Plasmids conferring resistance to extended-spectrum beta-lactamases including a rare IncN+IncR multireplicon carrying bla CTX-M-1 in Escherichia coli recovered from migrating barnacle geese ( Branta leucopsis)

Background: Increasing antimicrobial resistance (AMR) is a global threat and wild migratory birds may act as mediators of resistant bacteria across country borders. Our objective was to study extended-spectrum beta-lactamase (ESBL) and plasmid-encoded AmpC (pAmpC) producing Escherichia coli in barnacle geese using whole genome sequencing (WGS) and to identify plasmids harboring bla genes. Methods: Barnacle geese feces (n=200) were collected during fall 2017 and spring 2018 from an urban area in Helsinki, Finland. ESBL/AmpC-producing E. coli were recovered from nine samples (4.5%) and isolates were subjected to WGS on both short- and long-read sequencers, enabling hybrid assembly and determination of the genomic location of bla genes. Results: A rare multireplicon IncN+IncR was recovered from one isolate carrying bla CTX-M-1 in addition to aadA2b, lnu(F), and qnrS1. Moreover, rarely detected IncY plasmids in two isolates were found to harbor multiple resistance genes in addition to the human-associated bla CTX-M-15. Poultry-associated bla CMY-2 was identified from the widely distributed IncI1 and IncK plasmids from four different isolates. One isolate harbored an IncI1 plasmid with bla CTX-M-1 and flor. A chromosomal point mutation in the AmpC promoter was identified in one of the isolates. WGS analysis showed isolates carried multiple resistance and virulence genes and harbored multiple different plasmid replicons in addition to bla-carrying plasmids. Conclusions: Our findings suggest that wild migratory birds serve as a limited source of ESBL/AmpC-producing E. coli and may act as disseminators of the epidemic plasmid types IncI1 and IncK but also rarely detected plasmid types carrying multidrug resistance. Human and livestock-associated ESBL enzyme types were recovered from samples, suggesting a potential for interspecies transmission. WGS offers a thorough method for studying AMR from different sources and should be implemented more widely in the future for AMR surveillance and detection. Understanding plasmid epidemiology is vital for efforts to mitigate global AMR spread.


