Evaluation of reduction efficiencies of pepper mild mottle virus and human enteric viruses in full-scale drinking water treatment plants employing coagulation-sedimentation–rapid sand filtration or coagulation–microfiltration

•Reduction of PMMoV and human enteric viruses in four full-scale DWTPs was evaluated.•The developed virus concentration method achieved high recovery (>30%) of PMMoV.•CS–RSF and C–MF reduced PMMoV by about 1–3-log10.•Low coagulant dosage led to low reduction ratios (1-log10) of PMMoV in C–MF.•Reduction ratios of EVs and HuNoV GII were higher than those of PMMoV in CS–RSF. Here, we evaluated the reduction efficiencies of indigenous pepper mild mottle virus (PMMoV, a potential surrogate for human enteric viruses to assess virus removal by coagulation-sedimentation–rapid sand filtration [CS–RSF] and coagulation–microfiltration [C–MF]) and representative human enteric viruses in four full-scale drinking water treatment plants that use CS–RSF (Plants A and B) or C–MF (Plants C and D). First, we developed a virus concentration method by using an electropositive filter and a tangential-flow ultrafiltration membrane to effectively concentrate and recover PMMoV from large volumes of water: the recovery rates of PMMoV were 100% when 100-L samples of PMMoV-spiked dechlorinated tap water were concentrated to 20 mL; even when spiked water volume was 2000 L, recovery rates of >30% were maintained. The concentrations of indigenous PMMoV in raw and treated water samples determined by using this method were always above the quantification limit of the real-time polymerase chain reaction assay. We therefore were able to determine its reduction ratios: 0.9–2.7-log10 in full-scale CS–RSF and 0.7–2.9-log10 in full-scale C–MF. The PMMoV reduction ratios in C–MF at Plant C (1.0 ± 0.3-log10) were lower than those in CS–RSF at Plants A (1.7 ± 0.5-log10) and B (1.4 ± 0.7-log10), despite the higher ability of MF for particle separation in comparison with RSF owing to the small pore size in MF. Lab-scale virus-spiking C–MF experiments that mimicked full-scale C–MF revealed that a low dosage of coagulant (polyaluminum chloride [PACl]) applied in C–MF, which is determined mainly from the viewpoint of preventing membrane fouling, probably led to the low reduction ratios of PMMoV in C–MF. This implies that high virus reduction ratios (>4-log10) achieved in previous lab-scale virus-spiking C–MF studies are not necessarily achieved in full-scale C–MF. The PMMoV reduction ratios in C–MF at Plant D (2.2 ± 0.6-log10) were higher than those at Plant C, despite similar coagulant dosages. In lab-scale C–MF, the PMMoV reduction ratios increased from 1-log10 (with PACl [basicity 1.5], as at Plant