Integrating laboratory and field data to evaluate the effects of experimental microplastic exposure on Acanthogobius flavimaus.

Microplastic pollution is an escalating environmental concern with broad ecological and health implications. This study examines whether current levels of microplastic contamination in aquatic environments affect the metabolism of estuarine gobies (Acanthogobius flavimanus) by integrating laboratory and field metabolomics data through a data-driven, embedding-based framework. In the laboratory, gobies were exposed to polyethylene microplastics at concentrations reflecting natural environmental levels under a defined feeding regime and within a controlled environment. Metabolomic profiling of fish muscle tissue using nuclear magnetic resonance spectroscopy revealed no significant metabolic alterations relative to control specimens. To contextualize these experimental results, a large field dataset was integrated using dimensionality reduction, density-based clustering, and machine learning techniques including random forest and Bayesian network analysis. The field data were segregated into two distinct metabolic clusters. Laboratory-treated fish were predominantly grouped within one natural cluster, suggesting limited metabolic disruption under current exposure conditions. Moreover, Bayesian network analysis revealed overlapping metabolic features between laboratory and field samples, demonstrating the emerging potential of embedding-based methods for extrapolating ecological insights from controlled experiments. Collectively, these findings indicate that present-day microplastic exposure, as simulated in this study, may not elicit marked metabolic effects in estuarine gobies, while also highlighting the potential of integrative approaches to link laboratory results with natural ecosystem dynamics.