A multi-model assessment of the plant-physiological response to high and low carbon dioxide concentrations

Rising atmospheric carbon dioxide (CO₂) concentrations alter the vegetation composition indirectly through climate change and directly through plant physiological modifications. Both responses modulate climate through changed energy, moisture, and carbon fluxes between land and atmosphere. This makes accurate estimates of the responses important for future vegetation and climate projections. Yet, large inter-model differences regarding the magnitude of the direct response persist, leading to uncertain future projections. Here, we quantify the impact of CO₂ changes on the vegetation and climate in three Earth system models (ESMs) of varying complexity. The direct and indirect responses are separated using factorization experiments and statistical emulators. While most previous studies focus on either low or high CO₂ concentrations, we cover a large range from 150ppm to 1200ppm.

We find that plant function type (PFT) specific responses often follow a logarithmic shape except when threshold crossings create breakpoints. However, the grid box mean responses can differ from PFT-specific responses, indicating substantial modulations of the shape and amplitude by competition between PFTs. While competition amplifies the response for some variables, it dampens the response for others. For example, changes of biophysical properties like leaf area index and canopy height are amplified by competition, contributing to stronger plant-physiological impacts on some components of the terrestrial hydrological cycle than the radiative effect of rising CO₂ concentrations. The simulated long-term vegetation impacts can currently not be evaluated against present-day observations or manipulation experiments. Instead, we compare the model results with global compilations of paleobotanical data. Preliminary results for the Last Glacial Maximum indicate a model-dependent overestimation of the plant-physiological response. Future research aims at leveraging these comparisons to calibrate the modeled direct response of vegetation to CO₂, which would provide constraints for the long-term impacts of future emission scenarios on natural ecosystems.