Comparison of Particle-Associated Bacteria from a Drinking Water Treatment Plant and Distribution Reservoirs with Different Water Sources.

This study assessed the characteristics of and changes in the suspended particles and the associated bacteria in an unchlorinated drinking water distribution system and its reservoirs with different water sources. The results show that particle-associated bacteria (PAB) were present at a level of 0.8-4.5 × 10(3) cells ml(-1) with a biological activity of 0.01-0.04 ng l(-1) ATP. Different PAB communities in the waters produced from different sources were revealed by a 16S rRNA-based pyrosequencing analysis. The quantified biomass underestimation due to the multiple cells attached per particle was ≥ 85%. The distribution of the biologically stable water increased the number of cells per particle (from 48 to 90) but had minor effects on the PAB community. Significant changes were observed at the mixing reservoir. Our results show the characteristics of and changes in suspended PAB during distribution, and highlight the significance of suspended PAB in the distribution system, because suspended PAB can lead to a considerable underestimation of biomass, and because they exist as biofilm, which has a greater mobility than pipe-wall biofilm and therefore presents a greater risk, given the higher probability that it will reach the customers' taps and be ingested.

Sensitivity of quantitative microbial risk assessments to assumptions about exposure to multiple consumption events per day.

Quantitative microbial risk assessments (QMRAs) of contaminated drinking water usually assume the daily intake volume is consumed once a day. However, individuals could consume water at multiple time points over 1 day, so the objective was to determine if the number of consumption events per day impacted the risk of infection from Campylobacter jejuni during short-term contamination events. A probabilistic hydraulic and risk model was used to evaluate the impact of multiple consumption events as compared to one consumption event on the health risk from the intake of contaminated tap water. The fraction of the population that experiences greater than 10(-4) risk of infection per event at the median dose was 6.8% (5th-95th percentile: 6.5-7.2%) for one consumption event per day, 18.2% (5th-95th: 17.6-18.7%) for three consumption events per day, and 19.8% (5th-95th: 14.0-24.4%) when the number of consumption events varied around 3.49 events/day. While the daily intake volume remained consistent across scenarios, the results suggest that multiple consumption events per day increases the probability of infection during short-term, high level contamination events due to the increased coincidence of a consumption event during the contamination peak. Therefore, it will be important to accurately characterize this parameter in drinking water QMRAs.

Pyrosequencing reveals bacterial communities in unchlorinated drinking water distribution system: an integral study of bulk water, suspended solids, loose deposits, and pipe wall biofilm.

The current understanding of drinking water distribution system (DWDS) microbiology is limited to pipe wall biofilm and bulk water; the contributions of particle-associated bacteria (from suspended solids and loose deposits) have long been neglected. Analyzing the composition and correlation of bacterial communities from different phases helped us to locate where most of the bacteria are and understand the interactions among these phases. In the present study, the bacteria from four critical phases of an unchlorinated DWDS, including bulk water, pipe wall biofilm, suspended solids, and loose deposits, were quantified and identified by adenosine triphosphate analysis and pyrosequencing, respectively. The results showed that the bulk water bacteria (including the contribution of suspended solids) contributed less than 2% of the total bacteria. The bacteria associated with loose deposits and pipe wall biofilm that accumulated in the DWDS accounted for over 98% of the total bacteria, and the contributions of bacteria in loose deposits and pipe wall biofilm were comparable. Depending on the amount of loose deposits, its contribution can be 7-fold higher than the pipe wall biofilm. Pyrosequencing revealed relatively stable bacterial communities in bulk water, pipe wall biofilm, and suspended solids throughout the distribution system; however, the communities present in loose deposits were dependent on the amount of loose deposits locally. Bacteria within the phases of suspended solids, loose deposits, and pipe wall biofilm were similar in phylogenetic composition. The bulk water bacteria (dominated by Polaromonas spp.) were clearly different from the bacteria from the other three phases (dominated by Sphingomonas spp.). This study highlighted that the integral DWDS ecology should include contributions from all of the four phases, especially the bacteria harbored by loose deposits. The accumulation of loose deposits and the aging process create variable microenvironments inside loose deposits structures for bacteria to grow. Moreover, loose deposits protect the associated bacteria from disinfectants, and due to their mobility, the associated bacteria reach taps easily.

Exploring global Cryptosporidium emissions to surface water.

