Major efforts to improve the microbiological quality of bathing waters are being made in many countries. Regulations regarding bathing water quality (11, 53) have led to improvement in the microbiological quality of coastal waters and to a reduction in the incidence of waterborne diseases related to recreational water contact. However, water-related illnesses, particularly those caused by viral pathogens, are still of concern. These viruses cause a variety of waterborne diseases such as gastroenteritis, infectious hepatitis, meningitis, respiratory diseases, and eye infection (28). Moreover, many studies have reported the isolation of viruses in bathing waters that comply with the bacteriological criteria (15, 27, 34, 35, 42). These viruses can be transmitted by contact with polluted recreational waters (6, 8, 13).

New regulations for bathing waters are under consideration in the United States and the European Union. The U.S. Water Quality Standards for Coastal and Great Lakes Recreation Waters Proposed Rule (55), the proposal for a directive of the European Union Parliament and of the council concerning the quality of bathing waters (12), and the WHO guidelines for safe recreational waters (58) only include bacterial indicators as microbiological quality parameters. However, there is some concern due to the fact that these current bacterial indicators are not always reliable indicators of the presence of human pathogenic viruses, which is reflected in expert comments on the proposed regulations. Thus, a group of experts of the European Union (the Scientific Committee on Toxicity, Ecotoxicity, and Environment) recommended further research into viral indicators (46). Also, the Public and Scientific Affairs Board Committee on Environmental Microbiology of the American Society for Microbiology has submitted comments about the appropriateness of adding other indicators to the present bacterial indicators and specifically mentions bacteriophages (3).

Enteroviruses (31), adenoviruses (21, 41), and reoviruses (35) were mentioned for their potential to indicate the virological quality of costal waters. However, up to now, only the present European directive 76/160/ECC contemplates monitoring for viruses, stipulating a standard of 0 enteroviruses per 10 liters (11). Moreover, at present, enteroviruses are the group of viruses that can be most easily detected and are cultivated with feasible methods. They account for an estimated 5 to 15 million symptomatic infections each year in the United States (52). The range of diseases associated with enterovirus infections includes poliomyelitis, aseptic meningitis, encephalitis, neonatal enteroviral disease, myocarditis, pericarditis, hand-foot-and-mouth disease, upper respiratory disease, and others (28). Enterovirus infections can be recreationally waterborne (29). Some recent developments in cultivable enterovirus concentration and counting methods, such as VIRADEN, which combines concentration and detection (32, 39), offered the opportunity of having realistic measurements of enteroviruses in volumes of up to 10 liters of bathing water.

The inclusion of a bacteriophage parameter as a quality standard has been widely discussed, but there is no consensus as to the most appropriate group of bacteriophages. Somatic coliphages, F-specific RNA bacteriophages, and bacteriophages infecting Bacteroides fragilis have been proposed as model viruses in water quality control (17). Feasible standardized methods for their enumeration are available for all of them (18-20, 54). They have been reported in both seawaters and freshwaters used for bathing all over the world (15, 38). However, many of these studies were carried out either on only some of the phage group or with methods not yet standardized. In addition, studies with viruses and the three groups of phages are scarce. This clearly hinders the comparison of data on the occurrence of the above-mentioned group of phages and viruses.

We report here a method that improves the detection of cultivable enteroviruses in volumes of up to 10 liters. The report also compares values of cultivable enteroviruses obtained by these methods with the values of various bacterial indicators and groups of bacteriophages proposed as indicators in coastal water from beaches that comply with European Union regulations in the metropolitan area of Barcelona. In addition, in situ die-off experiments were run with enteroviruses and various bacterial indicators and bacteriophage groups in order to explain, at least partially, the relative abundance of the various indicators and viruses, which differ from those found in the fecal residues affecting the area.

