0
Meta Analysis ? AI-assigned paper type based on the abstract. Classification may not be perfect — flag errors using the feedback button. Tier 1 ? Systematic review or meta-analysis. Synthesizes findings across many studies. Strongest evidence. Environmental Sources Reproductive & Development Sign in to save

Global Meta-AnalysisIntegrated with Machine LearningAssesses Context-Dependent Microplastic Effects on Soil MicrobialBiomass Carbon and Nitrogen

Figshare 2025
Yangzhou Xiang (11266728), Matthias C. Rillig (130475), Josep Peñuelas (9016145), Luca Nizzetto (1624969), Jordi Sardans (2878262), Jian Long (385924), Jiachang Zhang (289752), Rui Li (4631), Ying Liu (18461), Yang Luo (350920), Bin Yao (4375), Yuan Li (67017)

Summary

This global meta-analysis of 90 studies found that microplastics in soil can increase microbial activity and affect carbon and nitrogen cycles, particularly biodegradable plastics which had the strongest effects. While focused on soil health rather than direct human impact, these changes could affect the quality of crops grown in contaminated soil and the broader food system.

Polymers
Body Systems
Study Type Review

Microplastics (MPs) in soil can paradoxically stimulate microbial biomass in a highly context-dependent manner, potentially inducing decomposition and affecting carbon and nitrogen cycles. We conducted a global meta-analysis with 90 studies (710 observations of microbial biomass carbon (MBC), 354 of microbial biomass nitrogen (MBN)) integrated with machine learning to quantify MPs effects on soil microbial biomass. Field studies showed no significant effects, contrasting with controlled experiments where MPs increased MBC by 9.6% (95% CI: 7.2–11.9%) and MBN by 10.4% (6.8–14.0%). Biodegradable plastics (PBAT, PLA) induced stronger effects (36.1–67.6%) than conventional polymers (PE, PP, PS, PVC). Temperature emerged as the dominant factor, with a contrasting MPs effect on MBC (positive) and MBN (negative) at higher temperatures, suggesting potential decoupling of carbon and nitrogen cycles under warming conditions. Machine learning models (XGBoost, R2 = 0.62) significantly outperformed linear regressions (R2 = 0.02–0.05), revealing nonlinear responses and threshold effects. Stimulatory effects were most significant for medium-sized MPs (30–90 μm), at high concentrations (>10 g kg–1), and in soils with intermediate fertility, highlighting context-dependent risks to soil carbon and nitrogen cycling.

Share this paper