The protozoan parasite Cryptosporidium is a major cause of diarrhoea worldwide. This paper presents the first model-based inventory with 0.5 by 0.5 degree resolution of global Cryptosporidium emissions for the year 2000 from humans and animals to surface water. The model is based on nutrient distribution modelling, because the sources and transport of oocysts and nutrients to the surface water are comparable. Total emissions consist of point source emissions from wastewater and nonpoint source emissions by runoff of oocysts in manure from agricultural lands. Results indicate a global emission of 3 × 10(17) oocysts per year, with comparable contributions from point and nonpoint sources. Hot-spot areas for point sources are big cities in China, India and Latin America, while the area with the largest nonpoint source emissions is in China. Uncertainties in the model are large. Main areas for further study are (i) excretion rates of oocysts by humans and animals, (ii) emissions of humans not connected to sewage systems, and (iii) retention of oocysts to determine surface water pathogen concentrations rather than emissions. Our results are useful to health organisations to identify priority areas for further study and intervention.

Inactivation of bacteriophage MS2 upon exposure to very low concentrations of chlorine dioxide.

This study investigates the effects of very low concentrations of ClO(2) applied in drinking water practice on the inactivation of bacteriophage MS2. Concentrations of 0.5 mg/L, 0.1 mg/L and 0.02 mg/L ClO(2) inactivated at least 5 log units of MS2 after an exposure time of approximately 20, 50 and 300 min respectively. When the ClO(2) concentration was as low as 0.005 mg/L, inactivation of 1 log unit MS2 was observed after 300 min exposure. Increasing the contact time to 24 h did not increase the inactivation any further. Non-linear inactivation kinetics (tailing) were observed for all conditions tested. Repeated addition of MS2 to the reactor showed that tailing was not caused by a reduction of the biocidal effect of ClO(2) during disinfection. The Modified Chick-Watson, the Efficiency Factor Hom (EFH) model and the Modified Cerf model, a modification of the two-fraction Cerf model, were fitted to the non-linear inactivation curves. Both the EFH and the modified Cerf model did fit accurately to the inactivation data of all experiments. The good fit of the Modified Cerf model supports the hypothesis of the presence of two subpopulations. Our study showed that ClO(2) is an effective disinfectant against model organism MS2, also at the low concentrations applied in water treatment practice. The inactivation kinetics followed a biphasic pattern due to the presence of a more ClO(2)-resistant subpopulation of MS2 phages, either caused by population heterogeneity or aggregation/adhesion of MS2.

Practical applications of quantitative microbial risk assessment (QMRA) for water safety plans.

The absence of indicator organisms in drinking water does not provide sufficient guarantee for microbial safety. Therefore the water utilities are implementing water safety plans (WSP) to safeguard drinking water quality. Quantitative microbial risk assessment (QMRA) can be used to provide objective quantitative input for the WSP. This study presents several applications of treatment modelling in QMRA to answer the risk managers questions raised in the WSP. QMRA can estimate how safe the water is, how much the safety varies and how certain the estimate of safety is. This can be used in the WSP system assessment to determine whether treatment is meeting health-based targets with the required level of certainty. Quantitative data analysis showed that short events of only 8 hours per year can dominate the yearly average health risk for the consumer. QMRA also helps the design of physical and microbial monitoring. The study showed that the required monitoring frequency increases with increasing treatment efficacy. Daily monitoring can be sufficient to verify a treatment process achieving 2 log reduction of pathogens, but a process achieving 4 log reduction needs to be monitored every 15 minutes. Similarly, QMRA helps to prepare adequate corrective actions by determining the acceptable 'down time' of a process. For example, for a process achieving 2.5 log reduction a down time of maximum 6 hours per year is acceptable. These applications illustrate how QMRA can contribute to efficient and effective management of microbial drinking water safety.

GAC adsorption filters as barriers for viruses, bacteria and protozoan (oo)cysts in water treatment.

