Replicable simulation of distal hot water premise plumbing using convectively-mixed pipe reactors

M Storme Spencer; Abraham C Cullom; William J Rhoads; Amy Pruden; Marc A Edwards

doi:10.1371/journal.pone.0238385

. 2020 Sep 16;15(9):e0238385. doi: 10.1371/journal.pone.0238385

Replicable simulation of distal hot water premise plumbing using convectively-mixed pipe reactors

M Storme Spencer ¹, Abraham C Cullom ¹, William J Rhoads ¹, Amy Pruden ¹, Marc A Edwards ^1,^*

Editor: Zhi Zhou²

PMCID: PMC7494094 PMID: 32936810

Abstract

A lack of replicable test systems that realistically simulate hot water premise plumbing conditions at the laboratory-scale is an obstacle to identifying key factors that support growth of opportunistic pathogens (OPs) and opportunities to stem disease transmission. Here we developed the convectively-mixed pipe reactor (CMPR) as a simple reproducible system, consisting of off-the-shelf plumbing materials, that self-mixes through natural convective currents and enables testing of multiple, replicated, and realistic premise plumbing conditions in parallel. A 10-week validation study was conducted, comparing three pipe materials (PVC, PVC-copper, and PVC-iron; n = 18 each) to stagnant control pipes without convective mixing (n = 3 each). Replicate CMPRs were found to yield consistent water chemistry as a function of pipe material, with differences becoming less discernable by week 9. Temperature, an overarching factor known to control OP growth, was consistently maintained across all 54 CMPRs, with a coefficient of variation <2%. Dissolved oxygen (DO) remained lower in PVC-iron (1.96 ± 0.29 mg/L) than in PVC (5.71 ± 0.22 mg/L) or PVC-copper (5.90 ± 0.38 mg/L) CMPRs as expected due to corrosion. Further, DO in PVC-iron CMPRs was 33% of that observed in corresponding stagnant pipes (6.03 ± 0.33 mg/L), demonstrating the important role of internal convective mixing in stimulating corrosion and microbiological respiration. 16S rRNA gene amplicon sequencing indicated that both bulk water (P_adonis = 0.001, R² = 0.222, P_betadis = 0.785) and biofilm (P_adonis = 0.001, R² = 0.119, P_betadis = 0.827) microbial communities differed between CMPR versus stagnant pipes, consistent with creation of a distinct ecological niche. Overall, CMPRs can provide a more realistic simulation of certain aspects of premise plumbing than reactors commonly applied in prior research, at a fraction of the cost, space, and water demand of large pilot-scale rigs.

Introduction

Opportunistic pathogens (OPs), such as Legionella, non-tuberculous mycobacteria, and Pseudomonas aeruginosa, now account for the primary source of tap-water associated disease in the U.S. and much of the world [1, 2]. Because these organisms grow in the premise (i.e., building) plumbing environment, rather than originating from fecal-contaminated source water, substantial attention has been expended towards identifying factors that stimulate their growth [3]. However, such efforts are hampered by lack of a suitable laboratory test apparatus that accurately represents premise plumbing conditions, while also being replicable to provide sufficient statistical power for evaluating various factors of interest [4].

There are numerous challenges to effectively simulating premise plumbing conditions in the laboratory. Existing approaches inevitably involve compromise in terms of complexity, cost, replicability, and/or ability to achieve relevant hydraulic and water chemistry regimes. Annular reactors and CDC biofilm reactors [5, 6] are designed with the intention of enabling continuous or semi-continuous flow with coupons spinning internally to produce hydraulic sheer stress, such as that experienced by a biofilm on the interior pipe wall [4, 7, 8]. However, these and other bench-scale plumbing simulations still do not achieve realistic plumbing flow patterns, hinder replication through their large size and cost, and are characterized by large amounts of unrepresentative surface area comprised of materials not used in plumbing including glass and plastic. More realistic pilot-scale plumbing simulations are large, costly, and require very large volumes/flows of water, which also makes influent water chemistry conditions difficult to precisely control [9, 10]. Pilot-scale studies examining OPs also typically require direct connection to premise plumbing of the study facility and cannot be sampled within the protection of a biological safety-level (BSL) 2 certified cabinet, elevating potential for exposure of workers to pathogen-containing aerosols during sampling and thus requiring appropriate institutional approvals.

Hydraulic conditions create distinct ecological niches, e.g., with mostly stagnant versus continuously flowing pipes representing two extremes, by controlling the temperature and delivery of disinfectants (e.g., chloramine, chlorine, copper) or nutrients (e.g., oxygen, organic carbon, nitrogen, phosphorus) to biofilms [11, 12]. Alternating periods of flow and stagnation create extreme fluctuations in temperature, disinfectant concentrations, nutrients, and metabolic products in the water that control microbial growth rates and can select for certain organisms [13]. Field measurements of water drawn from premise plumbing following overnight stagnation have documented over 3-log increases in total heterotrophic bacterial counts as a result of depleted disinfectant residuals [14, 15].

On the other hand, recent research has revealed that portions of premise plumbing from which consumers are not drawing water are not truly stagnant during periods of non-use, but can be subject to rapid internal convective mixing due to temperature gradients [16]. Convective mixing is characteristic of certain distal reaches of hot water premise plumbing, resulting in gentle circulation and sustained warm temperatures known to be ideal for growth of L. pneumophila and other OPs. Orientations in which hot water flows upwards and cools drives convective mixing and has correspondingly been observed to affect microbial community composition [17]. In another illuminating experiment, intermittent delivery of hot water at 51°C followed by rapid cooling to room temperature during stagnation produced two orders of magnitude higher levels of L. pneumophila compared to a situation with more frequent flow events delivering 51°C [10]. This is to be expected based on the common practice of heating water samples prior to culture in order to select for heat-tolerant Legionella versus conventional heterotrophs.

Here we introduce the convectively-mixed pipe reactor (CMPR) as a simple, replicable, and realistic system for premise plumbing simulations to evaluate factors contributing to the growth of OPs. The CMPR consists of off-the-shelf materials used in real-world plumbing systems. Capped pipe segments, with one end submerged in a hot water bath and the other contacting the cooler ambient air, simulate the premise plumbing riser from a hot water recirculation loop that connects to distal, stagnant outlets (Fig 1A). This temperature gradient recreates natural internal flow via convective currents (Fig 1B), constantly circulating water without use of pumps or mechanical mixers, substantially reducing the cost, maintenance requirements, and issues associated with inevitable mechanical malfunction. The simple design supports closed system operation of replicates to enable statistical rigor in experimental design. This study evaluates the overall reproducibility of physicochemical properties and microbial community compositions produced by CMPRs using three pipe materials (PVC, PVC-copper, and PVC-iron) as compared to the same configurations maintained under stagnant, constant temperature conditions. The findings are put into context with other studies to evaluate relative advantages and disadvantages of alternative premise plumbing simulations.

Fig 1 — (A) Capped ends of PVC-copper, PVC-iron, and PVC pipes are submerged at 60° in a hot water bath to simulate hot water recirculation, with plugged, accessible ends exposed to room temperature to simulate a stagnant distal outlet. Fifty-four pipes, configured into 6 rows of 9 pipes, were operated for this study. In-line ultraviolet light is incorporated for disinfection of water bath for secondary containment/disinfection of pathogens in case of leak to support BSL2 level experiments. (B) This configuration of the pipes induces convective mixing in the interior bulk water.

Materials and methods

Convectively mixed pipe reactor (CMPR) design and operation

Passive, convective mixing was achieved by submerging the capped end of sealed 4-ft pipe segments (CMPRs) into a heated water bath housing unit at a 60° angle with the other half exposed to the ambient room temperature (Fig 1A). A 60° angle was chosen because it maximizes convective mixing (Fig 1B, S1 Fig and S1 Table). An internal circulating water velocity of ~40 cm/s was determined by injecting neutrally buoyant Rhodamine dye pre-heated to 37°C into a clear PVC CMPR and timing the mixing velocity via video and stopwatch. Each end of the CMPRs were sealed with silicone stoppers to prevent changes in pH associated with atmospheric CO₂ transfer.

The CMPRs were composed either entirely of ¾” PVC (Silver-line Plastics, Ashville, NC) (PVC pipes/CMPRs), half ¾” PVC and half type M copper pipe (McMaster-Carr, Elmhurst, IL) (PVC-copper pipes/CMPRs), or 3” of mild steel pipe (McMaster-Carr, Elmhurst, IL) attached to 45” of ¾” PVC (PVC-iron pipes/CMPRs). These three materials exhibited distinct heat conduction properties and would in turn create distinct convective mixing patterns and velocities if submerged in the water bath. Thus, for purposes of this experiment, PVC segments were submerged in the water bath to normalize convective mixing across the conditions. To ascertain a high-resolution view of the variability of chemical and biological water quality parameters among CMPRs, 18 replicates of each material were used. In addition, triplicates of each pipe were placed at a 60° angle against a shelf in a constant temperature room averaging 38°C (ideal for growth of Legionella and other OPs) to compare the effects of convective mixing to fully stagnant, heated pipes.

The influent water to both CMPRs and stagnant pipes was initially seeded with backwash water from a granular activated carbon filter that had been in operation in a premise plumbing drinking water system for > 2 years to establish a mature microbial community. Water changes were conducted once weekly for the first 2 weeks of operation to facilitate colonization of the pipe surfaces. Water changes were then increased to twice weekly to better simulate an infrequent use pattern characteristic of distal outlets in large buildings, as a possible worst-case scenario for OPs control, until the end of the experiment (week 10). These periodic manual dump and fill water changes served to recreate turbulent intermittent flow and complete changeover of water, as occurs at infrequently used distal outlets. In order to mitigate changes in pH due to headspace and exchange with the atmosphere, and to prevent spills to facilitate testing of pathogens, equal volumes of water were measured and poured into the pipes to maintain consistent headspace (~1”) between CMPRs. Influent water was prepared with de-chloraminated Blacksburg, VA tap water via breakpoint chlorination to destroy residual ammonia. The remaining chlorine residual was quenched with sodium thiosulfate, and pH was adjusted to 7.5 using 1 M hydrochloric acid. A summary of influent water characteristics is provided in S2 Table.

The water bath in which the CMPRs were partly immersed was heated by a 22.7 L (6 gal) water heater and continuously recirculated through the housing unit (a 48” x 30” x 15”, 307 L (81 gal) fiberglass tank) to achieve a steady-state temperature of approximately 45–48°C (which is typical of lower end of water heater temperature settings) as well as an in-line ultraviolet disinfection unit (Viqua, Guelph, ON) to prevent microbial proliferation in the water bath and for biosafety in the event of a spill, as well as achieve the target bulk water temperature in the CMPRs. One recirculation pump delivered water from the heater to the CMPR housing unit and another pump delivered water from the CMPR housing unit back into the water heater. The unit is enclosed with a high-temperature, ultra-high molecular weight plastic cover (McMaster-Carr, Elmhurst, IL) reinforced with steel rods and fixed in position by clamps. Water was supplied to the heater using a valve connection to periodically change and/or refill the housing unit.

Water quality analysis

Temperature, dissolved oxygen (DO), pH, total organic carbon (TOC), and total and dissolved metals were measured at weeks 2 and 9 of the experiment. These timepoints were selected to profile the pipes following initial microbial colonization and at a later period when differences in biological water quality parameters would be expected based on pipe material [18, 19]. Samples were collected by inverting the pipe three times and decanting the contents into a sterile 1 L polypropylene bottle. pH and temperature were measured using a pH 110 meter with automatic temperature correction (Oakton Research, Vernon Hills, Il). DO was measured using a polarized DO probe (Thermo Fisher Scientific Orion 3-star meter, Waltham, MA). TOC was measured using a Sievers Model 5300C autosampler according to Standard Method 5310 C. Total and soluble copper (MDL = 0.08 ppb, MRL = 1.00 ppb) and iron (MDL = 0.36 ppb, MRL = 5.00 ppb) were measured following acidification with 2% v/v nitric acid using inductively coupled plasma mass spectroscopy (iCAP RQ ICP-MS; Thermo Fisher Scientific, Waltham, MA). Samples for soluble metal analysis were immediately filtered through a 0.45-μm nylon filter (Whatman, Maidstone, UK) prior to acidification and measurement.

Biological sampling

Bulk water total cell counts were taken during weeks 1, 2, 5, and 9, with an additional sampling of a random subset of CMPRs and pipes during week 6. Counts were measured using quantitative flow cytometry (BDAccuri C6; BD Biosciences, San Jose, CA) following staining with SYBR Green I fluorescent nucleic acid stain using previously developed methods [20, 21].

After 10 weeks, one pipe volume (~400 mL) was collected in a sterile, 1 L, polypropylene bottle following three inversions of the pipe for mixing, and then filter concentrated onto 0.22-μm mixed nitrocellulose ester membranes (Millipore, Billerica, MA). PVC caps on the ends of the pipes were removed using pipe cutters and swabbed and the entire inner surfaces (~13 cm²) of these endcaps were swabbed with a sterile, cotton-tip applicator (Puritan, Guilford, ME) for biofilm sample collection in a single circular motion, making contact with the full length of the cotton portion of the applicator. The PVC cap surfaces were selected for swabbing to ensure a consistent recovery of biofilm-associated microbes across conditions, as corrosion/deposition phenomena on the metal surfaces were anticipated to interfere with biofilm recovery. DNA was extracted from fragmented filters or swab tips using a FastDNA Spin Kit (MP Biomedicals, Solon, OH) following manufacturer’s protocols. Filter, swab, lab (autoclaved DI water exposed to laboratory conditions), and extraction blanks were included with sample processing. Quantitative polymerase chain reaction (qPCR) targeting the bacterial 16S rRNA gene to quantify total bacteria DNA was performed as described previously [22, 23]. Briefly, qPCR reactions were carried out in triplicate 10 μL reactions containing 1x SsoFast Evagreen Supermix (Bio-Rad, Hercules, CA), 400 nM of forward and reverse primers, and UV sterilized, molecular grade water with 1 μL of DNA template. Samples were diluted 1:10 to minimize inhibition and a standard curve was generated for each run using 10-fold serial dilutions of custom gBlock (Integrated DNA Technologies, Coralville, IA) gene fragments with a quantification limit of 1000 gene copies per reaction. Samples were considered quantifiable if at least 2 of 3 triplicates were above the limit of quantification.

16S rRNA amplicon sequencing

Sample preparation for 16S rRNA amplicon sequencing was performed following the Earth Microbiome Project protocol for amplification of the V4-V5 region of the 16S rRNA gene using the 515F/926R primer pair. Combined triplicate PCR products were pooled to 240 ng each and purified using the QIAquick PCR purification kit (Qiagen, Valencia, CA). Sequencing was performed on the Illumina MiSeq platform by the Biocomplexity Institute at Virginia Tech (300-bp, paired end reads). Sequencing reads were processed using the QIIME2 pipeline (v. 2019.1) [24]. Sequences were quality filtered and dereplicated using the DADA2 pipeline with forward reads truncated at 297 bp and reverse reads truncated at 200 bp [25]. Resulting amplicon sequence variants (ASVs) were maintained for downstream analyses. Remaining sequences were taxonomically classified using the Scikit-learn classifier [26] and a pre-trained Silva 16S rRNA (release 132) database [27] for 99% similarity (515F/926R region, seven-level taxonomy). Singleton ASVs were removed and taxonomy-based filtering to exclude ASVs identified as either mitochondria or chloroplast was performed. A total of 4,631,690 sequencing reads were maintained across the 137 samples with a mean of 33,807 reads, a minimum of 1,139 reads, and a maximum of 55,749 reads/sample. Sample reads were rarefied to 10,177 randomly-selected reads using the phyloseq package (v.1.28.0 [28]), which excluded 11 samples below the threshold read count (lab blank, extraction blank, a filter blank, and 8 samples). Influent water samples, seeded influent water samples, along with filter, swab, and extraction blanks were included in the sequencing lane. Sequence data have been deposited in the National Center for Biotechnology Information (NCBI) Sequence Read Archive (SRA) under BioProject ID PRJNA609991.

Statistical analyses

Statistical analyses of chemical and biological parameters were performed in JMP Pro 14 using the Wilcoxon test for comparing two groups or the Kruskal-Wallis test for comparing more than two groups, followed by the Dunn test for multiple comparisons. Statistical testing for microbial community data was performed in R Studio (v.1.2.1335) using R version 3.6.0. Alpha diversity in microbial community data was carried out with the phyloseq package (v.1.28.0 [28]) using the pairwise wilcoxon test (pairwise.wilcox.test) using a Benjamini-Hochberg procedure. Beta diversity was analyzed by applying the permutational multivariate analysis of variance (Adonis) [29] found in the vegan package (v.2.5.5) [30] to the Bray-Curtis dissimilarity matrix generated using the phyloseq package. Multivariate homogeneity of group dispersions analysis (betadisper) [31] was also applied to check the homogeneity of dispersion among groups assumption for Adonis. Non-metric multi-dimensional scaling (NMDS) using the Bray-Curtis Dissimilarity matrix was generated using the phyloseq package to ordinate data for comparison. Differential abundance of ASVs was determined utilizing DESeq2 using a Wald test for significance (p<0.01) [32].

Results and discussion

Trends and variance among the CMPRs

Physicochemical consistency among the CMPRs

Temperature. was highly consistent across CMPRs, as indicated by a coefficient of variation (CV) for each pipe type of <2% for both weeks 2 and 9 (Table 1). The experimental design successfully achieved ideal OP growth temperatures across all pipe types, with 37.6 ± 0.7°C during week 2 and 39.6 ± 0.6°C during week 9 (Fig 2D). Temperature is considered to be an overarching factor impacting OP growth and proliferation in premise plumbing systems [3, 33, 34] ().

Table 1. Water quality parameters for replicate CMPRs during week 2 (acclimation, once weekly water change), week 9 (differentiation, twice weekly water change) or week 10 (qPCR data).

