Drought-Induced Soil Carbon Dynamics in Subtropical Forests: Emergent Divergence from Model Structures

Abstract. Accurately quantifying drought impacts on terrestrial carbon cycling is essential for advancing predictions of climate-carbon feedbacks. However, current biogeochemical models exhibit limited capability in simulating drought-induced transformations of soil organic carbon (SOC), particularly regarding microbial processes. Here, we conducted a systematic comparative evaluation of three prevailing SOC modeling structures, including conventional three-pool partitioning scheme (SM1), mineral and particulate- associated carbon partitioning scheme (SM2) and Michaelis-Menten regulated carbon-stabilization scheme (SM3), to elucidate their capacity in simulating soil carbon dynamics under decadal drought scenarios in a subtropical forest. We found divergent effects of drought in soil C input (SM1, 66%; SM2, 10%; SM3, -4%) and mean residence time (MRT; SM1, -31%; SM2, -14%; SM3, 65%), which lead to the predicted SOC substantial accumulation for both SM1 and SM3 (+39.5% and +56.9%, respectively) and moderate depletion (-6.1%) for SM2. The different C input directly affect the passive SOC (SM1) and mineral-associated organic carbon (SM2 and SM3). In comparison, the drought effects on passive SOC (SM1), microbe biomass (SM2) and MAOC (SM2 and SM3), lead to notable spread in MRT. These findings highlight critical model structural dependencies in simulating drought-affected soil carbon dynamics and emphasize the necessity for models to integrate microbial-physicochemical interactions for improved climate-carbon coupling projections.