Conditions Necessary for the Transfer of Antimicrobial Resistance in Poultry Litter

Animal manures contain a large and diverse reservoir of antimicrobial resistance (AMR) genes that could potentially spillover into the general population through transfer of AMR to antibiotic-susceptible pathogens. The ability of poultry litter microbiota to transmit AMR was examined in this study. Abundance of phenotypic AMR was assessed for litter microbiota to the antibiotics: ampicillin (Ap; 25 μg/mL), chloramphenicol (Cm; 25 μg/mL), streptomycin (Sm; 100 μg/mL), and tetracycline (Tc; 25 μg/mL). qPCR was used to estimate gene load of streptomycin-resistance and sulfonamide-resistance genes aadA1 and sul1, respectively, in the poultry litter community. AMR gene load was determined relative to total bacterial abundance using 16S rRNA qPCR. Poultry litter contained 108 CFU/g, with Gram-negative enterics representing a minor population (<104 CFU/g). There was high abundance of resistance to Sm (106 to 107 CFU/g) and Tc (106 to 107 CFU/g) and a sizeable antimicrobial-resistance gene load in regards to gene copies per bacterial genome (aadA1: 0.0001–0.0060 and sul1: 0.0355–0.2455). While plasmid transfer was observed from Escherichia coli R100, as an F-plasmid donor control, to the Salmonella recipient in vitro, no AMR Salmonella were detected in a poultry litter microcosm with the inclusion of E. coli R100. Confirmatory experiments showed that isolated poultry litter bacteria were not interfering with plasmid transfer in filter matings. As no R100 transfer was observed at 25 °C, conjugative plasmid pRSA was chosen for its high plasmid transfer frequency (10−4 to 10−5) at 25 °C. While E. coli strain background influenced the persistence of pRSA in poultry litter, no plasmid transfer to Salmonella was ever observed. Although poultry litter microbiota contains a significant AMR gene load, potential to transmit resistance is low under conditions commonly used to assess plasmid conjugation.


Introduction
The United States produces~20 million tons of poultry manure each year [1,2]. Birds are raised on wood shavings and other plant products as bedding, on which the animals defecate. With time, this bedding material is broken down by microbial activity. The material, referred to as poultry litter, is a major by-product generated in poultry meat production and is highly valued as fertilizer. Animal manures, including poultry litter, are often used as soil amendments on organic produce farms [3]. However, application of animal manures has public health risks, as they harbor zoonotic pathogens [4][5][6]. Over the past 40 years, there has been a significant increase in foodborne outbreaks associated with the consumption of produce [7][8][9] and there have been several outbreaks tied directly to the application of animal manures to fields [10][11][12].
Animal manures also contain a diverse and abundant antimicrobial resistome [13]. Poultry litter is a complex microbial community consisting of more than just fecal bacteria, bacteria from the experimental litter and <2.99 Log 10 for commercial sources. Similarly, ampicillin resistance also varied between the two litter sources. The poultry litter sources also differed in the level of Gram-negative enterics (<3 vs. 4 Log 10 CFU/g). Because of between-group (farms) and within-group variability, as determined by ANOVA, litter samples were subsequently pooled to assess transferability of AMR from litter to a recipient Salmonella strain. With the high level of resistance genes and phenotypes, we expected facile transfer of antimicrobial resistance (AMR) from poultry litter bacteria to a susceptible bacterial strain. However, we did not detect transfer of antimicrobial resistance between the poultry litter microbiota and an antimicrobial-sensitive Salmonella Typhimurium strain, regardless of the abundance of Gram-negative enterics present in the litter community in filter matings ( Table 2). We hypothesized that the AMR transfer may occur in specific conditions optimal for conjugative plasmids present in the litter community. Therefore, we tested multiple conditions that varied media strength, growth temperature, and length of contact time in filter matings in an attempt to identify conditions enabling AMR transfer from litter bacteria to Salmonella. (2) built-up litter of commercial broiler chicken farms (n = 4). b Colony counts on tryptic soy agar, grown overnight at 37 • C. c Total colony counts on MacConkey agar incubated overnight at 37 • C. d Cm-chloramphenicol (25 µg/mL); Ap-ampicillin (25 µg/mL); Sm-streptomycin resistance (100 µg/mL); and Tc-tetracycline (25 µg/mL). e Copy # determined from standardized curve using Escherichia coli MC4100 containing plasmid R100 as the standard. DNA concentration was normalized