China is facing increasing marine microplastic pollution. Despite the fact that the South China Sea is the largest marine area in China, the ecological danger and present state of microplastic contamination in this region have not been systematically and comprehensively investigated. This study analyzed the abundance, distribution, and characteristics of microplastics in different environmental media and biological samples from the Xisha Islands in the South China Sea, and then the ecological risk assessment of microplastic pollution in this area was conducted. The findings indicated that the quantities of sediments, soil, water, fish, and birds were 41.56 ± 19.12 items/kg, 92.94 ± 111.05 items/kg, 2.89 ± 1.92 items/L, 2.57 ± 2.12 items/ind, and 1.702 ± 1.50 items/ind, respectively. By evaluating the pollution load index (PLI), polymer hazard index (PHI), and potential ecological risk index (PERI), the PLI of the Xisha Islands in the South China Sea as a whole indicated that the hazard level was slightly polluted, the PHI was at a high-risk level, and the PERI samples were at no risk, except for the soil and seawater, which were at a medium-risk level.

The agricultural sector plays a pivotal role in Hainan Province, China; therefore, the utilization of pesticides is indispensable.

Numerous organic chemicals are present in aquatic environments, yet systematically identifying and prioritizing these compounds remains a major challenge, particularly in tropical island watersheds where data are scarce. In this study, we applied high-resolution mass spectrometry (HRMS)-based non-target screening to comprehensively characterize emerging organic pollutants in three major rivers of Hainan Province-Changhua, Wanquan, and Nandu. A total of 177 high-confidence compounds were identified, spanning pharmaceuticals, industrial additives, pesticides, and natural products. To apportion pollutant sources, we employed non-negative matrix factorization (NMF), which revealed distinct anthropogenic signatures across the rivers, including domestic sewage, pharmaceutical discharges, and agricultural runoff. Using an integrated Toxicological Priority Index (ToxPi), we further prioritized 29 substances of elevated concern (ToxPi > 4.41, i.e., Mean + SD), such as stearic acid, tretinoin, and ethyl myristate, based on their detection frequency, relative abundance, bioconversion half-life, bioconcentrating factor, bioaccumulation factor, and predicted no-effect concentrations (PNECs).This study presents one of the first systematic applications of HRMS-based non-target screening combined with machine learning (NMF) and semi-quantitative risk scoring (ToxPi) in a tropical island river system. The findings offer novel insights into the chemical fingerprint of such understudied ecosystems and establish a replicable framework for pollution assessment and prioritization under data-limited conditions. This work provides a critical foundation for targeted pollution control, helping to balance ecological protection with ongoing regional development.