Impressive pan-genomic diversity of E. coli from a wild animal community near urban development reflects human impacts

Summary Human and domesticated animal waste infiltrates global freshwater, terrestrial, and marine environments, widely disseminating fecal microbes, antibiotics, and other chemical pollutants. Emerging evidence suggests that guts of wild animals are being invaded by our microbes, including Escherichia coli, which face anthropogenic selective pressures to gain antimicrobial resistance (AMR) and increase virulence. However, wild animal sources remain starkly under-represented among genomic sequence repositories. We sequenced whole genomes of 145 E. coli isolates from 55 wild and 13 domestic animal fecal samples, averaging 2 (ranging 1–7) isolates per sample, on a preserve imbedded in a human-dominated landscape in California Bay Area, USA, to assess AMR, virulence, and pan-genomic diversity. With single nucleotide polymorphism analyses we predict potential transmission routes. We illustrate the usefulness of E. coli to aid our understanding of and ability to surveil the emergence of zoonotic pathogens created by the mixing of human and wild bacteria in the environment.


INTRODUCTION Major gaps in wild animal studies
Despite its fame (or infamy), our understanding of the life of Escherichia coli that exists outside of human and domesticated animal guts remains in its infancy. 1 Of the few investigations of E. coli in wild animals, surprising diversity has been revealed, 2 increasingly with genes conferring clinically important antimicrobial resistance (AMR). 3While E. coli diversity is under-sampled in wild animals, 1 it has also been obscured by selective plating methods, where stool samples are directly subjected to antibiotics before typing, thereby eliminating substantial non-resistant diversity, or even potentially creating resistances during the laboratory methods. 4Shallow sampling methods within and between individuals are also to blame, as the recurrent presumption that a single isolate per host is representative of the within-host E. coli diversity overlooks substantial within-individual and within-host species diversity. 5The lack of genomic studies on E. coli from wild animals is critical, especially as wild animals may serve as reservoirs of harmful bacteria or even ''melting pots'' that could facilitate novel genetic combinations of virulence and AMR genes that could threaten global public health.

Problems with pathogenic E. coli
Two major categories of pathogenesis have been defined in E. coli: intra-intestinal pathogenic E. coli (InPEC) or diarrheagenic E. coli, which include the well-known Shiga-toxigenic E. coli (STEC); and extraintestinal pathogenic E. coli (ExPEC), which causes infections outside of the gut and is responsible for the death of over 2 million people per year. 6ExPEC strains include uropathogenic E. coli (UPEC), the leading cause of urinary tract infections (UTIs), and meningitis-associated E. coli (NMEC), which can cause bloodstream infections and is the most common cause of neonatal meningitis. 7Just 7 foodborne pathogens, one of which is E. coli, have been estimated to cost the United States anywhere between US$ 6.5 billion and US$ 35 billion annually. 8Though there is no evidence that wild animals are a significant or repeated source of food contamination by pathogenic E. coli, there are occasional reports of wild animals contaminating people or their food sources. 9,10In response to fear, farmers often resort to environmentally detrimental methods of wildlife exclusion, including trap-and-kill and removal of non-crop vegetation surrounding croplands. 11It is exceptionally challenging to source trace when outbreaks do occur, in part due to limitations of the E. coli genome database, but also because E. coli is notoriously genetically diverse.From genomic studies on isolates from humans, the core genome, or those genes found in every isolate of the species E. coli, has been estimated to be just 6% of the pan-genome, i.e.,

Research objectives
We investigated E. coli genomes from wild and domestic animals in an interacting community imbedded in an urban landscape to (1) capture the genetic and functional diversity of E. coli both among the community and within an individual, (2) assess strain-sharing between hosts to predict how E. coli might move through this network, (3) evaluate the prevalence of pathogenic (to humans) E. coli, AMR E. coli, and humanassociated sequence types (STs) in domestic and wild animals, and (4) consider the potential for the guts of wild animals to act as melting pots of novel genetic combinations that could threaten public health.We hypothesize that the genetic repertoire of wild animal E. coli will reflect the presence of domestic animals, as well as historical and present human impacts on the preserve and its waterways, and that wild animal species are differentially impacted depending on their ecology and life-history attributes.

