Biofilms shelter a diverse community of microorganisms, including bacteria, fungi, algae, and protozoa
[57]. Protozoa are important components or predators of the biofilms, affecting their structure and their internal complex feeding dynamics
[58]. Murga et al.
[59] demonstrated that
Legionella spp. are able to persist in a laboratory biofilm of
Pseudomonas aeruginosa,
Klebsiella pneumoniae, and a
Flavobacterium sp.; however, they are not able to replicate without the presence of
Hartmannella vermiformis. The work by Declerck et al.
[60] shows that the presence of the amoebae
Acanthamoeba castellanii is important to spread
L. pneumophila in a laboratory-simulated biofilm (from water distribution pipes) in a rotating annular reactor. Recently, Shahen et al.
[44] proposed an interesting model for the association of
Legionella–amoebae–biofilms. At first, biofilms and free-living amoebae growths are positively linked, and amoebae feed on (non-pathogen) bacteria in the biofilm. When the nutritional options become scarce and the ratio of amoebae to
Legionella increases, amoebae enter a ‘must-feed-on-
L. pneumophila’ mode, undergoing the formerly described growth/release-to-the-water cycles, liberating high concentrations of
L. pneumophila in the bulk water. This model, in a broader sense, seems to corroborate the conclusions of van der Kooij et al.
[61], who observed that
L. pneumophila proliferation depends on host protozoan, and found out that pathogen growth was dependent on the biofilm concentration—reduced
Legionella growth was also observed when biofilm concentration decreased. Additionally, the work by Kuiper et al.
[62] shows that the intercellular growth of
L. pneumophila in
Hartmannella vermiformis, in a batch laboratory system, was the main proliferation mechanism in the biofilm. Very interestingly, the authors concluded that 90% of
H. vermiformis was present in the biofilm and observed a positive relationship between the
Legionella concentration in the system and the attached biomass amount, suggesting that controlling biofilm build-up can limit
L. pneumophila proliferation.
On the other hand, studies with other biofilm models indicate that
L. pneumophila might use the exogenous products (e.g., amino acids) of other environmental bacteria to support its replication
[40][63]. Surman et al.
[37] used a model water system to investigate whether
L. pneumophila would replicate without a host protozoan. The authors’ conclusions suggest that intracellular replication is not mandatory for
Legionella replication ‘as long as there are other bacterial species present’. This supports the findings of Taylor et al.
[64], which highlighted the role and complexity of the different survival mechanisms that
Legionella seems to be able to use, adapt, and persist in the water systems.
An exhaustive overview of the link between
Legionella and biofilms or between
Legionella and protozoa is out of the scope of the present review. The reader might find complementary important information about these topics in former works
[30][31][35][64][48][58].
32.3. Bottlenecks of Real-Field Legionella Control
Whether
Legionella can replicate in the biofilm without a host protozoan or not, it is consensual that biofilms are relevant sites for
Legionella settlement in man-made water systems
[27]. As a consequence of the biofilm life cycle or as a result of operational dynamics of the water system, part of the biofilms colonized with
Legionella might be dislodged and, upon aerosolization, cause legionellosis events
[23][27]. Furthermore, biofilm shelters its microbial community against external aggressions such as temperature changes or biocides
[65]. For example, Giao et al.
[66] used a two-stage chemostat to grow heterotrophic biofilms from drinking water and studied the effect of increasing chlorine dosages on
L. pneumophila planktonic and sessile (biofilm) cells. The authors found that, regardless of chlorine presence (tested concentrations of 0.2 and 1.2 mgCl
2/l),
L. pneumophila could represent up to 25% of the total attached microbial community and that the total cell numbers of
Legionella in the biofilm were not affected by the residue’s concentrations of biocide. These results agree with the conclusion of Wright et al.
[67], who found that sessile populations were more resistant to the two tested biocides (Kathon and Bronopol) as compared with planktonic cells, emphasizing the extra protection conferred by biofilms
[67]. Furthermore, the biofilm’s physical stability is highly relevant for the success of cleaning and disinfection procedures
[68][69].
