The exchange of inorganic phosphorus between soil solution and matrix might largely affect the model predictions of terrestrial carbon cycle

Phosphorus (P) availability may influence the response of terrestrial ecosystems to environmental and climate change. Soil biogeochemical (organic) and geophysical (inorganic) P cycling processes are the key players in this regulation. There has been a continuous effort to include P cycling processes into terrestrial biosphere models (TBMs) and many modelling studies agreed on the significance of organic P cycling processes to terrestrial ecosystems. However, the role of inorganic P cycling processes remains unclear. Although the model representations of inorganic P cycling in most TBMs are similar, their parameterisations differ greatly, and none of TBMs have been validated against soil P measurements.

In this study, we developed a new algorithm based on the two-surface Langmuir isotherm to describe the inorganic P exchange between soil solution and soil matrix in the QUINCY TBM, and tested both the novel and conventional models at five beech forest sites in Germany along a soil P stock gradient, which are the main study sites of the German Research Foundation (DFG) funded priority programme 1685.

We conducted a literature review on Langmuir P sorption parameters, which indicates that the P sorption capacity (S_max) is strongly correlated with soil texture and the Langmuir coefficient (k_m) is strongly correlated with soil pH and organic matter (OM) content. We divided soil P sorption sites into the OM-rich clay and silty sites and OM-poor sandy sites and extracted empirical equations to calculate their S_max and k_m.

The two-surface Langmuir isotherm approach was implemented to QUINCY, and both the novel and conventional (one-surface Langmuir isotherm) models were applied to the study sites. The models were evaluated with observed soil inorganic P fractionations, foliar N and P contents, and normalized vegetation carbon (C) without calibration. The novel model significantly improved the goodness of model fit to P fractionation measurements at all sites. Both models were able to adequately capture the observed foliar N and P contents, but only the novel one reproduced the observed pattern of vegetation C along the soil P gradient.

We further tested the effect of both models on the responses to CO₂ addition, P addition and C&P addition at all study sites. The conventional model showed stronger ecosystem responses to P and C&P additions than the two-surface Langmuir one, especially at P-poor sites. It is probably due to that plants store more added P in the conventional model than the novel one. We also tested the sensitivity of both models to the P cycling parameterisation at one low-P site. Despite better model fit to the observed soil P fractionation, the novel model also produced higher and more robust gross primary production, foliar P content and vegetation C than the conventional one.

In summary, we showed that the two-surface Langmuir isotherm approach adequately reproduced the observed soil P fractionations and the pattern of vegetation C along a soil P gradient, owing to its better representation of inorganic P cycling and thus C-P interactions, particularly at low-P ecosystems.