Species diversity of campylobacteria in a wild bird community in Sweden


J. Waldenström, Section for Zoonotic Ecology and Epidemiology, Department of Biology and Environmental Sciences, Kalmar University, SE 391 82 Kalmar, Sweden. E-mail: jonas.waldenstrom@hik.se


Aims:  To analyse the occurrence and host species distribution of campylobacteria species in shorebirds, geese and cattle on grazed coastal meadows in Sweden.

Methods and Results:  Species identification was performed through a polyphasic approach, incorporating Amplified Fragment Length Polymorphism (AFLP) profiling, 16S RNA gene sequence analysis together with extensive phenotypic characterization. From 247 sampled birds and 71 cattle, we retrieved 113 urease positive thermophilic Campylobacter (UPTC) and 16 Campylobacter jejuni ssp. jejuni isolates. Furthermore, 18 isolates of Helicobacter canadensis, and five isolates that potentially represent a new genus of micro-aerophilic, spiral and Gram-negative bacteria were isolated. The distribution of bacterial species on hosts was uneven: all H. canadensis isolates were retrieved from geese, while all but one of the Campylobacter lari UPTC isolates were found in shorebirds. AFLP type distribution of Camp. lari UPTC isolates among individual, resampled and breeding-paired Redshank birds generally indicated a constant shift in strain populations over time and absence of geographical clustering.

Conclusions:  The large number of isolated campylobacteria, including species that are zoonotic enteropathogens, indicates that these wild birds potentially may serve as reservoirs of human infections. However, despite a common environment, the different host species largely carried their own campylobacteria populations, indicating that cross-species transmission is rare.

Significance and Impact of the Study:  Our study is one of few that provide data on the occurrence of campylobacteria in wild animals, adding information on the ecology and epidemiology of micro-organisms that are of public health concern.


The term ‘campylobacteria’ has been used to refer to the genera Campylobacter, Helicobacter, Arcobacter, Lawsonia and Anaerobiospirillum, in light of their morphological, physiological and ecological similarities rather than phylogenetic relationships (On 1996). Several species from this group of bacteria affect human and animal health. For example, Campylobacter jejuni ssp. jejuni is the major cause of bacterial diarrhea in humans, Campylobacter fetus ssp. fetus causes bovine abortion, Helicobacter pylori is the causative agent for gastritis and gastric lymphoma in humans, and Anaerobiospirillum succiniciproducens can cause septicemia in immunocompromized patients (Skirrow 1994; Altekruse et al. 1999).

Epidemiological knowledge is scarce for most species of campylobacteria, but generally they seem to have zoonotic potential. Several species have been isolated both from humans and animal hosts, e.g. Camp. jejuni ssp. jejuni, Campylobacter coli, Campylobacter lari, Camp. fetus ssp. fetus, Campylobacter upsaliensis, Arcobacter butzleri, Arcobacter cryaerophilus, H. pylori, Helicobacter pullorum, Helicobacter cinaedi and An. succiniciproducens (Skirrow 1994; On 1996). The occurrence of campylobacteria in wild animals is, however, probably understated, as few attempts of identifying animal reservoirs have been conducted. By far most attention has been paid to Camp. jejuni, particularly the occurrence of this bacterium in wild birds and the subsequent public health ramifications (Hubalek 1994; Broman et al. 2002; Waldenström et al. 2002).

We have previously studied the occurrence and host species distribution of thermophilic Campylobacter species in migrating birds in Sweden, where a high prevalence of Camp. lari and Camp. jejuni ssp. jejuni was found in several bird species, notably thrushes and shorebirds (Waldenström et al. 2002; Broman et al. 2004). Here, we investigate these aspects during the birds’ breeding season. We sampled breeding shorebirds (of the Scolopacidae and Charadriidae families) and geese (family Anseridae) stopping over at some grazed coastal meadows on the island of Gotland, SE Sweden. Additional samples were collected from cattle in one of the pastures. Isolates were identified through a stratified polyphasic approach that used Amplified Fragment Length Polymorphism (AFLP) profiling as the front-line identification and genotyping method, the efficacy of which for both purposes is well documented (On and Harrington 2000). Species identifications were supported by extensive phenotypic analysis and 16S rRNA gene sequence comparisons where prudent. Our findings are presented from both epidemiological and ecological perspectives.

