The Chinese mitten crab (Eriocheir sinensis) industry is currently facing the challenges of origin fraud, as well as a lack of precision and interpretability of existing traceability methods. Here, we propose a high-precision origin traceability method based on a combination of stable isotope analysis and interpretable machine learning. We sampled Chinese mitten crabs from six origins representing diverse aquatic environments and farming practices, and analyzed their δ13C, δ15N, δ2H, and δ18O stable isotope compositions in different sexes and tissues (hepatopancreas, muscle, and gonad). By comparing the classification performance of Random Forest, XGBoost, and Logistic Regression models, we found that the Random Forest model outperformed the others, achieving high accuracy (91.3%) in distinguishing samples from different origins. Interpretation of the optimal Random Forest model, using SHAP (SHapley Additive exPlanations) analysis, identified δ2H in male muscle, δ15N in female hepatopancreas, and δ13C in female hepatopancreas as the most influential features for discriminating geographic origin. This analysis highlighted the crucial role of environmental factors, such as water source, diet, and trophic level, in origin discrimination and demonstrated that isotopic characteristics of different tissues provide unique discriminatory information. This study offers a novel paradigm for stable isotope traceability based on explainable machine learning, significantly enhancing the identification capability and reliability of Chinese mitten crab origin traceability, and holds significant implications for food safety assurance.
This study established novel and high-throughput strategies for the simultaneous analysis of 30 fluorinated emerging pollutants in different matrices from the shrimp aquaculture system in eastern China using UHPLC-MS/MS. The parameters of SPE for analysis of water samples and of QuEChERS methods for sediment and shrimp samples were optimized to allow the simultaneous detection and quantitation of 17 per- and polyfluoroalkyl substances (PFASs) and 13 fluoroquinolones (FQs). Under the optimal conditions, the limits of detection of 30 pollutants for water, sediment, and shrimp samples were 0.01–0.30 ng/L, 0.01–0.22 μg/kg, and 0.01–0.23 μg/kg, respectively, while the limits of quantification were 0.04–1.00 ng/L, 0.03–0.73 μg/kg, and 0.03–0.76 μg/kg, with satisfactory recoveries and intra-day precision. The developed methods were successfully applied to the analysis of multiple samples collected from aquaculture ponds in eastern China. PFASs were detected in all samples with concentration ranges of 0.18–0.77 μg/L in water, 0.13–1.41 μg/kg (dry weight) in sediment, and 0.09–0.96 μg/kg (wet weight) in shrimp, respectively. Only two FQs, ciprofloxacin and enrofloxacin, were found in the sediment and shrimp. In general, this study provides valuable insights into the prevalence of fluorinated emerging contaminants, assisting in the monitoring and control of emerging contaminants in aquatic foods.