Antibiotic resistance, virulence, and phylogenetic analysis of Escherichia coli strains isolated from free-living birds in human habitats

Wild birds can be colonized by bacteria, which are often resistant to antibiotics and have various virulence profiles. The aim of this study was to analyze antibiotic resistance mechanisms and virulence profiles in relation to the phylogenetic group of E. coli strains that were isolated from the GI tract of wildfowl. Out of 241 faecal samples, presence of E. coli resistant to a cephalosporin (ESBL/AmpC) was estimated for 33 isolates (13,7%). Based on the analysis of the coexistence of 4 genes encoding ESBLs/AmpC (blaCTX-M, blaTEM, blaSHV, blaAmpC) and class 1 and 2 integrons genes (intI1, intI2) a subset of two resistance profiles was observed among the investigated E. coli isolates carrying blaAmpC, blaSHV, and blaCTX-M, blaTEM, class 1 and 2 integrons, respectively. The E. coli isolates were categorized into 4 phylogenetic groups A (39.4%), B2 (24.25%), D (24.25%) and B1 (12.1%). The pathogenic B2 and D groups were mainly typical for the Laridae family. Among the 28 virulence factors (Vfs) detected in pathogenic phylogenetic groups B2 and D, 7 were exclusively found in those groups (sfa, vat, tosA, tosB, hly, usp, cnf), while 4 VFs (fecA, fyuA, irp2, kspMTII) showed a statistically significant association (P≤0.05) with phylogroups A and B1. Our results indicated that strains belonging to commensal phylogroups A/B1 possess extensive iron acquisition systems (93,9%) and autotransporters (60,6%), typical for pathogens, hence we suggest that these strains evolve towards higher levels of virulence. This study, which is a point assessment of the virulence and drug resistance potential of wild birds, confirms the importance of taking wild birds as a reservoir of strains that pose a growing threat to humans. The E. coli analyzed in our study derive from different phylogenetic groups and possess an arsenal of antibiotic resistance genes and virulence factors that contribute to their ability to cause diseases.


Introduction
The relationship between humans and farm animals has already been proven many times, but there are still only a few reports on the association of infections in humans with a potential source among wild birds [20][21][22][23]. Meanwhile, nowadays, the assessment of the amount of various anthropogenic pollutants entering the environment is of key importance. Wild birds, especially waterfowl are often taken into account as indicators of ecosystem health, reflecting changes in habitats, increased disease incidence, and exposure and effects of biological and chemical pollution [24][25][26][27]. Moreover, migratory birds can transport microorganisms to distant sites where different bird species often congregate in large numbers, i.e. at stopover sites or wintering areas. In such places, horizontal transmission might occur, especially among individuals residing in water bodies that host dense flocks, which waterfowl often do [26,28]. Hence, migrating wild birds can play an important epidemiological role in spreading bacteria resistant to antibiotics in the environment which could potentially become a hazard to human and animal health [29][30][31][32][33].
There is little known about the occurrence of drug-resistant bacteria in wild birds living in Poland as, so far, no extensive research has been carried out on this topic. Hence, the aim of this study was to determine the prevalence of ESBL/AmpC E. coli strains in wild birds living close to human habitats in northern and central part of Poland. E. coli strains isolated from the gastrointestinal tract (GI) of wild-fowl were screened for their antibiotic resistance mechanisms and virulence profiles.