Materials and methods

Sampling and isolation

Fieldwork was conducted at five locations on the southern part of the Swedish island of Gotland in late spring and early summer 2001. The name and location of the investigated meadows can be seen in Fig. 1. The coastal meadows studied were grazed by cattle from late May until October. The meadows were under a regime of high grazing pressure, resulting in low grass vegetation (0–10 cm). In general, the lower parts of these meadows (near the shoreline) are flooded in winter, and rarely also in spring and summer. Pools are filled with water in winter and early spring, or by heavy rainfall, and are often not dried out until June or July.

Figure 1.

 Map of Scandinavia and the island of Gotland. The study locations are noted in the figures: (a) Närsholmen, (b) Stapeludden, (c) Aurriv, (d) Eastern Näsudden and (e) Western Näsudden.

During May–June 2001, adult Redshanks (Tringa totanus, n = 122; of which ten samples came from resampled individuals), Dunlins (Calidris alpina schinzii, n = 4), Lapwings (Vanellus vanellus, n = 3), Common Sandpiper (Actites hypoleucos, n = 1) and Common Snipe (Gallinago gallinago, n = 1) were captured in a specially constructed trap when they visited their nests. Samples were collected by insertion of a sterile swab approximately 1 cm into the cloacae of the birds. The Redshanks in this study were monitored as part of an investigation on breeding performance and habitat utilization of breeding shorebirds (Ottvall 2004). Trapped birds were therefore individually banded with colour bands and followed during the whole breeding season. Thus, information such as which individuals that made up pairs was known and was treated as epidemiological data in the analyses.

Apart from the breeding birds, samples were also collected from a flock of Barnacle Geese (Branta leucopsis) that, for several weeks in May 2001, used two of the meadows (Western and Eastern Näsudden) as staging sites during their northbound spring migration. Geese samples (n = 116) were obtained by collection of fresh droppings with sterile swabs. To get additional information on campylobacteria in this environment, we also collected 71 samples from grazing cattle from Eastern Näsudden. For the cattle, 11–20 samples were collected at five occasions by collecting fresh fecal matter with sterile swabs. Droppings from the same animal might have been taken at different sampling occasions, as the herd consisted of c. 25 animals.

The samples from shorebirds, geese and cows were stored in charcoal transport medium (Transwab, BioDisc, Solna, Sweden) at +4–8°C until cultured in the lab. The number of days between sampling and primary cultivation at the lab varied from zero to five, with a median value of 2 days. We tested whether transport longevity had any effect on culturability with likelihood ratio tests, when applicable.

Each sample was plated onto a Campylobacter-selective, blood-free medium [containing 45·5 g l−1 of Campylobacter-selective agar base LAB 112 (Lab M, Bury, England) and two ampoules of cefoperazone–amphotericin X 112 per 1000 ml (Lab M)]. Incubation was performed at +42°C in a micro-aerobic atmosphere (85% N2, 10% CO2, 5% O2) for 72 h, at which time the media were examined for bacterial growth. One colony per positive plate was used for further identification procedures. However, we included also atypical isolates, i.e. those which differed slightly in colony – and cell morphology to that of normal campylobacters. A limited phenotypic characterization based on cell morphology, and reactions in oxidase, catalase and hydrolysis tests was made to identify isolates as putative Camp. jejuni, or Campylobacter or non-Campylobacter spp.

Species identification

Chromosomal DNA was purified from bacterial cultures using the Puregene DNA Isolation Kit (Gentra Systems, MS, USA) according to the manufacturer's specifications. AFLP profiles were determined by digestion of 25 ng bacterial DNA with BglII and Csp6I, ligation of adapters and subsequent PCR amplification, and detection of fluorophore-labelled fragments by use of an ABI 377 GeneScan sequencer (Applied Biosystems, Foster City, CA, USA), as described in detail previously (On and Harrington 2000). Presumptive species identification of strains was preformed by comparison of the bird isolate AFLP profile with those in a database containing patterns from all extant species and subspecies of the genus Campylobacter (On and Harrington 2000) using the program BioNumerics v. 2·5 (Applied Maths, Kortrijk, Belgium). In this way, the identity of a strain was resolved according to which species of Campylobacter they clustered together with in the resulting UPMGA-tree.

