- Email: [email protected]
An ATP-based method for monitoring the microbiological drinking water quality in a distribution network
ARTICLE IN PRESS
Water Research 37 (2003) 3689–3696
An ATP-based method for monitoring the microbiological drinking water quality in a distribution network E. Delahayea,*, B. Welte! b, Y. Levic, G. Leblond, A. Montielb a
SAGEP, Plate-forme de Recherche, 4 Avenue Pierre Mend!es France, 94340 Joinville le Pont, France b SAGEP, DQE, 9 rue Schoelcher, 75014 Paris, France c Laboratoire Sant!e Publique-Environnement, Facult!e de Pharmacie, Universit!e Paris Sud XI, France d Institut de G!en!etique et de Microbiologie, Universit!e Paris Sud XI, Orsay, France Received 31 May 2002; received in revised form 31 January 2003; accepted 14 April 2003
Abstract The titration of adenosine triphosphate (ATP) by bioluminescence permits rapid evaluation of the quantity of viable micro-organisms present in a water sample. During two sampling campaigns, Soci!et!e Anonyme de Gestion des Eaux de Paris (SAGEP) tested a new extraction and titration system of bacterial ATP in the Paris drinking water distribution network. As far as the entire set of results of analyses of water in the network is concerned there is a linear relationship between log [ATP] and log(HPC-R2A/ml). Furthermore, as regards the drinking water originating from treatment of surface waters, some of the results obtained indicate a slight change as regards the Paris network in the microbiological quality. This is certainly linked to the distance travelled from the production location as well as to a reservoir effect observed on a site. Conversely, no change is apparent with regard to waters of underground origin. Lastly, despite changes in temperature and chlorine residual, no significant influence has been observed, essentially because of the very low density of culturable bacteria. r 2003 Elsevier Science Ltd. All rights reserved. Keywords: Drinking water; ATP; Storage tank; Distribution network; Bacteria; Diagnosis
1. Introduction Despite the presence of disinfectant, bacterial enumerations are liable to increase in all drinking water distribution networks due to the presence of fixed and complex microbial ecosystems known as biofilms [1]. Biofilms are made up of micro-organisms, the majority of which are bacteria. The whole is included in a polymeric-type gel [2] secreted by the bacteria themselves and mineral elements. This enveloping structure can have reductive effects and inhibit the action of the disinfectants [3,4]. The presence of biofilms in distribution networks can result in numerous disagreeable problems for the user such as microbial contamination *Corresponding author. E-mail address: [email protected] (E. Delahaye).
of the water distributed, induced corrosion or the production of sapid compounds. Soci!et!e Anonyme de Gestion des Eaux de Paris (SAGEP) is the sole producer of drinking water for the City of Paris (France). Every day, this company abstracts, produces and transports 750,000 m3 of drinking water to meet the needs of the City of Paris. More precisely, 50% of the drinking water supply originates from underground waters captured within a radius of 80–150 km from Paris. These spring waters are conveyed to Paris by gravity over the aqueducts of the Avre (AA), the Loing (LA) and the Vanne (VA). The remaining 50% comes from surface waters treated by three water treatment works (WTW): Ivry (IP) and Orly (OP) on the river Seine, and Joinville le Pont (JP) on the river Marne. Each of these three WTWs possesses a maximum production capacity of 300,000 m3/day.
0043-1354/03/$ - see front matter r 2003 Elsevier Science Ltd. All rights reserved. doi:10.1016/S0043-1354(03)00288-4
ARTICLE IN PRESS 3690
E. Delahaye et al. / Water Research 37 (2003) 3689–3696
This study sets out to evaluate titration of ATP by bioluminescence for the rapid monitoring of the microbiological quality of the drinking water during its distribution in the Paris network.
2. Materials and methods 2.1. The Paris drinking water distribution network The tests were mainly conducted within the Paris distribution network. This is a highly complex gridded network in which the average water residence time is 6 h. Because of the complexity of the network it is impossible to clearly determine linear portions on which there could be a succession of sampling points at fully identified distances and residence times. Two sampling campaigns were carried out: *
*
the first (campaign 1) in the spring of 2001, a period in which spring waters and surface waters have similar temperatures, the second (campaign 2) in the autumn of 2001, when the waters are of different temperatures depending on their origins.
Each sampling campaign took place at some four dozen points (Fig. 1). All points sampled had facilities to allow the sample to be taken from the bulk flow. Some
of them are supplied by spring waters and others by potable waters produced from treated river waters. Depending on the instance, they are located at the entry to the network or, on the contrary, are at a distance from the production sites and on occasion after passing through one or two intermediate storage tanks. The nomenclature used to designate the points is as follows: *
*
*
the first two letters indicate the origin of the water: OP, JP and IP for the surface water treatment plants at Orly, Joinville and Ivry, respectively; AA and VA for the underground waters travelling across the Avre and Vanne aqueducts, respectively; VA/LA for the points supplied by a mixture of water between the underground waters of the Vanne and Loing aqueducts, the following figure indicates (as far as the complexity of the network allows) the position of the sampling point in relation to the production site (the plants are numbered 0) or of the place of arrival of the water in Paris (the aqueduct arrivals are numbered 0), finally, the last two letters specify, as regards each point, whether a storage tank (ST), or merely a sampling point (SP), is involved.
2.2. The ATP titration kit (Profile 1, New Horizons Diagnostics, imported into France by the D!evelopment Marketing Services company) The kit is made up of: * *
* *
*
Fig. 1. Localisation of the sampling points used for the two campaigns. (OP: supplied by Orly WTW, IP: supplied by Ivry WTW, JP: supplied by Joinville WTW, VA: supplied by Vanne aqueduct, VA/LA: supplied by Vanne aqueduct and Loing aqueduct, AA: supplied by Avre aqueduct, ST: storage tank, SP: sampling point).
a micro-luminometer (mains or battery supply), small doses of a freeze-dried mixture of luciferin/ luciferase and their reconstitution liquid, filtration devices (pore diameter 0.4 mm), a leaching reagent of the non-bacterial ATP and inhibiting substances (SRA reagent), a reagent ensuring extraction of the ATP of bacterial origin (BRA II reagent).
Stoichiometrically speaking, at the time of the reaction between the ATP and the luciferin–luciferase enzymatic system, one photon is produced for every one molecule of ATP consumed. This enables a linear relationship to be established between the quantity of light emitted and the quantity of ATP initially present. Indeed, the average ATP load of a bacterium is of the order of 7 mmol/g of dry weight for cells in exponential growth phase [5], or approximately 3.5 mg of ATP for 2.3 g wet weight that is to say 1 fg per bacterial cell if one considers that the wet weight of a cell is of the order of 6.7 1013 g. For water samples from the network after filtration (syringe fitted with a filtration device) of 10 ml of
ARTICLE IN PRESS E. Delahaye et al. / Water Research 37 (2003) 3689–3696
25000 20000
RLU
4.00 3.50 3.00 y = 0.97x + 0.61 R2 = 0.67
2.50 2.00 1.50 1.50
2.00
2.50 3.00 log (HPC-R2A / ml)
3.50
4.00
Fig. 3. Validation of the ATP titration kit on surface water samples (HPC-R2A vs. RLU).