to 30 ng in qPCR. AMR gene abundance was presented as Log 10 ratio of AMR gene copies to 4.2 16S genes per bacterial genome. [43]. f Limit of detection. g BG-bacterial genomes. h ANOVA p > 0.05. i ANOVA p < 0.05. j ANOVA p < 0.01. k Comparison of poultry litter from commercial vs. experimental sources. recipient. In filter matings, the earliest detection of R100 plasmid transfer to Salmonella was six hours at 37 • C ( Figure 1). Peak plasmid transfer was observed for overnight incubation at 37 • C, where the conjugation frequency was 0.48. Therefore, all subsequent filter conjugation experiments were performed with an incubation of at least 24 h. Growth temperature (25 • C vs. 37 • C) and length of filter incubation (24 or 72 h) were then varied to identify parameters that may result in the highest poultry litter AMR transfer rate using E. coli with R100 as a donor control (Table 3; Figure 2). AMR transfer was observed using the E. coli plasmid-donor control in filter matings at 37 • C for a minimum of 24 h. No plasmid transfer of the R100 donor control was observed at 25 • C incubation for up to 72 h ( Figure 2). Another potential factor in plasmid transfer is media composition and richness where bacterial growth rate influences gene transfer [44][45][46]. To address this possibility, tryptic soy broth and agar were diluted 10-, 100-, and 1000-fold and used in filter matings at 37 • C ( Figure 3). Transconjugant abundance correlated with media strength ( Figure 3A) and ultimately recipient abundance ( Figure 3B). Dilute media (100-to 1000-fold) only reduced conjugation frequency of E. coli R100 by~10-fold. Therefore, subsequent experiments used a variety of conditions to detect litter AMR transfer to Salmonella.
2.2. What conditions are optimal for transfer of AMR from poultry litter resistome to Salmonella?
Escherichia coli containing conjugative plasmid R100-1 served as a plasmid donor control in experiments to optimize AMR transfer to the Salmonella Typhimurium LT2 pSLTrecipient. In filter matings, the earliest detection of R100 plasmid transfer to Salmonella was six hours at 37 °C ( Figure 1). Peak plasmid transfer was observed for overnight incubation at 37 °C, where the conjugation frequency was 0.48. Therefore, all subsequent filter conjugation experiments were performed with an incubation of at least 24 h. Growth temperature (25 °C vs. 37 °C) and length of filter incubation (24 or 72 h) were then varied to identify parameters that may result in the highest poultry litter AMR transfer rate using E. coli with R100 as a donor control (Table 3; Figure 2). AMR transfer was observed using the E. coli plasmid-donor control in filter matings at 37 °C for a minimum of 24 h. No plasmid transfer of the R100 donor control was observed at 25 °C incubation for up to 72 h ( Figure 2). Another potential factor in plasmid transfer is media composition and richness where bacterial growth rate influences gene transfer [44][45][46]. To address this possibility, tryptic soy broth and agar were diluted 10-, 100-, and 1000-fold and used in filter matings at 37 °C ( Figure 3). Transconjugant abundance correlated with media strength ( Figure 3A) and ultimately recipient abundance ( Figure 3B). Dilute media (100-to 1000-fold) only reduced conjugation frequency of E. coli R100 by ~10-fold. Therefore, subsequent experiments used a variety of conditions to detect litter AMR transfer to Salmonella. Optimal plasmid transfer incubation time for incFII plasmid R100-1 at 37 °C in filter matings with rifampicin-resistant Salmonella as a recipient. Filter matings were performed between the plasmid donor, nalidixic-acid-resistant E. coli MC4100 containing R100, and the Salmonella recipient S. Typhimurium LT2 pSLTat a 1:10 ratio. Salmonella recipient and transconjugants were Figure 1. Optimal plasmid transfer incubation time for incFII plasmid R100-1 at 37 • C in filter matings with rifampicin-resistant Salmonella as a recipient. Filter matings were performed between the plasmid donor, nalidixic-acid-resistant E. coli MC4100 containing R100, and the Salmonella recipient S. Typhimurium LT2 pSLT − at a 1:10 ratio. Salmonella recipient and transconjugants were enumerated by plating 10-fold dilutions onto rifampicin (64 µg/mL) alone or with chloramphenicol (25 µg/mL). Conjugation frequency was calculated as #transconjugants/#recipients. Salmonella LT2 + E. coli pR100 0.00 0.00 0.00 1.43 × 10 −1 Salmonella LT2 + Litter 1 b 0.00 0.00 0.00 0.00 a #transconjugants/#recipients. Recipients and transconjugants were enumerated by plating 10-fold dilutions onto media with rifampicin (64 µg/mL) alone or combination chloramphenicol (25 µg/mL) and rifampicin (64 µg/mL), respectively. b Poultry litter collected from 3rd successive research chicken flock, raised on built-up litter.

