Safe drinking water is delivered to the consumer through kilometres of pipes. These pipes are lined with biofilm, which is thought to affect water quality by releasing bacteria into the drinking water. This study describes the number of cells released from this biofilm, their cellular characteristics, and their identity as they shaped a drinking water microbiome. Installation of ultrafiltration (UF) at full scale in Varberg, Sweden reduced the total cell count to 1.5 × 10 3 ± 0.5 × 10 3 cells mL −1 in water leaving the treatment plant. This removed a limitation of both flow cytometry and 16S rRNA amplicon sequencing, which have difficulties in resolving small changes against a high background cell count. Following installation, 58% of the bacteria in the distributed water originated from the pipe biofilm, in contrast to before, when 99.5% of the cells originated from the treatment plant, showing that UF shifts the origin of the drinking water microbiome. The number of bacteria released from the biofilm into the distributed water was 2.1 × 10 3 ± 1.3 × 10 3 cells mL −1 and the percentage of HNA (high nucleic acid) content bacteria and intact cells increased as it moved through the distribution system. DESeq2 analysis of 16S rRNA amplicon reads showed increases in 29 operational taxonomic units (OTUs), including genera identified as Sphingomonas , Nitrospira , Mycobacterium , and Hyphomicrobium . This study demonstrated that, due to the installation of UF, the bacteria entering a drinking water microbiome from a pipe biofilm could be both quantitated and described.

While slow sand filters (SSFs) have produced drinking water for more than a hundred years, understanding of their associated microbial communities is limited. In this study, bacteria in influent and effluent water from full-scale SSFs were explored using flow cytometry (FCM) with cytometric histogram image comparison (CHIC) analysis; and routine microbial counts for heterotrophs, total coliforms and Escherichia coli. To assess if FCM can monitor biofilm function, SSFs differing in age and sand composition were compared. FCM profiles from two established filters were indistinguishable. To examine biofilm in the deep sand bed, SSFs were monitored during a scraping event, when the top layer of sand and the schmutzdecke are removed to restore flow through the filter. The performance of an established SSF was stable: total organic carbon (TOC), pH, numbers of heterotrophs, coliforms, E. coli, and FCM bacterial profile were unaffected by scraping. However, the performance of two newly-built SSFs containing new and mixed sand was compromised: breakthrough of both microbial indicators and TOC occurred following scraping. The compromised performance of the new SSFs was reflected in distinct effluent bacterial communities; and, the presence of microbial indicators correlated to influent bacterial communities. This demonstrated that FCM can monitor SSF performance. Removal of the top layer of sand did not alter the effluent water from the established SSF, but did affect that of the SSFs containing new sand. This suggests that the impact of the surface biofilm on effluent water is greater when the deep sand bed biofilm is not established.