Phylogenetic and pan-genomic analyses
Eight major phylogroups of E. coli have been described, each associated with distinct characteristics and ecological niches. 37The original Jasper Ridge Biological Preserve (JRBP) E. coli collection encompassed all major phylogroups, determined by a PCR-based phylogrouping method, 5 however, we did not successfully sequence an isolate belonging to phylogroup C.After removing 2 genomes that did not pass quality control, the resulting phylogeny of 143 de novo constructed E. coli genomes from wild and domestic animals at JRBP showed a broad distribution of genotypes across host status, AMR, and pathotype categories, and mostly agreed with the mono-phyletic groupings, with a few exceptions (Figure 1).Some mixing was observed between phylogroups D and E, suggesting potential mutation in the phylogroup-specific gene primers used in ClermonTyper to assign strains to phylogroups based on the differential presence of a few select genes.Three isolates of phylogroup A, all from horses, did not cluster with the rest of phylogroup A, potentially for the same reason.
Pan-genome analysis of 143 E. coli whole genomes from JRBP identified 32,028 gene families (over 7 times the average genome size), only 8% of which were shared among > 95% of the isolates.Just 17% of those were shared between > 15% isolates (Figure 2A).In total, 53 fimH types, 92 serotypes (O:H antigens), and 88 sequence types (STs; assigned by Achtman 7 gene MLST) were represented.Nine of these STs were assigned sequentially in EnteroBase (ST13010 through ST13018), indicating that these were novel to the database.
Principal component analysis (PCA) of the count matrix combining all PROKKA assignments (to a gene name, COG, or EC number) resulted in a PC1 axis that explained 24.5% of the total variance and phylogroup B2 being the most distinctly separated from other phylogroups (Figure 2B).Phylogroups A and B1 primarily separate in PC2, suggesting a more similar gene repertoire, while phylogroup E nests within phylogroup D. Interestingly, Escherichia isolates belonging to the cryptic clades separate clearly from the phylogroups, suggesting that they harbor a unique gene repertoire that distinguishes them, thus supporting their distinction from E. coli sensu stricto (Figure S1).Functional analyses were severely limited as 46% of genes in the pan-genome were unassigned to a functional COG (cluster of orthologous genes) category (Figure S2).

Within-scat diversity supports melting pot hypothesis
We sampled more than 1 E. coli isolate (averaging 3) from the same scat for 41 different individual samples, capturing substantial within-host genetic diversity.Though we did not sample each scat to saturation at the phylogroup or ST level, which would have involved constructing rarefaction curves and sampling additional isolates until the discovery of new diversity was exhausted, we still found that 28 of those individual scats sampled to n > 1 isolates carried > 1 phylogroup.Moreover, 29 scats carried > 1 ST, 13 of which also carried different STs of the same phylogroup (Table S1).Notably, 4 different STs were sequenced from each of a single opossum, puma, turkey, coyote, deer, and horse scat, as well as the goat community.This suggests that with-host diversity could be far greater, as these results were limited by the number of isolates sequenced per scat sample.

Possible recent transmission events
Both our sampling methods and the proximity of animals in this community allowed us to capture 62 potential clonal relationships between pairs of isolates as defined by the observation of % 100 single nucleotide polymorphisms (SNP) between genomes.Thirty-six of these were pairs of isolates taken from the same scat sample and 7 pairs were derived from different scat samples but from the same host species.Due to Figure 1.Genetic and ecological structure of E. coli from JRBP RECOPHY phylogeny of 143 de novo constructed E. coli genomes from animals at JRBP and including reference genomes MG1655 and O157:H7, rooted by E. fergusonii.Colored strips from inside out indicate host status, E. coli phylogroup, AMR, and pathotype (some isolates met criteria for multiple pathotype assignments).Graphic created in iTOL v5. the opportunistic nature of our sample collections, we were unable to confirm whether these samples came from the same or different individual(s).So, this could either indicate clonal relationships of E. coli occurring between different individuals of the same host species, or persistence of a clone within the gut of an individual for at least the amount of time that occurred between sample collections.However, 19 potential clonal pairs were derived from scat samples of different host species, so for these, we were certain they came from different individuals.Fourteen of those 19 pairs had % 10 SNP differences between them, a conservative measure to indicate a probable strain-sharing event.These 14 A B  clonal pairs represented 6 STs shared between 13 individuals.ST10 clones were isolated from each of a puma, a bobcat, and a coyote, all 3 scat samples from which were collected on the same day (Figure 3).A potential clonal pair of ST1304 isolates from a vole and a weasel were also collected on the same day.Notably, 2 potential clonal relationships were of ExPEC strains, and 1 crossed the wildland-urban interface (WUI).Table S2 summarizes the isolate metadata, SNP rate, and SNP count for all 62 pairs of strains identified with potential clonal relationships.

