Received 7 May 2010/Accepted 9 November 2010

APPLIED AND ENVIRONMENTAL MICROBIOLOGY, Feb. 2011, p. 882 888 Vol. 77, No. 3 0099-2240/11/$12.00 doi:10.1128/aem.01111-10 Copyright 2011, American Society for Microbiology. All Rights Reserved. Antimicrobial Resistance in Generic Escherichia coli Isolates from Wild Small Mammals Living in Swine Farm, Residential, Landfill, and Natural Environments in Southern Ontario, Canada Samantha E. Allen, 1 Patrick Boerlin, 1 Nicol Janecko, 2 John S. Lumsden, 1 Ian K. Barker, 1 David L. Pearl, 2 Richard J. Reid-Smith, 2,3 and Claire Jardine 1 * Department of Pathobiology 1 and Department of Population Medicine, 2 Ontario Veterinary College, University of Guelph, Guelph, Ontario, Canada, and Laboratory for Foodborne Zoonoses, Public Health Agency of Canada, Guelph, Ontario, Canada 3 Received 7 May 2010/Accepted 9 November 2010 To assess the impacts of different types of human activity on the development of bacteria in the feces of wild small mammals, we compared the prevalences and patterns of antimicrobial resistance and resistance s in ric Escherichia coli and Salmonella enterica from fecal samples collected from wild small mammals living in four environments: swine farms, residential areas, landfills, and natural habitats. Resistance to antimicrobials was observed in E. coli from animals in all environments: 25/52 (48%) animals trapped at swine farms, 6/69 (9%) animals trapped in residential areas, 3/20 (15%) animals trapped at landfills, and 1/22 (5%) animals trapped in natural habitats. Animals trapped on farms were significantly more likely to carry E. coli with resistance to tetracycline, ampicillin, sulfisoxazole, and streptomycin than animals trapped in residential areas. The resistance s sul2, aada, and tet(a) were significantly more likely to be detected in E. coli from animals trapped on farms than from those trapped in residential areas. Three S. enterica serotypes (Give, Typhimurium, and Newport) were recovered from the feces of 4/302 (1%) wild small mammals. All Salmonella were pansusceptible. Our results show that swine farm origin is significantly associated with the presence of bacteria and resistance s in wild small mammals in southern Ontario, Canada. However, fecal bacteria were found in small mammals living in all environments studied, indicating that environmental exposure to antimicrobials, antimicrobial residues, bacteria, or resistance s is widespread. Downloaded from http://aem.asm.org/ Antimicrobial resistance (AMR) is a global concern for animal and human health (35, 44). Recently, there has been increasing interest in bacteria and resistance s isolated from wildlife and the environment. Bacteria to antimicrobials have been detected in a variety of wildlife species (8, 9, 33), and wildlife have been implicated as potential reservoirs of bacteria and resistance s (4). In the bacteria of humans and domestic animals, AMR develops primarily as a consequence of exposure to antimicrobials (40, 45, 51). Colonization or infection by a bacterial strain can occur as the result of consumption of antimicrobials or antimicrobial residues or by direct transfer of bacteria or resistance s through direct contact with bacteria or from consumption of contaminated food and water (4, 40). Wild animals are not expected to be exposed directly to antimicrobials, and the source of AMR in the bacteria of wild animals is not clear. Resistant bacteria have been found at high prevalence in the intestinal bacteria of wild rodents living in proximity to livestock (20% and 54%) (22, 20) and among Escherichia coli recovered from the feces of wild Canada geese (Branta canadensis) and choughs (Pyrrhocorax pyrrhocorax) that were feeding and/or living in close proximity to livestock waste (72% and 21% to amoxicillin-clavulanic acid) (4, 7). Proximity to humans has also been associated with higher prevalence of AMR. For example, bacteria were detected more frequently in baboons feeding on human refuse than in animals living in more remote areas with no human contact (37). These studies suggest that exposure to anthropogenic factors, such as human refuse and livestock farming, may contribute to the development and maintenance of AMR in the bacteria of wildlife. However, the relationship between anthropogenic factors and AMR is not always clear. Gilliver et al. found that resistance to antibiotics is widespread (90%) in at least some wild rodent populations living in areas with no known exposure to antibiotics, whereas Osterblad et al., sampling in natural environmental areas, did not detect widespread bacteria (1 to 24%) in wildlife (13, 31). The main objective of this study was to assess the impacts of different types of human activity on the development of AMR in wildlife by comparing the prevalence and patterns of antimicrobial resistance and resistance s in ric E. coli and Salmonella from fecal samples collected from wild small mammals living on swine farms and in residential areas, landfills, and natural habitats. on March 16, 2019 by guest * Corresponding author. Mailing address: Department of Pathobiology, University of Guelph, Guelph, Ontario N1G 2W1, Canada. Phone: 519-824-4120, ext. 54656. Fax: 519-824-5930. E-mail: cjardi01@uoguelph.ca. Published ahead of print on 3 December 2010. MATERIALS AND METHODS Trapping and sampling. All procedures for trapping and handling small mammals were approved by the Animal Care Committee, University of Guelph, following the guidelines of the Canadian Council on Animal Care. We targeted wild mammals with small home ranges in this study so that we could be confidant 882

