Campylobacter jejuni is responsible for the majority of bacterial foodborne gastroenteritis in the US, usually due to the consumption of undercooked poultry. Research on which avian species transmit the bacterium is limited, especially in the US. We sampled wild birds in three families—Anatidae, Scolopacidae, and Laridae—in eastern North America to determine the prevalence and specific strains of Campylobacter. The overall prevalence of Campylobacter spp. was 9.2% for all wild birds sampled (n = 781). Campylobacter jejuni was the most prevalent species (8.1%), while Campylobacter coli and Campylobacter lari prevalence estimates were low (1.4% and 0.3%, respectively). We used multilocus sequence typing PCR specific to C. jejuni to characterize clonal complexes and sequence types isolated from wild bird samples and detected 13 novel sequence types, along with a clonal complex previously only associated with human disease (ST-658). Wild birds share an increasing amount of habitat with humans as more landscapes become fragmented and developed for human needs. Wild birds are and will remain an important aspect of public health due to their ability to carry and disperse emerging zoonotic pathogens or their arthropod vectors. As basic information such as prevalence is limited or lacking from a majority of wild birds in the US, this study provides further insight into Campylobacter epidemiology, host preference, and strain characterization of C. jejuni.
Three species of Campylobacter, Campylobacter jejuni, Campylobacter coli, and Campylobacter lari, are of greatest concern to the poultry industry (Evans and Powell 2008). All three species can be isolated from poultry, but the most pathogenic in terms of human disease is C. jejuni (Evans and Powell 2008). Campylobacter jejuni is responsible for the majority of bacterial foodborne gastroenteritis in the US (Abulreesh et al. 2006), and infections have been linked to increased risk for Crohn's disease, ulcerative colitis, and other conditions such as reactive arthritis (Hill et al. 2003; Kahng 2009; Nielsen et al. 2009; Schonberg-Norio et al. 2009).
Campylobacter bacteria can exist in the avian gut as commensal organisms and are a component of the natural microbiota of some bird species (Joens 2004). In mature poultry, ducks, and turkeys, Campylobacter infections are nonpathogenic and generally asymptomatic (Wright et al. 2008), making detection of the bacterium challenging and therefore posing a potential threat to consumers (Newell and Fearnley 2003). In the US, little research has been conducted on wild avian species that have the potential for transmitting Campylobacter spp., yet this could be an important pathway for transmission. Keller et al. (2011) evaluated the prevalence of Campylobacter spp. in a subset of wild bird species, including herons, waterfowl, raptors, gulls, and various songbirds and found that C. jejuni prevalence was greatest in the Corvidae and Laridae. Yogasundram et al. (1989) estimated C. jejuni prevalence in 13 orders of wild birds in the US and found that the Psittaciformes, Galliformes, and Anseriformes had the highest prevalence. In Japan, Ito et al. (1988) found high C. jejuni prevalence in Japanese crows (34%), with an overall prevalence of 14% in all sampled bird species. Waldenström et al. (2002) found Campylobacter in 21.6% of migrating birds in various taxa, while Ramos et al. (2010) isolated Campylobacter from 10% of sampled gull chicks. The vast majority of these studies occurred outside the US, further supporting the need to establish Campylobacter spp. prevalence in wild birds in the US.
We determined the prevalence of Campylobacter jejuni, C. coli, and C. lari in wild birds (Anatidae [waterfowl], Laridae [gulls], and Scolopacidae [shorebirds]) to identify DNA sequence similarities and differences in clonal complexes (CCs) and sequence types (STs) through multilocus sequence typing polymerase chain reaction (MLST-PCR). Information from this study contributes to the expansion of the global Campylobacter isolate database developed by Keith Jolley, sited at the University of Oxford (Jolley and Maiden 2010), giving further insight into epidemiology and how it relates to specific species of wild birds. Gulls are efficient carriers of zoonotic pathogens because they are attracted to altered human environments, including garbage dumps, untreated sewage, and manure (Reed et al. 2003). Canada Geese (Branta canadensis), and Snow Geese (Chen caerulescens) can also be found in proximity to human habitat, as well as on agricultural fields, predisposing them for transmission of zoonotic pathogens to humans. Shorebirds can carry high Campylobacter loads (79.6%) in Europe (Waldenström et al. 2002), but nothing is known about the prevalence of Campylobacter in these birds in North America.
