expansion, infrastructure development, and climate change are commonly cited drivers, but their relative contributions remain contested. Regional-scale analyses are needed to capture spatiotemporal changes, disentangle environmental and anthropogenic drivers, and inform decision-making. This study examined fire regime change in Ethiopia’s Lower Omo, a region of semi-arid pastoral lowlands and agrarian highlands undergoing rapid development from dams and agricultural megaprojects. Using 18 years of Landsat-derived BA data and Mann–Kendall trend analysis, we quantified changes and applied Generalized Additive Models with open-access remote sensing and GIS datasets to model nonlinear drivers of BA. Contrary to continental and national-level trends of declining BA, large areas of the region exhibited significant increases. Trends reflected the combined effects of environmental and anthropogenic drivers. Abiotic and biotic drivers were highly influential, with increasing effective rainfall, declining grassland, and increasing vegetation cover, consistent with bush encroachment, associated with BA declines. Anthropogenic drivers had varying effects; urban population centres were linked to higher BA, likely due to ignition pressure, while road expansion and large-scale commercial cropland conversion were associated with reduced BA, likely reflecting landscape fragmentation and a shift from extensive grassland burning to more spatially constrained agricultural fire regimes. Conversely, smallholder cropland expansion showed no significant effect. This research highlights the spatial heterogeneity of fire regime change, suggesting that coarse-resolution BA products and generalized population or human development proxies can obscure distinct anthropogenic mechanisms and interactions. It further demonstrates the value of fine-grain landscape-level assessments to uncover diversity and enable more precise discrimination of driver effects.