Drinking Water Distribution Systems

APPLIED AND ENVIRONMENTAL MICROBIOLOGY, Sept. 1981, p. 506-512 0099-2240/81/090506-07$02.00/0 Vol. 42, No. 3 Coliform Inhibition by Bacteriocin-Like Substances in Drinking Water Distribution Systems EDWARD G. MEANS AND BETTY H. OLSON* Environmental Analysis, Program in Social Ecology, University of California, Irvine, California 92717 Received 19 September 1980/Accepted 1 June 1981 Bacterial isolates from an unchlorinated potable groundwater system and a chlorinated surface water system were screened by an agar overlay method for the ability to produce bacteriocin-like substances (BLS) inhibitory to the growth of Escherichia coli, Klebsiella sp., and Enterobacter aerogenes. The production of coliform-specific BLS by noncoliform bacteria varied with the site and date of isolation as well as the genus of the producer strain. A total of 448 bacterial isolates were screened from the chlorinated system, and 22.1% produced BLS specific for at least one of the three coliforms. In the unchlorinated system, 7.9% (n = 696) possessed this ability. Flavobacterium/Moraxella comprised 57.1% of all bacteria (from both systems) producing BLS. The possibility that BLS interfere with coliform detection in standard bacteriological water quality tests is discussed. Bacteriocins were discovered in 1925 when tween the phenomena of coliform injury (3) and Andre Gratia (10) observed that the supernatant death. from a broth culture of one strain of the bacterium Escherichia coli would not support the (12) were corroborated by data from the Na- The experimental results of Hutchinson et al. growth of another strain, even when diluted tional Water Supply Study (15) as analyzed by 1,000-fold. Further experimentation by Gratia Geldreich et al. (8). Coliform detection in the and others, notably Frederiqc (7), led to the water distribution systems studied was stable as discovery of many different colicins (bacteriocins produced by E. coli) and the development 500 cells/ml. At densities of 1,000 noncoliform noncoliform bacterial numbers increased up to of a typing scheme. cells/ml, incidences of coliform detection decreased in frequency. Other researchers have Early research indicated that bacteriocins were fatal to the cell once adsorption to the cell reported similar findings (19). surface occurred; however, this effect was reversible if the bacteriocin was cleaved with trypera as being "antagonistic" toward coliform bac- Various studies have implicated specific gensin within 2 h of bacteriocin adsorption at the teria. These include Pseudomonas, Sarcina, sensitive cells' receptor (17). It has been suggested that some bacteriocins may be bacteriolus, and Actinomycetes (6, 13, 21). Micrococcus, Flavobacterium, Proteus, Bacilstatic in nature (4). Three basic phenomena may explain the interaction of noncoliform bacteria with coliform The influence of noncoliform microorganisms on the survival and recovery of coliform bacteria bacteria and the subsequent suppression of the has been well documented in the literature. In coliforms in standard MPN tests: nutritional freshwater environments, antagonism of E. coli competition, injury, or chemical antibiosis. The by noncoliform bacteria has been shown both phenomenon of bacterial injury (3) might exacerbate the effects ofboth nutritional competition directly and indirectly. Hutchinson et al. (12) isolated bacteria capable of inhibiting E. coli in and antibiosis sensitivity. Of primary interest in 8, 11, and 37% of water samples from spring, this paper is chemical antibiosis, specifically, the groundwater, and surface sources, respectively. interaction of coliforms with bacteriocin-like By adjusting the densities of these antagonistic substances (BLS) or related inhibitory substances produced by noncoliform bacteria. bacteria, they were able to show a reduction in the recovery of E. coli after incubation for 12 h The mechanism of coliform suppression has at 10 C. Enumerations of E. coli in these suspensions by the most probable number (MPN) cins or BLS might be responsible for the occur- not been fully determined. However, bacterio- technique (1) were reduced by 28 to 97%. However, these experiments did not differentiate be- gas production in the presumptive medium) in rence of false-negative results (growth without 506

