Occurrence of Antimicrobial-Resistant Escherichia coli in Marine Mammals of the North and Baltic Seas: Sentinels for Human Health

Antimicrobial resistance is a global health threat that involves complex, opaque transmission processes in the environment. In particular, wildlife appears to function as a reservoir and vector for antimicrobial-resistant bacteria as well as resistance genes. In the present study, the occurrence of antimicrobial-resistant Escherichia coli was determined in marine mammals and various fish species of the North and Baltic Seas. Rectal or faecal swabs were collected from 66 live-caught or stranded marine mammals and 40 fish specimens. The antimicrobial resistance phenotypes and genotypes of isolated E. coli were determined using disk diffusion tests and PCR assays. Furthermore, isolates were assigned to the four major phylogenetic groups of E. coli. Additionally, post mortem examinations were performed on 41 of the sampled marine mammals. The investigations revealed resistant E. coli in 39.4% of the marine mammal samples, while no resistant isolates were obtained from any of the fish samples. The obtained isolates most frequently exhibited resistance against aminoglycosides, followed by β-lactams. Of the isolates, 37.2% showed multidrug resistance. Harbour porpoises (Phocoena phocoena) mainly carried E. coli isolates belonging to the phylogenetic group B1, while seal isolates were most frequently assigned to group B2. Regarding antimicrobial resistance, no significant differences were seen between the two sampling areas or different health parameters, but multidrug-resistant isolates were more frequent in harbour porpoises than in the sampled seals. The presented results provide information on the distribution of antimicrobial-resistant bacteria in the North and Baltic Seas, and highlight the role of these resident marine mammal species as sentinels from a One Health perspective.


Introduction
Antimicrobial-resistant bacteria increasingly pose a global challenge for the successful treatment of bacterial infections in both humans and animals [1][2][3]. Besides the development of new antimicrobial drugs [4], existing antimicrobials need to be used appropri-

Study Area
The present study investigated samples of marine mammals and fishes of two different seas, the North Sea and the Baltic Sea. Both are connected via the Skagerrak, Kattegat, and Danish straits. While the North Sea is a marginal sea of the eastern Atlantic Ocean, the Baltic Sea is a brackish inland sea of the Atlantic. The North Sea has a high water exchange rate with the Atlantic Ocean, to which it is connected via the English Channel in the South and the Norwegian Sea in the North. At the German North Sea coast, there is a usual tidal range of around two to three meters. In comparison to this, the Baltic Sea has a low water exchange rate with the Atlantic Ocean, as it is almost completely surrounded by dry land, with only the above-mentioned natural connection via the North Sea to the Atlantic Ocean. The tidal range in the Western Baltic Sea is usually around 30 cm, and decreases to zero moving eastwards. Both Seas are densely populated by humans and are subject to highly diverse anthropogenic impacts.