Sampling and collection of E. coli avian strains
The bird faeces samples were collected from the cloaca of 241 birds belonging to the Anatidae families (Mallard Anas platyrhynchos-128, Mandarin Duck Aix galericulata-1, Mute Swan Cygnus olor -5), Laridae (Black-headed Gull Chroicocephalus ridibundus -38, Common Gull Larus canus -12, European Herring Gull Larus argentatus -16), Rallidae (Eurasian Coot Fulica atra-29, Common Moorhen Gallinula chloropus -1) and Corvidae (Hooded Crow Corvus cornix-5, Jackdaw Corvus monedula-4, Rook Corvus frugilegus-2) between January 2017 and October 2018 (S1 Table). The birds were sampled in northern and central Poland on different urban water reservoirs and on municipal beaches of the Baltic Sea Coast. All the bird handling and sample collecting was performed by persons with a valid ringing license, issued by the Ornithological Station of the Polish Academy of Sciences approved by the General Directorate for Environmental Protection in Poland (approval number: DZP-WG.6401.102.2020.TL) and the Ministry of the Environment (approval number: DL-III.6713.11.2018.ABR). Cloacal swabs were collected during the routine procedure of bird ringing without any unnecessary harm to the animals. After the sampling, the birds were immediately released. After the collection, the faecal samples were immersed in Amies transport medium and transported in an isothermal box to the laboratory. Within 3-4 h, the samples were analyzed for the presence of ESBL/AmpC-producing isolates with the use of cefotaxime-supplemented MacConkey agar (MCA) medium (2 mg/l). If a sample scored positive for the presence of antibiotic resistant bacteria, one lactose positive colony grown on MCA was selected, grown to pure culture and further analyzed. Isolated antibiotic-resistant coliform colonies were identified using MALDI-TOF MS (MALDI biotyper; Bruker Daltonics, USA) according to the manufacturer's instructions. After that, only the strains identified as E. coli were used for further analysis (n = 33). The purified isolates were grown on LA (Graso, Poland) medium and then stored at -80˚C in 20% glycerol stocks.

Bacterial DNA isolation
The genomic DNA of bacterial isolates (n = 33) was purified using the GENOMIC DNA KIT (BLIRT S.A., Gdansk, Poland) according to the manufacturer's protocol. The concentration of extracted DNA was measured by NanoDrop ND-100 (Thermo Fisher Scientific, Wilmington, USA). For the detection of the plasmid-born genes, DNA was isolated from bacterial strains by standard alkaline lysis and ethanol precipitation [35].

Antibiotic resistance genes and integrons detection
The presence of gene coding for antimicrobial resistance mechanisms was verified in ESBL/ AmpC-positive strains using molecular methods. Simplex polymerase chain reaction (PCR) was used for the detection of the ESBL gene's phenotype (bla CTX-M , bla SHV , bla TEM ), and AmpC β-lactamase gene (ampC). The presence of integrons was detected by PCR amplification of the integrase gene intI1 (for class 1 integrons) and intI2 (for class 2 integrons). Amplicons of bla CTX-M were randomly selected for sequencing (Genomed S.A. Warsaw, Poland). Specific primers and the PCR conditions were as described previously [36,37].

E. coli molecular typing
The polymerase chain reaction melting profiles (PCR MP) procedure was used to assess the genomic diversity of the isolated E. coli. The PCR MP was carried out as described in Krawczyk et al. [38]. Amplified genomic DNA fragments were separated using polyacrylamide gel electrophoresis (6%) in 1×TBE buffer and stained with ethidium bromide. Images of the gels were analyzed and archived using the Versa Doc Imaging System version 1000 (Bio-Rad Laboratories, Hercules, USA). Dendrogram was generated by the Dice Coefficient (DC) with UPGMA method and with a setting of 1% band tolerance (FPQuest TM software, BioRad; version 4.5).

Phylogenetic analysis
The phylogenetic group of each E. coli strain was determined according to the method developed by Clermont et al [15], by multiplex PCR of the genes chuA, yjaA and the DNA fragment TSPE4.C2.

Statistical analysis
Comparisons of the relative proportions of virulence genes were tested using Fisher's exact test or Pearson's χ2 test. Fisher's exact test was used to compare frequencies between groups. The threshold for statistical significance was a P value <0.05. The virulence scores of the strains are reported as the sum of the virulence genes they possessed. The Mann-Whitney U test was used to compare virulence scores among groups. Cluster analysis was based on the presence and absence of virulence genes and resistance genes. A binary matrix was used to determine similarities using the Euclidean distance and complete linkage, and strains were grouped using R package stats [42]. Dendrograms were constructed using the packages ggplot2 [43] and ggdendro [44] R version 3.5.3.

