Can models adequately reflect how long-term nitrogen enrichment alters the forest soil carbon cycle?

Abstract. Changes in the nitrogen (N) status of forest ecosystems can directly and indirectly influence their carbon (C) sequestration potential by altering soil organic matter (SOM) decomposition, soil enzyme activity, and plant-soil interactions. However, model representation of linked C-N cycles and SOM decay are not well-validated against experimental data. Here, we use extensive data from the Fernow Experimental Forest long-term, whole-watershed N fertilization study to compare the response to N perturbations of two soil models that represent decomposition dynamics differently (first-order decay versus microbially-explicit reverse Michaelis-Menten kinetics). These two soil models were coupled to a common vegetation model which provided identical input data. Key responses to N additions measured at the study site included a shift in allocation to favor woody biomass over belowground carbon inputs, reductions in soil respiration, accumulation of particulate organic matter (POM), and an increase in soil C:N ratios. The vegetation model did not capture the often-observed shift in allocation with N additions, which resulted in poor predictions of the soil responses. We modified the plant C allocation scheme to favor wood production over fine root production with N additions, which significantly improved the vegetation and soil respiration responses. To elicit an increase in the soil C stocks and C:N ratios with N additions, as observed, we also modified the decay rates of the particulate organic matter (POM) in the soil models. With all of these modifications, only the microbially explicit model captured a positive soil C stock and C:N response in line with observations. Our results highlight the importance of accurately representing plant-soil interactions, such as rhizosphere priming, and their responses to environmental change.