Virulence factor profiles
The number of distinct virulence factors (VFs) per genome in the JRBP E. coli isolates ranged from 11 to 46 (Table S3), meaning zero isolates qualified as ''low pathogenicity'' by our criteria (Table S4).The prevalence of ExPEC strains was 41.3%, and of InPEC strains, 4.9% (Table S5).No isolates were positive for the gene bfpA (bundle-forming pilus), so those containing the gene coding for the enteropathogenic E. coliassociated intimin protein, eae, were considered atypical; aEPEC. 38Of the wild hosts, at least one pathogen was obtained from 65.3% of individuals, as well as the squirrel midden and bat roost.Of the domestic hosts, a pathogen was found in 18.2% of individuals (just 2 horses), and the goat herd.
All phylogroup B2 isolates qualified as pathogens (ExPEC or InPEC) by our classification criteria, followed by phylogroup G (66.7%), phylogroup D (64.3%), and the cryptic clades (50%), though the sample sizes for phylogroup G and the cryptic clades were low (Table S5).Conversely, no ExPEC or InPEC isolates belonged to phylogroups E or F, and phylogroups A and B1 had less than 10% pathogen prevalence (Table S5).Principal component analyse (PCA) on the count matrix of VFs per genome showed phylogroup B2 separating from all other phylogroups with most top PC loadings associated with the direction of the phylogroup B2 cluster.However, lpfA (long polar fimbrial subunit A), associated with mobility, was identified in the top loadings in the direction of all other phylogroups (Figure S3).

Antimicrobial resistance gene profiles
In total, 16.1% of all isolates carried at least one antimicrobial resistance gene (ARG); 70.6% of those were multi-drug resistant (MDR), defined by the presence of > 2 ARGs.These isolates derived from 17 scat samples of 7 host species: bobcat (57.1% of scats tested), coyote (60.0%), gray fox (16.7%), opossum (100%), turkey (25.0%), horse (42.9%), and the goat herd.This represents 23.6% of all wild hosts and 33.3% of all domestic hosts, or 28.6% of domestic-sourced isolates and 13% of wild-sourced isolates.Remarkably, all AMR isolates from both horses and coyotes were MDR.Phylogroup G (66.7%), followed by B1 (34.2%), contained the most AMR isolates (Table S5).Collection sites of scat samples with AMR were concentrated on the horse trail along the northern border of the preserve and on the most heavily foot-trafficked trail encircling the reservoir (Figure S4).
Of 72 total ARGs identified among JRBP E. coli isolates, 20 were distinct, with 9 exclusively found in wild animals, 4 exclusively in domestic animals, and 7 (35%) shared between domestic and wild animals (Figure 4A).Furthermore, 84% of ARGs were plasmid-based, representing 6 classes of commonly prescribed antibiotics (Figure 4B).Fosfomycin resistance was the only ARG found solely in the chromosome.The plasmid replicon types of the majority of plasmid-based ARGs were unknown to the database (n = 30), but plasmid IncQ1 was associated with 12 ARGs and plasmid rep_cluster_2335 was associated with 18 ARGs (Table S6).The latter assignment to a non-descript replicon type ("rep cluster") indicates that the plasmid has been identified previously, but is poorly researched to-date.

