Rosa Araujo, Departament de Microbiologia, Universitat de Barcelona, Av. Diagonal 645, 08028 Barcelona, Spain. E-mail: firstname.lastname@example.org
Aims: To determine the prevalence of Campylobacter in surface waters of a highly populated Mediterranean area.
Methods and Results: Surface water and wastewater samples were collected from an area in the north-east of Spain during a 2-year study. All the samples were analysed using the MPN method and Multiplex PCR to quantify and identify Campylobacter. It was detected in 82% of the samples from the Llobregat River with a mean of 1·3 MPN 100 ml−1. The lowest counts were obtained in summer. Campylobacter coli was the predominant species in this river. The bacteria were isolated from marsh water but not from seawater samples. The highest counts of campylobacters were found in poultry wastewater where Camp. jejuni was the predominant species, as in urban sewage. In pig slurry, Camp. coli was the only species detected.
Conclusions: Campylobacter jejuni and Camp. coli are present and widely distributed in the surface water of the studied area. The two species co-exist, with Camp. coli being predominant. In river water, campylobacter counts presented a seasonal distribution. No relationship with faecal indicators was found.
Significance and Impact of the Study: This study provides the first data on the occurrence and concentrations of thermotolerant campylobacter species in surface water in a Mediterranean area.
Thermotolerant Campylobacter, especially Campylobacter jejuni and Campylobacter coli, are the most common bacterial species causing gastroenteritis in humans worldwide (WHO, 2000). In Spain, they are the main cause of gastroenteritis in children under age 5 (Perez-Ciordia et al. 2001), and in adults, the number of Campylobacter isolations reported to the Spanish Microbiological Information System has been growing over recent years. In fact, in 2006, the number of Campylobacter isolations was higher than that of Salmonella (Anonymous 2006), which historically has been more prevalent in Spain.
Our study was conducted in a highly populated region near the Mediterranean Sea in northeastern Spain, an area that suffers from water availability problems. Countries bordering the Mediterranean Sea, as well as South Africa, Chile or California, are examples of areas possessing a Mediterranean climate, which typically includes a drought season (summer) followed by a rainy season (autumn). Rivers are greatly affected by this type of climate. Their amounts of water are usually low throughout the year, except during the rainy season, and their water temperature varies depending on the season, being higher in summer. This study is focused on the Llobregat River, one of the most important rivers in our area. It rises in the Pyrenees and flows through agricultural regions and industrial and urban areas over a distance of 170 km. In the lower part of the river, there is a deltaic area that is one of the three most significant humid zones found in Catalonia. Indeed, it is also regarded as significant to the European Union as a whole, containing marshes that are an important bird habitat. The river discharges into the Mediterranean Sea, the beaches of which are intensely used for recreational purposes. The river water is used for agricultural purposes and it is made potable and used for drinking and supplying local industries in Barcelona and its surroundings. It suffers pollution from numerous sources: the large surrounding population, the many local industries and agricultural run-off and farming waste, especially that produced by swine and poultry, the most abundant livestock in this area (Anonymous 2007).
The aim of the study was to establish the densities of thermotolerant Campylobacter and the occurrence of their species in the Llobregat River and in waters related to it, such as seawater and water from the Llobregat Delta marshes. In addition, their relationship with physicochemical parameters and the prevalence of sources of pollution, such as urban sewage, poultry wastewater and pig slurry, which may contribute to both loading and persistence of the pathogen, was determined.
Material and methods
Seventy-two water samples were collected from several sampling points in Catalonia, in north-eastern Spain (Fig. 1): Fifty-five river water samples were collected from the Llobregat River. They were taken from two different sampling sites: one in the location of Abrera de Llobregat and the other in Molins de Rei. Nine marsh samples were collected from the Llobregat Delta, an area impacted by migratory birds. Eight seawater samples were collected from the beach at Gavà-Viladecans, from a depth of 0·5 m down from the surface in an area where the water level was around 1·5 m deep.