VOL. 77, 2011 ANTIMICROBIAL RESISTANCE IN WILD SMALL MAMMALS 883 Environment TABLE 1. Numbers of sites and trap nights and prevalences of E. coli recovery from species from different environments No. of sites Sampling dates No. of trap nights Peromyscus sp. No. of animals from which E. coli was isolated/no. of animals trapped M. musculus B. brevicauda M. pennsylvanicus T. striatus Rattus norvegicus Swine farm 9 May to July 2598 7/13 43/52 0/1 0/1 1/1 1/1 52/69 Natural 9 June to October 2646 15/31 0 1/8 0/2 6/7 0 22/48 Landfill 8 June to August 2332 10/18 6/6 2/6 1/2 1/1 0 20/33 Residential 10 June to August 2953 42/109 4/7 9/17 0/1 14/18 0 69/152 Total 36 10529 74/171 53/65 12/32 1/6 22/27 1/1 163/302 Total that their home ranges fell within the sampling environment where we were trapping. Small mammals were trapped at nine swine farms, nine natural habitats, eight landfills, and 10 residential areas in southern Ontario, Canada, from May to October 2008 (Table 1). All sampling sites were within a 100-km radius of Guelph, Ontario, which is located within one of the most densely populated areas of Canada. All of the swine farms were active, and traps were set around farm buildings, hedgerows, and lagoons within 150 m of a barn. Natural woodland habitats were selected in conservation areas administered by the Grand River Conservation Authority, Conservation Halton, and the Credit Valley Conservation Authority. The traps at these sites were placed in wooded areas that were 100 m in from park edges and away from buildings. Most landfills had open wet waste and active landfill pits, although in one case the active landfill pit was enclosed within a building. At landfills, traps were placed as near as possible to wet waste and the active landfill pits. All residential areas were in established urban/suburban neighborhoods ( 10 years old) and consisted of a cluster of 4 to 5 houses. These sites were excluded if they were within 100 m of a different environment area. For example, if a swine farm was adjacent to a residential area, both were excluded. At each site, 100 Sherman live traps (H. B. Sherman Traps, Inc., FL) were placed along 10 transects with 10-m spacing between traps. The traps were baited with peanut butter and oats, and cotton was provided as bedding. At each site, traps were set in the evening and checked early the following morning for three consecutive nights. Traps that were disturbed were removed from the calculation of the number of trap nights (number of trap nights number of sites (number of traps set number of traps disturbed) number of nights). Appropriate precautions were taken by workers to avoid exposure to zoonotic agents (27). Animals were identified to the genus level (in the case of Peromyscus, either Peromyscus maniculatus (deer mouse) or Peromyscus leucopus (whitefooted mouse) or to the species level (for all other animals). Peromyscus sp., Mus musculus (house mouse), Microtus pennsylvanicus (meadow vole), and Blarina brevicauda (short-tailed shrew) were euthanized with an overdose of halothane (MTC Pharmaceuticals, Ontario, Canada) for use in other studies. Other species were released at the point of capture without being handled. Feces were collected directly from the traps of all animals by using clean forceps. All fecal samples were placed in individual containers with 5 ml of brain heart infusion (BHI) broth (Difco, Ontario, Canada) containing 20% glycerol. The samples were kept for up to 2hinacooler in the field and then mixed thoroughly by vortexing them at the laboratory, where they were refrigerated prior to being frozen at 70 C (within 12 h of sample collection) for up to 1 week. E. coli isolation. The samples were thawed, and 3 ml was transferred to a tube containing 3 ml of buffered peptone water (BPW) (Becton Dickinson, Ontario, Canada), which was incubated at 37 C overnight. Equal amounts of 2 E. coli broth (Difco) and BPW were added to the tubes, and they were then incubated at 37 C overnight. After overnight incubation, E. coli broth samples were streaked onto MacConkey s agar (Becton Dickinson) and incubated at 37 C overnight. If lactose-fermenting colonies were obtained the next day, six different colonies were selected at random, restreaked onto MacConkey s agar (Becton Dickinson), and grown overnight at 37 C. The next day, each colony was streaked onto Trypticase soy agar (TSA) plates and incubated at 37 C overnight. Presumptive E. coli identifications were confirmed by indole and oxidase tests. Three indole-positive and oxidase-negative colonies per fecal sample were frozen at 70 C in BHI containing 20% glycerol for later testing. Only fecal samples that yielded E. coli were included in the study. Isolation of Salmonella spp. The samples were preenriched in BPW and incubated at 37 C for 24 h; 100 l of the enrichment culture