The strength of CO2 fertilization in forests inferred from the eddy-covariance record

Carbon sequestration, particularly in forests, is assumed to increase with increasing atmospheric CO₂ concentrations due to the CO₂ fertilization effect on photosynthesis (CFE). Estimating the contemporary effect of increasing atmospheric CO₂ on continuous measurements of gross primary production (GPP) in forest stands is still lacking as it is challenging to disentangle the CFE from other effects on GPP acting on long time scales such as climate variability, succession, land-cover change, and nutrient deposition. Here we introduce a statistical method, i.e., the “GPP residual” method, to estimate the effect of climate on GPP based on short-term variability, to remove it from the long-term signal, yielding the GPP residual that can be, at least partly, attributed to CFE.

We validate the applicability of this “GPP residual” method by testing whether it can accurately identify the CFE in simulations of the process-based land surface model QUINCY (QUantifying Interactions between terrestrial Nutrient CYcles and the climate system). We compare (i) the difference in the simulated GPP between historical simulations forced with transient and constant CO₂ concentrations with (ii) the non-climatic GPP variations determined when applying the “GPP residual” method to the transient-CO₂ simulation, and find encouraging agreement.

We next apply our approach to eddy-covariance derived GPP at 32 forested sites located in Europe and the US to quantify the CFE for each site and month-of-year in growing seasons. The median CFE across all site-months is 22 ± 6 % per 100 ppm change in CO₂. We note that other effects, such as nitrogen deposition and land management, also influence the GPP residual and could be incorrectly attributed to CFE. Assuming that these more site-specific effects may partially cancel out across sites as random effects, the estimated median value still reflects the strength of CFE. However, causal research will be needed to disentangle these long-term effects which cannot be separated by time scale.

In summary, our study derives for the first time an observation-based estimation of the CO₂ fertilization effect across forested ecosystems based on the eddy covariance record. Our results encourage future work to reconcile the uncertainties of the effect of increasing CO₂ on the global carbon cycle as determined from models, experiments and observations.