Forty-seven wastewater samples were collected: Twenty-eight samples were collected from inflowing raw urban sewage water at the treatment plant of Gava-Viladecans. This plant is in close proximity to urban inputs, so sewage is relatively fresh. Thirteen poultry wastewater samples were collected from the primary sedimentation tanks of two slaughterhouses situated in Amposta and Tortosa, two villages located in southern Catalonia. The origin of the faecal pollution for both sites stems from poultry and turkeys. Finally, six samples of pig slurry were collected from a storage tank at a pig farm located in the village of Arbeca (Catalonia).
Samples were collected in sterile plastic bottles, transported to the laboratory in isothermic containers and analysed within 4 h. Samples from river and urban sewage were taken monthly from April 2004 to December 2006. Poultry sewage water samples were taken from December 2004 to August 2005, pig slurry samples from March 2006 to May 2007, sea samples between July 2006 and May 2007 and marsh samples between December 2005 and December 2006.
Campylobacter jejuni (DSMZ 4688) and Campylobacter coli (DSMZ 4689) were used as controls. Campylobacter strains were stored at −80°C in Brain Heart Infusion (Difco, Sparks, MD, USA) with 20% glycerol. Campylobacter was cultured on Karmali agar (Scharlab, Barcelona, Spain) plates at 42°C for 2 days under microaerobic conditions generated in a jar using the Oxoid CampyGen™ system (Oxoid, Basingstoke, UK).
Detection of Campylobacter
To detect and quantify Campylobacter, all samples were analysed by the most probable number method (MPN) with Preston enrichment broth (Hijnen et al. 2004). A three-tube MPN analysis was carried out. Appropriate tenfold dilution series were tested. When necessary, the water samples were concentrated by centrifugation at 7700 g for 20 min. Each pellet was resuspended in 1 ml of phosphate buffered saline (PBS) and inoculated in the enrichment broth. For river, seawater and marsh samples, volumes from 1000 to 1 ml of water were added to 9-ml tube of Preston broth (Oxoid, Basingstoke, UK). For wastewater (both urban and animal sewage samples), volumes from 10 to 0·0001 ml of water were tested. Tubes were incubated at 42°C for 48 h under microaerobic conditions. A subculture was carried out by streaking a loopful of growth from a Preston broth tube to a plate of Karmali and incubated for 48 h at 42°C under microaerobic conditions. Plates were examined for the presence of Campylobacter colonies. MPN values were obtained from MPN tables. The minimum detection limit was 4 × 10−2 MPN 100 ml−1 for surface water and 4 × 100 × MPN 100 ml−1 for wastewater. A positive control tube culture of Camp. jejuni DSMZ 4688 was added to all MPN analyses.
Campylobacter identification by Multiplex PCR
Identification of Camp. jejuni, Camp. coli and non-jejuni/coli thermotolerant Campylobacter (Camp. spp) was made by the Multiplex PCR as described previously by Wong et al. (2004). The primers sequences used were as follows: Therm 1M Forward 5′-AAATTGGTTAATATTCCAATACCAACATTAG-3′ and Therm 2M Reverse 5′-GGTTTACGGTACGGGCAACATTAG-3′ for the detection of thermotolerant Campylobacter spp., LpxA Forward 5′-CCGAGCTTAAAGCTATGATAGTGGAT-3′ and LpxA Reverse 5′-TCTACTACAACATCGTCACCAAGTTGT-3′ for the detection of Camp. jejuni and CeuE Forward 5′-CATGCCCCTAAGACTTAACGATAAAGTT-3′ and CeuE Reverse 5′-GATTCTAAGCCATTGCCACTTGCTAG-3′ for the detection of Camp. coli. Primers were purchased from Invitrogen (Barcelona, Spain).
