Deciphering response effect and underlying mechanism of anammox-based nitrogen removal process under exposures to different antibiotics via big data analysis

Abstract

Recently, little research has been devoted to the systematic investigation regarding the effect of different antibiotics on anammox-based system and its underlying mechanism. In this study, a critical inhibition concentration to the anammox-based system was obtained: 2, 0.5 and 5 mg/L for OTC, TC and SM, respectively. However, SPM had no significant inhibition. Furthermore, Exp model and Monod model were capable to describe the inhibition period, while Gauss model was suitable for the recovery period. A universal machine learning model could accurately predict the NRR (R² over 0.9), especially when biomass information data was introduced. As a qualitative analysis, the inhibition effect of TC and OTC was strongest. The abundance of nitrogen functional genes was negatively correlated with antibiotics, while antibiotic resistance genes showed the opposite trend. Overall, the inhibition ratios of OTC, TC, SPM and SM on anammox process were calculated to be 91%, 82%, 50% and 30%, respectively.

Keywords: Anammox; Antibiotic; Kinetic; Machine learning; Nitrogen.