Prevalence of multidrug resistance in Pseudomonas spp. isolated from wild bird feces in an urban aquatic environment

Abstract Antimicrobial resistance (AMR) has been detected in the microbiota of wildlife, yet little is known about the origin and impact within the ecosystem. Due to the shortage of nonepizootic surveillance, there is limited understanding of the natural prevalence and circulation of AMR bacteria in the wild animal population, including avian species. In this surveillance study, feces from wild birds in proximity to the River Cam, Cambridge, England, were collected and Pseudomonas spp. were isolated. Of the 115 samples collected, 24 (20.9%; 95% CI, 12.6%‒29.2%) harbored Pseudomonas spp. of which 18 (75%; 95% CI, 58%‒92%) had a multiple antibiotic resistance (MAR) index greater than 0.2. No Pseudomonas spp. isolate in this study was pansusceptible. Resistance was found among the 24 isolates against ciprofloxacin (87.5%; 95% CI, 74.3%‒100%) and cefepime (83.3%; 95% CI, 68.4%‒98.2%), both of which are extensively used to treat opportunistic Pseudomonas spp. infections. The prevalence of Pseudomonas spp. in the wild bird feces sampled during this study is greater than previous, similar studies. Additionally, their multidrug resistance profile provides insight into the potential risk for ecosystem contamination. It further highlights the importance of a One Health approach, including ongoing surveillance efforts that help to develop the understanding of how wildlife, including avifauna, may contribute and disperse AMR across the ecosystem.

Despite AMR being commonly associated with high antibiotic usage, this is generally not true in wildlife; therefore, its presence may be used as an indicator of anthropogenic activities affecting the whole ecosystem. For instance, sewage water treatment facilities, farm manure, and slurry are all important habitats for birds and other animals, but they can become contaminated with AMR bacteria, antibiotics (and/or their metabolites), and other elements that can act as selective drivers of AMR. Although assigning the directionality of this dissemination process is an extremely challenging task, previous work has noted the association between resistance patterns and the physical proximity of wild animals to humans (Laurance et al., 2014;Muehlenbein, 2013;Nadimpalli et al., 2020). Wild birds in particular may act as vectors of AMR bacteria by acquiring them from contaminated environments such as rivers receiving sewage effluent and, subsequently, contaminating other environments such as livestock grazing areas and urban environments through fecal shedding. This process potentially facilitates the dissemination of enteric pathogens of public health concern including, but not limited to, Pseudomonas spp., Salmonella spp., Klebsiella spp., Campylobacter spp., and Staphylococcus aureus (Benskin et al., 2009;Navarro-Gonzalez et al., 2020).
The Pseudomonas genus contains over 60 species of Gramnegative, aerobic, nonspore forming, rod-shaped, and motile organisms. Pseudomonas spp. are also capable of protecting other organisms by sheltering them from unfavorable conditions within biofilm formations (Puga et al., 2018) as well as employing cooperation mechanisms such as quorum sensing (Venturi et al., 2010).
Pseudomonas are a metabolically versatile genus with typically large genome sizes varying from 3 to 7 Mbp (Hesse et al., 2018), known to contain several genetic mobile elements including megaplasmids (Cazares et al., 2020) as well as intrinsically and extrinsically acquired resistance mechanisms (Lister et al., 2009). These biological properties allow Pseudomonas spp. to survive in a multitude of environments, including community reservoirs such as soil and rhizosphere, swimming pools, and other infrastructures within urban settings (Nadimpalli et al., 2020). Carbapenems, cephalosporins, fluoroquinolones, and aminoglycosides are the most frequently used antipseudomonal antibiotics but resistance, including multidrug and pandrug resistance, has been reported in both veterinary and human medicine (Cabassi et al., 2017;Cazares et al., 2020;Haenni et al., 2015Haenni et al., , 2017. The resistance mechanisms used by Pseudomonas spp. are varied and facilitated by their genomic plasticity including multidrug efflux systems, outer membrane protein loss, target mutations, and enzyme production (Cabassi et al., 2017).
One species within the Pseudomonas genus, Pseudomonas aeruginosa, is the causative agent of several diseases ranging from external otitis to fatal pneumonia in a range of hosts including dogs, rabbits, birds, and humans. It is rarely a member of the normal microbial flora in healthy humans or animals (Lister et al., 2009). However, in humans, severe P. aeruginosa infections usually occur in immunocompromised patients and in healthcare settings. In the United Kingdom, between April 2019 and March 2020, P. aeruginosa was the second most frequent nosocomial infection with 24.8% of all reported cases (n = 4,336) leading to death (Public Health England, 2021). Other species in the Pseudomonas genus include prolific plant and aquaculture pathogens (Beaton et al., 2018), food spoilage-associated organisms (Stellato et al., 2017), and useful biocontrol agents against plant pathogens (Gómez-Lama Cabanás et al., 2018;Haas & Défago, 2005;Kuzmanović et al., 2018).
Previous studies exploring the natural prevalence of bacteria associated with birds investigated microorganisms that present a threat to either human or domesticated/production animal health (Haesendonck et al., 2016;Laurance et al., 2014;Navarro-Gonzalez et al., 2020). The majority of these studies provide little information on the nonepizootic prevalence of Pseudomonas spp. in avifauna as they tend to rely on postmortem examinations and/or data collected as a result of a disease outbreak causing high mortality (Gómez, 2006;Vasconcelos et al., 2017;Vidal et al., 2017;Walker et al., 2002).
Ongoing AMR surveillance studies of Pseudomonas spp. in wildlife, including birds, represent an underexplored area that may pose risks to humans, other animals, and the environment (Hernando-Amado et al., 2019;O'Neill, 2016;Pornsukarom & Thakur, 2017). The present study reports the nonepizootic prevalence of Pseudomonas spp.
in feces from wild birds in an urban aquatic setting. Additionally, the isolates' antimicrobial susceptibility was determined in order to estimate potential risks to other elements of the ecosystem, which is the overarching purpose of this research.