	Week 2			Week 9/10
	Copper	Iron	PVC	Copper	Iron	PVC
*Physicochemical Parameters*
(% Coefficient of Variation, 100 * Standard Deviation / Mean)
Temperature (°C)	1.71%	1.64%	1.92%	1.29%	1.51%	1.72%
pH	2.15%	2.96%	0.88%	2.30%	2.80%	0.04%
DO (mg/L)	2.60%	11.60%	12.00%	6.36%	14.70%	3.85%
TOC (mg/L)	5.26%	7.54%	28.50%	11.70%	30.20%	8.86%
Total Cu (mg/L)	32.20%	-	-	30.40%	-	-
Soluble Cu (mg/L)	21.50%	-	-	35.40%	-	-
Total Fe (mg/L)	-	18.80%	-	-	14.90%	-
Soluble Fe (μg/L)	-	361%	-	-	232%	-
*Biological Parameters*
Total Cell Counts (events/μL)	231%	43.80%	17.80%	25.40%	26.40%	18.20%
Bulk Water 16S rRNA (log [gc/mL])	-	-	-	2.75%	6.33%	1.58%
Biofilm 16S rRNA (log[gc/cm²])	-	-	-	7.02%	6.33%	11.60%

Open in a new tab

Fig 2 — Comparison of (A) DO, (B) TOC, (C) pH, and (D) temperature among CMPRs (random subset of n = 6 for DO, n = 18 for all others) and stagnant incubator room pipes (n = 3) after 9 weeks of aging. Matching letters indicate groupings based on Dunn’s test results (p < 0.05). Statistical groupings are independent for each panel.

pH was consistent with known trends for each type of pipe and reproducible. The targeted influent pH of 7.50 ± 0.05 was achieved. In week 2, PVC-copper pipes had a final pH of 7.66 ± 0.16 and PVC-iron pipes 7.72 ± 0.23, with a lower pH of 6.85 ± 0.06 in the PVC pipes (n_each = 18, p<0.0001 compared to each PVC-copper and PVC-iron CMPRs), presumably due to increased CO₂ production via cellular growth and respiration [35, 36] in PVC CMPRs and corrosion in metallic CMPRs [37, 38]. An increase in bulk water pH from the service line to outlets has been previously observed in a residential building featuring copper plumbing [39]. The CV remained low in all pipes through week 9 (Table 1), when the difference in pH as a function of pipe type decreased, with average pH of 7.45 ± 0.17 in PVC-copper, 7.27 ± 0.20 in PVC-iron, and 7.28 ± 0.04 in PVC (Fig 2C). PVC-copper pipes had a slightly higher pH on average than both PVC (n_each = 18, p = 0.0115) and PVC-iron (n_each = 18, p = 0.0006). pH is important, as it can strongly shape microbial community composition [40] and also influence key aspects of water quality affecting OP proliferation, especially disinfectant efficacy [41–43] and rate of release of pipe corrosion products, which can either be bactericidal and/or serve as micronutrients [21].

DO. also shifted during stagnation as a function of pipe material. DO in the PVC-iron pipes (1.96 ± 0.29 mg/L) was 34% that of PVC (5.71 ± 0.22 mg/L, n_each = 6, p = 0.0331) and 33% that of PVC-copper (5.90 ± 0.38 mg/L, n_each = 6, p = 0.0043) pipes in week 9 samples (Fig 2A). This is expected due to corrosion of iron pipes consuming oxygen over time [44–46]. DO tended to be lower in the PVC pipes than in the PVC-copper pipes during week 2 (n_copper = 9, n_PVC = 6, p = 0.0018), although there was no difference between PVC-copper and PVC pipes following 9 weeks of acclimation. This is likely due to the aging of PVC-copper pipes, resulting in increased cell growth due to the decreased release of antimicrobial copper as the pipes aged [19, 47], which in turn allowed more cellular respiration to consume DO, as was observed earlier in the experiment for the PVC pipes.

TOC. in the influent averaged 0.81 ± 0.12 ppm and increased following incubation in the CMPRs. Week 2 effluent TOC was fairly high; averaging 6.99 ± 0.37 mg/L in PVC-copper, 4.03 ± 0.30 mg/L in PVC-iron, and 3.68 ± 1.05 mg/L in PVC pipes. After 9 weeks, the TOC decreased and stabilized at 2.04 ± 0.24 mg/L in PVC-copper, 1.16 ± 0.35 mg/L in PVC-iron, and 1.21 ± 0.11 mg/L in PVC pipes (Fig 2B). This decrease from week 2 to week 9 was likely due to washout of TOC leaching from new materials. The PVC-copper condition in particular required extra epoxy to join the PVC and copper pipe junctions. The largest variability in TOC was observed in the PVC-iron pipes during week 9, which also had the highest variability in DO with a CV of 14.7%; however, the standard deviation was low in magnitude (0.35 mg/L). This indicates that pipes aged and formed scale with time as occurs in real-world plumbing.

Total copper and iron. concentrations in the respective CMPRs also decreased with pipe aging (Table 1). Consistent with typical levels of copper in new pipes with time [48], average total copper levels were initially 0.66 ± 0.21 mg/L after 2 weeks of aging, but decreased to 0.35 ± 0.10 mg/L by week 9. Variability in copper release was fairly high, as is characteristic in new pipes [49, 50], with CVs of 32.2% during week 2 and 30.4% during week 9. This could be due to differential aging rates in individual pipes or differential release of pipe scale during the sampling procedure. Average soluble copper was stable at 0.23 ± 0.05 mg/L at week 2 and 0.29 ± 0.10 mg/L at week 9. Average total iron decreased from 24.9 ± 4.68 mg/L at week 2 to 10.9 ± 1.62 mg/L at week 9, consistent with scale formation [51]. The variability in iron release was expected based on previous studies [46, 52, 53]. Between weeks 2 and 9, the CV decreased from 18.8% to 14.9% across CMPRs, consistent with iron release becoming more uniform as pipe-scale formed. The average soluble iron was 110 μg/L at week 2 and 90 μg/L at week 9, indicating that the vast majority of iron in the bulk water was in particulate form (>99%).

Microbiological characteristics of the CMPRs

Microbiological profiling indicated somewhat greater variability among replicate CMPRs than the physicochemical parameters, particularly for the metal pipe materials (Table 1). PVC was the least variable pipe type with regard to total cell counts, for both week 2 (CV = 17.8%) and week 9 (CV = 18.6%). Total cell counts were initially widely variable for PVC-copper pipes (CV = 231%), possibly due to different aging rates and toxicity impacts of variable copper release rates (CV = 32.2%). More variability was also observed in PVC-iron pipes relative to PVC (CV = 43.8%), which could also be due to differential rates of initial aging in the pipes given that pipes were new at the outset of the experiment (soluble iron CV = 361%). However, by week 9 the variability in total cell counts in PVC-copper CMPRs decreased and became similar to the variability displayed by PVC-iron pipes, despite soluble iron remaining variable (CV = 232%). We hypothesize that the greater initial variability in PVC-copper pipes was due to cupric ions being released at different rates, resulting in different rates of microbial inactivation. In PVC-iron pipes, high DO consumption by corrosion may explain the initial variability of total cell counts (S2 Fig).

Total bacteria measured by qPCR. targeting the 16S rRNA gene in the bulk water during week 10 of the experiment (Fig 3A) was less variable than total cell count metrics measured during week 9 (CV = 1.58–6.33%) (Table 1). The only pipe type with a range of variability more than one order of magnitude were PVC-iron CMPRs, with a range of 1.24 log(gc/mL). This could be related to the heterogeneous water chemistry produced by the iron condition due to insoluble iron oxide formation.

Biofilm total bacterial density, as estimated by 16S rRNA gene copy numbers, followed a similar trend as that in the bulk water, except there was somewhat more variance in the PVC-copper and PVC conditions (7.02% and 11.6% respectively). Similar levels of 16S rRNA genes were observed across the pipe materials in the biofilm at week 10 (Fig 3B; Table 1).

Effect of pipe type on bacterial growth with convective mixing

Total cell counts were measured on a weekly basis to track expected changes in microbial numbers with pipe age (Fig 4A) [18, 54–56]. During week 1, pipe type had a pronounced effect, with total cell counts in PVC-copper CMPRs being lower than in PVC and PVC-iron CMPRs (n_each = 18, p<0.0001 and p = 0.0015, respectively) and PVC being higher than PVC-iron (n_each = 18, p = 0.0027) CMPRs. The same trend was observed during week 2. This trend was consistent with typical biocidal activity of new copper pipes [19]. Iron also removes DO from the water through redox reactions, leaving less for use by cells for respiration and growth [44, 46]. This trend occurred until week 9, when no difference was observed between PVC and PVC-copper total cell counts; although they still had higher total cell counts than PVC-iron CMPRs (n_each = 18, p<0.0001 for both). This indicates that the copper pipes had aged to a point where the biocidal activity had decreased [18], potentially due to the formation of a more insoluble scale on the pipe wall. PVC-iron pipes; however, were found to still have ~1/3 of the DO of the other two conditions, potentially leading to lower cell growth due to nutrient limitations. There were no differences found among pipe types when comparing biofilm or bulk water qPCR measurements of 16S rRNA genes, which is likely due to inherent differences in the targets of the methods (DNA versus particles) [23].

16S rRNA gene amplicon sequencing. showed limited variability in microbial community composition among replicate CMPRs according to NMDS analysis. Blank samples were low in DNA content, as would be expected, with only one out of four blanks (one of the filter blanks) remaining in the analysis pool after rarefying to 10,177 reads across samples. This blank was taxonomically similar to samples (R² < 0.05) and was excluded under the assumption that it was cross-contaminated (S3 and S4 Figs). Statistically, there was a difference in beta diversity between the influent water samples and bulk water samples from all CMPRs (P_adonis = 0.001 for both weighted and unweighted analysis); however, the magnitude of the difference was very small (R² = 0.093; R² = 0.068, unweighted and weighted, respectively). Differences were also noted between the bulk water and biofilm communities, although the magnitude was low (P_adonis = 0.001, R² = 0.098, P_betadis = 0.011). This difference is consistent with a previous study comparing bulk water and biofilm taxonomic composition in a continuous flow premise-plumbing rig comparing different pipe materials [17].

Pipe type strongly influenced the composition of the microbial community that established in both the bulk water and biofilm, as has been observed in drinking water distribution systems [57], as well as other premise plumbing models such as pilot-scale rigs [58], simulated water heaters [59], and CDC biofilm reactors [60]. For CMPR bulk water samples, pipe type was found to be a major driver of microbial community composition (P_adonis = 0.001, R² = 0.355, P_betadis = 0.257) (S5A Fig). The microbial communities’ shift in response to pipe material may explain why total cell count variation decreased in PVC-copper and PVC-iron pipes between weeks 2 and 9, while the putative drivers of total cell count variation in these pipes remained more consistent. Very few taxa were found to be significantly enriched in metal CMPR bulk waters, with eight unique ASVs enriched in PVC bulk water. Two ASVs of the order Sphingobacteriales and one of the order Solibacterales, which have been found to be elevated in sediment samples [61], were enriched in PVC-iron. One ASV of the order Lactobacillales was enriched in PVC-copper CMPRs. The tendency for ASVs to be enriched more in the PVC bulk water may have to do with either nutrients leaching from PVC, the release of copper in PVC-copper CMPRs, or the reduced DO in PVC-iron CMPRs. The same relationship with pipe type was found for CMPR biofilm samples (P_adonis = 0.001, R² = 0.309, P_betadis = 0.001); but there was more heterogeneity of variance in the measurements (S5B Fig). Conversely, 30 unique ASVs were found to be enriched in the biofilm of PVC-iron CMPRs, while none were enriched in PVC-copper CMPRs and only one ASV of the order Sphingomonadales was enriched in the PVC CMPR’s biofilm. This may be due to the adherence of bacteria to the particulate iron on the pipe wall of PVC-iron CMPRs creating a niche environment for bacteria [62]. There was also no apparent difference in how PVC and PVC-copper biofilm samples clustered according to NMDS analysis, with the PVC-iron biofilm samples displaying the greatest variance (S5B Fig). This heterogeneity may be due to the age of the biofilm being only 10 weeks, while it is thought that at least one year is required to develop a truly mature biofilm on a copper pipe surface [63–65]. The greater variability in PVC-iron CMPRs may also be due to the influence of the sediments observed in the effluent of those pipes on the microbial ecology [62].

Alpha diversity based on the Shannon index was found to be lowest in the bulk water of PVC-copper CMPRs (1.60±0.37) and highest in PVC CMPRs (2.44±0.48), with PVC-iron CMPRs measuring in between (2.03±0.32) (n_each = 18, PVC-copper vs PVC-iron, p = 0.0018; PVC-copper vs PVC, p<0.0001; PVC-iron vs PVC, p = 0.0018). This is likely due to a less diverse set of microbes being able to survive in the presence of the copper ions, as well as possible nutrients being released by the PVC pipe. The decrease in DO in the PVC-iron pipes may also contribute to lower diversity of microbes being able to thrive, leading to some taxa dominating over others. This trend was not observed in the biofilm, with no difference between PVC-copper and PVC pipes and the Shannon index being slightly higher in the PVC-iron pipes than in PVC-copper and PVC (n_each = 18, p = 0.0253, p = 0.0027, respectively). The Shannon index was 1.43 ± 0.60 in PVC, 2.23 ± 0.55 in PVC-iron, and 1.73 ± 0.44 in PVC-copper CMPRs. This may be due to all samples being swabbed from the PVC portion of the biofilm, or the relatively low age of the biofilm itself.

CMPRs versus stagnant pipes

Physicochemical differences

Following 9 weeks of aging, DO was consistently lower in the CMPRs than stagnant pipes, with PVC-iron pipes having the most pronounced difference (Fig 2A). The DO in PVC-iron CMPRs was 33% that of the PVC-iron stagnant pipes (1.96 mg/L vs 6.03 mg/L, respectively) (n_CMPR = 6, n_stagnant = 3, p = 0.020), yet there was no measurable difference between total iron levels in PVC-iron CMPRs (10.9 mg/L) compared to stagnant (10.8 mg/L) pipes. The DO in the stagnant PVC-iron pipes was comparable to that of the other two stagnant pipe conditions, indicating that the loss of DO to iron oxidation was reduced by stagnation. This suggests that the convective mixing current subjected the mild steel pipe section to more corrosive conditions. Convective mixing likely increased the interactions of the bulk water with suspended solids and biofilm, increasing mass transfer and reactions with corrosion byproducts that consume DO. Since water in the stagnant pipes does not actively mix, there are mass transfer limitations to biofilm and corrosion reactions that consume DO. The effect of lower DO in the CMPRs is desired in this case, as DO depletion is a common phenomenon in “real-world” water systems where galvanized iron pipes have corroded following the loss of the protective zinc layer [46].

There was no difference in TOC levels between CMPRs and corresponding stagnant pipes (Fig 2B). Following 9 weeks of aging, the pH was slightly higher on average in CMPRs than in stagnant pipes (7.34 vs. 7.06, n_CMPR = 18, n_stagnant = 3, p<0.0001) (Fig 2C). In terms of temperature, CMPRs inherently experienced a gradient between water bath (45–48°C) and ambient (19–20°C) temperatures, with an average of the mixture measured at 39.6°C. The average measured temperature for the stagnant pipes was 35.2°C, within the target optimal growth range for OPs (Fig 2D).

Differences in microbial numbers

In the stagnant pipes, total cell counts exhibited a different pattern from that of the CMPRs and there was no difference among the pipe materials, except PVC-copper had slightly higher total cell counts than in PVC-iron stagnant pipes starting at week 6 (n_each = 3, p = 0.034) (Fig 4). This was the opposite of the trend displayed by the CMPRs, where PVC-copper pipe total cell counts were initially highly variable and much lower relative to PVC or PVC-iron pipes (n_each = 18, p<0.0001, p = 0.0027, respectively). Interestingly, at week 2, although cell counts were ~2.5 logs lower in PVC-copper CMPRs than in PVC-copper stagnant pipes, there was no difference in concentration of total copper (0.66 mg/L vs. 0.46 mg/L) or soluble copper (0.23 mg/L vs. 0.33 mg/L). This was presumably due to convective mixing enhancing the interaction between cells and copper.

Differences in total cell counts between stagnant and CMPR PVC-iron conditions are likely explained by the differences in DO available to the bacteria (2.17 mg/L versus 6.03 mg/L at week 2, respectively). By week 9 the total cell counts were approximately equal in PVC-iron CMPRs and stagnant pipes, perhaps due to the water changes being increased to twice weekly (Fig 4).

PVC pipes had higher total cell counts in CMPRs than in corresponding stagnant pipes from week 5 onwards (Fig 4). This could be the result of recirculation of nutrients that is enhanced by convective mixing. Total bacterial 16S rRNA gene copy numbers were also higher in PVC CMPRs than in the stagnant PVC pipes ((6.36 log[gc/mL] vs. 5.78 log[gc/mL], respectively) (n_CMPR = 18, n_stagnant = 3, p = 0.007) during week 10 (Fig 3A). There were no differences in total cell counts for either PVC-copper or PVC-iron pipes in CMPRs relative to stagnant pipes at week 9, indicating diminishing pipe aging effects in the metal pipe CMPRs with time.

There was 4 times the density of bacterial 16S rRNA gene copies in stagnant PVC-copper pipe biofilm than in PVC-copper CMPR biofilm (n_CMPR = 18, n_stagnant = 3, p = 0.035), potentially due to less delivery of copper to the biofilm under stagnant conditions (Fig 3B). This difference was not found when comparing PVC or PVC-iron CMPRs and stagnant pipes.

The trend of higher total cell counts in PVC pipes, along with the trends observed in the metal pipes, is indicative of the impact that convective mixing has on the bulk water reactions in the interior of the pipe. Without internal convective mixing, the redox reactions and biocidal activity of the metals appear to have little to no effect on bacterial growth rates within the pipes of this particular water chemistry, although redox reactions may still have substantial effects under other circumstances. It also appears that more nutrients are also made available in the CMPRs than in the stagnant pipes. Overall, the results demonstrate that convective mixing can play an important role in bacteria-pipe-water interactions.