previous testing on 57 surface water samples for validation purposes. In this case, for the line of regression (Fig. 3) between log [RLUs] and log (HPC-R2A/ml) the value of R2 is 0.67 (R ¼ 0:82). The statistical analysis of this Pearson’s product–moment correlation value proves that there is a highly significant linear relationship with a p-value of 7.99 1015 and a 95% confidence interval for R of 0.71–0.89. 2.3. Temperature and free chlorine Measurement of these parameters was carried out in the field at the time the samples were taken. The chlorine was dosed in accordance with the N,N-Diethylphenylene-1.4 diamine titrimetric French standard (AFNOR NF T 90-037). 2.4. Heterotrophic plate counts on R2A agar (HPC-R2A) The R2A agar (DIFCO) was used in order to estimate the density of culturable bacteria in the network water samples. This permits the growth of stressed bacteria which do not develop in other mediums [7]. For each sampling, two seedings were carried out. Incubation lasted 7 days at 22 C. The tables indicate that the results correspond to the average of the two seeded plates. 2.5. Heterotrophic plate counts on PCA agar (HPC-PCA 22 C and 37 C)
y = 182.43x - 482.33 R2 = 0.99
15000
4.50
log ( RLU)
sample, the filtration device was rinsed twice in succession (double washing recommended by the manufacturer for chlorinated samples) using 4 drops of SRA reagent, with, on each occasion, evacuation of the reagent from the filtration device by application of air pressure. After introduction of the filtration device into the luminometer, two drops of BRA II reagent were added immediately followed by 50 ml of the previously reconstituted luciferin/luciferase mixture (kept in ice). After rapid homogenisation of the whole, several successive measurements were carried out until a maximum was obtained in terms of relative luminescence units (RLUs). The time required for successful titration was about 2 min. Campaigns 1 and 2 having been conducted using different micro-luminometers. Standard curves were established prior to each campaign on standard solutions of ATP, 100 ml of which were directly deposited in a filtration device before addition of 50 ml of the reagent containing the luciferin–luciferase mixture. This made it possible to express the results in quantities of ATP rather than in RLU (a factor that may depend on the apparatus used). An error calculation concerning the value of the slope grade was made for each of the two linear straight regression lines [6] at a confidence coefficient of 95%. Fig. 2 presents the results in URL achieved for different concentrations of standard ATP solutions prior to campaigns 1 and 2. In both cases, the measurements carried out on the entire range covered by the luminometer (0 to 20000 URL) showed high correlation (R2 > 0:99) with the ATP concentrations. For campaign 1, consumption of a picogram of ATP corresponds to 158 (712) URL while the conversion factor used for campaign 2 is 182 (720) URL per picogram of ATP. When account is taken of the errors on the slope grades, it becomes apparent that the conversion factors used between one campaign and the next are not significantly different. Furthermore, before they are put to use in the distribution network, the equipment and the reagents kit designed for titration of the ATP have undergone
3691
y = 158.72x + 172.09 R2 = 0.99
For each sampling and incubation temperature, the analyses were carried out in duplicate in accordance with the NFEN ISO 62 22 standard.
10000 5000
campaign 1
campaign 2
0 0
20
40
60 80 100 ATP (pg / 100 µl)
120
2.6. Coliforms, thermoresistant coliforms and enterococci
140
Fig. 2. Calibration curves ATP vs. RLUs for each of the two campaigns.
For each of the three groups of bacteria tested, the analyses were conducted according to the corresponding French standards (NF T 90-414, XP T 90-416).
ARTICLE IN PRESS E. Delahaye et al. / Water Research 37 (2003) 3689–3696
3692
3. Results 3.1. First campaign: zone-by-zone analysis As far as the network covered by the Orly WTW is concerned and irrespective of the path followed by the water once it leaves the treatment plant, a very slight increase in the quantities of HPC-R2A and the quantities of ATP was observed (Table 1). This observation could be in relationship with the distance travelled from the treatment plant and the residence time in the network. In parallel with this increase in microbiological activity, a perceptible reduction in free chlorine concentration was noted. The results obtained in that part of the network supplied by the Ivry WTW (Table 1) revealed a degradation of the microbiological quality of the water
after storage in the Montmartre storage tank (IP9ST) with an increase from 996 to 4120 fg of ATP/ml and from 150 to 6660 HPC-R2A/ml. More generally, that part of the network situated furthest from the Ivry WTW offers the highest ATP and HPC-R2A values (points IP3-SP, IP4-SP, IP5ST, IP6-SP, IP7-SP, IP8-SP, IP9ST and IP10-SP). Table 1 also shows the results for that portion of the network supplied by the Joinville WTW. At the outlet to the M!enilmontant storage tank and on the branch directly served by the latter (points JP2-SP, JP3-SP and JP4-SP) no microbiological change in the quality of the water was detected. On the other hand a microbiological degradation was observed at the outlet to the Belleville storage tank (JP5ST). This degradation was also observed at certain of the points located downstream, such as JP9-SP.
Table 1 First campaign (spring (2001)), biological and physicochemical parameters for the Orly WTW distribution system, the Ivry WTW distribution system and the Joinville WTW distribution system Sampling point Orly plant (OP0) OP1-ST OP2-SP OP3-SP OP4-SP OP5-SP OP6-SP OP7-SP Ivry plant (IP0) IP1-SP IP2ST IP3-SP IP4-SP IP5ST IP6-SP IP7-SP IP8-SP IP9ST IP10-SP Joinville plant JP0 JP1ST JP2-SP JP3-SP JP4-SP JP5ST JP6-SP JP7-SP JP8-SP JP9-SP
ATP (fg/ml)
HPC-R2A/ml
HPC-PCA/ml (22 C)
HPC-PCA/ml (37 C)
Temperature ( C)
Free chlorine (mg Cl2/l)
7 8 2 71 254 134 223 422
0 10 10 20 80 500 390 740
0 0 0 0 0 0 0 0
0 0 0 0 0 0 0 0
14.2 12.8 12.0 11.7 12.0 11.9 12.1 11.6
0.56 0.32 0.33 0.04 0.04 oDL 0.04 oDL
76 36 63 918 650 374 731 1060 996 4120 802
0 0 0 70 35 605 2380 200 150 6660 920
0 0 0 0 1 0 0 0 0 3 4
0 0 0 0 1 1 1 0 1 1 1
12.8 11.4 12.2 11.7 11.8 12.1 11.8 11.8 11.3 10.9 11.8
0.58 0.15 0.07 oDL 0.05 oDL 0.11 oDL oDL oDL oDL
426
0
0
0
12.4
0.68
756 250 239 369 1760 78 96 154 2140
60 10 0 40 3720 80 170 560 3680
0 0 0 0 0 0 0 5 1
0 0 0 1 0 0 0 1 0
11.2 11.3 11.2 11.6 10.9 11.5 12.3 11.7 11.4
0.11 0.15 0.10 0.05 0.09 0.05 oDL 0.08 0.06
OP: supplied by Orly WTW, IP: supplied by Ivry WTW, JP: supplied by Joinville WTW, ST: storage tank, SP: sampling point, oDL: less than the detection limit.