| Fecal sampling
A total of 115 bird fecal samples were collected using Amies Plain Transwabs ® (MWE, Wiltshire). Swabs were transported to a containment level 2 laboratory where they were stored at 4°C for up to 24 hr before culturing. The bird fecal swabs were collected over a 24-month period from 6 locations along a 16-km stretch of the River Cam, Cambridge, England, and up to 0.8 km away from the riverbank ( Figure 1). The sampled locations include leisure locations (e.g., nature reserve, country park, rowing club) that are popular with human pursuits such as boating, swimming, and bird-watching. Although swabs were obtained regularly from each location throughout the sampling period (at least twice), samples were collected during all seasons while maintaining a minimum dry weather timespan of 48 hr to avoid rainfallassociated fecal microbiota changes (Shehane et al., 2005). Care was taken to sample freshly defecated specimens from areas where Passeriformes, Columbiformes, Anseriformes, and Charadriiformes were observed to be inhabiting/visiting (Shehane et al., 2005).

| Sample culture and isolation of Pseudomonas spp.
Within 24 hr of collection, the bird fecal swabs were streaked onto Pseudomonas cetrimide agar (Oxoid, Basingstoke) and incubated aerobically at 37°C for 24 hr. Plates with confluent growth were subcultured for isolation to avoid the risk of competitive inhibition. Putative Pseudomonas spp. colonies from the primary isolation media were subcultured, and isolated colonies were subject to Gram staining followed by biochemical testing for cytochrome c oxidase production and lactose fermentation via commonly used microbiological techniques (Public Health England, 2015). At least three passages were done and tested as described to obtain pure cultures.