To further evaluate the taxonomic status of the isolates, additional genotypic and phenotypic analyses were performed on selected isolates. First, we determined the 16S rRNA sequence from representatives of each species cluster by methods described previously (Angen et al. 1998; On et al. 1998). However, primer 1392r was replaced with primer 1509rx (5′-GTTACCTTGTTACGACTTACA-3′) in the initial PCR. The 16S rRNA sequences of the selected bird isolates were compared with sequences representing all named Epsilobacteria, and organisms that closely resemble them, using methods described previously (On 2001). Second, extensive phenotypic analysis was performed on selected isolates from different clusters with over 60 standardized tests included in a probabilistic identification system for Campylobacter, Arcobacter and Helicobacter (On et al. 1996). The identity of putative Camp. jejuni isolates, i.e. those giving a positive reaction in the hippurate hydrolysis test, was confirmed by use of a multiplex PCR assay (20) for species identification of Camp. coli and Camp. jejuni, using conditions validated previously (On and Jordan 2003), or by AFLP analysis as described above.


Species identification

Of the 318 collected faecal samples, 152 were found to be culture-positive for campylobacteria (Table 1). The numerical analysis of AFLP profiles identified several clusters in the dendrogram where either the bird isolates grouped separately or together with profiles from known bacterial species. Two of the clusters involved profiles from characterized species of the genus Campylobacter.

Table 1.   Isolation of campylobacteria species
 No. investigatedCampylobacter jejuni ssp. jejuniCampylobacter lari UPTCHelicobacter canadensisAvispirillum spp.
  1. UPTC, Urease positive thermophilic Campylobacter.

Barnacle Goose1166185
Common Sandpiper11 
Common Snipe1 

The species identity of 16 isolates in the first cluster, comprized of nine isolates from cattle, six from geese and one from a redshank, was confirmed as Camp. jejuni based on the results of a multiplex PCR (Vandamme et al. 1997). One isolate (C59) did not yield a product in this assay, nor in additional PCR tests for Camp. jejuni or Camp. coli based on the hipO and putative aspartokinase genes, respectively. Although unusual, negative results for these highly specific PCR assays have been described previously (On and Jordan 2003) and the AFLP and phenotypic profiles for C59 were in agreement with its identity as Camp. jejuni ssp. jejuni.

One hundred and nine isolates (108 from birds and one from cattle) formed a heterogeneous group that nonetheless clustered together with reference profiles of Camp. lari Urease Positive Thermophilic Campylobacter (UPTC) group. An additional two clusters comprising two isolates each from Redshank birds, also showed affiliation to UPTC reference profiles and 16S rRNA gene sequences of representative strains showed highest BLAST matches (2646–2777 bits) to Camp. lari sequences in Genbank (data not shown). Phenotypic analysis of these strains revealed that the key characteristics (production of urease, micro-aerobic growth on 1% glycine medium and at 42°C, sensitivity to 32 mg l−1 nalidixic acid) were typical of the Camp. lari UPTC group (On 2005). Thus the total number of UPTC isolates from the avian species examined was 112 (Table 1).