Introduction
As antimicrobial resistance (AMR) continues to increase unevenly worldwide (CDC, 2019;EFSA & ECDC, 2020;WHO, 2020), it is becoming increasingly urgent to study the transmission routes of resistant bacteria and mobile genetic elements harboring resistance genes.The drivers behind increasing AMR levels in different niches, including humans, animals, and the environment, have been studied (Holmes et al., 2016), but occurrence and transmission routes of resistant bacteria need to be continuously monitored to enable early mitigation efforts.
Regarding AMR in humans, studies have indicated human-tohuman contact as the main route of transmission of ESBLproducing bacteria (Day et al., 2019;Mughini-Gras et al., 2019).However, resistant bacteria may also spread via animals, food, and the environment (EFSA, 2011;Mughini-Gras et al., 2019) and their role should be continuously monitored as part of a One Health approach.
When studying bacteria resistant to antibiotics in different environments, it is important to assess the true threat any finding possesses to human or animal health.Whole genome sequencing (WGS) offers an efficient method for comparative epidemiological analysis.Different species and locations may have their unique set of bacterial sequence types (STs) and typical resistance genes but overlap between human-or animalassociated resistant bacteria and plasmids has been shown (Carattoli, 2009;Carattoli, 2011;Rozwandowicz et al., 2018).Combining short-read and long-read sequence technology provides a more accurate assembly of sequence data, which is especially important when determining the presence and structure of AMR-encoding plasmids.Plasmids occurring in multiple different regions, bacterial species, and sources can be considered epidemic (Carattoli, 2009).Certain traits, such as pilus formation and effective conjugation machinery in IncI1 and IncN type plasmids or plasmid addiction systems, may aid in the successful spread of plasmids (Carattoli, 2009;Carattoli, 2013).
One of the drivers of AMR has been identified to be international travel (Holmes et al., 2016;Lääveri et al., 2018;van der Bij & Pitout, 2012;Woerther et al., 2017).As people and goods travel across country borders, so do wild animals.Migratory birds are able to travel across continents, potentially carrying resistant bacteria with them which they have picked up from anthropogenic waste sources (Bonnedahl & Järhult, 2014;Dolejska & Papagiannitsis, 2018).Wild birds have been shown to carry and spread ESBL-producing E. coli effectively between individuals in a flock (Sandegren et al., 2018), emphasizing the potential rapid spread of AMR.Waterfowl, especially barnacle geese (Branta leucopsis), have become a regular sight in many countries.Thousands of geese feed and defecate in densely human-populated areas, such as near housing and recreational parks (Elmberg et al., 2017).Hundreds of thousands of barnacle geese migrate over Finland each year and feed on crop fields (SYKE, 2016).In addition to direct contact with bird feces, food-producing animals may come into contact with crops contaminated with fecal matter, as barnacle geese increasingly forage on maintained grasslands and pastures (Jensen et al., 2018).
Barnacle geese are a protected species under the European Union Birds Directive (2009/147/EC) and listed under Annex I, which has enabled them to grow exponentially in number during recent decades, reaching a total population size of 1,390,000 in the 2010s (Fox & Leafloor, 2018).The populations are divided into three groups: east Greenland / Scotland & Ireland, Svalbard / southwest Scotland, and Russia / Germany & the Netherlands (Jensen et al., 2018).From the 1980s to 2010s, the Russia / Germany & the Netherlands population size increased by 30 times, with birds breeding now also in the Baltic and North Sea area (Jensen et al., 2018).Originally, barnacle geese only migrated through the Baltic Sea and bred in the Arctic zone, but since 1985 increasing numbers are also breeding in coastal areas of Finland and partly inland, especially the Turku archipelago region and the capital area (SYKE, 2019;Yrjölä et al., 2017).The geese included in this study were expected to belong to this latter group.
The increasing number of birds breeding in and migrating through new areas has resulted in birds being in closer contact with humans.Increased numbers of birds have also resulted in worries of contaminated recreational and drinking water, crop damage, and conflict in cities between humans and birds.Regarding AMR in birds, gulls in particular have been studied previously, and they have been found to harbor ESBLproducing bacteria frequently, although with varying prevalence between different countries (Stedt et al., 2015).The less studied geese and other waterfowl have also been shown to carry multiple pathogens (Elmberg et al., 2017), such as Campylobacter jejuni in barnacle geese in Finland (Llarena et al., 2015) and Yersinia spp. in barnacle geese in Sweden (Niskanen et al., 2003), but studies on the role of barnacle geese in AMR dissemination and potential transmission to humans or food-producing animals is limited.Barnacle geese breeding in Svalbard and wintering in the UK were found to carry intestinal E. coli with a high level of resistance against colistin (100%) but only a low level against ceftazidime (2%) (Hatha et al., 2013).Waterfowl in the Netherlands and Poland have been found to have different levels of resistance against amoxicillin, enrofloxacin, and tetracycline in one study (Kuczkowski et al., 2016), and the difference was speculated to be partly attributed to differences in the environment and the proximity to humans.The resistance patterns in wild birds can mirror those found in humans in different geographic locations (Stedt et al., 2015).As technologies have advanced, hybrid sequencing now offers a powerful tool for in-depth analysis of resistance genes along with plasmid replicons, virulence genes, and phylogeny between samples extracted from different sources.The goal of the present study was to study ESBL/AmpC-producing E. coli in barnacle geese residing in close proximity to humans, to identify resistance and virulence genes, STs, and plasmid replicons, and to conduct comparative phylogenetic analysis to identify potential epidemic plasmids.

Sampling
Fecal samples from barnacle geese (n=200) were collected during one day on two occasions, September 2017 (samples H1-H100) and May 2018 (samples H101-H200), in Helsinki, Finland.The sample location, a recreational park area, was the same on both occasions.During both sampling days, a flock of approximately 500 barnacle geese were present in the area.Samples were selected on the basis of collecting fresh droppings from the area, each sample being located approximately 0.5 -2 meters from the previous sample, totaling at covering a sampling area of approximately 100 x 100 meters.Samples were aseptically collected into individual 1 l plastic bags and transported to the laboratory for further analysis within 1 h.

