Examining the Effect of Small-Amplitude Transients on Bacterial Community Composition in Water Distribution Pipes ^†

Abstract

This research explores the microbial ecology within drinking water distribution systems (DWDS). The aim of the paper was to examine the impact of small-amplitude transients on the bacterial composition in bulk water and biofilms. The findings highlight the dominance of Proteobacteria and the significant transformation of microbial communities in response to changes in environmental conditions. The study underscores the complex interplay between operational strategies and microbial dynamics in DWDS and emphasizes the necessity for ongoing research to enhance water quality management amidst climate change challenges.

1. Introduction

Recent investigations into the microbial ecology of water distribution systems have illuminated the significant impact of various factors on the characteristics of biofilm microbial communities. The dynamic interplay between hydraulic management [1,2], pipe material [3,4], chlorination levels [5], and water temperature [2] shapes these communities, and influences water quality and system resilience. Biofilm formation and its dynamics can be greatly influenced by the shear stress exerted by fluid flow over the surfaces where biofilms develop [6]. To date, no one has examined how short-lived accelerations caused by small-amplitude transients influence the bacterial community composition in bulk water and biofilms of water distribution pipes. The aim of the paper is to examine the influence of such small-amplitude transients in determining the bacterial community composition in the bulk water and biofilms adhered to drinking water pipes.

2. Material and Methods

Experiments were performed in the Drinking Water Distribution Laboratory (DWDL) at Queen’s University, a unique research facility in North America with two full-scale pipe loops (A and B), capable of partially reproducing operational conditions of drinking water networks. For these experiments, a fixed volume of local drinking water was recirculated in both pipe loops at steady flows of 0.6 L s⁻¹. But in Pipe loop B, a small transient was introduced by periodically closing a solenoid valve for 30 s each hour, temporarily reducing the flow rate to 0.54 L s⁻¹. These hydraulic conditions were maintained during 28 days. Biofilm growth was aided by the inoculation of the water with an initial community of microorganisms harvested from local drinking water and the continuous addition of small concentrations of Nutrient Broth No. 3 (0.60 mg L⁻¹ day⁻¹). Bulk water samples for bacterial community composition assessment were collected on days 1, 14, and 28 of the experiment and during a flushing operation at the end of the experiment. To achieve minimal quantities of DNA for analysis, large sample volumes were concentrated with a series of centrifugation steps. In sequence, DNA was extracted using QIAGEN’s DNeasy PowerSoil Pro Kit and sequenced using NextGen 16s rRNA and MiSeq System. Specification down to bacteria genus level was achieved for most of the samples.

3. Results

The initial microorganisms harvested from local drinking water that were used to start the experiments were identified as predominantly Gram-negative Proteobacteria (94.4%), more specifically belonging to the genus Acinetobacter (56%), Massilia (17%), Pseudomonas (10%), Brevundimonas (5.6%), Delftia (3.8%), and Agrobacterium (1.3%). Additional relevant microorganisms in this starting batch included Bacillus (1.5%), Chryseobacterium (1.2%), Prevotella (0.6%), and Sediminibacterium (0.5%). Bacillus, the sole Gram-positive bacteria from the Firmicutes Phylum, was detected alongside Gram-negative bacteria from the Bacteroidetes Phylum.

After the inoculation of the start microorganisms in the pipe loops, substantial changes were observed in the community composition of bacteria during the experiments. Figure 1 details the predominant bacterial genera identified, accounting for over 90% of the microbial composition observed. The top 15 bacterial genera are shown as circular charts for four different experimental stages (Days 1, 14, 28, and Flushing) and for both pipe loops A (no transients) and B (transients). A quite dramatic change in the bacteria community composition from the start of inoculum was already evident on the first day after the inoculation (inner circles from A and B in Figure 1), possibly due to microbe selection caused by the new environment of the pipe loops. One day after the inoculation (Day 1), the initial diversity was slightly reduced to only three predominant bacteria genera (Acinetobacter, Pseudomonas, and Bacillus). In particular, the depletion of Massilia bacteria was evident at this stage of the experiment. Differences between bacterial proportions were also observed on Day 1, showing that Acinetobacter fully dominated in pipe loop A (81%) in comparison to pipe loop B (47.1%). In addition, a higher proportion of Pseudomonas bacteria were identified in pipe loop B (35%) than in pipe loop A (12%). Pseudomonas are known for their capacity to produce extra-cellular polymeric substance (EPS) [5], and their initially higher numbers in pipe loop B may be connected to the presence of small-amplitude transients produced in this system. From the initial bacterial composition, the abundant Acinetobacter was depleted by Day 28, while Pseudomonas fractions were reduced but prevailed until the flushing stage and became more abundant in pipe loop A than in pipe loop B after Day 14. Bacterial diversity increased as the experiments wore on. The predominant bacteria in the later stages of the experiment included (i) Sediminibacterium (widely found in pipe loop B); (ii) Legionella (predominant in pipe loop A); and two abundant bacteria (Unknown #1 and Unknown #2) especially thrived in pipe loop B, classified to family Burkholderiales and Chloroplast, respectively.

