Coupling of Phosphorus Processes With Carbon and Nitrogen Cycles in the Dynamic Land Ecosystem Model: Model Structure, Parameterization, and Evaluation in Tropical Forests

摘要

The biogeochemical processes of carbon (C), nitrogen (N), and phosphorous (P) are fully coupled in the Earth system, which shape the structure, functioning, and dynamics of terrestrial ecosystems. However, the representation of P cycle in terrestrial biosphere models (TBMs) is still in an early stage. Here we incorporated P processes and C‐N‐P interactions into the C‐N coupled Dynamic Land Ecosystem Model (DLEM‐CNP), which had a major feature of the ability in simulating the N and P colimitation on vegetation C assimilation. DLEM‐CNP was intensively calibrated and validated against daily or annual observations from four eddy covariance flux sites, two Hawaiian sites along a chronosequence of soils, and other 13 tropical forest sites. The results indicate that DLEM‐CNP significantly improved simulations of forest gross and net primary production (R2: 0.36–0.97, RMSE:1.1–1.49?g C m?2?year?1 for daily GPP at eddy covariance flux sites; R2?=?0.92, RMSE?=?176.7?g C m?2?year?1 for annual NPP across 13 tropical forest sites). The simulations were also consistent with field observations in terms of biomass, leaf N:P ratio and plant response to fertilizer addition. A sensitivity analysis suggests that simulated results are reasonably robust against uncertainties in model parameter estimates and the model was very sensitive to parameters of P uptake. These results suggest that incorporating P processes and N‐P interaction into terrestrial biosphere models is of critical importance for accurately estimating C dynamics in tropical forests, particularly those P‐limited ones. Plain Language Summary Terrestrial carbon dynamics are usually regulated by nitrogen and phosphorus availability. However, most current terrestrial biosphere models or land surface models have ignored C, N, and P interactions, resulting in a large uncertainty in estimating the response of terrestrial ecosystems to environmental changes. Through coupling the P processes into an existing C‐N model, we have improved the model performance in simulating carbon dynamics in tropical forests. Model evaluation against field observations indicates that inclusion of P processes and N‐P colimitation on carbon assimilation is critical for tropical forests of different nutrient limits.