The occurrence of Campylobacter subtypes in environmental reservoirs and potential transmission routes


J.A. Hudson, ESR Ltd, Christchurch Science Centre, PO Box 29-181, Ilam, Christchurch, New Zealand (e-mail:


Aim:  To identify potential reservoirs and transmission routes of human pathogenic Campylobacter spp.

Methods and Results:  An enrichment PCR method for the detection and identification of Campylobacter jejuni and/or Campylobacter coli in faecal, food and river water samples was applied to 1450 samples of 12 matrix types obtained from a defined geographical area. PCR-positive samples were cultured to yield isolates for typing, and the data for 616 C. jejuni isolates obtained.

Serotyping and SmaI macrorestriction profiling using pulsed field gel electrophoresis revealed a high level of diversity within the isolates from each matrix. Campylobacter jejuni and C. coli subtypes indistinguishable from those obtained from human cases were detected in most of the matrices examined. No Campylobacter isolates were isolated from possum faeces.

Conclusions:  Ten of the 12 matrices examined may be involved in the transmission of human campylobacteriosis as they contained Campylobacter subtypes also isolated from clinical cases.

Significance and Impact of the Study:  Results indicate that, for this rural population, a range of potential transmission routes that could lead to campylobacteriosis exist. Their relative importance needs to be assessed from an exposure assessment standpoint.


The most frequently reported notifiable enteric disease in New Zealand is campylobacteriosis, with rates that are two to three times higher than other developed countries and many times higher than the USA. For example, in 2003 the reported incidence of campylobacteriosis in selected states of the USA was 12·6 cases per 100 000 (Centers for Disease Control 2004) in comparison with New Zealand where the incidence was 334·1 cases per 100 000 in 2002 (Anonymous 2002). While differences in reporting systems might partially explain this observation, the increased incidence of campylobacteriosis in New Zealand was not considered to be an artefact of reporting or improved methodology in a previous analysis (Lane and Baker 1993). Records of the incidence of campylobacteriosis in New Zealand have been maintained since the disease became notifiable in 1980, and since then the trend in the number of reported cases is upwards.

Human infection with Campylobacter caused through food-borne transmission has been estimated to cost NZ $40 136 000 annually, which is 73% of the total economic cost of food-borne infectious intestinal disease in New Zealand (Scott et al. 2000). This figure is based on estimated numbers of reported and unreported cases and an estimate that 65% of human cases are caused by food-borne transmission (Lake et al. 2000).

The consumption of undercooked chicken has been identified as an important source of campylobacteriosis in New Zealand (Eberhardt-Phillips et al. 1997). However, there is little doubt that other exposure(s) might also contribute to the disease burden in New Zealand (Ikram et al. 1994), but the reasons for New Zealand's reported elevated rates compared with other developed countries are not known.

To enable a comprehensive ‘farm to fork’ approach to the control of campylobacteriosis be taken, a good understanding of the microbial ecology of Campylobacter in environmental reservoirs and transmission routes is required. Questions surrounding the relative significance of reservoirs and transmission routes are important and remain largely unanswered. This lack of information is because of three primary reasons: Campylobacter transmission routes are complex; studies reported in the scientific literature tend to deal with small aspects of transmission in isolation and rarely involve a cross-disciplinary approach; and, until recently, little research has been conducted into campylobacteriosis in New Zealand.

One New Zealand study has provided some subtyping data for Campylobacter isolated in the summer and winter from human and veterinary cases, raw milk and poultry, and recreational water (Hudson et al. 1999). Some differences in the subtypes isolated in the two seasons were noted, and some subtypes were only isolated from human cases. The most frequently isolated subtype occurred in the summer and these isolates were obtained only from poultry and human cases.

The work described in this paper was undertaken to type Campylobacter isolates from reservoirs and transmission routes, and provide data that could be used to assess their relative importance to human disease with the aid of additional exposure information.

Materials and methods

Bacterial isolates

Campylobacter jejuni (NCTC 11351T), Campylobacter coli (NZRM 2607T) and Escherichia coli (ATCC 25922) were used for PCR controls. Campylobacter spp. were grown on Columbia blood agar base (CBA) (Merck, Darmstadt, Germany) supplemented with 5% sheep's blood at 42°C under microaerophilic conditions generated by the Oxoid CampyGenTM system (Basingstoke, UK). Escherichia coli was grown overnight in brain–heart infusion (BHI; Difco, Detroit, MI, USA) broth at 35°C.