VOL. 42, 1981 the MPN method (1). The failure of coliform bacteria to produce gas from lactose fermentation in the presumptive portion of the test and the recovery of this ability in the confirmed test may be a function of bacteriocin adsorption and release. This "growth but no gas" phenomenon has been reported for seawater (18), freshwater, and drinking water (6) samples. Suppression of coliform bacteria can result in a serious underestimation of the bacteriological quality of the water supply. To assess the extent and significance of BLS production by noncoliform bacteria in water distribution systems, bacterial isolates from a chlorinated and nonchlorinated water distribution system were screened for their ability to inhibit the growth of three members of the coliform group: Escherichia coli, Klebsiella sp., and Enterobacter aerogenes. In addition, bacterial isolates from false-negative and presumptive MPN tubes, and from positive confirmatory tubes, were identified and screened for their ability to inhibit the predominant coliform isolated from the same tubes. MATERLALS AND METHODS Description of the water distribution systems. The Irvine Ranch Water District (IRWD) distribution system receives fully treated and chlorinated surface water. This water is a 60:40 blend of Colorado River and northern California waters. A residual chlorine level of approximately 0.5 mg/liter is maintained in the system. Seventy-five thousand individuals receive water from the IRWD system. The nonchlorinated water system of the city of Garden Grove, Calif., is supplied with groundwater pumped from 30 wells in the area. Surface water (fully treated and chlorinated) from the Metropolitan Water District of southern California is imported in the summer to accommodate peak water usage. This imported water accounts for approximately 10% of the total annual supply. Normally, there is no treatment of water before its distribution; however, when imported water is used to supplement groundwater supplies, a chlorine residual of 0.1 mg/liter can be detected in certain areas of the city. Isolation of the bacteria. The two systems were sampled from January 1979, through September 1979. A total of 2,033 isolates were obtained and identified during January, April, May, July, and September of 1979. Samples were collected from fire hydrants by using sterile 250-ml wide-mouth glass bottles containing 0.3 ml of a 10% solution of sodium thiosulfate. Hydrants were flushed for 1 min at a rate of 756 liters/ min before sample collection. At each sampling time, approximately 100 bacteria from each of six hydrant sites were isolated from plate count (PC) agar (Difco Laboratories). Plates selected for bacterial isolation had been spread with 1 ml of distribution system water and incubated at 35 C for 48 h. The isolates were purified on PC agar, transferred to PC slants, stored BACTERIOCIN-LIKE SUBSTANCES IN DRINKING WATER 507 at 4'C, and identified within 3 months of collection by the Lassen method (14). Date of isolation, identification, and site of isolation were recorded for each isolate. Source of coliform test organisms. The Klebsiella sp. and E. aerogenes used in the screening procedure were obtained from samples from the Garden Grove water distribution system and isolated from the membrane filter m-endo (Difco) test (1). The E. coli strain was obtained from D. Wulff, University of Califomia, Irvine. All three coliform organisms were identified by the Lassen method (14). Screening for BLS production by the isolates. The isolates were screened according to the method of Frederiqc (7) with minor modifications. Tubes containing 3 ml of PC broth were inoculated and incubated for 96 h at room temperature. The tubes exhibiting growth were blended, and 0.01 ml of broth culture was spotted onto PC agar plates in triplicate. Six to nine organisms were placed on each plate. The plates were then incubated for 48 h at room temperature. In a fume hood, the 48-h plates were inverted over a plastic petri dish containing a standard glass slide upon which several drops of reagent-grade chloroform (Mallinckrodt) were placed. This procedure simplified Bauernfeind and Burrows' (2) modification of Gillies and Govan's (9) method. After the isolates had been exposed for 30 min, additional drops of chloroform were placed on the slide. The total exposure time was 1.5 h. After removal from the chloroform source, each of the three replicate plates was overlaid with 3 ml of warm (45 C) PC broth containing 0.5% agar and 0.01 ml of a 24-h PC broth culture of E. coli, Klebsiella sp., or E. aerogenes. After the agar overlay, solidified plates were incubated for 24 h at 35 C and then examined for zones of inhibition in the areas of the coliform lawn surrounding the test isolate. Zone size in each of the three coliform lawns was recorded in millimeters for each isolate tested. Positive results were confirmed in replicated screening. Failure of bacterial isolates to survive subculture for BLS screening was also recorded and is referred to as loss of viability. Analysis of positive and false-negative presumptive MPN tubes. A