Samples and sampling area. Samples were collected from two beaches in the metropolitan area of Barcelona, which has 39.5 km of coastline. About 23 km of this coastline is bathing beaches that are visited by several million bathers a year. The area is mostly affected by fecal pollution of humans (from 3.5 to 4.0 million) through two small rivers mostly affected by secondary effluents of municipal sewage treatment plants and the secondary effluents of three municipal wastewater treatment plants that discharge through three underwater outfalls. The outfalls discharge at different distances from the coast and at various depths; the smallest is 1,500 m from the coastline at a depth of 16 to 19 m, and the largest is 3,500 m at a depth of 55 to 60 m. Influence of agricultural, stockbreeding, and wildlife runoff inputs, if any, is negligible. Thus, during the period studied, the water in this area received treated sewage, accidental storm urban water runoff, and combined sewer overflows or fortuitous spills of nontreated wastewater. The densities of the microorganisms studied are quite steady in raw sewage and secondary effluents. Their numbers in raw sewage have been reported in several publications (26) and are (in log₁₀ units per 100 ml) as follows: Escherichia coli (6.7 to 7.1), enterococci (6.1 to 6.7), somatic coliphages (6.6 to 6.9), F-specific RNA bacteriophages (5.4 to 5.8), phages infecting strain RYC2056 of Bacteroides fragilis (3.8 to 4.2), phages infecting strain GA17 of Bacteroides thetaiotaomicron (3.8 to 4.3), and enteroviruses (1.8 to 3.3). Their numbers in secondary effluents are between 2 and 5% of these values (26).

In the metropolitan area of Barcelona, 14 beaches with 24 sampling points are sampled weekly (17 times) from 1 June to 15 September every year; information can be found at the Generalitat de Catalunya website (http://mediambient.gencat.net/aca/ca//medi/aigues_litorals). In the period studied (summer 2002), 100% of the beaches in the area complied with the bathing water quality criteria of European Union directive 76/160/CEE (information found at the Generalitat de Catalunya website mentioned above). The average numbers of bacterial indicators in seawater of the Barcelona coastline during the study period were as follows: 1.43 (±1.25) log₁₀ units per 100 ml for total coliform, 0.94 (±1.19) log₁₀ units per 100 ml for fecal coliform, and 0.78 (±0.99) log₁₀ units per 100 ml for fecal enterococci.

All samples were collected at two sampling points located at two different beaches (beaches A and B) during a period starting at the beginning of June and finishing at the end of October, which roughly corresponds to the bathing season. Independently, we had been routinely sampling weekly the two beaches studied during the summers (17 samples in triplicate each summer of 2000, 2001, and 2002).

All samples were collected 20 cm deep in 20-liter sterile polypropylene containers, transported to the laboratory at 4°C, and processed within 4 h of collection.

Bacterial determinations. Bacterial indicators were counted by membrane filtration according to standards of the American Public Health Association (1). For routine monitoring, total coliforms, fecal coliforms, and fecal streptococci were determined by membrane filtration of 1 and 10 ml of sample for each and further plating on mEndo-Agar-LES (Becton-Dickinson, Sparks, MD), mFC agar (Becton-Dickinson, Sparks, MD), and mEnterococcus agar (Becton-Dickinson, Sparks, MD) plus confirmation on bile esculin agar (Becton-Dickinson, Sparks, MD). For the specific study reported here, enterococci were determined as done for the routine monitoring (filtering 10 and 100 ml of sample), and Escherichia coli was detected and counted on Fluorocult LMX agar (Merck, Darmstadt, Germany); fluorescent colonies were confirmed with the indole test.

Bacteriophage detection and counting. Somatic coliphages were detected and counted by the double-layer technique, with 10 ml of sample analyzed according to a standard of the International Organisation for Standardisation (19).

For F-specific bacteriophages and phages infecting Bacteroides fragilis RYC2056 and Bacteroides thetaiotaomicron GA17, 200 ml of seawater samples was concentrated by adsorption to 0.22-μm-pore-size cellulose esters membrane filters (GSWP; Millipore Corp., Bedford, MA), followed by elution with a solution of 1% beef extract, 0.05 M NaCl, and 3% Tween 80 and ultrasonication according to a method described previously by Mendez et al. (30). The entire eluted volumes were counted by the double-layer technique according to standards of the International Organisation for Standardisation (18, 20), and the bacteriophages still retained in the membranes were counted by placing the membrane face down onto a host monolayer after the membrane was cut into fragments.

Determination of the genotype of F-specific RNA bacteriophages. Plaque transfer, fixation onto nylon membranes, and hybridization were carried out as described elsewhere previously (44). The specific oligonucleotides used to genotype F-specific RNA bacteriophages were those described previously by Beekwilder et al. (2).

Enterovirus detection and counting. Three procedures were assayed. Enteroviruses were concentrated from 10 liters by adsorption to and elution from glass powder, as described previously by Lucena et al. (25). The eluted viruses were then counted either by the standard plaque assay (49) or by the double-layer plaque assay (33) by using Buffalo green monkey (BGM) cells (European Collection of Animal Cell Cultures accession number 90092601).