Enrichment and PCR conditions

M-Exeter broth was prepared as outlined in Wong et al. (2004) which contains nutrient broth no. 2 and is supplemented with lysed horse blood, sodium metabisulphite, sodium pyruvate and iron sulphate. The selective agents added include cefaperazone and Oxoid antibiotic supplement SR204E.

Primary enrichment broths were incubated at 37°C for a minimum of 4 h under microaerophilic conditions and then transferred to an incubator at 42°C and incubation continued for a total of 48 h. Following incubation, 0·1 ml of the primary enrichment broth was inoculated into 10 ml m-Exeter broth and incubated at 42°C for 24 h. This secondary enrichment step was performed to reduce the possibility of detecting nonviable cells by PCR. Volumes (1 ml) of the secondary enrichments were centrifuged at 4000 g for 20 min to pellet the cells. The supernatant was discarded and the pellet washed three times and resuspended in sterile phosphate-buffered saline (BR14; Oxoid). The suspension was heated (100°C for 12 min) to lyse the cells, centrifuged at 11 750 g for 10 min and 10 μl of the supernatant added directly to the PCR premix. Multiplex PCR conditions for the simultaneous identification of C. jejuni and C. coli and visualization of PCR products were performed as previously described (Wong et al. 2004). Positive controls of C. jejuni (NCTC 11351T) and C. coli (NZRM 2607T) DNA and negative controls of uninoculated broth and E. coli (ATCC 25922) DNA were included.

Sources of samples for prevalence determination and isolates for subtyping

The overall plan was to screen enrichments of a variety of sample types for the presence of C. jejuni and C. coli by PCR, and to isolate colonies from these enrichments by conventional plating. Isolates would then be identified by PCR and typing performed by Penner serotyping and SmaI macrorestriction profiling via pulsed field gel electrophoresis (SmaI-PFGE) analysis. Samples tested comprised river water, whole chicken carcass rinses, beef, pig and sheep liver, fresh faeces from beef cattle, dairy cattle, sheep and ducks, and specimens of human faeces from people infected with Campylobacter. All samples were obtained from the Ashburton area, which was selected for the study because the South Canterbury Health District is consistently among those health districts with higher than average rates of campylobacteriosis in New Zealand. Also, Ashburton has one primary reticulated water source, and its geographical remoteness makes it likely that most of its inhabitants live and work in the area. Consequently, exposure to foods contaminated by Campylobacter is likely to be from foods bought locally.

Human faecal samples were obtained from people who were identified as campylobacteriosis cases and notified to the local public health service (Crown Public Health). Clinical laboratories were asked to store chilled any clinical faecal samples from the Ashburton area which tested positive for Campylobacter spp. These specimens were sent weekly to the ESR laboratory for testing.

Three water collection sites were sampled on a regular basis. One site was upstream of the region's major township (population of 15 000) and downstream of the infiltration gallery for the township's drinking water system. The township's treated drinking water was largely derived from this source, and was only supplemented by bore water during dry conditions. The other two sites were upstream of this site, on two major tributaries that join to become the main branch of the Ashburton River. All three sites were downstream of significant areas of dairy cattle, beef cattle and sheep farming.

Faeces from farm animals were collected from farms adjacent to the river and upstream of the water sites. Farms were chosen to reflect the diversity of farmed animals (sheep, dairy cow and beef cattle) represented within the study area. Although some of the farms stocked both sheep and cattle at the same time, only one animal faecal type from those farms was collected. Faecal samples from mallard ducks were collected from two parks within the boundary of the town. Each faecal sample from each type of animal was a composite from five different randomly collected stools.

Offal was sampled as it yields a higher prevalence of Campylobacter compared with meat derived from the same animal type (Kramer et al. 2000). Whole chickens were the other food type sampled. Food sampling frequency was proportional to the volume of food traded by the retailers in Ashburton.

Farmed animal faecal samples were collected on alternate weeks from January to December 2001. Human faecal samples were collected for a slightly longer time period – from January 2001 through to January 2002 to allow time for ingestion of the pathogen, reporting to a primary care physician and sample submission to occur.

Once 4 months of sampling had been completed, an assessment of Campylobacter prevalence in each of the matrices was used to adjust sampling frequency to ensure that sufficient isolates were obtained from each matrix.