Isolation and confirmation of Escherichia coli
From each sample, 1.0 ± 0.1 g of feces was enriched in 9 ml sterile buffered peptone water (Oxoid, Basingstoke, UK) by incubating at 37°C overnight.Subsequently, 10 μl of the pre-enrichment was streaked onto selective MacConkey agar plates (Oxoid, Basingstoke, UK) supplemented with 1 mg/l cefotaxime and incubated at 44°C for 18-22 h.One colony from each plate with bacterial growth was re-streaked onto MacConkey agar plates with 1 mg/l cefotaxime supplement and incubated at 37°C for 18-22 h.If a plate had morphologically different colonies, a representative colony from each different growth was streaked onto an agar plate.
After achieving a pure bacterial culture, the isolate was streaked onto a bovine blood agar plate and incubated at 37°C overnight for bacterial species determination with a matrixassisted laser desorption ionization-time of flight mass spectrometry (MALDI-TOF MS) based Bruker Biotyper (Bruker Daltonics).A score value of 2.0-3.0 was considered highconfidence and was set as the criteria.All isolates identified as E. coli were stored at -70°C for further characterization.
DNA extraction and whole genome sequencing Short-read sequencing.All ESBL/AmpC-producing E. coli isolates (n=9) were subjected to WGS with Illumina to study the presence of AMR and virulence genes and plasmid replicons, as well as to assess the multilocus ST.
Bacterial DNA was extracted and purified with the Pure-Link Genomic DNA Mini Kit (Cat# K182002, Invitrogen, Thermo Fischer Scientific, Carlsbad, CA, USA) according to the manufacturer's instructions.The assessment of DNA quality was carried out using a NanoDrop ND-1000 spectrophotometer (Thermo Fischer Scientific, Wilmington, DE, USA) and DNA quantity was measured using a Qubit 2.0 fluorometer (Invitrogen, Life Technologies, Carlsbad, CA, USA).Optical density OD260/280 of 1.8-2.0 and concentration of ≥ 50 ng/μl were set as thresholds.Library preparation was performed with Illumina Nextera XT and sequencing with Illumina Novaseq 6000 (Center for Genomics and Transcriptomics, Tübingen, Germany) with paired-end reads.Samples were sequenced with 100 × coverage, 2 × 100 bp read length, and a minimal phred quality score of 30.
Long-read sequencing.To study the complete sequences and to identify plasmid replicons carrying bla genes, all ESBL/AmpCproducing E. coli isolates (n=9) were additionally long-read sequenced.DNA extraction and purification were performed as described above.DNA extracts from all isolates were multiplexed in a random order with either SQK-LSK 108 or SQK-LSK 109 ligation sequence kit (Oxford Nanopore Technologies, Oxford, UK), depending on the availability of the respective flow cells, as described in more detail in the following sentences.DNA extracts from four geese isolates (H11, H21, H29, and H163) were multiplexed using the SQK-LSK108 ligation sequence kit (Oxford Nanopore Technologies, Oxford, UK) according to the manufacturer's protocol.Libraries were loaded onto FLO-MIN106D R9.4.1 MinION flow cells (Oxford Nanopore Technologies, Oxford, UK) used with the MinION Mk1B sequencing device and sequenced with MinKNOW software v19.06.8 for 48 h.For five isolates (H5, H58, H68B, H98, and H193) DNA extracts from three or two isolates at a time were multiplexed using the SQK-LSK109 ligation sequence kit (Oxford Nanopore Technologies, Oxford, UK) according to the manufacturer's protocol.Libraries were loaded onto FLO-FLG001 R9.4.1 Flongle flow cells (Oxford Nanopore Technologies, Oxford, UK) used with the MinION Mk1B sequencing device and sequenced with MinKNOW software v19.06.8 for 20-24 h.
The plasmid sequences were annotated with Prokka v1.13 (Seemann, 2014) and manually curated with BLASTn/BLASTp.Plasmid structures were compared with previously published reference plasmids for each different replicon type identified and visualized in BRIG v0.95 (Alikhan et al., 2011).The structure of plasmid pZPK-H11 was visualized with SnapGene software (from Insightful Science, available at https:// www.snapgene.com).Visualization of the plasmid structure can also be achieved with BRIG v0.95 (Alikhan et al., 2011), for example.Studied plasmids were compared using Bac-Compare (Liu et al., 2019) with previously published plasmids found with a BLASTn search against the National Center for Biotechnology Information (NCBI) database and the 20 best matches with available metadata for each studied incompatibility type (for IncY plasmids only six similar, previously published plasmids with available metadata were found) were used to build a core genome MLST (cgMLST) based tree with 95% occurrence for discriminatory loci.The minimum spanning tree was visualized using GrapeTree v1.5 (Zhou et al., 2018).Information on included previously published plasmids with available metadata from NCBI GenBank are provided in Table 1.