ESBL/AmpC-producing E. coli strains detection
Over the past decades multi-resistant ESBL-producing E. coli have become a prototype species for the spread of antimicrobial resistance into wildlife [45]. A total of 241 faecal samples from cloacal swabs of wild birds were screened for reduced sensitivity to third-generation cephalosporins. Of these 52 (21.5%) strains showed ESBL/AmpC phenotype and 33 isolates were confirmed as E. coli. These 33 strains were recovered from birds belonging to 6 different avian species (11 Mallards; 4 Black-headed Gull; 3 European Herring Gulls; 4 Common Gulls; 10 Eurasian Coots; 1 Jackdaw) (S1 Table).
The results showed, there is a statistically significant (P = 0.005) difference between four of the studied families in gut colonisation with ESBL/AmpC strains (Table 1).

Distribution of genes coding for ESBL resistance, AmpC lactamase and integrons class 1 and 2 among E. coli isolated from birds
In Enterobacterales and other Gram-negative bacteria, ESBLs often possess one of three types of genes: bla CTX-M , bla TEM and bla SHV , which enable them to mediate resistance to β-lactams.
Based on an analysis of the coexistence of six genes encoding ESBLs/AmpC resistance and class 1 and 2 integrons, the relationship between the strains was estimated (Fig 1).
A subset of two different patterns of genes (P1, P2) was observed among the investigated E. coli isolates. Profile P1 grouped together bla AmpC and bla SHV genes was mainly associated with clade I. The second profile (P2) included bla CTX-M and bla TEM genes, as well as both the class 1 Table 2 and class 2 integrase genes, and was also more common in clade I. We also observed the complete absence of class 2 integrons in clade II and III. This finding suggests that the resistance profile of the bacterial isolates in this study may not only be linked to integrons.

Avian ESBL/AmpC-producing E. coli belong to diverse phylogenetic groups
All isolates were grouped to phylogenetic clades according to Clermont et al. [15]. The E. coli isolates belonged to the phylogenetic groups A (39.4%), B2 (24.25%), D (24.25%) and B1 (12.1%). Phylogenetic typing demonstrated that strains isolated from the Anatidae were mainly classified in A phylogroup (64%), isolates from the Rallidae were classified in the B2 (50%) or A phylogroup (40%), while strains isolated from Laridae belonged to all phylogroups (Table 3). Only one ESBL/AmpC E. coli strain isolated from Corvidae was assigned to group D. The ESBL/AmpC E. coli strains that belonged to phylogroup B2 were isolated from only two families (Laridae, Rallidae), which is statistically significant (P = 0.041).

Distribution of ESBLs encoding genes according to phylogenetic groups
The occurrence and coexistence of bla CTX-M , bla TEM , bla SHV and bla AmpC genes in various phylogenetic groups (A, B1, B2 and D) were analyzed (Fig 2).
The bla CTX-M gene, similarly to bla TEM , appeared most often in the commensal group A (39.4% and 24.2%, respectively). Their co-occurrence (bla CTX-M + bla TEM ) in this group was also the most often recorded (24.2%). The bla CTX-M + bla TEM gene combination was equally distributed (12.1%) among phylogroups B1, B2 and D. The other β-lactamase gene, bla SHV, was more frequently observed in the group of B2 (6.1%) and D (9.1%) pathogenic strains. Only 3% of the strains carrying the bla SHV gene were assigned to group A, while the bla SHV gene was not found in group B1. The coexistence of three bla CTX-M + bla TEM + bla SHV genes was identified in 4 isolates belonging to the B2 (6.1%), A and D (3% each) phylogroups. However, no such gene set was found in group B1. Our research showed that the phylogenetic group B2 is the richest in genes associated with ESBL resistance.