Isolation and identification of Campylobacter

Samples were enriched and screened by PCR as described above. Secondary enrichments which tested positive by PCR were spread-plated to m-Exeter agar and incubated for 48 h at 42°C, and colonies characteristic of Campylobacter purified by two further subcultures on CBA containing 5% (v/v) defibrinated sheep blood. To identify the species present, a single isolated colony was resuspended in 27·25 μl of double-distilled water in a 0·5-ml thin-walled PCR tube. This was then heated for 3 min at 100°C and cooled to 4°C. PCR premix was added to the tube to a final volume of 50 μl, and PCR analysis performed as described above. A further colony was typed as described below.


Penner serotyping of C. jejuni isolates was performed by the passive haemagglutination technique. Antisera were produced at the Enteric Reference Laboratory (ESR-KSC, Wellington, New Zealand) by the methods described by Penner and Hennessy (1980) using their reference isolates for antisera production.

Preparation of agarose plugs for SmaI-PFGE analysis was performed as described by Gibson et al. (1994). Conditions for running the SmaI-PFGE gels were an initial switch time of 10 s, final switch time of 35 s; 120° for 22 h.

To ensure accurate normalization for inter-gel comparisons, agarose gels were run with molecular weight ladders of Lambda Ladder PFGE Marker (no. N0340S; New England Biolabs Inc., Beverly, USA). Once electrophoresis was completed the DNA gels were stained with ethidium bromide, scanned and the data saved in a Tiff format. The images were analysed by Bionumerics, version 2·5 software (Applied Maths, Saint-Martens-Latem, Belgium). Following the conversion and normalization of the gels, the degrees of similarity of DNA profiles were determined by the Dice coefficient and dendrograms were generated by the unweighted pair group method (upgma) with an optimization setting of 1 and 1·2% position tolerance. The SmaI-PFGE type identity of new strains was determined by matching their normalized DNA profiles against the constructed computerized SmaI-PFGE libraries. Clonally related types were defined as those differing by one band shift, or where a large molecular weight band was replaced by two smaller molecular weight bands which approximated in total the molecular weight of the original band.


Prevalence of Campylobacter in the matrices tested

Prevalence data of C. jejuni and C. coli in the matrices tested are shown in Table 1. Overall, C. jejuni was the predominant species identified in human faeces and in most of the other sample types. The one exception to this was pork offal, from which recovery of both species was similar. The composite samples, namely animal faeces and water, showed a high prevalence of C. jejuni. Sheep faeces yielded the highest percentage of C. coli-positive samples of all of the matrices, but this preponderance of C. coli was not reflected in sheep offal samples which, in contrast, yielded the highest prevalence of C. jejuni of all the offal types. While C. jejuni was more prevalent in beef cattle faeces than in sheep faeces, in the corresponding offal types the situation was reversed.

Table 1.  Prevalence data for Campylobacter coli and Campylobacter jejuni in the Ashburton samples tested as assessed by PCR detection
MatrixNumber of samples testedNumber of positive samplesNo. (%) positive for
C. jejuniC. coli
  1. *Composite of five individual samples.

Faeces from human campylobacteriosis cases696157 (82·6)6 (8·7)
Water293162162 (55·3)12 (4·1)
Duck faeces*926060 (65·2)1 (1·1)
Dairy cow faeces*918989 (97·8)9 (9·9)
Beef cattle faeces*877373 (83·9)14 (16·1)
Sheep faeces*876652 (59·8)41 (47·1)
Rabbit faeces72101 (1·4)
Possum faeces197000
Beef offal1781615 (9·0)1 (0·6)
Sheep offal1626563 (38·9)6 (3·7)
Pork offal187159 (4·8)9 (4·8)
Chicken carcass2045656 (27·5)2 (1·0)
Total1450664637 (43·9)101 (7·0)

Campylobacter jejuni serotypes

Thirty-two heat-stable (HS) serotypes of C. jejuni were identified among the 616 isolates tested (Table 2). To aid clarity, the five most frequently occurring serotypes among the isolates from each of the matrices are shown in Table 3.