Prevalence of human-associated sequence types
A handful of STs belonging to phylogroups B2 and D are responsible for the majority of ExPEC infections and are largely human-associated.In phylogroup B2, these are ST73, ST95, and ST131, and in phylogroup D, ST69. 39ST131 is the most prevalent ExPEC clone, mainly colonizing humans and human-associated animals, 40 and was identified in the bat roost near JRBP.This isolate belonged to clade B (O25:H4, fimH22) and classified as a UPEC.PlasmidFinder suggested that it may carry up to 3 plasmids (IncFII, IncXI, and pEC4115), but no ARGs were identified, suggesting that it is related to, but not the same MDR clone that has gained much recognition for its rapid global dissemination in the past decade. 41ST73, isolated here from a domestic horse, has previously been identified in companion animals 39 and has been shown to be shared between canine and human members of a household experiencing UTIs. 42A predominant cause of trimethoprim-sulfamethoxazole-resistant UTIs across the United States, 43 E. coli ST69 was isolated once from each of 3 different host species: puma, coyote, and opossum.The ST69 isolate from a coyote also carried 7 ARGs.Other important human-associated STs identified in wild animals at the preserve include ST127, isolated from an opossum, which is over-represented in human pneumonia isolates 44 and has been described as ''a recently emerged global pathogen,'' 45 ST117, isolated from 2 coyotes, one which also a UPEC with 3 ARGs, and ST155 46 also from a coyote and carrying 3 ARGs.

Diversity within ST10
The most common ST at JRBP was ST10, comprising 11.2% of all isolates.Substantial genetic diversity was observed within these 16 isolates, encompassing a pan-genome themselves of 7,080 gene families, just 51.5% shared among > 95% of isolates, representing 10 distinct serotypes and 5 different fimH types.This is evidence corroborating previous findings that ST10 has huge antigenic diversity and greater genetic diversity overall than other STs assessed. 47ST10 was present in all domestic host species, but in only 3 of 14 wild species (puma, bobcat, and coyote).PCA on the count matrix of genes in only the ST10 isolates showed substantial overlap in the gene repertoires of isolates from wild and domestic animals, however greater functional gene diversity was encompassed by those from wild animals.This could suggest potential strain-sharing between domestic and wild animals on the preserve (Figure S5).Including those listed above, human-associated STs composed 21.7% of all JRBP isolates, but their prevalence was higher among domestic hosts (28.6%) than wild hosts (20.0%).Controlling for some individual hosts contributing multiple isolates, 50% of domestic hosts and 36.4% of wild hosts carried at least 1 of these human-associated STs.The phylogenetic relatedness, phylogroup, host source, and major VF and ARG repertoires of these human-associated STs are summarized in Figure 5.

DISCUSSION
To what extent are wild melting pots a legacy of human impact?
Among the JRBP animal community, host-associated E. coli carried a vast repertoire of virulence-associated genes; nearly half were known pathogens to humans.Additionally, numerous STs with strong human associations were present.These reached higher prevalence in the domestic animals, indicating their guts may experience higher levels of human impact, but wild animals were not unexposed.Genes encoding clinically relevant AMR were also present in both wild and domestic-sourced isolates, but resistance was higher among domestic samples.Approximately one-third of distinct ARGs were shared between wild and domestic animals, suggesting that domestic animals may provide a source of either continuous selective pressure (via antibiotics in their excrement) or inoculation (by ARG-containing bacteria), thus contributing to the detection of these ARGs in wildlife.Most of the geographic locations of the scats carrying ARGs were concentrated along the horse trail on the northern border of the preserve and on the eastern side where the cattle and goats were present (Figure S4).Other scats with high ARG prevalence were picked up along the most heavily trafficked foot trail circling Searsville Reservoir.The streams flowing into the preserve, the reservoir, and its drainage, could all be sources of ARGs, as these waters travel through agricultural areas and residential areas that use septic tanks, potentially acquiring various forms of environmental pollution such as agrichemicals and human and domestic animal waste before entering the preserve.The animals on the preserve appear to be differentially impacted by such environmental pollution, as no ARGs were identified in any deer or puma, despite some of the largest scat sample sizes from these species.And too, no resistance was detected in any small mammals, though these sample sizes were smaller.Conversely, 6 of 10 coyotes carried MDR E. coli.One probable explanation for this observation is their tendency for coprophagy, especially as most of these scat samples were collected along the horse trail, and 3 of 7 horse scat samples also contained MDR E. coli.It could also be explained by the higher propensity for coyotes to enter residential areas, increasing their contact with the human environment.
Our results indicate that substantial genetic diversity is encompassed by wild animal-derived E. coli, even within a small geographic area.The fact that we identified 14 potential clonal pairs shared between individuals of this community, amid such diversity illuminates the role of transmission and rapid rate at which it is likely occurring within this system.Indeed, 2 of those were ExPEC strains shared between predator and prey and 1 crossed the WUI.In addition to identifying potential transmission events within this community, we also captured considerable within-scat diversity, indicating that within-host E. coli genetic diversity remains underappreciated in wild animals, but reinforcing the likelihood that the vertebrate gut may serve as a melting pot of novel genetic combinations that have the potential to heighten the risk of zoonotic spillback of novel pathogens.