positive and a false-negative presumptive MPN tube, both of which had confirmed in brilliant green bile broth (BGB; Difco), were selected for study. Volumes of 0.01 ml of medium from the respective BGB tubes were streaked onto eosin methylene blue agar (Difco) and incubated for 24 h at 35 C. Examination of the eosin methylene blue plates revealed atypical pink mucoid colonies from both tubes. Representative colonies were then purified and identified as Enterobacter cloacae by the Lassen method (14). These isolates were inoculated into lauryl tryptose broth (LTB; Difco) and incubated for 24 h at 35 C. Both isolates produced acid and gas reactions typical of members of the coliform group. The respective positive LTB tube and its corresponding BGB tube, along with the false-negative tube and its corresponding BGB tube, were serially diluted in phosphate buffer (1), and the dilutions were plated on PC agar to yield approximately 50 colonies per plate. All colonies from a given plate were purified,

508 MEANS AND OLSON placed on PC agar slants, and identified by the Lassen method (14). After identification, the isolates were tested for their ability to produce BLS specific against the associated predominant coliform isolated from the corresponding MPN tube, using the modification of Frederiqc's method described earlier. Additional modification was required, as extremely mucoid colony types (primarily Aeromonas hydrophila) were present which were resistant to inactivation by chloroform. Colony growth after soft agar overlay prevented zone measurements. To circumvent this problem, the colonies were carefully scraped off the agar surface with sterile spatulas before exposure to the chloroform. This effectively reduced the overgrowth problem yet left undisturbed any BLS which had diffused into the agar adjacent to the colony. The previously described overlay procedure was followed to complete the screening. RESULTS A total of 2,033 bacteria were isolated and identified from the two water distribution systems investigated. Of these isolates, 1,144 were screened by the agar overlay method for their ability to produce BLS specific for an E. coli, Klebsiella sp., and E. aerogenes strain (Fig. 1). A total of 44.0% of the isolates were not examined because they were no longer viable at the time of BLS screening. To consider the effect of loss of viability (dieoff) during storage of isolates on the reliability of the observed BLS producer frequencies, mortalities by type, date, and site were computed (Table 1). Garden Grove had a higher die-off among all genera (except Enterobacter) than did the IRWD system. This difference was greatest for Pseudomonas, Acinetobacter, and Klebsiella. The Flavobacterium/Moraxella genera displayed similar mortality in both systems. The date of bacterial isolation affected the APPL. ENVIRON. MICROBIOL. die-off in the IRWD and Garden Grove systems similarly (Table 1). The smallest loss was observed from the most recent sampling time (11 October 1979, 8.4% in the IRWD system and 11.7% in the Garden Grove system). However, low die-off also occurred in Irvine in late January (12.9%). Die-off percentages by date (Table 1) ranged from 8.4 to 64.2% in the IRWD system and from 11.7 to 61.2% in the Garden Grove system. In addition, a Pearson correlation coefficient of r = 0.41 was obtained between length of isolate storage and loss of culture viability, indicating that storage time did to some degree FIG. 1. Representative zones of growth inhibition produced by BLS on an agar overlay lawn of one of the three coliform test organisms used in this study (Klebsiella sp.). The two BLSproducer strains shown (2624 and 2625) were identified as belonging to the Flavobacterium/Moraxella group. TABLE 1. Percentage of loss of viability of bacteria isolates during storage at 40C by genera and date of isolation % Loss' Distribution system Acineto- Pseudomo- Pasteu- Pseudomo- Flavobac- Enterobacter nas rella nas/alca- teriuml Klebsiella Serratta bacter ligenes Moraxella Irvine 20.2 6.25 0.0 22.6 36.2 23.8 0.0 25.0 (99) (16) (106) (296) (63) (6) (4) Garden Grove 42.1 62.5 30.4 35.9 40.4 38.9 11.1 0.0 Irvineb - 12.9 62.9-39.4 64.2 8.4 (201) (196) (94) (106) (119) Garden Groveb 59.7-61.2 52.9 46.9-11.7 (313) (214) (278) (296) (230) a Sample size is given in parentheses; -, no sample collected. ball cultures were tested for viability and BLS production in November 1979. Length of storage at 40C, therefore, varied according to the date of bacterial isolation: Acinetobacter, 2 January 1979; Pseudomonas, 25 January 1979; Pasteurella, 26 April 1979; Pseudomonas/Alcaligenes, 24 May 1979; Flavobacterium/Moraxella, 19 July 1979; Klebsiella, 26 July 1979; Serratia 11 October 1979.