This puts great emphasis on proper biofilm management as part of an integrated approach to mitigate legionellosis incidence. Therefore, biofilm (often linked to dirtiness) control techniques are important components of legionellosis prevention
[23][24][25]. However, at this point, a paradigmatic aspect typically arises: although effective water treatment programs against
Legionella should focus on biofilms and planktonic bacteria
[24], the indicative threshold action levels are only set for bacteria in the water (
Legionella spp.)
[23][24]. In practice, this might result in
Legionella management procedures that are essentially grounded in occasional
Legionella water sampling results, which follow an underlying logic of ‘non-detected’ vs. ‘detected’. Through this perspective, a ‘non-detected
Legionella spp. result’ might be interpreted as ‘everything is OK’, while a positivity might indicate that something must be done or adjusted
[10][70].
Grounding
Legionella management on discrete planktonic heterotrophic bacteria counts and
Legionella spp. screening is probably one of the main weaknesses of current preventive real-field practices. Counteracting and over-relying on such information biases the interpretation of the microbiological status of the system
[10][27][28]. Firstly, water samples do not give representative information about the number of microorganisms in the system nor about the extent or location of the biofilm
[71]. For example, Flemming et al.
[57] estimated that 95% of the biomass present in drinking water distribution systems is attached to the walls rather than in the water. The under-representativeness of water samples is further illustrated in the works of Bonadonna et al.
[72]. Bonadonna et al.
[72] showed that the concentration of legionellae in biofilms from hot water networks was more than three orders of magnitude higher than the one recovered from the bulk water.
This point is further aggravated by discrete sampling, i.e., single snapshots in time of the microbiological status of the system
[28]. For example, Bentham
[27] found that in 25 of the 28 cooling towers sampled, there was no statistical relationship between
Legionella culture results taken 2 weeks apart, demonstrating that the microbiological status of the system changes within a small timeframe (as compared to routine water sampling).
32.4. The Scientific Perpetuation of a Water Legionella-Sampling Approach
Not surprisingly,
Legionella sampling in the water has been perpetuated in real-field practices, but also in scientific studies. Despite the limitations previously discussed, routine
Legionella screening in the water provides an output that has a call-to-action significance (especially for culture methods) that is very relevant to assess the efficacy of proper
Legionella water safety management
[23][25][28].
Culture methods, such as the international standard ISO 11731 (
ISO 11731 ‘Water quality. Detection and enumeration of Legionella’), have been standardized for several decades and are still considered the gold standard for
Legionella screening in some reference documents
[24]. Although they provide retrospective information (10 to 14 days to obtain a result) and underestimate the number of
Legionella present in the water sample
[73], the historical datasets and knowledge gained upon the use of culture methods over several decades (in distinct situations, including the investigation of legionellosis outbreak events) allowed the establishment of indicative thresholds of action according to the concentration of
Legionella spp. in the water
[25].
The advent of molecular techniques such as qPCR is providing an important boost to the study of
Legionella ecology as they overcome some culture limitations
[29][74]. These culture limitations are mostly linked to the following issues
[29]: (i)
Legionella cultivability is affected by the fastidious nature of the bacteria’s growth; (ii) the presence of other colonizing bacteria in the water sample may negatively affect the capacity of
Legionella to grow in laboratory medium; (iii)
Legionella VBNC cells
[38] or
Legionella inside vesicles (expelled from protozoa) are not detected; (iv) holding times between sampling collection and processing can lead to cultivability loss. On the other hand, qPCR detects DNA fragments that might belong to culturable, VBNC, and inactivated or even dead organisms, failing to distinguish between live and dead cells
[29]. Due to the presence of inhibitory compounds, some water samples in CTs might also show qPCR inhibition, leading to false-negative results. Young et al.
[29] estimated (based on five independent studies in CTs) that the inhibition fractions might be around 10%. Despite these limitations, the works of Young et al.
[29] and Collins et al.
[74] suggest that
Legionella spp. qPCR is a good tool to use in routine monitoring, and they propose action and alert levels that can help to interpret GU (genomic units) of
Legionella spp. per liter. More conservatively, Fisher et al.
