Procedures for trapping and handling raccoons were approved by the Animal Care Committee at the University of Guelph following the guidelines of the Canadian Committee on Animal Care. For a detailed description and map of the study area as well as trapping and sampling procedures, please refer to Bondo et al. (27) and (29). From May 2011 to November 2013, raccoons were live-trapped on swine farms and conservation areas near the cities of Guelph and Cambridge in southern Ontario, Canada. Individual animals were identified by ear and transponder tags and were sampled only once per monthly trapping week; however, additional samples were collected from the same individual if they were caught in subsequent months. Rectal fecal swabs were collected using Cary-Blair swab applicators [BBL CultureSwab, (BD) Becton, Dickinson and Company, Annapolis, MD, USA]. At each swine farm, one lagoon (i.e., manure pit) sample was collected on the first day of each trapping week as previously described (29). Briefly, to collect lagoon samples, a 24′ Nasco Swing Sampler (Conbar, Monroeville, NJ, USA) was used to collect three sub-samples from three locations around the pit, and up to two depths (i.e., the top 1/3, and mid depth of the storage), for a total of six sub-samples. The samples were then pooled into one sample for analysis. All samples were kept refrigerated or on ice until processing. The median and average number of days between collection and processing was three; the maximum number of days for raccoon swabs was eight whereas for manure pit samples it was 11.

Isolation of thermophilic Campylobacter species was performed as previously described (34). For samples collected prior to October 2011, the “conventional method” (enrichment followed by direct plating onto selective media) was used; the “membrane method” (enrichment followed by passive membrane filtration onto selective media) was used thereafter. Briefly, fecal swabs were immersed in 20 mL of Bolton broth (BB) containing BB selective supplement (SR0183, Oxoid), (20 mg/L cefoperazone, 20 mg/L vancomycin, 20 mg/L trimethoprim, and 50 mg/L cyclohexamine) and 5% laked horse blood (SR0048) and mixed rapidly for 20 s. Lagoon samples were mixed prior to adding 1 mL of sample to BB with supplement. Bolton broth cultures were incubated at 42°C in microaerobic conditions (10% CO₂, 5% O₂, 85% N₂) for 24 h and subsequently streaked onto modified blood-free charcoal cefoperazone deoxycholate agar supplemented with 32 mg/L cefoperazone and 10 mg/L amphotericin B (mCCDA) or passively filtered for 15 min through 0.65 μM cellulose acetate membrane filters onto the surface of mCCDA. All mCCDA plates were incubated for 48 h under the same microaerobic conditions. Three to five Campylobacter-like colonies were subcultured to blood agar and incubated for 24–48 h. Presumptive Campylobacter spp. colonies were identified on the basis of growth on mCCDA media, colony morphology, and oxidase tests. DNA was extracted from purified cultures using the EZ1 DNA tissue kit (Qiagen) according to the manufacturer's instructions for subsequent PCR speciation of presumptive Campylobacter isolates and subtyping of confirmed Campylobacter isolates.

Presumptive Campylobacter isolates were confirmed by multiplex PCR targeting a Campylobacter genus-specific region of the 16S rRNA gene, and mapA and ceuE genes for C. jejuni and C. coli identification, respectively (35). Amplicons were visualized using a QIAxcel capillary electrophoresis instrument with the DNA Screening kit. The AM320 separation method was used along with a 15–3,000 bp alignment marker and the QX 100–2.5 kb DNA size marker. Data were analyzed and visualized using the BioCalculator v. 3.0 software.

Campylobacter spp. isolates were subtyped by Comparative Genomic Fingerprinting (CGF) as previously described (36). Briefly, CGF consists of 8 multiplex PCR reactions that together assess the presence or absence of a set of 40 accessory gene targets found to have variable carriage in the Campylobacter population and that are used to generate a highly discriminatory binary fingerprint. Products from the CGF PCRs were visualized on the QIAxcel as previously described (36) and were scored positive (1) or negative (0) based on presence or absence of each target amplicon using a combination of the BioCalculator software's binary peak calling and confirmed with visual curation. The resulting binary fingerprints were assigned a three-digit CGF subtype (e.g., 0923.002.001) derived from cluster membership in the Canadian Campylobacter CGF database (C3GFdb). Fingerprints identical to those already existing in the database were assigned the appropriate CGF subtype, while novel fingerprints were assigned a CGF subtype based on their similarity to existing fingerprints in the database. Isolates from the same sample with identical subtypes were assumed to be derived from a single clone and only one representative isolate was included in further analyses.

