Carbon source heterogeneity modulates microbial co-occurrence networks and stochastic assembly to drive divergent polycyclic aromatic hydrocarbons degradation in biostimulated soils

Remediating polycyclic aromatic hydrocarbon (PAH)-contaminated soils requires understanding how biostimulant types regulate functional microbial communities. This study systematically compared the mechanisms of plant-derived (rice straw [RS] and sodium lignosulfonate [LN]) versus animal manure-derived (pig manure [PM] and cow dung [CD]) biostimulants in reshaping microbial ecological processes during PAH degradation. Through 90-day microcosm experiments, we integrated high-throughput sequencing, qPCR, and co-occurrence network analysis to assess microbial community responses. Results showed plant-derived biostimulants achieved significantly higher PAH degradation efficiency (76.84 % and 73.43 % for high-molecular-weight PAHs [HMW PAHs] in RS and LN, respectively) compared to animal manure-derived ones (24.35 % in CD). The RS treatment specifically enriched nine keystone PAH-degrading genera (e.g., Sphingobium, Micromonospora, Olivibacter). Network analysis revealed RS enhanced positive correlations to 71.09 %, whereas CD reduced network complexity and stability. Null model analysis indicated stochastic processes dominated community assembly under plant-derived biostimulants, while deterministic selection prevailed with animal manure-derived inputs. These findings demonstrate that carbon source characteristics fundamentally shape microbial interaction patterns and assembly processes, providing a theoretical basis for precision biostimulation. Plant-derived biostimulants are preferred for long-term remediation of HMW PAH-contaminated sites, while animal manure-derived options may suit acute pollution scenarios requiring rapid degrader enrichment.