Table 2. Campylobacter jejuni serotypes isolated from ten matrices in the Ashburton area
Heat stable serotypeMatrix
Human faecesWaterDuck faecesDairy cattle faecesBeef cattle faecesSheep faecesBeef offalSheep offalPork offalChicken carcassTotal
Frequency of isolation
  1. *Includes serotypes where only one isolate was obtained. These were: 3, 12 and 25. Totals may differ from those in Table 1 as enrichments testing positive by PCR did not necessarily always yield growth on modified Exeter agar. Data for possum and rabbit faeces not given.

1,44 5 15 1 5 2 2 212  4 48
217  8 21217 5 110210 84
4 Complex 8  7 31614 8 3 91 3 72
5  15 9   6 1 5  2 38
6 2 10 4  2 5    1 24
8,17  10 8     21 2 23
10 2   3  4  2   11
11 4  4  4 3 1  1   17
15 1  4 1         6
18 2  2    1      5
19 2  4 2  6 1 2 6   23
21   4 1 1    1 11 18
22 2  1   1   1    5
23,36 3  6 34 8 4 1 72 1 66
24   2          2
27   2    4      6
31 1  2       3  6
33   3 1  1 1      6
35 1  4  3 6  2 13  20
37   3 8 1 1 1     14
41   1   1       2
42   1 1   1    2  5
44   1    1      2
45     2 1      3
52 1  2   1    1  5
53   1  3        4
55   2          2
57 1  3   1     1  6
not done   1          1
untypable 4 3713 5 6 1 2 6 15 89
Table 3.  Comparison of the five most frequently Campylobacter jejuni isolated serotype from each matrix
Human faecesWaterDuck faecesDairy cow faecesBeef cattle faecesSheep faecesBeef offalSheep offalPork offalChicken carcass
  1. –, No single fifth serotype.

Five most frequently isolated serotypes (% of isolates from matrix of that serotype)
2 (29·8)1,44 (9·9)5 (15·5)23,36 (39·0)2 (23·9)4c (16·7)4c (20·0)1,44 (19·0)35 (33·3)21 (19·6)
4c (14·0)5 (9·9)8,17 (13·8)4c (18·4)4c (19·7)5 (12·5)1,44 (13·3)2 (15·9)2 (22·2) 2 (17·9)
1,44 (8·8)6 (6·6)37 (13·8)2 (13·8)23,36 (11·3)2 (10·5)19 (13·3)4c (14·3)23,36 (22·2) 1,44 (7·1)
11 (7·0)8,17 (6·6)6 (6·9)1,44 (5·7)19 (8·5)6 (10·5)35 (13·3)23,36 (11·1)4c (11·1) 4c (5·4)
23,36 (5·3)2 (5·3)4c (5·2)11 (4·6)35 (8·5)19 (9·5)8,17 (11·1)31 (5·4)

Almost half of the human isolates were HS 2 (29·8%) or HS 4 complex (14·0%). These two serotypes were also common in isolates from beef cattle faeces (43·6%), dairy cattle faeces (32·2%), sheep faeces (27·2%), sheep offal (30·2%) and chicken carcass rinses (23·3%). In contrast, only 8·6% of the duck faecal isolates and 9·9% of water isolates typed as HS 2 or HS 4 complex. Only 7% of beef offal isolates were HS 2 compared with 23·9% of isolates from beef faeces. Serotype HS 1,44, identified in 8·8% of the isolates from human campylobacteriosis, was isolated from all nonhuman matrices except pork offal, with most of the isolates of this serotype being derived from river water (9·9%) and sheep offal (19·0%). HS 23,36 occurred frequently in dairy cow faeces (39·0%) and at lower prevalence in other farm animal faeces. Sheep and pork offal also contained HS 23,36, but there was a low prevalence of this serotype in human faecal isolates (5·3%) and water isolates (3·9%). HS 21 was isolated with the highest frequency from chicken carcasses (19·6%) but no human faecal isolates of this serotype were identified. HS 2 was the serotype isolated at the second highest frequency (17·8%) from chicken and this was a common serotype in human cases.

Among the water isolates HS 8,17, HS 6, HS 5 and HS 1,44 were common, while HS 23,36, HS 4 complex and HS 2 were rarely isolated. Duck faeces contained HS 37 and HS 5 most frequently, the latter also being commonly obtained from river water.

A comparison of the profiles of the most frequently isolated serotypes from humans with those from the other matrices (Table 3), showed that the distribution profile was most similar to that of dairy cow faeces, as the five most frequently isolated serotypes were the same in these matrices. The least similar were the isolates from duck faeces where only one of the five most frequently isolated serotypes from this source was also isolated from human cases.