| Bacterial identification by 16S rRNA gene sequencing
Once pure cultures of bacteria were isolated and established, DNA was extracted using DNeasy Blood & Tissue Kit (Qiagen) as per the manufacturer's instructions. The DNA of individual putative Pseudomonas spp. isolates was used as a template to amplify the region encoding the 16S rRNA gene (1.5 kb) in a thermal cycler (Techne, Loughborough) using previously described primers complementary to the conserved regions of the 5′ and 3′ ends of 16S rRNA gene (Lane et al., 1991). A reaction mixture consisting of 1× MyTaq™ Red Mix (Bioline, London), 1 μl of template DNA (50-100 ng), and 0.5 μM of each primer in a final reaction volume of 25 μl was used.
PCR conditions were as follows: 95°C for 5 min, 34 cycles of 30 s at 95°C, 30 s at 54°C, and 2 min at 72°C were carried out followed by a terminal elongation step at 72°C for 10 min. The amplicons were separated electrophoretically through a 1% agarose gel in 1× TAE buffer and visualized using GelRed ® Nucleic Acid Gel Stain (Biotium, California) under ultraviolet illumination. Gel images were captured using GeneSys (Syngene, Cambridge), and the products were identified by molecular weight comparison with the markers of a 1 kb Plus DNA ladder (Invitrogen, Loughborough). The amplified PCR products were purified using the QIAquick PCR Purification Kit (Qiagen, Germany) and partially sequenced at the Department of Biochemistry, University of Cambridge, UK (https://www.bioc.cam. ac.uk/), using the forward primer (Lane et al., 1991). The sequences' similarity were determined by comparing with the GenBank database using BLAST (Altschul et al., 1990).
To determine the evolutionary relatedness among the Pseudomonas spp. isolates obtained in this study, a phylogenetic analysis was performed, employing MEGA version X (Kumar et al., 2018).
The sequences were compiled and aligned using ClustalW embedded in MEGAX. The phylogenetic tree was constructed using the Neighbor-Joining method (Saitou & Nei, 1987), and evolutionary distances were inferred using the Tamura-Nei model (Tamura & Nei, 1993). The reliability of the tree was evaluated by bootstrap resampling technique with 1,000 bootstrap replications. Salmonella enterica ATCC 13314 (NR 041696) sequence retrieved from GenBank was used as an outgroup.

| Antibiotic susceptibility testing
According to the European Committee on Antimicrobial Susceptibility Testing (EUCAST), antibiotic susceptibility testing was carried out using the disk diffusion method (The European Committee on Antimicrobial Susceptibility Testing, 2019). The tested antibiotics (Oxoid, Basingstoke) were meropenem (10 µg), cefepime (30 µg), gentamicin (30 µg), ciprofloxacin (5 µg), and levofloxacin (5 µg). After aerobic incubation at 37°C for 24 hr, the diameters of the zones of F I G U R E 1 (a) Schematic map of East Anglia as legally defined in the Nomenclature of Territorial Units for Statistics (NUTS) 2. The map illustrates key geographic locations surrounding the sample collection area (delineated by the dotted square); Scale bar, 40 km. (b) Trace map of the River Cam, Cambridge, alongside the geographic coordinates of each sample collection site (expressed as latitude and longitude, in degrees, minutes, and seconds, according to the global positioning system); Scale bar, 2 km inhibition, in millimeters, were measured for each antibiotic and the isolates were classified as resistant or susceptible (The European Committee on Antimicrobial Susceptibility Testing, 2019). Isolates classed as "susceptible, increased exposure" by EUCAST were included in the susceptible category. Multiple antibiotic resistance (MAR) indices were then determined for each isolate by using the formula MAR = a/b, where a represents the number of antibiotics to which the isolate was resistant to and b represents the total number of antibiotics to which the isolate has been tested for susceptibility (Krumperman, 1983).

| Statistical analysis
Statistical analysis was performed using SPSS Statistics 26 (IBM, Chicago). To investigate the differences in the number of isolates collected between the sampled locations, a chi-square test was calculated. A chi-square test was also performed to compare the number of resistant and susceptible isolates, for each antibiotic. In the sample collection locations where Pseudomonas spp. were isolated, a twoway chi-squared test was used to determine whether there was an association between the sample location and the resistance profile of the isolates obtained. Statistical significance was deemed as α = 0.05.