The third method assayed was the VIRADEN method (39), with some modifications so as to apply it to higher volumes than those originally described. Briefly, the water sample was clarified by filtration through 0.65-μm-pore-size low-protein-binding polyvinylidene difluoride membranes, as described elsewhere previously (32), and amended by adding MgCl₂ · 6H₂O to a final concentration of 0.05 M MgCl₂. The water sample was then filtered through a 90-mm-diameter 3-μm-pore-size cellulose esters membrane filter at a flow rate never exceeding 0.1 liters · min⁻¹. The viruses adsorbed on the membrane were counted on a BGM cell monolayer as follows. The growth medium in a 90-mm-diameter tissue culture dish with a confluent monolayer was discarded. Next, 260μl of a suspension of BGM cells in minimal essential medium supplemented with extra antibiotics containing 1.75 × 10⁷ ± 0.25 × 10⁷ cells · ml⁻¹ was placed in the center of the tissue culture dish. The membrane with the adsorbed viruses was then carefully placed upside down on top of the cell suspension and the cell monolayer; and finally, 15 ml of overlay medium was poured slowly onto the center of the membrane filter and spread all over the plate. Next, as was done for the standard plaque assay, the agar was allowed to set, and the tissue culture dishes were incubated at 37°C in the presence of 5% CO₂ at a relative humidity of more than 80% for 96 h. Afterwards, the agar and the membrane were removed simultaneously and the monolayer was stained with 0.1% crystal violet-formaldehyde solution.

A combination of two different cell lines can be used, as described elsewhere previously (33). When a second cell line was used in the membrane filter, the method was as follows: after filtering virus suspension, 260 μl of the second cell line (CaCo2; ATCC HTB37) with the same cell concentration was placed in the center of the plate on a BGM confluent monolayer, and the membrane filter was placed on top.

Viruses from suspected plaques were isolated before staining with crystal violet in formaldehyde solution. They were then regrown on BGM cells and tested by reverse transcription (RT)-PCR with primers specific to enteroviruses EP1 and EP4 (16) according to the RT-PCR method described previously by Spinner and Di Giovanni (50).

Detection of enterovirus by RT-PCR. Enteroviruses were concentrated from 500 ml of seawater by filtration through 0.22-μm-pore-size 47-mm-diameter cellulose esters membrane filters. They were eluted by placing the filter with 2 ml of glycine buffer (0.25 N, pH 9.5) in orbital agitation for 30 min. After this, 2 ml of 2× PBS (pH 7.1) was added. Viruses were concentrated again using Amicon Ultra Filter units (Millipore Corp., Bedford, MA) and recovered in 140 μl of 1× PBS (pH 7.1). RNA was extracted with the QIAamp Viral RNA Mini kit (QIAGEN GmbH, Germany), according to the manufacturer's instructions, and nested PCR was performed, as described previously by Pina et al. (41), with primers EP1-EP4 (16) and nEP1-nEP4 (22).

In situ natural inactivation experiments. In situ inactivation experiments were performed on bacteria (fecal coliforms, E. coli, and fecal streptococci) and bacteriophages (somatic coliphages, F-specific RNA bacteriophages, and bacteriophages infecting strain GA17 of B. thetaiotaomicron) naturally occurring on sewage and on echovirus 6 and coxsackievirus B5 grown in the laboratory. It is not feasible to obtain the concentrations of naturally occurring enteroviruses needed to perform this kind of experiment.

Raw sewage was diluted 1:50 in seawater from the site where the inactivation experiment was performed, and echovirus 6 and coxsackievirus B5 grown in the laboratory were added at a final concentration of 4 × 10³ ± 2 × 10³ PFU per ml. The microorganisms mixed as indicated above were placed in a dialysis tube, with a cutoff of 14,000 Da, and were sealed and placed at a 20- to 25-cm depth in a marina in the area of study.

Samples were taken after 1, 2, 3, and 7 days, and the various microbes were counted as described above. Fecal coliforms were detected and counted on mFC agar. These experiments were performed in the summer season. Water characteristics were measured between 11 and 12 a.m. Temperature ranged from 24 to 25°C, the pH ranged from 7.7 to 8.1, salinity ranged between 36 and 37‰, and turbidity ranged from 0.8 to 1.8 nephelometric turbidity units.