Virulence factors (VFs) and their prevalence in phylogenetic groups
Genes encoding the VFs associated with the production of adhesins, siderophores and other iron uptake systems, toxins, autotransporters and genes associated with capsule and biofilm synthesis were sought. The most frequent virulence genes such as: fepA (siderophore uptake transmembrane transporter activity), entB (enterobactin synthase component B), fimH (type 1 fimbriae) and fecA (the outer membrane receptor), were observed in �60% of the isolates. In contrast, 13 genes, including usp (uropathogenic specific protein), iha (putative adhesin-siderophore) and papG (P fimbriae), were present in less than 10% of the isolates, whereas 5 virulence genes (afa/ Dr, ibeA, focG, pssA, boa) were not detected among the 33 isolates. The median virulence score was 7 (range 1-16) for all isolates. Taking into account the division into commensal and pathogenic strains (according to the criteria of Clermont et al. 2007) [15] it was observed that out of the 31 examined virulence genes only fecA gene was significantly associated with the commensal group of isolates (A, B1) while, three (kspMTII, irp2, fyuA) were significantly associated with the pathogenic group of isolates (B2, D) (S2 Table). Virulence genotypes in relation to the individual phylogenetic group among 33 avian E. coli isolates were also analyzed. The median virulence score was higher for D and B2 groups (8 VFs (range 5-11) and 6.5 VFs (range 4-16), respectively) than VFs detected among strains from phylogroups A and B1 (6 VFs (range 1-14) and 5.5 VFs (range 4-12), respectively). It is worth noting that, among the 28 VFs detected in phylogroup B2, six were exclusively detected in that group (vat, tosA, tosB, hly, usp and sfa). Five VFs (fecA, fyuA, aida, irp2, kspMTII) showed a statistically significant association (P�0.05) with phylogenetic groups A, B1, B2 and D. In Table 4 only statistically significant results are presented.

Phylogenetic analysis based on the virulence-associated gene (VAG) profile
On the basis of virulence genes presence, the relationship between E. coli strains was assessed. On the cluster analysis, strains were distributed into five groups (Fig 3). Group I is only represented by a single isolate (IS 20) assigned to the phylogenetic group B2, which possesses most of the virulence genes. This isolate is the only one that has the sfa gene encoding S-fimbriae and a set of genes typical for uropathogenic strains (usp, hly, cnf1, tosA and tosB). This strain was isolated from a sample of birds representing the Rallidae family. Group III has the least VFs. These isolates are assigned to phylogenetic group A and considered as a group of commensal strains. Groups IV and V have a comparable amount of VFs but are phylogenetically diverse. The phylogenetic groups D and B2 were predominant in Group IV. There were only 3 isolates in Group II and they have more virulence genes than isolates from Groups IV and V, but less than the isolate from Group I. We found that fyuA, irp2 and kspMTII virulence genes are significant in differentiation between the pathogenic strains and the commensals. The results of this study also show that all strains, in general, had at least one virulence gene related to iron acquisition. Furthermore, the strains belonging to B1 and D phylogroups had predominantly at least one gene related to the autotransporter. Neither the gene coding capsule nor the toxin were included in A and B1 phylogroups.

PLOS ONE
The fourth clade (CIV) was assembled from 4 isolates, Is1, Is25, Is24 and Is28, with three genotypes: Gp14, Gp15 and Gp16, with A and B subtypes, respectively. Is24 and Is28 differed in colony morphology and size, and were obtained from a single swab collected from a Blackheaded Gull. It is noteworthy that both isolates representing the same genotype belonged to different phylogroups, A and B2, respectively, and showed unique virulence and antimicrobial resistance profiles. Group B2 E. coli is not common among commensal intestinal flora and the presence within this group significantly more virulence factors is caused mainly by horizontal gene transfer (HGT).
Isolates Is2, Is4 and Is14, associated with genotypes Gp17 (2×) and Gp18, respectively, belong to the fifth clade (CV). Is2 and Is4 were recovered from two European Herring Gulls sampled at different locations at an interval of five weeks.
Strains carrying the same genotype are assigned to the same phylogroup and often share the same antibiotics resistance profile. However, some subtypes of the same genotype belong to different phylogroups and do not correspond to the same antibiotics resistance profile, which suggests HGT of the resistance genes.