ARTICLE IN PRESS E. Delahaye et al. / Water Research 37 (2003) 3689–3696
3693
Table 2 First campaign (spring (2001)), biological and physicochemical parameters for the Vanne aqueduct distribution system, the Vanne & Loing aqueducts distribution system and the Avre aqueduct distribution system Sampling point Vanne aqueduct (VA0) VA1-SP VA2-SP VA3-SP VA4-SP Loing aqueduct (LA0) VA/LA1ST VA/LA2-SP VA/LA3-SP VA/LA4-SP VA/LA5-SP VA/LA6-SP VA/LA7-SP VA/LA8-SP VA/LA9-SP VA/LA10-SP Avre aqueduct (AA0) AA1-SP
ATP (fg/ml)
HPC-R2A/ml
HPC-PCA/ml (22 C)
HPC-PCA/ml (37 C)
Temperature ( C)
Free chlorine (mg Cl2/l)
6
50
0
0
11.9
0.15
19 201 13 12
10 10 0 80
0 0 0 0
0 0 0 0
11.8 11.7 11.9 11.2
oDL 0.08 0.05 0.06
13
10
0
0
11.8
0.09
23 4 3 4 8 2 3 3 4 4
0 0 10 0 130 10 110 0 0 0
0 0 0 0 0 0 0 0 0 0
1 0 0 0 0 0 0 1 0 0
11.3 11.5 11.9 11.8 11.7 12.1 11.9 11.8 12.0 12.0
0.12 0.10 0.10 0.08 0.11 0.11 0.11 0.10 0.11 0.12
6
0
0
0
12.0
0.52
90
60
0
0
11.3
0.04
VA: supplied by Vanne aqueduct, VA/LA: supplied by Vanne aqueduct and Loing aqueduct, AA: supplied by Avre aqueduct, ST: storage tank, SP: sampling point, oDL: less than the detection limit.
Table 2 shows the results of the entire part of the distribution system supplied by the Vanne and Loing aqueducts. Just like the water leaving the three treatment plants, these underground waters are of good microbiological quality. They feed the Montsouris reservoir (VA/LA1ST) where no degradation in the water quality was apparent. More generally, all this part of the network supplied by the waters from the Vanne and Loing aqueducts exhibited considerable stability from the microbiological point of view.
3.2. First campaign: comparison between the different zones in the distribution network
Whereas the intra- and inter-zone temperatures are relatively homogeneous, there are substantial differences in the quantities of HPC-R2A and of ATP from one zone to another or within one and the same zone. Underground waters have good microbiological quality, which is reflected in low HPC-R2A and ATP values. As regards that part of the network supplied by surface waters (IP and JP among others) considerable variations in the number of HPC-R2A and the quantities of ATP were observed. This duality between surface waters and underground waters in terms of microbiological quality is well known at SAGEP and has already been reported elsewhere [8]. 3.3. Comparison between campaign 1and campaign 2
Fig. 4 compares the different supply zones in the network in terms of their averages regarding HPC-R2A, quantity of ATP, temperature of the water and free chlorine (by assimilating the non-detectable values to zero values without taking into account the values existing when leaving the plants and arriving at the aqueducts). In these figures are presented the minimum and maximum values, the interval within which are situated 75% of the values observed (white rectangle) while the black horizontal lines form the medial values.
Bacteriological parameters corresponding to European potability standards were carried out during the two campaigns. These tests have always produced results that reflect the good quality of the water distributed. No coliforms (total or thermoresistant) were to be found. The values observed in terms of HPC-PCA at 22 C did not exceed 10 CFU/ml, 1 CFU/ ml for HPC-PCA at 37 C. These values are low compared with certain of those obtained on R2A agar.
ARTICLE IN PRESS 3694
E. Delahaye et al. / Water Research 37 (2003) 3689–3696
Fig. 4. Distribution of temperature, free chlorine, HPC-R2A and ATP for each area during the first campaign (spring 2001). OP: Orly WTW distribution system, IP: Ivry WTW distribution system, JP: Joinville WTW distribution system, VA/LA: Vanne aqueduct and Loing aqueduct distribution system, AA: Avre aqueduct distribution system.
Table 3 First campaign vs. second campaign for average values of biological and physicochemical parameters measured on the Paris distribution system
ATP (fg/ml) HPC-R2A (CFU/ml) Temperature ( C) Free chlorine (mg Cl2/l)
Campaign 1
Campaign 2
3757725 46071200 11.870.5 0.1270.16
1407210 40780 14.971.0 0.1770.12
This suggests that the standardised PCA agar reveals only a fraction of the culturable bacteria. The results of campaign 2 are not shown here in detail since no degradation of the microbiological quality was observed despite an average temperature higher than that observed for campaign 1 (Table 3). On the other hand, the average free chlorine quantity in the network is higher during campaign 1 than campaign 2. This perhaps accounts for the lower values observed for campaign 2 in terms of HPC-R2A as well as the quantity of ATP.
4. Discussion 4.1. The Paris network The initial conclusion to be drawn from this study is that, for the two campaigns taken together, Paris water is of good quality and in compliance with the standards in force. This is reflected in low values, both from the point of view of standardised parameters and those of finer-scale measurements like the titrations of ATP or seedings on R2A agar. In effect, as regards HPC-R2A content, campaign 1 offers values comparable to, indeed lower than, those observed in other networks [9–11]. Even if the hygienic significance of these bacteria is nonexistent, their absence or presence in small quantities makes it possible to diagnose good general water quality. This is confirmed by the total absence of indicators of faecal contamination. It is certainly, in parts, the low residence time of the water in the distribution network (6 h on average) which permits maintenance of the microbiological quality at an excellent level. Other studies [12,13] stress that for residence times of less than 12 h, the water quality in the
ARTICLE IN PRESS E. Delahaye et al. / Water Research 37 (2003) 3689–3696
network remains very similar to that observed when the water leaves the plant. Nevertheless, the zone of the distribution network supplied by the Orly WTW reveals a slight degradation in the microbiological quality of the water and a decrease in the free residual chlorine content. This is certainly a function of the distance travelled and/or the residence time. Identical phenomena have already been reported [14]. Moreover, two of the storage tanks studied (IP9ST and JP5ST) show a change in the microbiological quality of the water. Several studies carried out in different distribution networks [15–17] stressed at the time the degradation of the microbiological quality of water due to storage tanks. It is incidentally the existence of a large number of intermediate reservoirs in that part of the network supplied by the treatment plants that might explain the sharp difference in quality between treated surface waters and underground waters observed during the first campaign. One of the hypotheses would then be to say that a lower water speed in the reservoirs than in the pipes, together with the existence of a water–atmosphere contact that is not to be found in the pipes could result in contamination [18]. However, it is also perhaps in part the greater constancy in the free chlorine content on that part of the network supplied by underground waters (Fig. 4) which provides an explanation for this difference in regard to surface waters, even if the effect of the chlorine as a stabiliser of the microbiological quality of a particular water for concentrations of less than 0.5 mg of free chlorine/l is not always clearly apparent [12,11]. The geographical variations in water quality are not solely dependent on the underground waters/surface waters factor. Effectively, that portion of the network supplied by the Orly WTW offers overall better water quality than that supplied by the Ivry and Joinville WTWs. To the extent that these three plants have identical final steps (ozonisation, granular-activated carbon filters and chlorination) and that no notable differences in the parameters measured with regard to the waters produced are apparent, several explanations can be put forward to account for these differences at the network level. On the one hand, there are numerous interconnexions between those parts of the distribution system supplied by the Ivry and Joinville WTWs. This results in the course of the frequent alternations of waters differing in origin which may also partly explain the more intense microbiological activity observed in these zones. On the other hand, the Orly WTW was built in the 1970s whereas the older WTWs at Ivry and Joinville were only revamped in the 1990s. Thus, Orly was for many years the most efficient of the three surface water treatment plants and may be the zones of the network served by Ivry and Joinville WTWs are more highly colonised with biofilm. Observation of pieces of
3695
pipe in place for the past several decades could be a means of verifying this hypothesis. 4.2. Differences between the two campaigns For the entire set of results no correlation could be demonstrated between the ATP or HPC-R2A content and the chlorine and temperature factors. It is difficult to explain the differences in the overall microbiological quality of the water in the Paris network between the two sampling campaigns carried out. The absence of a temperature effect was stressed during the zone-by-zone analysis of campaign I and the higher residual chlorine content during the second campaign (Table 3) does not represent a satisfactory explanation. Indeed, the residual chlorine values remain for the most part considerably lower than 0.5 ppm. The qualitative variation in the composition of the organic matter from one season to another may be another explanation of these differences in values between campaigns. Effectively [19] the seasonal evolution in dissolved organic matter is a well-known phenomenon, especially as regards such highly assimilable compounds as amino acids whose maximum concentrations in surface waters can be observed in the springtime (campaign 1). 4.3. The ATP titration kit For the line of regression (Fig. 5) of log (HPC-R2A/ ml) on log [ATP] (using the 64 paired results collected during the two sampling campaigns without zero values) R2 ¼ 0:36 (R ¼ 0:60). In spite of this low value, the statistical analysis proves that there does exist a significant linear relationship with a p-value=1.19 107 and a 95% confidence interval for R of 0.42– 0.73. Nevertheless, the relationship obtained on the network is lower than the one obtained during our preliminary test phase in the laboratory on surface water samples (Fig. 3). This is certainly because of the very low density of culturable bacteria in the network. So, the use of the ATP titration kit within the distribution network must be envisaged above all as an alarm system in the event of considerable microbiological drift.