Data processing and statistical analysis. Data comparisons were performed using analysis of variance and the Student's t test. Differences and significance of the regression lines of the inactivation experiments were tested with the line regression comparison test. The receiver operating characteristic (ROC) curve (ROC analysis) was used to compare sensitivity and specificity of different microbial indicators to predict the number of enteroviruses. ROC analysis is frequently used to establish sensitivity and specificity of procedures and hence to compare procedures. All tests were done using the Statistical Package for Social Science (51).

Enterovirus detection method. Table 1 shows how the VIRADEN method using 10-liter samples provided greater recoveries than concentration by adsorption to and elution from glass powder, whichever enterovirus quantification method (standard plaque assay or double-layer plaque assay) was used. This difference is seen both in the numbers of positive tests and in the numbers of viruses detected when both methods gave positive isolation.

TABLE 1. Comparison of various methods for concentration and enumeration of enteroviruses in seawater

As described previously, VIRADEN allows the use of two different cell lines in the two cell layers formed (33). The use of CaCo2 cells in the upper layer was also tested, but the recoveries did not differ significantly (Student's t test; P > 0.05) from the recoveries obtained with BGM cells, although they gave some more positive detections (60% versus 40%). Since manipulation of BGM cells is easier than manipulation of CaCo2 cells, we decided to proceed with only BGM cells. Therefore, the results reported below are those obtained with BGM cells.

In some samples, very few plaques were obtained. In this case, all suspected plaques were isolated, regrown on BGM cells, and confirmed as enterovirus with a specific RT-PCR for enteroviruses. When the number of plaques was high, a percentage of randomly chosen plaques was tested as indicated above.

Enteroviruses, bacteriophages, and bacterial indicators at two selected beaches in the Barcelona area. Results for the routine monitoring are shown in Table 2. The levels of conventional bacterial indicators were quite steady, without significant differences (analysis of variance; P > 0.05) during the 3years, with most values near 0 CFU per 100 ml and less than 10% of values greater than 100 CFU per 100 ml. These outlier values were usually found after heavy rainstorms.

TABLE 2. Values of bacterial indicators found in routine monitoring of tested beachesa

Results of the specific microbiological monitoring are given in Table 3. The seawater samples tested had the salinity, pH, color, and turbidity characteristic of the area (see Materials and Methods), with the exception of sample 3, which was quite turbid and dark brown. Values of E. coli showed that the beaches were suitable for bathing under the new proposed European Union regulations. In addition, the values of enterococci showed which beaches were suitable for bathing according to the old and the proposed European Union and U.S. bathing waters regulations and WHO guidelines. However, cultivable enteroviruses in 10 liters were detected in 10 of the 18 (55%) samples from which results were obtained, since two samples were toxic for the cell cultures. Values ranged from 1 to 158 PFU per 10 liters. Suspected plaques were isolated, regrown on BGM cells, and tested with a specific RT-PCR for enteroviruses, as indicated above. All suspected plaques tested were confirmed as enteroviruses by this method. In contrast, enteroviruses were detected by nested RT-PCR only in 4 of the 20 samples which did not coincide with the samples with the higher counts of cultivable enteroviruses.

TABLE 3. Microbiological determinations in seawater samples

Coliphages were detected in all samples but one (95%), although only 10 ml was tested, with numbers clearly higher that the numbers found with bacterial indicators. The geometric mean (standard deviation) for somatic coliphages was 63 (6.1), whereas those for E. coli and enterococci were 9.4 (8.7) and 7.0 (3.9), respectively. Values of E. coli and enterococci for these samples fully agree with values of fecal coliforms and enterococci in the area (see Materials and Methods). Given the number of positive samples (95% for somatic coliphages, 60% for E. coli, and 80% for enterococci) and the densities, we can conclude that coliphages were significantly more abundant than either E. coli or enterococci.

F-specific RNA bacteriophages (concentration of 200 ml) were only detected in three (15%) of the tested samples, with values more than 2 log₁₀ units lower than the numbers of somatic coliphages. The genotypes of the F-specific RNA phages detected were only I (18%) and II (82%).

Regarding bacteriophages infecting Bacteroides, those infecting strain GA17 were detected in 70% of the samples, with numbers between 1 and 2 log₁₀ units lower than those of somatic coliphages but higher than those of F-specific RNA phages. However, phages infecting strain RYC2056 were only detected in 16% of the samples.