| RE SULTS
Pseudomonas spp. were detected in 24 of the 115 bird fecal samples collected and tested during this study (20.9%; 95% CI, 12.6%-29.2% prevalence). Samples were collected from all six locations but were unevenly distributed across different sites (p < .001) ( Table 1). The distribution of Pseudomonas spp. isolates was also significantly different between sample collection locations (p < .05), with one (location 3) yielding no isolates (Table 1).
Of the 24 Pseudomonas spp. isolates, 12 exhibited a 16S rRNA gene similarity of <98.7% with any Pseudomonas spp. and, thus, were classified only to the genus level. Several isolates (n = 9) were identified as P. koreensis, and one isolate (n = 1) was identified as P. aeruginosa. The remaining isolates belonged to the P. fulva (n = 1) and P. fluorescens (n = 1) species. Based on the phylogenetic analysis, a monophyletic tree was obtained (Figure 2) with the sequences grouped within two major clades, which were distantly separated from the outgroup (Salmonella enterica ATCC 13314). P. aeruginosa formed a single clade, with all the remaining sequences clustered separately with different subclades.
Antibiotic sensitivity screening revealed that the majority of Pseudomonas spp. isolates (75%; 95% CI, 58%-92%; n = 18/24) were resistant to more than one antibiotic (MAR index > 0.2), with MAR indices ranging from 0.2 to 0.6 ( Table 2). No Pseudomonas spp. isolate was susceptible to all the antibiotics tested. The distribution of the AMR findings according to their collection source is presented in Table 1. Isolates with MAR indices ≤ 0.2 may be categorized as being from low risk sources of contamination, while isolates with MAR indices > 0.2, therefore multidrug resistant (MDR), may be high-risk sources (Krumperman, 1983). MDR strains were isolated from all collection areas where Pseudomonas spp. were detected. Excluding location 3, a two-way chi-squared analysis revealed that there was no association between the sample site and the proportion of resistant isolates, with p = .995.