MATERIALS AND METHODS
We collected wild bird fecal and cloacal swab samples at 26 sites throughout Delaware, Pennsylvania, and New Jersey from July 2009 to July 2011, using standardized procedures similar to those of Keller et al. (2011). We identified large, single-species flocks to sample adult geese and gulls and observed them prior to flushing to ensure we collected fresh fecal samples. We used cloacal swabbing to collect samples from hatch-year geese, gulls, and shorebirds while the birds were being handled by New Jersey Fish and Wildlife Services (NJ FWS) officials or the Delaware Shorebird Project, respectively.
We collected cloacal or fecal samples from Snow Geese, Canada Geese, Ring-billed Gulls (Larus delawarensis), Herring Gulls (Larus argentatus), Greater Black-backed Gulls (Larus marinus), Laughing Gulls (Leucophaeus atricilla), Ruddy Turnstones (Arenaria interpres), Red Knots (Calidris canutus), Sanderling (Calidris alba), and Semipalmated Sandpipers (Calidris pusilla). We collected all Snow Goose samples from one location in Pennsylvania on several sampling occasions during late winter (February–March). We collected Canada Goose samples from flocks in agriculture fields and ponds near human habitat, as well as during banding drives during the summer (June) with NJ FWS. We collected adult gull samples from urbanized habitats such as parking lots and from beaches. We collected hatch-year gull samples with NJ FWS during the summer (July) on several sampling occasions. We analyzed gull species collectively given the number of samples per species (Table 1).
We collected all samples with sterile standard cotton swabs and immediately placed them in 1.5-mL screw-cap microcentrifuge tubes containing a specialized medium (2x LB containing tryptone, NaCl, and yeast extract, with 2x salt containing K2HPO4, KH2PO4, Na-citrate, MgSO4 × 7H2O, [NH4]2SO4, with 87% glycerol; Gergen et al. 1979). We stored the microcentrifuge tubes containing the swabs on ice (4–7 C) until transferred to a −70 C freezer. For each sample, we assigned an identification number, a record of age or sex if known, and date and location the sample was taken.
We cultivated samples by vortexing and plating on Campylobacter-selective blood-free charcoal-based agar plates supplemented with cefoperazone and amphotericin B (Oxoid, Manchester, Connecticut, USA). Plates were incubated 48 hr at 42 C in a microaerobic environment (85% N, 10% CO2, 5% O2) using CampyGen Packs (Oxoid) and BBL Campy EZ Gas Paks (BD, Sparks, Maryland, USA) in AnaeroJars (Oxoid) and GasPak Anaerobic Systems Jars (BD). Bacteria that showed characteristic Campylobacter colony growth, were positive for catalase and oxidase reactions, and could not grow in oxygen-rich environments were observed under the light microscope. Gram-negative curved rods were preliminarily considered to be Campylobacter species and were further isolated on 5% sheep blood tryptic soy agar (TSA) plates incubated microaerobically at 37 C for 24–48 hr. Isolates from these plates were frozen at −70 C in the collection medium.
DNA extraction and PCR
We thawed culture isolated samples and grew them on 5% sheep blood TSA plates. We prepared overnight cultures by scraping the 5% sheep blood TSA plates into a test tube containing 2 mL of Mueller-Hinton broth (BD) and incubating overnight at 42 C. DNA was extracted from bacterial cells using the Qiagen Gentra Puregene Yeast/Bacterial Kit for Gram-negative bacteria (Qiagen, Germantown, Maryland, USA). Following DNA extraction, samples were stored at −70 C.