Comparison of microbial communities in stagnant and CMPR Pipes

There was an overarching difference in the microbial community composition of the bulk water of the CMPRs versus stagnant pipes based on the Bray-Curtis dissimilarity matrix (P_adonis = 0.001, R² = 0.222, P_betadis = 0.785), as well as the biofilm between the CMPRs and the stagnant pipes (P_adonis = 0.001, R² = 0.119, P_betadis = 0.827) (Fig 5). There was also a difference in the microbial community composition when considering the unweighted Jaccard similarity matrix in the bulk water (P_adonis = 0.001, R² = 0.148, P_betadis = 0.966) and biofilm (P_adonis = 0.001, R² = 0.088, P_betadis = 0.695). Comparing PVC-copper pipes, there were 84 unique ASVs enriched in the bulk water of stagnant PVC-copper pipes compared to only one enriched in PVC-copper CMPRs. This suggests stronger biocidal effects in copper CMPRs than stagnant pipes. For PVC-iron pipes, 5 ASVs were enriched in stagnant pipes and 21 in CMPRs. For PVC pipes, 12 ASVs were enriched in stagnant pipes and 15 in CMPRs. Overall, there were 58 enriched ASVs in the biofilm of stagnant pipes compared to CMPRs, whereas only 9 were enriched in the biofilm of CMPRs. Since biofilm was collected from the bottom of the PVC portion of each pipe, it is possible that the bacteria were more likely to settle in the stagnant pipes than the CMPRs, allowing more adherence to the pipe wall at that location.

Fig 5 — Nonmetric multidimensional scaling (NMDS) plot, generated from Bray-Curtis dissimilarity matrix using the phyloseq package in R for 16S rRNA gene amplicon sequences, comparing CMPR and stagnant pipe (A) bulk water and (B) biofilm taxonomic microbial community composition at the end of week 10. Bulk water and biofilm NMDS plots were generated independently. There is a difference in microbial community when comparing both the bulk water P_adonis = 0.001, R2 = 0.222, P_betadis = 0.785) and biofilm (P_adonis = 0.001, R2 = 0.119, P_betadis = 0.827) of CMPRs vs stagnant pipes.

Alpha diversity in the stagnant pipes was not found to be affected by pipe type for either the bulk water or the biofilm. Overall, the Shannon index was 2.92 ± 0.30 for bulk water samples and 2.79 ± 1.10 for the biofilm samples. This is in contrast to the trend measured in CMPRs noted above, where alpha diversity was lowest in PVC-copper pipes, indicating again that convective mixing enhanced the effects of metal pipes.

CMPRs versus other premise plumbing simulation methods

CMPRs were designed to deliver nutrients (e.g., oxygen, organic carbon) to biofilms in the low-nutrient environment of drinking water while maintaining ideal temperatures for OP growth throughout the pipe volume with controlled replication. In this study, the CMPRs achieved a relevant premise plumbing water chemistry conditions while using more representative pipe materials than the simulated glass water heaters (SGWHs) or CDC biofilm reactors, and allowing for more precise control of conditions than pilot-scale premise plumbing systems (Table 2).

Table 2. Comparison of the CMPR to alternative reactors for simulating premise plumbing based upon available pricing and other information for required equipment components and personal knowledge and experience of the authors.

Y = Yes, N = No, D = difficult/expensive.

	CDC Biofilm Reactor	Simulated Glass Water Heaters (SGWHs)	Pilot-Scale Premise Plumbing Pipe Rigs	Convective Mixing Pipe Reactors (CMPRs)
Precise control of influent water conditions?	D	Y	N	Y
Ease of operation under premise plumbing conditions (simulates extremes in water quality/flow changes)?	N	Y	Y	Y
Realistic flow conditions?	N	N	Y	N
Standardized protocol and methods?	Y	Y	N	Y
Multiple conditions simultaneously?	D	Y	D	Y
Multiple pipe materials simultaneously?	D	Y	D	Y
Replicate reactors practical?	N	Y	N	Y
Ease of replication?	Y	Y	N	Y
Economical to build?	N	Y	N	Y
Off the shelf components only?	N	Y	Y	Y
Mostly representative materials?	N	N	Y	Y
High pipe surface area/volume ratio?	N	N	Y	Y
Low manual labor?	D	N	N	N
Small footprint?	D	Y	N	Y
Low cost to operate/maintain?	N	Y	N	Y
Amenable to biosafety protections?	D	Y	N	Y
Cost, USD (S3 Table)	$2,480 /reactor	$3 /bottle	~$4,000 /rig	~$1,200 /rig
Representative Cost, USD per replicate reactor	$2,480	$3	$450^*	$22

Open in a new tab

*Assuming each distal line from recirculation rig is a separate reactor, otherwise $4,000

While costs of any system will vary with time and location, the test apparatus described herein, including the housing unit, cost only ~$1,200 to accommodate 54 parallel operated CMPRs (S3 Table). Pilot-scale premise plumbing rigs constructed in prior studies were estimated to cost approximately ~$4,000 and accommodate fewer replicates (e.g., 36 pipes) and thus had more inherent variability because each distal line is not a perfect replicate [10]. CMPRs are also much less costly than the CDC biofilm reactor, with a cost of $2,480 USD per reactor with stir plate and temperature controls, not including a peristaltic pump needed for each reactor (Biosurface Technologies, Bozeman, MT). This makes testing multiple pipe conditions in parallel expensive and often cost prohibitive, as an individual reactor with full controls will be needed for each pipe type tested, while also introducing a greater chance of mechanical failure of at least one component. Moreover, when a failure of a component inevitably occurs in a CMPR, true replication is maintained because it will affect every pipe equally. Although SGWHs are the most economical option (~$3 USD/reactor), the vast majority of the available surface is glass, unless steps are taken to equip with extended pipe surface area. SGWHs are also largely stagnant, unless placed on a shaker table, but the extent to which RPMs correspond to real-world premise plumbing conditions has not been determined.

The ultimate basis of comparison, however, should be the ability of these systems to replicate real-world premise plumbing conditions. Here, we have shown that the CMPRs replicate phenomena observed in some important real-world settings. Unfortunately, it is not possible to compare our water quality findings to those of other full-scale systems because every water has a unique chemistry and every sampling location in a building is unique in terms of water use patterns, temperature and disinfectant time profiles. Such a comparison would be useful but it would be a major undertaking in terms of personnel and resource requirements, but would be highly valuable because it is important to obtain a better understanding of the real-world strengths and limitations of each laboratory simulation method. A particular strength of CMPRs is an ability to be modified to accommodate a wide range of replicates, pipe conditions, temperature regimes, and adjustments to influent water chemistry. We judge that CMPRs are a useful, novel and cost-effective test method that can help to better understand a wide variety of microbiological and physicochemical premise plumbing research issues.

Conclusions

Herein we created and tested a novel reactor design that takes advantage of a natural convective mixing phenomenon to facilitate replicated testing of realistic premise plumbing conditions for purposes such as evaluating conditions that support pathogen growth or disinfection. Several key aspects of the CMPRs were validated:

CMPRs can maintain temperatures ideal for the growth of OPs.
Little variability in physical and biological parameters were observed across replicate CMPRs beyond that which is inherent to new pipe materials, allowing for relatively homogenous testing conditions.
Without convective mixing, there was little to no effect of redox reactions or biocidal activity normally expected when there are interactions with metal pipes, demonstrating that such mixing can be an important factor for bacterial growth and pipe interactions when present.
CMPRs are a cost-effective system that can simulate a substantial range of premise plumbing conditions.
CMPRs were able to induce and isolate certain effects of key test variables, e.g., pipe material effect on water chemistry and microbiology, that have been observed as trends in other premise plumbing simulations and full-scale buildings. These include chemical phenomena such as the observed differences in pH and DO, the variability and gradual reduction in copper and iron release, as well as biological differences such as total cells counts and microbial community composition in the bulk water and biofilm. This enables future testing to explore impacts of a wider range of chemistries and experimental conditions while rigorously testing for statistical confidence between true replicate reactors.

Overall, the CMPR provides a simple, replicable reactor capable of simulating some important naturally occurring phenomenon representative of premise plumbing.

Supporting information

S1 Fig. Temperature profile of clear PVC pipe filled with RO water after allowed to equilibrate with the room and incubator temperatures to undergo convective mixing at different angles.

Incubator consisted of a box using a light bulb as a heat source. Temperatures were measured using an infrared thermometer temperature gun at various points along the pipe.

(DOCX)

Click here for additional data file.^{(18.5KB, docx)}

S2 Fig. Comparison of total cell counts, DO, and total iron in CMPR effluent in weeks 2 and 9.

No significant correlation was found between total cell counts and either DO or total iron (Spearman rank correlation).

(DOCX)

Click here for additional data file.^{(226.3KB, docx)}

S3 Fig. Phylum level taxonomic microbial community profiles of each CMPR and each stagnant pipe organized by pipe type and location.

Naming convention: the first letter is a nominal indicator representing a set of 9 pipes representing a row of pipes in the CMPR housing unit, the second letter indicates pipe type (C = copper, F = PVC-iron, P = PVC), and the number represents the replicate number within the set of 9 pipes. All samples were collected during Week 10 with the exception of influent and seeded influent samples (Week 0).

(DOCX)

Click here for additional data file.^{(352.2KB, docx)}

S4 Fig. Nonmetric multidimensional scaling (NMDS) plot, generated from Bray-Curtis dissimilarity matrix using the phyloseq package in R for 16S rRNA gene amplicon sequences rarefied to 1139 randomly selected reads to maintain all blanks and samples to compare taxonomic microbial community composition between blanks and samples.

A single blank (filter blank) clusters near the majority of samples. This is assumed to be the result of cross-contamination during the preparation process given that all other blanks cluster separately. The microbial community compositions of blanks were found to be different than that of both bulk water and biofilm samples (P_adonis < 0.01 for both unweighted and weighted metrics).

(DOCX)

Click here for additional data file.^{(77.3KB, docx)}

S5 Fig

Nonmetric multidimensional scaling (NMDS) plot, generated from Bray-Curtis dissimilarity matrix using the phyloseq package in R for 16S rRNA gene amplicon sequences, comparing CMPR (A) bulk water and (B) biofilm taxonomic microbial community composition. Bulk water and biofilm NMDS plots were generated independently. Pipe type was an important factor influencing the microbial community in both the bulk water P_adonis = 0.001, R² = 0.355, P_betadis = 0.257) and biofilm (P_adonis = 0.001, R² = 0.309, P_betadis = 0.001).

(DOCX)

Click here for additional data file.^{(65.4KB, docx)}

S1 Table. Test data determining the maximum mixing velocity based on pipe angle using neutrally buoyant rhodamine dye.

(DOCX)

Click here for additional data file.^{(14.5KB, docx)}

S2 Table. Physicochemical characteristics of influent water.

(DOCX)

Click here for additional data file.^{(13.2KB, docx)}

S3 Table. Cost estimates for each type of premise plumbing simulation method.

(DOCX)

Click here for additional data file.^{(14.9KB, docx)}

S1 File. Manuscript data.

Data used for analysis in this manuscript.

(XLSX)

Click here for additional data file.^{(91.9KB, xlsx)}

S2 File

(DOCX)

Click here for additional data file.^{(2.1MB, docx)}

Acknowledgments

The authors acknowledge Advanced Research Computing at Virginia Tech for providing computational resources and technical support that have contributed to the results reported within this paper. The authors would like to thank Sophia Lee for their assistance with maintenance of the CMPRs and data collection.

Data Availability

Sequence data have been deposited in the NCBI Sequence Read Archive (BioProject ID PRJNA609991). Additional relevant data are within the manuscript and its Supporting Information files.

Funding Statement

This work was supported by the National Science Foundation (CBET award number 1706733, nsf.gov) and the Center for Disease Control and Prevention (contract number 75D30118C02905, cdc.gov), and a National Science Foundation Graduate Fellowship to Abraham Cullom. The funders had no role in study design, data collection and analysis, decision to publish, or preparation of the manuscript.

