The study site is a swine finishing CAFO located in the Mid-Atlantic United States. The CAFO consists of two tunnel-ventilated swine houses built atop 12-ft deep concrete pits where swine waste is stored before periodic siphoning into a transport truck for off-farm disposal by land application. Each house has the capacity to hold 2,500 hogs; however, during the sampling period approximately 1,500 hogs were being housed in each building. Air sampling at the swine facility was conducted on 9 December 2003 and 5 January 2004.

Air samples were collected at a calibrated flow rate of 12.5 L/min using all-glass impingers (AGI-30; Ace Glass, Vineland, NJ) designed to collect respirable particles, including bioaerosols, with an aerodynamic diameter < 5 μm. Impingers were autoclaved and filled with 20 mL phosphate-buffered saline (PBS) before sampling. On the first sampling day, sampling was conducted over a 30-min period. On the second sampling day, sampling was conducted for 60 min in order to increase yield. For the longer sampling period, the impinger solution was replenished with distilled deionized H₂O to maintain the sampler collection efficiency (Lin et al. 1997) and avoid increasing liquid salinity. All sampling equipment was placed on top of a table (1.5 m from the ground) within an empty swine stall located within the facility approximately 30 m from the south wall of the swine facility where air exits through ventilation fans. At the time of sampling, four of eight 32-inch ventilation fans were in operation to maintain a farm-operator–designated target temperature of 21°C within the facility. Temperature and relative humidity were monitored throughout the sampling periods and were 22°C ± 1°C and 76 ± 4% respectively. Impingers were stored and transported back to the laboratory at 4°C.

Approximately 3 hr after the last air sample was collected, impinger liquid samples were analyzed in the laboratory. Because no standard method exists regarding the isolation of Enterococcus from air, the standard methods used for the isolation of Enterococcus from recreational water were modified to accommodate the air samples [U.S. Environmental Protection Agency (EPA) 2000]. All broths and agars were obtained from Becton Dickinson (Sparks, MD). Three 10-fold dilutions (using PBS as the diluent) of the impinger samples were plated (100 μL/plate) in duplicate on mE agar. Negative control plates were made by plating 100 μL of both the replenishing fluid that was transported to the site and the dilution liquid. All plates were incubated for 48 hr at 41.5°C under aerobic conditions. All resulting colonies were counted, and counts from dilution plates containing 30–300 CFU were used in back-calculations to determine the concentration of isolated bacteria per cubic meter of air within the swine CAFO. Colonies from sample dilution plates that ranged from pink to red in color (indicative of Enterococcus colonies) were streaked onto Enterococcosel agar and incubated for 24 hr at 41.5°C under aerobic conditions. CFUs characteristic of Enterococcus that formed a black precipitate on the Enterococcosel agar plates were considered presumptive Enterococcus (U.S. EPA 2000). Presumptive Enterococcus isolates were archived in a 20% glycerol, tryptic soy broth solution at –80°C for subsequent speciation and antimicrobial susceptibility testing.

All presumptive Enterococcus isolates, as well as the quality control strains E. faecium 19434 and Enterococcus faecalis 29212 (American Type Culture Collection, Manassas, VA), were streaked from –80°C archived stocks onto both tryptic soy agar and tryptic soy agar No. 2 with 5% defibrinated sheep blood (Quad Five, Ryegate, MT) and incubated for 24 hr at 37°C. All of the media formulations and test interpretations used have been described previously (Murray et al. 2003). Gram stains were prepared on all isolates to verify the presence of gram-positive cocci. Each isolate was tested for the production of catalase in the presence of 3% hydrogen peroxide. Catalase-positive isolates were identified as Staphylococcus species (except for one isolate, which was further tested for oxidase activity and identified as Micrococcus luteus). Each Staphylococcus isolate was inoculated onto 0.5 mL rabbit plasma (Becton Dickinson) to test for the production of coagulase. Catalase-negative isolates were differentiated further by pyrrolidonyl-arylamidase activity using Remel’s PYR kit (Remel, Lenexa, KS). The following biochemical tests were performed on the isolates displaying pyrrolidonyl-arylamidase activity: mannitol, arabinose, sorbitol, raffinose, lactose, and sucrose carbohydrate fermentation tests; arginine deamination; acidification of methyl-α-d-glucopyranoside; pyruvate utilization; and isolate pigmentation.