Enrichment broths tubes that presented culturable campylobacters were processed by PCR method. Total volume of the tube was centrifuged at 7700 g for 20 min at 4°C, and the pellet was resuspended in 1 ml of PBS. Total DNA was extracted by heat lysis (100°C for 10 min) and thermic shock (10 min at −20°C), followed by centrifugation at 10 000 g for 5 min (Van Eys et al. 1989).
Ten microlitre of the DNA extraction was added directly to the PCR premix. Multiplex PCR conditions were performed as described by Wong et al. (2004). Positive controls for Camp. jejuni DSMZ 4688 and Camp. coli DSMZ 4689 DNA, as well as negative controls DNA, were included.
PCR products were analysed by electrophoresis on a 2% agarose gel (Ecogen, Barcelona, Spain) for approximately 1 h 40 min at 100 V. The gel was stained with ethidium bromide (0·5 μg ml−1), and the resulting amplimers were visualized with an ImageMaster®VDS (Pharmacia Biotech, Barcelona, Spain). A 100-bp DNA Ladder (Promega, Madrid, Spain) was included at both ends of the gel as a molecular weight marker.
Microbial and physical analysis
Water samples were analysed for the following faecal indicators: faecal coliforms (FC), Escherichia coli (EC) and somatic coliphages (SC). Culture media and incubation conditions were as follows: FC agar (Difco, Sparks, USA) incubated at 44 ± 1°C for 24 h for FC; rapid E. coli agar (Sharlab, Barcelona, Spain) incubated at 37 ± 1°C for 24 h for EC; and bacteriophages were analysed according to ISO 10705-2:2000 (Anonymous 2000). Turbidimetry (T) and pH were measured in all samples using a Ratio/xr turbidimeter (Hach Lange France S.A.S, Noisy le Grand, France) and with a Micro pH 2000 pH-meter (Crison Instruments S.A, Barcelona, Spain), respectively. The temperature was recorded for all river, seawater and marsh samples.
Tables were prepared using excel software (Microsoft® EXCEL XP). To calculate the geometric mean, zero counts were substituted by a value that is one significant digit less than the detection limit. All bacterial counts were adjusted by the addition of 1 to convert zeros to positive numbers and then transformed to log10. Statistical analyses were performed using the spss 11.5 and statgraphics plus 5.1 software. A nonparametric Spearman rank-order correlation coefficient was calculated between the different parameters analysed. anova and Kruskal–Wallis tests were also performed to determine differences between microbial counts depending on season. All tests were applied at a 95% level of confidence.
Occurrence of Campylobacter spp. in polluted waters
Campylobacter spp. were quantified simultaneously with faecal indicators, and physicochemical parameters in different aquatic environments affected directly or indirectly by faecal inputs (Table 1). The occurrence of campylobacters varied from 100% in poultry wastewater and pig slurry to 0% in seawater. Table 1 shows the summary statistics for all results.
Table 1. Summary statistics of Campylobacter, faecal indicators and physicochemical parameters by water type
Campylobacter (MPN 100 ml−1)
Faecal coliforms (CFU 100 ml−1)
E. coli (CFU 100 ml−1)
Somatic coliphages (PFU 100 ml−1)
G. mean, Geometric mean; Max, Maximum; Min, Minimum; % positive, Percentage of positive samples; NA, nonanalysed; Temp (°C), Water temperature.