Samples collected from water, duck faeces and chicken carcasses had the largest proportions of isolates that could not be typed, and the overall percentage of isolates that could not be typed was 14·4%.

SmaI-PFGE types isolated

A total of 131 SmaI-PFGE types were noted among the C. jejuni isolates. Of these, 60 subtypes were represented by only one isolate, and a further 20 subtypes by only two isolates. Only 13 SmaI-PFGE types were represented by more than 10 isolates (Table 4). Some subtypes were frequently isolated from particular matrices. For example, SmaI-PFGE type 19b was isolated on 42 occasions, of which 27 (64%) were from beef cattle faeces. Similarly, 88% of subtype 21 isolates were from water. Types 21 and 236 isolates were primarily from duck faeces and water, with no isolates of these types originating from human faeces.

Table 4.  Most frequently isolated Campylobacter jejuniSmaI-PFGE types from ten matrices sampled in the Ashburton area
PFGE typeMatrix
Human faecesWaterDuck faecesDairy cattle faecesBeef cattle faecesSheep faecesBeef offalSheep offalPork offalChicken carcass
Frequency of Isolation
32 1 4411528
3d  5  31 2  
162 5   1 3  
19b2 1 2753 3 1
21 15  11    
254113 136   6
25b 121       
333   47116  
34  1 1193151 
354 3  43     
60a  4  1     7
221  75  1    
236  67       

Thirty-seven different C. coliSmaI-PFGE types were identified. Table 5 shows data for those SmaI-PFGE types isolated on more than two occasions. Isolates of C. coliSmaI-PFGE type 1 were isolated mostly from ruminant faeces, but this type was not isolated from human faeces, and a similar observation could be made for SmaI-PFGE type 11. SmaI-PFGE type 17 was only detected in human faeces and water.

Table 5.  Most frequently isolated Campylobacter coliSmaI-PFGE types from matrices sampled in the Ashburton area
PFGE typeMatrix
Human faecesWaterDairy faecesCattle faecesSheep faecesSheep offal
Frequency of isolation
 1  23 91
 33    7 
101 4  5 
11 1 1122
331    12

Comparison of indistinguishable isolates

As serotyping data were only available for C. jejuni isolates, this section contains information for that species only. Isolates which were indistinguishable by both serotyping and SmaI-PFGE typing are considered to be closely related for the purposes of this comparison. Indistinguishable subtypes are identified as serotype: SmaI-PFGE (HS:P) type. As 186 serotype: SmaI-PFGE types were isolated, Table 6 shows only data for those subtypes where an isolate was obtained from a sample of human faeces, and subtypes where more than ten isolates of that subtype were identified.

Table 6.  Subtypes of Campylobacter jejuni isolates from human campylobacteriosis and for subtypes where more than 10 isolates were identified
SerotypePFGE typeMatrix
Human faecesWaterDuck faecesDairy cow faecesBeef cattle faecesSheep faecesBeef OffalSheep OffalPork OffalChicken Carcass
Frequency of isolation
  1. Nineteen isolates from human cases, all of different subtypes, were isolated only once. Six isolates from human cases were untypable by one of the methods.

1,44  3a1     15  
1,44 332     14  
2  1c2         
2  31 1 3311417
2 1621     2  
2 18a4 1       
2 331   371    
2 541   11 1  
4 Complex  11      1 2
4 Complex 34 1  973131 
4 Complex 521   3 1    
4 Complex2431    1    
8,17236 66       
10 181   2 2 1  
10 18b1    1    
11 3543  43     
15 60b14        
19 121   2     
21 60a 3  1     7
22 282         
23,36 19b21 2553 3 1
23,36 221   1 12  
31 251 1      1
35 1012    2   

The most frequently occurring subtype was HS23, 36:P19b (40). Of these isolates, 75% were from dairy cow and cattle faeces, and only two were human clinical isolates. Twenty-five isolations were made of the next most frequent subtype, HS4 complex:P34 with most isolates originating from ruminant faeces or offal, and none of the isolates originating from human faeces. Subtypes HS2:P3 and HS2:P33 were also common in ruminant faeces and a single isolation from human faeces occurred in each instance. All 12 isolates of subtype HS8,17:P236 were from duck faeces or water.