5. Conclusion During the course of the study, the Paris network provided good-quality waters with bacteriological parameters matching the potability standards in force (HPCPCA at 22 C and 37 C, total and thermoresistant coliforms, enterococci) and also with non-standardised parameters offering increased detection of viable or
ARTICLE IN PRESS E. Delahaye et al. / Water Research 37 (2003) 3689–3696
3696 4 3.5 log [ATP]
3 2.5 2 y = 0.58x + 0.93 R2 = 0.36
1.5 1 0.5 0 0
1
2
3
4
5
log (HPC-R2A / ml)
Fig. 5. Relationship between HPC-R2A and ATP values (Paris network samples). [ATP]: ATP concentration (fg/ml).
culturable bacteria (HPC-R2A and ATP). Further, whatever the season, underground waters presented a high microbiological stability and showed no degradation during distribution. In early spring some of the waters originating from surface water treatment were liable to slight degradation during transit through the network. Different factors could well be responsible for this change in quality such as the distance travelled from the production site or the passage through intermediate storage tanks. Conversely, the effects of chlorine or temperature were not clearly apparent. The Paris network is heterogeneous with at least two zones corresponding to surface waters and underground waters. Moreover, surface waters may also constitute an heterogeneous set because of the differences obtained during the first campaign between the part supplied by Orly WTW and those supplied by Joinville and Ivry WTWs. It also appears necessary to observe pipes already in place and to measure the potential for the formation of biofilm at the exit from these three plants. The ATP titration kit provided globally satisfactory results that are correlated with the enumerations on R2A agar. Simple and rapid, its use within the distribution network can be envisaged above all as an alarm system in the event of considerable microbiological drift.
Acknowledgements This study has been greatly facilitated by the cooperation of Jean Louis Wagner and Charles Cervin (DMS, importer into France of Profile 1).
References [1] Paquin JL, Block JC, Haudidier K, Hartemann P, Colin F, Miazga J, Levi Y. Effect of chlorine on the bacterial colonisation of a model distribution system. Rev Sci Eau 1992;5:399–414.
[2] Flemming HC, Wingender J. Relevance of microbial extracellular polymeric substances (EPSs)—Part I: structural and ecological aspects. Water Sci Technol 2001;43:1–8. [3] De Beer D, Srinivasan R, Stewart PS. Direct measurement of chlorine penetration into biofilms during disinfection. Appl Environ Microbiol 1994;60:4339–44. [4] Lu W, Kiene L, Levi Y. Chlorine demand of biofilms in water distribution systems. Water Res 1998;33(3):827–35. [5] Neuhard J, Nygaard P. Purines and pyrimidines. In Escherichia coli and Salmonella typhimurium, vol. 1. Neidhardt (ed.) American Society for Microbiology. ASM Press, Washington DC ISBN: 0-914826-89-1, 1987; p. 445–73. [6] Miller JC, Miller JN. Statistics for analytical chemistry, 3rd ed. Chichester: Ellis Horwood Limited, ISBN: 0-13030990-7. [7] Reasoner DJ, Geldreich EE. A new medium for the enumeration and subculture of bacteria from potable water. Appl Environ Microbiol 1985;49:1–7. [8] Zacheus O, Martikainen P. Occurrence of heterotrophic bacteria and fungi in cold and hot water distribution systems using water of different quality. Can J Microbiol 1995;41:1088–94. [9] Carter JT, Rice EW, Buchberger SG, Lee Y. Relationships between levels of heterotrophic bacteria and water quality parameters in a drinking water distribution system. Water Res 2000;34:1495–502. [10] Gatel D, Servais P, Block JC, Bonne P, Cavard J. Microbiological water quality management in the Paris suburbs distribution system. J Water Supply Res TechAQUA 2000;49:231–43. [11] Mathieu L, Paquin JL, Block JC, Randon G, Maillard J, Reasoner D. Parameters governing bacterial growth in water distribution systems. Rev Sci Eau 1992;5:91–112. [12] Abouzaid H, El Mghari Tahib M, Outair A, Laraki L, Benabdallah S. Development of water quality in a distribution network: a case study from the Bouznika centre, Morocco. Water Supply 1996;14(3–4):453–71. [13] Piriou P, Dukan S, Levi Y, Jarrige PA. Prevention of bacterial growth in drinking water distribution systems. Water Sci Technol 1997;35(11/12):283–7. [14] Van Der Koo.ıj D. Potential for biofilm development in drinking water distribution systems. J Appl Microbiol Symp. Suppl. 1999;85:39S–44S. [15] Bailey IW, Thompson P. Monitoring of water quality after disinfection in distribution systems. Water Supply 1995; 13(2):35–48. [16] Clark RM, Goodrich JA, Wymer LJ. Effect of the distribution system on drinking water quality. J Water Supply Res Tech-AQUA 1993;42(1):30–8. [17] Maul A, El-Shaarawi AH, Block JC. Heterotrophic bacteria in water distribution systems. I. Spatial and temporal variation. II. Sampling design for monitoring. Sci Tot Environ 1985;44:201–24. [18] Kern!eis A, Nakache Feguin A, Feinberg M. The effects of water residence time on the biological quality in a distribution network. Water Res 1995;29(7):1719–27. [19] Dossier Berne F, Merlet N, Cauchi B, Legube B. Evolution of amino acids and dissolved organic matter in a drinking water treatment plant: correlations with biodegradable dissolved organic carbon and long-term chlorine demand. Rev Sci Eau 1996;1:115–33.