Of the remaining clusters, one consisted of 17 isolates from Barnacle Geese, with AFLP profiles containing few bands. A polyphasic identification approach identified these strains as Helicobacter canadensis (Waldenström et al. 2003). The profile of an additional ungrouped goose isolate (B66) was closely related to the other strains, but was nevertheless distinct from type and reference H. canadensis strains, and it also differed phenotypically from these strains in its sensitivity to carbenicillin (32 mg l−1) and cefoperazone (64 mg l−1) when nutrient agar was used as the basal medium. The 16S rRNA gene sequence of this isolate showed a high similarity to other H. canadensis isolates and it was therefore considered as this species, increasing the total number of H. canadensis strains to 18 (Table 1). The remaining cluster, represented by five isolates, could not be attributed to any known genera of campylobacteria based on their AFLP profiles. These isolates, all from Barnacle Geese, had very complex AFLP profiles. A sequence-based phylogeny of the 16S rRNA gene of these isolates and sequences from selected members of the Epsilobacteria, indicated that they most likely represent a new genus of bacteria (Fig. 2), with closest resemblance to Anaerobiospirillum.

Figure 2.

 Dendrogram showing relationships between the Avispirillum genus and several other genera of the Epsilobacteria, as inferred by comparison of 16S rRNA gene sequences and neighbour-joining clustering. The scale bar indicates sequence similarity. Gaps and unknown bases were not considered in the analysis. Lawsonia intercellularis was used as an outgroup.

Isolation frequencies and distribution among hosts

Isolation frequency of campylobacteria was highest among Redshanks, which also was the most frequently sampled host species. Redshanks had a total campylobacteria prevalence of 86·1% (n = 122), compared with 24·1% in the Barnacle Goose, the second most frequently sampled study species, and 14·1% in cattle. There was no effect of transportation time on the culturability of campylobacteria in the Redshank samples (Likelihood ratio 9·05, d.f. = 5, P = 0·11), even though time from sampling to primary culture varied from 0 to 5 days for these samples. In contrast, samples from Barnacle geese endured transport periods of between 1 and 2 days, with recovery rates from samples examined after 2 days transit significantly different (χ2 = 11·06, d.f. = 2, P = 0·004). The effect of transportation was not investigated in the other host species, because of low sample sizes or isolation frequencies.

The bacterial species distribution on hosts was uneven, with all but one of the Redshank isolates, 99·1%, belonging to the Camp. lari UPTC group. Seven of nine samples from other shorebirds (Dunlins, Lapwings, Common Snipe and Common Sandpiper) were also positive for UPTC. In geese, however, UPTC was not isolated, and was found in only one sample from cattle. In contrast, the dominating bacterial species in Barnacle Geese, comprised 64·3% of the isolates, was H. canadensis, which was not isolated from Redshanks, other shorebirds or cattle (Table 1).

Isolation of Camp. jejuni ssp. jejuni was comparatively infrequent in this data set, with nine isolates from cattle, six from Barnacle Geese and one from a Redshank (Table 1).

Spatial and temporal effects on prevalence and epidemiology

The UPTC collection from Redshanks allowed for a more detailed inspection on spatial and temporal effects on prevalence and epidemiology. There was no difference in UPTC prevalence between the three meadows with the largest samples of Redshanks (Aurriv 88·2%n = 34, Western Näsudden 84·2%n = 38, Eastern Näsudden 83·7%n = 43; χ2 = 0·353, n = 115, d.f. = 2, P = 0·838). Intrastrain relationships between these isolates as inferred by cluster analysis of AFLP profiles did not indicate any clustering respective to the meadow of origin. In addition, AFLP profiles of UPTC isolates from birds that had been trapped and sampled twice during the course of the study showed that only strains from two individuals showed highly similar AFLP profiles (>90% similarity) at both samplings (Fig. 3). The time between samplings of recaptured birds ranged from 7 to 24 days. In one case, a Redshank was positive for UPTC on May 14 and for Camp. jejuni ssp. jejuni on May 31.

Figure 3.

 UPMGA dendrogram of AFLP profiles of resampled Redshank individuals. Each individual is indicated by a letter (a–i). Individual e switched from Campylobacter lari UPTC to Campylobacter jejuni (the basal strain in the dendrogram), all other individuals carried Campylobacter lari UPTC strains at both sampling occasions.

A similar analysis on differences in AFLP profiles of UPTC isolates between Redshank breeding pair members yielded only one pair of 16 that had similar profiles (data not shown).


