Reparameterization Required After Model Structure Changes From Carbon Only to Carbon‐Nitrogen Coupling.

Prediction of carbon (C) sequestration in terrestrial ecosystems under climate change depends on C‐nitrogen (N) interactions. While many newly developed C‐N coupling models predict lower C storage than their C‐only counterpart models, it has not been carefully examined whether or not such model predictions would be supported by real‐world observations. This study is to retrieve knowledge on N cycle from one field warming experiment in an alpine meadow of China so that we can better predict land C sequestration. We estimated two sets of parameters with one C‐only model and one C‐N coupling model, respectively, with the data assimilation technique. Our results showed that estimated process rates (e.g., senescence and decomposition rates) of organic C from almost all pools except standing litter and leaf are higher with the C‐N coupling model than those with the C‐only model. The allocation coefficient of C to root estimated with the C‐only model was 68.0% and 56.6% smaller than that with the C‐N coupling model under the ambient and warming treatments, respectively. Both the C‐only and C‐N coupling models simulate similar C pool sizes as observed at either the ambient or warming treatment with their respective parameter estimates. Meanwhile, the warming treatment increased the slow soil organic matter (SOM) pool due to decrease in estimated decomposition rate of the slow SOM via parameter fitting and increase in the C input from surface litter. In general, our results suggest that reparameterization is required when we add N processes to a C cycle model to realistically predict the ecosystem dynamics in response to future changes in N availability. Plain Language Summary: Models with nitrogen (N) module usually predict lower carbon (C) storage than their C‐only counterpart models, but the C storage in the real‐world ecosystems won't change according to whether or not a model considers N processes. We estimated two sets of parameters with one C‐only model and one C‐N coupling model, respectively, from a field warming experiment in an alpine meadow of China with the data assimilation technique. The technique integrates observations with model via parameter estimation. We found that a model structure change from the C‐only to C‐N coupling influences model parameterization but doesn't influence the simulations. However, warming increased soil C storage due to the larger C input and the smaller decomposition rate than the control. We suggest when we add N processes to a C cycle model, the new models should be re‐parameterized, ideally with data assimilation, to better predict the ecosystem C dynamics under chaing N availability. Key Points: Model structure changes from carbon only to carbon‐nitrogen coupling often result in lowered ecosystem carbon storage and lowered carbon sequestration, which may not reflect the reality in the real‐world ecosystemsOur study with data assimilation suggested that parameter values change with model structures whereas simulated ecosystem carbon dynamics are similar under either ambient and warning conditionsNitrogen (with other elements as well) regulates carbon cycle processes mainly through its influences on parameters in this study [ABSTRACT FROM AUTHOR]