**Background:** Agricultural overexpansion has led to widespread land degradation and cropland abandonment globally, including 83 million hectares abandoned from 1992–2015. The China Danxia is an ecologically fragile region in South China characterized by acidic, nutrient-poor soils with thin layers and exposed parent material. The Grain for Green Program has been implemented to restore these degraded lands through vegetation restoration. Soil extracellular enzyme activities (EEA) and enzyme stoichiometry (EES) are sensitive indicators of microbial nutrient limitation and soil nutrient availability. This study aimed to compare microbial metabolic characteristics and soil multifunctionality under different vegetation restoration types in the China Danxia.
**Methods:** The study was conducted in the Danxia Mountain World Geopark, Guangdong Province (mean annual temperature 19.9°C, precipitation 1,512.3 mm). Three ecosystem types were compared: abandoned cropland (AC) without restoration, naturally restored secondary forests (NF, >30 years), and artificially planted forests (AF, >30 years). In 2021, three 10m×10m plots were randomly selected per ecosystem type. Soil samples were collected from 0–10 cm and 10–20 cm depths. Six extracellular enzymes were measured: β-1,4-glucosidase (BG), cellobiohydrolase (CBH), β-1,4-xylosidase (BX), β-1,4-acetylaminoglucosidase (NAG), leucine aminopeptidase (LAP), and alkaline phosphatase (AP). Microbial nutrient limitation was quantified using vector analysis of enzyme stoichiometry. Soil ecosystem multifunctionality (EMF) was calculated from 12 indicators (C, N, and P pools and enzyme activities) using Z-standardization. Statistical analyses included ANOVA, partial least squares path modeling (PLS-PM), Mantel tests, and variance partitioning analysis (VPA).
**Key Results:** Vegetation restoration significantly altered soil properties. NF soils showed the highest pH (6.21), SOC (17.14 g kg⁻¹), STN (1.41 g kg⁻¹), and MBN (28.22 mg kg⁻¹). At 0–10 cm depth, C_EEA and P_EEA increased by 15.16% and 79.10% respectively in NF compared to AC, while N_EEA decreased. Vector analysis revealed that AC soils exhibited microbial N-limitation, while both NF and AF soils shifted to P-limitation (vector angles >45°). Vector length decreased from 1.06 to 0.81 at 10–20 cm depth after restoration, indicating reduced C-limitation. Soil C:N_mic showed weak homeostasis (1/H=1.18), while C:P_mic was strongly homeostatic. Vegetation restoration significantly increased soil EMF, with NF showing the highest values. PLS-PM revealed that microbial biomass had the strongest direct effect on EMF (path coefficient 0.97). EMF was significantly correlated with vector angle under N-limitation (P<0.001) but not under P-limitation (P=0.762). VPA showed that pH, soil nutrients, and microbial biomass jointly explained 86.66% of EEA and EES variation, with their interaction accounting for 44.93%.
**Clinical Implications:** This study provides evidence that vegetation restoration, particularly natural secondary forest succession, alleviates N-limitation and improves soil ecosystem multifunctionality in degraded China Danxia soils. The shift from microbial N-limitation to P-limitation after restoration highlights the need for balanced nutrient management strategies. The findings suggest that strengthening soil N management is more beneficial for ecosystem restoration in this region, as N-limitation was the primary constraint on soil multifunctionality. The strong influence of biotic (microbial biomass) and abiotic (pH, nutrients) interactions on enzyme stoichiometry indicates that both factors must be considered in restoration planning. These insights can inform nutrient management strategies for ecological restoration not only in the China Danxia but also in other ecologically degraded areas undergoing restoration.