Of the two subtypes most frequently isolated from human clinical cases (four isolates of each) namely, subtypes HS2:P18A and HS11:P35, only one other isolation of HS2:P18A was made from duck faeces, while isolates of HS11:P35 were made from water and bovine faeces.


PCR detection

The enrichment PCR method gave a small proportion of false-positive results as C. jejuni was isolated from 616 of the 637 samples scored as positive by PCR. This may have been due to several reasons. There may have been some loss of viability in some cultures occurring in the delay between obtaining the PCR results and plating to agar. Also the primer sets employed may have imperfect specificity, although previous data did not show this (Wong et al. 2004). It is possible too that viable but nonculturable cells (Rollins and Colwell 1986) were detected by the PCR, and by definition, were unable to grow on the agar.

Human Isolates

While all of the human clinical specimens should have yielded a Campylobacter isolate, because they had been confirmed as containing the organism, a few did not. It is possible that low numbers of organisms may have been present in some of the original samples which lost viability during storage and transport, or samples could have been stored inappropriately or for too long. In addition, the screening laboratories may have isolated species other than C. jejuni and C. coli and these would not have been detected by the PCR method used.

The ratio of C. jejuni to C. coli isolates (close to 90 : 10) was expected based on our previous experience and other reports (Lawson et al. 1999).


The prevalence of Campylobacter in the matrices was largely as expected from previous work in New Zealand and internationally. However, in contrast to a previous New Zealand study where 56·6% of chicken samples were positive (Hudson et al. 1999), only 27·5% were positive for C. jejuni in this study. All the chicken was supplied by one company, which may produce chicken with a lower prevalence of Campylobacter than that previously reported, and as only whole chickens were sampled, it may also be that whole chickens are less frequently contaminated than the chicken portions tested by Hudson et al. (1999).

The observed prevalence of around 50% for Campylobacter in river water samples was expected (Savill et al. 2001). High prevalence of Campylobacter were also found in ruminant faeces, but these data were derived from composite samples and the prevalence is likely to be higher than those for individual animals. Highly variable carriage rates in ruminants have been reported, from a few percent in adult animals to 42·1% in calves (Nielsen 2002). In New Zealand, prevalence of between 12 and 31% have been noted for dairy cattle (Meanger and Marshall 1989). Similarly duck faecal samples were composite samples from five birds, and so the prevalence (65·2%) is likely to be higher than for individual birds. International data indicate prevalences of 30–40% (Luechtefeld et al. 1980; Fallacara et al. 2001). Ducks were sampled from two parks with significant human impact which may also have artificially increased the prevalence. Campylobacter jejuni was not detected in feral rabbit or possum faecal samples, and only one rabbit sample was positive for C. coli.

Of the types of offal tested, sheep offal was the most frequently contaminated (40·1%), followed by beef (9·0%) and pork (8·0%). Pork offal was contaminated at the same prevalence by both Campylobacter species.

Campylobacter coli was isolated most frequently from ruminant faeces, and almost half of the composite sheep faeces samples contained this species. This was unexpected as C. coli is normally associated with pigs (Nesbakken et al. 2002). However, the high prevalence of C. coli in sheep faeces was not reflected in data for sheep offal.


Serotypes HS2 and HS4 complex occurred frequently in human isolates, and they were also common in dairy and beef cattle faeces. Previous serotyping data are available for three New Zealand studies (Nicol and Wright 1998; Hudson et al. 1999). In the first study, HS 2 comprised 26% of human clinical isolates and was also isolated from bovine and ovine sources. HS 4 was also isolated from bovine sources and human isolates. In the second study of human isolates, frequent serotypes were HS 2 (16%), HS 4 (8%) and HS 8 (8%). In the third study (Hudson et al. 1999), the most frequently isolated serotypes from human and veterinary cases, raw chicken and water were HS 4 complex (14·8%), HS 2 (14·2%), HS 33 (9·9%), HS 6 (8·6%) and HS 12 (7·4%). In the winter, most human campylobacteriosis was caused by HS 4 complex (52·6%) and HS 2 (21·7%), but in the summer the most prevalent serotypes were HS 2 (26·1%), HS 33 (21·7%) and HS 6 (15·2%). In contrast to previous New Zealand data, no HS 8 or HS 33 isolates were obtained from human faeces, although they were isolated from other matrices. At the level of serotype, therefore, the range of isolates described in the present work is in accordance with that observed previously in New Zealand.