We diluted extracted DNA to approximately 25 ng/µL and used multiplex PCR (MX-PCR; Vandamme et al. 1997) for the initial discrimination of C. jejuni and C. coli. We used 25-µL reactions using 1 µL of template DNA (∼25 ng/µL), 5 µL of 5X Q solution, 2.5 µL of MgCl2 (25 mM), and 2.5 µL of 10X PCR buffer containing 15 mM MgCl2, 1 µL of dNTPs (10 mM), 1 µL each of the C. coli and C. jejuni forward and reverse primers (10 µM), 0.1 µL of Taq DNA polymerase (5 U/µL), and 8.9 µL of double-distilled (dd) H2O for each reaction. All reagents used were from the Qiagen Taq PCR core kit, and the initial denaturization step was 1 min at 94 C. We used a PCR developed by Lawson et al. (1998) for detection of C. lari on samples that did not show appropriate bands at 350 bp for C. coli. We used 25-µL reactions using 1 µL of template DNA (∼25 ng/µL), 5 µL of 5X Q solution, 2.5 µL MgCl2 (25 mM), and 2.5 µL of 10X PCR buffer containing 15 mM MgCl2, 0.5 µL of dNTPs (10 mM), 2 µL each of the C. lari forward and reverse primer (10 µM), 0.1 µL of Taq DNA polymerase (5 U/µL), and 9.4 µL of dd H2O for each reaction. All reagents used were from the Qiagen Taq PCR core kit, and the initial denaturization step was 1 min at 94 C. We ran 1–2% agarose gels to visualize PCR products.
We used MLST PCR for C. jejuni on the samples that were negative for both the C. coli or C. lari PCRs or were positive for C. jejuni with the MX-PCR. For C. jejuni, the MLST scheme uses DNA sequences of internal fragments of seven housekeeping genes (Dingle et al. 2001). We used protocols (including primers) from the Campylobacter Multi Locus Sequence Typing website (Jolley and Maiden 2010). Reaction conditions were that of denaturization for 1 min at 94 C, annealing for 1 min at 50 C, and extension for 1 min at 72 C. This was followed by a final extension at 72 C for 10 min. We ran 25-µL reactions using 1 µL of template DNA (∼25 ng/µL), 5 µL of 5X Q solution, and 2.5 µL of 10X PCR buffer containing 15 mM MgCl2, 0.5 µL of dNTPs (10 mM), 2 µL each of the designated housekeeping gene's forward and reverse primer (10 µM), 0.1 µL of Platinum Taq (Invitrogen, Carlsbad, California, USA) (5 U/µL), and 11.9 µL of dd H2O for each reaction. Starred primers as indicated by the database usually give the best results and were used initially (Jolley and Maiden 2010). If a sample was not positive, the alternatively listed primers were used for each of the housekeeping genes. We visualized PCR products with 1–2% agarose gels.
PCR purification and sequencing
We purified samples for which the PCR product was positive for all seven housekeeping genes using a QiaQuick PCR purification system employing spin-columns (Qiagen). Between 7 and 7.5 µL of PCR product was used. Following PCR purification, we stored samples at −70 C. Purified samples and sequencing primers were sent to the University of Pennsylvania School of Medicine (Philadelphia, Pennsylvania, USA) DNA Sequencing Facility for DNA sequencing. Samples were run with the starred and double-starred primers, which usually give the best results, or have been shown to be more efficient with wild bird samples. If a sample did not amplify, the alternatively listed primers were used for each of the housekeeping genes as needed. For each sample, seven sequencing reactions were performed, each corresponding to one of the seven housekeeping genes. A distinct allele was assigned for each housekeeping gene. The most recent database query was performed on 20 February 2012.
