Coupled soil biogeochemical and plant responses to experimental warming control emergent soil carbon stocks

Soil organic carbon (SOC) responses to atmospheric warming depend on soil biogeochemical and plant responses and their interactions. These processes occur over a wide range of time scales and are spatially heterogeneous, leading to difficulties in predicting emergent ecosystem carbon dynamics. Here we explore coupled soil and plant responses to warming using a mechanistic model (ecosys) and observations from a whole-soil warming experiment in the Sierra Nevada mountains of California (Blodgett Forest Whole Soil Warming Experiment). ecosys represents the coupled hydrological, thermal, soil biogeochemical, and plant processes that affect ecosystem carbon cycling, and has been applied and tested in dozens of ecosystems. We briefly describe the model components relevant to the warming study and then show that it accurately simulates observed soil moisture and temperature, SOC stocks, root biomass, and experimental warming effects on soil T and moisture. Using the simulations, we then show that the emergent short-term effect of warming on respiration losses is dominated by heterotrophic respiration vs changes in plant inputs. However, multi-season to multi-year responses strongly depend on changes in N availability and plant N uptake, leading to increased soil surface CO₂ emissions from heterotrophic and autotrophic respiration, increases in aboveground biomass, and relatively small changes in SOC stocks from increased litter inputs. We also explore the roles of microbial thermal adaptation and soil moisture on the SOC stock changes. Finally, we compare the short- and long-term responses to warming to evaluate how well experimental soil warming manipulations directly inform ecosystem carbon dynamics under expected long-term climate change warming. Our work highlights the need to evaluate climate change impacts on the carbon cycle for the integrated plant-soil system, and points to needed improvements in current large-scale land models.