Prioritized pollutant data collection and screening

The PI was used to generate a priority pollutant list. This list is of great significance for the protection of the Yangtze River and the continuous improvement of water environmental quality. 1.1 Data Collection 1.1.1 Persistence and Bioaccumulation The persistence of pollutants is evaluated by the biodegradation coefficient (BIOWIN), with a smaller BIOWIN value indicating stronger persistence and greater difficulty in degradation of the pollutants in the environment.

The bioaccumulation is assessed by the octanol-water partition coefficient (Kow), which can reflect the migration ability of the pollutants between the aqueous phase and organisms. A larger Kow value indicates stronger hydrophobicity and higher bioaccumulation of the pollutants. The BIOWIN value is estimated by the BIOWIN 3 model of the EPI Suite 4.1 software [52], and the Kow value is obtained by the KOWWIN model of the EPI Suite 4.1 software.

In cases where there is no experimental data, the estimated data is used as the Kow value. 1.1.2 Ecotoxicity The impact of pollutants on the ecosystem is evaluated using PNEC (ng·L−1). PNEC is primarily calculated using chronic toxicity data (NOEC) and assessment factors (AF). Based on the availability of nutrient level data for algae, crustaceans, and fish, the AF values are 100, 50, and 10 respectively.

If there is no NOEC, PNEC is calculated using the half-effect concentration (EC50) or half-lethal concentration (LC50) and the AF is set at 1000. The toxicity data (NOEC, EC50, LC50) are sourced from the ECOTOX toxicity database of the United States. During the process of screening priority pollutants, when multiple NOEC values are available, the lowest value is adopted to best prevent potential toxicity risks. For OMPs without experimental measured ecological toxicity data, the NOEC for three trophic levels (i.e., algae, crustaceans, and fish) is obtained by the ECOSAR model of the EPI Suite 4.1 software. 1.2 Priority of Pollutants 1.2.1 Data Processing and Standardization The data preprocessing work involves the following steps: The OMPs concentrations below the method detection limit (MDL) are replaced by half of the MDL.

The median concentration is used for the prioritization of pollutants because it is less affected by outliers and ND responses. Due to the wide range of data distribution, the PNEC and concentration values are respectively converted to logarithms with base 10 and base 2. The data is normalized to a dimensionless value ranging from 0 to 1 using the min-max normalization method [54], which retains the distribution characteristics and relative magnitude relationship of the original data while reducing the interference of outliers on the overall data structure. 1.2.2 Multivariate Analysis Method Principal Component Analysis (PCA) is a classic dimensionality reduction method.

PCA constructs a comprehensive evaluation variable by extracting the principal component (PC1) with the highest variance contribution. This variable has been used to characterize the persistence, bioaccumulation, and ecotoxicity of OMPs. In this study, PCA analysis was conducted on the standardized data of the persistence, bioaccumulation, and ecotoxicity of pollutants.

PC1 was defined as the comprehensive hazard potential (HP) to reflect the cumulative effect of multi-dimensional hazard effects, and the HP threshold was set at 0.4 to screen out pollutants. For the standardized data of the exposure factors of pollutant concentration and detection rate, an exposure potential (EP) variable was constructed through PC1 to quantify the exposure risk, and the EP threshold was set at 0.5 to screen out pollutants. 1.2.3 Priority Convert EP and HP into dimensionless items within the range of 0 to 1.

By multiplying the normalized EP and the normalized HP, the pollutant PI value is determined. Since the five main factors related to priority ranking have been normalized into dimensionless variables, they can represent the independent characteristics of the pollutants.