Water, Air, & Soil Pollution

, Volume 211, Issue 1, pp 251–259

Tetracycline-Resistant Escherichia coli in a Small Stream Receiving Fish Hatchery Effluent


  • Matthew Stachowiak
    • Penn State Harrisburg
  • Shirely E. Clark
    • Environmental EngineeringPenn State Harrisburg
    • Penn State Harrisburg
  • Katherine H. Baker
    • Environmental MicrobiologyPenn State Harrisburg

DOI: 10.1007/s11270-009-0296-1

Cite this article as:
Stachowiak, M., Clark, S.E., Templin, R.E. et al. Water Air Soil Pollut (2010) 211: 251. doi:10.1007/s11270-009-0296-1


We examined the impact of the effluent discharged from a freshwater (trout and related species) fish hatchery on the presence of antibiotic-resistant microorganisms in a small stream. There had been no documented use of antibiotics in the hatchery for at least 6 months prior to our study, although a variety of biocides were employed routinely for cleaning. Heterotrophic bacteria and Escherichia coli were isolated from both water column and sediment samples at sites above and below the discharge of the hatchery effluent as well as from the hatchery effluent itself. Randomly chosen isolates (≥96 isolates per site) were tested for their resistance to ampicillin, cephalexin, erythromycin, and tetracycline. Resistance to at least one antibiotic was found in greater than 30% of both the heterotrophic isolates and the E. coli isolates from each of the sites. There were no significant differences among the sites in the proportion of the heterotrophic isolates resistant to any specific antibiotic. The proportion of E. coli isolates resistant to tetracycline in the hatchery effluent and in both the downstream water and sediment samples was significantly higher than in either the upstream water or sediment. These results support the possibility of the hatchery as a source of tetracycline-resistant microorganisms even in the absence of recent use of this antibiotic.


Public healthMicrobiological processesEnvironmental impactsAntibiotic resistance

1 Introduction

The spread of antibiotic-resistant microorganisms in the environment is recognized widely as an important public health issue with physicians concerned about their future ability to treat infectious diseases (Schmidt 2002). For example, methicillin-resistant Staphylococcus aureus now is responsible for more deaths annually in the US than human immunodeficiency virus/acquired immunodeficiency syndrome (Camargo and Gilmore 2008). The useful life span of an antibiotic is limited by the emergence and spread of resistant bacteria with early concerns about resistant bacteria focused primarily on clinical settings and nosocomial infections. Over the last several decades, the focus of research has expanded with the realization that resistant organisms are widespread in the environment. The imprudent use of agricultural—and other food production-associated applications of—antibiotics selects for resistant organisms (da Costa et al. 2008; Wegener 1999; Sørum and L'Abée-Lund 2002). When these organisms enter the environment, the resistance genes they carry can be transferred to other organisms, thus amplifying the possibility of resistance in pathogenic species (Hastings et al. 2004).

Resistant bacteria have been isolated from rivers and streams worldwide (Ash et al. 2002; Schmidt et al. 2000, 2001; Rhodes et al. 2000; Goñi-Urriza et al. 2000; Kelch and Lee 1978; Niemi et al. 1983; Weber et al. 1994). Ash et al. (2002) reported high frequencies of resistance to ampicillin in bacteria isolated from water samples obtained from rivers across the USA. In addition, they found that a high proportion of these isolates (40%) carried plasmids coding for resistance and thus potentially were capable of transferring resistance to other organisms. French et al. (1987) reported on the isolation of high numbers of multiple-antibiotic-resistant Escherichia coli from fecally contaminated streams in Hong Kong. Mukherjee and Chakraborty (2007) found that the majority of resistant bacteria isolated from river water carried R-plasmids, which were transferable to other organisms.