We detected Campylobacter spp. in 17% of wild bird samples (n = 781) through culturing methods (Table 2). Using PCR-based methods, all species except Greater Black-backed Gulls were positive for Campylobacter. Prevalence of C. coli and C. lari was low in these wild bird samples (0.9% and 0.3%, n = 781). Snow Geese had the greatest prevalence of C. coli (6.3%, n = 111); Herring Gulls were the only species to carry C. lari (3%, n = 66).
Across all samples, 8.1% (n = 781) were positive for C. jejuni, and 10.6% amplified at all seven C. jejuni housekeeping genes but did not provide full sequences at all seven loci. We detected 17% C. jejuni prevalence in Snow Geese (n = 111), amplifying at all seven housekeeping genes for the C. jejuni MLST scheme. All except two Snow Goose samples gave complete ST profiles when sequenced. We detected a 4.7% C. jejuni prevalence in Canada Geese (n = 274), and most samples gave complete ST profiles. Forty-six percent of the positive Canada Goose samples were isolated from hatch-year geese, 15% from adults, and 39% from Canada Geese of unknown age. We detected 6.3% C. jejuni prevalence in gulls (n = 127), and all but one sample provided sequences for all seven housekeeping genes. Half of these positive samples were from hatch-year birds. We detected 25% C. jejuni prevalence in Ruddy Turnstones (n = 124), and 61% of these samples provided complete ST profiles. We detected an 11% C. jejuni prevalence in Semipalmated Sandpipers (n = 62), and five of the seven positive samples gave complete ST profiles (Table 3).
Seventy-eight percent of culture-positive Campylobacter spp. samples were positive for C. jejuni by MX-PCR, while full STs were obtained from 76% of positive C. jejuni samples. Of these, 34 were assigned a clonal complex, and 63 were at least assigned a ST, 29 of which had an unknown CC. More than one third of the samples (n = 25) did not yet have a ST assigned and were sorted into 13 novel STs by the curator of the Campylobacter Multi Locus Sequence Typing website (Table 3).
The most common CC was the ST-1287 complex, followed by the ST-1275 and ST-1034 complexes. Two Snow Goose samples were isolated from the ST-45 complex, and one Snow Goose sample was isolated from the ST-658 complex. For samples where no CC is currently assigned, ST-5705 was the most common, followed by ST-710. Sequence types 691, ST 693, ST 2514, and ST 3961 each occurred in two samples. Sequence types 2349, ST 3112, and ST 4535 each occurred in one sample. Sequence types 5879, ST 5880, ST 5881, ST 5883, ST 5889, and ST 5890 each occurred in one sample and had never been isolated before this study.
The overall prevalence of Campylobacter spp. for all wild birds in this study was 9.2%, similar to previously conducted studies investigating wild birds in the US (Yogasundram et al. 1989; Keller et al. 2011). Keller et al. (2011) estimated C. jejuni prevalence at 7.2% from six orders, while Yogasundram et al. (1989) estimated a 10.1% prevalence of C. jejuni from 13 orders of wild birds. In this study, samples were heterogeneous in terms of prevalence at all levels, including order, family, genus, and species, ranging from 0 to 43.1%. Campylobacter jejuni was the most prevalent species of Campylobacter, while C. coli and C. lari prevalence estimates were low (8.1%, 1.4%, 0.3%, respectively).
There was an uneven age distribution in samples analyzed among species, but also within species. Hatch-year Canada Geese and gulls had a higher prevalence of Campylobacter spp. than after-hatch-year birds. Dobbin et al. (2005) found Campylobacter in a subset of 2–3-wk-old Double-crested Cormorant chicks (Phalacrocorax auritus). Double-crested Cormorants nest in dense colonies similar to those of gulls, where bacteria may be easily shared. Hatch-year birds may come into contact with different bacteria such as Campylobacter as their immune system is developing, suggesting a higher prevalence in this age group.