Water Research 37 (2003) 3689–3696
An ATP-based method for monitoring the microbiological drinking water quality in a distribution network E. Delahayea,*, B. Welte! b, Y. Levic, G. Leblond, A. Montielb a
SAGEP, Plate-forme de Recherche, 4 Avenue Pierre Mend!es France, 94340 Joinville le Pont, France b SAGEP, DQE, 9 rue Schoelcher, 75014 Paris, France c Laboratoire Sant!e Publique-Environnement, Facult!e de Pharmacie, Universit!e Paris Sud XI, France d Institut de G!en!etique et de Microbiologie, Universit!e Paris Sud XI, Orsay, France Received 31 May 2002; received in revised form 31 January 2003; accepted 14 April 2003
Abstract The titration of adenosine triphosphate (ATP) by bioluminescence permits rapid evaluation of the quantity of viable micro-organisms present in a water sample. During two sampling campaigns, Soci!et!e Anonyme de Gestion des Eaux de Paris (SAGEP) tested a new extraction and titration system of bacterial ATP in the Paris drinking water distribution network. As far as the entire set of results of analyses of water in the network is concerned there is a linear relationship between log [ATP] and log(HPC-R2A/ml). Furthermore, as regards the drinking water originating from treatment of surface waters, some of the results obtained indicate a slight change as regards the Paris network in the microbiological quality. This is certainly linked to the distance travelled from the production location as well as to a reservoir effect observed on a site. Conversely, no change is apparent with regard to waters of underground origin. Lastly, despite changes in temperature and chlorine residual, no significant influence has been observed, essentially because of the very low density of culturable bacteria. r 2003 Elsevier Science Ltd. All rights reserved. Keywords: Drinking water; ATP; Storage tank; Distribution network; Bacteria; Diagnosis
1. Introduction Despite the presence of disinfectant, bacterial enumerations are liable to increase in all drinking water distribution networks due to the presence of fixed and complex microbial ecosystems known as biofilms [1]. Biofilms are made up of micro-organisms, the majority of which are bacteria. The whole is included in a polymeric-type gel [2] secreted by the bacteria themselves and mineral elements. This enveloping structure can have reductive effects and inhibit the action of the disinfectants [3,4]. The presence of biofilms in distribution networks can result in numerous disagreeable problems for the user such as microbial contamination *Corresponding author. E-mail address: [email protected] (E. Delahaye).
of the water distributed, induced corrosion or the production of sapid compounds. Soci!et!e Anonyme de Gestion des Eaux de Paris (SAGEP) is the sole producer of drinking water for the City of Paris (France). Every day, this company abstracts, produces and transports 750,000 m3 of drinking water to meet the needs of the City of Paris. More precisely, 50% of the drinking water supply originates from underground waters captured within a radius of 80–150 km from Paris. These spring waters are conveyed to Paris by gravity over the aqueducts of the Avre (AA), the Loing (LA) and the Vanne (VA). The remaining 50% comes from surface waters treated by three water treatment works (WTW): Ivry (IP) and Orly (OP) on the river Seine, and Joinville le Pont (JP) on the river Marne. Each of these three WTWs possesses a maximum production capacity of 300,000 m3/day.
0043-1354/03/$ - see front matter r 2003 Elsevier Science Ltd. All rights reserved. doi:10.1016/S0043-1354(03)00288-4
ARTICLE IN PRESS 3690
E. Delahaye et al. / Water Research 37 (2003) 3689–3696
This study sets out to evaluate titration of ATP by bioluminescence for the rapid monitoring of the microbiological quality of the drinking water during its distribution in the Paris network.
2. Materials and methods 2.1. The Paris drinking water distribution network The tests were mainly conducted within the Paris distribution network. This is a highly complex gridded network in which the average water residence time is 6 h. Because of the complexity of the network it is impossible to clearly determine linear portions on which there could be a succession of sampling points at fully identified distances and residence times. Two sampling campaigns were carried out: *
*
the first (campaign 1) in the spring of 2001, a period in which spring waters and surface waters have similar temperatures, the second (campaign 2) in the autumn of 2001, when the waters are of different temperatures depending on their origins.
Each sampling campaign took place at some four dozen points (Fig. 1). All points sampled had facilities to allow the sample to be taken from the bulk flow. Some
of them are supplied by spring waters and others by potable waters produced from treated river waters. Depending on the instance, they are located at the entry to the network or, on the contrary, are at a distance from the production sites and on occasion after passing through one or two intermediate storage tanks. The nomenclature used to designate the points is as follows: *
*
*
the first two letters indicate the origin of the water: OP, JP and IP for the surface water treatment plants at Orly, Joinville and Ivry, respectively; AA and VA for the underground waters travelling across the Avre and Vanne aqueducts, respectively; VA/LA for the points supplied by a mixture of water between the underground waters of the Vanne and Loing aqueducts, the following figure indicates (as far as the complexity of the network allows) the position of the sampling point in relation to the production site (the plants are numbered 0) or of the place of arrival of the water in Paris (the aqueduct arrivals are numbered 0), finally, the last two letters specify, as regards each point, whether a storage tank (ST), or merely a sampling point (SP), is involved.
2.2. The ATP titration kit (Profile 1, New Horizons Diagnostics, imported into France by the D!evelopment Marketing Services company) The kit is made up of: * *
* *
*
Fig. 1. Localisation of the sampling points used for the two campaigns. (OP: supplied by Orly WTW, IP: supplied by Ivry WTW, JP: supplied by Joinville WTW, VA: supplied by Vanne aqueduct, VA/LA: supplied by Vanne aqueduct and Loing aqueduct, AA: supplied by Avre aqueduct, ST: storage tank, SP: sampling point).
a micro-luminometer (mains or battery supply), small doses of a freeze-dried mixture of luciferin/ luciferase and their reconstitution liquid, filtration devices (pore diameter 0.4 mm), a leaching reagent of the non-bacterial ATP and inhibiting substances (SRA reagent), a reagent ensuring extraction of the ATP of bacterial origin (BRA II reagent).
Stoichiometrically speaking, at the time of the reaction between the ATP and the luciferin–luciferase enzymatic system, one photon is produced for every one molecule of ATP consumed. This enables a linear relationship to be established between the quantity of light emitted and the quantity of ATP initially present. Indeed, the average ATP load of a bacterium is of the order of 7 mmol/g of dry weight for cells in exponential growth phase [5], or approximately 3.5 mg of ATP for 2.3 g wet weight that is to say 1 fg per bacterial cell if one considers that the wet weight of a cell is of the order of 6.7 1013 g. For water samples from the network after filtration (syringe fitted with a filtration device) of 10 ml of
ARTICLE IN PRESS E. Delahaye et al. / Water Research 37 (2003) 3689–3696
25000 20000
RLU
4.00 3.50 3.00 y = 0.97x + 0.61 R2 = 0.67
2.50 2.00 1.50 1.50
2.00
2.50 3.00 log (HPC-R2A / ml)
3.50
4.00
Fig. 3. Validation of the ATP titration kit on surface water samples (HPC-R2A vs. RLU).
