Trait-based projections of climate change effects on global biome distributions

Aim: Climate change will likely modify the global distribution of biomes, but the magnitude of change is debated. Here, we followed a trait-based, statistical approach to model the influence of climate change on the global distribution of biomes. Location: Global. Methods: We predicted the global distribution of plant community mean specific leaf area (SLA), height and wood density as a function of climate and soil characteristics using an ensemble of statistical models. Then, we predicted the probability of occurrence of biomes as a function of the three traits with a classification model. Finally, we projected changes in plant community mean traits and corresponding changes in biome distributions to 2070 for low (RCP 2.6; +1.2°C) and extreme (RCP 8.5; +3.5°C) future climate change scenarios. Results: We estimated that under the low climate change scenario (sub)tropical biomes will expand (forest by 18%–22%, grassland by 9%–14% and xeric shrubland by 5%–8%), whereas tundra and temperate broadleaved and mixed forests contract by 30%–34% and 16%–21%, respectively. Our results also indicate that over 70%–75% of the current distribution of temperate broadleaved and mixed forests and temperate grasslands is projected to shift northwards. These changes become amplified under the extreme climate change scenario in which tundra is projected to lose more than half of its current extent. Main conclusions: Our results indicate considerable imminent alterations in the global distribution of biomes, with possibly major consequences for life on Earth. The level of accuracy of our model given the limited input data and the insights on how trait–environment relationships can influence biome distributions suggest that trait-based correlative approaches are a promising tool to forecast vegetation change and to provide an independent, complementary line of evidence next to process-based vegetation models.