Background

Human campylobacteriosis has been the most commonly reported zoonosis in the European Union (EU) since 2005, with 214,779 confirmed cases in 2013 according to the European Food Safety Authority (EFSA) [1]. The disease burden was calculated at 35,000 disability-adjusted life years (DALYs) per year and the annual cost in the EU at around €2.4 billion [2]. The global number of DALYs was calculated to be 7,541,000 per year [3]. The cause of campylobacteriosis is Campylobacter spp. (primarily C. jejuni and C. coli)—a Gram-negative, spiral, microaerophilic bacterium and a common commensal inhabitant of the intestinal microflora of food production animals such as cattle, pigs and poultry [4]. It is estimated that 50–80 % of Campylobacter spp. strains infecting humans originate from the chicken reservoir, 20–30 % from the cattle reservoir and a small proportion from other reservoirs including wild animals [5]. As a consequence, the entire meat production chain and end products may be contaminated with C. jejuni or C. coli. In the EU, the pathways to humans are mainly through food, though environmental transmission and direct animal contact are also possible [6]. Therefore, reducing the occurrence of campylobacteriosis in the EU is a food safety issue of high priority, yet one which presents challenges [7].

According to a recent and extensive systematic review of 95 published studies of Campylobacter spp. sources around broiler farms [8], several wild animals (including wild birds) are known to be carriers. However, only a small number of the reviewed studies had a primary focus on wild birds living in close proximity to the farms. On a broiler farm in Athens GA, USA, 10 % (of 124) wild birds—mainly House Sparrow (Passer domesticus) and Common Starling (Sturnus vulgaris)—carried C. jejuni [9]. Colles et al. [10] found C. jejuni in 50.2 % of droppings from 331 Canada Goose (Branta canadensis) and Greylag Goose (Anser anser), and in 29.9 % of 954 Common Starling on a free-range broiler farm. Concerning cattle farms, a study in central Iowa, USA sampled 188 wild birds on dairy cattle, sheep and goat farms and found Campylobacter spp. in 4.8 % [11].

During the past decade, source attribution studies including multilocus sequence typing (MLST) have been conducted to compare the similarity of C. jejuni strains from wild birds with those from chicken and cattle [10–15] and with isolates from human disease [10, 12, 13, 15–17]. The overall conclusion is that the vast majority of C. jejuni strains are highly host specific. However, the studies also all identified a small proportion of strains with genotypes overlapping wild birds, farm animals [10–15] and human disease isolates [10, 13, 15–17].

Several studies on Campylobacter spp. carriage rates in wild birds in urban areas report a prevalence from 0–90 % [18–24]. Although it would appear that wild birds living in cities (mainly sparrows, pigeons, doves and starlings) have low carriage rates [19, 20, 22], French et al. [16] suggested that wild birds in city parks could contribute to campylobacteriosis in preschool children. The overall highest reported carriage rates have been found in gulls and crows foraging on refuse dumps in urban areas of Norway, Sweden, England, Japan, Spain and USA [18–21, 23–25].

Some of the large discrepancies in wild bird Campylobacter spp. prevalence between different studies may be attributed to host taxonomy or differences in the ecological guilds present. Bird ecological guilds are groupings of birds that exploit environmental resources in a similar way [26, 27]. The significance of different ecological guilds on the carriage rates of Campylobacter spp. was shown in a study of 1794 birds (the majority of which were migratory), sampled at Ottenby Bird Observatory on the island Oeland, Sweden [28]. The highest prevalence of Campylobacter spp. was found among ground-foraging guilds of short-distance migratory birds wintering in Europe.

The aim of our study was to estimate the prevalence of Campylobacter spp. in farm related wild bird species. Additionally, to investigate an association between Campylobacter spp. contaminated farm environments and wild birds around cattle, pig and poultry farms by performing an analysis of factors associated with Campylobacter spp. carriage of the wild birds.