Gauging the implications: A grain of salt?
It is important to note that resistances detected in our study tended to be to primarily older antibiotic agents.Previous studies suggest that E. coli of animal origin frequently show resistances to older antimicrobial agents, including tetracyclines, phenicols, sulfonamides, trimethoprim, and fosfomycin, attributed to the longer time frame these drugs have had to reach higher prevalence in the environment. 48Natural microbial communities carry a diversity of genes conferring AMR that also serve other such functions as efflux, cell-to-cell signaling, biocontrol of symbionts, or resistances to other environmental toxins, like heavy metals; known as the ''intrinsic resistome''. 34Indeed, there is natural variation in the intrinsic AMR levels among E. coli strains, and some level of resistance occurs even in the absence of known history of antibiotic exposure. 49However, high levels of resistance frequently and rapidly evolve in bacterial communities exposed to antibiotics, either by mutation or HGT of ARGs.Because resistance usually comes at a metabolic cost (strains that carry them typically have reduced fitness), it is usually purged in the absence of continued antibiotic selective pressure and/or persistent introduction of the resistance genes to the bacterial environment. 49Criteria for ARGs posing ''high risk'' to public health include gene mobility, host pathogenicity, and the enrichment of specific resistances in anthropogenically impacted environments compared to non/low-human-impacted areas. 50For instance, chromosomal mutations leading to intrinsic resistance to colistin (a last-resort antibiotic) were documented decades ago, but the low HGT potential for those genes limited their ability to spread and thus had minimal clinical impact. 51However, the plasmid-mediated and thus highly mobile colistin resistance gene, mcr-1, has already spread to 7 different pathogenic species in 31 countries 52 and occurs at reportedly high prevalence in human and livestock feces. 53,54athogenesis in E. coli might be a by-product of its genomic plasticity, thus enabling adaptation to a broad range of environments.Dubbed ''the coincidental hypothesis for VFs'', it suggests that ''virulence genes'' evolved and are selectively maintained to serve other purposes in the bacterium's ecology, especially in commensalism, as many ''virulence factors'' play roles in colonization of the gut and defense against other intestinal organisms. 37However, of over 1,400 infectious organisms identified in a comprehensive literature review, an alarming 75% were of zoonotic origin, 55 underscoring the value of monitoring wild-animal carriage of human pathogens in order to protect public health.Such methodology is advocated by the ''One Health'' approach, which recognizes that human health is tightly linked to the health of animals and ecosystems. 56An extensive 2018 literature review on the occurrence of STEC in wild animals identified just 79 studies todate, only 10% of which investigated the prevalence of STEC in wild animals in conjunction with livestock and humans, leading to the conclusion that the research had majorly overlooked the ''One Health'' approach, thus undercutting our potential to understand transmission routes between these spheres. 57