Discussion
Wild birds live in most habitats around the world and their faeces, containing potentially harmful bacteria, are freely dispersed in the environment. Bacteria causing urinary tract infections, sepsis and respiratory infections, as well as those associated with food poisoning, are common in the faeces of all birds [46]. A large recipient of antibiotic-resistant bacteria that come from different sources (e.g., bird droppings, medical waste, municipal wastewater, agricultural applications) is the aquatic environment. It also appears that birds roosting on surface water have an adverse effect on the microbiological quality of the water [47]. In connection with this, there is an increasing body of research demonstrating that the aquatic environment is an ideal setting for the acquisition and dissemination of antimicrobial resistance genes and virulence factors harboured by commensal and pathogenic bacteria [46].
Based on our research, we wanted to answer the following questions: 1. Are extended-spectrum β-lactamase-positive Escherichia coli common among wild birds living in Poland? 2. What VFs are carried by E. coli strains isolated from wild birds in Poland? 3. Which phylogenetic groups are typical for E. coli strains isolated from the GI-tract of wild birds? 4. Is there an association between phylogenetic groups with multidrug resistance and VFs in E. coli isolates from wild birds? 5. What is the genetic diversity of E. coli strains isolated from the GI-tract of wild birds living in the habitats of people?

Wild birds as a reservoir of ESBL/AmpC E. coli
Among Enterobacterales, ESBL/AmpC-producing E. coli has been increasingly reported in Europe and other areas of the world. The CTX-M-type β-lactamases are the most common ESBLs among isolates of human and veterinary origin worldwide [48][49][50][51]. In our study a comparatively high proportion (13.7%) of wild birds were carriers of ESBL or/and AmpC-producing E. coli strains (S1 Table). This percentage correlates with results obtained from studies performed in Europe [52][53][54][55]. According to our report the highest share of ESBL strains was found in rallids and the lowest in corvids and ducks. This may reflect differences in diet and feeding among these groups. However, the rallids were captured in sites different from those where the other species with ESBL strains were caught. Hence, the region where the ringing of the birds takes place may be essential for the observed differences between bird families in ESBL/AmpC carriage. The most commonly identified resistance determinants were CTX-M and TEM. Characterization of the β-lactamase genes showed a dominance of the bla CTX-M gene present in 87,9% of the ESBLs-producing E. coli isolates. Direct sequencing of amplicons, identified CTX-M-15 as the main CTX-M-type β-lactamase. A significant presence of CTX-M-producing species among strains isolated from wild birds has been reported in other studies [36,54]. TEM-and SHV-type ESBLs were mostly prevalent among nosocomial E. coli isolates and not community-acquired pathogens [55][56][57]. The enormous spread of CTX-Mtype ESBLs is rapidly changing ESBL epidemiology and, in many parts of the world, these enzymes are now the most prevalent ESBLs in Enterobacterales. In Europe, the CTX-M type has already taken the dominant position from TEM and SHV types [54,55].
By identifying the same clones in humans and wild/domestic animals, it has been proven that the transmission of ESBL-producing strains between species is possible. The key factor for dissemination of ESBLs between bacteria, in both hospital and community settings, is HGT [48,51]. A particular role in HGT play integrons. Many Enterobacterales that inhabit the intestinal tracts of humans and animals carry integrons as a consequence of the high consumption of antibiotics in human and veterinary medicine [58]. Class 1 integrons identified in a great number of bacterial genera appear to be prevalent. About 30% of all resistant E. coli recovered from cattle, pigs and poultry carried a class 1 integron [59]. The presence of integrons among resistant E. coli isolates from wild birds is rather low in contrast to hospital E. coli strains [60][61][62][63], however, in our study, 72% of the isolates of E. coli contained the class 1 integrase gene and as many as 51% of the isolates tested contained the class 2 integrase gene. Taking into consideration that wild birds do not naturally come into contact with antibiotics, it is remarkable that class 1 and 2 integrons were frequently detected among the isolates in this study.