ROC curves were done using the criteria “numbers of enteroviruses in 10 liters of seawater” and numbers of both bacterial and viral indicators. Only ROC curves with enterococci and somatic coliphages produce areas under the curve (the surface of the areas being 0.7 and 0.63, respectively) that could be considered sufficiently high to make methods comparable. This means that the numbers of enterococci and somatic coliphages predict more accurately the numbers of cultivable enteroviruses than the other indicators tested.

In situ natural inactivation experiments. The results of the in situ natural inactivation experiments of bacterial (fecal coliforms, E. coli, and enterococci) and viral (somatic coliphages, F-specific RNA phages, and phages infecting strain GA17 of Bacteroides thetaiotaomicron naturally occurring in sewage) indicators and laboratory-grown enteroviruses diluted in seawater are shown in Fig. 1.

FIG. 1. Survival of the various indicators studied in the in situ inactivation experiments. E. coli was below the detection limit at 48 h. F-specific phages with RNA genome (FRNAPH) were below the detection limit at 72 h. Fecal coliforms (FC) and fecal enterococci (more ...)

The regression lines of inactivation values [log(N_t/N₀), where N_t is the number of microorganisms at time t and N₀ is the number of microorganisms at time zero] of the various microorganisms studied were calculated from data of six independent experiments (Table 4). Slopes and R² values of the regression lines are shown in Table 4, as are the values of T₉₀ and T₉₉ calculated from the regression lines.

TABLE 4. Slopes and R2 values of the regression lines obtained from inactivation data of the in situ inactivation experimentsa

The microorganisms studied can be clustered in four groups according to the inactivation regression lines. There were significant differences (line regression comparison tests; P < 0.01) between these four groups. E. coli, the only component of the first group, was the least persistent, whereas somatic coliphages and phages infecting strain GA-17 of Bacteroides thetaiotaomicron were the most persistent microorganisms in the inactivation experiments. The two enteroviruses tested, echovirus 6 and coxsackievirus B5, were less persistent than these two groups of bacteriophages but more persistent than fecal coliforms, enterococci, and F-RNA phages.

The method described here to detect enteroviruses in 10 liters of seawater performs better, both in positive detection and in numbers detected, than the other methods tested, which in turn compete successfully with other traditional methods based on adsorption-elution (47). The method is feasible (technically and economically), at least for studies like this one, which seeks as much information as possible about the fate of enteroviruses and their surrogate indicators in waters. Of course, improving the detection method gives valuable information about both the presence of enteroviruses and their relationship to indicators. All the isolates characterized were enteroviruses, whereas other authors have described the frequent isolation of reoviruses in surface waters when BGM cells are used (after cytopathic effect in liquid culture for 2 weeks). The nondetection of reoviruses may be due to either the inability of the method to detect them because of short incubation or their absence because of the source of the fecal contamination. In our study, the source of fecal contamination is predominantly human, whereas in the other studies, it may have been a mixture of human and animal fecal contamination.

Enteroviruses were detected in 55% of samples from beaches complying with the present regulations and with future regulations regarding the occurrence and levels of bacterial indicators. Detection of enteroviruses by cell culture in marine water and freshwater that comply with the regulations based on bacterial indicators has already been reported (7, 27, 34, 35, 56, 57). Percentages of isolation reported here and in other publications are relatively similar, but numbers per 10 liters reported here are higher than numbers reported previously (27, 56, 57). This difference is probably due to the better performance of the method used here. In our view, the kinds of data obtained in this study allow better comparison with other parameters.

Only 4 (20%) of the 20 samples analyzed gave positive amplification of enterovirus RNA by nested RT-PCR. Moreover, the positive samples were not those with more cultivable enteroviruses or any of the other indicators tested. Results reported in the literature regarding detection of enteroviruses and other human viruses by genomic techniques are very variable, ranging from 7 to 70%, and are also inconsistent with the values of other indicators (9, 36, 41). Regardless of which group or groups of viruses should be tested, several factors limiting the use of genomic methods for monitoring water quality need to be considered. First, the detection of viral genomes in water by PCR or RT-PCR methods does not provide information about the infectivity of the viruses in question, although some methods that may help to circumvent this problem are arising (37). This impedes a meaningful risk evaluation if positive results are obtained (14). Second, the high sensitivity of these techniques needed in environmental studies is likely to contribute PCR artifacts (cross- and carryover contamination, nonspecific amplification, inhibition by environmental inhibitors of reverse transcriptase and DNA polymerase, etc.). Several multicenter studies for the detection of viruses by genomic methods revealed problems of standardization (23), which increase still further with environmental samples because of the need for a virus concentration step. Given the sensitivity of the method we used to determine cultivable enteroviruses and the problems still pending with genomic techniques, we think that counting cultivable enteroviruses is still the best approach to both determine the risks and study which potential viral indicators best protect users of recreational waters against pathogenic viruses.

