Foliar temperature acclimation improves model performance while suppressing global land carbon uptake

Background/Question/Methods

Photosynthesis and respiration on land represent the largest carbon flux between the Earth’s surface and the atmosphere, and parameters associated with these processes represent the large sources of uncertainty in Land Models (LMs). The incorporation of temperature acclimation of foliar photosynthesis and respiration has been proposed as a way to reduce this uncertainty and possibly reduce future negative carbon-climate interactions. Here we tested these predictions by first examining the influence of contemporary acclimation formulations on a LM’s ability to reproduce flux tower data, then by using both multiple techniques to explore and understand how these formulations affect simulations of carbon uptake. To test model performance, we simulated net ecosystem exchange at 15 FLUXNET sites spanning 3 biomes and compared simulated rates to observed rates. To determine the spatial and temporal patterns of the influence of acclimation on LMs, we calculate the ratio of acclimated to unacclimated rates of maximum Rubisco carboxylation (V_cmax; ratio: E_v), maximum rate of Ribulos-1,5-bisphosphate regeneration (J_max; ratio: E_j), and dark respiration (R_d; ratio: E_r) globally from 1861-2100. Finally, to further explore these results in a LM, we simulated land carbon uptake from 1861-2100 at the same 15 sites under the RCP 8.5 scenario.

Results/Conclusions

We found that incorporating acclimation processes improved model performance, particularly in tropical areas. However, incorporating a full complement of acclimation formulations limited past and contemporary global carbon uptake compared to formulations without acclimation. This limitation was found to be primarily due to decreased V_cmax and increased R_d in the coolest regions and time periods. This limitation slowly decreases as the Earth warms, but continues for several decades and will likely persist beyond the end of this century. Our results indicate that the inclusion of acclimation results in more realistic simulations by shifting the temperature response of V_cmax, J_max, and R_d towards more ideal parameterizations. Interestingly, these shifts result in a decrease in carbon uptake globally compared to mean parameterizations based primarily on warm temperate species, which are commonly used in LMs and were used here to represent non-acclimation functions. This indicates that global-scale LMs are likely overestimating land carbon uptake potential.