Methods

Study design and selection of farms

The study covered four cattle farms, four slaughter pig farms, and four free-range poultry farms in Denmark, together with the wild bird populations living inside production buildings or within a 100 m radius from the farms. The study was conducted during January and February (winter) and during August and September (summer) in 2001. Two farms were sampled per week, and visited every weekday in order to get as many wild bird samples as possible. The cattle and pig farms were initially selected for a project investigating the occurrence of Salmonella in wildlife near Danish cattle and pig farms during 2001 and 2002 [29], while the poultry farms were included in this study only. The sampling schemes for Campylobacter spp. and Salmonella were conducted simultaneously in 2001.

Sampling

Wild birds

Birds were caught and ringed following the EURING system (http://www.euring.org/) by licensed ringers with mist-nets, traps, or by hand, thus ensuring that each bird was only sampled once per sampling event. The birds were released again after sampling. To ensure that a sufficient number of birds were caught during the winter months, several feeding places were established at each herd, using sterilised birdseed. We sampled as many birds as possible, and data on the estimated age, fat score, gender and exact place of capture were noted. Cloacal swab samples were obtained from the wild birds, using slim aluminum cotton swabs (DANSU, Ganløse, Denmark) and placed in Brain Heart Infusion (BHI) transport medium (DIFCO, Sparks, MD, USA) containing 5 % (v/v) calf blood (National Veterinary Institute, Copenhagen, Denmark) and 0.5 % agar (Oxoid Ltd., Basingstoke, Hampshire, UK).

Production animals

To detect Campylobacter spp. in cattle and pig herds, manure was collected at numerous places in the livestock facilities or among herds in pasture, and mixed into approximately twenty 200 ml containers (Dispatch Container Nunc, Life Technologies, Nærum, Denmark) per herd in each sampling round (i.e. 5–10 manure samples per container equalling 150–180 ml of manure) in order to obtain a representative measure of the within-herd Campylobacter spp. status. In order to sample poultry flocks, material from the litter surface was collected on a pair of boot socks whilst walking through the flock’s resting house [30].

Bacteriological examination and species characterisation

All samples were transported to the laboratory on the sampling day at ambient temperature, refrigerated overnight between 2 and 4 °C, and Campylobacter spp. cultivation was initiated the following day. For the number of samples tested, see Table 1.

Table 1 Campylobacter spp. prevalence and species distribution
Full size table

Cloacal swabs

Campylobacter spp. were isolated by streaking a swab with the faecal material directly on to modified Charcoal Cefoperazone Deoxycholate Agar (mCCDA) (CM0739, SR0155) (Oxoid) [31], and the plates were incubated under microaerobic conditions (6 % O2, 6 % CO2, in 88 % N2) at 42 °C for 48 h. Campylobacter spp.-like colonies were purified on blood agar and identified to species level using standard procedures including tests for hippurate and indoxyl acetate hydrolysis, catalase production and susceptibility to cephalotin and nalidixic acid according to NMKL 119 [32]. Campylobacter spp. isolates were identified as C. jejuni, C. coli, C. lari, C. upsaliensis, C. hyointestinalis or Campylobacter spp.

Manure

The manure was diluted to 1 g per 9 ml of buffered peptone water (CM1049, Oxoid), and 10 µl of the suspended material was streaked on mCCDA and incubated as described above.

Boot socks

Each pair of boot socks was placed in a stomacher bag, and after being diluted in 1:10 w/w in buffered peptone water (CM1049, Oxoid), faeces were released by gentle manipulation and 10 µl of the suspension was spread on mCCDA and incubated as described above.

Data analysis

The dependent variable was defined as a positive isolation of Campylobacter spp. from a wild bird. Descriptive statistics were performed using bivariate analysis [33] on Campylobacter spp. positive samples from wild birds. The association between independent variables was assessed using the Chi square test with a statistical significance threshold of P < 0.05. The evaluation of a possible association between Campylobacter spp. positive samples in the wild birds and in the herd was carried out separately for the two seasons (winter and summer).