Species diversity of campylobacteria

We analysed the occurrence of different campylobacteria species in a bird community that co-exists temporally on pastured coastal meadows, using an established AFLP protocol (Kokotovic and On 1999; On and Harrington 2000). This genotyping technique estimates variation across the genome using both restriction enzyme analysis and specific PCR, and has proved to delineate species as well as clonally related lineages of Campylobacter (On and Harrington 2000). Although the isolation procedures in our study were designed primarily to detect thermophilic Campylobacter species, we included isolates that differed slightly in colony – and cell morphology to these species in the identification procedures that followed. Numerical analysis of the AFLP profiles identified several clusters, including two that grouped with reference profiles of extant Campylobacter species (Camp. lari UPTC, and Camp. jejuni ssp. jejuni) and some that exhibited low similarities to any of the profiles in the databank. Sequence analysis of the 16S rRNA gene and phenotypic testing of representatives from these unidentified clusters, revealed their identity as Camp. lari UPTC, H. canadensis, and one novel taxon sharing closest similarity to the genus Anaerobiospirillum, a Gram-negative spiral bacterium phylogenetically distinct from Campylobacter yet similarly associated with gastroenteritis in humans (Malnick 1997). Pending further analyses to clarify the taxonomic position of this group, we propose the generic name Avispirillum spp. to refer to these organisms.

Despite the high prevalence of diverse campylobacteria in the investigated bird species, the results should be regarded as minimum values. In the UPTC collection from the Redshank samples we could not find any differences in recovery rate in samples even after 5 days after initial sampling, but there was a significant effect in samples from Barnacle geese. This discrepancy could be because of UPTC organisms, which were the most commonly isolated bacterial species in Redshanks, being better adapted to survival outside their warm-blooded host compared with the bacterial species that were found in the geese samples (H. canadensis, Camp. jejuni ssp. jejuni and ‘Avispirillum’ spp.). UPTC strains exhibit better resistance to temperature and ultraviolet light-mediated stresses in water compared with Camp. coli and Camp. jejuni (Obiri-Danso et al. 2001), and the predominance of tolerance to 2% NaCl among UPTC strains (On 1996) suggest increased resistance to drying compared with the other species. In any case, some isolates may have died before the samples were processed in the laboratory, and some low-grade infections may have been missed because of the lack of enrichment. Given that the isolation criteria used were designed for detecting thermophilic Campylobacter, it is not unlikely that we could have obtained more campylobacteria isolates using additional isolation techniques for fastidious or hydrogen gas-dependent species.

Epidemiological implications

In our dataset, UPTC organisms were the most commonly isolated species of campylobacteria. All but one of the 113 samples positive for UTPC were from shorebirds (Table 1) and the AFLP profile diversity within this group was large. Campylobacter lari UPTC strains have been isolated from a variety of environmental sources, such as mussels and oysters in the Netherlands (Endtz et al. 1997) and seagulls in Northern Ireland (Kaneko et al. 1999), suggesting that these, together with other Camp. lari taxa, are common in many aquatic environments. The UPTC group is highly diverse; using multilocus enzyme electrophoresis (MLEE) Matsuda et al. (2003) concluded that UPTC isolates were hypervariable and clustered separately from Camp. lari, which also been noted with the use of AFLP (On and Harrington 2000). The taxonomic relationships within Camp. lari and its biovars are not fully established, but most likely represent distinct subspecies (Duim et al. 2004).

UPTC organisms have only rarely been associated with human illness (Megraud et al. 1988), but its occurrence could have been underestimated. The frequent isolation of UPTC strains in this study suggests that they have limited effects on these avian hosts, which is further illustrated by the fact that all sampled birds appeared to be healthy with no signs of illness and that their breeding activities continued as normal. Interestingly, UPTC strains were not detected among the geese, and only one isolate was obtained from a cow. There are some obvious ecological differences between the sampled bird species. The shorebird species belong to the local breeding fauna, whereas the Barnacle Geese use the area as a stopover and staging area during the migration from their winter quarters in the Netherlands to their breeding areas on the Siberian tundra. While the geese generally are herbivores, the waders feed on invertebrates, mollusks and insects. Furthermore, there are also slight differences in prey choice and microhabitat used when foraging between the different shorebird species. For example, Lapwings forage more often on the drier areas of the meadow at some distance from the shoreline, whereas Redshanks most often feed at the water edge (Cramp and Simmons 1983). Intriguingly, this suggests differences in feeding areas, diet or other ecological factors of the host, could influence what species of campylobacteria they become colonized with.