Poultry Litter Bacteria Are Not Inhibiting Plasmid Transfer between Escherichia coli and Salmonella
We hypothesized that failure to observe AMR transfer from litter bacteria to Salmonella might be due to interference. Conjugation requires physical cell-to-cell contact for plasmid transfer to occur. Salmonella is often present at low levels in poultry litter [47]. If litter bacteria that possess conjugative AMR plasmids and antibiotic-susceptible to Salmonella at 37 • C. Tryptic soy agar (TSA) was used as per manufacturer's recommendations (1×) or diluted 10-(0.1×), 100-(0.01×), or 1000-fold (0.001×); to this, MgSO 4 (10 mM) and agar (1.5%) was added. Rifampicin-resistant S. Typhimurium LT2 pSLT − and nalidixic-acid-resistant E. coli containing R100 or poultry litter bacteria, in a 1:10 recipient to donor ratio, were filtered onto 0.45 µm filers. Poultry litter bacteria were isolated from litter collected from the 3rd successive research chicken flock raised on built-up poultry litter. Conjugation frequency of R100 to Salmonella was calculated as #transconjugants/#recipients. Recipients and transconjugants were enumerated by plating 10-fold dilutions onto media with rifampicin (64 µg/mL) alone or with ampicillin (25 µg/mL), chloramphenicol (25 µg/mL), streptomycin (100 µg/mL), or tetracycline (25 µg/mL). No transconjugants were observed for any of the antibiotic combinations in filter matings with Salmonella and poultry litter bacteria. (B) Linear correlation between transconjugants and recipient abundance in filter matings.
No transfer of plasmids from poultry litter bacteria was observed in these conditions. However, plasmid transfer was observed when the E. coli R100 plasmid donor was included with the poultry litter bacteria in filter matings at 37 • C ( Table 2). Salmonella abundance and R100 conjugation frequency were reduced~2 log 10 using commercial poultry litter compared to litter bacteria from the research flock (p < 0.01). No ampicillin resistance was observed in transconjugants, despite~5% of the commercial litter bacteria being resistant; furthermore, only resistance phenotypes linked to R100 were observed, suggesting that no other plasmids were transferred to Salmonella.

Poultry Litter Bacteria Are Not Inhibiting Plasmid Transfer between Escherichia coli and Salmonella
We hypothesized that failure to observe AMR transfer from litter bacteria to Salmonella might be due to interference. Conjugation requires physical cell-to-cell contact for plasmid transfer to occur. Salmonella is often present at low levels in poultry litter [47]. If litter bacteria that possess conjugative AMR plasmids and antibiotic-susceptible recipient cells are both minor populations, the probability of transfer would be low. In order to investigate the role of donor abundance, conjugation frequency was examined using donor E. coli containing R100, starting at 10 6 cell density and diluting 10-fold to extinction. Filter matings were performed with high (10 8 ) or low (10 6 ) levels of poultry litter bacteria relative to recipient concentrations (10 6 vs. 10 8 , respectively) with decreasing abundance of the E. coli donor ( Figure 4). R100 plasmid transfer (conjugation frequency: 6.06 × 10 −6 ) to Salmonella was observed even at levels where the donor and recipient strains were at their lowest starting cell density (10:10 6 ) relative to the total poultry litter bacteria (10 8 ). Transconjugant abundance positively correlated (R 2 = 0.9632) with the initial starting cell density of the plasmid donor. Therefore, while AMR gene abundances are high in poultry litter, these results suggest that the resistance genes may not reside on conjugative, mobile genetic elements capable of transfer to, persistent in, or expressed in Salmonella.

