Modelling soil responses to nitrogen and phosphorus fertilization along a soil phosphorus stock gradient

In this study, we investigate the responses of soil organic carbon (C) to nitrogen (N) and phosphorus (P) additions along a soil P stock gradient of five beech forest stands in Germany, using a modelling approach. Two different soil models with coupled C, N, and P cycles are used to simulate fertilization experiments conducted at the study sites. The first model, the stand-alone soil module of QUINCY (QUINCY-soil, Thum et al. 2019), is a conventional soil model that uses first-order kinetics to describe soil organic matter (SOM) turnover and represents microbial biomass only implicitly. The second model, the Jena Soil Model (JSM) (Yu et al. 2020), is a novel microbial soil model, which explicitly simulates microbial dynamics and describes the turnover of SOM as the consequence of several interactive processes, such as microbially mediated depolymerisation of litter and SOM, organo-mineral association, and vertical transport. We applied both site-level models to five study sites and compared the modeled soil profile with observations. In addition, model scenarios were conducted to simulate the fertilization of N and P, and we further evaluate the effect of soil P stock, plant litter quality, and SOM CNP stoichiometry, on the responses of soil (heterotrophic) respiration (Rs) to nutrient addition. We found that the fitness between simulated and observed SOM profiles (defined as normalized root mean square ratios, Knrmsr) were generally better in JSM than in QUINCY-soil (Knrmsr larger by 0.03±0.10 to 0.16±0.06 for various soil measurements at all sites); The general pattern of observed Rs responses to nutrient fertilization, that N addition decreases Rs whereas P addition increases it, can be reproduced by JSM but not by QUINCY-soil. Our results indicated that detailed explicit description of microbial processes and organo-mineral association is required to model plant-soil-microbial interactions, thus to better reproduce SOM profiles and their responses to nutrient additions. It highlights the need to better represent these processes in future model developments.