Discharges from wastewater treatment facilities and runoff from animal production and agricultural land are recognized as important sources of both resistant microorganisms and resistance genes (Beier et al. 2008; Kim and Aga 2007; Kim and Carlson 2007; Garcia et al. 2007; Edge and Hill 2007; Ferreira da Silva et al. 2007; Davies 1994; Levy 2005; Levy and Marshall 2004). The commercial production of fish has impacts similar to other animal husbandry, and recent studies have demonstrated the increased incidence of resistance of environmental isolates and fish pathogens associated with these facilities (Petersen et al. 2002; Alcaide et al. 2005; Miranda et al. 2003; Huys et al. 2001). Commercial production of fish for both human consumption and for the enhancement of recreational fishing through stream stocking is rapidly increasing worldwide. Recent studies have indicated a fourfold increase in industrial aquaculture worldwide over the past several decades (Naylor and Burke 2005; Cabello 2006; Goldburg and Naylor 2005).

While the overall use of antibiotics in fish production has increased in recent years, not all facilities routinely use antibiotics. In many instances, use is intermittent, with these agents only employed to treat active disease outbreaks. At such facilities, the time between uses can be long, on the order of months. Even in the absence of continuous usage, there is still the possibility that resistant microorganisms persist in protected reservoirs such as sediments or the intestinal tracts of fish (Schmidt et al. 2000; Petersen et al. 2002; Niemi and Taipalinen 1982). Furthermore, several studies have shown that the use of biocides may select for organisms resistant to these compounds with cross-resistance to antibiotics associated with efflux pumps (Levy 2002). Biocides such as hydrogen peroxide, aldehydes (e.g., formalin), and quaternary ammonium compounds (Roccal) are used during routine maintenance activities in fish production facilities. Therefore, commercial fish production facilities could be a source of antibiotic-resistant microorganisms to receiving waters at times when there is no active use of antibiotics as a result of cross-resistance induced by biocides. This study examines the possibility of long-term discharge of resistant microorganisms from a commercial fish hatchery not currently using antibiotics and its impact on the small stream receiving this discharge.

2 Materials and Methods

The stream studied was the Yellow Breeches Creek in south-central Pennsylvania. This stream, which drains a 567.2-km2 (219 mi2) watershed, is designated a High-Quality Cold Water Fish Stream and a Scenic River by the Commonwealth of Pennsylvania (HRG 2005). This study focused on the upper reaches of the stream in an area where the dominant land use is a mixture of forested and agricultural land and there are no known municipal point source discharges (SRBC 2007). Prior studies on the stream have indicated good overall water quality. A bacteriological survey of the stream (2006) documented generally low concentrations of indicator organisms (E. coli, Enterococcus [sic], and fecal coliforms) with recreational water quality guidelines exceeded only rarely (SRBC 2007). There was no correlation between high discharge levels and high bacteria levels. In fact, the highest bacterial concentrations were associated with the lowest discharge. This, coupled with the lack of nutrient (nitrogen and phosphorous) concentrations typically associated with domestic sewage or agricultural wastes, suggests that septic systems or agricultural and urban runoff are not the primary sources of bacterial contamination (SRBC 2007). Instead, the available information points to a continuous low input of bacteria from an unknown source or sources (SRBC 2007).

The only significant point source discharge into the reach studied is a fish hatchery. The chief fish export of the hatchery is a variety of trout as well other species of fish including striped bass, channel catfish, and tiger muskellunge. Annual production of trout reaches about 915,000 adults (roughly 231,000 kg). Stable isotope (13C) studies have demonstrated an impact of the hatchery discharge on both sediments and biota (macroinvertebrates) for several kilometers downstream (HRG 2005). Discussions with hatchery personnel confirmed the hatchery had used (oxy)tetracycline within the past year, but not within the 6 months previous to the study. In addition, the workers confirmed that several biocides, including hydrogen peroxide and hypochlorous acid, were used in the facility (personal communication). We were unable to ascertain if other antibiotics or biocides were used in the hatchery operations. No studies have specifically examined the microbiological impact of the hatchery discharge on the stream.

Three locations along the stream were sampled for microbiological analysis. The upstream location was approximately 1.6 km above the fish hatchery in an agricultural area. The effluent location was at the confluence of the discharge from the fish hatchery and the stream. Mixing of the receiving water and the hatchery effluent at this location was not complete, and the bulk of the discharge at this sampling location was hatchery effluent. The downstream location was approximately 1 km below the hatchery discharge and well outside the area of active mixing. Therefore, samples obtained from this location were considered indicative of the impact of the hatchery discharge on the stream.