Limitations of the study
It has been reported that humans host an average of 3.5 E. coli phylogroups and likely many more transiently over their lifetime. 58,59However, such an average has yet to be estimated for any non-human host.Here, even within a single phylogroup from a single scat, we sequenced multiple STs, indicating that substantial diversity exists below the phylogroup level within a single gut environment.It follows that we have also yet to establish an ideal sample size to capture the within-host diversity of E. coli.Even results from such studies in humans have varied in the proposed number of isolates required to exhaust new strain discovery and to what level of probability that sample size has of obtaining minor strains. 60In a previous study, we estimated the total phylogroup diversity at the host species level and found that it positively corresponded with host body mass and is likely also influenced by other factors such as host diet and proximity to human impacts. 5This suggests that the ''optimal'' sample size to obtain the within-host diversity of E. coli will likely vary greatly across the animal kingdom.
Though our sample size of 143 genomes was relatively small compared to some previous studies such as one on 1,294 E coli isolates from humans, poultry, wild animals, and water on the Australian continent, our PCAs on the pan-genome closely resembled one another, with the major phylogroups clustering similarly and their PC1 axis accounting for 23.6% of the variance compared to 24.5% in our analysis. 61This could suggest that we obtained a good representation of the breadth of functional diversity documented in E. coli within just 68 animal hosts living in a small nature preserve.However, it could also indicate an underlying shortcoming of the reference database, as a key limitation of studies like these lies in the lack of description of the functional roles corresponding to the genetic diversity within bacterial genomes.Even the well-known laboratory model organism, E. coli K-12, still has 35% of its genes lacking experimental evidence of function and another 5% only known as pseudo-or phantom genes. 62This dramatic gap in the functional database was recapitulated here, as a total of 46% of the pan-genome of JRBP E. coli was undescribed in the COG database.Determining the function of a gene is highly involved and requires substantial time and laboratory work, 63 but the fact that we have yet to understand the functional and metabolic roles of nearly half of the genes in a well-studied laboratory strain hinders attempts to unravel the eco-evolutionary dynamics of E. coli in diverse ecosystems.For example, though pipelines to predict AMR via genetic analysis, like ResFinder, are typically highly accurate, 64 genotypic and phenotypic observations do not always agree.Does observing a resistant phenotype on antibiotic-containing agar represent what would happen in the gut, or could the context of the environment dictate gene expression?Deciphering the functional role of genes would also retroactively inform current WGS studies that have generated ample genetic data, but which are not as informative as they could be considering the gaps in gene-function databases.
Furthermore, much remains unknown surrounding the ability of pathogenic E. coli strains that make humans sick to also sicken wild animals.This is partially due to the deficit of studies in wild animals, as well as the challenges of determining a host's well-being at the time of collection, especially if collections are taken opportunistically.Avian pathogenic E. coli (APEC) is a well-known pathogen of birds, 65 but has likely received more attention due to its major impacts on the poultry industry.Future research should endeavor to address the limitations of opportunistic sampling by adopting methods of catch-and-release and tracking individuals over time to obtain important health and lifestyle metrics.We should also strive to sample a greater diversity of hosts and individuals from each host species to paint a more thorough picture of the ecology and movement of E. coli within a wild animal network.Such work should simultaneously aim to expand the database of E. coli genomes from a larger diversity of wild hosts, especially in the vicinity of agricultural lands.Such improvements to monitoring wildlife gut microbial communities will support source tracking efforts during outbreaks, enhancing our ability to prevent contamination, thus avoiding economic losses from crop contamination and expenses associated with medical treatment.This will also inform the propensity for wild animals to spread pathogens, as highly symptomatic individuals would shed more pathogens into the environment, and those carrying a greater diversity of strains run a higher risk of facilitating novel pathogen emergence.It may also assist in removing false blame placed on wild animals, thus bolstering wildlife conservation efforts.