Six potential factors associated with Campylobacter spp. carriage were included: (1) age (old, young); (2) herd type (cattle, pig, poultry); (3) proximity (in stable, around stable); (4) ecological guild with ≥10 samples (i.e. aerial insectivorous, foliage-gleaners, insectivorous seedeaters, open-land insectivorous, tit-like birds, sparrows, passerine seedeaters, terrestrial and low fly-catching feeders and thrushes); (5) fat score (0–8) [34], and (6) gender (male, female, not determined).

Based on the characteristic behaviour patterns of each ecological guild, the following five factors were selected: (1) feed (animal, mix, vegetable); (2) forage area (aerial, ground, vegetation); (3) proximity to stables (in stable, around stable); (4) contact with slurry (no, yes), and (5) migration range (long, medium, short, partial, none). This analysis included only the summer sampling, as more guilds were present, and the birds exhibited a wider range of behavioural patterns during the summer season than in winter.

Multivariate analyses [33] were carried out in all sampled wild birds organised in an ecological guild structure based on Gotellia et al. [27], using SAS Enterprise guide ver. 3.0.2. The logistic regression analyses were carried out using SAS PROC GENMOD. The modelling procedure assumed a binomial distribution and used logit as the link function. Goodness of fit was assessed by likelihood ratio statistics. The model was adjusted for overdispersion using the PSCALE option. In the analysis, non-significant variables were removed using stepwise backwards elimination. Statistical significance of the covariates was assessed using the likelihood ratio test based on P  ≤  0.05. The odds ratio (OR) and the 95 % confidence interval were reported for statistically significant variables.

In order to evaluate the impact of different herd types and season on the C. jejuni and C. coli carriage rates, sparrows (n = 616) were selected for the analysis, since this guild of non-migratory wild birds was the only one to be caught in a sufficient number on all farms during both winter and summer sampling. Correlation coefficients (R2) were calculated between the prevalence (%) of C. jejuni and C. coli in sparrows and the proportions (%) C. jejuni and C. coli in manure from each of the three herd types.

Results

Campylobacter spp. prevalence in sampled wild birds

In total, 1607 wild birds were sampled. The overall Campylobacter spp. carriage rate was significantly lower in winter (15.9 %, 112 positive samples out of a total of 706) than in summer (20.0 %, 180 positive samples out of a total of 901; OR = 1.32, 1.02–1.71, P = 0.03). For the species of Campylobacter spp. detected in each farm type, and the carriage rate among wild birds in winter and summer, see Table 1. For the prevalence of Campylobacter spp. in each bird species, see Table 2 and grouped in ecological guilds , see Table 3.

Table 2 The prevalence of Campylobacter spp. in wild birds and the allocation of bird species to ecological guild
Full size table
Table 3 Campylobacter spp. prevalence in ecological guilds
Full size table

The Campylobacter spp. carriage rates varied considerably between ecological guilds. The highest prevalence was found within two guilds: thrushes with 61.8 % (110/178) positive samples and sparrows with 21.3 % (131/616) positive samples (Table 2). Combined, these guilds were responsible for 82.5 % (241 out of 292) of the positive wild bird samples. The main bird species of these two guilds were the Common Blackbird (Turdus merula; n = 174), House Sparrow (n = 366) and Eurasian Tree Sparrow (Passer montanus; n = 250). They were also the most frequently sampled wild birds on the farms. Other birds that were frequently present were the Barn Swallow (Hirundu rustica; n = 128), Great Tit (Parus major; n = 129), European Greenfinch (Carduelis chloris; n = 90) and Common House Martin (Delichon urbica; n = 83), all of which had a low Campylobacter spp. prevalence (Table 2).

Factors associated with Campylobacter spp. carriage in wild birds

Analysis of the six selected risk factors for Campylobacter spp. carriage in wild birds (age, herd type, proximity, ecological guild, fat score and gender) revealed that the ecological guild was significantly associated with Campylobacter spp. carriage during both winter and summer (Table 3). Thrushes and open-land insectivorous birds were more likely to carry Campylobacter spp. than sparrows (used as a reference guild), whereas all other guilds had lower odds than sparrows. In general, herd type, fat score, gender and age were not significantly associated with Campylobacter spp. prevalence in wild birds (all sampled birds). Proximity was significant in summer (see proximity to stables in Table 4) but not in winter (data not shown).