References

1.Collier SA, Stockman LJ, Hicks LA, Garrison LE, Zhou F, Beach MJ. Direct healthcare costs of selected diseases primarily or partially transmitted by water. Epidemiology & Infection. 2012;140(11):2003–13. [DOI] [PMC free article] [PubMed] [Google Scholar]
2.Falkinham JO III, Hilborn ED, Arduino MJ, Pruden A, Edwards MA. Epidemiology and ecology of opportunistic premise plumbing pathogens: Legionella pneumophila, Mycobacterium avium, and Pseudomonas aeruginosa. Environmental Health Perspectives. 2015;123(8):749–58. 10.1289/ehp.1408692 [DOI] [PMC free article] [PubMed] [Google Scholar]
3.Falkinham JO, Pruden A, Edwards M. Opportunistic premise plumbing pathogens: increasingly important pathogens in drinking water. Pathogens. 2015;4(2):373–86. 10.3390/pathogens4020373 [DOI] [PMC free article] [PubMed] [Google Scholar]
4.Pruden A, Edwards MA, Falkinham JO. State of the science and research needs for opportunistic pathogens in premise plumbing: Water Research Foundation; 2013. [Google Scholar]
5.Donlan R, Piede J, Heyes C, Sanii L, Murga R, Edmonds P, et al. Model system for growing and quantifying Streptococcus pneumoniae biofilms in situ and in real time. Appl Environ Microbiol. 2004;70(8):4980–8. 10.1128/AEM.70.8.4980-4988.2004 [DOI] [PMC free article] [PubMed] [Google Scholar]
6.Goeres DM, Loetterle LR, Hamilton MA, Murga R, Kirby DW, Donlan RM. Statistical assessment of a laboratory method for growing biofilms. Microbiology. 2005;151(3):757–62. [DOI] [PubMed] [Google Scholar]
7.Camper AK, Jones WL, Hayes JT. Effect of growth conditions and substratum composition on the persistence of coliforms in mixed-population biofilms. Appl Environ Microbiol. 1996;62(11):4014–8. 10.1128/AEM.62.11.4014-4018.1996 [DOI] [PMC free article] [PubMed] [Google Scholar]
8.Camper AK, Brastrup K, Sandvig A, Clement J, Spencer C, Capuzzi A. Effect of distribution system materials on bacterial regrowth. Journal‐American Water Works Association. 2003;95(7):107–21. [Google Scholar]
9.Douterelo I, Sharpe R, Boxall J. Influence of hydraulic regimes on bacterial community structure and composition in an experimental drinking water distribution system. Water Research. 2013;47(2):503–16. 10.1016/j.watres.2012.09.053 [DOI] [PubMed] [Google Scholar]
10.Rhoads WJ, Ji P, Pruden A, Edwards MA. Water heater temperature set point and water use patterns influence Legionella pneumophila and associated microorganisms at the tap. Microbiome. 2015;3(1):67. [DOI] [PMC free article] [PubMed] [Google Scholar]
11.Leon Ohl A, Horn H, Hempel D. Behaviour of biofilm systems under varying hydrodynamic conditions. Water Science and Technology. 2004;49(11–12):345–51. [PubMed] [Google Scholar]
12.Liu Z, Lin Y, Stout J, Hwang C, Vidic R, Yu V. Effect of flow regimes on the presence of Legionella within the biofilm of a model plumbing system. J Appl Microbiol. 2006;101(2):437–42. 10.1111/j.1365-2672.2006.02970.x [DOI] [PubMed] [Google Scholar]
13.National Research Council. Drinking water distribution systems: assessing and reducing risks : National Academies Press; 2007. [Google Scholar]
14.Brazos B, O'Connor J, Abcouwer S, editors. Kinetics of chlorine depletion and microbial growth in household plumbing systems. Proceedings AWWA 1985 Water Quality Technology Conference, Houston; 1985.
15.Nguyen C, Elfland C, Edwards M. Impact of advanced water conservation features and new copper pipe on rapid chloramine decay and microbial regrowth. Water Research. 2012;46(3):611–21. 10.1016/j.watres.2011.11.006 [DOI] [PubMed] [Google Scholar]
16.Rhoads WJ, Pruden A, Edwards MA. Convective mixing in distal pipes exacerbates Legionella pneumophila growth in hot water plumbing. Pathogens. 2016;5(1):29. [DOI] [PMC free article] [PubMed] [Google Scholar]
17.Ji P, Rhoads WJ, Edwards MA, Pruden A. Impact of water heater temperature setting and water use frequency on the building plumbing microbiome. The ISME Journal. 2017;11(6):1318–30. 10.1038/ismej.2017.14 [DOI] [PMC free article] [PubMed] [Google Scholar]
18.Lehtola MJ, Miettinen IT, Keinänen MM, Kekki TK, Laine O, Hirvonen A, et al. Microbiology, chemistry and biofilm development in a pilot drinking water distribution system with copper and plastic pipes. Water Research. 2004;38(17):3769–79. 10.1016/j.watres.2004.06.024 [DOI] [PubMed] [Google Scholar]
19.Van der Kooij D, Veenendaal HR, Scheffer WJ. Biofilm formation and multiplication of Legionella in a model warm water system with pipes of copper, stainless steel and cross-linked polyethylene. Water Research. 2005;39(13):2789–98. 10.1016/j.watres.2005.04.075 [DOI] [PubMed] [Google Scholar]
20.Hammes F, Berney M, Wang Y, Vital M, Köster O, Egli T. Flow-cytometric total bacterial cell counts as a descriptive microbiological parameter for drinking water treatment processes. Water Research. 2008;42(1–2):269–77. 10.1016/j.watres.2007.07.009 [DOI] [PubMed] [Google Scholar]
21.Rhoads WJ, Pruden A, Edwards MA. Interactive effects of corrosion, copper, and chloramines on Legionella and mycobacteria in hot water plumbing. Environ Sci Technol. 2017;51(12):7065–75. 10.1021/acs.est.6b05616 [DOI] [PubMed] [Google Scholar]
22.Suzuki MT, Taylor LT, DeLong EF. Quantitative analysis of small-subunit rRNA genes in mixed microbial populations via 5′-nuclease assays. Appl Environ Microbiol. 2000;66(11):4605–14. 10.1128/aem.66.11.4605-4614.2000 [DOI] [PMC free article] [PubMed] [Google Scholar]
23.Wang H, Bedard E, Prevost M, Camper AK, Hill VR, Pruden A. Methodological approaches for monitoring opportunistic pathogens in premise plumbing: A review. Water Research. 2017;117:68–86. 10.1016/j.watres.2017.03.046 [DOI] [PMC free article] [PubMed] [Google Scholar]
24.Bolyen E, Rideout JR, Dillon MR, Bokulich NA, Abnet CC, Al-Ghalith GA, et al. Reproducible, interactive, scalable and extensible microbiome data science using QIIME 2. Nature Biotechnology. 2019;37(8):852–7. 10.1038/s41587-019-0209-9 [DOI] [PMC free article] [PubMed] [Google Scholar]
25.Callahan BJ, McMurdie PJ, Rosen MJ, Han AW, Johnson AJA, Holmes SP. DADA2: high-resolution sample inference from Illumina amplicon data. Nature Methods. 2016;13(7):581 10.1038/nmeth.3869 [DOI] [PMC free article] [PubMed] [Google Scholar]
26.Pedregosa F, Varoquaux G, Gramfort A, Michel V, Thirion B, Grisel O, et al. Scikit-learn: Machine learning in Python. The Journal of Machine Learning Research. 2011;12:2825–30. [Google Scholar]
27.Quast C, Pruesse E, Yilmaz P, Gerken J, Schweer T, Yarza P, et al. The SILVA ribosomal RNA gene database project: improved data processing and web-based tools. Nucleic Acids Research. 2012;41(D1):D590–D6. [DOI] [PMC free article] [PubMed] [Google Scholar]
28.McMurdie PJ, Holmes S. phyloseq: an R package for reproducible interactive analysis and graphics of microbiome census data. PLOS One. 2013;8(4). [DOI] [PMC free article] [PubMed] [Google Scholar]
29.Anderson MJ. A new method for non‐parametric multivariate analysis of variance. Austral Ecology. 2001;26(1):32–46. [Google Scholar]
30.Oksanen J, Kindt R, Legendre P, O’Hara B, Stevens MHH, Oksanen MJ, et al. The vegan package. Community Ecology Package. 2007;10:631–7. [Google Scholar]
31.Anderson MJ. Distance‐based tests for homogeneity of multivariate dispersions. Biometrics. 2006;62(1):245–53. 10.1111/j.1541-0420.2005.00440.x [DOI] [PubMed] [Google Scholar]
32.Love MI, Huber W, Anders S. Moderated estimation of fold change and dispersion for RNA-seq data with DESeq2. Genome Biology. 2014;15(12):550 10.1186/s13059-014-0550-8 [DOI] [PMC free article] [PubMed] [Google Scholar]
33.National Academies of Sciences, Engineering, and Medicine. Management of Legionella in water systems: National Academies Press; 2020. [PubMed] [Google Scholar]
34.Yee RB, Wadowsky RM. Multiplication of Legionella pneumophila in unsterilized tap water. Appl Environ Microbiol. 1982;43(6):1330–4. 10.1128/AEM.43.6.1330-1334.1982 [DOI] [PMC free article] [PubMed] [Google Scholar]
35.Wurts WA, Durborow RM. Interactions of pH, carbon dioxide, alkalinity and hardness in fish ponds. 1992. [Google Scholar]
36.Feely RA, Alin SR, Newton J, Sabine CL, Warner M, Devol A, et al. The combined effects of ocean acidification, mixing, and respiration on pH and carbonate saturation in an urbanized estuary. Estuarine, Coastal and Shelf Science. 2010;88(4):442–9. 10.1016/j.ecss.2010.05.004. [DOI] [Google Scholar]
37.Stumm W, Morgan JJ. Aquatic chemistry: chemical equilibria and rates in natural waters: John Wiley & Sons; 2012. [Google Scholar]
38.Deberry DW, Kidwell JR, Malish D. Corrosion in Potable Water Systems–Final Report. Washington, DC: United States Environmental Protection Agency, Office of Drinking Water; 1982. [Google Scholar]
39.Salehi M, Odimayomi T, Ra K, Ley C, Julien R, Nejadhashemi AP, et al. An investigation of spatial and temporal drinking water quality variation in green residential plumbing. Building and Environment. 2020;169:106566 10.1016/j.buildenv.2019.106566. [DOI] [Google Scholar]
40.Pinto AJ, Xi C, Raskin L. Bacterial community structure in the drinking water microbiome is governed by filtration processes. Environ Sci Technol. 2012;46(16):8851–9. 10.1021/es302042t [DOI] [PubMed] [Google Scholar]
41.Butterfield C, Wattie E, Megregian S, Chambers C. Influence of pH and temperature on the survival of coliforms and enteric pathogens when exposed to free chlorine. Public Health Reports (1896–1970). 1943:1837–66. [PubMed] [Google Scholar]
42.Butterfield C, Wattie E. Influence of pH and temperature on the survival of coliforms and enteric pathogens when exposed to chloramine. Public Health Reports (1896–1970). 1946:157–92. [PubMed] [Google Scholar]
43.Yu-sen EL, Vidic RD, Stout JE, Victor LY. Negative effect of high pH on biocidal efficacy of copper and silver ions in controlling Legionella pneumophila. Appl Environ Microbiol. 2002;68(6):2711–5. 10.1128/aem.68.6.2711-2715.2002 [DOI] [PMC free article] [PubMed] [Google Scholar]
44.Zhang Y, Griffin A, Edwards M. Effect of nitrification on corrosion of galvanized iron, copper, and concrete. Journal AWWA. 2010;102(4):83–93. 10.1002/j.1551-8833.2010.tb10093.x [DOI] [Google Scholar]
45.McNeill LS, Edwards M. IRON PIPE corrosion IN DISTRIBUTION SYSTEMS. Journal AWWA. 2001;93(7):88–100. 10.1002/j.1551-8833.2001.tb09246.x [DOI] [Google Scholar]
46.Sarin P, Snoeyink V, Bebee J, Jim K, Beckett M, Kriven WM, et al. Iron release from corroded iron pipes in drinking water distribution systems: effect of dissolved oxygen. Water Research. 2004;38(5):1259–69. 10.1016/j.watres.2003.11.022 [DOI] [PubMed] [Google Scholar]
47.Lytle DA, Liggett J. Impact of water quality on chlorine demand of corroding copper. Water Research. 2016;92:11–21. 10.1016/j.watres.2016.01.032 [DOI] [PubMed] [Google Scholar]
48.Edwards M, Powers K, Hidmi L, Schock M. The role of pipe ageing in copper corrosion by-product release. Water Science and Technology: Water Supply. 2001;1(3):25–32. [Google Scholar]
49.Masters S, Parks J, Atassi A, Edwards MA. Inherent variability in lead and copper collected during standardized sampling. Environ Monit Assess. 2016;188(3):177 10.1007/s10661-016-5182-x [DOI] [PubMed] [Google Scholar]
50.Rushing JC, Edwards M. The role of temperature gradients in residential copper pipe corrosion. Corrosion Science. 2004;46(8):1883–94. 10.1016/j.corsci.2003.11.001. [DOI] [Google Scholar]
51.Tang Z, Hong S, Xiao W, Taylor J. Characteristics of iron corrosion scales established under blending of ground, surface, and saline waters and their impacts on iron release in the pipe distribution system. Corrosion Science. 2006;48(2):322–42. [Google Scholar]
52.Lytle DA, Sarin P, Snoeyink VL. The effect of chloride and orthophosphate on the release of iron from a cast iron pipe section. Journal of Water Supply: Research and Technology-Aqua. 2005;54(5):267–81. 10.2166/aqua.2005.0026 [DOI] [Google Scholar]
53.McNeill Laurie S, Edwards M. Phosphate Inhibitors and Red Water in Stagnant Iron Pipes. Journal of Environmental Engineering. 2000;126(12):1096–102. 10.1061/(ASCE)0733-9372(2000)126:12(1096) [DOI] [Google Scholar]
54.Silhan J, Corfitzen CB, Albrechtsen HJ. Effect of temperature and pipe material on biofilm formation and survival of Escherichia coli in used drinking water pipes: a laboratory-based study. Water Science and Technology. 2006;54(3):49–56. 10.2166/wst.2006.447 [DOI] [PubMed] [Google Scholar]
55.Schwartz T, Hoffmann S, Obst U. Formation and bacterial composition of young, natural biofilms obtained from public bank-filtered drinking water systems. Water Research. 1998;32(9):2787–97. 10.1016/S0043-1354(98)00026-8. [DOI] [Google Scholar]
56.Kerr C, Osborn K, Robson G, Handley P. The relationship between pipe material and biofilm formation in a laboratory model system. J Appl Microbiol. 1998;85(S1):29S–38S. [DOI] [PubMed] [Google Scholar]
57.Lin W, Yu Z, Chen X, Liu R, Zhang H. Molecular characterization of natural biofilms from household taps with different materials: PVC, stainless steel, and cast iron in drinking water distribution system. Applied Microbiology and Biotechnology. 2013;97(18):8393–401. 10.1007/s00253-012-4557-3 [DOI] [PubMed] [Google Scholar]
58.Ji P, Parks J, Edwards MA, Pruden A. Impact of Water Chemistry, Pipe Material and Stagnation on the Building Plumbing Microbiome. PloS one. 2015;10(10):e0141087–e. 10.1371/journal.pone.0141087 . [DOI] [PMC free article] [PubMed] [Google Scholar]
59.Proctor CR, Dai D, Edwards MA, Pruden A. Interactive effects of temperature, organic carbon, and pipe material on microbiota composition and Legionella pneumophila in hot water plumbing systems. Microbiome. 2017;5(1):130 10.1186/s40168-017-0348-5 [DOI] [PMC free article] [PubMed] [Google Scholar]
60.Lu J, Buse HY, Gomez-Alvarez V, Struewing I, Santo Domingo J, Ashbolt NJ. Impact of drinking water conditions and copper materials on downstream biofilm microbial communities and Legionella pneumophila colonization. J Appl Microbiol. 2014;117(3):905–18. 10.1111/jam.12578 [DOI] [PubMed] [Google Scholar]
61.Qin K, Struewing I, Domingo JS, Lytle D, Lu J. Opportunistic pathogens and microbial communities and their associations with sediment physical parameters in drinking water storage tank sediments. Pathogens. 2017;6(4):54. [DOI] [PMC free article] [PubMed] [Google Scholar]
62.Liu G, Bakker G, Li S, Vreeburg J, Verberk J, Medema G, et al. Pyrosequencing reveals bacterial communities in unchlorinated drinking water distribution system: an integral study of bulk water, suspended solids, loose deposits, and pipe wall biofilm. Environ Sci Technol. 2014;48(10):5467–76. 10.1021/es5009467 [DOI] [PubMed] [Google Scholar]
63.Boe-Hansen R, Albrechtsen H-J, Arvin E, Jørgensen C. Dynamics of biofilm formation in a model drinking water distribution system. Journal of Water Supply: Research and Technology—AQUA. 2002;51(7):399–406. [Google Scholar]
64.Buse HY, Lu J, Struewing IT, Ashbolt NJ. Preferential colonization and release of Legionella pneumophila from mature drinking water biofilms grown on copper versus unplasticized polyvinylchloride coupons. International Journal of Hygiene and Environmental Health. 2014;217(2–3):219–25. 10.1016/j.ijheh.2013.04.005 [DOI] [PubMed] [Google Scholar]
65.Martiny AC, Jørgensen TM, Albrechtsen H-J, Arvin E, Molin S. Long-term succession of structure and diversity of a biofilm formed in a model drinking water distribution system. Appl Environ Microbiol. 2003;69(11):6899–907. 10.1128/aem.69.11.6899-6907.2003 [DOI] [PMC free article] [PubMed] [Google Scholar]

PLoS One. doi: 10.1371/journal.pone.0238385.r001

Decision Letter 0

Zhi Zhou

12 Jun 2020

PONE-D-20-13252

Replicable simulation of distal hot water premise plumbing using convectively-mixed pipe reactors

PLOS ONE

Dear Dr. Edwards,

Thank you for submitting your manuscript to PLOS ONE. After careful consideration, we feel that it has merit but does not fully meet PLOS ONE’s publication criteria as it currently stands. Therefore, we invite you to submit a revised version of the manuscript that addresses the points raised during the review process.

It's an interesting study to use the new reactor to study plumbing systems. As mentioned by the reviewer, more experimental details can be provided and experimental conditions can be better justified. Showing how the new CMPRs represent or be similar to real plumbing systems would greatly strengthen the paper. Please address other detailed comments from the reviewer as well.

Please submit your revised manuscript by Jul 27 2020 11:59PM. If you will need more time than this to complete your revisions, please reply to this message or contact the journal office at plosone@plos.org. When you're ready to submit your revision, log on to https://www.editorialmanager.com/pone/ and select the 'Submissions Needing Revision' folder to locate your manuscript file.

Please include the following items when submitting your revised manuscript:

A rebuttal letter that responds to each point raised by the academic editor and reviewer(s). You should upload this letter as a separate file labeled 'Response to Reviewers'.
A marked-up copy of your manuscript that highlights changes made to the original version. You should upload this as a separate file labeled 'Revised Manuscript with Track Changes'.
An unmarked version of your revised paper without tracked changes. You should upload this as a separate file labeled 'Manuscript'.

If you would like to make changes to your financial disclosure, please include your updated statement in your cover letter. Guidelines for resubmitting your figure files are available below the reviewer comments at the end of this letter.

If applicable, we recommend that you deposit your laboratory protocols in protocols.io to enhance the reproducibility of your results. Protocols.io assigns your protocol its own identifier (DOI) so that it can be cited independently in the future. For instructions see: http://journals.plos.org/plosone/s/submission-guidelines#loc-laboratory-protocols

We look forward to receiving your revised manuscript.

Kind regards,

Zhi Zhou, Ph.D.

Academic Editor

PLOS ONE

Journal Requirements:

When submitting your revision, we need you to address these additional requirements.

1. Please ensure that your manuscript meets PLOS ONE's style requirements, including those for file naming. The PLOS ONE style templates can be found at https://journals.plos.org/plosone/s/file?id=wjVg/PLOSOne_formatting_sample_main_body.pdf and https://journals.plos.org/plosone/s/file?id=ba62/PLOSOne_formatting_sample_title_authors_affiliations.pdf

Additional Editor Comments (if provided):

[Note: HTML markup is below. Please do not edit.]

Reviewers' comments:

Reviewer's Responses to Questions

Comments to the Author

1. Is the manuscript technically sound, and do the data support the conclusions?

The manuscript must describe a technically sound piece of scientific research with data that supports the conclusions. Experiments must have been conducted rigorously, with appropriate controls, replication, and sample sizes. The conclusions must be drawn appropriately based on the data presented.

Reviewer #1: Partly

**********

2. Has the statistical analysis been performed appropriately and rigorously?

Reviewer #1: Yes

**********

3. Have the authors made all data underlying the findings in their manuscript fully available?

The PLOS Data policy requires authors to make all data underlying the findings described in their manuscript fully available without restriction, with rare exception (please refer to the Data Availability Statement in the manuscript PDF file). The data should be provided as part of the manuscript or its supporting information, or deposited to a public repository. For example, in addition to summary statistics, the data points behind means, medians and variance measures should be available. If there are restrictions on publicly sharing data—e.g. participant privacy or use of data from a third party—those must be specified.

Reviewer #1: Yes

**********

4. Is the manuscript presented in an intelligible fashion and written in standard English?

PLOS ONE does not copyedit accepted manuscripts, so the language in submitted articles must be clear, correct, and unambiguous. Any typographical or grammatical errors should be corrected at revision, so please note any specific errors here.

Reviewer #1: Yes

**********

5. Review Comments to the Author

Please use the space provided to explain your answers to the questions above. You may also include additional comments for the author, including concerns about dual publication, research ethics, or publication ethics. (Please upload your review as an attachment if it exceeds 20,000 characters)

Reviewer #1: This paper describes a new approach to simulating domestic plumbing systems that are influenced by convective mixing. The introduction was interesting and informative. Some parts of the methods section were well done, but other parts were lacking. Specifically, some important experimental details were included in the results section rather than in the methods and some of the reasons for some of the experimental design decisions were not explained (e.g. Why were some parameters measured weekly and others measured only in week 2 and week 9? Why was the water changeover schedule changed from once a week to twice a week partway through the experiment?) The results section is comprehensive but somewhat disorganized. For the most part, the results supported the conclusions at the end of the paper, but it could be better organized and written to emphasize how each set of experiments informed the final conclusions.

General comments:

My main problem with this work is that the authors failed to put their results into context or to prove that their CMPRs were representative of full scale domestic plumbing systems, or comparable/superior to alternative pilot systems (e.g. pilot scale pipe rigs, model showers), or existing bench-scale reactors (e.g. CDC reactors). This made it impossible to evaluate whether their CMPRs were representative of real world conditions and/or more appropriate than alternative pilot-scale and bench-scale options. In theory, the best way to explore this would be by conducting parallel studies with the different simulation options and/or in real domestic hot water systems. This may not be feasible at this point, but the authors are part of a large, well established laboratory and, based on other papers published out of this group, should have access to data and expertise that could substantially improve this aspect of the paper.

The section devoted to comparing the CMPRs with other simulation options focused mostly on cost rather than water quality and biological impacts. I think it would be much stronger if they focused more on the latter, as described in the specific comments later in this review.

There seems to have been a large amount of variability between the individual CMPRs within each condition (i.e. each pipe type) for some parameters (Table 1). This contradicts one of the conclusions at the end of their paper. The coefficient of variation for temperature was low for all conditions, so presumably the variability was not related to temperature management. I don’t feel like the authors convincingly explained why replicate new pipes exposed to identical conditions would have such different results.

Finally, the writing was often convoluted with small but frequent grammatical errors. A careful review of the text should be sufficient to remedy this. The flow of the paper could be tightened by ensuring that the results section is organized such that the different sections, including the final section comparing the CMPRs to other simulation options, clearly support the final conclusions of the study.

Specific comments:

Abstract Is it standard to include n, p, and R2 values in abstracts in PLOS ONE? It seemed odd to me when I was reading through the paper, but if it is normal for this journal there is no need to change it.