VFs are carried by E. coli strains isolated from wild birds in Poland
The results of this study showed that wild birds may be colonized by both commensal and pathogenic E. coli strains. To discriminate between pathogenic and commensal E. coli strains, the VFs irp2, fyuA, kspMTII and fecA that distinguished both pathotypes were studied. The irp2-fyuA gene cluster has been confirmed to be located on a high pathogenicity island (HPI) detected originally in Yersinia spp. [64]. Both genes are frequently found in pathogenic strains of E. coli [65][66][67][68][69]. Studies carried out in Brazil suggest that irp2 could be a candidate for predicting the pathogenicity of avian E. coli strains, and fyuA may be related to virulence [70]. In our studies, genes irp2 and fyuA were more often associated with pathogenic rather than commensal strains (81.3% and 29%, respectively), with differences statistically significant (P value 0.004 for both) (S2 Table).
On the basis of our results, we also infer that kspMTII may be a candidate for predicting the degree of E. coli strains pathogenicity. As reported, prevalence rates of the kspMTII gene for avian pathogenic E. coli isolates have tended to be low: 15.7% and 0% for American [71] and Zimbabwean isolates [72]. In our study the kspMTII gene was found more frequently among pathogenic group B2/D (37.5%) when compared with commensal strains from group A/B1 (0%), with differences statistically significant (P value 0.007).

The phylogenetic groups of E. coli strains isolated from the GI-tract of wild birds
Based on the genetic structure of the E. coli population in animals and humans, the association between the phylogroup strain and host and environmental factors has been demonstrated [73][74][75]. Most of avian E. coli strains are assigned to phylogroups A, B1 while the majority of the extraintestinal pathogenic E. coli isolates of human origin are associated with phylogenetic type B2 and to a lesser degree to group D [24,25]. Diet, gut morphology and body mass appear to be crucial predictors of the distribution of the phylogenetic groups [76]. In animals, the main factor influencing the genetic structure of the E. coli population in the intestine is domestication. Domestic animals have a lower percentage of B2 strains than their wild counterparts (from 30% in wild animals to 14% and 11% in livestock and zoo animals, respectively) and a higher percentage of strains belonging to the phylogenetic group A [77][78][79]. In our research, we conclude that the wild birds were colonized both by non-pathogenic strains (with the advantage of A group) and by pathogenic strains belonging to groups B2 and D (24,24% each). The most commonly detected pathogenic human ExPEC strains belong to group B2, which suggests that avian E. coli had a hospital origin.
The strains isolated from gulls (Laridae) were mostly classified as D or B2 phylogroups, similarly, the strains isolated from Rallidae predominantly belong to the B2 phylogroup (62,5%). Gulls are mainly omnivorous and in the non-breeding season they are associated with human settlements where they often feed on litter of human origin [76,80] and may have been in contact with human ExPEC strains [37]. Rallidae, on the other hand, remain only on bodies of water and have limited contact with anthropogenic waste. It seems that the presence of B2 ExPEC strains is not related to environmental selectivity or the feeding ecology of these bird families, but rather these strains are widespread in the urban environment as they were found in the samples collected in different urban areas.