[75] advise the use of qPCR for rapid
Legionella screening, where a PCR-negative result suggests no
Legionella presence, and a positive output should require confirmation via culture method. Hopefully, the potentialities of molecular approaches will push the development of new methods for
Legionella detection and quantification in situ and the design of simple-to-use and portable solutions for industrial application
[76].
The lack of standard practices for biofilm sampling and analysis
[32][33], even for research purposes, also contributes to this water screening perpetuation. Swabbing the surface is often used with the aim of analyzing
Legionella at the biofilms
[77][78], yet the scope of the standard application does not include biofilm sampling. Swab sampling is usually based on the international standard ISO 18593-2004 (
ISO 18593:2004 ‘Microbiology of food and animal feeding stuffs—Horizontal methods for sampling techniques from surfaces using contact plates and swabs’). However, swab sampling aims to assess the microbial load on surfaces (mostly for food safety purposes) rather than sample or examine the biofilm in industrial water systems. Swab sampling destroys the biofilm structure, and measuring the swabbed area is often an unfeasible task
[25]. However, in the absence of a more suitable approach, it is recommended for surface screening purposes related to
Legionella [23].
Legionella’s specific environmental monitoring is still very limited and does not reflect the complex interactions within biofilms and protozoa. Why, however, does this still happen? Why is research so reluctant to bridge this gap and start including protozoa and biofilms in standard
Legionella works? Do we have the tools and methods, but are they still not fully explored/understood? Or do we have to find new solutions for old problems? This dilemma is very well illustrated when the added value of online biofilm monitoring tools is compared with their effective use.
32.5. Online Biofilm Monitoring—An Unmet Need or an Unexplored Solution?
Online, continuous, non-destructive biofilm (and other deposits) monitoring appears as an important tool to assess, and prevent in a timely manner, build-up/detachment events, as well as to evaluate the efficacy of the applied countermeasures
[79].
The works of Janknecht and Melo
[80], Flemming
[34], and Nivens et al.
[81] provide interesting insights into biofilm monitoring approaches, discussing available techniques, their physical principles, and their advantages and disadvantages. Among the extended list of technologies reported in the literature, several are suitable for online monitoring in industrial systems
[80]. Furthermore, some of these state-of-the-art technologies have been successfully tested and are commercially available for implementation in real-field water systems
[82]. Despite the potentialities associated to each biofilm monitoring technique and their contribution to improved early-warning biofouling management, the water treatment industry/sector does not seem to have a clear strategy for their adoption (authors’ personal experiences). This happens because interpreting the sensor’s output information is often complex, requires specialized know-how
[34], and becomes a serious barrier for their integration into the water system process. If integration in real-field systems fails, the monitoring potential for the water management program vanishes and it becomes just another setting that a system’s manager must supervise. This agrees with Flemming’s
[34] arguments that the industry is still not committed to the optimization and validation of such early-warning tools, which, as explained, require a long timeframe and interdisciplinary work for their validation. At the end of the day, legislation might impose the adoption of online biofilm sensors but, to do so, science must strengthen the arguments about the potentials of complementary surface monitoring, not only for biofilm management but also for legionellosis prevention. Thus, following for example the conclusions of Kuiper et al.
[62], if the biofilm is under continuous supervision and control, legionellosis prevention increases.
Reflecting on the questions previously enunciated, we might conclude that the tools are there and they have intrinsic potential, but academia and industry are not able to coordinately collaborate and fully demonstrate their added value. Following this rationale, the next section will discuss some ideas on how to build an integrated approach that allows a complementary study of
Legionella ecology in real-field systems, which can be optimized and used in the future to enhance prevention in engineered water systems.