Sample Collection
Samples of marine mammals were obtained either as swabs of fresh faeces from the ground or as rectal swabs of alive or dead animals (Sterile transport swabs, Heinz Herenz, Germany). Used swabs were directly stored in a transport gel medium. Faecal samples were taken at haul-out sites of seals by inserting the sterile, dry swab into the excrement and turning it a few times to gain enough material. Rectal swabs of live seals were taken during regular seal catches for health assessments, while two live harbour porpoises were accidentally caught in a fishing net and rectal swabs were obtained before release. Furthermore, rectal swabs were taken of stranded dead marine mammals that showed a decomposition status of three or better, with the exception of two animals with decomposition status four. Dead marine mammals were either stranded deceased or were mercy killed on the beach by authorized and specially trained members of the marine mammal stranding network of Schleswig-Holstein. Ringed seals in Sweden were legally hunted in accordance with the hunting quota by Swedish hunters. For rectal samples, sterile swabs were inserted into the rectum and subsequently turned several times on the mucosa. All marine mammal swabs were taken between 30 March 2017 and 8 June 2019.
Fish samples were partly taken during research surveys carried out by the Johann Heinrich von Thünen Institute in both the North and Baltic Seas (sample ID [74][75][76][77][78][79][80][81][82][83][84][85][86][100][101][102][103][104][105][106]. The sampling in the Baltic Sea was performed in March 2018 during the cruise Solea SB746, BITS (Baltic International Trawl Survey), the one in the North Sea took place in June 2019 during the cruise Solea SB764 HERAS (the 2019 ICES Coordinated Acoustic Survey in the Skagerrak and Kattegat, North Sea, West of Scotland and the Malin Shelf area). All other fishes were obtained from local fishermen or private anglers. Samples were gained as rectal swabs or, in the case of fish that were too small, as intestinal swabs. Fish species were identified on the basis of morphological parameters [66]. Fish samples were taken between 26 November 2017 and 15 July 2019.
A map created with ArcGIS (ESRI, Version 10.6.1) indicates all sample locations for the marine mammals as well as for the fishes (Figure 1). A more detailed map for the samples of the North and Baltic Sea coasts of Schleswig-Holstein can be found in the Supplementary Material (Supplementary Figure S1). Information on all samples, including species, sample area, sampling date and method, age class, and sex of sampled animals, as well as if a necropsy was performed, are listed in the Supplementary Materials (Supplementary Table S1).
Antibiotics 2022, 11, x FOR PEER REVIEW 4 of 23 stranding network of Schleswig-Holstein. Ringed seals in Sweden were legally hunted in accordance with the hunting quota by Swedish hunters. For rectal samples, sterile swabs were inserted into the rectum and subsequently turned several times on the mucosa. All marine mammal swabs were taken between 30 March 2017 and 8 June 2019. Fish samples were partly taken during research surveys carried out by the Johann Heinrich von Thünen Institute in both the North and Baltic Seas (sample ID 74-86, [100][101][102][103][104][105][106]. The sampling in the Baltic Sea was performed in March 2018 during the cruise Solea SB746, BITS (Baltic International Trawl Survey), the one in the North Sea took place in June 2019 during the cruise Solea SB764 HERAS (the 2019 ICES Coordinated Acoustic Survey in the Skagerrak and Kattegat, North Sea, West of Scotland and the Malin Shelf area). All other fishes were obtained from local fishermen or private anglers. Samples were gained as rectal swabs or, in the case of fish that were too small, as intestinal swabs. Fish species were identified on the basis of morphological parameters [66]. Fish samples were taken between 26 November 2017 and 15 July 2019.
A map created with ArcGIS (ESRI, Version 10.6.1) indicates all sample locations for the marine mammals as well as for the fishes (Figure 1). A more detailed map for the samples of the North and Baltic Sea coasts of Schleswig-Holstein can be found in the supplementary material (Supplementary Figure S1). Information on all samples, including species, sample area, sampling date and method, age class, and sex of sampled animals, as well as if a necropsy was performed, are listed in the Supplementary Materials (Supplementary Table S1).

Necropsies
The stranding network of the German federal state of Schleswig-Holstein retrieves stranded marine mammals off the coast of Schleswig-Holstein, which then undergo post

Necropsies
The stranding network of the German federal state of Schleswig-Holstein retrieves stranded marine mammals off the coast of Schleswig-Holstein, which then undergo post mortem examinations at the Institute for Terrestrial and Aquatic Wildlife Research (ITAW), University of Veterinary Medicine Hannover, Foundation, Büsum, Germany [67][68][69]. In the present study, 41 of the 66 sampled marine mammals underwent post mortem examinations, namely nine harbour seals (six from the North Sea, three from the Baltic Sea), two grey seals (one from the North Sea, one from the Baltic Sea), 16 ringed seals (all from the Baltic Sea) and 14 harbour porpoises (two from the North Sea, 12 from the Baltic Sea). Of these 41 animals, 17 were stored frozen after sampling and prior to necropsy, all others were necropsied directly after recovery. The decomposition states varied from fresh, through good, to moderate, except of two animals with poor decomposition status (sample ID 1 and 148).
Pathological investigations were performed in accordance with a standardised protocol [70,71]. All carcasses underwent a full post mortem investigation. Harbour porpoises were classified into the following age classes depending on their total length: neonate (<1.0 m), juvenile (1.0-1.3 m) and adult (>1.3 m). Seals were classified in accordance with their habitus, their stranding date, and the known pupping season into the three classes neonate, juvenile and adult. The nutritional status of the carcasses was judged depending on the blubber thickness and muscle development [70]. Furthermore, samples were taken for histopathological, parasitological, and microbiological investigations. These samples were taken and investigated as previously described [70]. The extent of histopathological investigations depended on the decomposition status and macroscopical findings. Histopathological samples were obtained from at least the main organs (lung, liver, spleen, kidney, and intestine) in all but four carcasses (sample ID 1, 81, 83, and 85), and additionally from organs and tissues with morphological changes. They were fixed in 10% buffered formalin, and subsequently embedded in a paraffin wax, cut into 3 µm slices, and stained with haematoxylin and eosin. Parasites were classified on the basis of their morphological characteristics. From 38 carcasses, tissues submitted for microbiological testing included at least the lung, liver, spleen, kidney, and intestine in all but two animals (sample ID 1 and 81).