There is a broad agreement among data from developed countries that HS 2 and 4 are the serotypes most often associated with human disease (McMyne et al. 1982; Jones et al. 1984). Other serotypes tend to contribute to a small proportion of human infection with C. jejuni individually, but represent a large component when combined.

Serotypes 2 and 4 complex were frequently isolated from ruminant faeces and offal samples tested in the work described here. In a British study cattle yielded mostly HS 4 complex, 2, 1, 11 and 27,31 (Jones et al. 1984). Of the beef isolates tested in another British study, 40% were HS 2 (Fricker and Park 1989). Sheep yielded mostly HS 1, 2, 4 complex, 6,7 and 23 (Jones et al. 1984). Only a few isolates from lamb meat were serotyped by Fricker and Park (1989), and these were mostly HS 2, 4, 9 and 48. The same authors reported a predominance of HS 2 and 4 in offal, although the animal of origin was not recorded.

HS 21 was the serotype most frequently isolated from chicken carcasses in this study, followed by HS 2. HS 21 was also the most frequently isolated serotype from chicken portions in a Christchurch study (Hudson et al. 1999), but was not detected in human cases in either this or the Christchurch study. However 14% of Wellington/Hutt Valley (North Island) human isolates of Campylobacter typed in 1997 were identified as HS 21 (C. Nicol, unpubl. obs.).

Isolates from Danish poultry products were of serotypes similar to those reported here, with the five most frequently isolated serotypes being HS 1,44, 2, 3, 4 complex and 5 (Nielsen and Nielsen 1999). Chicken isolates were mostly HS 1, 4 complex, 6,7 and 2 in a British study (Jones et al. 1984). HS 1, 2 and 4 complex were also isolated from chicken intestinal contents by Munroe et al. (1983), along with HS 3, 5 and 31. HS 4 complex was the most frequent chicken isolate serotype in another British study, but no single serotype, or group of serotypes, predominated (Hood et al. 1988). Fricker and Park (1989), again in a British survey, detected mostly HS 8, 1 and 2 in chicken isolates.

More recent Danish work compared serotype distributions in wildlife, human campylobacteriosis cases and broilers. Human campylobacteriosis was most often associated with HS 2 and 4 complex, a similar pattern to our results. Isolates from feral animals (e.g. hedgehogs, squirrels and ducks) were predominantly HS 4 complex and HS 12. HS 1,44 was confined to human cases and broiler flocks. It was concluded that wildlife did not contribute to serotypes causing human disease or infecting broilers, but that there was significant overlap of broiler and human clinical serotypes (Petersen et al. 2001).

In the present study, Ashburton river water yielded a diverse range of serotypes, as might be expected given the numerous inputs into a river system. Unlike other studies (Jones et al. 1984; Fricker and Park 1989) HS 2 was only the fifth most frequently isolated serotype from Ashburton waterways.

Eighty-nine (14·4%) of the 616 C. jejuni isolates were untypable by serotyping. This proportion is similar to that reported elsewhere. Figures as low as 1·5% have been reported (Nielsen and Nielsen 1999) but only after repeated subculture and testing.

SmaI-PFGE types

As it is not possible to compare SmaI-PFGE data from this study directly to those published elsewhere, as typing systems and nomenclature vary between researchers, comparisons are drawn between this and a previous New Zealand study where the same typing methodology was used (Hudson et al. 1999). Because of the small number of C. coli isolates, comment on SmaI-PFGE data alone (Table 4) is confined to C. jejuni isolates.

Most of the SmaI-PFGE types reported by Hudson et al. (1999) from Christchurch were also identified in isolates from Ashburton (the two being 95 km apart). The most predominant type isolated in Christchurch, P25, came from human cases, raw chicken and water samples. This was also the case with the P25 isolates identified in the current study, but with additional isolations from duck and ruminant faeces.

SmaI-PFGE types 18, 18a, 18b (clonally related subtypes) and P3 isolates, were identified quite frequently in the Christchurch study but not from water. P18, 18a, 18b and 18c isolates were not obtained from water in the present study, and P3 isolates, which were a large group (n = 28), were not isolated from water either. The absence of these clonally related SmaI-PFGE types from water samples is notable given that, from both studies, more than 60 isolates of this type were obtained. Possibly, isolates of this group are poor survivors in water.

