An operational decision tree strategy optimizing the pairing of receptor model and classification approach in source...

Microplastics (MPs) are characterized by multiple attributes such as morphology and polymers that enable diverse classification approaches. The source apportionment of MPs is significantly different from that of traditional pollutants, whose classification relies solely on chemical composition. Current research lacks quantitative methods in microplastic source apportionment, with receptor models' reliability unverified and optimal classification approaches unclear. In this study, the performances of receptor models (Principal Component Analysis-Multiple Linear Regression (PCA-MLR) and Positive Matrix Factorization (PMF)) combined with classification approaches (polymeric and morphological) across source types (primary and secondary) and complexities (2-6 sources) were evaluated based on 56 scenarios. Results showed that PCA-MLR using morphological classification failed to resolve actual source profiles in all scenarios, with the Pearson correlation coefficient (r) between simulated and true profiles below 0.3. Under polymeric classification, PMF consistently yielded simulated source profiles significantly correlated with true values (p < 0.01) across all scenarios, with r ranging from 0.823-1.000. For ≤ 4 sources, PMF using morphological classification also performed well, potentially exceeding polymer-based results. The models exhibited distinct preferences in evaluating source impact intensities: PCA-MLR better resolved primary sources (r = 0.990 ± 0.013) than secondary sources (r = 0.973 ± 0.037), while PMF excelled for secondary sources (r = 0.990 ± 0.010) more than primary sources (r = 0.959 ± 0.055). Building on these findings, we developed an operational decision tree strategy optimizing model-classification pairing upon target source type and complexity. This framework significantly enhances the efficiency and accuracy of microplastic source apportionment, providing critical support for precise source-specific risk mitigation in aquatic environments.