At the time of this analysis, the C3GFdb consisted of 4,847 distinct CGF profiles obtained from 19,141 Campylobacter isolates collected from across Canada, primarily from the last decade. These included 23.3% isolates derived from human clinical origin, 32.0% from poultry sources, 20.9% from cattle sources and 14.6% from environmental surface water samples. To compare epidemiologic attributes of subtypes observed among study isolates, we compiled summary statistics based on the composition of host-sources observed in the C3GFdb for each subtype in this study. We classified CGF subtypes observed in this study (n = 114) into four ecological range categories (ERC) based on the composition of sources from which they have been historically observed in the C3GFdb. Subtypes from the present study included: (I) Raccoon Exclusive (RE; n = 44); (II) Raccoon and Environmental Water (REW; n = 16); (III) Swine/Swine Lagoon (SSL; n = 10); and (IV) Mixed Host (MH; n = 44). Phylogenetic and source association analysis was performed using the R language for statistical computing (v.3.13) (37) and the following packages: tidyverse (38) and ggtree (39). A dendrogram was constructed from the 40-gene CGF profile using the “hamming” distance and “average linkage” clustering from the dist and hclust functions, respectively, and visualized using “ggtree.” Clade designations were made based on a tree height “h = 10,” which produced stable, genetically distinct groupings. For each of the CGF subtypes observed in the present study we identified the associated host-sources from the C3GFdb. For associations between CGF subtypes and MLST Clonal Complexes, we used an in-house database created from in silico CGF and MLST predictions generated from publicly available WGS data analyzed using the program Microbial in silico Typer (40); a table of associations is provided as Supplementary Table S1.

Forty-three percent (n = 272/628) of the raccoons in this study were captured on multiple occasions, with 18.5% (n = 116/628) captured three or more times, up to maximum of eight captures. These data were analyzed to examine strain dynamics at the level of individual animals as described below.

CGF subtype diversity was assessed using the Simpson's Diversity Index (41) with a 95% confidence interval (42). A chi-square test statistic was used to explore the hypothesis that there was no significant difference in the distribution of isolates from prevalent Campylobacter clades (isolate count n > 9) between each of the location types (e.g., swine farm sites and conservation sites). Chi-square test statistics and post-hoc follow-up calculations for computing adjusted standardized residuals and z-scores for each clade were performed as described by Sharpe (43).

From May 2011 to November 2013, we collected 1,096 fecal swabs from 628 raccoons trapped on conservation areas (n = 687) and swine farms (n = 409), and 50 manure pit samples from the environment of the swine farm sites. The prevalence of Campylobacter spp. in raccoon fecal samples was 46.3% (508/1,096) (Table 1). There were no significant differences in Campylobacter prevalence from raccoon fecal swabs with a sample-to-test interval of 1–2 days (46.7%, n = 467) vs. a 3–4 days (46.0%, n = 363), 5–6 days (48.5%, n = 227), or 7–8 days interval (33.3%, n = 39). Similarly, for swine manure pit samples, there were no significant differences in Campylobacter prevalence between sample-to-test intervals of 1–2 days (40.0%, n = 10) vs. 3–4 (38.1%, n = 21), 5–6 (33.3%, n = 9), or 7–11 days (22.2%, n = 9). Among the Campylobacter positive raccoon samples, 502 (98.8%) were positive for C. jejuni, six (1.2%) for Campylobacter spp. (unidentified Campylobacter species), and one for C. coli (Table 1). A single sample was found to harbor mixed Campylobacter species, testing positive for both C. jejuni and an undefined Campylobacter spp. Among swine manure pit samples 34.0% (17/50) were positive for Campylobacter, of which 94.1% (16/17) were positive for C. coli (Table 1). One sample tested positive for Campylobacter spp.

A total of 1,555 confirmed Campylobacter isolates derived from Campylobacter-positive raccoon fecal swabs were subtyped by CGF. A further 58 isolates were recovered from swine manure pit samples. After accounting for potentially clonal isolates (i.e., isolates from the same sample sharing the same CGF subtype), 610 isolates remained; this included 581 raccoon isolates and 29 manure pit sample isolates. Among the Campylobacter positive raccoon samples (n = 508), a single CGF subtype was recovered for the majority of the samples (443/508; 87.2%), two CGF types were recovered in 11.2% of samples (57/508), and three subtypes in 1.6% of samples (8/508). We observed a broad genetic diversity in the Campylobacter population, with a total of 114 distinct CGF subtypes identified among study isolates. Subtype clusters ranged in size from 1 to 36 isolates, with 13 subtypes comprising over 50% of isolates in the dataset (n = 329/628) and 51 subtypes detected in single instances. Forty-six subtypes, representing 21.3% of study isolates (n = 130/610), were novel and had not been previously observed in the C3GFdb. There were 96 subtypes observed among raccoon isolates (C. jejuni = 91; Campylobacter spp. = 4; C. coli = 1) and 18 subtypes observed among lagoon isolates (C. coli = 17; Campylobacter spp. = 1); no subtypes were shared between raccoons and manure pit samples. Clustering of the 114 CGF binary profiles revealed 8 major lineages, designated clades A–H (Figure 1). The clades segregated by Campylobacter species: clades A–C, G, and H comprised C. jejuni isolates; clade F comprised C. coli isolates; and clade D comprised non-jejuni/coli Campylobacter species isolates (i.e., undetermined). The majority of raccoon isolates were observed in clade A (52.0%; n = 302), followed by clades B and C, with 30.5% (n = 177) and 14.5% (n = 84) of isolates, respectively. A small number of raccoon isolates (3.1%; n = 18) were found distributed among clades D (n = 5), E (n = 1), F (n = 1), G (n = 2), and H (n = 9). Among swine lagoon isolates, the majority were found in clade F (89.7%; n = 26), with the remaining isolates (10.3%; n = 3) in clade D. Similar levels of diversity were observed between swine farms and conservation areas at the subtype level (Simpson's Index of 0.9648 and 0.9628, respectively). Although overall genotypic richness was similar between swine farms and conservation areas, there was a significant relationship between clade and location type. For clade A, more isolates were identified from swine farm sites compared to conservation sites (p < 0.001). In contrast, for clades B and C, fewer isolates were identified from swine farm sites compared to conservation sites (p < 0.01).