previous testing on 57 surface water samples for validation purposes. In this case, for the line of regression (Fig. 3) between log [RLUs] and log (HPC-R2A/ml) the value of R2 is 0.67 (R ¼ 0:82). The statistical analysis of this Pearson’s product–moment correlation value proves that there is a highly significant linear relationship with a p-value of 7.99 1015 and a 95% confidence interval for R of 0.71–0.89. 2.3. Temperature and free chlorine Measurement of these parameters was carried out in the field at the time the samples were taken. The chlorine was dosed in accordance with the N,N-Diethylphenylene-1.4 diamine titrimetric French standard (AFNOR NF T 90-037). 2.4. Heterotrophic plate counts on R2A agar (HPC-R2A) The R2A agar (DIFCO) was used in order to estimate the density of culturable bacteria in the network water samples. This permits the growth of stressed bacteria which do not develop in other mediums [7]. For each sampling, two seedings were carried out. Incubation lasted 7 days at 22 C. The tables indicate that the results correspond to the average of the two seeded plates. 2.5. Heterotrophic plate counts on PCA agar (HPC-PCA 22 C and 37 C)
y = 182.43x - 482.33 R2 = 0.99
15000
4.50
log ( RLU)
sample, the filtration device was rinsed twice in succession (double washing recommended by the manufacturer for chlorinated samples) using 4 drops of SRA reagent, with, on each occasion, evacuation of the reagent from the filtration device by application of air pressure. After introduction of the filtration device into the luminometer, two drops of BRA II reagent were added immediately followed by 50 ml of the previously reconstituted luciferin/luciferase mixture (kept in ice). After rapid homogenisation of the whole, several successive measurements were carried out until a maximum was obtained in terms of relative luminescence units (RLUs). The time required for successful titration was about 2 min. Campaigns 1 and 2 having been conducted using different micro-luminometers. Standard curves were established prior to each campaign on standard solutions of ATP, 100 ml of which were directly deposited in a filtration device before addition of 50 ml of the reagent containing the luciferin–luciferase mixture. This made it possible to express the results in quantities of ATP rather than in RLU (a factor that may depend on the apparatus used). An error calculation concerning the value of the slope grade was made for each of the two linear straight regression lines [6] at a confidence coefficient of 95%. Fig. 2 presents the results in URL achieved for different concentrations of standard ATP solutions prior to campaigns 1 and 2. In both cases, the measurements carried out on the entire range covered by the luminometer (0 to 20000 URL) showed high correlation (R2 > 0:99) with the ATP concentrations. For campaign 1, consumption of a picogram of ATP corresponds to 158 (712) URL while the conversion factor used for campaign 2 is 182 (720) URL per picogram of ATP. When account is taken of the errors on the slope grades, it becomes apparent that the conversion factors used between one campaign and the next are not significantly different. Furthermore, before they are put to use in the distribution network, the equipment and the reagents kit designed for titration of the ATP have undergone
3691
y = 158.72x + 172.09 R2 = 0.99
For each sampling and incubation temperature, the analyses were carried out in duplicate in accordance with the NFEN ISO 62 22 standard.
10000 5000
campaign 1
campaign 2
0 0
20
40
60 80 100 ATP (pg / 100 µl)
120
2.6. Coliforms, thermoresistant coliforms and enterococci
140
Fig. 2. Calibration curves ATP vs. RLUs for each of the two campaigns.
For each of the three groups of bacteria tested, the analyses were conducted according to the corresponding French standards (NF T 90-414, XP T 90-416).
ARTICLE IN PRESS E. Delahaye et al. / Water Research 37 (2003) 3689–3696
3692
3. Results 3.1. First campaign: zone-by-zone analysis As far as the network covered by the Orly WTW is concerned and irrespective of the path followed by the water once it leaves the treatment plant, a very slight increase in the quantities of HPC-R2A and the quantities of ATP was observed (Table 1). This observation could be in relationship with the distance travelled from the treatment plant and the residence time in the network. In parallel with this increase in microbiological activity, a perceptible reduction in free chlorine concentration was noted. The results obtained in that part of the network supplied by the Ivry WTW (Table 1) revealed a degradation of the microbiological quality of the water
after storage in the Montmartre storage tank (IP9ST) with an increase from 996 to 4120 fg of ATP/ml and from 150 to 6660 HPC-R2A/ml. More generally, that part of the network situated furthest from the Ivry WTW offers the highest ATP and HPC-R2A values (points IP3-SP, IP4-SP, IP5ST, IP6-SP, IP7-SP, IP8-SP, IP9ST and IP10-SP). Table 1 also shows the results for that portion of the network supplied by the Joinville WTW. At the outlet to the M!enilmontant storage tank and on the branch directly served by the latter (points JP2-SP, JP3-SP and JP4-SP) no microbiological change in the quality of the water was detected. On the other hand a microbiological degradation was observed at the outlet to the Belleville storage tank (JP5ST). This degradation was also observed at certain of the points located downstream, such as JP9-SP.
Table 1 First campaign (spring (2001)), biological and physicochemical parameters for the Orly WTW distribution system, the Ivry WTW distribution system and the Joinville WTW distribution system Sampling point Orly plant (OP0) OP1-ST OP2-SP OP3-SP OP4-SP OP5-SP OP6-SP OP7-SP Ivry plant (IP0) IP1-SP IP2ST IP3-SP IP4-SP IP5ST IP6-SP IP7-SP IP8-SP IP9ST IP10-SP Joinville plant JP0 JP1ST JP2-SP JP3-SP JP4-SP JP5ST JP6-SP JP7-SP JP8-SP JP9-SP
ATP (fg/ml)
HPC-R2A/ml
HPC-PCA/ml (22 C)
HPC-PCA/ml (37 C)
Temperature ( C)
Free chlorine (mg Cl2/l)
7 8 2 71 254 134 223 422
0 10 10 20 80 500 390 740
0 0 0 0 0 0 0 0
0 0 0 0 0 0 0 0
14.2 12.8 12.0 11.7 12.0 11.9 12.1 11.6
0.56 0.32 0.33 0.04 0.04 oDL 0.04 oDL
76 36 63 918 650 374 731 1060 996 4120 802
0 0 0 70 35 605 2380 200 150 6660 920
0 0 0 0 1 0 0 0 0 3 4
0 0 0 0 1 1 1 0 1 1 1
12.8 11.4 12.2 11.7 11.8 12.1 11.8 11.8 11.3 10.9 11.8
0.58 0.15 0.07 oDL 0.05 oDL 0.11 oDL oDL oDL oDL
426
0
0
0
12.4
0.68
756 250 239 369 1760 78 96 154 2140
60 10 0 40 3720 80 170 560 3680
0 0 0 0 0 0 0 5 1
0 0 0 1 0 0 0 1 0
11.2 11.3 11.2 11.6 10.9 11.5 12.3 11.7 11.4
0.11 0.15 0.10 0.05 0.09 0.05 oDL 0.08 0.06
OP: supplied by Orly WTW, IP: supplied by Ivry WTW, JP: supplied by Joinville WTW, ST: storage tank, SP: sampling point, oDL: less than the detection limit.
