Dynamic patterns and drivers of carbon accrual under different forest restoration approaches

Forest restoration represents a critical nature-based solution for climate change mitigation through enhanced carbon accumulation. While recognized for its ecological potential, the temporal trajectories of carbon accumulation across restoration approaches and their underlying mechanisms remain poorly quantified. This study analyzes multi-decadal (26 years) of carbon storage dynamics within a chronosequence framework to elucidate the mechanisms linking carbon accumulation patterns with drivers and management legacies. Restoration strategies diverged markedly: old-growth forests (OF; ≥40 years old in 1990; n = 25) sustained persistent carbon accumulation, whereas secondary forests (SF; <40 years old in 1990; n = 30) protection exhibited marked temporal variability in carbon gain and loss. Reforestation (RF, n = 30) yields 2.2-13.5 times higher carbon gains (1.3-2.7 vs 0.2-0.6 Mg C ha^-1 yr^-1) than natural recovery (NR, n = 50) in regions in which forests have been removed. Structural equation modeling (SEM) revealed initial C stocks emerged as the most important regulators of changes in carbon stocks (ΔC stocks) when considering direct and indirect effects (p < 0.001). Tree diversity (species richness) and stand structure attributes (stand density in terms of tree per ha, age and tree size) (p < 0.05) exhibited temporal divergence in both effect size and relevance on carbon accumulation. Notably, tree size effects displayed context-dependent reversals: correlations with ΔC stocks shifted from positive to negative when baseline stocks exceeded 23.05 Mg C ha^-1. These results underscore the optimization of carbon accumulation requires targeted strategies aligned with conservation priorities and restoration objectives.