The spatial distribution of C3 and C4 grasses in North America through the next century

Abstract:

C₄ grasses currently cover ~18% of the earth’s surface and are economically important as food sources, but their distributions are likely to change with future climate changes. As a result of the opposing impacts of atmospheric CO₂ and temperature on C₃ and C₄ physiology, future changes to the productivity and distributions of these grasses have remained unclear. We have used past and present tooth enamel, collagen, and bone carbon isotope ratios (δ¹³C) of Bison and Mammoth grazers to record the δ¹³C values of their diet, and the abundance of C₃ and C₄ vegetation in these habitats. Thus, the δ¹³C values of bison and mammoth tissues serve as a proxy for vegetation composition across North America through time. We combine these isotope data with ensemble CMIP5 climate model outputs, eight different climatic and fire predictor variables and advanced statistical techniques to model the spatial distribution of C₃ and C₄ grasses up through the year 2100 for two different emissions scenarios. Using the Random Forest algorithm, our model explains 91% of the spatial and temporal isotopic variability in bison and mammoth tissues and infers that mean summer temperature is the strongest predictor of all climate variables. For the emission scenario RCP4.5, in which atmospheric CO₂ levels are predicted to rise to ~540 ppm by 2100, we find decreases in the abundance of C₄ grasses of up to 30% in the south-central Great Plains and the Florida peninsula, and increases of up to 50% in the northern Great Plains. For the RCP8.5 scenario, in which atmospheric CO₂ levels are expected to rise to ~930 ppm by 2100, our model predicts minor decreases in the abundance of C₄ grasses in Texas and Oklahoma, but increases of 30-50% over the majority of the Great Plains. The overall effect of these changes is a homogenization of the Great Plains ecoregion in terms of grassland type distributions, and the loss of the highest abundance of C₄ ecosystems of the panhandles of Texas, Oklahoma and western Kansas. These results have important implications for future changes to insect and mammalian biodiversity and trophic interactions across the Great Plains of North America.