was inoculated into modified semisolid Rappaport-Vassiliadis medium (MRSV) (Becton Dickinson). The plates were incubated at 42 C for 24 to 72 h and examined for a typical migration pattern. All subsequent incubations and tests were done at 37 C. Presumptive positive samples from MSRV plates were streaked onto MacConkey s agar, followed by subcultures of three non-lactose-fermenting colonies on TSA. Confirmatory testing for Salmonella was conducted using triple-sugar iron, Christenssen s urea agar, and Salmonella O antiserum Poly A-I and Vi (Becton Dickinson). Susceptibility testing. Up to three from each sample were submitted for susceptibility testing to the Laboratory for Food-Borne Zoonoses, Public Health Agency of Canada, Guelph, Ontario, Canada. The broth microdilution method was used (Sensititre System; Trek Diagnostics, OH), following the protocols of the Canadian Integrated Program for Antimicrobial Resistance Surveillance (14, 30). E. coli from each sample were tested for the following antimicrobial agents (breakpoints are indicated in parentheses): amikacin ( 64 g/ml), amoxicillin-clavulanic acid ( 32 and 16 g/ml, respectively), ampicillin ( 32 g/ml), cefoxitin ( 32 g/ml), ceftiofur ( 8 g/ml), ceftriaxone ( 2 g/ ml) (11), chloramphenicol ( 32 g/ml), ciprofloxacin ( 4 g/ml), gentamicin ( 16 g/ml), kanamycin ( 64 g/ml), naladixic acid ( 32 g/ml), streptomycin ( 64 g/ml), sulfisoxazole ( 512 g/ml), tetracycline ( 16 g/ml), and trimethoprim-sulfamethoxazole ( 4 and 76 g/ml, respectively). For descriptive analysis, the antimicrobials were grouped according to the Canadian Veterinary Drug Directorate s categorization of antimicrobial drugs based on their importance to human medicine (category I is the most important, and category III is the least important) (49). Antimicrobial resistance detection. Two E. coli per sample were selected randomly to test for resistance s. E. coli lysates were prepared as described previously (28). Three or four colonies were grown in 500 l BHI overnight, and 20 l of this culture was transferred to 200 l lysis buffer (0.1 M Tris-HCl, ph 8.5, 0.05% Tween 20, 0.24 mg/ml proteinase K). The sample was incubated at 60 C for 1 h and heated at 97 C for 15 min to inactivate the proteinase K. The major resistance s for sulfonamides (sul1, sul2, and sul3), tetracycline [tet(a), tet(b), and tet(c)], aminoglycosides [aadb, apha1, apha2, aac(3)iv, stra/strb, and aada], phenicols (cmla, flor, and cata1), and betalactams (bla TEM, bla SHV, bla CMY-2, and bla CTX-M ) were tested with a set of multiplex and single PCR protocols using Qiagen multiplex PCR kits (Qiagen, Ontario, Canada). All multiplex PCRs were done in 25- l volumes using Qiagen multiplex PCR master mixture, 1 Q solution, and 1 primer mixture according to the manufacturer s instructions. PCR protocols and primers are described by Travis et al., Kozak et al., and Khashayar (19, 20, 46). Statistical analysis. We initially attempted to include the site as a random variable, but due to problems with model convergence, we pooled sites within each environment. Logistic regression models were constructed using ralized estimating equations with an exchangeable correlation structure to account for multiple from the same animal (STATA10 statistical software). The factors examined for association with resistance to 1 antimicrobial were environment (farm, residential, landfill, and natural) and host species. Univariable models were built first, and then a multivariable model was constructed. Exact logistic regression models were used to look for associations between the environment, specific resistance phenotypes, and resistance s at the animal level (STATA10 statistical software). We analyzed results at the animal level, since other methods used for clustered data would not converge due to small sample sizes. Associations were considered significant at a P value of 0.05. From these models, the odds ratio (OR) and 95% confidence interval (CI) of each variable were reported. RESULTS Recovery of E. coli and Salmonella. E. coli was isolated from 163 of 302 fecal samples (54%). Table 1 summarizes the sam-

884 ALLEN ET AL. APPL. ENVIRON. MICROBIOL. TABLE 2. Prevalences of resistance to individual antimicrobials in E. coli from wild small mammals trapped on swine farms and in natural habitats, landfills, and residential areas Prevalence in wild small mammals from: Swine farm Natural habitats Landfills Residential areas Total Antimicrobial agent a (n 155) (n 52) (n 63) (n 22) pling design and the number of small mammals of each of six species trapped in each environment of interest, as well as the E. coli recovery from each source. Salmonella enterica was isolated from 4 of 302 fecal samples (1%; 2 were obtained from 1 of the 4 samples). Two Salmonella enterica serotype Give were recovered from two separate M. musculus mice trapped on a landfill site, two Salmonella enterica serotype Newport were recovered from one Peromyscus sp. mouse trapped at natural site, and one Salmonella enterica serotype Typhimurium isolate was recovered from one Tamias striatus individual trapped on a farm. Prevalences of E. coli and Salmonella. The prevalences of resistance in and animals from swine farms, residential