E. coli Isolation and Identification
Obtained swabs were stored at room temperature for between one week and almost nine months. After initial selective enrichment for Enterobacteriacae in Mossel bouillon, samples were subsequently transferred to selective plates containing antimicrobials as sample processing aimed directly on the isolation of only resistant E. coli. Mossel bouillon (Carl Roth GmbH & Co. KG, Karlsruhe, Germany) was blended, and Blood agar (Carl Roth GmbH & Co. KG) and Gassner agar (Sigma Aldrich Chemie GmbH, Steinheim am Albuch, Germany) plates were poured in accordance with the manufacturers' recommendations. Furthermore, MacConkey agar (Carl Roth GmbH & Co. KG) plates with the addition of eight different antimicrobials were poured in accordance with the manufacturer's information using the following antibiotics: ampicillin (30 mg/L; Carl Roth GmbH & Co. KG), cephalothin (30 mg/L; TCI Europe N.V., Zwijndrecht, Belgium), chloramphenicol (10 mg/L; Carl Roth GmbH & Co. KG), ciprofloxacin (1 mg/L; Alfa Aeser, Thermo Fischer Scientific, Waltham, MA, USA), colistin (2 mg/L; AppliChem GmbH, Darmstadt, Germany), gentamicin (10 mg/L; Carl Roth GmbH & Co. KG), sulfamethoxazole (512 mg/L; Sigma Aldrich Chemie GmbH), and tetracycline (15 mg/L; AppliChem GmbH). First, each swab was streaked on Blood and Gassner agar before being placed into Mossel bouillon. Agar plates and bouillon were incubated for 24 h at 37 • C. Here, detection of colonies on blood and Gassner agar plates, as well as blurring of the bouillon, were used as references for bacterial growth. Subsequently, 100 µL of the Mossel bouillon was streaked on each of the eight different antimicrobial agar plates, which were then incubated for 24 h at 37 • C. From the antimicrobial containing agar plates that showed bacterial growth, five colonies were sampled. In cases where less than five colonies grew on one plate, all colonies were sampled. The selected colonies were then suspended in LB medium (Carl Roth GmbH & Co. KG) and incubated for 24 h at 37 • C. After incubation, 870 µL of the LB medium were compounded with 130 µL glycerine, vortexed and stored at −80 • C. Colonies were selected based on their macroscopic appearance. On MacConkey agar, the typical colour of E. coli colonies (among others, e.g., Klebsiella and Enterobacter) is red or pink with a hazy medium surrounding the colony.
A preselection of potential E. coli was performed using selective plating on Chromocult (Merck KGaA, Darmstadt, Germany), Gassner and MacConkey agar. As E. coli ferments lactose, the typical appearance of its colonies is blue to violet on Chromocult, blue on Gassner, and red or pink on MacConkey agar. For genomic DNA isolation, presumptive E. coli isolates were cultured overnight, suspended in 300 µL bidistilled water, then heated at 99 • C for 15 min and centrifuged (13,000× g) for 2 min. Species confirmation of isolates was performed as previously described via gadA-PCR [72]. Isolates that did not react in this PCR-assay were analysed via MALDI-TOF for species identification. By this means, 116 E. coli isolates were confirmed. As each sample was streaked on eight different plates and up to five colonies per plate were collected, one sample delivered one to five isolates.
For each isolate, a bacterial suspension in NaCl with 0.5 on the MacFarland scale, corresponding to approximately 10 8 CFU/mL, was prepared and streaked on Mueller-Hinton (Carl Roth GmbH & Co. KG) agar plates. Subsequently, up to six antimicrobial disks were placed on the agar surface and incubated for 18 h at 35 • C ± 2 • C. Inhibition zones were evaluated in accordance with CLSI standards [74,75]. For subsequent analyses, isolates with intermediate susceptibility and resistant isolates were grouped together. Potential colistin-resistant isolates were further tested using MacConkey agar plates supplemented with colistin (2 mg/L), representing an additional antimicrobial class. Copy isolates were avoided by using only one of the isolates that were obtained from the same swab and showed the same resistance profile.