Table 4 Factors associated with Campylobacter spp. carriage and specific bird behaviour during summer
Full size table

Patterns of behaviour in summer

Concerning the impact of particular patterns of behaviour in summer (i.e. feed, forage area, proximity to stables, contact with slurry and migration range), there was significantly increased odds for Campylobacter spp. carriage in birds eating food of animal or mixed animal and vegetable origin foraging on the ground and in vegetation close to the production buildings (Table 4). No association was found between Campylobacter spp. carriage and contact with slurry or migration range (data not shown).

Herd type and Campylobacter species distribution

C. jejuni was the most commonly isolated Campylobacter species in wild birds on all farm types, comprising 78.3 % (58 out of 74) of wild bird isolates on cattle farms, 75.9 % (101 out of 133) on pig farms and 89.9 % (80 out of 89) on poultry farms (Table 1). The remaining isolates were almost entirely C. coli, of which 46 out of 48 isolates were found at the winter sampling.

Looking at the proportions of Campylobacter species in herd manure and the prevalence in wild birds at each of the 12 individual farms revealed a strong correlation between the prevalence of C. jejuni in both wild birds and the proportions in manure on cattle farms (R2 = 0.92), and a moderate correlation on poultry farms (R2 = 0.54). Likewise, a moderate correlation was found between C. coli in both wild birds and in pig manure (R2 = 0.62; Fig. 1). In contrast, no correlation was seen between C. coli in wild birds and in manure on cattle and poultry farms, or between C. jejuni in wild birds and in manure in pig herds (Fig. 1).

Fig. 1
figure 1

Correlation between the prevalence (%) of Campylobacter jejuni and C. coli in sparrows and the proportions (%) of C. jejuni and C. coli in manure from cattle, pig and poultry herds. The prevalence, proportion and correlation coefficients (R2) on the regression lines are shown in red (poultry farms), blue (cattle farms) and green (pig farms) circles (C. jejuni) and triangles (C. coli)

Full size image

Discussion

A seasonal peak in the prevalence of Campylobacter spp. in wild birds was observed in summer. This was also found in a study of farm related Common Starling in the UK [12], and a study of Black-headed Gull (Larus ridibundus) in Sweden [23]. The underlying causes of seasonality in the epidemiology of Campylobacter spp. are not fully understood. However, seasonality is also a recognised factor in the pattern of Campylobacter spp. infections in poultry [2], and in the occurrence of human campylobacteriosis [1]. The vast majority (82.5 %) of Campylobacter spp. in wild birds in our study was isolated from thrushes and sparrows (Tables 2, 3), representing some of the most common wild bird species in Denmark (i.e. Common Blackbird, House Sparrow and Eurasian Tree Sparrow).

The Campylobacter spp. carriage rates of the farm-related wild birds were found to be closely associated with the ecological guild (Table 3). Studies from Sweden [28] and Italy [35] have reported results for ecological guilds sampled at bird stations. The Swedish study found the highest Campylobacter spp. prevalence in wagtails, Common Starling and thrushes [28], in agreement with the results presented here. Common bird species such as the European Greenfinch, European Robin (Erithacus rubecula), Great Tit and Common Chaffinch (Fringilla coelebs) showed low Campylobacter spp. prevalence in both the Swedish study and the present study (Table 2). Our analysis identified feeding habit, forage area and proximity to stables as factors significantly associated with the carriage of Campylobacter spp. in wild birds (Table 4). This is in line with the results of the Italian study [35], where feeding habit was considered an important factor, and carnivorous birds foraging on the ground showed the highest prevalence of Campylobacter spp. A Japanese study [20] examined the correlation between the crop and actual stomach content and the prevalence of C. jejuni, and found a negative correlation between vegetable stomach content and C. jejuni colonisation. Several other studies have reported that omnivorous birds such as crows and gulls foraging close to areas with human garbage and sewage have a particular risk of high carriage rates [19, 20, 24, 25].