Conclusions
The prevalence of pathogenic and AMR E. coli in the JRBP animal community suggests that they experience direct or indirect anthropogenic impacts, and the similarity of some of these strains to one another across host species suggests transmission may be happening in situ.The occurrence of human-associated STs indicates that the surrounding human population and its legacy of impact have influenced the gut microbiomes of wild animals.We know home ranges of many species studied here extend beyond the bounds of ''protection'' into heavily populated areas.Human-wildlife conflict does occur here, for example, puma have reportedly preyed on domestic dogs and encounters between coyotes and residents in the area are common.Although there are no sewage treatment facilities in the immediate vicinity, the large number of septic tanks and leach fields, horse paddocks and chicken coops, may impact wildlife on the preserve.JRBP is probably unexceptional in this, as what was found here likely represents to varying degrees the impact of human populations around the globe on wild animals and their microbiomes.We highlight how harmful E. coli could be a result of our actions (or inactions) that introduce our waste, antibiotics, and AMR bacteria into the environment, where the microbes may then further evolve within the guts of wild animals.Thus, we recommend several interventions that seek to decrease the distribution and rapid evolution of harmful E. coli.These include drastically reigning in antibiotic use, properly disposing of antibiotics and other human waste to decrease our impact on the environment, improving water sanitation methods to better remove drugs, ensuring that compost reaches and maintains proper temperature to kill pathogenic E. coli, 66 increasing protected spaces for wild animals to exist away from humans, and constructing wildlife corridors to reduce human-wildlife encounters.
We promote using E. coli as a model for understanding bacterial ecology and evolution in wild animal populations.With genomic data, we can assess the rate of carriage and propagation of AMR and pathogenic strains in the wild, evaluate the risk for zoonotic disease emergence, and potentially predict spillover and spillback events.All of this has strong implications for global public health, especially in the Anthropocene.Life on the planet is now enmeshed, highly interconnected, and faced with unprecedented environmental change.Surging human-wildlife interactions, alarming rise of AMR globally, and increased pollution are exerting strong selective pressures on bacterial communities to develop and maintain resistance mechanisms to heavy metals, agrichemicals, and antibiotics.These genes are incorporating into bacterial genomes and rapidly disseminating in the environment with poorly described and likely detrimental effects on human, animal, and ecosystem health.

STAR+METHODS
Detailed methods are provided in the online version of this paper and include the following:  81 Mutation rates in E. coli have been reported from 0.2 3 10 À10 to 5 3 10 À10 nucleotides/generation. [94][95][96] Its high mutation rate suggests that there is an extremely small likelihood of near-perfect genome sharing unless derived from an identical source.Though clonal stability within E. coli STs has only been described to a limited extent, 47 the criteria we used to identify potential strain-sharing events of % 10 SNPs is comparable to the 9 SNP differences observed in a recurrent outbreak of ST10 clones in a broiler unit 97 and the % 6 SNP differences observed in within-host intra-clonal E. coli isolates from human UTIs and feces. 98We also report the cut-off of % 100 SNPs often seen in the literature. 99

Figure 2 .
Figure 2. Gene and functional diversity of E. coli from JRBP (A) Percentages of gene families out of the total pan-genome belonging to the core and accessory (shell and cloud) genomes.(B) E. coli phylogroup functional similarity using principal component analysis (PCA) of 135 E. coli genomes (excluding cryptic clade isolates; n = 8).Each dot corresponds to a genome in the first 2 principal components (PC).Vectors show the top 5 annotations with the largest PC1 and PC2 values in both positive and negative directions.

Figure 3 .
Figure 3. Possible recent transmission at JRBP Red arrows indicate potential clonal relationships identified among E. coli from different host species as defined by % 10 SNP differences between pairs of isolates.The sequence types (STs) of those isolates are indicated.Gray arrows suggest that transmission is occurring throughout the JRBP ecosystem and across the WUI.# Same collection day; *Likely a novel ST; y ExPEC.

Figure 4 .
Figure 4. Ecological and genomic contexts of AMR at JRBP (A) Of 20 unique ARGs identified by ResFinder among E. coli genomes from JRBP, 9 were unique to isolates from wild animals, 4 were unique to isolates from domestic animals, and 7 were shared between wild and domestic animals; n = number of genomes per group.(B) Sankey diagram detailing the genomic locations (plasmid vs. chromosome) of all 72 ARGs identified across 143 genomes linked to their predicted resistance phenotype and when plasmid-based, their plasmid replicon type.

Figure 5 .
Figure 5. Phylogenetic and metadata analysis of human-associated STs at JRBP A maximum-likelihood tree was based on RAxML-NG under GTR+GAMMA with default parameters on the core gene alignment (Roary) of 31 human-associated STs.Presence or absence of the 5 ExPEC-associated VFs, 4 UPEC-associated VFs, and the 38 other major VFs (gray) and ARGs (purple) are detailed for each isolate.Domestic hosts are shown in bold.

TABLE
METHOD DETAILSPotential clonal relationships were defined as pairs of genomes with both % 10 clonal SNPs and % 10 recombined SNPs in the mix model matrix output by RECOPHY.
d RESOURCE AVAILABILITY B Lead contact B Materials availability B Data and code availability d