Granular Activated Carbon (GAC) adsorption filtration is commonly used in drinking water treatment to remove NOM and micro-pollutants and on base of the process conditions a certain capacity to eliminate pathogenic micro-organisms was expected. The experiences with the mandatory quantitative microbial risk assessment of Dutch drinking water revealed a lack of knowledge on the elimination capacity of this process for pathogens. The objective of the current study was to determine the capacity of GAC filtration to remove MS2, Escherichia coli and spores of Clostridium bifermentans as process indicators for pathogens and more directly of (oo)cysts of Cryptosporidium parvum and Giardia lamblia. Challenge tests with fresh and loaded GAC were performed in pilot plant GAC filters supplied with pre-treated surface water at a contact time which was half of the contact time of the full-scale GAC filters. MS2 phages were not removed and the removal of E. coli and the anaerobic spores was limited ranging from < or =0.1-1.1 log. The (oo)cysts of C. parvum and G. lamblia, however, were removed significantly (1.3-2.7 log). On base of the results of the experiments and the filtration conditions the removal of the indicator bacteria and (oo)cysts was largely attributed to attachment. The model of the Colloid Filtration Theory was used to describe the removal of the dosed biocolloids in the GAC filters, but the results demonstrated that there is a lack of quantitative knowledge about the influence of collector characteristics on the two major CFT parameters, the single collector and the sticking efficiency.

Improved methods for modelling drinking water treatment in quantitative microbial risk assessment; a case study of Campylobacter reduction by filtration and ozonation.

Quantitative microbial risk assessment (QMRA) is increasingly applied to estimate drinking water safety. In QMRA the risk of infection is calculated from pathogen concentrations in drinking water, water consumption and dose response relations. Pathogen concentrations in drinking water are generally low and monitoring provides little information for QMRA. Therefore pathogen concentrations are monitored in the raw water and reduction of pathogens by treatment is modelled stochastically with Monte Carlo simulations. The method was tested in a case study with Campylobacter monitoring data of rapid sand filtration and ozonation processes. This study showed that the currently applied method did not predict the monitoring data used for validation. Consequently the risk of infection was over estimated by one order of magnitude. An improved method for model validation was developed. It combines non-parametric bootstrapping with statistical extrapolation to rare events. Evaluation of the treatment model was improved by presenting monitoring data and modelling results in CCDF graphs, which focus on the occurrence of rare events. Apart from calculating the yearly average risk of infection, the model results were presented in FN curves. This allowed for evaluation of both the distribution of risk and the uncertainty associated with the assessment.

How can the UK statutory Cryptosporidium monitoring be used for Quantitative Risk Assessment of Cryptosporidium in drinking water?

Quantitative Microbiological Risk Assessment (QMRA) is increasingly being used to complement traditional verification of drinking water safety through the absence of indicator bacteria. However, the full benefit of QMRA is often not achieved because of a lack of appropriate data on the fate and behaviour of pathogens. In the UK, statutory monitoring for Cryptosporidium has provided a unique dataset of pathogens directly measured in large volumes of treated drinking water. Using this data a QMRA was performed to determine the benefits and limitations of such state-of-the-art monitoring for risk assessment. Estimates of the risk of infection at the 216 assessed treatment sites ranged from 10(-6.5) to 10(-2.5) person(-1) d(-1). In addition, Cryptosporidium monitoring data in source water was collected at eight treatment sites to determine how Cryptosporidium removal could be quantified for QMRA purposes. Cryptosporidium removal varied from 1.8 to 5.2 log units and appeared to be related to source water Cryptosporidium concentration. Application of general removal credits can either over- or underestimate Cryptosporidium removal by full-scale sedimentation and filtration. State-of-the-art pathogen monitoring can identify poorly performing systems, although it is ineffective to verify drinking water safety to the level of 10(-4) infections person(-1) yr(-1).

Estimation of the consumption of cold tap water for microbiological risk assessment: an overview of studies and statistical analysis of data.

The volume of cold tap water consumed is an essential element in quantitative microbial risk assessment. This paper presents a review of tap water consumption studies. Study designs were evaluated and statistical distributions were fitted to water consumption data from The Netherlands, Great Britain, Germany and Australia. We conclude that the diary is to be preferred for collecting water consumption data. If a diary is not feasible, a 24 h recall would be the best alternative, preferably repeated at least once. From the studies evaluated, the mean daily consumption varies from 0.10 L to 1.55 L. No conclusions could be drawn regarding the effects of season, age and gender on tap water consumption. Physical activity, yearly income and perceived health status were reported to influence water consumption. Comparison of the different statistical probability distribution functions of the datasets demonstrated that the Poisson distribution performed better than the lognormal distribution as suggested by Roseberry and Burmaster. For quantitative microbiological risk assessment (QMRA) it is recommended to use country-specific consumption data and statistical distributions, if available. If no country specific data are available we recommend to use the Australian distribution data from the Melbourne diary study (Poisson, lambda=3.49 glasses/d) as a conservative estimate.