VOL. 42, 1981 BACTERIOCIN-LIKE SUBSTANCES IN DRINKING WATER 509 affect successful subculture of bacterial isolates. The loss of bacterial viability during storage at 40C effectively caused observed frequencies of BLS producers to underestimate actual frequencies. Table 2 shows figures for mortality by site. Excluding well 23/9 because of small sample size (n = 40), die-off percentages ranged from 28.1% at Rockinghorse to 64.6% at well 6064. BLS production by bacterial isolates. A site-specific phenomenon occurred in the production of BLS by the isolates (Table 3). At Paseo Picasso, 88 of 196 isolates (34.9%) produced BLS specific for E. coli, Klebsiella sp., or E. aerogenes. Of this 34.9%, a total of 75 strains (85.0%) were Flavobacterium/Moraxella. Less than 8.0% of the isolates from the other five sites examined were capable of producing substances inhibitory for the three coliforms. The data in Table 3 show that the percentage of producers varied between sample times, with a low of zero in the 26 July sample to a maximum TABLE 2. Percentage of loss of viability during storage of isolates by site of isolation % of nonvia- Total no. of Site ble organisms isolates Irvine Source 38.5 319 Paseo Picasso 34.8 387 Garden Grove Well 6064 64.6 365 Well 6321 56.2 457 Well 2319 15.0 40 Rockinghorse 28.1 469 of 22.1% in the 11 October sample. A small percentage of the bacterial population was capable of producing an inhibitory substance in the majority of samples. Producer occurrence peaked in January, July, and October. Of the isolates from the chlorinated IRWD system, 22.1% were capable of inhibiting at least one of the three coliform bacteria tested, compared with 7.9% for the unchlorinated Garden Grove system. Combining the data from both systems, 13.4% of the isolates examined were capable of producing BLS specific for at least one of the three coliform test organisms. Sample location, date of sample collection, and genus of the producer strain significantly influenced the degree and frequency of inhibition of the three coliforms (Tables 4 and 5). However, due to the relatively small Cramers V association coefficients, the relationships were significant but not extremely strong. This statistical relationship implies that other factors not controlled for here may be more important in governing the occurrence of BLS-producing bacteria in water distribution systems. The size of the zone of inhibition in lawns of E. coli varied according to the site from which the BLS-producing bacteria were isolated (F = 20.7, P = < 0.001). Date of isolation appeared to be of equal importance for both E. coli and Klebsiella sp. (F 7.5, P = = < 0.001), but of lesser importance for E. aerogenes (F 2.1, P = < 0.05). Table 6 shows the number of bacteria capable of inhibiting the growth of the three test organisms. Of the isolates tested, 3.9, 5.8, and 7.6% were capable of inhibiting E. aerogenes, Kleb- TABLE 3. Effect of sample location and sampling time on the bacterial population producing BLS Isolates producing inhibitor specific for: % of isolates pro- Site or date of No. tested ducing coliform- E. coli Klebsiella sp. E. aerogenes isolation inhibitory substancea No. % No. % No. % Site Irvine Source 196 5.6 3 1.5 3 1.5 5 2.6 Paseo Picasso 252 34.9 55 21.8 40 15.9 25 9.9 Well 6064 129 4.6 4 3.1 3 2.3 0 0.0 Well 6321 196 14.3 15 7.5 12 6.0 8 4.0 Well 23/9 34 2.9 1 2.9 0 0.0 0 0.0 Rockinghorse 337 5.9 9 2.7 8 2.4 7 2.1 Date 2 Jan. 1979 126 8.7 4 3.2 4 3.2 6 4.8 25 Jan. 1979 175 14.8 15 8.6 8 4.6 7 4.0 26 April 1979 148 8.5 6 3.9 7 4.6 6 3.9 24 May 1979 131 4.6 3 2.3 2 1.5 1 0.8 19 July 1979 214 13.6 15 7.0 13 6.1 6 2.8 26 July 1979 38 0.0 0 0.0 0 0.0 0 0.0 11 Oct. 1979 312 22.1 44 14.1 32 10.3 19 6.1 a The sum of the row percentages do not always equal the percentage of isolates producing coliform-inhibitory substances because of BLS cross-specificity.