River water n = 55
1·3 × 100
4·1 × 103
1·2 × 103
1·0 × 104
1·1 × 103
7·8 × 104
2·2 × 104
5·4 × 105
<4·0 × 10−2
9·5 × 101
4·9 × 101
8·0 × 102
Marsh water n = 9
5·5 × 10−1
3·6 × 101
2·6 × 101
4·6 × 101
3·8 × 102
7·2 × 102
<4·0 × 10−2
2·0 × 100
5·0 × 10−1
Seawater n = 8
<4·0 × 10−2
2·7 × 100
4·1 × 100
2·9 × 101
1·0 × 101
3·2 × 101
6·0 × 101
1·0 × 100
0·5 × 100
1·0 × 101
Urban sewage water n = 28
2·5 × 102
2·0 × 107
6·7 × 106
5·4 × 106
1·5 × 104
3·8 × 108
9·5 × 107
2·6 × 108
<4·0 × 100
1·5 × 106
5·0 × 104
8·2 × 105
Poultry wastewater n = 13
6·0 × 105
1·0 × 107
8·4 × 106
5·2 × 106
1·1 × 107
5·1 × 107
2·0 × 107
7·4 × 107
9·0 × 103
1·9 × 106
1·9 × 106
1·1 × 106
Pig slurries n = 6
4·9 × 105
3·5 × 107
1·5 × 107
8·6 × 107
2·0 × 104*
4·6 × 106
7·5 × 108
1·7 × 108
1·1 × 109
3·0 × 104
2·4 × 104
2·3 × 106
1·5 × 106
2·6 × 107
6·5 × 103
The Llobregat River was analysed monthly during two years in two sampling points separated by 30 km. A total of 55 samples were analysed, 32 collected in Abrera and 23 in Molins de Rei. Although samples were taken from two sites, all were analysed together because no significant differences between the means of the micro-organisms in the two points were found. Campylobacter was isolated in 82% of the 55 samples and the counts in the samples ranged from <0·4 × 10−2 to 1·1 × 103 MPN 100 ml−1 of water, with a mean of 1·3 MPN 100 ml−1. The means for the faecal coliforms, E. coli and somatic coliphages showed that this fluvial system was moderately polluted (Table 1).
The water from a marsh area of the Llobregat Delta was less faecally polluted, as shown by the means for faecal indicators. Campylobacter was detected in 78% of the samples, with a mean of 0·55 MPN 100 ml−1.
Campylobacter was not isolated from any of the eight seawater samples, where faecal indicators were low even though they were collected near an area with secondary effluent inputs from a wastewater treatment plant and 1·5 km away from the mouth of the Llobregat River. However, DNA of thermotolerant Campylobacter was detected in two samples, one from October and the other from December of 2006 (data not shown).
As expected, urban sewage and wastewaters of zoonotic origin contained very high numbers of faecal indicators. In the urban sewage, Campylobacter was detected in 24 of 28 samples analysed (86%) and the counts varied from <4·0 × 100 to 1·5 × 104 MPN 100 ml−1, with a mean of 2·5 × 102 MPN 100 ml−1. The wastewaters of zoonotic origin contained the highest counts of Campylobacter spp. and the lowest variation. In poultry wastewater, the mean was 6·0 × 105 MPN 100 ml−1 and in pig slurry samples, it was 4·9 × 105 MPN 100 ml−1.
Percentages of thermotolerant Campylobacter species in different types of water
Figure 2 shows the distribution of thermotolerant species identified by multiplex PCR in each matrix analysed. In the Llobregat River, Camp. coli was the most abundant of the thermotolerant species identified. It was detected in 43% of the positive samples alone or mixed with other species. In contrast, Camp. jejuni was detected in 25% of the samples. It was not possible to identify the corresponding thermotolerant species in 54% of the samples. These data show that no single species was predominant in the Llobregat River, in fact, different species co-existed in 20% of the samples. Seven of the nine marsh samples analysed were positive for the presence of thermotolerant Campylobacter. A mixture of species was detected in 14.3% of the marsh samples. As in the river, the dominant species was Camp. coli (43%), followed by Camp. jejuni (28%).
In urban sewage, Camp. jejuni was the most abundant thermotolerant species (48%), followed by Camp. coli (38%). A mixture of Campylobacter species was detected in 4.8% of the samples. The wastewaters of zoonotic origin contained campylobacters in all the samples analysed. These matrices presented a different pattern of Campylobacter species compared with the other environments. One species was dominant, with no mixtures of species detected in any sample. In poultry waste, the dominant species was Camp. jejuni (92%) whereas in pig slurry, the single species was Camp. coli (100%).