Line 44 Emanating is a strange word choice here. Perhaps originating would be more appropriate.

Lines 55-56 Impeded from replication is an awkward way to say that it is difficult to include large numbers of replicates. I suggest that you reword this.

Lines 59-60 Would biological safety standards be difficult to maintain because the pilot equipment is likely to be too large for a standard laboratory? If a pilot system was small enough to fit in a laboratory, would it still be difficult to maintain biological safety standards?

Lines 82-85 This sentence is convoluted and too long. I suggest that you break it into two or three

smaller and more focused sentences.

Lines 111-112 The copper + PVC CMPRs referred to as copper pipes throughout the text but the iron + PVC pipes are referred to PVC-iron. I suggest that you refer to the copper + PVC pipes as PVC-copper throughout the paper to emphasize that, like the iron pipes, the copper pipes were connected to a PVC portion.

Line 115 Aren’t the different properties of the materials the point of the study?

Lines 115-116 Why did you include so many replicates?

Lines 119-120 I don’t think that once or twice weekly water changes are likely to replicate water usage patterns in real domestic plumbing systems where water is used many times per day. Also, why did you change from once a week to twice a week partway through your experiment?

Line 125 Could you include some basic information about the “base” (influent?) water that you used in your experiments (TOC, iron, etc.) in a table in the supplementary information? Also, you refer to this water as base water in some parts of the paper and influent water elsewhere. Choose one term and stick to it.

Line 128 No need to put the word seeded in quotes here or elsewhere.

Line 129 Is the bacterial community in GAC backwash water likely to be representative of that found in a domestic plumbing system? If yes, could you explain why?

Lines 143-144 Why were these parameters only measured in week 2 and week 9? Why not measure them more frequently?

Line 169 By circular motion do you mean that you swabbed the entire inner circumference of the pipe or that you took a swab from one part of the pipe (e.g. the bottom surface)?

Line 194 Please describe the different blanks mentioned here.

Lines 219-220 You could remove this sentence.

Line 231 What are the expected pH trends for each type of pipe? Please describe briefly and provide citation(s). If these expected trends were observed in full scale domestic water heating systems it might be worth mentioning this in the final section of the results to support the idea that your CMPRs are comparable to real world systems.

Lines 280 Why would individual iron pipes age at different rates if they are all new and all subjected to the same conditions?

Lines 283-286 Break this sentence into two sentences, one about copper and one about iron.

Lines 284-285 Why would different individual copper pipes release different amounts of cupric ions if they were all new and all subjected to the same temperature conditions and water changeout regimes? This potentially points to a serious weakness in your approach or the execution of the study.

Lines 312-314 Add a citation to support this hypothesis.

Lines 314-315 Have other studies found that iron pipes have lower DO than copper or PVC pipes?

Lines 322-324 This should also be mentioned in the methods section.

Lines 324-325 The fact that new pipes were used does not indicate that pipe aging occurred, but the fact that you started with new pipes and your cell counts changed over time does suggest that pipe aging occurred. I suggest that you reword this sentence to clarify.

Lines 325 Have other studies shown that pipe aging impacts cell counts? If yes, cite them here (you cited refs 37 and 38 to support a similar statement elsewhere in the paper). Depending on what kind of apparatus was used in these other studies, this could also be discussed in the final part of the results section.

Line 324 Why did you change your water changeover regime in week 2?

Line 330 Remove the word inherently.

Line 334 When you say “bulk water from the CMPRs”, do you mean that all of the data from all of the conditions (PVC, iron-PVC, copper) were grouped together and compared to the influent water? If yes, I would mention this in the text.

Lines 341-350 Has this relationship between microbial community and pipe material been observed in real world domestic hot water systems or in studies conducted with other lab scale apparatus? If yes, this would be an interesting thing to bring up in the final section of the results.

Line 344 Remove the word however.

Line 361 Did you actually observe sediments in your pipes?

Line 372 Why were all of the swabs taken from the PVC section of the pipes?

Lines 382-385 I don’t think that your DO findings, on their own, support all of this. I suggest that you tone down your language and bit and break this sentence into two sentences. The first sentence could begin with “this suggests that” rather than “thus” to emphasize that you are, to some degree at least, speculating about the causes underlying your DO results.

Lines 386-389 This is an interesting point. Perhaps you could move it to the final section of the results to better support your assertions that the CMPRs can simulate real world conditions.

Line 418 Here and elsewhere in the paper you should write times, rather than x.

Lines 459-482 This section is, in my opinion, the heart of the paper, but it contains a number of unsupported or poorly supported claims by the authors and as a result it is ultimately unconvincing. For example, reference 10 was used to support the statement that pilot-scale plumbing rigs had higher inherent variability than your CMPRs, but when I quickly reviewed that paper, I saw no mention of high variability.

In general, this section would be more effective if you focused on comparing your actual water quality findings (e.g. DO, cell counts, etc.) to the results of studies that examined full scale domestic hot water systems or that used CDC reactors, pilot pipe rigs, etc.. This would show that your CMPRs were comparable or superior to existing methods. There is some of this in various parts of the results section (lines 338-340, lines 386-389, etc.) and this content could be moved to this final section to

better support your claims.

The cost information is interesting but ultimately costs will differ from one jurisdiction to another and over time, so I wouldn’t make it the main focus here.

Also, assuming that stagnant pipe reactors like those used in this study are widely used to simulate domestic hot water systems, they should be included in Table 2 and discussed in this section.

Line 460 You didn’t measure nitrogen or discuss it in the rest of the paper, so I suggest that you omit it from this list.

Line 461 This is the first time that your base / influent water is described as oligotrophic. This term may not be familiar to your readers, so I think it would be more effective to briefly describe the important characteristics of your water matrix (low TOC, moderate pH, dechlorinated, etc.)

Line 476 Are mechanical failures truly unavoidable with the CDC reactors? Couldn’t the UV light, recirculation pump, heater, etc. in the CMPR set-up fail as well?

Line 489 Some of your results, especially from week 2, do not support this conclusion.

Figure 1 What does the checked box represent? Also, it might be nice to include a schematic showing the stagnant pipe apparatus for comparison.

Figure 2 I must admit to initially being totally confused by the letters on these plots because the way that you’ve indicated statistical differences is unfamiliar to me and, in my opinion, counterintuitive. I’m used to statistical differences being indicated by different letters, rather than by the same letters. So, for example, if condition 1 and condition 2 were statistically different, they would be labeled A and B, but if they were indistinguishable, they’d both be labeled A because they would be in the same group. This is in line with the outputs of all of the statistical programs that I’ve ever worked with. I think that the approach that I’ve described here is likely to be more familiar and intuitive to readers, so I suggest that you adopt it here.

Also, why do some plots only have capital letters and some only have lower case letters? Weren’t the same statistical tests run on all four datasets?

Figure 4 It would be interesting to see DO, cell counts, and iron plotted together (or on three faceted plots in a single figure) to emphasize how they influence one another.

Table 2 This table is interesting but the authors don’t support their assumptions with any citations and it isn’t always clear how they came to their conclusions about different factors. For example, why wouldn’t it be possible create a standardized protocol for a pilot-scale pipe rig? Why would the CMPRs cost less to operate than a CDC reactor, especially seeing as the CMPR apparatus includes a UV lamp and heater, both of which would consume electricity? Why would it be more difficult or expensive to ensure biosafety protections for CDC biofilm reactors? Also, the authors should specify that their costing is USD.

**********

6. PLOS authors have the option to publish the peer review history of their article (what does this mean?). If published, this will include your full peer review and any attached files.

If you choose “no”, your identity will remain anonymous but your review may still be made public.

Do you want your identity to be public for this peer review? For information about this choice, including consent withdrawal, please see our Privacy Policy.

Reviewer #1: No

[NOTE: If reviewer comments were submitted as an attachment file, they will be attached to this email and accessible via the submission site. Please log into your account, locate the manuscript record, and check for the action link "View Attachments". If this link does not appear, there are no attachment files.]

While revising your submission, please upload your figure files to the Preflight Analysis and Conversion Engine (PACE) digital diagnostic tool, https://pacev2.apexcovantage.com/. PACE helps ensure that figures meet PLOS requirements. To use PACE, you must first register as a user. Registration is free. Then, login and navigate to the UPLOAD tab, where you will find detailed instructions on how to use the tool. If you encounter any issues or have any questions when using PACE, please email PLOS at figures@plos.org. Please note that Supporting Information files do not need this step.

PLoS One. 2020 Sep 16;15(9):e0238385. doi: 10.1371/journal.pone.0238385.r002

Author response to Decision Letter 0

27 Jul 2020

Our responses are also uploaded in a file with better formating to view each comment. We also produce it below.

We thank the editor and reviewer for their time and effort in evaluating our manuscript and offering suggestions for improvement. Attached are the original copy of the manuscript and a revised version using tracked changes. Responses to comments are provided below with the line numbers referring to the track changes version of the manuscript. Portions of quotations are bolded to emphasize particular changes to the editors and do not reflect bold formatting in the manuscript.

Reviewer 1:

General comments:

1-1 My main problem with this work is that the authors failed to put their results into context or to prove that their CMPRs were representative of full scale domestic plumbing systems, or comparable/superior to alternative pilot systems (e.g. pilot scale pipe rigs, model showers), or existing bench-scale reactors (e.g. CDC reactors). This made it impossible to evaluate whether their CMPRs were representative of real world conditions and/or more appropriate than alternative pilot-scale and bench-scale options. In theory, the best way to explore this would be by conducting parallel studies with the different simulation options and/or in real domestic hot water systems. This may not be feasible at this point, but the authors are part of a large, well established laboratory and, based on other papers published out of this group, should have access to data and expertise that could substantially improve this aspect of the paper.

Response: We agree with the reviewer that such testing would be ideal, but that it is not feasible to do in a meaningful way because every feed tap water is inherently different and the experiments with different designs would have to be done in parallel. This is a main message of the manuscript, that no existing system reasonably allows for such testing in parallel at the scale of the CMPR. We note that such testing has also not previously been reported for any of the alternative laboratory simulation highlighted in the manuscript.

In response to this comment, we point out some of the obvious challenges that would have to be overcome for such testing in lines 754-760.

Additionally, we followed the advice of the reviewer to the extent possible with new text in lines 393-395, 439-440, 444-446, 500-501, 505-506, and 558-561, with the purpose of better contextualizing our work with comparisons to previous studies. Overall, changes have been made to emphasize that our water quality findings correspond reasonably well to those of other simulation methods or full-scale buildings, but that it would be valuable to have a large-scale study to try and compare strengths and weaknesses of every lab simulation method to portions of actual premise plumbing in buildings.

Context has also been provided for the inherent variability of iron and copper release, which are a key concern for this reviewer. Briefly, new metallic pipes are expected to be variable in metal release rates. See lines 439-440 and 444-446 and responses to Comments 1-23 and 1-25 below for additional details.

1-2 The section devoted to comparing the CMPRs with other simulation options focused mostly on cost rather than water quality and biological impacts. I think it would be much stronger if they focused more on the latter, as described in the specific comments later in this review.

Response: See response to Comment 1-1 above and Comment 1-41 below. The reality is that cost and scale are major limitations to realistic premise-plumbing simulations. This study proposes and evaluates a new design, while also providing context to existing alternatives.

1-3 There seems to have been a large amount of variability between the individual CMPRs within each condition (i.e. each pipe type) for some parameters (Table 1). This contradicts one of the conclusions at the end of their paper. The coefficient of variation for temperature was low for all conditions, so presumably the variability was not related to temperature management. I don’t feel like the authors convincingly explained why replicate new pipes exposed to identical conditions would have such different results.

Response: This reviewer concern has been addressed in detail below in response to Comment 1-23 and Comment 1-25. Additionally, we discuss the expectation that there is inherent variability in premise plumbing studies in response to Comment 1-12.

1-4 Finally, the writing was often convoluted with small but frequent grammatical errors. A careful review of the text should be sufficient to remedy this. The flow of the paper could be tightened by ensuring that the results section is organized such that the different sections, including the final section comparing the CMPRs to other simulation options, clearly support the final conclusions of the study.

Response: All co-authors have re-read and revised the manuscript to address grammatical errors and improve clarity. We have worked to better present the results, as suggested, and to ensure that the results support the conclusions, as seen in our responses to Comments 1-22, 1-23, 1-25, 1-27, 1-28, 1-29, 1-30, 1-34, 1-38, and 1-39. Additionally, we have added a new co-author, Abraham Cullom, who was previously an acknowledgement for assisting in testing the CMPRs, but now aided in independently re-analyzing the data and revising the manuscript with fresh perspective. Abraham is also applying the CMPR design in his current research towards answering some of the questions raised in the current study.

Specific comments:

1-5 Abstract Is it standard to include n, p, and R2 values in abstracts in PLOS ONE? It seemed odd to me when I was reading through the paper, but if it is normal for this journal there is no need to change it.

Response: We have evaluated other PLOS ONE papers and have determined that it is common to report statistical details in the abstract. Thus, we have maintained the statistical details in the abstract.

1-6 Line 44 Emanating is a strange word choice here. Perhaps originating would be more appropriate.

Response: ‘Emanating’ has been replaced with ‘originating’ in line 55.

1-7 Lines 55-56 Impeded from replication is an awkward way to say that it is difficult to include large numbers of replicates. I suggest that you reword this.

Response: The phase ‘are impeded from replication by size and cost’ has been replaced by ‘hinder replication through their large size and cost’ in lines 66-67.

1-8 Lines 59-60 Would biological safety standards be difficult to maintain because the pilot equipment is likely to be too large for a standard laboratory? If a pilot system was small enough to fit in a laboratory, would it still be difficult to maintain biological safety standards?

Response: A primary difficulty associated with pilot-scale systems of any size is that they require direct connection to the actual premise plumbing of a public building and associated institutional permissions and oversight for automated operation. By design, pilot-scale systems are continuous-flow and produce large volumes of water that must be assumed to be contaminated with pathogens and thus appropriately disinfected and disposed of. Typically, this requires inline disinfection and hard plumbing to the drain. Note that most of the relevant pathogens of interest to premise plumbing can be spread via inhalation of bioaerosols, creating a potential health risk during sampling. Because a pilot-scale reactor cannot fit within the confines of a biological safety level (BSL)-2-approved cabinet, sampling must be carried out in the open. To safely conduct sampling, a BSL-2 plan must be developed and approved by the institution, typically with the laboratory evacuated and the workers wearing N-95 masks.

Lines 70-82 have been amended to clarify the biological safety concerns of pilot-scale systems. In particular, CMPRs can be sampled safely within the confines of a BSL-2- approved cabinet.

‘Pilot-scale studies examining OPs also typically require direct connection to premise plumbing of the study facility and cannot be sampled within the protection of a biological safety-level (BSL) 2 certified cabinet, elevating potential for exposure of workers to pathogen-containing aerosols during sampling and thus requiring appropriate institutional approvals.’

1-9 Lines 82-85 This sentence is convoluted and too long. I suggest that you break it into two or three smaller and more focused sentences.

Response: This sentence has been divided into two in lines 129-132:

‘The CMPR consists of off-the-shelf materials used in real-world plumbing systems. Capped pipe segments, with one end submerged in a hot water bath and the other contacting the cooler ambient air, simulate the premise plumbing riser from a hot water recirculation loop that connects to distal, stagnant outlets (Fig 1A).’

1-10 Lines 111-112 The copper + PVC CMPRs referred to as copper pipes throughout the text but the iron + PVC pipes are referred to PVC-iron. I suggest that you refer to the copper + PVC pipes as PVC-copper throughout the paper to emphasize that, like the iron pipes, the copper pipes were connected to a PVC portion.

Response: References to ‘copper CMPRs/pipes’ have been changed to ‘PVC-copper CMPRs/pipes’ throughout the text.

1-11 Line 115 Aren’t the different properties of the materials the point of the study?

Response: Differences in heat conduction properties of these materials were not the focus of this particular study, although examining how pipe material’s effects on internal mixing velocities affect a multitude of factors could be an important focus of a future project. This study attempted to isolate the properties of the three pipe materials that would influence bulk water chemical and biological parameters while maintaining the same internal mixing velocity. The text has been modified in lines 215-218 for clarification:

‘These three materials exhibited distinct heat conduction properties and would in turn create distinct convective mixing patterns and velocities if submerged in the water bath. Thus, for purposes of this experiment, PVC segments were submerged in the water bath to normalize convective mixing across the conditions.’

1-12 Lines 115-116 Why did you include so many replicates?

Response: A major limitation of all prior premise plumbing OPs research, including our own, is the insurmountable costs of obtaining true replication. That limitation reduces statistical confidence and increases the likelihood of drawing false conclusions or yielding statistically insignificant results. A strength of the CMPR approach is that replicates can be added are relatively low cost and, frankly, we were very curious as to how reproducible reactors and chemical and biological conditions could be with a large number of replicates using an approach that minimized most sources of variability to the extent possible. We suspect that if all prior work (including our own) had required a minimum n = 5 (which is cost prohibitive for most current premise plumbing simulations) that a high inherent variability of prior systems would become much more apparent. Since this is the first test of the CMPR concept, we wanted to also do the first test using a very large number of replicates to explore this issue, in an approach in which most contributors to variability were minimized. The variability of CMPRs was therefore a key focus of this study.

This sentence has been prefaced with ‘to ascertain a high-resolution view of the variability of chemical and biological water quality parameters among CMPRs’ in lines 218-219.

1-13 Lines 119-120 I don’t think that once or twice weekly water changes are likely to replicate water usage patterns in real domestic plumbing systems where water is used many times per day.

Response: We agree that twice weekly water changes do not represent the use frequency of many outlets. However, as stated in lines 119-122 of the original manuscript, this system is operated in a manner to simulate the use of distal outlets, and we have now changed the wording to say “infrequently used” distal outlets. Future work could choose to operate the CMPRs with more frequent water changes, if desired. The text has been expanded to emphasize these points in lines 238-246:

‘The influent water to both CMPRs and stagnant pipes was initially seeded with backwash water from a granular activated carbon filter that had been in operation in a premise plumbing drinking water system for > 2 years to establish a mature microbial community. Water changes were conducted once weekly for the first 2 weeks of operation to facilitate colonization of the pipe surfaces. Water changes were then increased to twice weekly to better simulate an infrequent use pattern characteristic of distal outlets in large buildings, as a possible worst-case scenario for OPs control, until the end of the experiment (week 10). These periodic manual dump and fill water changes served to recreate turbulent intermittent flow and complete changeover of water, as occurs at infrequently used distal outlets.’

