Identifying urban river antibiotic sources by integrating fluorescence characterization and path modeling as an...

This study investigates urban rivers in Ningbo, a coastal city in southeastern China, to assess how varying levels of sewage infrastructure affect antibiotic contamination and to evaluate an integrated source-tracing framework combining partial least squares structural equation modeling (PLS-SEM) and excitation-emission matrix-parallel factor analysis (EEM-PARAFAC). Antibiotic concentrations ranged from 352.19 ± 151.24-2082.35 ± 865.50 ng·L^-1, with substantially higher levels observed in rivers with less complete sewer networks. Dehydrated erythromycin (ETM-H₂O) was the dominant antibiotic, accounting for 45-94 % of total concentrations. While PMF attributed pollution to livestock, aquaculture, and pharmaceutical industry sources, these attributions are inconsistent with the actual industrial profile and anthropogenic activities in the region. In contrast, PLS-SEM revealed a strong direct effect of tryptophan-like DOM, indicative of domestic sewage, on both total antibiotics (β = 0.721, p < 0.05) and ETM-H₂O (β = 0.789, p < 0.01), identifying sewage leakage as the primary source. Tryptophan-like components also indirectly influenced fluoroquinolones and roxithromycin via phosphorus mediation (β = 0.264 and 0.552, respectively), suggesting additional contributions from manure fertilization in riparian vegetable fields. This study demonstrates the value of coupling DOM fluorescence and path modeling for identifying antibiotic sources in urban waters.