At both the upstream and downstream locations, separate water and sediment samples were aseptically collected. Because of logistical limitations, only water samples were collected at the hatchery discharge. Samples were collected under standardized discharge conditions to prevent the effects of variable stormwater runoff as a sample bias. Specifically, all samples were collected at least 2 days, but not more than 4 days, after a rainfall of between 12.5 and 25 mm. Each site was sampled over a 3-month (summer) period to obtain sufficient isolates (≥96) for comparison. The data presented are based on site-specific composites of the isolates obtained. All samples were placed on ice (4°C) immediately after collection and transported to the laboratory for analysis. Microbiological analysis was conducted within 8 h of sample collection.

Heterotrophic microorganisms were isolated by plating onto R2A agar (BBL, Cockeysville, MD, USA). Plates were incubated for 5 days at 25°C. E. coli was isolated using the membrane filtration technique with mTEC agar (HACH, Loveland, CO, USA) as the selective medium. Plates were incubated for 2 h at 42°C followed by an additional 22 h at 35°C. Isolated colonies of heterotrophs and E. coli were selected randomly and picked from the initial plates using sterile wooden toothpicks. The isolates were cultured on 0.1X trypticase soy broth (0.1X tryptic soy broth (TSB); BBL, Cockeysville, MD, USA) and stored at −80°C (0.1X TSB + 1% glycerol) until they were characterized for antibiotic sensitivity.

Each isolate was tested for sensitivity to four antibiotics (Sigma-Aldrich, Inc.): ampicillin (10 µg ml−1), erythromycin (15 µg ml−1), tetracycline (30 µg ml−1), and cephalexin (30 µg ml−1). Frozen isolates were thawed and transferred to fresh 0.1 TSB. They were cultured overnight at 37°C. Approximately 200 μl of each isolate was transferred to an individual well in a 96-well microplate (Corning Cell Wells) to yield a master plate. Separate master plates were prepared for each location and substrate sampled.

The master plates were used to inoculate separate microtiter plates containing 0.1 TSB supplemented with individual antibiotics. The antibiotic test plates were incubated overnight at 35°C. Individual isolates were scored as sensitive (no growth) or resistant (growth) using the presence of visible turbidity as an indicator.

Because of the categorical nature of the data (resistant vs. not resistant), comparisons among sampling sites were made using contingency analysis (χ2) test. All data analysis was performed on raw data using Prism 4.0 Statistical software (GraphPad, Inc., La Jolla, CA, USA). Percentages were used for graphical presentation to facilitate visual comparisons of sites.

3 Results

3.1 Heterotrophic Microorganisms

The majority of the heterotrophic microorganisms isolated on R2A medium in all water and sediment samples were resistant to at least one antibiotic (Fig. 1). There were no significant differences between the upstream and downstream sites in the proportion of isolates in any of the resistance categories (χ2 = 5.006, df = 8, p = 0.7570). Inclusion of the effluent isolates within the analysis indicated statistically significant differences (χ2 = 19.48, df = 8, p = 0.0125) between the effluent and any of the stream sites with the proportion of the effluent isolates, either showing resistance to multiple antibiotics or to no antibiotics reduced compared to the stream sites, and the proportion of isolates resistant to a single antibiotic increased.
Fig. 1

Antibiotic-resistant heterotrophic microorganisms. Percent of total isolates (N = 131 upstream water, 144 upstream sediment, 98 effluent, 203 downstream water, and 183 downstream sediment) showing resistance to a single antibiotic or to multiple antibiotics. The values shown are the raw data converted into a percent of the total isolates for each site

Within the resistant heterotrophic microorganisms, there were no significant differences (χ2 = 8.64, df = 12, p = 0.73) among the sites in the proportion of isolates resistance to any particular antibiotic (Fig. 2).
Fig. 2

Antibiotic resistance patterns of heterotrophic microorganisms. Percent of resistant isolates per specific antibiotic. The values shown are the raw data converted into a percent of the resistant isolates (N = 71 upstream water, 78 upstream sediment, 73 effluent, 112 downstream water, and 108 downstream sediment). In the case of isolates with resistance to multiple antibiotics, each antibiotic was separately scored as positive and thus, the total for all of the antibiotics at a particular site may exceed 100%