The Contribution of Donor Plasmid and Donor Strain Type to Antibiotic Resistance Transmission in Poultry Litter
The poultry house environment is generally maintained at 75 • F (25 • C) when rearing chickens. Poultry house temperatures are slightly elevated to 30 • C during brooding when hatchlings are first placed in the flock house [48]. In an effort to better mimic the flock house environment, gene transfer studies were performed at 25 • C. However, no R100 plasmid transfer to Salmonella was observed in the poultry litter at 25 • C, despite the persistence of both the donor E. coli and recipient Salmonella in the poultry litter microcosm for 7 days ( Figure 5). While there were between-group differences (donor alone vs. donor and recipient) in E. coli counts on nalidixic acid alone (strain or strain with R100) or with chloramphenicol (donor with plasmid) (ANOVA p < 0.01), no in-group differences were observed. Failure to observe transfer of pR100 was, therefore, not due to its loss but rather related to temperature, as transfer readily occurred at 37 • C, and not 20 • C, in vitro. Therefore, several AMR conjugative plasmids, belonging to different plasmid incompatibility groups, were examined for their ability to transfer resistance at 25 • C. IncI and IncW plasmids were capable of transferring resistance to Salmonella at 25 • C with transfer frequencies of 10 −4 to 10 −9 (Table 4). Conjugative plasmid pRSA was chosen for further study due to derepression of the tra operon and broad-host range [49,50]. Because the E. coli donor strain MC4100 contains several mutations including recA1, thi01, and relA1 that may place it at a metabolic disadvantage in poultry litter, we substituted it for E. coli 1932, a wild-type prototroph. Escherichia coli 1932 containing pRSA was mixed with Salmonella in a poultry litter microcosm at 25 • C. However, the E. coli 1932 strain did not persist in the litter past 3 days, while the Salmonella recipient strain persisted up to 7 days ( Figure 6A). No transconjugants were observed using donor strain 1932 and plasmid pRSA in poultry litter. E. coli 1932 is a human isolate and apparently less able to adapt to the litter environment than poultry isolates. Plasmid pRSA was then moved into the nalidixicresistant chicken E. coli isolate, 5651 [51]. Repeating the previous experiment, E. coli 5651 containing pRSA was used as the plasmid donor. This E. coli strain persisted longer than the human isolate; however, it was not detected on day 7 ( Figure 6B). There were no significant differences in E. coli counts on media containing nalidixic acid or kanamycin (Student's t-test p > 0.05). No transconjugants were observed over the 7-day mating period in the poultry litter microcosm.

Discussion
There was a sizable poultry litter resistome, especially with regard to phenotypic resistances to streptomycin and tetracycline and a corresponding abundance of aadA1 (streptomycin resistance) and sul1 (sulfonamide resistance). Poultry litter has been reported to contain an abundant and diverse resistome [18,53,54]. Many AMR genes are also linked to mobile genetic elements (MGE) [53,55]; however, integrons and some transposons are not self-transmissible and depend on conjugative plasmids, phages or coli poultry isolate 5651 served as plasmid pRSA donor. Black, closed circle with solid line: Salmonella LT2 pST − ; gray, closed circle with solid line: nalidixic-acid-resistant E. coli; and gray, closed circle with dashed line: kanamycin-resistant E. coli (pRSA). There were no significant differences (Student's t-test, p > 0.05) in E. coli counts on nalidixic acid or kanamycin. No transconjugants were detected. As low to no E. coli were detected in poultry litter at 7 days, the experiment was terminated after this time point. Poultry litter was obtained from four commercial broiler farms and pooled.

Discussion
There was a sizable poultry litter resistome, especially with regard to phenotypic resistances to streptomycin and tetracycline and a corresponding abundance of aadA1 (streptomycin resistance) and sul1 (sulfonamide resistance). Poultry litter has been reported to contain an abundant and diverse resistome [18,53,54]. Many AMR genes are also linked to mobile genetic elements (MGE) [53,55]; however, integrons and some transposons are not self-transmissible and depend on conjugative plasmids, phages or natural transformation for dissemination. In this study, we did not detect transfer of antimicrobial resistance between the poultry litter bacteria and an antimicrobial-sensitive S. Typhimurium recipient strain, despite the abundance of Gram-negative enterics present in the litter community. Numerous AMR genes and associated MGE are shared among disparate bacterial members of the litter microbiota, including Salmonella [18]. Comparative genome analyses of various multi-drug-resistance (MDR) plasmids bare out a common origin amongst different bacterial species harboring the same or similar conjugative MDR plasmids [38,56,57].
