Responses of Carbon, Nitrogen, and Phosphorus Export to Drought-Induced Forest Dieback Based on Multi-Substance Hydrological Modelling

Severe and prolonged summer-droughts and subsequent bark beetle outbreaks have increased forest vulnerability and mortality, altering the export of carbon (C), nitrogen (N), and phosphorus (P) from forested catchments. Currently, comparative field studies of how these three substances respond to forest dieback, as well as evaluation of process-based model simulations, remain limited. This study aims to analyze and compare the impacts of drought-induced forest dieback on catchment C, N, and P export and their underlying drivers, and the capability of a process-based hydrological water quality model to simulate those matter fluxes under rapid forest change. We applied a modified dynamic HYPE model to three headwater catchments in the Harz Mountains (Germany) with different land-use compositions. The conifer-dominated catchments Warme Bode and Rappbode experienced severe forest dieback of approximately 57% and 75%, respectively, between the 2018 drought until 2024, whereas the agricultural catchment Hassel showed a lower forest loss of about 15%. The model was calibrated by simultaneously optimizing hydrological and C–N–P process parameters using long-term discharge and water-quality observations. Model performance was overall acceptable, with good performance for hydrology and N simulations (mean NSE = 0.80 for discharge and 0.71 for nitrogen), moderate performance for C (mean NSE = 0.68), and the weakest performance for P (mean NSE = 0.53). Results showed clear increases in C, N, and P exports in the forest-dominated catchments after forest dieback, whereas changes in the agricultural catchment were minor. Among the three substances, N showed the strongest increase after forest dieback, driven by increased nitrogen availability associated with reduced plant uptake and enhanced soil mineralization. The increase in C export resulted from elevated organic carbon availability in surface soils, and was also controlled by changes in hydrological processes. P showed a relatively weaker response to forest dieback, with changes primarily driven by increased runoff magnitude and intensity, as well as enhanced flushing due to the loss of vegetation cover. Differences in simulation capabilities of these three substances further indicate distinctions in their generation and transport mechanisms. N dynamics are mainly governed by subsurface flow paths and biogeochemical availability, whereas C export depends more on surface runoff and flow-path connectivity, which are relatively well represented in the model. P export relies more on high-flow events, which are roughly generalized and simplified in the model parameterization. Therefore, the simultaneous optimization of C, N, and P points towards a more realistic representation of the various runoff components and biogeochemical processes in the model. Overall, this study advances the understanding of forest dieback impacts on catchment nutrient and carbon exports, reveals the limitations of multi-substance modeling, and provides suggestions for model development.