Ethics statement
The study did not include any handling or disruption of animals, and therefore ethical approval of the research was not needed.

Escherichia coli
Out of 200 samples, 98 (49%) yielded bacterial growth on MacConkey agar plates supplemented with cefotaxime (1 mg/l).Of these, 55 (56%) samples were collected in fall 2017 and 43 (44%) in spring 2018.One colony was tested from each sample by using MALDI-TOF, revealing E. coli from nine (4.5%) samples.Seven (78%) of these originated from fecal samples collected on the first sampling period and two (22%) from the second sampling period.All E. coli samples originated from fresh, wet droppings.

Antimicrobial susceptibility testing.
All nine isolates subjected to antimicrobial susceptibility testing were resistant to thirdgeneration cephalosporins (cefotaxime and ceftazidime) (Table 3).According to the combination disk diffusion test, four isolates were phenotypically ESBL producers, four isolates AmpC producers, and one isolate both an ESBL and AmpC producer.Antimicrobial susceptibility testing results are shown in Table 3.
Whole genome sequencing and sequence analysis WGS revealed seven different E. coli STs in the sequenced isolates (Table 4).Each sequenced isolate harbored from one to six different plasmid replicons.Three isolates carried an IncI1 plasmid replicon.Multidrug resistance was found in all of the sequenced isolates.Four of the isolates harbored genes conferring resistance to seven different antibiotic classes.In addition to beta-lactams, resistance genes were detected against aminoglycoside, fluoroquinolone, macrolide, lincosamide, streptogramin B, phenicol, sulfonamide, tetracycline, and trimethoprim.No fosfomycin or rifampicin resistance genes were detected.
Human-associated resistance gene bla CTX-M-15 was found in two isolates sharing the same sequence type ST453 and plasmid replicon IncY.Additionally, four and two isolates harbored bla CMY-2 and bla CTX-M-1 , respectively, which are commonly found from poultry (EFSA, 2011;Rozwandowicz et al., 2018).Interestingly, no beta-lactamase genes were recovered from isolate H68B, although this isolate was phenotypically an AmpC producer.The phenotype correlated with the genotype in all but two isolates-the aforementioned isolate H68B and isolate H193-which was phenotypically both an AmpC and an ESBL producer, but only bla CMY-2 was recovered from the WGS analysis.In addition, isolate H21 was phenotypically an AmpC producer but harbored both bla CMY-2 and bla TEM-32 .
The isolates harbored a wide variety of different virulence genes with 38 different genes found altogether.The increased serum survival gene iss, tellurium ion resistance gene terC,  and glutamate decarboxylase gene gad were found in all isolates, and the long polar fimbriae gene lpfA in seven isolates.Enterobactin siderophore receptor gene iroN and temperature sensitive hemagglutinin gene tsh were both found in three isolates.Two of the isolates harbored the adhesin intimin coding eae gene, which is associated with enteropathogenic E. coli (Frankel et al., 1998;Müller et al., 2016), but no Shiga toxin coding genes were found.