510 MEANS AND OLSON TABLE 4. Analysis of variance: size of the zone of inhibition for E. coli, Klebsiella spp., and E. aerogenes by site, date, and genus Site Date Genus Inhibition zone diam of: F P F P F P E. coli 20.7 <0.001 7.5 <0.001 5.9 <0.001 Klebsiella sp. 6.7 <0.001 7.5 <0.001 5.0 <0.001 E. aerogenes 8.6 <0.001 2.1 <0.05 4.6 <0.001 TABLE 5. Coliform Chi-square analysis: frequency of coliform-specific BLS producers by site, date, and bacterial genus Site Date Genus x2 P V x pp 2 V E. coli 118.9 <0.001 0.24 90.5 <0.001 0.21 91.7 <0.001 0.21 Klebsiella sp. 79.8 <0.001 0.16 55.4 <0.001 0.16 69.3 <0.001 0.18 E. aerogenes 43.7 <0.001 0.15 27.1 <0.001 0.12 53.3 <0.001 0.16 TABLE 6. Numerical breakdown of bacterial genera producing BLS for Enterobacter, Aerogenes, Klebsiella spp., or E. coli Inhibition of E. aero- Inhibition of Klebsiella Inhibition of E. coli genes sp. Group Total No. in- Total No. in- Total No. inno. hibit- % no. hibit- % no. hibit- % tested ing tested ing tested ing Actinomycetes 44 1 2.2 44 2 4.5 44 1 2.2 Acinetobacter spp. 178 0 0.0 178 8 4.5 178 8 4.5 Pseudomonas spp. 18 0 0.0 18 0 0.0 18 2 11.1 Pasteurella spp. 16 0 0.0 16 4 25.0 16 5 31.3 Pseudomonas/Alcaligenes spp. 268 4 1.5 267 2 0.7 267 5 1.9 Flavobacterium/Moraxella spp. 351 28 8.0 353 37 10.5 353 53 15.0 Klebsiella spp. 114 4 3.5 114 8 7.0 114 3 2.6 Serratia spp. 14 0 0.0 14 0 0.0 14 2 14.3 Enterobacter spp. 7 2 28.6 7 2 28.5 7 2 28.6 Unknown' 64 5 7.8 64 2 3.1 64 6 9.4 Unidentifiedb 70 1 1.4 70 1 1.4 70 0 0.0 Total 1,144 45 3.9 1,145 66 5.8 1,145 87 7.6 aunidentifiable by Lassen method. 'Not yet identified. siella sp., and E. coli, respectively. Of the Enterobacter species examined, approximately 28.0% produced BLS which were specific for E. aerogenes, E. coli, and Klebsiella sp. Coliform isolates from the distribution system generally exhibited strongest production of BLS (gauged by zone size) specific for the coliform test microorganisms (E. aerogenes, E. coli, and Klebsiella sp.). The Flavobacterium/Moraxella group was an important noncoliform inhibitor for all three types of coliforms. Of these isolates, 8, 10, and 15% produced substances inhibitory for E. aerogenes, Klebsiella sp., and E. coli, respectively. Some representatives of all the genera tested were capable of inhibiting at least one of the three coliforms. Relatively high percentages of the Serratia (13.3%), Flavobacterium/Moraxella (8.4%), and Pseudomonas (8.3%) groups APPL. ENVIRON. MICROBIOL. appeared to produce BLS for two of the three coliform groups. Members of eight different genera had the ability to inhibit two of the three coliform test organisms. Only four of the bacterial genera examined were active against all three coliforms. These included the Flavobacterium/Moraxella (0.4%), Klebsiella (0.6%), Enterobacter 25.0%), and unidentified genera (0.5%). Flavobacterium/Moraxella comprised 57.1% of the bacteria producing BLS determined by this method, whereas Acinetobacter, Klebsiella, and Pseudomonas/Alcaligenes groups comprised 9.1, 7.8, and 7.1%, respectively. Composition of the MPN tube genera. The predominant coliform in both the false-negative and positive BGB tubes was E. cloacae. E. cloacae was present in such low numbers in the

VOL. 42, 1981 positive LTB tube that we could not isolate it at a dilution of 10-6 (Table 7). The only difference between the positive and false-negative MPN tubes was the presence of the Flavobacteriumr Moraxella group in the latter. Although the screening of representative isolates from the MPN tubes failed to demonstrate any BLS activity against the predominant coliforms from the same tubes, members of the Flavobacterium/Moraxella group have been previously shown to be common coliform antagonists (16). DISCUSSION The single most important factor which could introduce sample bias in this study was the differential die-off of bacterial genera between the IRWD and the Garden Grove systems. The experimental design of this study precluded the determination of whether the die-off of stored isolates was completely random with respect to the isolates' ability to produce