Climate-driven changes in fire disturbance could potentially alter tree species regeneration catalyzing abrupt changes in forest composition, structure, and function. Our ability to predict species’ responses under novel conditions is dependent on the development of ecosystem models that can represent complex interactions between disturbance, climate, species competition, forest demography, and increasing atmospheric carbon dioxide. Dynamic Global Vegetation Models (DGVMs) are process-based models that represent these interactions and are therefore widely used to predict climate change impacts on terrestrial vegetation, frequently at the continental to global scale. DGVMs require input parameters including physiology, biochemistry, structure, and allocation to characterize generalized plant functional types. The potential for modeling vegetation at the landscape scale with DGVMs is promising, but presents a challenge for parameterizing individual species with limited data. Yet when accomplished, the species-level approach is exciting because it enables investigation of interactions between species composition, forest demography, and fire dynamics. Here we present results from parameterization of the dominant tree species in the Greater Yellowstone Ecosystem (GYE) for the DGVM LPJ-GUESS to explore the causes of recent changes in plant productivity detected from satellite-derived vegetation indices. We hypothesized that increased post-fire regeneration of the dominant species, lodgepole pine, explains areas of increased productivity.
Results/Conclusions
For the dominant tree species in the GYE, bioclimatic limit parameters were extracted from Daymet gridded climate data using Forest Inventory Analysis (FIA) occurrence data. Other parameter values were gathered from literature and the TRY Plant Trait Database. Original parameter values resulted in unrealistic modeled species compositions, such as a relatively rare tree species dominating the landscape. Sensitivity analysis revealed the importance of parameters dictating growing degree-day requirements and photosynthesis temperature limits. Parameter optimization based on satellite-derived MODIS leaf area index (LAI) for these influential parameters resulted in improved agreement in modeled species compositions and FIA plot data. Using the newly parameterized regional plant species improved agreement with both MODIS LAI and FIA plot data compared to using the generic global plant types. Results from fractional factorial model simulations indicate a strong carbon dioxide fertilization effect on post-fire forest regeneration, while climate change is reducing post-fire productivity. These results suggest that the ability of plants to benefit from increasing atmospheric carbon dioxide depends on the bioclimatic context and the physiology of individual species and demonstrates the utility of process-based DGVMs for modeling species responses to future novel climate and disturbance conditions.