Quantifying the impact of multiple stressors on microbial communities in Dianshan Lake sediments using Random Forest...

Alluvial plain lakes, characterized by low hydrodynamic activity and fine sediments, host microorganisms whose diversity and community structure are strongly shaped by the presence of nutrients and pollutants. However, quantifying the complex impacts of these multiple stresses is challenging. This study focused on Dianshan Lake, a human-impacted plain lake in Shanghai, China, to quantify the contributions of sediment properties, pollution levels, and nutrients on microorganisms, using classical statistical methods and Random Forest (RF) analysis. The RF model achieve a good fit (R² = 0.80-0.87). Feature contribution analysis indicated that contaminants were the primary factors shaping microbial diversity in sediments, accounting for approximately 40% of the total contribution. However, sediment redox conditions emerged as the most influential single factor. In lightly to moderately polluted freshwater lakes, benthic microorganisms displayed common dominance, but the contributions of influencing factors varied. Proteobacteria and Chloroflexi, the dominant phyla, were significantly impacted by pollutants, with contributions exceeding 50%. PAHs (Polycyclic Aromatic Hydrocarbons) primarily suppressed genera within Proteobacteria, while Anaerolineaceae in Chloroflexi exhibited strong tolerance to Cd. The dominant species in Dianshan Lake sediments were also strongly influenced by NO₃⁻-N and NH₄⁺-N, far exceeding the impact of various forms of phosphorus. This also highlights the issue of nitrate-driven eutrophication in the region. This study demonstrates that RF analysis effectively identifies key controlling factors in lightly to moderately polluted sedimentary environments, providing valuable insights into the ecological processes and a scientific foundation for ecological risk management in similar aquatic environments.