Coupling stress and fire to predict forest change
We designed the dynamic vegetation model MC to simulate complex interactions between climate and vegetation distribution, biogeochemical cycling, and fire. We have used the model to project potential vegetation shifts, change in C sequestration potential and fire risk using a variety of climate futures at multiple scales from National Parks to continents. Results have shown much geographic patchiness due to soil types, our first source of uncertainty, as well as temporal variability due to changes in rainfall seasonality.
While all climate models project warmer conditions in the 21st century, they differ in their projections of rainfall, which causes a wide variety of responses in vegetation and ultimately organisms that depend on it. Our model is sensitive to the water available for plant production and soil organic matter decomposition, fuel-build up and wildfire occurrence, to the extent that precipitation often drives the simulated ecosystem resilience to disturbance such as drought and fires. We can thus contrast climate futures in terms of high vs low risk of losing valuable ecosystem services following disturbance.
As we simulate vegetation shifts towards warmer types (e.g. warm subtropical grasslands replacing cool temperate grasslands), expansion of forest types is enhanced by the CO2 effect on tree water use efficiency and production, mitigating the decrease in soil water availability. Different magnitudes of the CO2 effect can thus generate different futures based on its magnitude and CO2 emission levels as well as the depth of future drought periods.
We have tried to quantify the role of human impacts in comparison to that of climate on ecosystem resilience to change. We looked in detail at the impacts of fire suppression on C stocks and fire risk. We also recently used a simple protocol to evaluate the relative influence of forest rotation, crop harvest, and urban expansion on ecosystem dynamics using downscaled land-use projections. Not only do we see changes in the size of C pools and in overall sequestration potential but we also see different results with the same climate scenario when potential forest land is used for crop, urban areas or short rotation tree plantations that do not sequester as much carbon nor build the fuel load that natural vegetation would. I will present results from a variety of projects and focus on shared conclusions as well as discrepancies based on challenges to climate change research linked to data limitation and local processes ignored in most simulations.