Because we did not detect AMR gene transfer to Salmonella, we investigated the role of donor strain, plasmid type, and environmental conditions on transfer rates and donor strain persistence. The physiological state, of donor and recipient, is an important parameter to plasmid transfer [44][45][46]. Shafieifini et al. reported higher conjugation frequency with lower strength Mueller-Hinton broth, indicating that slower growth rates may enhance transfer [45]. Salmonella virulence plasmid belonging to the incompatibility group IncFII was shown to be transferable only when the plasmid donor was grown in a minimal medium [58]. Fernandez-Astorga et al. reported a decline in transconjugants and conjugation frequency with diluted media [59]. These differences may reflect plasmid or donor/recipient strain type. However, varying temperature and media concentration did not result in increased AMR transfer.
Some conjugative plasmids respond to mate-sensing signal peptides produced by potential recipients. In this system, the donor also produces an antagonist to the signal, inhibiting plasmid transfer at high donor-to-recipient ratios [60]. At low cell densities, plasmid transfer occurs at a high rate until the donor and transconjugant population density produces enough antagonists to inhibit conjugation. Others have also found that the quorum-sensing autoinducer acyl homoserine lactone can increase plasmid transfer in a dose-dependent manner [61]. Autoinducers can affect bacterial motility [62,63]. Autoinducer AI-2 can even act as a chemoattractant [62] and induce biofilm formation [64], where plasmid transfer can occur [46]. Quorum-sensing has also been shown to contribute to plasmid transfer in microbial communities [65]. For these reasons, we also varied E. coli and litter bacteria donor concentrations in filter matings and used the litter microcosm itself in order to provide signaling molecules for the matings. Although these conditions did not result in plasmid transfer in our experiments, plasmid transfer to Gram-negative enterics has been well documented in vivo [42,[66][67][68][69][70][71] and ex vivo [72], including studies demonstrating acquisition of resistance from the resident microbiota [42,67,72]. The inability to document plasmid transfer from the poultry litter microbiome to Salmonella may be attributed to (1) the absence of AMR conjugative plasmids in the bacterial population [73]; (2) limited transfer potential of conjugative plasmids or plasmid replication outside their donor host, kin, or evolutionarily related bacteria [37,68,[74][75][76]; or (4) genetic barriers to plasmid acquisition [77], including exclusion by resident plasmids [78]. However, we did not detect plasmid transfer in the poultry litter microcosm even when an E. coli plasmid donor was included. In order to determine if failure of plasmid transfer was due to plasmid type, we utilized a number of conjugative plasmids of different incompatibility groups. IncF plasmid R100 transfer is optimal at 37 • C, and no transfer was ever detected at room temperature. IncI plasmids are commonly found in Salmonella from non-food animal sources and contain the full conjugation machinery for plasmid transmission [78]. The IncW plasmid pRSA was chosen based on its use in comparable studies [49,79] and its ability to transfer at 25 • C. However, transfer was not observed when E. coli with conjugative plasmid pRSA was used in poultry litter. In this case, E. coli strain type became a confounding factor, as E.
coli 1932 died off after 3 days. While switching to a poultry isolate, E. coli 1932, resulted in improved persistence, no plasmid transfer was detected. Others have also reported the influence of donor strain and plasmid type on AMR transfer [68,69,74,80,81]. While the focus here is on donor and plasmid, recipient background can significantly impact plasmid acquisition [74,78]; therefore, we selected Salmonella LT2 pST − which has been cured of the resident F plasmid that can act as barrier to plasmid transfer.