43. New Pathways to Build an Integrated and Effective Legionella Surveillance Strategy in Water Systems
Effective
Legionella management needs to be an integrated process
[23], adaptable to changes and grounded in consistent information about the water treatment critical issues. This process is conceived as a direct call to ‘keep an eye at the whole picture’, rather than just to ‘be focused on isolated pieces of the puzzle’. To meet the ambitious goal of building more integrated
Legionella prevention practices, a paradigm shift is needed. As previously discussed, the intricate level of interactions among
Legionella and the vast community of microorganisms in the bulk water and in the biofilm is scientifically very challenging and requires a ‘greater focus on total system ecology rather than on individual bacterial-protozoan interactions’
[64]. Some other authors
[8][20] emphasize that improvements in legionellosis mitigation practices at engineered systems are very dependent on a broader understanding of legionellae ecology.
43.1. An Integrated Monitoring Physical Model for Legionella Study and Control in Real Systems
One feasible approach to gain this knowledge, while tracking operational features of the systems, is the combination of complementary monitoring methods, which include (a) online, continuous information and discrete sensing; (b) surface and water monitoring; (c) biofilm and
Legionella analysis. Even though the development of such an idea can follow different pathways and certainly requires wider scientific reflection/discussion, we propose, for illustration purposes, an integrated monitoring model for
Legionella study at field-based systems (Figure 1). This model aims to catalyze a joint discussion on a renewed
Legionella management strategy, which can be optimized under the scope of field studies for later adoption at water utilities. Here, we will only focus on the macro perspective of the model rather than on overviewing specific methodologies, since those will depend on several items, including the sort of water system under study.
Figure 1. Integrated monitoring conceptual model for Legionella study and control in field-based systems. The model proposes four complementary sets of information: water (1 and 2) and biofilm (3 and 4) monitoring, discretely sampled (1 and 4) and continuously measured (2 and 3). Continuous information will enhance pro-active control and surveillance, based on early-warning information, while discrete information will allow to gain more specific information about Legionella ecology.
The conceptual model proposed in Figure 1 relies on the idea that
Legionella control will be as effective as we manage to gain a broader perspective on the overall ecology of
Legionella. Surveillance and pro-active control driven with online, continuous measurements are essential for effective
Legionella mitigation practices, and specific information is key for enhancing understanding about
Legionella overall ecology. Under these assumptions, four complementary sets of information were foreseen.
The first set of information is related to the routine monitoring approach, focused on periodic water sampling for physical, chemical, and microbiological characterization. This also includes
Legionella spp. and
L. pneumophila detection and quantification. Recently, Walker et al.
[83] reviewed current
Legionella testing methods, and LeChevallier
[84] proposed an interesting guidancefor the development of a
L. pneumophila monitoring plan for water utilities. Both works are of great importance to the implementation of improved routine
Legionella monitoring procedures. Furthermore, given the role of protozoa in the overall
Legionella ecology and virulence
[47], it seems to be very important to include their analysis under this first level of monitoring. This also embraces with the findings of Shaheen et al.
[85], who suggest that monitoring free-living amoebae can be useful to predict the ‘possible imminent high occurrence of
Legionella’ in engineered water systems. Protozoa are not detected through traditional bacteriological methods, and the detection of a large diversity of free-living protozoa can be a challenging and laborious task
[58]. This is demonstrated, for example, in the work of Valster et al.
[86], who found that different protozoan communities developed in duplicated samples (samples from different water settings). Nisar et al.
[87] discuss the relevance of molecular techniques such as PCR and fluorescence in situ hybridization (FISH) for
Legionella and protozoan screening in environmental water samples. In this work, the authors also came across the conclusion that, in potable water systems (including hospitals),
Vermamoeba and
Acanthamoeba were the hosts predominantly associated with
L. pneumophila. This also raises the possibility of selecting some specific protozoa indicators that might be linked to
L. pneumophila. For example, the review conducted by Lau et al.
[30] might be a great starting point for this discussion, since it systematizes the protozoa species (mostly amoebae) found to host
Legionella species in drinking water settings.
The second set of information is related to standard water treatment parameters that will directly or indirectly reflect the performance of the control measures
[88], including, for example, pH, conductivity, temperature, flow, critical pumps operation, and biocidal residue (if applicable). This also aligns with the WHO (World Health Organization) guidelines
[23], which state that ‘operationally, control measures, (…) should be monitored online’. The need to reinforce operational monitoring is also stressed in the recently revised European Directive (2020/2181) on the quality of water for human consumption
[26]. An online, real-time dataset of these parameters enables the timely identification and correction of punctual deviations to the established operational limits
[23][26][89], avoiding situations that can favor
Legionella proliferation. For example, Whiley et al.