Plasmid analysis.
Eight out of the nine hybrid sequenced E. coli isolates were found to harbor a plasmid replicon with a bla gene.All hybrid-assembled bla-carrying plasmids were found to be in one circular contig.
One isolate, H68B, did not carry any plasmids with bla genes, but instead a chromosomal point mutation in the AmpC promoter was identified.This isolate did, however, harbor multiple IncF type replicons and resistance genes mdf(A) and tet(A) (Table 4).
pZPK-H163 from isolate H163 and pZPK-H193 from isolate H193 were found to be identical with pairwise alignment with BLASTn.Both plasmids were 89.6 kb in size with a G+C content of 50.31% and predicted 101 CDSs.pMLST analysis identified the plasmids as ST23 CC-2.In addition to bla CMY-2 , the plasmids carried the macrolide-associated resistance gene mdf(A).Similar to pZPK-H5, the only virulence gene identified on these plasmids was cia.
The studied IncI1 plasmids demonstrated typical IncI plasmid backbones with maintenance and stability-related regions with parAB, impCAB, ssb, psiAB, ardA, and pndAC genes, and transfer-associated, shufflon, and pilus formation regions (Figure 1A-B).In pZPK-H5, flor and cia genes were located in the accessory module, whereas bla CTX-M-1 was integrated into the shufflon region.As previously described, bla CTX-M-1 was integrated into shufflon segment B by a copy of ISEcp1, although in pZPK-H5 this IS element contains an integration of ISKpn26 (Brouwer et al., 2015).In pZPK-H163 and pZPK-H193, bla CMY-2 was located in the accessory module after the replication region.

Illumina run accession
IncK type plasmids Two isolates, H21 and H58, harbored IncK type replicons with bla CMY-2 .Plasmid pZPK-H21 from isolate H21 was 90.9 kb in size with a G+C content of 52.51% and 108 CDSs.
A BLASTn search against the NCBI database indicated that the most similar previously published plasmids consisted of poultry-associated IncK2 or IncB/O/K/Z replicons harboring bla CMY-2 (Table 1).pZPK-H21 was also found to be similar with two IncK2 plasmids derived from E. coli from a human urinary tract infection in Switzerland, with 94% coverage and 99.81% identity with plasmid pDV45 (GenBank accession: KR905384.1)and 92% coverage and 99.88% identity with plasmid p5312.29 (GenBank accession: KR905385.1).A cgMLST-based minimum spanning tree was built with 64 discriminatory loci (95% occurrence) and is visualized in Figure 3.
IncY type plasmids Two isolates, H29 and H98, harbored IncY type replicons.pZPK-H29 from isolate H29 and pZPK-H98 from isolate H98 were highly similar when compared with pairwise alignment with BLASTn, resulting in 100% coverage and 99.94% identity.Both plasmids harbored bla CTX-M-15 and bla TEM-1B .In addition to bla genes, both plasmids carried the resistance genes qnrS1, aph( 6)-Id, aph(3")-Ib, sul2, dfrA14, and tet(A).Both plasmids were 98.5 kb in size, had a G+C content of 51.13% and 98 predicted CDSs in pZPK-H29, and had a G+C content of 51.12% and 99 predicted CDSs in pZPK-H98.These IncY plasmids were not found to harbor any virulence genes.BLASTn comparison with previously described IncY type plasmid pPGR46 (GenBank accession: KM023153) demonstrated 71% coverage with both pZPK-H29 and pZPK-H98, and 99.98% identity with pZPK-H29 and 99.94% identity with pZPK-H98 (Figure 1D).A BLASTn search against the NCBI database identified only six IncY type previously published plasmids with available metadata to be similar to pZPK-H29 and pZPK-H98 (Table 1).Three of the similar previously