This pattern of host-bacterium associations was found also for H. canadensis, which only was isolated from Barnacle Geese. The geographical distribution of this bacterium, with previous isolations from humans in Canada and Australia, and isolations from Barnacle geese in Sweden (this study), Canada geese from UK (Waldenström et al. 2003) and from two species of wild rodents in China (Goto et al. 2004) strongly suggests that it is widespread in nature. A full description of 17 of the 18 H. canadensis isolates of this study has been published separately (Waldenström et al. 2003).

Campylobacter jejuni ssp. jejuni was isolated from 16 samples, mainly from cattle and geese but also from one Redshank sample. All bird samples that were positive for Camp. jejuni had been collected from the meadows at Näsudden. This large meadow is divided into a western and an eastern part. The sampled cattle came only from the eastern part, while bird samples came from both areas. The six geese samples clustered separately from the other samples. The cattle isolates formed two different clusters, with one Redshank sample clustering at c. 90% similarity to a cow isolate. We cannot conclude, however, whether cattle, geese or some nonsampled host was the prime reservoir for Camp. jejuni on the meadow. The finding of a shared AFLP profile between a Redshank and a cattle sample is interesting, but it remains unknown if this has any implications on human health issues.

The AFLP and 16S rRNA gene sequence analyses results suggest five strains represent a novel bacterial genus that we tentatively refer to as Avispirillum spp. to reflect their spiral morphology and avian origins. We are carrying out a detailed taxonomic characterization of these isolates, which will be presented elsewhere.

Spatial and temporal effects on prevalence and epidemiology

We observed no trends in the distribution of AFLP types of UPTC isolates either from a geographic or temporal perspective. Carriage of the same or highly related UPTC strain in a single bird after resampling was found in only two instances. Among breeding pairs, only one of 16 pairs examined shared UPTC strains with the same AFLP type. These findings suggest that individual UPTC strains are frequently lost and acquired by Redshanks. Alternatively, a single bird may harbour several genetic strains of UPTC concomitantly and our analysis method, with only one investigated colony per positive plate, may have been suboptimal for this purpose. In any case, the genetic diversity of UPTC strains in this study, with no dominant AFLP profile, and the high prevalence in shorebirds, suggest that these organisms are probably very common in the birds’ environment.

Overall our study emphasizes the diversity and dynamics of campylobacterial populations among several wild bird species. There is no doubt much more to be learnt about the occurrence of campylobacteria in the environment and among wild animal populations. Such knowledge may help us to better understand the dynamics and transmission routes of zoonotic bacteria, especially where the zoonotic potential of the bacterial species found are known or suspected, as in this study. The public health implications for acquisition of campylobacterial disease through direct or indirect contact (e.g. via pollution of open waters) are self-evident and should be considered in light of the many questions that remain concerning epidemiology of the most well-studied Campylobacter species, as well as species whose prevalence in episodes of undiagnosed human gastroenteritis is presently unknown.


We are thankful for the assistance in the field by G. Gunnarsson, E. Trolle-Gunnarsson, K. Larsson and P. Jaxgård, and for the lab-work at Kalmar County hospital by I. Carlsson, and at the Danish Veterinary Institute in Copenhagen by P. Jordan. I. Eliasson at Kalmar County Hospital provided much logistical and methodological help. Funding for this study was generously put forward by the Uddenberg-Nordingska Foundation, the Elis Wide Foundation, the Alvin Foundation, the Oscar and Lili Lamm Foundation, the Swedish Research Council for Environment, Agriculture and Spatial Planning (FORMAS), the Research Council of Southeast Sweden (FORSS) and the Carl Trygger Foundation and the Crafoordska Foundation.