Combined use of microbiological and non-microbiological data to assess treatment efficacy.

The treatment efficacy for reducing Campylobacter concentrations by a drinking water treatment plant was assessed using a stochastic Monte Carlo model. The goal of the study was to reduce uncertainty of the results by combining microbiological and non-microbiological data in an advanced treatment assessment. Combining raw water Campylobacter and E. coli data reduced the uncertainty on raw water (peak) concentrations five-fold. Similar improvement was achieved for rapid sand filtration. Ozone disinfection was modelled based on ozone concentrations, contact time and temperature. Since this data was available, whereas most microbiological analyses at this point were negative, uncertainty was reduced three-fold. The slow sand filtration assessment could not be improved; however, since previous steps contained less uncertainty, this did not increase uncertainty by much. The study showed that using appropriate data for each treatment step can greatly reduce uncertainty in treatment assessment.

Modelling the length of microbiological protection zones around phreatic sandy aquifers in The Netherlands.

The aim of the current study was to calculate the size of protection zones around (sub)oxic and anoxic sandy aquifers without confining layers using a virus infection and transport model. The maximum allowable virus infection risk was 10(-4)/person/year at the 95% confidence level. Model results demonstrated that phreatic (sub)oxic sandy aquifers in The Netherlands required protection areas with a residence time of 43-117 d to ensure that the maximum virus infection risk would not be exceeded. This was 0.7-2 x the current guideline of 60d. In contrast, phreatic anoxic sandy aquifers without confining layers needed protection zones of 555-898d to stay below the maximum virus infection risk, 9.5-15 x the current guideline. A sensitivity analysis of the model demonstrated that the calculated protection zone was most sensitive for virus inactivation rate and collision efficiency. Values of both parameters were predicted from values obtained from previously published field and laboratory studies. At present, as it is unknown if these values can also be used at other locations, model results should be interpreted with care.

Inactivation of Escherichia coli by ozone under bench-scale plug flow and full-scale hydraulic conditions.

To determine the disinfection efficacy of ozonation, water companies can apply several disinfection calculation methods. The goal of this study was to evaluate the use of the T10 and continuous stirred tank reactor (CSTR) method to extrapolate inactivation rates of ozone sensitive microorganisms observed in laboratory tests to full-scale ozonation in drinking water treatment. The inactivation efficacy of the ozonation at the Amsterdam water treatment works was assessed by determining Escherichia coli concentrations in large volume samples before and after ozonation over a period of 1 year. The inactivation of dosed E. coli WR1 was tested in a bench-scale dissolved ozone plug flow reactor (DOPFR) on the same feed water as the full-scale ozonation in which a concentrated ozone solution in Milli-Q water was dosed. Applying the T10 method on the inactivation rates observed in the DOPFR strongly overestimated the inactivation capacity of the full-scale ozonation. The expected inactivation based on the CSTR method (LT2ESWTR) approached the observed inactivation at full-scale. Therefore, the CSTR method should be preferred to calculate inactivation of ozone sensitive organisms such as E. coli, viruses, Giardia and Campylobacter by full-scale ozonation.

Inactivation credit of UV radiation for viruses, bacteria and protozoan (oo)cysts in water: a review.