[90] reported real-time monitoring of the temperature and flow in the thermostatic mixing valves of water distribution networks as an interesting surveillance strategy to detect changes in water quality, as well as to identify hazardous situations regarding different opportunistic pathogens, including
Legionella. This continuous information would be an important complement to well-established water routine sampling, as discussed in previous sections since it raises the opportunity to keep continuously an eye on the system in between samplings and while microbiological analysis is being processed. This information would also serve for registration purposes (an essential asset of a proper
Legionella prevention plan)
[24].
As formerly discussed, the potential of online, continuous, non-destructive biofilm monitoring can be determinant to establish a proactive, informated-based water management
[34]. Flemming
[34] systematized the features of an ideal online, real-time biofouling monitoring sensor able to provide information about the biofilm: location and extent, quantity (mass, thickness), nature of the deposit (organic/inorganic, biological/non-biological, chemical composition), the kinetics of deposit formation, and removal. Additionally, such monitoring tools should be applied to a large monitoring area and should be low cost and easy to handle. Due to this long and very specific list of features, it is very unlikely that a unique sensor meets all these requirements at once. As such, combining different monitoring tools into an ‘all-in-one’ solution is probably the most feasible way to strengthen the arguments for their routine implementation. This ‘all-in-one’ setup should combine a selection of tools that are suitable for real-field operation and that provide distinct (but complementary) output information about biofilm deposits.
Regarding
Legionella prevention, it seems plausible to accept that both the biofouling extent and nature (biotic/abiotic) of the attached layers are important parameters to assess. Measuring biofilm build-up/removal kinetics can provide important insights on ‘how fast is the biofilm being formed/removed’ and ‘how far will the stabilization plateau be achieved’. This concept is somewhat similar to the ‘Biofouling Formation Potential’ described by van der Kooji et al.
[61], yet applied to a different measuring unit. Those two indicators (kinetics and maximum biofilm amount) will provide information about the biofilm formation potential of the system and the biofilm extent, respectively. Both the ‘stabilization plateau’ and ‘threshold of interference’
[91], as well as biofilm kinetics, depend on the particular water system and its specific operating conditions
[34]. As such, for a given system, at a given representative location, an increased build-up rate or an unexpected sloughing-off event (which can bring
Legionella back into the bulk water) are certainly examples of early-warning calls that something in the standard operation has changed (even though that can be a planned change). Similarly, removal rates can be used to assess the efficacy of implemented countermeasures. For example, Pereira et al.
[92] reported the use of a surface sensor technology
[93] to monitor in real-time the formation/removal of biofouling layers, identifying proactively processual changes in the bypass of a cooling water system.
Evaluating the nature (biotic/abiotic) of the biofilm layer can be important for assessing and adjusting the efficacy of microbial control programs
[79], with the aim of keeping microbial growth at the surface under control. For example, the commercially available Alvim sensor
[94]—an online, electrochemical sensor—was successfully used in industrial water settings to follow the biofilm growth and to optimize cleaning procedures. Monitoring the nature of the deposit will be particularly relevant in finding out how biotic and/or abiotic attached layers affect
Legionella persistence. Another promising tool is the OnGuard
TM analyzer, which has been successfully used to optimize the biocidal program of a cooling water system, based on the detection of biofouling formation/removal kinetics
[95]. This analyzer can also provide information about the nature of the attached deposit
[95].
To gain detailed information that can enhance
Legionella ecology understanding, surface online monitoring must be complemented with biofilm discrete sampling, followed by a detailed analysis and characterization, including
Legionella screening. For that, the inclusion of biofilm sampling probes (or coupons), which can be periodically removed over time, might be a suitable approach. Some overviews on biofilm formation devices suitable for industrial application can be found in the works of
[96] or
[97], for example. Some interesting solutions for biofilm formation studies are the Flow Cell system
[93][98][99] or the Modified Robbins Devices
[100], which are very well characterized in the laboratory in terms of operation and hydrodynamics and have been successfully used in the study of biofilms in full-scale water systems.