| DISCUSSION
to the potential risk to humans, animals, and the environment To the authors' knowledge, data on the AMR and MDR profiles of Pseudomonas spp. in wild birds, outside outbreak circumstances, are limited, even if the possibility of transmission to and from other species, including humans, is a concern that should warrant more attention.
Pseudomonas spp. and, in particular, P. aeruginosa, pose a serious therapeutic challenge for treatment due to their ability to develop resistance to multiple antimicrobial classes (Lister et al., 2009;Public Health England, 2021), as supported by this present study.
Here, the prevalence of resistance was highest toward ciprofloxacin (n = 21/24; 87.5%; 95% CI, 74.3%-100%), a widely prescribed fluoroquinolone that is part of the WHO's Essential Medicine List (Sharland et al., 2018). Fluoroquinolones have a favorable pharmacokinetic and pharmacodynamic profile that makes them first-line choice in the treatment of several community-acquired and nosocomial infections, including P. aeruginosa. With the widespread use of the drug, resistance has emerged and continues to rapidly escalate (Lister et al., 2009;Magiorakos et al., 2012).
Importantly, this study also reported that the lowest prevalence of resistance was found toward meropenem, a critically important member of the carbapenem antibiotic class. Meropenem is a lastline antipseudomonal that is strictly reserved to human medicine in order to preserve its effectiveness. However, the increasing usage of carbapenems in hospitals worldwide exerts a selective pressure that promotes the emergence of resistant Pseudomonas spp. clones in both clinical and community settings. Since susceptibility to this class of antibiotics is often not assessed or reported in animal studies due to their restricted use, this study has purposefully included a carbapenem to evaluate potential collateral consequences of human antibiotic usage in animal Pseudomonas spp. strains. The presence TA B L E 2 Distribution of zone of inhibition results (in mm) and multiple antibiotic resistance (MAR) indices (calculated as per Krumperman (1983)) for the 24 Pseudomonas spp. isolates obtained from wild bird feces Note: The antibiotics tested by disk diffusion method were meropenem (MEM, 10 μg), cefepime (FEP, 30 μg), gentamicin (CN, 30 μg), ciprofloxacin (CIP, 5 μg), and levofloxacin (LEV, 5 μg). The isolates' title is constituted of two components, the first number represents the sample collection location, and the second represents the temporal order of collection (for instance, strain 1.2 was the second to be collected from sample collection location 1). The cells have been colored red and green to demonstrate the isolates' antibiotic resistance and susceptibility, respectively. a Breakpoints for the tested antibiotics are as follows ( of resistance to meropenem only in a single isolate (of 24 collected in this study), suggests that meropenem-resistant Pseudomonas spp.
strains are not prevalent in this wild bird population. Nonetheless, one must still consider that several of the Pseudomonas spp. isolates were MDR, demonstrating that these bacteria may harbor different resistance mechanisms, posing a potential risk for all types of resistance, including carbapenems (Cazares et al., 2020;Lister et al., 2009).
Identifying these resistance mechanisms is key to strengthening the current understanding of evolutionary selective pressures and the potential AMR circulation and spillover routes  (Table 1); however, this study did not find a statistically significant association (p = .995). Conversely, the prevalence of Pseudomonas spp. was found to differ significantly between the sampled locations (p < .05). This was most likely a result of the varied number of samples collected from each location (Table 1), which introduced a potentially confounding variable to the study.
For instance, as a result of fieldwork limitations and bird behavior, only 6 swabs were collected from location 3 and no Pseudomonas spp. isolates were obtained. Additionally, several more samples were collected from other areas with more vegetation and human activity that encouraged foraging behaviors (e.g., picnicking and food littering). Nonetheless, this is a retrospective observation made by the authors and was not subject of direct analysis in this study.
From a wider, yet relevant, perspective, future work needs to focus not only on collecting/collating microbiological surveillance data from wild birds, but also on analysis, reporting, and dissemination of these data. Importantly, there must be a dedicated effort to devise interventions that are mindful of the entire ecosystem so that they can comprehensively inform the policymaking process (Wellcome Trust et al., 2018;O'Neill, 2016). For instance, as suggested in this paper, some of the drivers of AMR to wildlife may be anthropogenic; thus, reasonable interventions could include efforts to improve antimicrobial stewardship and policies regarding waste management.

| CON CLUS ION
In conclusion, this study provides useful information regarding the natural prevalence of MDR Pseudomonas spp. in wild bird fecal samples collected around the River Cam in Cambridge, England. The results have highlighted that wild birds could act as potential MDR bacterial reservoirs in areas where spillover of AMR to humans, other animals, or the environment could occur. Nevertheless, as described throughout this manuscript, determining the source and directionality of AMR dissemination is challenging which may, to a certain extent, explain why the currently available evidence is limited and/ or contradictory. It is our view that in order to overcome this, future work needs to be multidisciplinary and comprehensive of the whole ecosystem, including the often forgotten wildlife fauna. The findings in this study are significant and highlight the need for a One Health Approach to tackle AMR, including increased, ongoing, nonepizootic surveillance to prevent the spread and minimize potential risks to humans, animals, and the environment.

ACK N OWLED G M ENTS
This research was funded by internal funding from Anglia Ruskin University, which was obtained via a competitive process. We thank Imogen Duncan and Eliza-Alexandra Bujor for their valuable technical support. We also express our appreciation to the undergraduate students, Larissa Moldovan and Georgia Loweth, who contributed to the fieldwork portion of the study. Our special thanks are to Andrea Kovacs for her involvement in the final stages of the laboratory work and Gerbrandus Boots for reviewing the manuscript.

CO N FLI C T O F I NTE R E S T
The authors have no conflicts of interest to declare.

DATA AVA I L A B I L I T Y S TAT E M E N T
The partial 16S rRNA gene sequences were submitted to GenBank with accession numbers MN904979 to MN905002.