Birds in the Anatidae had >27% prevalence in European studies (Waldenström 2005) and specifically 15% in Canada Geese in Europe (Wahlstrom et al. 2003). In contrast, migrating Graylag Geese (Anser anser) in the Netherlands had low Campylobacter prevalence (Lillehaug et al. 2005). In North America, Canada Geese harbor C. jejuni infrequently, as Pacha et al. (1988) and Keller et al. (2011) described (5% and 0%, respectively). Snow Geese, a migratory goose species, had the highest C. jejuni prevalence rate of geese in our study, while the resident Canada Geese samples had a much lower prevalence rate. Seasonal variation of Campylobacter abundance could have played a role in these differences (Taylor et al. 2013). Besides varying prevalence among species due to migration status, temporal components such as seasonal variation in Campylobacter abundance could also have played a role in the patterns we detected (Broman et al. 2002; Taylor et al. 2013). We sampled Snow Geese in winter and Canada Geese throughout the year, with the majority of Canada Goose samples being collected during the summer.
All shorebird species sampled in this study were migratory. Ruddy Turnstones undergo long-distance complete migrations with no resident populations (Harrington and Flowers 1996), and they had the highest prevalence of C. jejuni (15.3%, n = 124). Hughes et al. (2009) found that European Ruddy Turnstones had a 0% (n = 27) prevalence. This may be because these birds are a different subspecies and have a different migration route. Like migratory Snow Geese, migratory shorebird species tended to have a higher Campylobacter prevalence than nonmigratory or less migratory wild bird species in this study. In addition, small flocks of Ruddy Turnstones often intermingle with flocks of Red Knots (Calidris canutus), Semipalmated Sandpipers, and Sanderling (Harrington and Flowers 1996). This interspecific flocking behavior offers opportunities for bacterial exchange on stopover sites such as the Delaware Bay, where these samples were taken. Semipalmated Sandpipers and Red Knots had the second highest prevalences of C. jejuni (8.1% and 7%, respectively). Semipalmated Sandpipers also undergo long-distance migrations between North and South America, with a limited number of stopover sites (Rodrigues 2000). Red Knots, one of the longest-distance migrating birds, make prolonged stops at migration staging sites to prepare for extreme long-distance flights (Piersma 1987). Differences in prevalence may relate to migration route, frequency and use of stopover sites, feeding behavior, ability to encounter the pathogen, species physiology, and exposure rate. We did not detect Campylobacter species in Sanderling, even though these birds share the same stopover points and occur in proximity to other species. There is a large range in Sanderling migration distances and durations as compared to the other shorebird species sampled (Myers et al. 1985). Opposite effects have been shown in songbirds (Passeriformes), where Campylobacter prevalence was higher in short-distance migrants (17.2%) and lower in long-distance migrants (31.5%) (Sensale et al. 2006). This may be attributed to differences in physiology, as shorebird migrations tend to be much more strenuous for individuals with fewer stopover points. Therefore, higher stress during the migration event could cause greater susceptibility to opportunistic commensal organisms. Greater prevalence of Campylobacter in shorebirds may also indicate a strong commensal relationship with the bacteria (Hughes et al. 2009).
There was a large variety of STs and CCs present from the two orders and three families of wild birds in this study, indicating that wild birds potentially harbor a wide array of pathogenic Campylobacter. Overall 5 CCs and 37 assigned STs, 13 of which were previously unassigned, were isolated. The majority of CCs we isolated had other wild bird or environmental water samples associated with them. It is generally thought that C. jejuni isolates that had only been isolated from wild birds may not be involved in human disease (Hughes et al. 2009). However, as isolates have been added to the database, human disease has been associated with complexes thought to only be associated with wild birds (Hughes et al. 2009; Sheppard et al. 2011). The ST-45 complex and the ST-658 complex, isolated from Snow Geese, are characteristic CCs associated with human disease. The ST-658 complex has only been isolated from human stool in Europe. As sample size increases in the database, wild bird samples are also being added to these complexes, also suggesting that wild birds are exposed to some strains thought to be exclusive to humans. Lu et al. (2009) argued that waterfowl can amplify and possibly transmit pathogens to humans by directly contaminating agricultural fields or surface waters. However, this process works both ways. Agricultural fields contaminated with untreated manure, for example, can directly infect wildlife. The ST-45 complex currently contains 464 isolates, approximately 55% of which have been isolated from human disease incidents. Two Snow Goose samples were isolated from this CC, which has been isolated from Macaroni Penguins (Eudyptes chrysolophus), even though C. jejuni persistence is rare among Antarctic animals (Griekspoor et al. 2009). This suggests that the strain could have been introduced through human disturbance. This supports the assertion that humans and poultry operations are contaminating wildlife with some human-dominated Campylobacter lineages.
