Faunal-microbial synergism reconfigures wetland microcosm ecosystems: Machine learning elucidates bioturbation-driven ecological resilience

The pervasive infiltration of microplastics (MPs) is undermining the decontamination capacity of nature-based treatment systems, yet most existing studies are limited to short-term observations that obscure long-term ecological dynamics. To address this gap, this study integrates a 360-day, tubifex-augmented wetland microcosm with advanced machine learning analytics to decouple the mechanisms underlying ecosystem resilience against 50 μm polystyrene MPs (at 100 and 1000 μg/L).In contrast to the functional decline typically observed in conventional systems, the bioturbation-enhanced regime exhibited exceptional homeostatic regulation and maintained high nitrogen removal even under high-load MP stress, sustaining robust nitrogen removal efficiencies exceeding 82% for total nitrogen, markedly surpassing the approximately 60% observed in controls even under high-load MP stress. Mechanistic analysis revealed that habitat engineering by tubifex mitigated contaminant toxicity by reshaping microenvironmental niches, thereby promoting the selective recruitment and enrichment of stress-tolerant functional taxa and restoring overall metabolic capacity. In addition, a Random Forest model effectively bridged temporal gaps in ecological monitoring, enabling high-accuracy prediction of sustainability from limited datasets (test set R² = 0.92). Collectively, these findings establish a new paradigm of faunal-microbial synergy and provide a robust, data-driven framework for designing resilient wastewater treatment infrastructure in the plastisphere era.