areas, landfills, and natural habitats are presented in Table 2. Eighty-three E. coli (17%) from 35 animals showed reduced susceptibility to antimicrobials (Table 2). Sixty-three were from 25 animals trapped on swine farms, 14 were from 6 animals trapped in residential areas, 5 were from 3 animals trapped in landfills, and 1 isolate was from 1 animal trapped in a natural habitat. Resistance to tetracycline was most common in animals from swine farms (58 E. coli from 25 animals) (Table 2). Resistance to ampicillin was observed most commonly in from residential areas (12 E. coli from 4 animals) (Table 2). In animals trapped at landfills, only resistance to ampicillin, kanamycin, and chloramphenicol was detected, while only resistance to chloramphenicol was detected in E. coli from natural-area animals (Table 2). All 5 Salmonella were susceptible to the 15 antimicrobials tested. Prevalence of resistance s. Resistance s were detected in 52 of 322 E. coli obtained from 28 animals (Table 3). Forty-two of these E. coli were from animals trapped on swine farms, eight were from animals trapped in residential areas, two were from animals trapped on landfills, and no resistance s were detected in E. coli from (n 57) (n 20) (n 200) (n 69) (n 475) (n 163) Ampicillin 28 (18.0) 10 (19.2) 0 0 3 (5.2) 1 (5.0) 12 (6.0) 4 (5.7) 43 (9.0) 15 (9.2) Amoxicillinclavulanic 11 (7.0) 4 (7.6) 0 0 0 0 10 (5.0) 4 (5.7) 21 (4.4) 8 (4.9) acid Cefoxitin 11 (7.0) 4 (7.6) 0 0 0 0 11 (5.5) 5 (7.2) 22 (4.6) 9 (5.5) Ceftiofur 11 (7.0) 4 (7.6) 0 0 0 0 9 (4.5) 3 (4.3) 20 (4.2) 7 (4.2) Ceftriaxone 10 (6.4) 4 (7.6) 0 0 0 0 9 (4.5) 3 (4.3) 19 (4.0) 7 (4.2) Streptomycin 24 (15.4) 13 (25.0) 0 0 0 0 4 (2.0) 2 (2.8) 28 (5.9) 15 (9.2) Kanamycin 2 (1.2) 1 (1.9) 0 0 3 (5.2) 1 (5.0) 2 (1.0) 1 (1.4) 7 (1.4) 3 (1.8) Gentamicin 7 (4.5) 3 (5.7) 0 0 0 0 2 (1.0) 1 (1.4) 9 (1.8) 4 (2.4) Sulfisoxazole 32 (20.6) 13 (25.0) 0 0 0 0 4 (2.0) 2 (2.8) 36 (7.5) 15 (9.2) Trimethoprimsulfamethoxazole 3 (1.9) 1 (1.9) 0 0 0 0 2 (1.0) 1 (1.4) 5 (1.0) 2 (1.2) Tetracycline 58 (37.4) 25 (48.0) 0 0 0 0 10 (5.0) 4 (5.7) 68 (14.3) 29 (17.7) Chloramphenicol 10 (6.4) 4 (7.6) 1 (1.6) 1 (4.5) 1 (1.8) 1 (5.0) 1 (0.5) 1 (1.4) 13 (2.7) 7 (4.2) a No resistance to amikacin, ciprofloxacin, or nalidixic acid was detected in E. coli from wild small mammals. animals trapped in natural habitats. The most frequently detected resistance s were tet(a), aada, bla TEM, sul2, and bla CMY-2 (Table 3). The most frequently observed resistance in bacteria from animals trapped on swine farms was tet(a), whereas in those from residential areas, bla TEM was the most frequently observed. There were discrepancies between the susceptibility test results and the genotyping results that were confirmed by repeated genotypic testing. Resistance s for streptomycin (aada and stra-strb) were detected in 38, but only 16 of these were classified as to streptomycin. In two of nine to chloramphenicol, the related resistance s were not detected, and resistance s for gentamicin were not detected in any of the seven showing resistance to gentamicin. In 2 of 18 to streptomycin, related resistance s were not detected, as was the case in 2 of 47 to tetracycline, in 1 of 26 to sulfisoxazole, and in 1 of 14 to cefoxitin. In all four to trimethoprim-sulfamethoxazole, the related sul resistance s were not detected; however, we did not test for dfr s, which can confer resistance to trimethoprim-sulfamethozaxole (15, 18). Associations between bacteria, host species, and environment. The univariable model indicated that the odds of finding resistance to antimicrobials were significantly higher in from swine farms than in from residential areas (Table 4). In addition, in our univariable model, the odds of detecting AMR were 13 times higher for M. musculus and eight times higher for B. brevicauda than for Peromyscus sp. (Table 4). The small number of samples collected from other species prevented their inclusion in the model. In the multivariable model, including the environment and host species, a significant association between the environment and overall AMR occurred, but there was no significant association between AMR and the host species. In the multivariable model, the odds of detecting AMR were 9 times

VOL. 77, 2011 ANTIMICROBIAL RESISTANCE IN WILD SMALL MAMMALS 885 TABLE 3. Prevalence of antimicrobial resistance s in E. coli from wild small mammals trapped in swine farms, natural, landfill and residential areas Prevalence in wild small mammals from: Swine farms Natural habitats Landfills Residential areas Total AMR a (n 104) (n 52) (n 43) (n 22) (n 39) (n 20) (n 136) (n 69) higher for from animals on swine farms than for from residential areas when the host species was controlled for (we were not able to include B. brevicauda in the model) (Table 4). In our multivariable model, we were unable to include natural habitats as an environmental variable and to test for interactions between the host species and the environment due to problems with model convergence. Associations between resistance to specific antimicrobials, resistance s, and environment. The odds of finding animals with resistance to ampicillin (OR 3.83; CI 1.02 to 17.82; P 0.046), streptomycin (OR 10.96; CI 