1-14 Lines 119-120, continued: Also, why did you change from once a week to twice a week partway through your experiment?

Response: The once weekly water changes during the first two weeks were selected to allow the microbes in the seed water to colonize the pipe walls and reduce potential for washout. Mention of the influent seeding has been moved to the beginning of this paragraph and the text has been changed in lines 240-244 in order to make this clearer:

‘Water changes were conducted once weekly for the first 2 weeks of operation to facilitate colonization of the pipe surfaces. Water changes were then increased to twice weekly to better simulate an infrequent use pattern characteristic of distal outlets in large buildings, as a possible worst-case scenario for OPs control, until the end of the experiment (week 10).’

1-15 Line 125 Could you include some basic information about the “base” (influent?) water that you used in your experiments (TOC, iron, etc.) in a table in the supplementary information? Also, you refer to this water as base water in some parts of the paper and influent water elsewhere. Choose one term and stick to it.

Response: Thank you. An additional table has been added to the SI (Table S. 2.) to provide this information. All references to ‘base water’ have been changed to ‘influent water.’

1-16 Line 128 No need to put the word seeded in quotes here or elsewhere.

Response: This change has been made throughout the text.

1-17 Line 129 Is the bacterial community in GAC backwash water likely to be representative of that found in a domestic plumbing system? If yes, could you explain why?

Response: Given that whole-house GAC filters are commonplace (i.e., 10% of homes in our experience), yes, we expect that the GAC backwash water with characteristically removed particles derived from the distribution system, is relevant to what enters a substantial fraction of domestic plumbing systems. Digging deeper into this comment, we would further expect that the microbial community composition would shift from the backwash water to the CMPR water, just as it would from a whole-house GAC unit to the residential plumbing. Still, the bulk of the microbes in premise plumbing would originate from the water supply, but various populations are enriched or diminished by the GAC filter and conditions thereafter.

Figure S. 4. Directly addresses this comment, with an NMDS plot directly comparing the microbial community composition of the seed, influent, CMPR, and stagnant pipe waters. Figure S. 4. Is referred to in line 539. The shifts in microbial community composition are consistent with the above expectations, as we observed a difference between the microbial community structures of the influent, which was from a tap water outlet separate from the GAC filter, and the CMPR pipe effluent, which, as mentioned above, had been seeded with GAC backwash weeks before, and this difference was small. We note these details in lines 539-542:

‘Statistically, there was a difference in beta diversity between the influent water samples and bulk water samples from all CMPRs (Padonis=0.001 for both weighted and unweighted analysis); however, the magnitude of the difference was very small (R2=0.093; R2=0.068, unweighted and weighted, respectively).’

We also emphasize that the GAC filter used here has been used in a pilot-scale premise-plumbing system and flushed with tap water three times daily for the 2+ years of its operation and thus would provide a mature seed of influent microbes. The text has been updated in lines 238-240 to demonstrate that it had been used in a premise plumbing drinking water system;

1-18 Lines 143-144 Why were these parameters only measured in week 2 and week 9? Why not measure them more frequently?

Response: The week 2 sampling was conducted to provide a profile of the CMPRs following their acclimation and during early stages of operation. Week 9 was selected as a representative time point when one would expect that the effects of pipe materials would be observable, based on prior studies referenced in line 279

While more time points may have been interesting, they were not necessary to achieve the primary goal of demonstrating the ability of the CMPRs to distinguish effects of pipe material on water chemistry in a repeatable fashion, and as is already noted, we had sufficient replicates (n=16) to draw high statistical confidence in a single sampling event.

The justification for sampling at week 2 and week 9 is now provided in lines 277-279:

‘These timepoints were selected to profile the pipes following initial microbial colonization and at a later period when differences in biological water quality parameters would be expected based on pipe material [18, 19].’

This is briefly mentioned again in the caption for Table 1, now reading:

‘Table 1. Water quality parameters for replicate CMPRs during week 2 (acclimation, once weekly water change), week 9 (differentiation, twice weekly water change) or week 10 (qPCR data)’

1-19 Line 169 By circular motion do you mean that you swabbed the entire inner circumference of the pipe or that you took a swab from one part of the pipe (e.g. the bottom surface)?

Response: This phrase has been expanded to ‘and the entire inner surfaces (~13 cm2) of these endcaps were swabbed’ to better explain our method in lines 300-301.

1-20 Line 194 Please describe the different blanks mentioned here.

Response: The included blanks are specified in lines 338: ‘…along with filter, swab, and extraction blanks…’. The list of blanks in line 307-308 has been updated to include the swab blank.

1-21 Lines 219-220 You could remove this sentence.

Response: We agree. The sentence has been removed.

1-22 Line 231 What are the expected pH trends for each type of pipe? Please describe briefly and provide citation(s). If these expected trends were observed in full scale domestic water heating systems it might be worth mentioning this in the final section of the results to support the idea that your CMPRs are comparable to real world systems.

Response: Corrosion on metallic pipes tends to raise pH, unless an insoluble metal hydroxide scale forms, in which case the pH will not change. Respiration is expected to lower the pH through the production of carbon dioxide, while the pH effect of other biological reactions will vary. Explanations of and references for these phenomena are now provided after the relevant results are discussed, in lines 379-395:

‘In week 2, PVC-copper pipes had a final pH of 7.66 ± 0.16 and PVC-iron pipes 7.72 ± 0.23, with a lower pH of 6.85 ± 0.06 in the PVC pipes (neach=18, p<0.0001 compared to each PVC-copper and PVC-iron CMPRs), presumably due to increased CO2 production via cellular growth and respiration [35, 36] in PVC CMPRs and corrosion in metallic CMPRs [37, 38]. An increase in bulk water pH from the service line to outlets has been previously observed in a residential building featuring copper plumbing [39].’

1-23 Lines 280 Why would individual iron pipes age at different rates if they are all new and all subjected to the same conditions?

Metal release from new iron and copper pipes has been shown to be highly variable, presumably due to small differences in the surface after manufacturing. We were funded by manufacturers to explore this issue and ended our work only showing that new copper pipes are often 50% variable from one another in terms of copper release. Unfortunately, we were never provided follow up funding to determine why. We are providing this unpublished report made to copper pipe manufacturers to the reviewer in supplemental documents.

This variability is almost never examined due to lack of replication in prior OPs studies (including our own), so we cannot cite statistics from prior research that proves this occurs. Supporting references are now provided in lines 439-440. Additionally, language was amended in lines 444-446 to both emphasize this expectation and highlight another instance in which the CMPRs replicated a known phenomenon in premise plumbing drinking water quality:

‘The variability in iron release was expected based on previous studies [46, 52, 53]. Between weeks 2 and 9, the CV decreased from 18.8% to 14.9% across CMPRs, consistent with iron release becoming more uniform as pipe-scale formed.’

1-24 Lines 283-286 Break this sentence into two sentences, one about copper and one about iron.

Response: This sentence was divided into two to better explain these two points in lines 471-474:

‘We hypothesize that the greater initial variability in PVC-copper pipes was due to cupric ions being released at different rates, resulting in different rates of microbial inactivation. In PVC-iron pipes, high DO consumption by corrosion may explain the initial variability of total cell counts (Fig S.2).’

1-25 Lines 284-285 Why would different individual copper pipes release different amounts of cupric ions if they were all new and all subjected to the same temperature conditions and water changeout regimes? This potentially points to a serious weakness in your approach or the execution of the study.

Response: This is the same question and response noted previously for iron. Copper pipes, even those from the same production batch, are highly variable in terms of copper release. It is speculated that this is due to variable oxide or carbon films on the pipe surface. Additional references for this point have been added in line 439-440:

‘Variability in copper release was fairly high, as is characteristic in new pipes [49, 50], with CVs of 32.2% during week 2 and 30.4% during week 9.’

This variability, along with that observed in iron release from iron pipes, is one reason we judge that higher replication is recommended for future premise plumbing simulation studies and partly motivates the development of the CMPR to efficiently accommodates this need. We have attached the industry funded report, Influence of Drawing Lubricant on Copper Release from Copper Pipes During NSF 61 Testing, for additional documentation of very high variability in copper release from copper pipes.

1-26 Lines 312-314 Add a citation to support this hypothesis.

Response: A reference has been added to line 509-510 (’… to a point where the biocidal activity had decreased [18]…’) which demonstrates copper pipes supporting similar cell growth to plastic pipes after sufficient aging.

1-27 Lines 314-315 Have other studies found that iron pipes have lower DO than copper or PVC pipes?

Response: References have been added to line 506 that demonstrate iron corrosion’s consumption of DO, as well as a direct comparison of DO to that in copper pipes:

‘Iron also removes DO from the water through redox reactions, leaving less for use by cells for respiration and growth [44, 46].’

1-28 Lines 322-324 This should also be mentioned in the methods section.

Response: More details are offered on the sampling time points for total cell count in lines 291-292:

‘Bulk water total cell counts were taken during weeks 1, 2, 5, and 9, with an additional sampling of a random subset of CMPRs and pipes during week 6.’

1-29 Lines 324-325 The fact that new pipes were used does not indicate that pipe aging occurred, but the fact that you started with new pipes and your cell counts changed over time does suggest that pipe aging occurred. I suggest that you reword this sentence to clarify.

Response: This language has been changed to ‘the observed change in cell counts over the course of the experiment is likely due to aging of new pipes’ in lines 531-532.

1-30 Lines 325 Have other studies shown that pipe aging impacts cell counts? If yes, cite them here (you cited refs 37 and 38 to support a similar statement elsewhere in the paper). Depending on what kind of apparatus was used in these other studies, this could also be discussed in the final part of the results section.

Response: References demonstrating changes in total cell counts and HPCs with pipe age have been added to line 500-501, as the line referred to above is a figure caption:

‘Total cell counts were measured on a weekly basis to track expected changes in microbial numbers with pipe age (Fig 4A) [18, 54-56].’

1-31 Line 324 Why did you change your water changeover regime in week 2?

Response: This is addressed in our response to the comment for lines 119-120.

1-32 Line 330 Remove the word inherently.

Response: This word was removed.

1-33 Line 334 When you say “bulk water from the CMPRs”, do you mean that all of the data from all of the conditions (PVC, iron-PVC, copper) were grouped together and compared to the influent water? If yes, I would mention this in the text.

Response: This phrase has been expanded to ‘between the influent water samples and bulk water samples from all CMPRs’ for added clarity in line 540.

1-34 Lines 341-350 Has this relationship between microbial community and pipe material been observed in real world domestic hot water systems or in studies conducted with other lab scale apparatus? If yes, this would be an interesting thing to bring up in the final section of the results.

Response: References to studies that have made similar observations are now provided in lines 558-561:

‘Pipe type strongly influenced the composition of the microbial community that established in both the bulk water and biofilm, as has been observed in drinking water distribution systems [57], as well as other premise plumbing models such as pilot-scale rigs [58], simulated water heaters [59], and CDC biofilm reactors [60].’

This point is again emphasized in the final section, in lines 793-796:

‘These include chemical phenomena such as the observed differences in pH and DO, the variability and gradual reduction in copper and iron release, as well as biological differences such as total cells counts and microbial community composition in the bulk water and biofilm.’

1-35 Line 344 Remove the word however.

Response: This word has been removed.

1-36 Line 361 Did you actually observe sediments in your pipes?

Response: Sediments were regularly observed in the effluent of PVC-iron CMPRs. Line 594-595 has been changed to reflect this:

‘The greater variability in PVC-iron CMPRs may also be due to the influence of the sediments observed in the effluent of those pipes on the microbial ecology [62].’

1-37 Line 372 Why were all of the swabs taken from the PVC section of the pipes?

Response: This was necessary in order to ensure consistent and comparable recovery of biofilms, thus directly comparing the effects of the water chemistry on the biofilms at a consistent sampling surface. Depending on the study, researchers could choose to examine biofilms directly on the pipe surface of interest, but they would need to do so fully aware that corroded metal surfaces impede biofilm recovery and would need to design the experiment and the biofilm recovery system accordingly. This is now discussed in the manuscript in lines 303-305:

‘The PVC cap surfaces were selected for swabbing to ensure a consistent recovery of biofilm-associated microbes across conditions, as corrosion/deposition phenomena on the metal surfaces were anticipated to interfere with biofilm recovery.’

1-38 Lines 382-385 I don’t think that your DO findings, on their own, support all of this. I suggest that you tone down your language and bit and break this sentence into two sentences. The first sentence could begin with “this suggests that” rather than “thus” to emphasize that you are, to some degree at least, speculating about the causes underlying your DO results.

Response: This section has been changed to acknowledge that the explanation for these results in lines 623-626 is speculative:

‘This suggests that the convective mixing current subjected the mild steel pipe section to more corrosive conditions. Convective mixing likely increased the interactions of the bulk water with suspended solids and biofilm, increasing mass transfer and reactions with corrosion byproducts that consume DO’

1-39 Lines 386-389 This is an interesting point. Perhaps you could move it to the final section of the results to better support your assertions that the CMPRs can simulate real world conditions.

Response: This point, along with other real-world phenomena observed in the CMPRs, are now briefly emphasized in lines 791-807.

1-40 Line 418 Here and elsewhere in the paper you should write times, rather than x.

Response: These changes have been made. ‘1x’ and ‘2x’ in reference to water change frequency have been changed to ‘once’ and ‘twice.’

1-41 Lines 459-482 This section is, in my opinion, the heart of the paper, but it contains a number of unsupported or poorly supported claims by the authors and as a result it is ultimately unconvincing. For example, reference 10 was used to support the statement that pilot-scale plumbing rigs had higher inherent variability than your CMPRs, but when I quickly reviewed that paper, I saw no mention of high variability. In general, this section would be more effective if you focused on comparing your actual water quality findings (e.g. DO, cell counts, etc.) to the results of studies that examined full scale domestic hot water systems or that used CDC reactors, pilot pipe rigs, etc. This would show that your CMPRs were comparable or superior to existing methods. There is some of this in various parts of the results section (lines 338-340, lines 386-389, etc.) and this content could be moved to this final section to

better support your claims.

Response: As the reviewer mentioned, we agree that this would be extremely valuable, but as the reviewer also acknowledges, it is not feasible in this study.

To our knowledge such a comparison has never been made for any of the other premise plumbing simulation systems, including CDC reactors, pilot rigs, etc. We now have added text giving a list of reasons why such a comparison would be difficult-- mainly because premise plumbing itself is essentially infinitely variable in flow patterns, temperature, and disinfectant profiles.

Our intention is that the heart of the paper is the introduction and initial evaluation of the CMPR as a new premise plumbing simulation system and a frank first-time discussion of some practical strengths and weaknesses that could help other researchers make better decisions regarding which test approaches to apply in their future work. In forthcoming research projects, we will further evaluate the CMPRs for addressing specific research questions. Note that a comprehensive research effort comparing the efficacy of lab simulations to replicate problematic niches in premise plumbing would be a massive undertaking and one that would be difficult to justify and secure appropriate funding to achieve.

In this study, we do provide substantial evidence that the CMPRs would be at least as comparable as currently available laboratory simulations, while also being frank in acknowledging advantages, disadvantages, and unknowns.

Exemplary statements and changes in the revised manuscript in support of our response to Comment 1-41 include the following:

• Lines 406-411: ‘DO tended to be lower in the PVC pipes than in the PVC-copper pipes during week 2 (ncopper=9, nPVC=6, p=0.0018), although there was no difference between PVC-copper and PVC pipes following 9 weeks of acclimation. This is likely due to the aging of PVC-copper pipes, resulting in increased cell growth due to the decreased release of antimicrobial copper as the pipes aged [19, 47], which in turn allowed more cellular respiration to consume DO, as was observed earlier in the experiment for the PVC pipes.’

• Lines 443-446: ‘Average total iron decreased from 24.9 ± 4.68 mg/L at week 2 to 10.9 ± 1.62 mg/L at week 9, consistent with scale formation [51]. The variability in iron release was expected based on previous studies [46, 52, 53].’

• Lines 505-506: ‘Iron also removes DO from the water through redox reactions, leaving less for use by cells for respiration and growth [44, 46].’

• Lines 558-561: ‘Pipe type strongly influenced the composition of the microbial community that established in both the bulk water and biofilm, as has been observed in drinking water distribution systems [57], as well as other premise plumbing models such as pilot-scale rigs [58], simulated water heaters [59], and CDC biofilm reactors [60].’

• Lines 628-630: ‘The effect of lower DO in the CMPRs is desired in this case, as DO depletion is a common phenomenon in “real-world” water systems where galvanized iron pipes have corroded following the loss of the protective zinc layer [46].’

1-42 Lines 459-482, continued: The cost information is interesting but ultimately costs will differ from one jurisdiction to another and over time, so I wouldn’t make it the main focus here.

Response: We agree and use the words “estimated” in line 734 (‘Pilot-scale premise plumbing rigs constructed in prior studies were estimated to cost…’) and “representative” in the bottom of Table 2 (‘Representative Cost, USD per replicate’). Additionally, we acknowledge they will differ in the updated caption for Table 2:

‘Table 2. Comparison of the CMPR to alternative reactors for simulating premise plumbing based upon available pricing and other information for required equipment components and personal knowledge and experience of the authors. Y = Yes, N = No, D = Difficult/Expensive’

And in line 732, we preface this paragraph with the phrase, ‘while costs of any system will vary with time and location.’

1-43 Lines 459-482, continued: Also, assuming that stagnant pipe reactors like those used in this study are widely used to simulate domestic hot water systems, they should be included in Table 2 and discussed in this section.

Response: We did not mean to imply that this approach is superior to other premise plumbing models, or that it replicates every niche in premise plumbing. This novel approach does recreate a plumbing niche very important to OPs. We have emphasized research demonstrating some specific, real-world phenomena that our system replicates.