The work of Azeredo et al.
[32] is a good starting point to choose which analytical techniques for biofilm characterization best fits a study’s purposes. Apart from the standard methods focused on biofilm physiology and the composition of the attached layers, we emphasize the role that structural characterization plays in the control of
Legionella. Several arguments support this suggestion: (a) protozoa have a significant impact on ‘shaping’ biofilm architectures
[58], (b) biofilm structure affects the efficacy of countermeasures
[101], (c) sloughing-off events are more likely to occur when heterogeneity increases
[102]. As such, evaluating structural changes in real-field systems can inform on biofilm and protozoa interactions, with a visible influence on
Legionella control.
43.2. Representativeness—Worst Case Scenario Conditions
A critical issue in the implementation of the conceptual model proposed herein is representativeness since most of the key points regarding biofilm build-up and
Legionella settlement are not accessible for sensor installation or sample collection. Engineering a bypass monitoring platform, combining the different monitoring sets of information, and operating under worst-case scenario conditions, can overcome this representativeness limitation. Worst-case conditions are accepted as part of
Legionella monitoring plans, in case it is impossible to overcome physical or processual limitations
[24]. For example, it is recommended that routine water sampling might be collected at the time (for example, before biocide dosage) and place (warmer temperatures) that represent the highest risk for
Legionella settlement in the system
[24]. The idea of a bypass monitoring platform relies on the assumption that if the water treatment favors (or not) biofilm formation/removal and
Legionella settlement, it will preferentially occur and be detected at the monitoring platform. As such, properly testing the worst-case conditions becomes a crucial step. Since both biofilms and
Legionella are affected by, for example, hydrodynamics, temperatures, and surface materials
[23][31], these parameters can be carefully chosen and set at the bypass monitoring platform to mimic the critical spots of the main system.
The complexity of this conceptual monitoring model demands a wise balance between a ‘perfect monitoring solution’ and a fit-to-purpose, real-field implementableone. The definition of consistent data flows (of process and biofilm indicators), and the ability to transform such data into meaningful information, can be a decisive step towards a successful approach. This would meet the expectation drawn by Fields et al.
[22], for example, that ‘Computer-based reporting systems may one day provide a means of conducting timely surveillance’. A final real-field implementable solution will have to bridge the gap between the approach (what should be done) and implementation (what can actually be done).
43.3. Final Disclaimer
While the ideas discussed in this final section might sound very exploratory, they aim to bring together existing tools and new elements to the discussion and studies around
Legionella management in man-made water systems. The conceptual monitoring model proposed in Figure 1 aims to encourage the strengthening of
Legionella monitoring procedures by integrating different approaches that can provide a broad perspective on
Legionella ecology and improve its surveillance in water systems. This model is especially important in the framework of real-field studies discussed in
Section 2, which are a great opportunity to bridge knowledge across disciplines while reinforcing scientific outputs towards new standardized and integrated methodologies. Integrated data monitoring and analysis, which can provide early-warning information, will certainly build more resilient real-field
Legionella control practices and strengthen field-based scientific outputs.
54. Conclusions
Legionella control at water systems is a multivariable problem. It is unfeasible to assume that Legionella might be eradicated from water systems; therefore, prevention assumes great relevancy. Field-based trials are an important component of Legionella study. However, these studies are traditionally focused on assessing Legionella ecology in the bulk water, often disregarding the role of protozoa and biofilms as critical ecological niches for Legionella growth, infectivity, and perseverance in water systems. Improved, consistent, and adaptable-to-change Legionella management procedures require a great focus on the total ecology of the system and a wider convergence between engineering tools and microbiological approaches. To boost this discussion, an integrated monitoring model for Legionella study and control at field-based systems is proposed here. This model is grounded in the combination of four complementary sets of information and is expected to bridge the gap between scientific approaches and real-field needs, so as to enhance Legionella understanding and pro-active surveillance in the water systems.