Molecular Typing
PCR assays were used to assign all E. coli isolates to one of the four major phylogenetic groups (A, B1, B2, and D) by targeting two genes (chuA and yjaA) [81] and the anonymous DNA fragment TSPE4.C2 [72]. Macrorestriction analyses with XbaI digestion according to the PulseNet protocol for Salmonella, Shigella, E. coli O157 and other Shiga toxin-producing E. coli [82] was used to investigate the genetic relatedness of the isolates. DNA fragments were separated by pulsed-field gel electrophoresis using a CHEF DR II system (BioRad, Hercules, CA, USA) with the following settings: 6 V, initial time 6.

Statistical Analysis
RStudio version 1.4.1103 (R version 4.1.2) with the packages readxl and DescTools was used to perform statistical analyses. The significance of the association between (i) the number of samples with resistant E. coli and (ii) the number of samples with multidrugresistant E. coli and the variables area (North and Baltic Sea) and marine mammal species was tested via Fisher's exact test because of the small sample size. Furthermore, for necropsied animals, the number of samples with resistant E. coli was tested against the variables overall health status, body condition, age, sex, sepsis, and (broncho)pneumonia. The level of significance was set at p < 0.05.

Occurrence of Antimicrobial-Resistant E. coli Isolates
For the present study, 66 samples were collected from marine mammals and 40 from different fish species. Of the marine mammal samples, 58 (87.9%) showed microbial growth on at least one of the antimicrobial-containing agar plates, as also observed for 16 (40%) of the fish samples. At least one antimicrobial-resistant E. coli isolate was obtained from 26 (39.4%) of the marine mammal samples, whereas the identity of the other collected isolates could not be confirmed as E. coli. No resistant E. coli was isolated from any of the fish samples. Resistant E. coli was obtained from 16 (44.4%) of the 36 marine mammal samples from the Baltic Sea as well as from 10 (33.3%) of the 30 marine mammal samples from the North Sea. Of the different species, the resistant isolates were most frequent in ringed seals, with 10 (all Baltic Sea) of 16 samples carrying resistant E. coli isolates. Nine (three Baltic Sea, seven North Sea) of 24 harbour seal samples, three (all North Sea) of 10 grey seal samples and four (three Baltic Sea, one North Sea) of 16 harbour porpoises' samples carried resistant E. coli isolates. The numbers of samples with resistant E. coli across all marine mammals as well as separated by species and seas are depicted in Figure 2. No significant differences were detected in the occurrence of resistant E. coli either between the two areas (p = 0.4504) or between the different species (p = 0.1602). Of the 116 E. coli isolates confirmed originally, 63 were identified as duplicates. Ten isolates did not show any resistance in the disk diffusion test. In total, 43 E. coli isolates with phenotypic (intermediate) resistances were obtained from the 66 marine mammal samples (Figure 3), while no isolate from the fish samples could be confirmed as E. coli. The 43 isolates were distributed among the species and sampling areas as follows: 10 isolates from four harbour porpoises (one isolate from the North Sea, nine from the Baltic Sea); 13 isolates from nine harbour seals (eight isolates from the North Sea, five from the Baltic Sea); three isolates from three grey seals (all from the North Sea); and 17 isolates from 10 ringed seals (all from the Baltic Sea). Of the 116 E. coli isolates confirmed originally, 63 were identified as duplicates. Ten isolates did not show any resistance in the disk diffusion test. In total, 43 E. coli isolates with phenotypic (intermediate) resistances were obtained from the 66 marine mammal samples (Figure 3), while no isolate from the fish samples could be confirmed as E. coli. The 43 isolates were distributed among the species and sampling areas as follows: 10 isolates from four harbour porpoises (one isolate from the North Sea, nine from the Baltic Sea); 13 isolates from nine harbour seals (eight isolates from the North Sea, five from the Baltic Sea); three isolates from three grey seals (all from the North Sea); and 17 isolates from 10 ringed seals (all from the Baltic Sea).