BLS. There is no evidence in the literature to suggest that the ability to produce BLS confers survival advantages or disadvantages in vitro. Therefore, it is likely that the loss of isolates during storage at 40C decreased the ranks of both BLS producers and nonproducers randomly and independently of genus, site location, or date of isolation. The inhibition zone size and the frequency of BLS production varied with the site of sample collection, the date of sample collection, and the generic status of the bacterial isolates. Thus, a variety of environmental factors, both physicochemical and biological, might influence the occurrence and frequency of bacteria which synthesize coliform-specific BLS in water distribution systems. Such factors may explain the marked differences observed in the percentage of BLS producers in the two systems studied. Part of this difference was undoubtedly due to the greater frequency of the Flavobacteriuml Moraxella group in the IRWD system. TABLE 7. Generic make-up of the isolates from positive and false-negative LTB and BGB tubes No. of isolates (x 106) Isolatesa LTB± BGB± LTB+ BGB+ Enterobacter cloacae 20 26-24 Aeromonas hydrophila 310 14 460 29 Klebsiella sp. _b 1 - - Serratia sp. - 3-15 Flavobacterium/Morax- 90 1 - - ella sp. Total 420 45 460 68 a No. of colonies isolated off plate count agar for identification: LTB±, 42; BGB±, 45; LTB+, 46; BGB+, 68. b -, No representatives of the listed genus were isolated at a 10-6 phosphate buffer dilution. BACTERIOCIN-LIKE SUBSTANCES IN DRINKING WATER 511 Mitchell and Yankofsky (16) concluded that the bactericidal properties of natural seawater for E. coli depended upon the native marine bacterial flora. However, others have proposed that antibiotic production by marine microorganisms was not a significant antagonist of coliform bacteria in natural seawater (5). Because of the comparatively low density of microorganisms in drinking water, it is unlikely that antibiotics would play a significant role in the suppression of coliform bacteria by noncoliform microorganisms. It is difficult to demonstrate experimentally that bacteria produce BLS under natural conditions and, at present, in situ synthesis has not been quantified. Studies supporting the single-hit hypothesis of bacteriocin function (i.e., one molecule of bacteriocin is sufficient to produce a death response as opposed to several molecules) (11) improve the possibility that low levels of bacteriocin production might conceivably cause coliform injury or death in situ. However, contact between a coliform cell and a coliform-specific BLS molecule must still occur. Assuming uniform distribution of bacteria in water distribution mains, then low concentrations of BLS in the system water would reduce the probability of such contact occurring. Bacterial colonization of pipe surfaces in water distribution systems has been demonstrated (20; M. Allen and E. Geldreich, in Technological Conference Proceedings, Water Quality, in press). Such colonization might create a situation where BLS inhibition is more plausible. These microcolonies could be important sites of production of coliform-specific BLS by noncoliform bacteria. Approximately 10% of the isolates in this study produced BLS specific for at least one of three coliforms tested. This suggests that the potential for bacteriocin-induced suppression of coliforms in these systems is considerable. False-negative MPN tubes (i.e., growth without gas production) have been recorded in 15 instances over the 27-month study period, representing 5.2% of the water samples collected. The cause of these false-negative results is unknown, although it is conceivable they may involve BLS. When large numbers of BLS producers are inoculated with coliform bacteria into MPN presumptive medium tubes, suppression of the coliform organisms might occur. Such inhibition might explain the decrease observed by Hutchinson et al. (12) in numbers of E. coli cells which had been exposed to a noncoliformn bacterial antagonist. The preliminary MPN experiments described herein failed to provide support for BLS-induced suppression of coliforms in MPN tubes, and additional related experiments must be performed to clarify this point. Nonetheless,