Penner serotyping: SmaI-PFGE subtypes

Isolates belonging to subtypes that were indistinguishable by both Penner serotyping and SmaI-PFGE analysis can be considered as being closely related. The most frequently isolated subtypes from human cases were HS2:P18a and HS11:P35. In the case of the former, only one other isolate of this type was detected, and that was in duck faeces. In the case of the latter, other isolates were from water, beef and dairy cattle faeces. This might indicate that bovine faeces are a source of human pathogenic campylobacters. This possibility is supported by the observation that isolates of the most frequently identified subtype, HS23,36:P19b (40 isolates) came from water, cattle faeces and two human cases. Evidence for such an association has been shown in a Swedish epidemiological study (Nygård et al. 2004) and from multilocus sequencing data (Colles et al. 2003; Manning et al. 2003). However, direction of transmission cannot be inferred from the data presented here and a third unidentified route may have infected both cattle and human cases.

Two subtypes, HS2:P1c and HS22:P28, were only isolated from human cases and so the source infection of these cases was possibly via a transmission route not considered.

The overlap between water and duck isolate subtypes shown by subtype H8,17:P236 would be expected from direct and indirect deposition of duck faeces into waterways. This subtype may be adapted to live in ducks, or the association may be an artefact of the dataset.

Interestingly, subtypes H1,44:P3a and H1,44:P33 were present in both offal and human faeces but were not detected in the faeces of ruminant animals. This might indicate that the source of the contamination of the offal and infection of the cases was the same; an unknown reservoir or transmission route, or again that it was simply an artefact of the dataset.

Eight of the ten subtypes from Christchurch (Hudson et al. 1999) were detected among the Ashburton isolates. The only group that the Christchurch study identified as occurring in both summer and winter was HS2:P18/18a. In the current study P18 was not isolated, but P18a and the clonally related P18c were, although only a few isolates of this subtype were obtained, four from human cases and one from duck faeces. Nineteen isolates of this subtype were detected in both surveys and 14 originated from human cases. The other sources were veterinary cases (2), duck faeces (1) and chicken (2).

HS2:P3 isolates were all from human cases in the Christchurch study. However this subtype, and its clonal relatives, were common in the isolates described here. Of the 35 isolates, 19 originated from ruminant faeces, seven from chicken carcasses, six from offal, one from duck faeces, and one from a human case. Possibly this subtype was present but not detected in the earlier work as faeces from healthy ruminants and offal were not tested.

HS33:P25 was the most frequently isolated single subtype of C. jejuni isolated in the Christchurch study (Hudson et al. 1999) and was only isolated from human cases and chicken portions. No isolates of this subtype were identified in the current study. This might be due to geographical or demographic reasons, fluctuations in extant subtypes, or possibly because an outbreak caused by this subtype occurred during the Christchurch study.

Because of the many subtypes which were isolated in this study only tentative conclusions can be drawn. However, hypotheses which warrant further investigation include; (i) cattle may act as a significant reservoir of C. jejuni subtypes capable of causing human campylobacteriosis, (ii) some subtypes may be associated with particular environmental reservoirs as has been suggested by genotyping (Moser et al. 2002; Colles et al. 2003; Manning et al. 2003) and never, or rarely, cause human disease, and (iii) not all reservoirs and transmission routes of human disease for this population were considered in this study.


This work was funded by the Ministry of Health. The authors would like to thank the following people and organizations for their support and advice: Crown Public Health: CPH Timaru, including Monika Hansen and Chris Ambrose; the Christchurch Public Health Laboratory; Ashburton District Council: Richard Durie, Dennis Burridge and Peter Thompson for sample collection; Sally Harrow (University of Canterbury) and Sue Walker (KSC, ESR) for the PFGE subtyping of isolates; Ruth May (KSC,ESR) for the serotyping of isolates; Lisa Lopez and Kylie Gilmour (KSC, ESR) for support with the notification data: Ruth Pirie (ESR) for GIS information and maps; Els Maas for help with methodology; Phil Carter (ESR) for PFGE expertise and Margaret Tanner (CSC) for her patience and support. We also thank all the farmers and retailers of the Ashburton District for their cooperation in the collection of samples.

We thank Rob Lake and Hillary Michie for their comments and help in preparing this manuscript.