We found a correlation between the prevalence of C. jejuni in wild birds and proportions in both manure on cattle and poultry farms, and between C. coli in wild birds and pig manure (Fig. 1). However, this correlation can only account for part of the Campylobacter spp. epidemiology on the farms, since some of the C. jejuni and C. coli detected in the wild birds (i.e. the C. jejuni in birds on pig farms and the C. coli in birds on the cattle farms) could not be explained by the correlation to farm manure (Fig. 1, Table 1). It is likely that bird-to-bird transmission, or sources not included in this study were responsible for the observed Campylobacter spp. It is also possible that the farm animals and the wild birds both acquired Campylobacter spp. from the same sources, but became colonised by different species adapted to their specific gut environments. An interesting aspect for further research would be to investigate why the isolation rate of C. coli in the wild birds during the summer sampling was so low on all farms, and why the proportion of C. coli in the pig manure was also lower in summer (60.2 %) than in winter (85.2 %; Table 1).

Our study showed that in summer, sparrows caught at poultry or pig farms were more likely to carry Campylobacter spp. than sparrows caught at cattle farms. The reason for this remains speculative, though the majority of cows were at pasture during the summer months, thus potentially resulting in minimal contact with the sparrows close to the farm buildings. Further investigation should be performed in order to evaluate this.

We anticipated that wild birds and livestock occupying very close living space might share strains locally and that this might be a key point to understand the epidemiology of Campylobacter spp. in wild birds on livestock farms. We realise however, that our study suffers from an inferior resolution depth, as we summarised our results at the Campylobacter species level and not the genotype level. We may therefore have emphasised farm factors over strain factors, which were not measured. More recent studies using MLST have shown a large degree of host specificity [12, 17, 36, 37] and minimal overlap in MLST profiles of Campylobacter spp. from wild birds and from poultry, cattle and humans. There was a greater similarity between the level of C. jejuni found in Common Starling in Sweden and Common Starling in the UK, than there was between C. jejuni from Swedish Common Starling and their Swedish environment [37]. This segregation between the Campylobacter spp. strains in wild birds and the livestock reservoir is supported by a host attribution study [38] investigating the host association in seven housekeeping loci in 2732 published C. jejuni isolates from a number of sources including chicken, farm ruminants, and wild birds (passerine birds, ducks and geese). The main finding was that phylogenetically distinct C. jejuni lineages were associated with distinct wild birds, whereas in the farm environment, phylogenetically distant farm animals shared several C. jejuni lineages. Likewise, a possible adaptation of certain clonal complexes to flocks of barnacle geese in Finland has been found in a recent study [39]. Some studies note that wild birds may have a minor role in transmitting pathogenic C. jejuni strains to cattle [11, 13, 15] and to humans [10, 13, 15, 16, 39], whereas others found no evidence of transmission [12]. A recent study [40] found wild bird C. jejuni strains to be a consistent source of human disease in the UK, suggesting the existence of some more obscure epidemiological pathways between the wild bird reservoir and humans. From 2003 to 2013, the burden of campylobacteriosis cases attributed to wild birds was estimated at 10,000 per year in the UK. Therefore, it appears that the development of methods to control the transmission of Campylobacter spp. between livestock, humans, and wild birds requires better elucidation and understanding of the dynamics of transmission.

Conclusions

Based on the findings in this study, we conclude that the carriage of C. jejuni and C. coli in wild birds on livestock farms is correlated to the proximity to stables, feeding habits and forage areas on the ground and in vegetation. Birds with forage areas further away from livestock buildings or in the air, carried less Campylobacter spp. These findings suggest that wild birds may play a role in sustaining the epidemiology of Campylobacter spp. on farms, although this study is not able to elucidate the direction of the transmission, and further studies including genotyping are required.