Relationships between Campylobacter and both faecal indicators and physical parameters
To determine whether a quantitative relationship existed between the pathogen and the various parameters analysed, multiple correlations were analysed for river water, urban sewage water and poultry wastewater. These correlation studies were not applied to the other matrices as the number of available samples was insufficient for carrying out meaningful statistical analyses.
The Spearman coefficients between all the parameters for each type of water demonstrated a significant correlation between FC and E. coli in Llobregat river water, urban sewage water and poultry wastewater samples, with coefficients of 0·80, 0·71 and 0·66, respectively (data not shown). No relationship was apparent between the numbers of any faecal indicator and Campylobacter, as it is shown in Table 2. However, a moderate correlation (Spearman coefficient of −0·5) between Campylobacter and temperature was observed in the case of Llobregat river samples. The water temperature fluctuated throughout the year with a maximum of 26°C in summer and a minimum of 0·5°C in winter.
Table 2. Correlation matrix of Campylobacter spp. with both faecal indicators and physical parameters in different water type
In each case, the Spearman correlation coefficient is shown.
FC, faecal coliforms; EC, E. coli; SC, somatic coliphages; NTU, Turbidity; Temp (°C), Water temperature; NA, nonanalysed.
*Significance at a P-value < 0·05.
Llobregat river water (n = 55)
Urban sewage water (n = 28)
Poultry wastewater (n = 13)
An analysis of variance was conducted to determine any possible seasonality in Campylobacter counts and faecal indicator counts of the Llobregat River (Fig. 3). The Kruskal–Wallis test showed that Campylobacter counts differed significantly with each season. Counts were higher in winter and autumn than in spring and summer. In contrast, none of the faecal indicators analysed exhibited any variations related to season. The test was also applied to urban sewage microbial counts, with no seasonal difference found (data not shown).
Although Campylobacter is an emergent pathogen in Spain, this is the first study of the densities and distribution of the human pathogenic species, Camp. coli and Camp. jejuni, in aquatic environments and their primary sources of faecal contamination in this country. The study shows the occurrence of thermotolerant campylobacters in the faecally polluted waters of the Llobregat River. The bacteria have been isolated from 82% of the river samples analysed during the 2-year study. In other fluvial systems, the percentage varied between 14% in a river in Finland (Hörman et al. 2004) to 100% in a river in the USA (Vereen et al. 2007). Thus, the Llobregat River contains one of the highest percentages of Campylobacter (Savill et al. 2001; Eyles et al. 2003; Devane et al. 2005). Despite this, most samples yielded low counts (mean of 1·3 MPN 100 ml−1), especially in summer, when many were negative. Campylobacter loads presented a seasonal distribution independent of faecal indicator counts, which remained quite constant throughout the year. Therefore, the low pathogen numbers noted in summer do not reflect any decrease in faecal contamination levels. Most likely, this decrease is caused by environmental factors, such as high sunlight and little rain, as previously reported (Thomas et al. 2002; Vereen et al. 2007). The negative correlation between Campylobacter and temperature in river samples is consistent with this hypothesis.
The presence of thermotolerant species in Llobregat river water was variable, and many samples revealed the co-existence of different species, some of which could not be identified by the method used. Further identification of the isolated strains has since been carried out in our laboratory. Our results indicate that Camp. coli predominates over Camp. jejuni, as Baffone et al. (1995) have previously found in river water from Italy. However, this conflicts with other reports (Carter et al. 1987; Popowski et al. 1997; and Hörman et al. 2004) in which Camp. jejuni was the most abundant species in surface waters.
Campylobacter was detected, but not isolated, in seawater samples only in the cold months. This may be because the sampling station, a beach regularly used for recreational purposes, contained low levels of faecal contamination although it was situated near the mouth of the river and the effluent from a wastewater treatment plant. These results are consistent with the fact that the presence of Campylobacter depends on faecal inputs and environmental conditions (Alonso and Alonso 1993).
