Brazil’s Cerrado region in Central Brazil is comprised of forest, savanna, and grassland ecosystems in which forest dominance increases near the Amazon region’s monsoon core, as does biodiversity and carbon stocks. Forest-savanna boundaries are thought to be maintained by fire locally and climatic gradients regionally, changing dynamically through time. Understanding conditions where forests expand into savannas is important for assessing regional biodiversity conservation and carbon sequestration potential. We attempt to quantify forest-savanna cover changes over the past several thousand years and contrast changes with existing precipitation records and recent anthropogenic land conversions at the Cerrado-Amazon transition.
Forest and savanna distribution changes can be reconstructed using δ13C values from soil organic matter (SOM) preserved in soil profiles. Tree- and grass-dominated systems within this region have distinct carbon isotope signatures due to differences in dominance of species with C3 (e.g., trees) and C4 (e.g., grasses) photosynthetic pathways. Soil depth is used as an age proxy using radiocarbon activity of SOM to generate age-depth models, which generally supports the assumption that average SOM age increases linearly with increasing depth. Through combining δ13CSOM data and soil age we can uncover whether and how forest-savannas changed their distribution. We can map these changes across broad spatiotemporal scales by combining reconstructions of vegetation shifts locally using individual soil profiles and regionally using proximity polygon interpolation of δ13C data.
Results/Conclusions
Soil δ13CSOM profiles were measured at savanna (n=14), forest (n=26), and forest-savanna transition zones (n=6). We found forests to be expanding in 83% of the sampled locations for the past two thousand years. At several sites, forest expansion has been occurring for approximately four thousand years, but only 33% of the sites span that time period. Additional data are being collected for a more comprehensive analysis. δ13CSOM interpolation maps indicate forest expansion strength depends upon core monsoon region proximity. Vegetation cover shifts permit inferences regarding past climatic conditions; increasing tree dominance and forest expansion imply wetter conditions whereas grass dominance denotes drier climates. To test this hypothesis, comparisons between our vegetation records and independent speleothem-based precipitation records are underway.
Besides understanding how forest-savanna dynamics affect local to regional carbon stocks and biodiversity, combing δ13CSOM data and corresponding interpolation maps may potentially influence land management and conservation policy. For example, incongruences between δ13CSOM topsoil data and land cover designations can be used to identify and prevent undesirable land conversions that cannot be determined on vegetation cover alone.