IncN+IncR multireplicon
From isolate H11, two plasmid replicons, IncN and IncR, were identified on the same contig.This multireplicon plasmid pZPK-H11 was 72.4 kb in size, with a G+C content of 50.91% and 78 predicted CDSs.Pairwise alignment with BLASTn indicated that the plasmid aligned with IncN reference plasmid R46 (GenBank accession: AY046276) with a coverage of 55% and 99.73% identity, whereas alignment with previously published IncR type plasmid pK245 (Gen-Bank accession: DQ449578) resulted in a coverage of 31% and 98.53% identity (Figure 1E).
The multireplicon plasmid harbored bla CTX-M-1 and additional resistance genes qnrS1, aadA2b, and lnu(F).No virulence genes were detected on the multireplicon plasmid.The IncN replicon in the plasmid was identified as ST1 with pMLST analysis.Top two hits from a BLASTn search against the NCBI database were IncF+IncN type multireplicons both isolated from E. coli of human origin, one in Italy with bla NDM-5 (Gen-Bank accession: MT199177.1;60% coverage and 99.97% identity) and the other in Switzerland carrying bla CTX-M-1 (GenBank accession: KJ484634.1;59% coverage and 99.97% identity) (Table 1).Other top hits with BLASTn included IncN plasmids isolated from various sources, such as uncultured bacteria from wastewater treatment plant effluent in Germany, carrying bla TEM-1 (GenBank accessions: JN102342.1,JN102344.1,and JN102343.1).A cgMLST-based minimum spanning tree was built with 34 discriminatory loci (95% occurrence) and is visualized in Figure 5.

Discussion
The purpose of this study was to identify ESBL/pAmpCproducing E. coli in migrating barnacle geese residing near humans and to investigate plasmids harboring bla genes.The bla genes were encoded on four different plasmid replicons from nine different samples: IncI1 and IncK, the rare IncY, and the multireplicon IncN+IncR.
IncI1 carrying bla CTX-M-1 was identified in one of the isolates in our study.The finding of two identical IncI1 plasmids harboring bla CMY-2 from two samples, pZPK-H163 and pZPK-H193, indicates very recent clonal transmission within the flock or the possibility of samples originating from the same goose individual.IncK type plasmids with bla CMY-2 were recovered from two isolates in our study.IncK type plasmids carrying bla CMY-2 and bla CTX-M-14 have been previously isolated mainly from E. coli from animal sources in Europe (Rozwandowicz et al., 2018).The most prevalent resistance genes in the isolates in our study were AmpC type bla CMY-2 followed by bla CTX-M-1 and bla CTX-M-15 .The finding of bla CMY-2 and bla CTX-M-1 on IncI1 plasmids is a common finding among poultry (Accogli et al., 2013;Leverstein-van Hall et al., 2011), and these genes and plasmid replicons have been identified previously also from other wild avian species, such as seagulls and pelicans in Florida (Poirel et al., 2012).Wild birds of different species in Catalonia have also been found to harbor bla CTX-M-15 and bla CMY-2 among other beta-lactamases, such as bla OXA-48 in a barn owl (Darwich et al., 2019).No carbapenemases were recovered in our study.In addition, in a study by Bonnedahl et al. (2009), 9.4% of wild yellow-legged gulls in the south of France were found to harbor ESBL-producing E. coli and 6% more specifically bla CTX-M-1 .The finding of beta-lactamases, especially in wild birds habituating aquatic environments, is worrisome because of the potential of bacterial transmission between humans and animals via surface water.Wild gulls in Sweden were found to harbor bla CTX-M-15 /bla CTX-M-14 -producing E. coli similar to human and surface water isolates (Atterby et al., 2017).Wild birds may acquire resistance genes from anthropogenic sources and circulate resistance determinants again back into the human population, although contact with wild birds has been estimated to contribute only 0.3% of community-acquired intestinal carriage of ESBL/pAmpC-producing E. coli in humans in the Netherlands (Mughini-Gras et al., 2019).
In addition to human health, wild migratory birds may play a role as disseminators of human-derived AMR to remote areas, potentially endangering the welfare of native animal species (Hernández & González-Acuña, 2016).
Interestingly, the finding of ESBL/AmpC-producing E. coli in 4.5% of the studied barnacle geese fecal samples is slightly less than the prevalence of ESBL/AmpC-producing E. coli and Klebsiella pneumoniae described in Finnish asymptomatic human carriers (6.3%) (Rintala et al., 2018) It has to be taken into account, however, that the sample size was limited.There have been studies of the effect of ultraviolet light on E. coli survival (Vermeulen et al., 2008;Whitman et al., 2004), but to assess the effect of sunlight and other environmental factors on E. coli or resistance genes and plasmid survival in bird feces, more studies would need to be conducted.Seven out of nine samples positive for ESBL/ AmpC-producing E. coli originated from samples collected in fall 2017, and temporal changes have been identified in ESBL-producing Enterobacteriaceae previously in urban Swedish mallards, with higher prevalence during warm summer months (Hessman et al., 2018).
To assess the actual risk of AMR in wildlife, bacteria, resistance genes, and mobile genetic elements have to be typed to a level where comparison between different sources, i.e., humans and animals, is plausible.WGS provides a powerful tool for studying phylogeny between different bacterial and plasmid reservoirs, and also enables the early detection of new resistance genes or successful plasmids.Long-read sequencing of the isolates allows for even more in-depth analysis of the location of resistance genes on plasmid replicons.
As E. coli belong to the normal intestinal microbiota of animals and virulence potential varies between strains, the public health relevance of ESBL/AmpC-producing E. coli derived from barnacle geese feces should be interpreted with caution.By examining virulence traits, however, it is possible to identify potential pathogens.The increased serum survival gene iss was found from all of the sequenced isolates.Iss has been identified in extraintestinal pathogenic E. coli (ExPEC) as well as in avian pathogenic E. coli (APEC) (Johnson et al., 2008;Tivendale et al., 2004).Iss and iroN have been found to be prevalent in E. coli, causing salpingitis in egg-laying hens (Poulsen et al., 2020), and have also been identified in E. coli in poultry production (Kim et al., 2020;Oikarainen et al., 2019).The glutamate decarboxylase gene gad was also prevalent in the sequenced isolates, and it has been identified frequently in APEC in broilers together with iss and iroN (Azam et al., 2020).These findings highlight the potential of AMR transmission via virulent E. coli strains between bird species, as well as between other animals and potentially humans.
Wildlife may serve as a reservoir for resistant bacteria (Carroll et al., 2015) and more than 60% of emerging infectious diseases since the 1940s have been zoonotic (Jones et al., 2008), which makes the continuous surveillance of wildlife pathogens extremely important.Our study highlights the fact that wild birds living in close contact with humans carry bacteria with plasmid-mediated beta-lactamases and may pose a potential risk of bacterial clonal spread or horizontal gene transfer through the environment.

