The widespread contamination of aquatic ecosystems by microplastics (MPs) and nanoparticles (NPs) poses a growing threat to organismal health; however, predictive frameworks that integrate ecotoxicological stress with biological outcomes remain limited. This study develops a predictive model linking MP and NP exposures to growth impairment and oral microbiome dysbiosis in the common carp ( Cyprinus carpio ), advancing current analyses beyond descriptive patterns. Individual and combined exposures significantly reduced weight gain ( p < 0.05). Notably, coexposure decreased the alpha diversity of oral bacteria (MP Chao1: 398.22 ± 211.05), whereas single-pollutant exposures increased it, indicating a complex, stressor-dependent ecological response. Bacterial diversity was positively correlated with host growth (Shannon index, R 2 = 0.501, p = 0.0018), which explains the 50% variance between growth and alpha diversity. An XGBoost machine learning model that reliably predicts individual fish growth from microbiome profiles alone was used to establish this relationship for ecological prediction. SHAP value interpretation identified key predictive bacterial taxa and diversity metrics, transforming the oral microbiome into a quantifiable biomarker of host health. Our findings indicate that MNP contamination results in predictable and mechanistically informative microbiome shifts.
Kakakhel et al. (Sun,) studied this question.