Despite our earlier findings of a low rate of plasmid transfer to Salmonella in vivo [42], the rate of acquisition of AMR from the litter microbiota was too low to detect in these experiments. This is probably most likely because the poultry litter used in this study did not support or inhibit Salmonella or E. coli growth. Alternatively, the poultry litter microbiota may produce factors that interfere with plasmid abundance or transfer [73]. Bacterial growth is central to AMR transfer [45], as energy is needed for plasmid transmission as well as DNA synthesis machinery needed to produce the complementary strand of the transferred plasmid DNA [82]. Because no increase in donor or recipient cell density was ever observed in poultry litter even at low cell densities, transfer may be a rare event. However, even in an E. coli strain that persisted in poultry litter, the plasmid was lost after 3 days of incubation in the microcosm. We previously reported similar findings for S. Newport AMR plasmid in chickens administered E. coli donors with antibiotic selection pressure [42]. Others have reported the impact of donor strain background on plasmid transfer [69,74,80,81]. Plasmid transfer has been previously reported in poultry bedding material where conjugation frequencies varied depending on bedding or presence of inhibitory chemical residues [83,84]. Guan et al. reported plasmid transfer but with a higher rate in chicken manure than compost where donor and recipient cell abundance was affected by compost temperature [85]. Tecon et al. reported that the frequency of cell-cell contact that led to plasmid transfer was a function of water activity; the drier the cell matrix, the more likely plasmid transfer occurs [86]. Therefore, the physical nature of the litter matrix may also impact on plasmid transfer indicating that litter management may have a large impact of the likelihood of resistance transfer to foodborne pathogens such as Salmonella. Plasmid transfer in litter requires (1) permissive microbiota, bacteria strain, and plasmid types; (2) bacteria growth; (3) water activity that favors biofilm formation; and antibiotic selection pressure that favors emergent AMR in pathogens inhabiting this environment. Future studies should reveal whether certain litter types or management practices such as top dressing, litter amendments, or deep litter systems are more likely to promote AMR dissemination.

Extraction of Bacteria from Litter
Bacteria were extracted from litter and separated from detritus using spin columns as previously described [87]. For each sample, one gram of litter was weighed out and placed in a 50 mL conical tube with 10 mL of 0.1% Tween 80 in sodium phosphate buffer (3.4 mM NaH 2 PO 4 , 46.6 mM Na 2 HPO 4 ; pH 7.4). The tube was then placed in the arm of a Fisherbrand™ "wrist-action" Flask Shaker (Fisher; Waltham, MA, USA) and processed at maximum speed for five minutes. Spin columns were constructed by removing the plunger and flaps from a 20 mL syringe then placing it within a 50 mL conical tube. Sterile 4 × 4-inch gauze pads were placed into the 20 mL syringe to form the column matrix. The columns were pre-wet with 10 mL 0.1% Tween 80 in 50 mM sodium phosphate buffer (pH 7.4). The litter samples were then poured into the spin columns which were centrifuged at 700× g for 1 min. The spin column filtrate was centrifuged a second time at 2000× g for 30 min at 4 • C to pellet the bacteria. The supernatant was discarded and the pellet was resuspended in 1 mL freezer stock medium (15% glycerol, 1% peptone). The sample was equally split into two sterile 1.5 mL capacity microfuge tubes and cells were pelleted by centrifugation at 7500× g for 15 min. The supernatant was discarded from each tube. One pellet was stored at −20 • C until DNA extractions. The second pellet was resuspended in 750 µL freezer stock media and transferred to 1.5 mL cryovials stored at −80 • C.
Poultry litter used in this study was obtained from two sources: experimental broiler flocks and commercial broiler chicken houses in the southeastern United States. Litter obtained from the experimental flock was collected following the 3rd successive chicken flock raised on built-up litter. Random grabs of poultry litter were collected from vacated experimental pens and pooled. Poultry litter was also obtained from four commercial broiler chicken farms submitted to us by a third party. The chicken breeds were typical Ross or Ross/Cobb hybrids used for the production of meat birds (broilers). Bedding for experimental and commercial flocks was pine shavings. Commercial broiler production in the US typically uses a built-up litter system: windrowing to compost litter before its spread, to which it is top dressed with fresh pine shavings before the placement of the next flock.