Conclusion
Wild migratory geese were found to carry ESBL/AmpCproducing E. coli with diverse bacterial STs and the assessment of samples with WGS revealed the location of resistance genes in specific plasmid replicons.A novel multireplicon, IncN+IncR, carrying bla CTX-M-1 was recovered, and humanassociated bla CTX-M-15 was found from two IncY plasmids.The findings indicate that wildlife carrying plasmids and resistance genes are potentially worrisome for public health.In addition to successful, more frequently identified plasmid replicons IncI1, IncK, and IncN, the rare IncY plasmid and IncN+IncR multireplicon plasmids were recovered in our study, indicating that wildlife carry epidemic plasmids and also serve as potential disseminators of plasmids currently viewed as uncommon in humans and livestock.Although the prevalence of ESBL/AmpC-producing E. coli was relatively moderate at 4.5%, the transmission potential should not be underestimated, especially in urban areas with dense human and animal populations.Continuous and prudent monitoring of resistance determinants in different environments is vital for understanding evolving resistance patterns and the epidemiology of AMR within and between the different One Health compartments.
The manuscript is easy to read and clear.All manuscript sections are high quality writing, detailed description of used methods and tools allows to repeat the research.The sequences are easily accessible at NCBI database.I do not have any remarks.Reviewer Expertise: WGS analysis of AMR (pathogenic) zoonotic bacteria I confirm that I have read this submission and believe that I have an appropriate level of expertise to confirm that it is of an acceptable scientific standard.
with disk diffusion method according to EUCAST (2017); EUCAST (2019).b Criteria for ESBL production: resistance to third-generation cephalosporins and ≥ 5 mm difference in combination disk diffusion test; criteria for AmpC production: resistance to thirdgeneration cephalosporins and cephamycin and < 5 mm difference in inhibition zones in combination disk diffusion test.c R = phenotypically resistant; S = phenotypically susceptible.

