Microplastics are ubiquitous in estuarine ecosystems, but evidence on ingestion mechanisms, particularly trait and behavior based drivers, remains limited, hindering a comprehensive understanding of factors and seasonal contrasts governing fish ingestion. We conducted field surveys in May and August at the Changjiang River Estuary in the East China Sea, targeting three representative estuarine coastal fish species: Larimichthys polyactis, Trichiurus japonicus, and Pampus argenteus, with incidental records of other species like Japanese jack mackerel and blue backed mackerel. Using Ivlev's selectivity index, chi-square tests, and multiple correspondence analysis (MCA), we characterized and visualized microplastic trait composition, while random forest (RF) models quantified ingestion selectivity and seasonal variations from May to August across shape, color, polymer type, and particle size. Fibers dominated environmental samples and fish ingestion (71.8% of ingested items), yet MCA revealed stronger structural organization in ingested microplastics, with the first two dimensions explaining 32% of variance (p < 0.001, Chi-squared test) and showing distinct trait bundling. In the upper right MCA quadrant, a clear cluster of green, polyethylene (PE), and linear microplastics emerged, suggesting frequent co-occurrence in fish. Conversely, microplastic traits in the water column were more heterogeneously distributed and lacked a distinct structure, as indicated by the lower variance explanation (13%). This suggests a more scattered and less organized availability of microplastic characteristics in the surrounding environment compared to the selective patterns observed in fish. This study provides a foundation for understanding behavioural selectivity in microplastic ingestion by estuarine fish. By utilizing the Random Forest (RF) model as a detection framework, this research identifies vulnerable seasonal hotspots in the Changjiang River Estuary and provides early warning indicators for specific recurring trait combinations, such as green, PE, and linear microplastics, which are key for targeted long-term monitoring.
Hu et al. (Sat,) studied this question.