UV disinfection technology is of growing interest in the water industry since it was demonstrated that UV radiation is very effective against (oo)cysts of Cryptosporidium and Giardia, two pathogenic micro-organisms of major importance for the safety of drinking water. Quantitative Microbial Risk Assessment, the new concept for microbial safety of drinking water and wastewater, requires quantitative data of the inactivation or removal of pathogenic micro-organisms by water treatment processes. The objective of this study was to review the literature on UV disinfection and extract quantitative information about the relation between the inactivation of micro-organisms and the applied UV fluence. The quality of the available studies was evaluated and only high-quality studies were incorporated in the analysis of the inactivation kinetics. The results show that UV is effective against all waterborne pathogens. The inactivation of micro-organisms by UV could be described with first-order kinetics using fluence-inactivation data from laboratory studies in collimated beam tests. No inactivation at low fluences (offset) and/or no further increase of inactivation at higher fluences (tailing) was observed for some micro-organisms. Where observed, these were included in the description of the inactivation kinetics, even though the cause of tailing is still a matter of debate. The parameters that were used to describe inactivation are the inactivation rate constant k (cm(2)/mJ), the maximum inactivation demonstrated and (only for bacterial spores and Acanthamoeba) the offset value. These parameters were the basis for the calculation of the microbial inactivation credit (MIC="log-credits") that can be assigned to a certain UV fluence. The most UV-resistant organisms are viruses, specifically Adenoviruses, and bacterial spores. The protozoon Acanthamoeba is also highly UV resistant. Bacteria and (oo)cysts of Cryptosporidium and Giardia are more susceptible with a fluence requirement of <20 mJ/cm(2) for an MIC of 3 log. Several studies have reported an increased UV resistance of environmental bacteria and bacterial spores, compared to lab-grown strains. This means that higher UV fluences are required to obtain the same level of inactivation. Hence, for bacteria and spores, a correction factor of 2 and 4 was included in the MIC calculation, respectively, whereas some wastewater studies suggest that a correction of a factor of 7 is needed under these conditions. For phages and viruses this phenomenon appears to be of little significance and for protozoan (oo)cysts this aspect needs further investigation. Correction of the required fluence for DNA repair is considered unnecessary under the conditions of drinking water practice (no photo-repair, dark repair insignificant, esp. at higher (60 mJ/cm(2)) fluences) and probably also wastewater practice (photo-repair limited by light absorption). To enable accurate assessment of the effective fluence in continuous flow UV systems in water treatment practice, biodosimetry is still essential, although the use of computational fluid dynamics (CFD) improves the description of reactor hydraulics and fluence distribution. For UV systems that are primarily dedicated to inactivate the more sensitive pathogens (Cryptosporidium, Giardia, pathogenic bacteria), additional model organisms are needed to serve as biodosimeter.

Elimination of viruses, bacteria and protozoan oocysts by slow sand filtration.

The decimal elimination capacity (DEC) of slow sand filters (SSF) for viruses, bacteria and oocysts of Cryptosporidium has been assessed from full-scale data and pilot plant and laboratory experiments. DEC for viruses calculated from experimental data with MS2-bacteriophages in the pilot plant filters was 1.5-2 log10. E. coli and thermotolerant coliforms (Coli44) were removed at full-scale and in the pilot plant with 2-3 log10. At full-scale, Campylobacter bacteria removal was 1 log10 more than removal of Coli44, which indicated that Coli44 was a conservative surrogate for these pathogenic bacteria. Laboratory experiments with sand columns showed 2-3 and >5-6 log10 removal of spiked spores of sulphite-reducing clostridia (SSRC; C. perfringens) and oocysts of Cryptosporidium respectively. Consequently, SSRC was not a good surrogate to quantify oocyst removal by SSF. Removal of indigenous SSRC by full-scale filters was less efficient than observed in the laboratory columns, probably due to continuous loading of these filter beds with spores, accumulation and retarded transport. It remains to be investigated if this also applies to oocyst removal by SSF. The results additionally showed that the schmutzdecke and accumulation of (in)organic charged compounds in the sand increased the elimination of microorganisms. Removal of the schmutzdecke reduced DEC for bacteria by +/-2 log10, but did not affect removal of phages. This clearly indicated that, besides biological activity, both straining and adsorption were important removal mechanisms in the filter bed for microorganisms larger than viruses.

Quantitative risk assessment of Cryptosporidium in surface water treatment.

Quantitative microbiological risk assessment requires quantitative data to assess consumer exposure to pathogens and the resulting health risk. The aim of this study was to evaluate data sets on the occurrence of Cryptosporidium oocysts in raw water and on the removal of model organisms (anaerobic spores, bacteriophages) to perform such a risk assessment. A tiered approach was used by first calculating approximate point estimates and when the point estimate was close to the required safety level (10(4) annual risk of infection), fitting the data to probability distributions and Monte Carlo analysis to calculate the distribution of the risk of infection. Sensitivity analysis showed that the variability in the Cryptosporidium data in raw water (largely introduced by the variability of the recovery efficiency of the detection method) determined most of the variance in the risk estimate.

Modelling the sewage discharge and dispersion of Cryptosporidium and Giardia in surface water.

