Restoration pathways reshape forest structure and multifunctionality in subtropical plantations and secondary forests

This is the authors' abstract. We don't add key points for this paper.

Forest restoration may enhance ecosystem multifunctionality (EMF) without improving all ecosystem services equally. We compared 36 plots in 32–39-year-old stands across four subtropical regions of southern China, representing pure Cunninghamia lanceolata plantations, C. lanceolata –broadleaved mixed plantations, and naturally regenerated secondary forests following plantation clear-cutting. Eight service indicators represented soil conservation, water conservation, canopy-based climate regulation, carbon storage, stand productivity, tree species diversity, regeneration capacity, and soil fertility maintenance. EMF was quantified using averaging and threshold-based approaches, while root mean square deviation measured pairwise service imbalance. Regression models, relative-importance analysis, and structural equation modeling were used to examine associations between environmental and stand attributes and EMF. Mixed plantations had the highest mean EMF (0.559), compared with secondary forests (0.403) and pure plantations (0.383), and significantly higher multifunctionality at the 50% and 75% thresholds. However, they also had the largest number of highly imbalanced service pairs. Differences at the 25% threshold were not significant. Regeneration capacity was relatively low across all pathways, whereas static water retention was higher in secondary forests than in either plantation type. Prominent correlates differed among pathways: soil total phosphorus and species mingling in pure plantations, diameter-based structural diversity and stand age in mixed plantations, and stand density and soil total nitrogen in secondary forests. Higher aggregate EMF therefore did not imply a more balanced service bundle. This observational comparison supports evaluating restoration through both aggregate multifunctionality and individual service profiles, with inference restricted to the sampled middle-aged stands.