The campylobacters found in surface waters could have different origins, one of the most important of which may be wild birds (Casanovas et al. 1995; Obiri-Danso et al. 2001). Samples from Llobregat Delta marshes were included in this study to determine the importance of birds in the contamination levels of the water. Low counts of Campylobacter were present in most samples. The low counts of campylobacters because of the presence of wild birds indicate that these animals might not be the main source of contamination in the water of the studied area. The distribution of thermotolerant species in these samples was similar to that of the Llobregat River, although in this case, the presence of mixed species was less marked than in river samples.
The area surrounding the Llobregat River is densely populated, and urban developments co-exist with industries, farms and agricultural land. Thus, wastewaters from either urban areas or animals could contribute to water contamination. Although by law wastewater must be treated, leaks can infiltrate and pollute river water. Moreover, the controlled use of animal faecal wastes as fertilizers is permitted (Anonymous 2001), which may contribute to water contamination via land run-off. Poultry and pigs are the most numerous livestock, with some 500 000 pigs and 70 000 poultry in the area of the Llobregat River catchment, and these generate large amounts of faecal waste. The highest Campylobacter counts were found in poultry wastewater, which is comparable to other published data (Koenraad et al. 1996). In pig slurry, Campylobacter counts were slightly lower, although their occurrence was higher compared with other studies (Watabe et al. 2003). Campylobacter jejuni was the predominant species isolated from poultry, and Camp. coli was the predominant species isolated from pigs. The thermotolerant species found in each type of wastewater were consistent with previously published studies (Aarestrup et al. 1997; Saenz et al. 2000). The high incidence of Camp. coli in pig slurry and the abundance of these wastes in the studied area may explain the higher proportion of this species in river water.
Another Campylobacter reservoir analysed was urban sewage, which contains faecal matter largely generated by the human populations of this region. In this matrix, Campylobacter was isolated in most of the samples with numbers some 1000-times lower than those of faecal indicators. Contrary to the expected data, urban sewage sample counts did not peak in either summer or autumn, as human enteritis does (Perez-Ciordia et al. 2001; Rodríguez et al. 2001; Luquero et al. 2007). Moreover, the proportion of species found, one Camp. jejuni for each Camp. coli, conflicts with existing clinical data, which has recorded ten Camp. jejuni for each Camp. coli (De Mateo 1997; Kramer et al. 2000; Saenz et al. 2000).
In our study, Campylobacter did not exhibit any correlation with faecal indicators in any of the matrices analysed, as found previously (Carter et al. 1987 and Hörman et al. 2004). Nevertheless, in all analysed samples, faecal indicators exceeded the numbers of the pathogen, thus fulfilling their predictive function.
In conclusion, Camp. jejuni and Camp. coli are widely distributed in the waters of the studied area. The higher loads corresponded to the samples from animal wastes, either poultry or pig slurry, which are probably the principal source of pathogen pollution in the surface waters. Further studies on the molecular biotyping of these strains may elucidate their origin in the sewage. In river waters and possibly in seawater, there is a seasonal distribution of the pathogen but not of faecal indicators. This may be because the occurrence of the Campylobacter species, as well as their concentration, may depend not only on the different contamination sources and on their contribution to superficial water pollution but also upon the pathogen’s ability to survive, which seems to be related to the climatic conditions of the given geographical area much more than to faecal indicators. This work has implications for the control of potential waterborne transmission.
This study was supported by the Generalitat de Catalunya (2005SGR00592) and the “Centre de Referència en Biotecnologia de la Generalitat de Catalunya (CeRBa)”. Sarah Rodríguez is the recipient of a fellowship from the University of Barcelona. We thank Dr B. Gilping from the Christchurch Science Centre of New Zealand and Dr G. Medema from the National Institut of Public Health and the Environment of the Netherlands for their advice about analytical methods. The authors specially thank J.A. Oliva for his kindly help in sampling.