Carbon–nitrogen coupling under three schemes of model representation: a traceability analysis

The interaction between terrestrial carbon (C) and nitrogen (N) cycles has been incorporated into more and more land surface models. However, the scheme of C-N coupling differs greatly among models, and how these diverse representations of C-N interactions will affect C-cycle modeling remains unclear. In this study, we explored how the simulated ecosystem C storage capacity in the terrestrial ecosystem (TECO) model varies with three different commonly-used schemes of C-N coupling. The three schemes (SM1, SM2, and SM3) have been used in three different coupled C-N models (i.e., TECO-CN 2.0, CLM 4.5, and O-CN, respectively). They differ mainly in the stoichiometry of C and N in vegetation and soils, plant N uptake strategies, pathways of N import, and the competition between plants and microbes for soil mineral N. We incorporated them into the C-only version of TECO model, and evaluated their impacts on the C cycle with a traceability framework. Our results showed that all of the three C-N schemes resulted in significant reductions in steady-state C storage capacity compared with the C-only version, but the magnitude varied with −23 %, −30 % and −54 % for SM1, SM2, SM3, respectively. The reduced C storage capacity is the combination of decreases in net primary productivity (NPP) by −29 %, −15 % and −45 % with changes of mean C residence time (MRT) by 9 %, −17 % and −17 % for SM1, SM2, and SM3, respectively. The divergent NPP are mainly attributed to the different assumptions on plant N uptake, plant tissue C:N ratio, down-regulation photosynthesis, and biological N fixation. In comparison, the alternative representations of the plant and microbe competition strategy and the plant N uptake, combining with the flexible C:N ratio in vegetation and soils, led to a notable spread MRT. These results highlight that the diverse assumptions on N process representation among different C-N coupled models could cause additional uncertainty to land surface models. Understanding their difference can help us to improve the capability of models to predict future biogeochemical cycles on land.