2.30 to 105.08; P 0.01), sulfisoxazole (OR 10.96; CI 2.30 to 105.08; P 0.01), and tetracycline (OR 13.60; CI 4.15 to 58.94; P 0.01) were significantly higher on swine farms than in residential areas. No association was found between the environment and resistance to amoxicillin-clavulanic acid, ceftiofur, ceftriaxone, cefoxitin, gentamicin, kanamycin, chloramphenicol, and trimethoprim-sulfamethoxazole. The odds of detecting a number of resistance s, sul2 (OR 6.00; CI 1.13 to 60.63; P 0.032), aada (OR 13.13; CI 2.06 to ; P 0.003); and tet(a) (OR 12.40; CI 3.31 to 70.05; P 0.01) were significantly higher in animals from swine farms than those from residential areas. No significant associations were found between the environment and the resistance s sul1, sul3, cmla, apha1, stra-strb, tet(b), tet(c), bla TEM, and bla CMY-2. DISCUSSION (n 322) (n 163) bla TEM 14 (13.4) 10 (19.2) 0 0 2 (5.1) 1 (5.0) 7 (5.1) 4 (5.7) 23 (7.1) 15 (9.2) bla CMY-2 8 (7.6) 4 (7.6) 0 0 0 0 6 (4.4) 3 (4.3) 14 (4.3) 7 (4.2) stra-strb 6 (5.7) 3 (5.7) 0 0 0 0 3 (2.2) 2 (2.8) 9 (2.7) 5 (3.0) aada 24 (23.0) 16 (30.7) 0 0 0 0 6 (4.4) 3 (4.3) 30 (9.3) 19 (11.6) apha1 1 (0.9) 1 (1.9) 0 0 2 (5.1) 1 (5.0) 2 (1.4) 1 (1.4) 5 (1.5) 3 (1.8) sul1 1 (0.9) 1 (1.9) 0 0 0 0 2 (1.4) 1 (1.4) 3 (0.9) 2 (1.2) sul2 14 (13.4) 8 (15.3) 0 0 0 0 3 (2.2) 2 (2.8) 17 (5.2) 10 (6.1) sul3 7 (6.7) 4 (7.6) 0 0 0 0 0 0 7 (2.1) 4 (2.4) tet(a) 33 (31.7) 19 (36.5) 0 0 0 0 5 (3.6) 3 (4.3) 38 (11.8) 22 (13.4) tet(b) 5 (4.8) 4 (7.6) 0 0 0 0 3 (2.2) 2 (2.8) 8 (2.4) 6 (3.6) tet(c) 4 (3.8) 3 (5.7) 0 0 0 0 0 0 4 (1.2) 3 (1.8) cmla 7 (6.7) 4 (7.6) 0 0 0 0 0 0 7 (2.1) 4 (2.4) a The resistance s aadb, apha2, aac(3)iv, bla SHV, flor, cata1, and bla CTX-M were not detected in wild small mammals in this study. TABLE 4. Logistic regression models built using ralized estimating equations a (GEE) showing associations between AMR occurrence (resistance to antimicrobials), trapping area type, and host species type Variable OR 95% CI P c Univariable b Environment Natural 1 (2) 0.21 0.009 4.99 0.57 Swine farm 63 (41) 9.47 3.59 24.98 <0.01 Landfill 5 (9) 1.2 0.24 6.49 0.78 Residential (reference) 14 (7) Host species M. musculus 59 (39) 13.3 4.43 40.13 <0.01 B. brevicauda 10 (29) 8.2 1.84 36.78 <0.01 Peromyscus sp.(reference) 10 (5) Multivariable b Environment Swine farm 63 (41) 9.32 1.40 61.95 0.02 Landfill 5 (9) 1.99 0.19 20.36 0.56 Residential 14 (7) (reference) Host species M. musculus 59 (39) 3.48 0.85 14.10 0.08 Peromyscus sp. (reference) 10 (5) a GEE models were constructed with an exchangeable correlation structure to account for multiple from the same animal. b T. striatus, M. pennsylvanicus, and Rattus norvegicus were excluded from both models and natural habitats and B. brevicauda were excluded from multivariable analysis due to problems with model convergence. c Numbers in boldface represent statistically significant (P 0.05) associations. We found fecal E. coli in wild small mammals living in all four environments that we examined in southern Ontario (swine farms, natural habitats, residential areas, and landfills). Overall, bacteria were detected most frequently in E. coli from small wild mammals trapped on swine farms. Since bacteria to these antimicrobials are commonly detected in feces, manure pits, and environmental samples from swine farms (5, 17, 47), wildlife living on these

886 ALLEN ET AL. APPL. ENVIRON. MICROBIOL. farms may come into contact with bacteria and/or antimicrobial residues that could provide selection pressure for the development of bacteria in wildlife that have contact with these areas (4, 7, 23) or acquisition of bacteria from farm environments. Although we found a low prevalence of bacteria in wild small mammals living in residential areas, the spectrum of resistance phenotypes included the same 12 antimicrobials that were detected on swine farms. Interestingly, there was no significant difference in the frequencies of resistance to amoxicillin-clavulanic acid, cefoxitin, ceftiofur, ceftriaxone, kanamycin, gentamicin, and trimethoprim-sulfamethoxazole between farms and residential areas. Resistance to these antimicrobials is not rally detected in E. coli from Canadian livestock (i.e., beef and swine) (14), and consistent with a previous study (20), none of these resistance phenotypes were detected in small mammals from natural habitats. Our results suggest that the development of resistance to these antimicrobials in wild small mammals in residential areas may be associated with proximity to anthropogenic sources of AMR other than livestock. The source of AMR for wildlife living in residential areas is not clear. Rolland et al. detected a higher frequency of AMR in baboons that had direct contact with human refuse than in those that did not (37). Since wildlife may come into direct contact with discarded antimicrobial medication, antimicrobial residues, or bacteria, samples were sought in animals living in residential areas and in the vicinity of landfills. There was no evidence of increased risk of AMR at landfills relative to residential areas. We detected resistance to only three antimicrobials (ampicillin, kanamycin, and chloramphenicol) from animals trapped on landfills at frequencies similar to