Among the four ecological range categories that we defined for the subtypes in this study (Raccoon Exclusive or RE; Raccoon and Environmental Water or REW; Swine/Swine Lagoon or SSL; and Mixed Host or MH), raccoon isolates were evenly distributed among RE, REW and MH subtypes (RE: 31.3%, n = 182; REW: 34.1%, n = 198; MH: 34.6%, n = 201). Swine lagoon isolates were similarly distributed among SSL and MH subtypes (SSL: 51.7%, n = 15; MH: 48.3%, n = 14). The distribution of ERC subtypes suggests a strong raccoon association for certain clades observed in our analysis (Figure 1). Clade C, which included 14.5% of raccoon isolates in the study, was strictly composed of RE subtypes, and these were novel to the C3GFdb. Clade A, which included the majority of raccoon isolates in the study (52.0%), comprised largely RE and REW subtypes (n = 33/38). Source frequencies from the C3GFdb (Table 2) indicate that a majority of isolates from clade A subtypes (n = 461) include raccoon isolates from this study (65.5%; n = 302) along with historical isolates from environmental water sources (26.2%; n = 121). A small number of isolates (6.9%; n = 32) comprised the remaining isolates in the clade and were derived from cattle, poultry, and human clinical sources (Table 2). Conversely, clades B and H, which included 30.5 and 1.6% of raccoon isolates, respectively, were of MH subtype composition (n = 23/29 and n = 6/7, respectively). Based on C3GFdb source association data, isolates from subtypes in clades B (n = 2,118) and H (n = 1,329) were primarily chicken-associated (clade B = 54.9%; clade H = 55.0%). Importantly, these clades included a large proportion of isolates from human clinical cases (14.4 and 33.3%, respectively), which is in contrast to clades A and C (1.7 and 0.0%, respectively). Unsurprisingly, the C. coli specific-clade F included 26/29 swine manure pit isolates and consisted of 9 SSL subtypes and 9 MH subtypes.

A total of 468 samplings (43%; n = 468/1,096) involved consecutive recapture events involving the same animal (Figure 2), with a majority (77%; n = 361/468) occurring within the same sampling year. The positive rate for Campylobacter was 67% (n = 242/361) and 69% (n = 74/107) among recaptures in the same and different sampling years, respectively (Figure 2). Among recaptures within the same sampling year, 58% (n = 209/361) reflected a Campylobacter status change, with either animals that acquired or lost Campylobacter (n = 139), and animals that acquired a different subtype (n = 70) between recaptures. This rate was 67% among recapture events in different sampling years (n = 72/107), including 52 cases of animals acquiring or losing Campylobacter between recaptures and 20 cases of animals acquiring a different subtype. Overall, 60% (n = 281/468) of recapture events reflected a change in Campylobacter status, although the rate rises to 89% (n = 281/316) when excluding recaptures in which the animal tested negative on both occasions (n = 152). There were 125 instances of consecutive positive results, with a median of 35 days between recaptures (minimum = 25 days, maximum = 617 days). Of these, 40.0% (n = 50/125) occurred within a window of 0–30 days (i.e., short), an additional 33.6% (n = 42/125) occurred in a window of 31–60 days (i.e., intermediate), and 26.4% (n = 33/125) occurred in a window of >60 days (i.e., long), which included 22 recaptures that occurred on different sampling years. Overall, 72.0% of consecutive recaptures that were positive on both occasions (n = 90/125) yielded a different subtype on consecutive samplings. Among the 28% of consecutive recaptures yielding the same subtype (n = 35/125), we observed a decreasing proportion of matching subtypes as the time between sampling events increased (Figure 3). Furthermore, when examined in the context of ecological range, we observed that a larger proportion of recaptures in which the same subtype was consecutively isolated involved raccoon-associated subtypes (i.e., categories RE and REW) compared to MH subtypes (Overall: 30/35; 0–30 days: 15/18; 31–60 days: 12/14; and >60 days: 3/3). Among recapture events yielding changes in Campylobacter status on consecutive observations, the ratio of instances of strain acquisition to strain loss was similar across the three major raccoon-containing clades (A: 126/90; B: 66/46; C: 41/24 and across the various ERCs (MH: 73/57; RE: 80/53; REW: 87/56), all of which were similar to the overall ratio of acquisition-to-loss (n = 240/166).

The authors declare that the research was conducted in the absence of any commercial or financial relationships that could be construed as a potential conflict of interest.