In Vitro Conjugation
Bacterial strains and plasmids used in this study are described in Table 5. Escherichia coli M4100 containing incFII conjugative plasmid pR100-1 or poultry litter bacteria served as donors in the initial conjugation experiments. The Salmonella enterica Typhimurium strain pSLT − was chosen as recipient based on high plasmid-transfer frequencies reported for this strain in filter matings with E. coli donors bearing IncFI or IncFII plasmids [78]. Salmonella recipient and E. coli donor strains were grown as standing, overnight cultures at 37 • C in 5 mL tryptic soy broth (TSB; Sigma Aldrich) supplemented with 10 mM MgSO 4 . The mating mix consisted of E. coli MC4100 with R100-1 (5 µL,~10 6 CFU) or poultry litter bacteria (5 µL;~10 6 CFU) with S. Typhimurium pSLT − recipient strain (50 µL;~10 7 CFU) in 5 mL of 10 mM MgSO 4 . Cells were collected on a 0.45 µm pore size cellulose filter membrane (Millipore Sigma; Burlington, MA, USA), which was aseptically placed, cell side up, on M9 agar containing 0.2% glucose [78]. Pure cultures with donor or recipient strain alone were similarly treated and included as controls. After overnight incubation at 37 • C, a cell suspension was made by vortexing the filter in 5 mL of 10 mM MgSO 4 . The cell suspension was diluted 10-fold in buffered saline gelatin (BSG) [88] and plated on TSA plates containing the appropriate antibiotic for enumerating Salmonella alone (rifampicin, 64 µg/mL) or Salmonella transconjugants from matings with E. coli pR100-1 (rifampicin, 64 µg/mL; chloramphenicol, 25 µg/mL) or litter bacteria (rifampicin, 64 µg/mL; ampicillin, chloramphenicol, streptomycin, or tetracycline at concentrations stated in Section 4.2). The conjugation frequency was determined from the number of transconjugants divided by recipients, averaging the results of triplicate matings [78].
The impact of temperature (25 • C vs. 37 • C) and medium strength (1×, 0.1×, 0.01×, 0.001×; TSB and TSA) on transfer of antimicrobial resistance to recipient Salmonella was assessed for poultry litter bacteria or E. coli MC4100 with pR100-1 as donors as follows. Bacteria were grown in regular-strength TSB, supplemented with MgSO 4 , and incubated overnight at 37 • C. Mating mixes were made as previously described. Filters were placed on regular strength or diluted (10-to-1000 fold) TSB with agar (1.5% wt./volume) (Fisher), supplemented with 10 mM MgSO 4 , and incubated overnight at 37 • C. Matings were also performed placing filters on standard TSA with 10 mM MgSO 4 and incubating plates at 25 • C or 37 • C overnight; filter matings were performed at 25 • C for 24 and 72 h. Recipients and transconjugants from matings with E. coli R100 or litter bacteria were enumerated by plating dilutions on TSA with antibiotic combinations and concentrations as stated previously. All dilutions were plated in triplicate. In order to determine the minimum Escherichia pR100 donor cell density with poultry litter bacteria sufficient to observe plasmid transfer, it was assessed as follows. Salmonella recipient and E. coli donor strains were grown overnight in TSB with 10 mM MgSO 4 . Recipient S. Typhimurium pSLT − strain (10 6 or 10 8 CFU/mL) was mixed with E. coli R100-1 plasmid donor that had been diluted 10-fold to cell densities ranging from 10 7 -10 0 CFU per ml in 5 mL 10 mM MgSO 4 with litter bacteria. The poultry litter bacteria were coincubated with S. Typhimurium pSLT-recipient at 10:1 (10 8 to 10 6 CFU/mL) or 1:10 (10 6 to 10 8 CFU/mL) ratios of poultry litter bacteria to Salmonella and E. coli MC4100 with pR100-1 (10 6 -10 0 CFU/mL). Cells were collected on 0.45 µm filters; filters were aseptically transferred to TSA; and plates were incubated overnight at 37 • C. The mating mix was plated on TSA with rifampicin (64 µg/mL) or rifampicin plus chloramphenicol (25 µg/mL) to enumerate recipients and transconjugants, respectively. All dilutions were plated in triplicate.

