Mixed effects and threshold regression modeling of forest carbon dynamics under disturbance in the FoRTE experiment

Forest disturbances increasingly influence terrestrial carbon dynamics, yet ecosystem responses depend on disturbance severity and post-disturbance environmental conditions. Using observations from the Forest Resilience Threshold Experiment (FoRTE), we develop a hierarchical mixed-effects modeling framework with threshold regression to investigate how disturbance severity shapes soil CO $$_2$$ flux and recovery trajectories. Results show that soil temperature is the dominant control on short-term flux variability, while soil moisture exerts a smaller positive influence after accounting for covariate interactions. Model selection identifies evidence for a potential high-severity disturbance threshold associated with altered flux responses; however, inference is conditional on the limited number of experimental plots and should be interpreted as exploratory. Recovery following disturbance is gradual, with higher soil moisture associated with faster modeled rebound. Leave-one-plot-out cross-validation indicates moderate predictive skill across spatial units, while diagnostic analyses highlight remaining structural uncertainty. These findings demonstrate the value of hierarchical threshold approaches for characterizing disturbance–recovery dynamics and provide a quantitative basis for future large-scale assessments of forest carbon resilience under increasing disturbance regimes.