Predicting distribution coefficients for antibiotics in a river water–sediment using quantitative models based on their spatiotemporal variations

Highlights: • Significant spatial variations along Liuxi River with more antibiotics at downstream • TMP and NFX are most frequently detected in water and sediment, respectively. • Distribution coefficient (Kd) of all 14 antibiotics were higher than reported values. • pH and specific surface area were important sediment properties in building models. • 12 models with high robustness and precision were developed to predict antibiotic Kd. -- Abstract: Antibiotics are widely used in humans and animals, but their presence in environmental matrices after use is of great concern. Distribution behavior of antibiotics in natural water–sediment systems is influenced by sediment properties, but how these properties, such as surface area, affect their distribution between water and sediment phases remains unclear. The concentrations of antibiotics also vary both spatially and temporally. In this study, a solid/liquid distribution coefficient Kd(pre) was proposed and evaluated in 12 quantitative predicting models based on aquatic field data compared with a bulk coefficient Kd. Results confirmed by the occurrence pattern, concentration levels and spatiotemporal distributions indicated that the characteristics of antibiotics pollution in rural northwestern Guangzhou were generally consistent with previous investigations, suggesting that this investigation was representative of the present aquatic pollution status of antibiotics. The median concentrations were < 100 ng·L⁻¹ and 220 ng·g⁻¹ (d.w.) in the water and sediments, respectively. The most pronounced high concentrations of total antibiotic residue found were 778.0 ng·L⁻¹ for sulfonamides (SAs) in water and 1596.9 ng·g⁻¹ (d.w.) for fluoroquinolones (FQs) in sediments at site 13 in December of 2016, probably due to its dense population, high frequency of antibiotic use and low water flow. Moreover, 12 quantitative models were established with a high level of robustness and ability to spatiotemporally predict the Kd for each of the 12 antibiotics. The models revealed that pH, organic matter and specific surface area of sediments played significant roles in influencing the adsorption of SAs, FQs, tetracyclines (TCs) and (macrolides) MLs. Our findings provide insights into the effects of physicochemical properties on distribution of antibiotics, predicting their fate and transport, as well as assessments of exposure and risk of these emerging pollutants to aquatic ecosystems.