3.2 E. coli

Greater than 40% of the E. coli isolated from water samples was resistant to at least one of the antibiotics (Fig. 3). There were significant differences (χ2 = 51.02, df = 8, p < 0.0001) among the sites sampled. Exclusion of the effluent sample did not remove these differences (χ2 = 38.74, df = 8, p = 0.001). When the upstream samples were removed from the data set, there were no significant differences (χ2 = 3.997, df = 8, p = 0.0857) among the effluent and the downstream samples. Furthermore, there were no significant differences (χ2 = 4.865, df = 8, p = 0.7719) between the upstream water and the upstream sediment samples when these were compared to each other.
Fig. 3

Antibiotic-resistant Escherichia coli. Percent of total isolates (N = 164 upstream water, 171 upstream sediment, 143 effluent, 234 downstream water, and 119 downstream sediment) showing resistance to a single antibiotic or to multiple antibiotics. The values shown are the raw data converted into a percent of the total isolates

There were significant differences (χ2 = 106.2, df = 12, p < 0.0001) in the patterns of resistance among the sampling sites (Fig. 4). As with overall resistance, there were no significant differences between the effluent and the downstream sites (χ2 = 4.592, df = 12, p = 0.97) or between the upstream water and the upstream sediment samples (χ2 = 5.194, df = 12, p = 0.95). Resistance to erythromycin and cephalexin was similar in all of the samples (χ2 = 1.804, df = 4, p = 0.77). Resistance to ampicillin in the upstream samples appeared slightly, but not statistically significantly, higher than in the downstream samples. The reduction in the proportion of isolates resistant to ampicillin in the downstream samples is consistent with both die-off and dilution of the ampicillin-resistant organisms. Only 37% of the E. coli isolates from the hatchery were resistant to ampicillin compared to between 55% and 60% of the isolates from the upstream samples indicating that the hatchery effluent was probably not adding a significant number of these organisms to the stream.
Fig. 4

Antibiotic resistance patterns of Escherichia coli. Percent of resistant isolates per specific antibiotic. The values shown are the raw data converted into a percent of the resistant isolates (N = 72 upstream water, 84 upstream sediment, 100 effluent, 152 downstream water, and 132 downstream sediment). In the case of isolates with resistance to multiple antibiotics, each antibiotic was separately scored as positive and thus, the total for all of the antibiotics at a particular site may exceed 100%

Fewer than 25% of the E. coli isolates from water and sediment samples collected from the upstream site were resistant to tetracycline. In water from both the hatchery effluent and the downstream water and sediment, the proportion of tetracycline-resistant E. coli was dramatically increased. For both of the downstream sampling sites, greater than 60% of the isolates were resistant. All of the antibiotic-resistant isolates from the hatchery were resistant to tetracycline.

4 Discussion

Antibiotic resistance is a growing concern among both clinical practitioners and public health professionals. In recent years, numerous studies have highlighted the importance of environmental reservoirs for resistant microorganisms. These reservoirs may harbor resistant organisms that subsequently can be transferred to other sites. In addition, microorganisms can transfer resistance to previously susceptible organisms via horizontal gene transfer and mobilization.

Resistant microorganisms can enter surface water from domestic sewage (Beier et al. 2008; Ferreira da Silva et al. 2007). In addition, agricultural runoff, particularly runoff associated with animal manure, can introduce large numbers of resistant microorganisms into surface waters (Edge and Hill 2007; Kim and Carlson 2007). The high proportion of the heterotrophic isolates from the upstream site resistant to at least one antibiotic supports the possible input of resistant organisms from agricultural runoff. It is also possible that antibiotic resistance genes from naturally occurring soil microorganisms (e.g., Streptomyces sp.) contributed to the resistant isolates found in the stream (Heuer et al. 2002; Nikolakopoulou et al. 2005, 2008; Zhang et al. 2009). More extensive sampling, including the isolation of organisms on additional media and molecular characterization of the specific resistance genes present in the isolates, is needed to conclusively demonstrate a connection between terrestrial runoff and the microorganisms in the stream.

