A Hybrid Modeling Framework for Predicting Spatiotemporal Dynamics of Antimicrobial Resistance in Coastal Waters

Abstract

Antimicrobial resistance (AMR) in aquatic environments poses a critical threat to both environmental and human health. This study presents a novel hybrid modeling framework that integrates a process-based hydrodynamic-environmental model with a data-driven approach to predict the spatiotemporal dynamics of AMR in coastal waters. Macrolide-related antimicrobial resistance genes (ARGs_Macro) were selected as representative markers. The model results were validated using data from a monthly sampling campaign conducted across Singapore’s coastal waters, yielding a mean coefficient of determination (R²) of 0.693, a Nash-Sutcliffe efficiency (NSE) of 0.589, and a root-mean-square deviation (RMSE) of 0.0257 GC/16S rRNA across 12 sampling points. Lincomycin, pH, dissolved oxygen, zinc and temperature were identified as significant influencers of ARGs_Macro. Although Lincomycin is not classified as a macrolide, it ranks as the most important driver of ARGs_Macro due to its shared resistance mechanisms with macrolides, potentially facilitating cross-resistance. The spatiotemporal model results revealed that coastal areas, particularly in the northern part of Singapore, are vulnerable to significant ARG accumulation, with monsoon seasons amplifying the spread of AMR due to hydrodynamic conditions. This study highlights the development of a robust modeling framework that provides valuable insights into the environmental drivers of AMR in coastal waters, offering a foundation for regulatory strategies and future research aimed at mitigating the risks of antimicrobial resistance in aquatic environments.