The bacterial diversity was again increased in the flushing stage, likely because biofilms were mobilized from the pipe wall and released into the bulk water. During the flushing stage, an additional 86 genera were identified in pipe loop A and an additional 67 genera were identified in pipe loop B, under the “others” category in Figure 1. Despite this, both pipe loops still shared a similar community composition of predominant bacteria. A large increase of Unknown genus #1 and Unknown genus #2 was observed between Day 28 and the flushing samples, suggesting that they were dominant in the biofilm during the conditioning and flushing stages. Nevertheless, a total of 9 out of the 15 predominant bacterial genera highlighted in Figure 1 were found exclusively from pipe loop samples (not a single copy of their genes was identified in the initial bacteria batch). It is hypothesized that certain bacteria, undetected in initial samples due to their low numbers, were initially present in the systems and thrived and outcompeted established microbes during the experiment. This suggests their strong adaptation to the pipe loop environment.

4. Discussion

The experimental results showed that besides the Gram-positive Bacillus bacteria, other predominant microbes were mainly Gram-negative Proteobacteria, which are known for being resilient to challenging environments and have also been widely reported in previous studies investigating drinking water microbiomes [7,8]. Among them, bacteria from the Acinetobacter and Pseudomonas genera were dominant in the experimental samples collected. Acinetobacter has been highlighted for its significant chlorine resistance [9] and potential for antimicrobial resistance, while Pseudomonas is noted for its biofilm formation abilities [5,7]. Their early presence and subsequent dominance suggest a foundational role in biofilm structure creation, promoting diverse microbial community integration and persistence in water distribution systems. In addition, the later appearance of Legionella in the experiments suggests that small concentrations of viable bacteria had survived the disinfection of local drinking water. Many of the additional bacteria genera found later in the experiment were also previously reported in drinking water systems worldwide, among them are possibly chlorine-resistant bacteria (e.g., Bosea, Hydrogenophaga), nitrifying bacteria (e.g., Niveispririllum), and photosynthetic bacteria (e.g., Rhodobacter), which highlight the versatile functioning capacity of these communities [5].

Observed differences in microbial compositions between the control (pipe loop A) and transient affected (pipe loop B) setups lacked complete explanations. The higher abundance of Sediminibacterium in pipe loop B bulk water in comparison to pipe loop A may suggest that these bacteria could not colonize pipe loop B pipe walls because of the presence of transients. In addition, besides a similar dominance of the bacterial genera between pipe loop A (no transients) and pipe loop B (with transients), a higher final diversity was clearly observed in the control pipe loop A (no transients). This suggests that the frequent, small-amplitude transients in pipe loop B may have contributed to reducing some of the bacteria that thrived in pipe loop A.

5. Conclusions

In summary, bacteria identified in the experiments were harvested from local drinking water and most of them have been previously found in other DWDS worldwide, including bacterial genera that are connected to chlorine and antimicrobial resistance, and possibly opportunistic pathogens. The production of small-amplitude transients in pipe loop B did not seem to have produced substantial differences in the bacterial community composition as compared to loop A, which experienced no transients. A slightly higher diversity was observed in pipe loop A (control loop with no transients), suggesting that the small-amplitude transients in loop B may have somehow had an inhibitory effect on biofilm development in this rig. As expected, a higher diversity of bacteria was found to occur at the flushing stage when biofilms were mobilized from pipe walls. Further studies are required to confirm this and clarify how unsteady regimes may alter biofilm occurrence in DWDS.