those in residential environments. Recreational water and wastewater have been identified as a potential source of bacteria in the environment (10, 16, 25), which could affect the gut bacteria of wildlife in residential areas. Silva et al. detected ampicillin- bacteria in the feces of urban pigeons (Columba livia) and hypothesized that sewer water may be the source (42). The small mammals in our study might have come into contact with antimicrobial residues and bacteria from human waste in waterways. Another possible source of AMR for wildlife living in residential areas is contact with raw meat scraps, since AMR was detected in Salmonella from raw-food diets purchased from Canadian pet stores (12). In addition, across Canada, bacteria have been detected in beef, chicken, and pork from retail meat samples (14). Further studies are required to investigate the association between proximity to humans and the occurrence of AMR and to identify potential sources of AMR for wildlife living in association with humans. The only resistance detected in bacteria from animals in natural habitats was to chloramphenicol, and it was detected in only a single isolate. Previous studies have reported varying frequencies of antimicrobial resistance in wildlife trapped in natural habitats (8, 26, 29, 32). Gilliver et al. detected a high prevalence (90%) of bacteria in wild bank voles and wood mice (Apodemus sylvaticus) that were trapped at two woodland localities in the United Kingdom (13). However, consistent with the results of this study, Osterblad et al. detected bacteria at low prevalences (1 to 24%) in feces recovered from wild moose (Alces alces), white-tailed deer (Odocoileus virginianus), and bank voles in Finland, and Kozak et al. detected a low prevalence (20%) of bacteria in wild small mammals from natural habitats in southern Ontario (20, 31). Unlike swine farms, natural habitats do not have an easily identifiable source of antimicrobials. Wild animals may come into contact with natural antibiotics produced by bacteria and fungi (32, 43, 49). In addition, since some antimicrobials used in human and veterinary medicine do not completely break down, they may spread through the environment in soil and wastewater (39). Resistant bacteria and antimicrobial residues originating from other environmental areas may contaminate areas perceived to be natural environments (2). The observed discrepancies in the prevalence of AMR in wildlife living in natural habitats in different geographic locations may reflect different levels of overall contamination in the regional environment (31). Also, wildlife, such as the black-headed gull (Larus ridibundus) or the Russian rook (Corvus frugilegus), could transport bacteria from other areas to natural habitats and then act as reservoirs, maintaining AMR resistance within natural ecosystems (4, 7, 9). In ral, there was good agreement between AMR phenotypes and the presence or absence of resistance s. For streptomycin, the disagreement between the presence of resistance s and resistance phenotypes was compatible with findings in other studies (20). This suggests that the breakpoints established to test for resistance to the antimicrobial may be too high for epidemiological purposes or that the is not being expressed (5, 20). The absence of relevant resistance s from that were phenotypically to other antimicrobials may reflect the fact that the observed phenotypic resistance was caused by a resistance other than those investigated here or by a nonspecific resistance mechanism (e.g., an efflux pump) (1, 3, 6, 24, 38, 41). The resistance s tet(a) and aada were detected most frequently in E. coli from animals from swine farms and were significantly associated with farm origin; sul2 was also significantly associated with farm origin, and although the resistance s sul3, tet(c), and cmla were not significantly associated with swine farm origin, they were detected solely in E. coli from wildlife trapped on swine farms. The resistance s bla TEM and apha1 were detected in from animals trapped at swine farms and landfills and in residential areas, suggesting that the may be more widely disseminated throughout the environment. The resistance apha1 was detected in from farms, residential areas, and landfills at roughly equal frequencies, suggesting that the may be associated with human activities or livestock-related exposure in residential areas. In addition, a number of were to multiple antimicrobials and carried more than one resistance. For example, the combination of cmla and tet(a) also was significantly associated with of swine farm origin. Because the wild small mammals targeted in this study have small home ranges and are unlikely to be purposefully treated with antibiotics, we suspect that they carried bacteria as a consequence of local environmental exposure to antimicrobials, antimicrobial residues, bacteria, or resistance s. Our ability to detect statistically significant asso-