However, to quantitatively compare water quality findings from CMPRs to those of other models, simultaneous studies using the same source water would need to be performed as discussed earlier. We agree that the best way would be to directly compare several systems to a real-world building all using the same water source. To our knowledge, no prior studies have statistically compared one reactor approach to another, much less to results of a full-scale building. An additional paragraph to elucidate this position has been added in lines 752-764:

‘The ultimate basis of comparison, however, should be the ability of these systems to replicate real-world premise plumbing conditions. Here, we have shown that the CMPRs replicate phenomena observed in some important real-world settings. Unfortunately, it is not possible to compare our water quality findings to those of other full-scale systems because every water has a unique chemistry and every sampling location in a building is unique in terms of water use patterns, temperature and disinfectant time profiles. Such a comparison would be useful but it would be a major undertaking in terms of personnel and resource requirements, but would be highly valuable because it is important to obtain a better understanding of the real-world strengths and limitations of each laboratory simulation method. A particular strength of CMPRs is an ability to be modified to accommodate a wide range of replicates, pipe conditions, temperature regimes, and adjustments to influent water chemistry. We judge that CMPRs are a useful, novel and cost-effective test method that can help to better understand a wide variety of microbiological and physicochemical premise plumbing research issues.’

Furthermore, the bullet point beginning on line 491 of the original manuscript has been moved to the end of that list and expanded to place more emphasis water quality phenomena CMPRs replicate. Lines 791-807 now read:

‘CMPRs were able to induce and isolate certain effects of key test variables, e.g., pipe material effect on water chemistry and microbiology, that have been observed as trends in other premise plumbing simulations and full-scale buildings. These include chemical phenomena such as the observed differences in pH and DO, the variability and gradual reduction in copper and iron release, as well as biological differences such as total cells counts and microbial community composition in the bulk water and biofilm. This enables future testing to explore impacts of a wider range of chemistries and experimental conditions while rigorously testing for statistical confidence between true replicate reactors.’

Additionally, earlier edits emphasize comparisons between our water quality findings to those from other pilot system and full-scale building studies. See responses to Comments 1-22, 1-23, 1-25, 1-27, 1-30, & 1-34.

Our emphasis on the differences between stagnant pipes and CMPRs is intended to demonstrate the possible importance of convective mixing in facilitating the interactions between the bulk water, biofilm, and pipe wall that would be expected to occur in premise plumbing, as well as their role in achieving conditions suitable for microbial growth for the study of OPs.

1-44 Line 460 You didn’t measure nitrogen or discuss it in the rest of the paper, so I suggest that you omit it from this list.

Response: Nitrogen has been removed in line 718.

1-45 Line 461 This is the first time that your base / influent water is described as oligotrophic. This term may not be familiar to your readers, so I think it would be more effective to briefly describe the important characteristics of your water matrix (low TOC, moderate pH, dechlorinated, etc.)

Response: We did not mean to imply that the water used in this study is more oligotrophic than other drinking waters are expected to be. ‘Oligotrophic drinking water’ has been changed to ‘the low-nutrient environment of drinking water’ in lines 718-719. A table has been added to the SI to provide information regarding the influent water.

1-46 Line 476 Are mechanical failures truly unavoidable with the CDC reactors? Couldn’t the UV light, recirculation pump, heater, etc. in the CMPR set-up fail as well?

Response: Our experience in running multiple parallel reactors, is that mechanical failures and snafus are unavoidable, and they occur randomly to all reactors at some rate over the months needed to do a representative study, and that as a result none of the reactors are truly replicated. Moveover, the beauty of the CMPR, is that when (not if) a failure occurs, it is not only less frequent because there is only one of everything, but it happens to all reactors equally. So replication between reactors is maintained, even during the failures.

We did not intend to imply that individual components of the CMPR system are less likely to fail than those found in other reactors. Rather, that the need for multiple peristaltic pumps and stir plates for each pipe material when using CDC biofilm reactors introduces many more chances for an individual component to fail. This sentence has been updated in lines 743-747 as follows:

‘This makes testing multiple pipe conditions in parallel expensive and often cost prohibitive, as an individual reactor with full controls will be needed for each pipe type tested, while also introducing a greater chance of mechanical failure of at least one component. Moreover, when a failure of a component inevitably occurs in a CMPR, true replication is maintained because it will affect every pipe equally.’

1-47 Line 489 Some of your results, especially from week 2, do not support this conclusion.

Response: The phrase ‘beyond that which is inherent to new pipe materials’ has been added to line 783 in order to better specify the observed results. This, along with the above changes to address the natural variability of new pipe sections, should indicate more precisely what was found.

1-48 Figure 1 What does the checked box represent? Also, it might be nice to include a schematic showing the stagnant pipe apparatus for comparison.

Response: The figure has been revised so that the manuscript’s text clarifies that the checked box is an interior view of convective mixing in the pipe. In-text references to the schematic portion of Figure 1 (Fig 1A) and the interior view portion (Fig 1B) are now in lines 132, 133, 206 and 207. Changes to the figure caption referencing these separate panels are made in lines 142-148:

‘Fig 1. Schematic of convective mixing pipe reactors (CMPRs). (A) Capped ends of PVC-copper, PVC-iron, and PVC pipes are submerged at 60° in a hot water bath to simulate hot water recirculation, with plugged, accessible ends exposed to room temperature to simulate a stagnant distal outlet. Fifty-four pipes, configured into 6 rows of 9 pipes, were operated for this study. In-line ultraviolet light is incorporated for disinfection of water bath for secondary containment/disinfection of pathogens in case of leak to support BSL2 level experiments.’

We feel a visual representation of the stagnant pipes is not necessary, as they are much simpler to construct than the CMPRs and lines 220-221 of the manuscript provide sufficient detail for their complete replication. Additional details (‘…triplicates of each pipe were placed at a 60° angle against a shelf …’) have been provided in lines 220-221 to describe our particular configuration and facilitate reader’s understanding of the set-up.

1-49 Figure 2 I must admit to initially being totally confused by the letters on these plots because the way that you’ve indicated statistical differences is unfamiliar to me and, in my opinion, counterintuitive. I’m used to statistical differences being indicated by different letters, rather than by the same letters. So, for example, if condition 1 and condition 2 were statistically different, they would be labeled A and B, but if they were indistinguishable, they’d both be labeled A because they would be in the same group. This is in line with the outputs of all of the statistical programs that I’ve ever worked with. I think that the approach that I’ve described here is likely to be more familiar and intuitive to readers, so I suggest that you adopt it here.

Also, why do some plots only have capital letters and some only have lower case letters? Weren’t the same statistical tests run on all four datasets?

Response: Figures 2 and 3 have been edited so that letters indicate grouping rather than differences. The captions for these figures have been updated accordingly.

1-50 Figure 4 It would be interesting to see DO, cell counts, and iron plotted together (or on three faceted plots in a single figure) to emphasize how they influence one another.

Response: We have created a new plot and added this figure to the SI (Figure S. 2), comparing DO, cell counts, and iron data from weeks 2 and 9. Reference to this figure is provided in line 474. While visually the three parameters appear to change in sync as the pipes age, we were unable to identify a statistically-significant correlation between cell counts and either DO or iron concentration. We suspect that this is because of the natural variability in these parameters, as well as the fact that DO and iron only have two sampling points as mentioned earlier.

1-51 Table 2 This table is interesting but the authors don’t support their assumptions with any citations and it isn’t always clear how they came to their conclusions about different factors. For example, why wouldn’t it be possible create a standardized protocol for a pilot-scale pipe rig? Why would the CMPRs cost less to operate than a CDC reactor, especially seeing as the CMPR apparatus includes a UV lamp and heater, both of which would consume electricity? Why would it be more difficult or expensive to ensure biosafety protections for CDC biofilm reactors? Also, the authors should specify that their costing is USD.

Response: The assessment summarized in Table 2 is “based upon available pricing and other information for required equipment components and personal knowledge and experience of the authors,” which is now explicitly stated in the title of Table 2 in lines 724-726. A full explanation of each value in this table would form a substantial review article on its own and is better suited for a study that directly compares these systems. On the issue of cost, for example, the CMPR only needs one pump, UV lamp and heater, whereas every replicate CDC reactor requires one pump and one stir plate. The scaling of replicative costs are linear in the case of the CDC reactor. Thus, we make the case that the assumptions built into Table 2 are reasonable based on available information, while also making it clear that more precise comparison would require a separate study. USD and other caveats have been added to the bottom line of Table 2.

Attachment

Submitted filename: Final Reviewers Comments Response v4.docx

Click here for additional data file.^{(75.9KB, docx)}

PLoS One. doi: 10.1371/journal.pone.0238385.r003

Decision Letter 1

Zhi Zhou

17 Aug 2020

Replicable simulation of distal hot water premise plumbing using convectively-mixed pipe reactors

PONE-D-20-13252R1

Dear Dr. Edwards,

We’re pleased to inform you that your manuscript has been judged scientifically suitable for publication and will be formally accepted for publication once it meets all outstanding technical requirements.

Within one week, you’ll receive an e-mail detailing the required amendments. When these have been addressed, you’ll receive a formal acceptance letter and your manuscript will be scheduled for publication.

An invoice for payment will follow shortly after the formal acceptance. To ensure an efficient process, please log into Editorial Manager at http://www.editorialmanager.com/pone/, click the 'Update My Information' link at the top of the page, and double check that your user information is up-to-date. If you have any billing related questions, please contact our Author Billing department directly at authorbilling@plos.org.

If your institution or institutions have a press office, please notify them about your upcoming paper to help maximize its impact. If they’ll be preparing press materials, please inform our press team as soon as possible -- no later than 48 hours after receiving the formal acceptance. Your manuscript will remain under strict press embargo until 2 pm Eastern Time on the date of publication. For more information, please contact onepress@plos.org.

Kind regards,

Zhi Zhou, Ph.D.

Academic Editor

PLOS ONE

Additional Editor Comments (optional):

Reviewers' comments:

Reviewer's Responses to Questions

Comments to the Author

1. If the authors have adequately addressed your comments raised in a previous round of review and you feel that this manuscript is now acceptable for publication, you may indicate that here to bypass the “Comments to the Author” section, enter your conflict of interest statement in the “Confidential to Editor” section, and submit your "Accept" recommendation.

Reviewer #1: All comments have been addressed

**********

2. Is the manuscript technically sound, and do the data support the conclusions?

Reviewer #1: Yes

**********

3. Has the statistical analysis been performed appropriately and rigorously?

Reviewer #1: Yes

**********

4. Have the authors made all data underlying the findings in their manuscript fully available?

Reviewer #1: Yes

**********

5. Is the manuscript presented in an intelligible fashion and written in standard English?

Reviewer #1: Yes

**********

6. Review Comments to the Author

Reviewer #1: Thanks for addressing all of my comments. It's unfortunate that it wasn't feasible to compare your results to full scale systems for this paper, but I suppose that just means that there is still some interesting work to be done in this space.

**********

7. PLOS authors have the option to publish the peer review history of their article (what does this mean?). If published, this will include your full peer review and any attached files.

If you choose “no”, your identity will remain anonymous but your review may still be made public.

Do you want your identity to be public for this peer review? For information about this choice, including consent withdrawal, please see our Privacy Policy.

Reviewer #1: Yes: Stephanie Gora

PLoS One. doi: 10.1371/journal.pone.0238385.r004

Acceptance letter

Zhi Zhou

20 Aug 2020

PONE-D-20-13252R1

Replicable simulation of distal hot water premise plumbing using convectively-mixed pipe reactors

Dear Dr. Edwards:

I'm pleased to inform you that your manuscript has been deemed suitable for publication in PLOS ONE. Congratulations! Your manuscript is now with our production department.

If your institution or institutions have a press office, please let them know about your upcoming paper now to help maximize its impact. If they'll be preparing press materials, please inform our press team within the next 48 hours. Your manuscript will remain under strict press embargo until 2 pm Eastern Time on the date of publication. For more information please contact onepress@plos.org.

If we can help with anything else, please email us at plosone@plos.org.

Thank you for submitting your work to PLOS ONE and supporting open access.

Kind regards,

PLOS ONE Editorial Office Staff

on behalf of

Dr. Zhi Zhou

Academic Editor

PLOS ONE

Associated Data

This section collects any data citations, data availability statements, or supplementary materials included in this article.

Supplementary Materials

S1 Fig. Temperature profile of clear PVC pipe filled with RO water after allowed to equilibrate with the room and incubator temperatures to undergo convective mixing at different angles.

Incubator consisted of a box using a light bulb as a heat source. Temperatures were measured using an infrared thermometer temperature gun at various points along the pipe.

(DOCX)

Click here for additional data file.^{(18.5KB, docx)}

S2 Fig. Comparison of total cell counts, DO, and total iron in CMPR effluent in weeks 2 and 9.

No significant correlation was found between total cell counts and either DO or total iron (Spearman rank correlation).

(DOCX)

Click here for additional data file.^{(226.3KB, docx)}

S3 Fig. Phylum level taxonomic microbial community profiles of each CMPR and each stagnant pipe organized by pipe type and location.

(DOCX)

Click here for additional data file.^{(352.2KB, docx)}

(DOCX)

Click here for additional data file.^{(77.3KB, docx)}

S5 Fig

(DOCX)

Click here for additional data file.^{(65.4KB, docx)}

S1 Table. Test data determining the maximum mixing velocity based on pipe angle using neutrally buoyant rhodamine dye.

(DOCX)

Click here for additional data file.^{(14.5KB, docx)}

S2 Table. Physicochemical characteristics of influent water.

(DOCX)

Click here for additional data file.^{(13.2KB, docx)}

S3 Table. Cost estimates for each type of premise plumbing simulation method.

(DOCX)

Click here for additional data file.^{(14.9KB, docx)}

S1 File. Manuscript data.

Data used for analysis in this manuscript.

(XLSX)

Click here for additional data file.^{(91.9KB, xlsx)}

S2 File

(DOCX)

Click here for additional data file.^{(2.1MB, docx)}

Attachment

Submitted filename: Final Reviewers Comments Response v4.docx

Click here for additional data file.^{(75.9KB, docx)}

Data Availability Statement

Sequence data have been deposited in the NCBI Sequence Read Archive (BioProject ID PRJNA609991). Additional relevant data are within the manuscript and its Supporting Information files.

[pone.0238385.ref001] 1.Collier SA, Stockman LJ, Hicks LA, Garrison LE, Zhou F, Beach MJ. Direct healthcare costs of selected diseases primarily or partially transmitted by water. Epidemiology & Infection. 2012;140(11):2003–13. [DOI] [PMC free article] [PubMed] [Google Scholar]

[pone.0238385.ref002] 2.Falkinham JO III, Hilborn ED, Arduino MJ, Pruden A, Edwards MA. Epidemiology and ecology of opportunistic premise plumbing pathogens: Legionella pneumophila, Mycobacterium avium, and Pseudomonas aeruginosa. Environmental Health Perspectives. 2015;123(8):749–58. 10.1289/ehp.1408692 [DOI] [PMC free article] [PubMed] [Google Scholar]

[pone.0238385.ref003] 3.Falkinham JO, Pruden A, Edwards M. Opportunistic premise plumbing pathogens: increasingly important pathogens in drinking water. Pathogens. 2015;4(2):373–86. 10.3390/pathogens4020373 [DOI] [PMC free article] [PubMed] [Google Scholar]

[pone.0238385.ref004] 4.Pruden A, Edwards MA, Falkinham JO. State of the science and research needs for opportunistic pathogens in premise plumbing: Water Research Foundation; 2013. [Google Scholar]

[pone.0238385.ref005] 5.Donlan R, Piede J, Heyes C, Sanii L, Murga R, Edmonds P, et al. Model system for growing and quantifying Streptococcus pneumoniae biofilms in situ and in real time. Appl Environ Microbiol. 2004;70(8):4980–8. 10.1128/AEM.70.8.4980-4988.2004 [DOI] [PMC free article] [PubMed] [Google Scholar]

[pone.0238385.ref006] 6.Goeres DM, Loetterle LR, Hamilton MA, Murga R, Kirby DW, Donlan RM. Statistical assessment of a laboratory method for growing biofilms. Microbiology. 2005;151(3):757–62. [DOI] [PubMed] [Google Scholar]

[pone.0238385.ref007] 7.Camper AK, Jones WL, Hayes JT. Effect of growth conditions and substratum composition on the persistence of coliforms in mixed-population biofilms. Appl Environ Microbiol. 1996;62(11):4014–8. 10.1128/AEM.62.11.4014-4018.1996 [DOI] [PMC free article] [PubMed] [Google Scholar]

[pone.0238385.ref008] 8.Camper AK, Brastrup K, Sandvig A, Clement J, Spencer C, Capuzzi A. Effect of distribution system materials on bacterial regrowth. Journal‐American Water Works Association. 2003;95(7):107–21. [Google Scholar]

[pone.0238385.ref009] 9.Douterelo I, Sharpe R, Boxall J. Influence of hydraulic regimes on bacterial community structure and composition in an experimental drinking water distribution system. Water Research. 2013;47(2):503–16. 10.1016/j.watres.2012.09.053 [DOI] [PubMed] [Google Scholar]

[pone.0238385.ref010] 10.Rhoads WJ, Ji P, Pruden A, Edwards MA. Water heater temperature set point and water use patterns influence Legionella pneumophila and associated microorganisms at the tap. Microbiome. 2015;3(1):67. [DOI] [PMC free article] [PubMed] [Google Scholar]

[pone.0238385.ref011] 11.Leon Ohl A, Horn H, Hempel D. Behaviour of biofilm systems under varying hydrodynamic conditions. Water Science and Technology. 2004;49(11–12):345–51. [PubMed] [Google Scholar]

[pone.0238385.ref012] 12.Liu Z, Lin Y, Stout J, Hwang C, Vidic R, Yu V. Effect of flow regimes on the presence of Legionella within the biofilm of a model plumbing system. J Appl Microbiol. 2006;101(2):437–42. 10.1111/j.1365-2672.2006.02970.x [DOI] [PubMed] [Google Scholar]

[pone.0238385.ref013] 13.National Research Council. Drinking water distribution systems: assessing and reducing risks : National Academies Press; 2007. [Google Scholar]

[pone.0238385.ref014] 14.Brazos B, O'Connor J, Abcouwer S, editors. Kinetics of chlorine depletion and microbial growth in household plumbing systems. Proceedings AWWA 1985 Water Quality Technology Conference, Houston; 1985.

