Divergent projections of future African biome shifts with process‐based and species distribution models

Future climate change is expected to influence ecosystem dynamics and the biogeographic distribution of biomes. Such shifts would have profound impacts on biodiversity and the provision of ecosystem services that are essential for humans. A robust understanding of potential future biome changes is therefore required to inform conservation and adaptation strategies. Here, we compared future biome changes in Africa modeled using a process‐based dynamic vegetation model (aDGVM) and species distribution models (SDMs) for different representative concentration pathway (RCP) scenarios, and assessed the impacts of plant‐physiological effects of elevated CO 2 on vegetation. We show that species distribution models can reproduce biome patterns simulated by the aDGVM for current climate conditions ( k values between 0.69 and 0.80 for different scenarios, indicating substantial agreement). However, future biome projections differed between modeling approaches. Projections from SDMs showed the lowest magnitude of biome shifts (14.1% of the area for RCP4.5) and were more similar to aDGVM simulations that excluded the effects of elevated CO 2 on vegetation. Projections from the aDGVM that included CO 2 effects showed the highest magnitude of biome changes (25% of the area for RCP4.5). The aDGVM showed mainly transitions towards more wood‐dominated biomes, whereas SDMs projected forest dieback, particularly in the RCP8.5 scenario. Annual precipitation, dry‐season precipitation and, in the aDGVM, CO 2 , were the main contributors explaining the most frequent biome transitions, followed by temperature. We conclude that all models project biome changes, primarily along biome boundaries, but the extent differs. CO 2 effects, as included in process‐based models, strongly influenced future vegetation. Different modeling approaches are necessary to quantify the range of possible future biome shifts and identify areas with high likelihood of undesired vegetation change.