Factors Shaping Young and Mature Bacterial Biofilm Communities in Two Drinking Water Distribution Networks

The presence of biofilms in drinking water distribution systems (DWDS) can affect both water quality and system integrity; yet these systems remain poorly studied due to lack of accessibility. We established two independent full-scale DWDS Testbeds (A and B) on two different campuses situated in a tropical urban environment and equipped them with online sensors. Testbed B experienced higher levels of monochloramine and lower water age than Testbed A within the campus. Based on long amplicon-sequencing of bacterial 16S rRNA genes extracted from the mature biofilms (MPB) growing on pipes and young biofilms (YSB) growing on the sensors, a core community was identified in the two testbeds. The relative abundances of operational taxonomic units at the family level, including Mycobacteriaceae, Methylobacteriaceae, Rhodospirillaceae, Nitrosomonadaceae, and Moraxellaceae, were consistent for MPB and YSB on each campus. The MPB community was found to be influenced by conductivity, sample age, and pipe diameter as determined by both canonical correlation analysis and fuzzy set ordination. MPB displayed higher α-diversity based on Hill numbers than YSB; in general, second order Hill numbers correlated positively with conductivity and sample age, but negatively with ORP and nitrite. Pseudomonas spp. together with Bacillus spp. likely initiated biofilm formation of YSB on Testbed A under conditions of reduced monochloramine and high water age. Significant levels of orthophosphate were detected in YSB samples at two stations and associated with higher levels of stagnation based on long-term differential turbidity measurement (DTM). Orthophosphate and DTM may act as indicators of the biofilm growth potential within DWDS. Highlights - Established two testbeds to study biofilms in full-scale distribution system - Biofilms on pipes and sensors had core community - Temporal effect and higher α-diversity for biofilms on pipes - Water chemistry was related to biofilm community differences Graphical Abstract


52
There is an increasing recognition of public health risks in drinking water distribution 53 (DWDS) and premise plumbing systems where water stagnation and decay of residual disinfectant    Two types of biofilm were collected at the same locations: 1) mature pipe biofilms (MPB; 159 extracted pipe wall coupons obtained during installation of the sensor nodes) and 2) young sensor biofilms (YSB; sampled during routine sensor maintenance). Figure S1D

266
The average live bacterial cell counts in the bulk water at locations in Testbed A were 267 significantly higher than measured in the city water at the inlet (see Table S1). Live cell counts Water (BQW) values at these stations (see Table S1). In contrast, there was a lower count of dead 273 cells within the network (A15, A16) compared to the inlet (A11).

274
The average inlet microbial concentration for Campus B and Campus A differed by a factor Campus B (stations B15 and B16) and in the incoming water (stations B13 and B14, see Table S1).

278
A comparison of the two testbeds ( Figure S1) revealed that Campus A had lower levels of residual  Table 2).
18     Similarly, the FSO analysis (Fig. 3b) suggested that conductivity, biofilm age (Table 2) Figure 5 Clustering of young sensor biofilms (YSBs) by biofilm age at OTU level: a) Biofilm sampled two times from A14 (Permanova, p = 0.001); b) Biofilm sampled two times at A16 (Permanova, p = 0.001); c), Clustering of all the YSB samples by sampling location and age. The dash lines indicate 95% confidence level. Permutational multivariate analysis of variance (PERMANOVA) was applied to assess the differences in bacterial community structure among samples and it shows that significant community difference is seen from YSB of different biofilm age from the same station A14, but also A16. The sample name shown in the MPD plot indicates different part of the sensor material.

455
CCA plots were generated to explore the relationship of environmental variables with the 456 community difference. Among these variables, Figure 6 shows that pH, water age, conductivity, 457 DTM, and YSB sample age all contributed to the community differences among YSB samples.

459
Multivariate analysis based on FSO showed that pH and sample age had a significant influence on 460 the YSB community difference with a correlation coefficient greater than 0.7 ( Figure S5). The     (Table 2).

500
Acinetobacter spp. were also identified in YSB samples from both campuses using the plate 501 count method with R2A medium, thus proving their viability. The genus Acinetobacter, 502 containing several species known to cause significant burdens of disease as nosocomial pathogens,  The age of bulk water as well as biofilm together with their respective location in the DWDS 529 had significantly higher influence on the community composition than the substrate for YSB. We thank the staff at the facilities management office of the two campuses for providing access to 548 the networks and information about operational changes within the studied periods. We are also 549 grateful to Ami Preis (Xylem) during the initiation, operation, and support of the online monitoring 550 platform for these two testbeds. We thank de Sessions Paola Florez for help with the sequencing service, and Eric Alm and Gu Xiaoqiong for their discussion of the Venn diagram. The authors 552 would like to thank Eric Dubois Hill for his invaluable help with the visualization of the data.

554
The authors declare no conflict of interest.  All data supporting the findings of this study are available from the corresponding author 565 upon request.