[pone.0238385.ref015] 15.Nguyen C, Elfland C, Edwards M. Impact of advanced water conservation features and new copper pipe on rapid chloramine decay and microbial regrowth. Water Research. 2012;46(3):611–21. 10.1016/j.watres.2011.11.006 [DOI] [PubMed] [Google Scholar]

[pone.0238385.ref016] 16.Rhoads WJ, Pruden A, Edwards MA. Convective mixing in distal pipes exacerbates Legionella pneumophila growth in hot water plumbing. Pathogens. 2016;5(1):29. [DOI] [PMC free article] [PubMed] [Google Scholar]

[pone.0238385.ref017] 17.Ji P, Rhoads WJ, Edwards MA, Pruden A. Impact of water heater temperature setting and water use frequency on the building plumbing microbiome. The ISME Journal. 2017;11(6):1318–30. 10.1038/ismej.2017.14 [DOI] [PMC free article] [PubMed] [Google Scholar]

[pone.0238385.ref018] 18.Lehtola MJ, Miettinen IT, Keinänen MM, Kekki TK, Laine O, Hirvonen A, et al. Microbiology, chemistry and biofilm development in a pilot drinking water distribution system with copper and plastic pipes. Water Research. 2004;38(17):3769–79. 10.1016/j.watres.2004.06.024 [DOI] [PubMed] [Google Scholar]

[pone.0238385.ref019] 19.Van der Kooij D, Veenendaal HR, Scheffer WJ. Biofilm formation and multiplication of Legionella in a model warm water system with pipes of copper, stainless steel and cross-linked polyethylene. Water Research. 2005;39(13):2789–98. 10.1016/j.watres.2005.04.075 [DOI] [PubMed] [Google Scholar]

[pone.0238385.ref020] 20.Hammes F, Berney M, Wang Y, Vital M, Köster O, Egli T. Flow-cytometric total bacterial cell counts as a descriptive microbiological parameter for drinking water treatment processes. Water Research. 2008;42(1–2):269–77. 10.1016/j.watres.2007.07.009 [DOI] [PubMed] [Google Scholar]

[pone.0238385.ref021] 21.Rhoads WJ, Pruden A, Edwards MA. Interactive effects of corrosion, copper, and chloramines on Legionella and mycobacteria in hot water plumbing. Environ Sci Technol. 2017;51(12):7065–75. 10.1021/acs.est.6b05616 [DOI] [PubMed] [Google Scholar]

[pone.0238385.ref022] 22.Suzuki MT, Taylor LT, DeLong EF. Quantitative analysis of small-subunit rRNA genes in mixed microbial populations via 5′-nuclease assays. Appl Environ Microbiol. 2000;66(11):4605–14. 10.1128/aem.66.11.4605-4614.2000 [DOI] [PMC free article] [PubMed] [Google Scholar]

[pone.0238385.ref023] 23.Wang H, Bedard E, Prevost M, Camper AK, Hill VR, Pruden A. Methodological approaches for monitoring opportunistic pathogens in premise plumbing: A review. Water Research. 2017;117:68–86. 10.1016/j.watres.2017.03.046 [DOI] [PMC free article] [PubMed] [Google Scholar]

[pone.0238385.ref024] 24.Bolyen E, Rideout JR, Dillon MR, Bokulich NA, Abnet CC, Al-Ghalith GA, et al. Reproducible, interactive, scalable and extensible microbiome data science using QIIME 2. Nature Biotechnology. 2019;37(8):852–7. 10.1038/s41587-019-0209-9 [DOI] [PMC free article] [PubMed] [Google Scholar]

[pone.0238385.ref025] 25.Callahan BJ, McMurdie PJ, Rosen MJ, Han AW, Johnson AJA, Holmes SP. DADA2: high-resolution sample inference from Illumina amplicon data. Nature Methods. 2016;13(7):581 10.1038/nmeth.3869 [DOI] [PMC free article] [PubMed] [Google Scholar]

[pone.0238385.ref026] 26.Pedregosa F, Varoquaux G, Gramfort A, Michel V, Thirion B, Grisel O, et al. Scikit-learn: Machine learning in Python. The Journal of Machine Learning Research. 2011;12:2825–30. [Google Scholar]

[pone.0238385.ref027] 27.Quast C, Pruesse E, Yilmaz P, Gerken J, Schweer T, Yarza P, et al. The SILVA ribosomal RNA gene database project: improved data processing and web-based tools. Nucleic Acids Research. 2012;41(D1):D590–D6. [DOI] [PMC free article] [PubMed] [Google Scholar]

[pone.0238385.ref028] 28.McMurdie PJ, Holmes S. phyloseq: an R package for reproducible interactive analysis and graphics of microbiome census data. PLOS One. 2013;8(4). [DOI] [PMC free article] [PubMed] [Google Scholar]

[pone.0238385.ref029] 29.Anderson MJ. A new method for non‐parametric multivariate analysis of variance. Austral Ecology. 2001;26(1):32–46. [Google Scholar]

[pone.0238385.ref030] 30.Oksanen J, Kindt R, Legendre P, O’Hara B, Stevens MHH, Oksanen MJ, et al. The vegan package. Community Ecology Package. 2007;10:631–7. [Google Scholar]

[pone.0238385.ref031] 31.Anderson MJ. Distance‐based tests for homogeneity of multivariate dispersions. Biometrics. 2006;62(1):245–53. 10.1111/j.1541-0420.2005.00440.x [DOI] [PubMed] [Google Scholar]

[pone.0238385.ref032] 32.Love MI, Huber W, Anders S. Moderated estimation of fold change and dispersion for RNA-seq data with DESeq2. Genome Biology. 2014;15(12):550 10.1186/s13059-014-0550-8 [DOI] [PMC free article] [PubMed] [Google Scholar]

[pone.0238385.ref033] 33.National Academies of Sciences, Engineering, and Medicine. Management of Legionella in water systems: National Academies Press; 2020. [PubMed] [Google Scholar]

[pone.0238385.ref034] 34.Yee RB, Wadowsky RM. Multiplication of Legionella pneumophila in unsterilized tap water. Appl Environ Microbiol. 1982;43(6):1330–4. 10.1128/AEM.43.6.1330-1334.1982 [DOI] [PMC free article] [PubMed] [Google Scholar]

[pone.0238385.ref035] 35.Wurts WA, Durborow RM. Interactions of pH, carbon dioxide, alkalinity and hardness in fish ponds. 1992. [Google Scholar]

[pone.0238385.ref036] 36.Feely RA, Alin SR, Newton J, Sabine CL, Warner M, Devol A, et al. The combined effects of ocean acidification, mixing, and respiration on pH and carbonate saturation in an urbanized estuary. Estuarine, Coastal and Shelf Science. 2010;88(4):442–9. 10.1016/j.ecss.2010.05.004. [DOI] [Google Scholar]

[pone.0238385.ref037] 37.Stumm W, Morgan JJ. Aquatic chemistry: chemical equilibria and rates in natural waters: John Wiley & Sons; 2012. [Google Scholar]

[pone.0238385.ref038] 38.Deberry DW, Kidwell JR, Malish D. Corrosion in Potable Water Systems–Final Report. Washington, DC: United States Environmental Protection Agency, Office of Drinking Water; 1982. [Google Scholar]

[pone.0238385.ref039] 39.Salehi M, Odimayomi T, Ra K, Ley C, Julien R, Nejadhashemi AP, et al. An investigation of spatial and temporal drinking water quality variation in green residential plumbing. Building and Environment. 2020;169:106566 10.1016/j.buildenv.2019.106566. [DOI] [Google Scholar]

[pone.0238385.ref040] 40.Pinto AJ, Xi C, Raskin L. Bacterial community structure in the drinking water microbiome is governed by filtration processes. Environ Sci Technol. 2012;46(16):8851–9. 10.1021/es302042t [DOI] [PubMed] [Google Scholar]

[pone.0238385.ref041] 41.Butterfield C, Wattie E, Megregian S, Chambers C. Influence of pH and temperature on the survival of coliforms and enteric pathogens when exposed to free chlorine. Public Health Reports (1896–1970). 1943:1837–66. [PubMed] [Google Scholar]

[pone.0238385.ref042] 42.Butterfield C, Wattie E. Influence of pH and temperature on the survival of coliforms and enteric pathogens when exposed to chloramine. Public Health Reports (1896–1970). 1946:157–92. [PubMed] [Google Scholar]

[pone.0238385.ref043] 43.Yu-sen EL, Vidic RD, Stout JE, Victor LY. Negative effect of high pH on biocidal efficacy of copper and silver ions in controlling Legionella pneumophila. Appl Environ Microbiol. 2002;68(6):2711–5. 10.1128/aem.68.6.2711-2715.2002 [DOI] [PMC free article] [PubMed] [Google Scholar]

[pone.0238385.ref044] 44.Zhang Y, Griffin A, Edwards M. Effect of nitrification on corrosion of galvanized iron, copper, and concrete. Journal AWWA. 2010;102(4):83–93. 10.1002/j.1551-8833.2010.tb10093.x [DOI] [Google Scholar]

[pone.0238385.ref045] 45.McNeill LS, Edwards M. IRON PIPE corrosion IN DISTRIBUTION SYSTEMS. Journal AWWA. 2001;93(7):88–100. 10.1002/j.1551-8833.2001.tb09246.x [DOI] [Google Scholar]

[pone.0238385.ref046] 46.Sarin P, Snoeyink V, Bebee J, Jim K, Beckett M, Kriven WM, et al. Iron release from corroded iron pipes in drinking water distribution systems: effect of dissolved oxygen. Water Research. 2004;38(5):1259–69. 10.1016/j.watres.2003.11.022 [DOI] [PubMed] [Google Scholar]

[pone.0238385.ref047] 47.Lytle DA, Liggett J. Impact of water quality on chlorine demand of corroding copper. Water Research. 2016;92:11–21. 10.1016/j.watres.2016.01.032 [DOI] [PubMed] [Google Scholar]

[pone.0238385.ref048] 48.Edwards M, Powers K, Hidmi L, Schock M. The role of pipe ageing in copper corrosion by-product release. Water Science and Technology: Water Supply. 2001;1(3):25–32. [Google Scholar]

[pone.0238385.ref049] 49.Masters S, Parks J, Atassi A, Edwards MA. Inherent variability in lead and copper collected during standardized sampling. Environ Monit Assess. 2016;188(3):177 10.1007/s10661-016-5182-x [DOI] [PubMed] [Google Scholar]

[pone.0238385.ref050] 50.Rushing JC, Edwards M. The role of temperature gradients in residential copper pipe corrosion. Corrosion Science. 2004;46(8):1883–94. 10.1016/j.corsci.2003.11.001. [DOI] [Google Scholar]

[pone.0238385.ref051] 51.Tang Z, Hong S, Xiao W, Taylor J. Characteristics of iron corrosion scales established under blending of ground, surface, and saline waters and their impacts on iron release in the pipe distribution system. Corrosion Science. 2006;48(2):322–42. [Google Scholar]

[pone.0238385.ref052] 52.Lytle DA, Sarin P, Snoeyink VL. The effect of chloride and orthophosphate on the release of iron from a cast iron pipe section. Journal of Water Supply: Research and Technology-Aqua. 2005;54(5):267–81. 10.2166/aqua.2005.0026 [DOI] [Google Scholar]

[pone.0238385.ref053] 53.McNeill Laurie S, Edwards M. Phosphate Inhibitors and Red Water in Stagnant Iron Pipes. Journal of Environmental Engineering. 2000;126(12):1096–102. 10.1061/(ASCE)0733-9372(2000)126:12(1096) [DOI] [Google Scholar]

[pone.0238385.ref054] 54.Silhan J, Corfitzen CB, Albrechtsen HJ. Effect of temperature and pipe material on biofilm formation and survival of Escherichia coli in used drinking water pipes: a laboratory-based study. Water Science and Technology. 2006;54(3):49–56. 10.2166/wst.2006.447 [DOI] [PubMed] [Google Scholar]

[pone.0238385.ref055] 55.Schwartz T, Hoffmann S, Obst U. Formation and bacterial composition of young, natural biofilms obtained from public bank-filtered drinking water systems. Water Research. 1998;32(9):2787–97. 10.1016/S0043-1354(98)00026-8. [DOI] [Google Scholar]

[pone.0238385.ref056] 56.Kerr C, Osborn K, Robson G, Handley P. The relationship between pipe material and biofilm formation in a laboratory model system. J Appl Microbiol. 1998;85(S1):29S–38S. [DOI] [PubMed] [Google Scholar]

[pone.0238385.ref057] 57.Lin W, Yu Z, Chen X, Liu R, Zhang H. Molecular characterization of natural biofilms from household taps with different materials: PVC, stainless steel, and cast iron in drinking water distribution system. Applied Microbiology and Biotechnology. 2013;97(18):8393–401. 10.1007/s00253-012-4557-3 [DOI] [PubMed] [Google Scholar]

[pone.0238385.ref058] 58.Ji P, Parks J, Edwards MA, Pruden A. Impact of Water Chemistry, Pipe Material and Stagnation on the Building Plumbing Microbiome. PloS one. 2015;10(10):e0141087–e. 10.1371/journal.pone.0141087 . [DOI] [PMC free article] [PubMed] [Google Scholar]

[pone.0238385.ref059] 59.Proctor CR, Dai D, Edwards MA, Pruden A. Interactive effects of temperature, organic carbon, and pipe material on microbiota composition and Legionella pneumophila in hot water plumbing systems. Microbiome. 2017;5(1):130 10.1186/s40168-017-0348-5 [DOI] [PMC free article] [PubMed] [Google Scholar]

[pone.0238385.ref060] 60.Lu J, Buse HY, Gomez-Alvarez V, Struewing I, Santo Domingo J, Ashbolt NJ. Impact of drinking water conditions and copper materials on downstream biofilm microbial communities and Legionella pneumophila colonization. J Appl Microbiol. 2014;117(3):905–18. 10.1111/jam.12578 [DOI] [PubMed] [Google Scholar]

[pone.0238385.ref061] 61.Qin K, Struewing I, Domingo JS, Lytle D, Lu J. Opportunistic pathogens and microbial communities and their associations with sediment physical parameters in drinking water storage tank sediments. Pathogens. 2017;6(4):54. [DOI] [PMC free article] [PubMed] [Google Scholar]

[pone.0238385.ref062] 62.Liu G, Bakker G, Li S, Vreeburg J, Verberk J, Medema G, et al. Pyrosequencing reveals bacterial communities in unchlorinated drinking water distribution system: an integral study of bulk water, suspended solids, loose deposits, and pipe wall biofilm. Environ Sci Technol. 2014;48(10):5467–76. 10.1021/es5009467 [DOI] [PubMed] [Google Scholar]

[pone.0238385.ref063] 63.Boe-Hansen R, Albrechtsen H-J, Arvin E, Jørgensen C. Dynamics of biofilm formation in a model drinking water distribution system. Journal of Water Supply: Research and Technology—AQUA. 2002;51(7):399–406. [Google Scholar]

[pone.0238385.ref064] 64.Buse HY, Lu J, Struewing IT, Ashbolt NJ. Preferential colonization and release of Legionella pneumophila from mature drinking water biofilms grown on copper versus unplasticized polyvinylchloride coupons. International Journal of Hygiene and Environmental Health. 2014;217(2–3):219–25. 10.1016/j.ijheh.2013.04.005 [DOI] [PubMed] [Google Scholar]

[pone.0238385.ref065] 65.Martiny AC, Jørgensen TM, Albrechtsen H-J, Arvin E, Molin S. Long-term succession of structure and diversity of a biofilm formed in a model drinking water distribution system. Appl Environ Microbiol. 2003;69(11):6899–907. 10.1128/aem.69.11.6899-6907.2003 [DOI] [PMC free article] [PubMed] [Google Scholar]

PERMALINK

Replicable simulation of distal hot water premise plumbing using convectively-mixed pipe reactors

M Storme Spencer

Abraham C Cullom

William J Rhoads

Amy Pruden

Marc A Edwards

Roles

Abstract

Introduction

Fig 1. Schematic of convective mixing pipe reactors (CMPRs).

Materials and methods

Convectively mixed pipe reactor (CMPR) design and operation

Water quality analysis

Biological sampling

16S rRNA amplicon sequencing

Statistical analyses

Results and discussion

Trends and variance among the CMPRs

Physicochemical consistency among the CMPRs

Table 1. Water quality parameters for replicate CMPRs during week 2 (acclimation, once weekly water change), week 9 (differentiation, twice weekly water change) or week 10 (qPCR data).

Fig 2. Physicochemical comparison between CMPRs and stagnant pipes.

Microbiological characteristics of the CMPRs

Fig 3. Quantitative PCR comparison between CMPRs and stagnant pipes.

Effect of pipe type on bacterial growth with convective mixing

Fig 4. Total cell counts in CMPRs and stagnant pipes as a function of aging.

CMPRs versus stagnant pipes

Physicochemical differences

Differences in microbial numbers

Comparison of microbial communities in stagnant and CMPR Pipes

Fig 5. Dissimilarity in microbiome compositions of samples between CMPRs and stagnant pipes.

CMPRs versus other premise plumbing simulation methods

Table 2. Comparison of the CMPR to alternative reactors for simulating premise plumbing based upon available pricing and other information for required equipment components and personal knowledge and experience of the authors.

Conclusions

Supporting information

Acknowledgments

Data Availability

Funding Statement

References

Decision Letter 0

Zhi Zhou

Roles

Author response to Decision Letter 0

Decision Letter 1

Zhi Zhou

Roles

Acceptance letter

Zhi Zhou

Roles

Associated Data

Supplementary Materials

Data Availability Statement

ACTIONS

PERMALINK

RESOURCES

Similar articles

Cited by other articles

Links to NCBI Databases