Antimicrobial susceptibility testing was conducted using the minimal inhibitory concentration (MIC) agar dilution method [National Committee for Clinical Laboratory Standards (NCCLS) 2002]. E. faecalis 29212 was used as the quality control reference strain. Susceptibility to erythromycin, clindamycin, virginiamycin (streptogramin A and B combination), tetracycline, and vancomycin was tested. Erythromycin, clindamycin, tetracycline, and vancomycin were obtained from Sigma (St. Louis, MO). Virginiamycin was obtained from Research Products International Corp. (Mt. Prospect, IL). Concentrations of antibiotics tested ranged from 0.5 μg/mL to 256 μg/mL for erythromycin and tetracycline, 0.03 μg/mL to 128 μg/mL for clindamycin, 0.03 μg/mL to 32 μg/mL for virginiamycin, and 0.03 μg/mL to 64 μg/mL for vancomycin.

In preparation for the agar dilution tests, the air sample isolates, as well as the MIC reference strain E. faecalis 29212, were streaked from –80°C archived stocks onto tryptic soy agar No. 2 with 5% defibrinated sheep blood (QuadFive, Ryegate, MT) and incubated for 24 hr at 37°C. After 24 hr, each isolate was suspended in 3 mL Mueller-Hinton broth with a sterile cotton swab and adjusted to a 0.5 McFarland standard using a Vitek colorimeter (Hach, Loveland, CO). Two hundred microliters of each suspension was transferred to a well within a Cathra replicator plate (Oxoid Inc., Ogdensburg, NY) and replicated with 1-mm pins in accordance with NCCLS guidelines onto Mueller-Hinton agar plates that were previously prepared with the appropriate concentrations of antibiotics (NCCLS 2002). Plates were incubated for 24 hr at 37°C under aerobic conditions. After 24 hr, the plates were read manually and MICs were determined. Specifically, the MIC was recorded as the minimum antibiotic concentration that completely inhibited bacterial growth. According to the MIC, isolates were categorized as susceptible, intermediate, or resistant to each antibiotic using the following MIC breakpoints established by the NCCLS for Enterococcus: erythromycin, susceptible ≤ 0.5 μg/mL, intermediate 1–4 μg/mL, and resistant ≥ 8 μg/mL; clindamycin, susceptible ≤ 0.5 μg/mL, intermediate 1–2 μg/mL, and resistant ≥ 4 μg/mL; virginiamycin, susceptible ≤ 1 μg/mL, intermediate 2 μg/mL, and resistant ≥ 4 μg/mL; tetracycline, susceptible ≤ 4 μg/mL, intermediate 8 μg/mL, and resistant ≥ 16 μg/mL; and vancomycin, susceptible ≤ 4 μg/mL, intermediate 8–16 μg/mL, and resistant ≥ 32 μg/mL (NCCLS 2002).

The mean concentration of presumptive Enterococcus present in the air of the swine CAFO on both 9 December 2003 and 5 January 2004 was 4 × 10⁴ CFU/m³. After bacterial speciation was completed on 137 presumptive Enterococcus isolates, only 47 out of 137 isolates (34%) were confirmed to be Enterococcus (Table 1). Forty-four isolates (32%) were identified as staphylococci, 45 isolates (33%) were viridans group streptococci, and 1 isolate was identified as Micrococcus luteus (Table 1).

Ninety-eight percent (121 of 124) of the bacterial isolates that grew successfully during the antimicrobial susceptibility tests were resistant to high levels of at least two antibiotics commonly used in swine production (erythromycin, clindamycin, virginiamycin, or tetracycline), and 93% of the isolates (115 of 124) were resistant to three antibiotics commonly used in swine production. Individually, 98% of the isolates were resistant to erythromycin, 94% were resistant to clindamycin, 90% were resistant to tetracycline, and 37% were resistant to virginiamycin. None of the isolates displayed resistance to vancomycin. Because none of the E. avium, E. pseudoavium, or E. raffinosus isolates (all belonging to the Enterococcus physiologic group I) grew successfully on the control or antibiotic-amended MIC plates after being suspended as 0.5 McFarland standard solutions, MIC data for these isolates were not determined. MIC distributions among all other isolates were similar for erythromycin, clindamycin, tetracycline, and vancomycin, regardless of bacterial genus or species (Tables 2 and 3). For instance, across all organisms, most isolates (96%) had MICs > 256 μg/mL for erythromycin (Tables 2 and 3). In contrast, resistance to virginiamycin was more prevalent among coagulase-negative staphylococci versus Enterococcus or Streptococcus isolates (Tables 2 and 3). Phenotypes of antibiotic resistance among the bacterial isolates appear in Table 4.