VOL. 77, 2011 ANTIMICROBIAL RESISTANCE IN WILD SMALL MAMMALS 887 ciations between the prevalence of resistance s and the environment type was affected by the small number of samples collected in some environments, particularly natural habitats and landfills. The host species was not significant in our multivariable model once we controlled for the environment; however, further examination of potential interaction effects were not possible due to issues associated with model convergence. More intensive sampling is required to define any association between the occurrence of bacteria and s in wildlife and their origins in natural habitats and landfills and to investigate the role of the host species in the epidemiology of AMR. In addition further studies are required to investigate potential sources of bacteria and resistance s for wildlife in all environments. ACKNOWLEDGMENTS We thank T. King for assistance with small mammal trapping and B. Jefferson and G. Chalmers for help in the laboratory. We thank D. DiDonato, L. Lee, and M. Martin for all of the primary isolations. We also thank B. Bloomfield and R. M. Friendship for their help in gaining access to farm sites. We thank all landowners for giving us permission to trap on their properties, including the Grand River Conservation Authority, Conservation Halton, and the Credit Valley Conservation Authority. REFERENCES 1. Adrian, P., C. Thomson, K. Klugman, and S. Amyes. 2000. New cassettes for trimethoprim resistance, dfr13, and streptomycin-spectinomycin resistance, aada4, inserted on a class 1 integron. Antimicrob. Agents Chemother. 44:355 361. 2. Al-Bahry, S., et al. 2009. Bacterial flora and antibiotic resistance from eggs of green turtles Chelonia mydas: an indicator of polluted effluents. Mar. Pollut. Bull. 58:720 725. 3. Aslam, M., M. Diarra, C. Service, and H. Rempel. 2009. Antimicrobial resistance s in Escherichia coli recovered from a commercial beef processing plant. J. Food Prot. 72:1089 1093. 4. Blanco, G., J. A. Lemus, and J. Grande. 2009. Microbial pollution in wildlife: linking agricultural manuring and bacterial antibiotic resistance in red-billed choughs. Environ. Res. 109:405 412. 5. Boerlin, P., et al. 2005. Antimicrobial resistance and virulence s of Escherichia coli from swine in Ontario. Appl. Environ. Microbiol. 71:6753 6761. 6. Bunny, K., R. Hall, and H. Stokes. 1995. New mobile cassettes containing an aminoglycoside resistance, aaca7, and a chloramphenicol resistance, catb3, in an integron in pbwh301. Antimicrob. Agents Chemother. 39:686 693. 7. Cole, D., et al. 2005. Free-living Canada geese and antimicrobial resistance. Emerg. Infect. Dis. 11:935 938. 8. Costa, D., et al. 2008. Mechanisms of antibiotic resistance in Escherichia coli recovered from wild animals. Microb. Drug Resist. 14:71 77. 9. Dolejska, M., A. Cizek, and I. Literak. 2007. High prevalence of antimicrobial- s and integrons in Escherichia coli from black-headed gulls in the Czech Republic. J. Appl. Microbiol. 103:11 19. 10. Edge, T. A., and S. Hill. 2005. Occurrence of antibiotic resistance in Escherichia coli from surface waters and fecal pollution sources near Hamilton, Ontario. Can. J. Microbiol. 51:501 505. 11. EUCAST. 2009. Cephalosporins EUCAST clinical MIC breakpoints. http: //www.srga.org/eucastwt/mictab/miccephalosporins_v2.htm. 12. Finley, R., et al. 2008. The occurrence and antimicrobial susceptibility of Salmonellae isolated from commercially available canine raw food diets in three Canadian cities. Zoonoses Public Health 55:462 469. 13. Gilliver, M. A., M. Bennett, M. Begon, S. M. Hazel, and C. A. Hart. 1999. Antibiotic resistance found in wild rodents. Nature 401:233 234. 14. Government of Canada. 2007. Canadian integrated program for antimicrobial resistance surveillance (CIPARS) 2005. Public Health Agency of Canada, Guelph, Ontario, Canada. 15. Grape, M., L. Sundstrom, and G. Kronvall. 2007. Two new dfr s in trimethoprim- integron-negative Escherichia coli. Antimicrob. Agents Chemother. 51:1863 1864. 16. Hamelin, K., et al. 2006. A virulence and antimicrobial resistance DNA microarray detects a high frequency of virulence s in Escherichia coli from Great Lakes recreational waters. Appl. Environ. Microbiol. 72:4200 4206. 17. Heuer, H., and K. Smalla. 2007. Manure and sulfadiazine synergistically increased bacterial antibiotic resistance in soil over at least two months. Environ. Microbiol. 9:657 666. 18. Jansson, C., and O. Skold. 1991. Appearance of a new trimethoprim resistance, dhfrix, in Escherichia coli from swine. Antimicrob. Agents Chemother. 35:1891 1899. 19. Khashayar. 2009. Prevalence of antimicrobial resistance and of selected beta-lactams resistance s in Escherichia coli from canine urinary tract infections in Ontario and Quebec, p. 35 41. M.Sc. thesis. University of Guelph, Guelph, Ontario, Canada. 20. Kozak, G. K., P. Boerlin, N. Janecko, R. J. Reid-Smith, and C. Jardine. 2009. Antimicrobial resistance in Escherichia coli from swine and wild mammals in the proximity of swine farms and in natural environments in Ontario, Canada. Appl. Environ. Microbiol. 75:559 566. 21. Literak, I., et al. 2010. Antimicrobial- faecal Escherichia coli in wild mammals in central Europe: multi Escherichia coli producing extended-spectrum beta-lactamases in wild boars. J. Appl. Microbiol. 108: 1702 1711. 22. Literak, I., et al. 2009. Highly variable patterns of antimicrobial resistance in commensal Escherichia coli from pigs, sympatric rodents, and flies. Microb. Drug Resist. 