DNA Extraction and qPCR
DNA was extracted from 10 8 bacteria cells using the ZymoBIOMICS DNA Miniprep Kit (Zymo Research, Irvine, CA, USA) according to the manufacturer's instructions. The length of bead-beating was optimized for DNA extraction from poultry litter bacteria by comparing incremental vortex time (0, 5, 10, 20, 30, or 40 min) with visual assessment of the decrease in bacterial cell density monitored by Gram stain and microscopy. DNA was quantified using a NanoDrop One (Thermo Fisher Scientific; Waltham, MA, USA). There was significant bacterial lysis only after 10 min of vortexing and DNA concentration plateaued at this time point as compared to the later time points. Ten minutes was therefore chosen as the optimal length for bead-beating using this kit. Aliquots of DNA were made in fresh tubes and were normalized to 10 ng/µL using molecular-grade water. DNA samples were stored at −20 • C. To estimate total bacterial genomes, 16S rRNA qPCR was used, and qPCR was used to quantify streptomycin-resistance, sulfonamide-resistance, and class 1 integron-integrase genes aadA, sul1, and intI1, respectively, in the poultry litter community ( Table 2). qPCR was performed in triplicate. A standard curve was generated using pDU202 for aadA1 and sul1 and E. coli LE392 for 16S rRNA, starting with 10 ng DNA template and a series of ten-fold dilutions to 1 picogram. Every PCR experiment included a tube without DNA template (no-template) in order to detect reagent or PCR contamination. Escherichia coli strain LE392 served as an additional negative control for the aadA1 and sul1 qPCR and, like the no-template control, was never positive. qPCR mixtures contained 5 µL of iQ SYBR Green Supermix (Bio-Rad; Hercules, CA, USA), 0.5 µM of forward and reverse primers, 1 µL of template DNA, and molecular-grade water for a final volume of 10 µL. The following thermocycler (Bio-Rad CFX96 Real Time System; C1000 Touch TM thermocycler) conditions were used: 3 min at 94 • C and 30 cycles of 30 s at 94 • C, 30 s at annealing temperatures stated in Table 6, and 2 min at 72 • C. This was followed by a melt curve with a temperature range of 55 to 95 • C and 0.5 • C increments. The peak melt curve for all positive litter samples overlapped with the positive controls for aadA1 and sul1. The melt curve for the 16S rRNA qPCR produced a distinct, non-overlapping peak for the litter samples compared to the E. coli LE392 control. The peaks for all the litter samples overlapped. This result was not unexpected as Gram-positive Firmicutes and Actinobacteria are the dominant phyla in poultry litter [14].

Poultry Litter Microcosm
Frozen poultry litter from pooled field samples were placed in Whirl-PAK bags (Nasco; Fort Atkinson, WI, USA) and pulverized with a Seward Stomacher ® 400 Circulator stomacher (Seward Ltd.; Norfolk, UK) (10× g for 5 min). Fifteen grams was subsequently placed in a 50 mL conical tube. Each tube received 12 mL of 10 mM MgSO 4 containing E. coli plasmid donor (10 7 or 10 4 CFU/mL), S. Typhimurium pSLT − recipient (10 7 or 10 6 CFU/mL), donor and recipient, or no bacteria added. Each tube was then vortexed for 1 min to ensure even distribution liquid into poultry litter. Samples were then incubated for a total of 7 or 14 days with 1 g sampling, performed in triplicate for each sample at days 0, 1, 3, 7, and 14. Bacterial extractions were preformed from each sample as previously described. The cell suspensions were diluted 10-fold and plated on TSA plates containing chloramphenicol (pR100-1; 25 µg/mL) or kanamycin (pRSA; 50 µg/mL) alone, rifampicin alone (64 µg/mL), or combination chloramphenicol (pR100-1) or kanamycin (pRSA) and rifampicin for enumerating donor, recipient, and transconjugants, respectively. All dilutions were plated in triplicate.

Statistical Analysis
Analysis of Variance (ANOVA) was used to determine differences between and within groups. Student's t-test was also used to determine significant differences between groups. Linear regression was used to determine correlation between transconjugant abundance and recipient, donor abundance, or media strength.

Conclusions
Transmittance of AMR within any microbial community is dependent on several factors. Key to this is that the resistance needs to reside on conjugative, stable MGE with a broad host range and a competent donor host capable of persisting in this environment. However, conjugation is just one mechanism by which AMR spreads among microbial communities. Phages and natural transformation also play an important role in the spread of AMR [92,93].
The reality is distribution of antimicrobial resistances and their associated resistance genes is uneven across the diverse microbial species that inhabit poultry litter [94]. While specific tetracycline resistance alleles such as tetM are found among phylogenetically diverse phyla in animal manures [95,96], other alleles have co-evolved with their bacterial host: tetQ in the Bacteroidia [95] or the tetracycline efflux pumps and associated tet alleles tetA, tetB, or tetC in the Enterobacteriaceae [96]. AMR do find their way into pathogens such as S. enterica, often on conjugative plasmids that vary in host range from the narrow incF to broad-host-range incQ conjugative plasmids [97]. Enteropathogens, such as Salmonella, do acquire AMR from community bacteria whether it is environmental or gut microbiota. The challenge is to prevent emergent resistance or spread of AMR pathogens in the environment.
The key might be litter management for pathogen-exclusion properties [98,99] or processes, such as composting, that reduce pathogens in poultry litter [100,101].