ARTICLE IN PRESS E. Delahaye et al. / Water Research 37 (2003) 3689–3696
3693
Table 2 First campaign (spring (2001)), biological and physicochemical parameters for the Vanne aqueduct distribution system, the Vanne & Loing aqueducts distribution system and the Avre aqueduct distribution system Sampling point Vanne aqueduct (VA0) VA1-SP VA2-SP VA3-SP VA4-SP Loing aqueduct (LA0) VA/LA1ST VA/LA2-SP VA/LA3-SP VA/LA4-SP VA/LA5-SP VA/LA6-SP VA/LA7-SP VA/LA8-SP VA/LA9-SP VA/LA10-SP Avre aqueduct (AA0) AA1-SP
ATP (fg/ml)
HPC-R2A/ml
HPC-PCA/ml (22 C)
HPC-PCA/ml (37 C)
Temperature ( C)
Free chlorine (mg Cl2/l)
6
50
0
0
11.9
0.15
19 201 13 12
10 10 0 80
0 0 0 0
0 0 0 0
11.8 11.7 11.9 11.2
oDL 0.08 0.05 0.06
13
10
0
0
11.8
0.09
23 4 3 4 8 2 3 3 4 4
0 0 10 0 130 10 110 0 0 0
0 0 0 0 0 0 0 0 0 0
1 0 0 0 0 0 0 1 0 0
11.3 11.5 11.9 11.8 11.7 12.1 11.9 11.8 12.0 12.0
0.12 0.10 0.10 0.08 0.11 0.11 0.11 0.10 0.11 0.12
6
0
0
0
12.0
0.52
90
60
0
0
11.3
0.04
VA: supplied by Vanne aqueduct, VA/LA: supplied by Vanne aqueduct and Loing aqueduct, AA: supplied by Avre aqueduct, ST: storage tank, SP: sampling point, oDL: less than the detection limit.
Table 2 shows the results of the entire part of the distribution system supplied by the Vanne and Loing aqueducts. Just like the water leaving the three treatment plants, these underground waters are of good microbiological quality. They feed the Montsouris reservoir (VA/LA1ST) where no degradation in the water quality was apparent. More generally, all this part of the network supplied by the waters from the Vanne and Loing aqueducts exhibited considerable stability from the microbiological point of view.
3.2. First campaign: comparison between the different zones in the distribution network
Whereas the intra- and inter-zone temperatures are relatively homogeneous, there are substantial differences in the quantities of HPC-R2A and of ATP from one zone to another or within one and the same zone. Underground waters have good microbiological quality, which is reflected in low HPC-R2A and ATP values. As regards that part of the network supplied by surface waters (IP and JP among others) considerable variations in the number of HPC-R2A and the quantities of ATP were observed. This duality between surface waters and underground waters in terms of microbiological quality is well known at SAGEP and has already been reported elsewhere [8]. 3.3. Comparison between campaign 1and campaign 2
Fig. 4 compares the different supply zones in the network in terms of their averages regarding HPC-R2A, quantity of ATP, temperature of the water and free chlorine (by assimilating the non-detectable values to zero values without taking into account the values existing when leaving the plants and arriving at the aqueducts). In these figures are presented the minimum and maximum values, the interval within which are situated 75% of the values observed (white rectangle) while the black horizontal lines form the medial values.
Bacteriological parameters corresponding to European potability standards were carried out during the two campaigns. These tests have always produced results that reflect the good quality of the water distributed. No coliforms (total or thermoresistant) were to be found. The values observed in terms of HPC-PCA at 22 C did not exceed 10 CFU/ml, 1 CFU/ ml for HPC-PCA at 37 C. These values are low compared with certain of those obtained on R2A agar.
ARTICLE IN PRESS 3694
E. Delahaye et al. / Water Research 37 (2003) 3689–3696
Fig. 4. Distribution of temperature, free chlorine, HPC-R2A and ATP for each area during the first campaign (spring 2001). OP: Orly WTW distribution system, IP: Ivry WTW distribution system, JP: Joinville WTW distribution system, VA/LA: Vanne aqueduct and Loing aqueduct distribution system, AA: Avre aqueduct distribution system.
Table 3 First campaign vs. second campaign for average values of biological and physicochemical parameters measured on the Paris distribution system
ATP (fg/ml) HPC-R2A (CFU/ml) Temperature ( C) Free chlorine (mg Cl2/l)
Campaign 1
Campaign 2
3757725 46071200 11.870.5 0.1270.16
1407210 40780 14.971.0 0.1770.12
This suggests that the standardised PCA agar reveals only a fraction of the culturable bacteria. The results of campaign 2 are not shown here in detail since no degradation of the microbiological quality was observed despite an average temperature higher than that observed for campaign 1 (Table 3). On the other hand, the average free chlorine quantity in the network is higher during campaign 1 than campaign 2. This perhaps accounts for the lower values observed for campaign 2 in terms of HPC-R2A as well as the quantity of ATP.
4. Discussion 4.1. The Paris network The initial conclusion to be drawn from this study is that, for the two campaigns taken together, Paris water is of good quality and in compliance with the standards in force. This is reflected in low values, both from the point of view of standardised parameters and those of finer-scale measurements like the titrations of ATP or seedings on R2A agar. In effect, as regards HPC-R2A content, campaign 1 offers values comparable to, indeed lower than, those observed in other networks [9–11]. Even if the hygienic significance of these bacteria is nonexistent, their absence or presence in small quantities makes it possible to diagnose good general water quality. This is confirmed by the total absence of indicators of faecal contamination. It is certainly, in parts, the low residence time of the water in the distribution network (6 h on average) which permits maintenance of the microbiological quality at an excellent level. Other studies [12,13] stress that for residence times of less than 12 h, the water quality in the
ARTICLE IN PRESS E. Delahaye et al. / Water Research 37 (2003) 3689–3696
network remains very similar to that observed when the water leaves the plant. Nevertheless, the zone of the distribution network supplied by the Orly WTW reveals a slight degradation in the microbiological quality of the water and a decrease in the free residual chlorine content. This is certainly a function of the distance travelled and/or the residence time. Identical phenomena have already been reported [14]. Moreover, two of the storage tanks studied (IP9ST and JP5ST) show a change in the microbiological quality of the water. Several studies carried out in different distribution networks [15–17] stressed at the time the degradation of the microbiological quality of water due to storage tanks. It is incidentally the existence of a large number of intermediate reservoirs in that part of the network supplied by the treatment plants that might explain the sharp difference in quality between treated surface waters and underground waters observed during the first campaign. One of the hypotheses would then be to say that a lower water speed in the reservoirs than in the pipes, together with the existence of a water–atmosphere contact that is not to be found in the pipes could result in contamination [18]. However, it is also perhaps in part the greater constancy in the free chlorine content on that part of the network supplied by underground waters (Fig. 4) which provides an explanation for this difference in regard to surface waters, even if the effect of the chlorine as a stabiliser of the microbiological quality of a particular water for concentrations of less than 0.5 mg of free chlorine/l is not always clearly apparent [12,11]. The geographical variations in water quality are not solely dependent on the underground waters/surface waters factor. Effectively, that portion of the network supplied by the Orly WTW offers overall better water quality than that supplied by the Ivry and Joinville WTWs. To the extent that these three plants have identical final steps (ozonisation, granular-activated carbon filters and chlorination) and that no notable differences in the parameters measured with regard to the waters produced are apparent, several explanations can be put forward to account for these differences at the network level. On the one hand, there are numerous interconnexions between those parts of the distribution system supplied by the Ivry and Joinville WTWs. This results in the course of the frequent alternations of waters differing in origin which may also partly explain the more intense microbiological activity observed in these zones. On the other hand, the Orly WTW was built in the 1970s whereas the older WTWs at Ivry and Joinville were only revamped in the 1990s. Thus, Orly was for many years the most efficient of the three surface water treatment plants and may be the zones of the network served by Ivry and Joinville WTWs are more highly colonised with biofilm. Observation of pieces of
3695
pipe in place for the past several decades could be a means of verifying this hypothesis. 4.2. Differences between the two campaigns For the entire set of results no correlation could be demonstrated between the ATP or HPC-R2A content and the chlorine and temperature factors. It is difficult to explain the differences in the overall microbiological quality of the water in the Paris network between the two sampling campaigns carried out. The absence of a temperature effect was stressed during the zone-by-zone analysis of campaign I and the higher residual chlorine content during the second campaign (Table 3) does not represent a satisfactory explanation. Indeed, the residual chlorine values remain for the most part considerably lower than 0.5 ppm. The qualitative variation in the composition of the organic matter from one season to another may be another explanation of these differences in values between campaigns. Effectively [19] the seasonal evolution in dissolved organic matter is a well-known phenomenon, especially as regards such highly assimilable compounds as amino acids whose maximum concentrations in surface waters can be observed in the springtime (campaign 1). 4.3. The ATP titration kit For the line of regression (Fig. 5) of log (HPC-R2A/ ml) on log [ATP] (using the 64 paired results collected during the two sampling campaigns without zero values) R2 ¼ 0:36 (R ¼ 0:60). In spite of this low value, the statistical analysis proves that there does exist a significant linear relationship with a p-value=1.19 107 and a 95% confidence interval for R of 0.42– 0.73. Nevertheless, the relationship obtained on the network is lower than the one obtained during our preliminary test phase in the laboratory on surface water samples (Fig. 3). This is certainly because of the very low density of culturable bacteria in the network. So, the use of the ATP titration kit within the distribution network must be envisaged above all as an alarm system in the event of considerable microbiological drift.