15:229 237. 23. Literák, I., R. Vanko, M. Dolejska, A. Cizek, and R. Karpiskova. 2007. Antibiotic Escherichia coli and Salmonella in Russian rooks (Corvus frugilegus) wintering in the Czech Republic. Lett. Appl. Microbiol. 45:616 621. 24. Livermore, D. M., et al. 2001. Antibiotic resistance in bacteria from magpies (Pica pica) and rabbits (Oryctolagus cuniculus) from west Wales. Environ. Microbiol. 3:658 661. 25. Miao, X.-S., M. F. Bishay, M. Chen, and C. D. Metcalfe. 2004. Occurrence of antimicobials in the final effluents of wastewater treatment plants in Canada. Environ. Sci. Technol. 38:3533 3541. 26. Millan, J., G. Aduriz, B. Moreno, R. A. Juste, and M. Barral. 2004. Salmonella from wild birds and mammals in the Basque Country (Spain). Rev. Sci. Tech. Off. Int. Epiz. 23:905 911. 27. Mills, J., et al. 1995. Guidelines for working with rodents potentially infected with hantavirus. J. Mammal. 76:716 722. 28. Miserez, R., et al. 1998. Detection of alpha- and epsilon-toxigenic Clostridium perfringens type D in sheep and goats using a DNA amplification technique (PCR). Lett. Appl. Microbiol. 26:382 386. 29. Nascimento, A. M. A., et al. 2003. Short communication: antibiotic- gram-negative bacteria in birds from the Brazilian Atlantic forest. Condor 105:358 361. 30. NCCLS. 2000. Performance standards for antimicrobial disk susceptibility tests, 7th ed. NCCLS document M2-M7. NCCLS, Wayne, PA. 31. Osterblad, M., K. Norrdahl, E. Korpimaki, and P. Huovinen. 2001. How wild are wild animals? Nature 409:37 38. 32. Poeta, P., et al. 2007. Detection of antimicrobial activities and bacteriocin structural s in faecal enterococci of wild animals. Microbiol. Res. 162: 257 263. 33. Poeta, P., et al. 2005. Characterization of antibiotic resistance s and virulence factors in faecal enterococci of wild animals in Portugal. J. Vet. Med. B Infect. Dis. Vet. Public Health 52:396 402. 34. Poppe, C., et al. 2001. Trends in antimicrobial resistance of Salmonella isolated from animals, foods of animal origin, and the environment of animal production in Canada. Microb. Drug Resist. 7:197 212. 35. Pruden, A., R. Pei, H. Storteboom, and K. H. Carlson. 2006. Antibiotic resistance s as emerging contaminants: studies in northern Colorado. Environ. Sci. Technol. 40:7445 7450. 36. Rajić, A., et al. 2004. Antimicrobial resistance of Salmonella isolated from finishing swine and the environment of 60 Alberta swine farms. Vet. Microbiol. 104:189 196. 37. Rolland, R. M., G. Hausfater, B. Marshall, and S. B. Levy. 1985. Antibiotic bacteria in wild primates: increased prevalence in baboons feeding on human refuse. Appl. Environ. Microbiol. 49:791 794. 38. Samaha-Kfoury, J. N., and G. F. Araj. 2003. Recent development in beta lactamases and extended spectrum beta lactamases. BMJ 327:1209 1213. 39. Sarmah, A. K., M. T. Meyer, and A. B. Boxall. 2006. A global perspective on the use, sales, exposure pathways, occurrence, fate and effects of veterinary antibiotics (VAs) in the environment. Chemosphere 65:725 759. 40. Sayah, R. S., J. B. Kaneene, Y. Johnson, and R. Miller. 2005. Patterns of antimicrobial resistance observed in Escherichia coli obtained from domestic and wild-animal fecal samples, human septage, and surface water. Appl. Environ. Microbiol. 71:1394 1404. 41. Shaw, K. J., P. N. Rather, R. S. Hare, and G. H. Miller. 1993. Molecular tics of aminoglycosides resistance s and familial relationships of the aminoglycoside-modifying enzymes. Microbiol. Rev. 57:138 163. 42. Silva, V. L., J. R. Nicoli, and T. C. Nascimento. 2009. Diarrheagenic Escherichia coli strains recovered from urban pigeons (Columba livia) in Brazil and their antimicrobial susceptibility patterns. Curr. Microbiol. 59:302 308. 43. Singer, R. S., M. P. Ward, and G. Maldonado. 2006. Can landscape ecology untangle the complexity of antibiotic resistance? Nat. Rev. Microbiol. 4:943 952. 44. Skurnik, D., et al. 2008. Characteristics of human intestinal Escherichia coli with changing environments. Environ. Microbiol. 10:2132 2137.

888 ALLEN ET AL. APPL. ENVIRON. MICROBIOL. 45. Skurnik, D., et al. 2006. Effect of human vicinity on antimicrobial resistance and integrons in animal faecal Escherichia coli. J. Antimicrob. Chemother. 57:1215 1219. 46. Travis, R. M., et al. 2006. Chloramphenicol and kanamycin resistance among porcine Escherichia coli in Ontario. J. Antimicrob. Chemother. 58:173 177. 47. Varga, C., et al. 2008. Comparison of antimicrobial resistance in ric Escherichia coli and Salmonella spp. cultured from identical fecal samples in finishing swine. Can. J. Vet. Res. 72:181 187. 48. Veterinary Drug Directorate Health Canada. 2009. Categorization of antimicrobial drugs based on their importance in human medicine. Government of Canada, Ottawa, Ontario, Canada. http://www.hc-sc.gc.ca/dhp -mps/consultation/vet/consultations/amr_ram_hum-med-rev-eng.php. 49. Waksman, S. 1961. The role of antibiotics in nature. Perspect. Biol. Med. 4:271 287. 50. White, D. G., et al. 2000. Characterization of chloramphenicol and florfenicol resistance in Escherichia coli associated with bovine diarrhea. J. Clin. Microbiol. 38:4593 4598. 51. Witte, W. 1998. Medical consequences of antibiotic use in agriculture. Science 279:996 997.