Modelling the discharge of parasitic protozoa into surface water and the dispersion in rivers and streams gives insight into the contribution of the different sources of environmental contamination and in the transmission of these organisms from the point of discharge to drinking water abstraction points and bathing sites. We tested the applicability of emission (PROMISE) and dispersion (WATNAT) models developed for chemical pollutants to describe the environmental behaviour of Cryptosporidium and Giardia in the Netherlands. The annual load of Cryptosporidium and Giardia in domestic wastewater was 3.2 x 10(13) and 3.8 x 10(14) respectively. The majority (85%) of the Cryptosporidium oocysts was discharged with effluent of wastewater treatment plants. while the majority (82%) of the Giardia cysts was discharged with untreated wastewater discharges and sewer overflows. The estimated annual import through the river Rhine and Meuse was 3.2 x 1014 Cryptosporidium oocysts and 2.1 x 10(15) Giardia cysts, of which the river Rhine contributed 87 and 66%, respectively. This outweighed the total load of the discharges of treated and untreated wastewater in the Netherlands. The combination of PROMISE and WATNAT predicted concentrations of Cryptosporidium and Giardia in surface water that were in the same order of magnitude as the concentrations that were observed at 5 of the 6 sites compared. At a site with primarily agricultural contamination, the models predicted concentrations that were 1 10log-unit lower than the observed concentrations. This is a first step in the direction of a quantitative description of the transmission cycle of Cryptosporidum and Giardia through water. The use of these models combines observational occurrence data and experimental data from laboratory survival studies into a single integrated description. The description needs further improvement by incorporation of agricultural run-off and increasing the number and time frame of input monitoring data.

Multiple barriers for Cryptosporidium and Giardia

Se analizan los factores a tener en cuenta para el diseño y la evaluacion de un sistema de tratamiento de agua efectivo en la remocion de Cryptosporidium y Giardia. Las compañias de agua y el gobierno de Holanda han iniciado un programa de investigacion para determinar la probabilidad de transmision de esaots parasitos

Sedimentation of free and attached Cryptosporidium oocysts and Giardia cysts in water.

Experimental analysis of the sedimentation velocity of Cryptosporidium parvum oocysts and Giardia lamblia cysts was compared with mathematical description of their sedimentation velocities by using measurements of (oo)cyst size and density and the density and viscosity of the sedimentation medium to determine if the sedimentation kinetics of freely suspended oocysts of C. parvum and cysts of G. lamblia can be described by Stokes' law. The theoretically calculated sedimentation kinetics showed a good agreement with the experimentally observed kinetics. Both showed a decline in sedimentation velocity over time, caused primarily by variation in (oo)cyst density. The initial apparent sedimentation velocities in Hanks balanced salt solution at 23 degrees C was 0.35 micron . s-1 for oocysts and 1.4 micron . s-1 for cysts. (Oo)cysts that enter the surface water environment by discharges of biologically treated sewage may be attached to sewage particles, and this will affect their sedimentation kinetics. Therefore, (oo)cysts were mixed with settled secondary effluent. (Oo)cysts readily attached to the (biological) particles in effluent; 30% of both cysts and oocysts attached during the first minutes of mixing, and this fraction increased to approximately 75% after 24 h. The sedimentation velocity of (oo)cysts attached to secondary effluent particles increased with particle size and was (already in the smallest size fraction [1 to 40 micron]) determined by the sedimentation kinetics of the effluent particles. The observed sedimentation velocities of freely suspended (oo)cysts are probably too low to cause significant sedimentation in surface water or reservoirs. However, since a significant proportion of both cysts and oocysts attached readily to organic biological particles in secondary effluent, sedimentation of attached (oo)cysts after discharge into surface water will probably be a significant factor in the environmental ecology of C. parvum and G. lamblia. Attachment to particles influences not only sedimentation of (oo)cysts in surface water but also their behavior in drinking water treatment processes.

Salivary levels of immunoglobulin A in triathletes.

We investigated whether the physical exercise of an olympic distance triathlon affected the salivary IgA excretion of triathletes, as a biomarker for mucosal immune defence. 42 triathletes participated in the study. It was found that the salivary flow rate was decreased significantly after the race, thereby resulting in a significant reduction of the total salivary IgA output. The salivary IgA concentration (mg IgA/ml) did not differ, but expressed as total salivary protein, a significant reduction was observed. This was on the account of the salivary protein concentration (mg protein/ml), which was significantly increased. In contrast to the IgA secretion, the salivary amylase activity was increased significantly after the race. Therefore, our data suggest that the exercise of a triathlon may decrease the level of IgA-mediated immune protection at the mucosal surface. As triathletes may during the race be exposed to micro-organisms present in the swimming water, a decreased IgA-mediated immunity during the race may pose triathletes at an increased risk of infections.