An association between phylogenetic groups with multidrug resistance and virulence in E. coli isolates from wild birds
Our research showed that most of the ESBL genes were found in the phylogenetic group B2, despite the fact that this group was not the most prevalent. Recent reports suggest that the E. coli strains belonging to the B2 phylogroup produce multiple VFs, which influences their pathogenic potential [16,56,81]. In our research 6 VFs (vat, tosA, tosB, hly, usp and sfa) out of 28 VFs have only been detected in phylogroup B2. tosA and tosB genes are considered as markers for the presence of other virulence genes, whereas, usp, hlyA, sfa genes are often detected in the UPEC strains responsible for UTI [16,82]. Taking into account the division into commensal and pathogenic strains it was observed that the fecA gene encoding the outer membrane receptor with siderophore uptake transmembrane transporter activity was significantly associated with the commensal group of isolates (A, B1). Three other genes-kspMTII, irp2 and fyuA were mainly associated with the pathogenic group of isolates (B2, D). irp2 and fyuA genes are also found in avian pathogenic Escherichia coli (APEC). A third gene-kspMTII, belongs to a group of genes responsible for the synthesis of capsules. Gene coding capsules are most frequently found in isolates from community environments compared to hospital samples [83]. The relationship between resistance and virulence traits is mainly observed among clinical strains in the hospital environment [81,84,85]. In our research, an intriguing phenomenon is that E. coli strains producing ESBL/AmpC showing a high virulence profile, were also present in the natural habitat of birds. The dissemination of these strains via birds may also suggest an increased contamination of the human environment.

Assessment of genetic diversity of E. coli strains isolated from the GI-tract by genotyping
The PCR MP technique is widely used in various aspects of epidemiological study due to its high discriminatory power [56,[86][87][88]. We decided to use this method for studying the diversity of ESBL/AmpC strains in the environment. We demonstrated the genetic heterogeneity and diverse origins of E. coli strains (24 genotypes with subtypes). ESBL/AmpC positive strains of the same genotype successfully colonized the birds of different species. Migrating birds can transfer them to other areas (and in turn to birds), so they are a transmission vector of ESBL strains between different water bodies. In our study two E. coli isolates collected from a Eurasian Coot, demonstrating phenotypic diversity in colony morphology and size, had the same genotypic profile. Interestingly, six weeks later, the same individual was caught again in the same area and once more two E. coli colonies, with distinct coloration and morphology, exhibiting an identical PCR-MP profile, were recovered. In both cases isolates were confirmed as positive for ESBL production but with a different genotypic profile. Repeated findings of ESBL/AmpC-positive isolates over time indicate that these strains are resident and circulate abundantly in the environment. We also found that two E. coli isolates recovered from two Mallards captured in different locations exhibited an identical PCR MP profile. This identification of a common clone at two geographically distant centres indicated the possibility of transmission of ESBL/AmpC producing E. coli in the absence of antibiotic selection. The residence time of resistant bacteria in the digestive tract of wild birds remains unclear.

Conclusion
Free-living animals may serve as a reservoir and source of antibiotic-resistant bacteria, however, they have been studied much less frequently than farm animals. Our work revealed that a broad range of avian species were identified as highly prevalent carriers of multi-resistant E. coli. Our research indicates that not only the B2/D phylogenetic group but also the commensal A1 group is equipped with virulence genes and fitness factors. The presence of VFs along with the cephalosporin resistance AmpC gene and genes encoding ESBL may indicate the appearance of so-called high-risk strains in the animal environment. The spread between bacteria of antibiotic resistance genes is associated with the presence of mobile genetic elements that enable HGT among bacteria. It leads to the genetic variation of microorganisms, playing a major role in the evolution of bacteria. Understanding the scale of occurrence of drug-resistance strains in the environment, and in particular among birds, will allow for the assessment of the transmission of hospital strains into the environment and provide an indication of the risk of acquiring these strains by people from particularly exposed groups. Depending on resistance profiles and the prevalence of virulence factors, the E. coli strains of our study could have a considerable potential to cause human disease.
Supporting information S1