An integrated modelling framework to predict the fate and transport of antimicrobial resistance in Singapore coastal waters

Predicting the transport and fate of antimicrobial resistance (AMR) in aquatic environments is crucial for managing this pressing environmental issue. We proposed a hybrid modeling framework that couples process-based and data-driven models to predict the spatiotemporal distribution of antibiotics and their related antibiotic resistance genes (ARGs) in Singapore's coastal waters (SCW). In this study, Lincomycin and its related ARGs were selected for analysis. Firstly, this study provides valuable insights into the complex dynamics of ARGs in coastal waters through the application of a meticulously constructed Random Forest (RF) model. This model helps identify key environmental drivers of ARGs, specifically Lincomycin, pH, zinc, DO and temperature, thereby illuminating the factors influencing ARG levels. Subsequently, we applied a process-based model using the Delft 3D suite to simulate the fate and transport of these key environmental drivers. Finally, the outputs from the process-based model were integrated with the RF model to predict ARGs. The modelling framework was calibrated and validated against monthly data collected from 12 sampling points around SCW during 2022-2023. The results revealed that the simulation performance provided 'reasonable prediction' results, with all modeled targets showing an R² above 0.7 and an NSE greater than 0.8. The research presented in this study not only enhances our understanding of the intricate interplay between environmental variables and ARG levels but also has significant implications for environmental and public health management. By emphasizing the importance of specific environmental factors, these models offer a proactive approach to addressing the urgent challenge of antibiotic resistance in coastal ecosystems. This ultimately contributes to the preservation of both the environment and public health.