Microplastic contamination and risk profiling in cultured freshwater prawns Litopenaeus vannamei (Boone, 1931) and Macrobrachium rosenbergii (De Man, 1879) from Southwestern India using a machine learning approach.

The presence of microplastic (MP) particles in freshwater species is a growing global concern due to potential impacts on food security and human health. This study investigated MP contamination in two commercially important freshwater species, Macrobrachium rosenbergii (giant freshwater prawn, GP) and Litopenaeus vannamei (white leg shrimp, WS) cultured in inland and semi coastal aquaculture systems in Kollam, southwest India. A total of 307 MP items were identified from pooled gastrointestinal tract (GIT) samples, with average concentrations of 0.709 ± 2 MPs/g GIT in GP and 1.015 ± 2 MPs/g GIT in WS. MPs ranged in size from < 250 µm to 5 mm, with particles in the 500 µm-1 mm range being most prevalent. Blue colored MPs and fiber morphotypes dominated in both species, accounting for 47.31% in GP and 46.28% in WS. ATR-FTIR and Confocal Raman microscope integrated with AFM identified five polymer types: polyethylene, polystyrene, acrylonitrile-butadiene-styrene (ABS), polycarbonate (PC), and polypropylene. To evaluate potential hazards, Polymer Hazard Index (PHI) values were computed, showing a higher hazard in GP (PHI: 29.06) due to the presence of ABS and PC, compared to WS (PHI: 22.92). However, an integrated Pollution Risk Index (PRI), incorporating MP load, PHI, and shape revealed higher ecological risk in WS (PRI: 15.12) due to greater ingestion rates, versus GP (PRI: 13.72). A random forest machine learning model classified MP risk levels with 91.3% accuracy, identifying species, MP density, and polymer type as key predictors. These findings present a novel, integrated framework for MP risk assessment in aquaculture systems.