The ST-1287 complex, which included Campylobacter from Ruddy Turnstones, Red Knots, and Semipalmated Sandpipers, had previously been isolated from other shorebirds throughout the world, such as Dunlin (Calidris alpina) and Sharp-tailed Sandpipers (Calidris acuminata). The same scheme holds for the ST-1275 complex, mainly isolated from gulls (Black-headed Gulls [Chroicocephalus ridibundus]), and the ST-1034 complex, isolated from geese. In our study, the ST-1034 complex was isolated from Snow Geese and Canada Geese, while it had previously been isolated mostly from unnamed goose species and a Snow Goose in Canada. This suggests a tight taxonomic relationship between specific C. jejuni clonal complexes and their preferred host, possibly even coevolution of commensal strains.
Thirteen STs, comprising 25 samples or 39.7% of all STs, were novel STs not previously isolated. This is similar to studies evaluating C. jejuni strains in wild birds (Colles et al. 2008; French et al. 2009; Hughes et al. 2009). The Snow Goose and Canada Goose samples amplified well with the developed PCR and sequencing primers, while amplifying and obtaining usable DNA sequences from these samples was more challenging. Several species of geese have domesticated varieties. Therefore, Campylobacter spp. typing may be more efficient with goose samples (Griekspoor et al. 2013). Like gulls, geese tend to come into more contact with humans and have the potential to pick up human-specific strains more frequently. Overall, however, Campylobacter prevalence in gulls was low and confined to a single CC (ST-1275), suggesting that, at least in terms of spreading the enteropathogen Campylobacter, after-hatch-year gulls are of less concern than previously suggested.
Wild birds share an increasing amount of habitat with humans as landscapes become fragmented and developed for human needs. Wild birds will remain important to public health because of their ability to carry emerging zoonotic pathogens or aid in the dispersal of arthropod vectors (Reed et al. 2003). Although wild birds have the potential to spread disease organisms such as Campylobacter, humans are also leaking pathogens into the environment that have the potential for infecting wild birds. Even though some groups of birds such as shorebirds tend to carry their own commensals and unique STs and CCs, ultimately finding human disease associated CCs in these birds could be used as an indicator for further spread of some pathogens into the environment. As basic information such as prevalence is still not available from a great majority of wild bird species in the US, our study provides further insight into Campylobacter epidemiology, host preference, and strain characterization of C. jejuni.
We thank the University of Delaware Research Foundation, the University of Delaware Department of Entomology and Wildlife Ecology, the University of Delaware Department of Animal Science, and the Avian Bioscience Center. We especially thank Jack Gelb for supporting this project. This study would not have been possible without the help of the Delaware Shorebird Project, the Pennsylvania Game Commission, New Jersey Fish and Wildlife Services, and several private land owners that allowed us to take samples. We used the Campylobacter Multi Locus Sequence Typing website (http://pubmlst.org/Campylobacter/) developed by Keith Jolley and sited at the University of Oxford (Jolley and Maiden 2010, BMC Bioinformatics, 11:595). The development of this site has been funded by the Wellcome Trust. Thanks go to D. Delaney, V. D'Amico, and J. Bowman for their valuable input throughout this study.