Figure 2 .
Figure 2. Minimum spanning tree based on core genome multilocus sequence typing of IncI1 type plasmids.Plasmids pZPK-H5, pZPK-H163, and pZPK-H193 compared with 20 previously published IncI1 plasmids from GenBank.Nodes are colored by the country of origin and node text includes information on the source, bla gene, and bacterial species if other than Escherichia coli.GenBank accession numbers are provided in parentheses.

Figure 3 .
Figure 3. Minimum spanning tree based on core genome multilocus sequence typing of IncK type plasmids.Plasmids pZPK-H21 and pZPK-H58 compared with 20 previously published IncK or IncB/O/K/Z plasmids from GenBank.Nodes are colored by the country of origin and node text includes information on the source, bla gene, and possible additional mcr-1.GenBank accession numbers are provided in parentheses.

Figure 4 .
Figure 4. Minimum spanning tree based on core genome multilocus sequence typing of IncY type plasmids.Plasmids pZPK-H29 and pZPK-H98 compared with six previously published IncY plasmids from GenBank.Nodes are colored by the country of origin and node text includes information on the source, bla gene, and bacterial species if other than Escherichia coli.GenBank accession numbers are provided in parentheses.

Figure 5 .
Figure 5. Minimum spanning tree based on core genome multilocus sequence typing of IncN+IncR multireplicon.Plasmid pZPK-H11 compared with 20 previously published plasmids from GenBank.Nodes are colored by the country of origin and node text includes information on the source, bla gene, or other resistance gene if the bla gene was not present, bacterial species if other than Escherichia coli, and replicon type if other than IncN.GenBank accession numbers are provided in parentheses.

Figure 6 .
Figure 6.Genetic structure of multireplicon plasmid pZPK-H11.Predicted coding sequences and their orientation are represented by arrows colored based on the function of the gene product.Size of the plasmid is depicted on the outer circle.
the work clearly and accurately presented and does it cite the current literature?YesIs the study design appropriate and does the work have academic merit?YesAre sufficient details of methods and analysis provided to allow replication by others?YesIf applicable, is the statistical analysis and its interpretation appropriate?Not applicableAre all the source data underlying the results available to ensure full reproducibility?YesAre the conclusions drawn adequately supported by the results?YesYes Are all the source data underlying the results available to ensure full reproducibility?Yes Are the conclusions drawn adequately supported by the results?Yes Competing Interests: No competing interests were disclosed.

Table 4 . Genomic characteristics of hybrid sequenced Escherichia coli isolates and plasmid replicons from barnacle geese.
(Simões et al., 2010)14)n wild birds previously, but this was the first time it was studied in migratory birds in Finland, and with WGS applied.The prevalence of 4.5% is relatively low compared with other studies conducted with different avian species.In a study conducted in the Netherlands(Veldman et al., 2013), ESBL/AmpC-producing E. coli were found in 15.7% of 21 different bird species tested, mainly originating from aquaticassociated species.In Spain, 15% of sampled birds were found to be positive for ESBL/pAmpC-producing E. coli(Alcalá  et al., 2016), whereas in birds of prey in Germany and Mongolia the prevalence was found to be 13.8% and 10.8%, respectively(Guenther et al., 2012).The lower percentage in the present study could be partly explained by the different bird species examined.Barnacle geese do not feed on landfills or other human-related waste areas like gulls in certain areas(Bonnedahl et al., 2014)or other bird or mammal species like birds of prey but, as a waterfowl, may come into contact with human wastewater.Although the prevalence of ESBL/AmpC-producing E. coli in our study was less than in some other birds studied earlier, barnacle geese that have habituated to densely human-populated areas, such as recreational parks and public beaches(Simões et al., 2010), pose the potential risk of AMR spread to humans.Parks and public beaches can quickly become contaminated with fecal matter while flocks of hundreds of birds feed and defecate in the area.Additionally, household pets and children may come into contact with fecal matter more easily.