512 MEANS AND OLSON it is clear from this research that coliform-specific bacteriocins have the potential to suppress coliform detection in MPN tests. Whether this is, in fact, a significant phenomenon is unknown at this time. What has previously been termed "bacterial antagonism" may be related to such bacteriocins or BLS produced by one bacterium and specific for another. ACKNOWLEDGMENTS We thank Jeff Garvey and Milton Aust of the Garden Grove Water Department and Bob McGrew of the Irvine Ranch Water District for their cooperation during the collection of the water samples. We are grateful to Carlos Martinez for technical assistance. This research was supported by grant R805680010 from the U.S. Environmental Protection Agency. LITERATURE CITED 1. American Public Health Association. 1976. Standard methods for the examination of water and wastewater, 14th ed. American Public Health Association, Inc., New York. 2. Bauernfeind, A., and J. R. Burrows. 1978. Suggested procedure allowing use of plastic petri dishes in bacteriocin typing. Appl. Environ. Microbiol. 35:970. 3. Bissonnette, G. K., J. J. Jezeski, G. A. McFeters, and D. G. Stuart. 1975. Influence of environmental stress on enumeration of indicator bacteria from natural waters. Appl. Microbiol. 29:186-194. 4. Bradley, D. 1967. Ultrastructure of bacteriophages and bacteriocins. Bacteriol. Rev. 31:230-314. 5. Carlucci, A. F., P. V. Scarpino, and D. Pramer. 1961. Evaluation of factors affecting survival of Escherichia coli in sea water. V. Studies with heat- and filter-sterilized sea water. Appl. Microbiol. 9:400-404. 6. Fischer, G. 1950. The antagonistic effect of aerobic sporulating bacteria on coli-erogenes group. Z. Immunol. Exp. Ther. 107:16-22. 7. Frederiqc, P. 1948. Actions antibiotiques reciproques chez les Enterobacteria-ceae. Rev. Belge Pathol. Med. Exp. 19(Suppl. 4):1-107. APPL. ENVIRON. MICROBIOL. 8. Geldreich, E. E., H. D. Nash, and D. J. Reasoner. 1972. The necessity for controlling bacterial populations in potable waters: community water supply. J. Am. Water Works Assoc. Sept. 596-602. 9. Gillies, R. R., and J. R. W. Govan. 1966. Typing Pseudomonas pyocyanea by pyocine production. Pathol. Bacteriol. 91:339-345. 10. Gratia, A. 1925. Sur un remarquable exemple d'antagonisome entre deux souches de Sap Colibacille. C. R. Soc. Biol. 93:1040-1041. 11. Hedges, A. J. 1966. An examination of single-hit and multi-hit hypotheses in relation to the possible kinetics of colicin adsorption. J. Theor. Biol. 11:383-410. 12. Hutchinson, D., R. H. Weaver, and M. Scherago. 1943. The incidence and significance of microorganisms antagonistic to Escherichia coli in water. J. Bacteriol. 45: 29. 13. Kligler, L. J. 1919. Non-lactose fermenting bacteria from polluted wells and sub-soil. J. Bacteriol. 4:35-42. 14. Lassen, J. 1975. Rapid identification of gram-negative rods using a three-tube method combined with a dichotomic key. Acta Pathol. Microbiol. Scand. Sect. B 83: 525-533. 15. McCabe, L. H., J. M. Symons, R. D. Lee, and G. G. Robeck. 1970. Survey of community water supply systems. J. Am. Water Works Assoc. 62:670-687. 16. Mitchell, R., and S. Yankofsky. 1967. Lysis of Escherichia coli by marine microorganisms. Nature (London) 215:891-892. 17. Nomura, M., and M. Nakamura. 1962. Reversibility of inhibition of nucleic acids and aprotein synthesis by colicin K. Biochem. Biophys. Res. Commun. 7:306-309. 18. Olson, B. H. 1978. Enhanced accuracy of coliform testing in seawater by a modification of the most-probablenumber method. Appl. Environ. Microbiol. 36:438-444. 19. Reitler, R., and R. Seligmann. 1957. Pseudomonas aeruginosa in drinking water. J. Appl. Bacteriol. 20: 145-150. 20. Ridgway, H. F., and B. H. Olson. 1981. Scanning electron microscope evidence for bacterial colonization of a drinking-water distribution system. Appl. Environ. Microbiol. 41:274-287. 21. Weaver, R. H., and T. Boiter. 1951. Antibiotic-producing species of Bacillus from well water. Trans. Ky. Acad. Sci. 13:183-188.