There was no change in the pattern of resistance in heterotrophic microorganisms isolated from either water or sediment with the introduction of effluent from the hatchery. This could be explained by the small volume of the total discharge contributed by the hatchery. The contribution of the hatchery effluent to the populations of heterotrophic microorganisms in the receiving stream may not have been detectable given the presence of the high background occurrence of resistance.

In contrast to the heterotrophic microorganisms, there were striking differences in the resistance of E. coli isolates associated with the discharge. In the fishery effluent as well as in both water and sediment samples downstream of the discharge, the incidence of tetracycline resistance in the E. coli isolates was dramatically higher than in the upstream isolates. Discussions with personnel at the hatchery indicated that there was no active use of antibiotics at the hatchery during the time of the study, nor had antibiotics been used in the 6 months prior to the study. Thus, our results indicate that commercial fish production facilities can introduce resistant organisms into receiving waters even when there is no active use of antibiotics in the facility.

E. coli usually is found in the intestinal tract of warm-blooded animals such as mammals and birds and normally is not considered a part of the intestinal microbiota of poikilothermic organisms such as fish. Several studies, however, have indicated that E. coli may become established in the gut of trout raised in hatcheries (Del Rio-Rodriquez et al. 1997; Niemi and Taipalinen 1982). E. coli present in pelleted fish food can provide the inoculum needed to establish this organism in the fish intestine and to periodically provide additional inocula (Kerry et al. 1995). Furthermore, Del Rio-Rodriquez et al. (1997) demonstrated the transfer of oxytetracycline-resistance from Aeromonas salmonicida to E. coli within trout intestines. Finally, E. coli excreted by trout may persist in sediments and biofilms within the hatchery providing a regular source of resistant organisms to the receiving water.

In addition to the possible persistence of resistant E. coli within the hatchery, it also is possible that the use of biocides contributed to the presence of antibiotic-resistant organisms. Biocides, such as chlorohexidine and quaternary ammonium compounds, can enhance the function of endogenous multidrug efflux pumps (Levy 2002). These pumps are involved in tetracycline resistance in several species of bacteria including Pseudomonas (Li et al. 1995) and E. coli (Fralich 1996). Thus, the tetracycline-resistant E. coli discharged in the hatchery effluent may be the result of the routine use of biocides in fish production.

The pattern of tetracycline-resistant E. coli found in our study is consistent with the hatchery as a source of organisms. While the high concentration of tetracycline-resistant E. coli in the effluent samples may reflect the incomplete mixing and hence the dominant influence of the hatchery waste at this site, the presence of elevated concentrations of resistant E. coli in water and sediment at the downstream site indicates that the introduced organisms are capable of at least limited persistence in this environment. Persistence of both antibiotics and resistant bacteria can be longer in sediments than in the overlying water. Tetracycline-resistant bacteria isolated from a cattle feedlot lagoon were shown to persist for at least 29 days in sediments in a laboratory microcosm (Engelmann et al. 2006). In addition, several researchers have documented the persistence of antimicrobial substances within marine sediments for months after use of the antibiotic was terminated (Husevag et al. 1991; Kerry et al. 1995, 1996; Gordon et al. 2007). Our results support these observations and underscore the need for long-term monitoring of commercial fish-farming operations to assess their impact on antimicrobial resistance in the environment.

5 Conclusions

This research underscores the widespread occurrence of antibiotic-resistant microorganisms in surface waters and sediments. Even in the upstream area where anthropogenic impact is relatively low, the majority of microbial isolates was resistant to at least one antibiotic. In addition, the continued introduction of tetracycline-resistant organisms from the hatchery to the stream even after a significant time period had elapsed since the use of antibiotics indicates the presence of reservoirs of organisms or unknown sources of resistance. In either case, it may not be appropriate to assume that terminating the use of antibiotics will lead to a rapid decrease in resistant organisms. Neither is it appropriate to rely on information concerning recent use of antibiotics in identifying potential sources of resistant organisms.


We would like to thank the School of Science, Engineering, and Technology at PSH for its support. The authors would also like to thank Ed Spayd for his assistance in collecting samples used for this research, and Danielle Harrow and Robert Currer for their critical reading of this manuscript.

Copyright information

© Springer Science+Business Media B.V. 2009