Antibiotic distribution and driving mechanisms in the island "river-estuary-nearshore" continuum

Current research on land-based pollutants entering marine environments often focuses on individual segments, limiting a comprehensive understanding of pollutant migration along the river-estuary-nearshore (REN) continuum. This study investigates the Nandu River continuum on Hainan Island by integrating positive matrix factorization (PMF) source apportionment with XGBoost-SHAP modeling to elucidate antibiotic distribution patterns and their key driving mechanisms. Results showed that antibiotic concentrations decreased along the gradient of river > estuary > nearshore. In the river section, medical (22.91%) and household medication (21.83%) were the dominant sources, whereas livestock and poultry breeding (21.47%) and aquaculture (20.07%) prevailed in the estuary section. In contrast, scientific, educational, and cultural (13.71%) were the primary contributors in the nearshore section. The XGBoost-SHAP model quantified the relative importance of driving factors as follows: emission sources (ES) (61.87 ng/L) > suspended particulate matter (SPM) (17.75 ng/L) > pH (12.18 ng/L). Antibiotic distribution along the continuum was jointly governed by intrinsic chemical responses to environmental gradients, hydrodynamically controlled transport and particle-mediated processes, and differentiated emission inputs. This study provides mechanistic insights into antibiotic behavior in small-to-medium river-marine continuums and offers scientific support for integrated land-sea environmental management.