Somatic coliphages were the most abundant bacteriophages. However, low numbers of F-specific RNA and phages infecting strain RYC2056 were detected. These results coincide with data reported previously for the same area (5). As in that previous study, the densities of somatic coliphages were higher than the densities of bacterial indicators and do not correlate with them. The ratios between somatic coliphages and F-RNA bacteriophages were much higher than those in raw sewage. The F-specific RNA belonged to genogroups (genotypes) I and II, whereas genogroups II and III are the most abundant genogroups in the raw sewage of the area (45). These results seem to confirm the results of previous reports indicating a higher persistence of genogroups (genotypes) I and II in river (43) and lake (24) water. In addition, these data plus the changes in the relative proportion between indicators point out that, as well as dilution, differential inactivation played an important role in the final conformation of the microbial populations in the area studied.

Data reported here are the first data on phages infecting B.thetaiotaomicron GA17 in seawater. This is a strain that has the ability to discriminate human from animal sources in our geographical area (40). The occurrence and levels of phages detected by this strain can be used for fecal source tracking in bathing areas. They clearly outnumbered F-specific RNA bacteriophages.

Enteroviruses, somatic coliphages, and phages infecting strain GA17 maintained the same proportions as in raw sewage, whereas E. coli and F-specific RNA phages had clearly lower proportions in seawater. Enterococci keep an intermediate position. Seawater data including only fecal coliforms, F-specific RNA phages, and enteroviruses reported in Holland keep an order of abundance similar to the ones reported here (56). These changes in the relative proportions are fully consistent with the die-off rates obtained in the in situ inactivation experiments, in which microorganisms were inactivated quite differently.

Somatic coliphages and bacteriophages infecting Bacteroides thetaiotaomicron were the most persistent among the naturally occurring viruses, whereas F-specific RNA phages were clearly less persistent in in situ inactivation experiments. These results coincide with results reported previously by Chung and Sobsey (4) and Sinton et al. (48). Somatic coliphages and phages infecting Bacteroides were also clearly more persistent than the laboratory-grown enteroviruses tested. However, persistence of enteroviruses may be underestimated, since they were grown in the laboratory. Moreover, other human viruses such as adenoviruses (10) and hepatitis A viruses (4) had been described to be more persistent than polioviruses in seawater, but Chung and Sobsey (4) also reported higher survival of phages infecting Bacteroides fragilis than polioviruses, hepatitis A viruses, or rotaviruses in seawater at 25°C. Therefore, the fact that somatic coliphages and phages infecting Bacteroides persist longer than some enteroviruses in seawater shows that they are good indicators of the more persistent viruses.

Of bacteria, E. coli was the parameter that was inactivated faster, whereas fecal coliforms and enterococci had an intermediate position between somatic coliphages and F-specific RNA phages. A similar order of persistence was previously described for seawater in winter in our area, even though persistence was longer for all parameters (5).

The detection of cultivable viruses at sites in the absence of high bacterial indicator counts reaffirms the need for additional pathogen or new indicator monitoring in order to properly assess human health risks in recreational waters affected by waste disposal. The maintenance of the ratios (numbers in sewage to numbers in bathing water), the ROC curve area values, their densities in bathing water samples, and the similarity of inactivation rates between cultivable enteroviruses and somatic coliphages and bacteriophages of strain GA17 of Bacteroides thetaiotaomicron make both of them good candidates for additional indicators in our area. Because of their densities in bathing waters and simplicity of detection and because their numbers seem very constant worldwide, somatic coliphages seem to the best indicators. Moreover, somatic coliphages can be determined in 4 h, which greatly helps assessment of the validity of predictive models of quality of bathing areas.

This study was supported by grant 2001 SGR 00099 from the Generalitat de Catalunya and CeRBa (Centre de Referència en Biotecnologia de la Generalitat de Catalunya). L.M.-L. was a recipient of a fellowship from the Generalitat de Catalunya.

We thank Susana Calle and Cristina Valdivieso for their excellent technical assistance.