5. Conclusion During the course of the study, the Paris network provided good-quality waters with bacteriological parameters matching the potability standards in force (HPCPCA at 22 C and 37 C, total and thermoresistant coliforms, enterococci) and also with non-standardised parameters offering increased detection of viable or
ARTICLE IN PRESS E. Delahaye et al. / Water Research 37 (2003) 3689–3696
3696 4 3.5 log [ATP]
3 2.5 2 y = 0.58x + 0.93 R2 = 0.36
1.5 1 0.5 0 0
1
2
3
4
5
log (HPC-R2A / ml)
Fig. 5. Relationship between HPC-R2A and ATP values (Paris network samples). [ATP]: ATP concentration (fg/ml).
culturable bacteria (HPC-R2A and ATP). Further, whatever the season, underground waters presented a high microbiological stability and showed no degradation during distribution. In early spring some of the waters originating from surface water treatment were liable to slight degradation during transit through the network. Different factors could well be responsible for this change in quality such as the distance travelled from the production site or the passage through intermediate storage tanks. Conversely, the effects of chlorine or temperature were not clearly apparent. The Paris network is heterogeneous with at least two zones corresponding to surface waters and underground waters. Moreover, surface waters may also constitute an heterogeneous set because of the differences obtained during the first campaign between the part supplied by Orly WTW and those supplied by Joinville and Ivry WTWs. It also appears necessary to observe pipes already in place and to measure the potential for the formation of biofilm at the exit from these three plants. The ATP titration kit provided globally satisfactory results that are correlated with the enumerations on R2A agar. Simple and rapid, its use within the distribution network can be envisaged above all as an alarm system in the event of considerable microbiological drift.
Acknowledgements This study has been greatly facilitated by the cooperation of Jean Louis Wagner and Charles Cervin (DMS, importer into France of Profile 1).
References [1] Paquin JL, Block JC, Haudidier K, Hartemann P, Colin F, Miazga J, Levi Y. Effect of chlorine on the bacterial colonisation of a model distribution system. Rev Sci Eau 1992;5:399–414.
[2] Flemming HC, Wingender J. Relevance of microbial extracellular polymeric substances (EPSs)—Part I: structural and ecological aspects. Water Sci Technol 2001;43:1–8. [3] De Beer D, Srinivasan R, Stewart PS. Direct measurement of chlorine penetration into biofilms during disinfection. Appl Environ Microbiol 1994;60:4339–44. [4] Lu W, Kiene L, Levi Y. Chlorine demand of biofilms in water distribution systems. Water Res 1998;33(3):827–35. [5] Neuhard J, Nygaard P. Purines and pyrimidines. In Escherichia coli and Salmonella typhimurium, vol. 1. Neidhardt (ed.) American Society for Microbiology. ASM Press, Washington DC ISBN: 0-914826-89-1, 1987; p. 445–73. [6] Miller JC, Miller JN. Statistics for analytical chemistry, 3rd ed. Chichester: Ellis Horwood Limited, ISBN: 0-13030990-7. [7] Reasoner DJ, Geldreich EE. A new medium for the enumeration and subculture of bacteria from potable water. Appl Environ Microbiol 1985;49:1–7. [8] Zacheus O, Martikainen P. Occurrence of heterotrophic bacteria and fungi in cold and hot water distribution systems using water of different quality. Can J Microbiol 1995;41:1088–94. [9] Carter JT, Rice EW, Buchberger SG, Lee Y. Relationships between levels of heterotrophic bacteria and water quality parameters in a drinking water distribution system. Water Res 2000;34:1495–502. [10] Gatel D, Servais P, Block JC, Bonne P, Cavard J. Microbiological water quality management in the Paris suburbs distribution system. J Water Supply Res TechAQUA 2000;49:231–43. [11] Mathieu L, Paquin JL, Block JC, Randon G, Maillard J, Reasoner D. Parameters governing bacterial growth in water distribution systems. Rev Sci Eau 1992;5:91–112. [12] Abouzaid H, El Mghari Tahib M, Outair A, Laraki L, Benabdallah S. Development of water quality in a distribution network: a case study from the Bouznika centre, Morocco. Water Supply 1996;14(3–4):453–71. [13] Piriou P, Dukan S, Levi Y, Jarrige PA. Prevention of bacterial growth in drinking water distribution systems. Water Sci Technol 1997;35(11/12):283–7. [14] Van Der Koo.ıj D. Potential for biofilm development in drinking water distribution systems. J Appl Microbiol Symp. Suppl. 1999;85:39S–44S. [15] Bailey IW, Thompson P. Monitoring of water quality after disinfection in distribution systems. Water Supply 1995; 13(2):35–48. [16] Clark RM, Goodrich JA, Wymer LJ. Effect of the distribution system on drinking water quality. J Water Supply Res Tech-AQUA 1993;42(1):30–8. [17] Maul A, El-Shaarawi AH, Block JC. Heterotrophic bacteria in water distribution systems. I. Spatial and temporal variation. II. Sampling design for monitoring. Sci Tot Environ 1985;44:201–24. [18] Kern!eis A, Nakache Feguin A, Feinberg M. The effects of water residence time on the biological quality in a distribution network. Water Res 1995;29(7):1719–27. [19] Dossier Berne F, Merlet N, Cauchi B, Legube B. Evolution of amino acids and dissolved organic matter in a drinking water treatment plant: correlations with biodegradable dissolved organic carbon and long-term chlorine demand. Rev Sci Eau 1996;1:115–33.