Resistance to at least one antimicrobial agent of three or more antimicrobial classes was defined as multidrug resistance and occurred in 37.2% (16/43) of the E. coli isolates. Of these, six were resistant to three classes, four to four classes, four to five classes, and two to six classes of the total of eight antimicrobial classes tested. Resistance to two antimicrobial classes occurred in 18.6% (8/43) of the isolates and resistance to one antimicrobial class in 44.2% (19/43). Comparing the two areas, the rate of multidrug resistance was higher in the isolates from the Baltic Sea (13/31, 41.9%) than in those from the North Sea (3/12, 25%). Interspecies differences were also observed. The highest rate of multidrug resistance was found in isolates carried by harbour porpoises (8/10, 80%), followed by those in harbour seals (5/13, 38.5%) and ringed seals (3/17, 17.6%). None of the three isolates from grey seals were multidrug-resistant. No statistically significant association was observed between the number of samples containing multidrug-resistant E. coli and the sampling locations (p = 0.3065). However, a statistically significant difference was found between marine mammal species regarding the number of multidrug-resistant E. coli isolates, with harbour porpoises (8/10) carrying a significantly higher number of multidrug-resistant isolates compared to the three seal species (8/33; p = 0.007231). A one-to-one comparison of the species revealed a significant difference between harbour porpoises and ringed seals (p = 0.03934), as well as between harbour porpoises and grey seals (p = 0.03497), but not between harbour porpoises and harbour seals (p = 0.0903). The number of multidrugresistant isolates is depicted in Figure 5 for the three species carrying them. Resistance to at least one antimicrobial agent of three or more antimicrobial classes was defined as multidrug resistance and occurred in 37.2% (16/43) of the E. coli isolates. Of these, six were resistant to three classes, four to four classes, four to five classes, and two to six classes of the total of eight antimicrobial classes tested. Resistance to two antimicrobial classes occurred in 18.6% (8/43) of the isolates and resistance to one antimicrobial class in 44.2% (19/43). Comparing the two areas, the rate of multidrug resistance was higher in the isolates from the Baltic Sea (13/31, 41.9%) than in those from the North Sea (3/12, 25%). Interspecies differences were also observed. The highest rate of multidrug resistance was found in isolates carried by harbour porpoises (8/10, 80%), followed by those in harbour seals (5/13, 38.5%) and ringed seals (3/17, 17.6%). None of the three isolates from grey seals were multidrug-resistant. No statistically significant association was observed between the number of samples containing multidrug-resistant E. coli and the sampling locations (p = 0.3065). However, a statistically significant difference was found between marine mammal species regarding the number of multidrug-resistant E. coli isolates, with harbour porpoises (8/10) carrying a significantly higher number of multidrugresistant isolates compared to the three seal species (8/33; p = 0.007231). A one-to-one comparison of the species revealed a significant difference between harbour porpoises and ringed seals (p = 0.03934), as well as between harbour porpoises and grey seals (p = 0.03497), but not between harbour porpoises and harbour seals (p = 0.0903). The number of multidrug-resistant isolates is depicted in Figure 5 for the three species carrying them. Depending on their phenotypic resistance, isolates were tested for a selection of corresponding resistance genes. By total numbers, the most commonly detected resistance Depending on their phenotypic resistance, isolates were tested for a selection of corresponding resistance genes. By total numbers, the most commonly detected resistance gene was bla TEM (n = 16), while bla OXA-1-like (n = 2) and bla SHV (n = 1) genes, also mediating resistance to non-extended spectrum β-lactams, occurred only rarely. The gene bla CMY-2 , which is indicative of AmpC β-lactamase production, was detected once. Higher numbers were found for genes mediating resistance to streptomycin (strA (n = 11), strB (n = 11), and aadA1 (n = 7)), for genes mediating resistance to sulfonamides (sul1 (n = 6) and sul2 (n = 10)) and tetracyclines (tet(A) (n = 7), tet(B) (n = 8), and tet(D) (n = 3)). More rarely found resistance genes included dfrA1 (n = 4) and dfrA5 (n = 3), mediating resistance to trimethoprim, floR (n = 3), catA1 (n = 2); catA2 (n = 1), mediating resistance to phenicols; and qnrS (n = 3), mediating resistance to quinolones.
Not all isolates with phenotypic or intermediate resistance carried one of the tested resistance genes. This was the case in 16/31 aminoglycoside-resistant isolates. For isolates showing phenotypic resistance to β-lactams, no resistance genes were detected in 7/26 isolates. Furthermore, 1/7 phenicol-resistant isolates, 1/13 sulfonamide-resistant isolates, and 7/10 quinolone-resistant isolates did not carry any of the corresponding tested genes. In all isolates resistant to trimethoprim (n = 7) and tetracycline (n = 14), a corresponding gene was detected. Fourteen of the isolates without gene detection (n = 32) showed only intermediate resistance. These included 10 isolates with resistance to aminoglycosides, three with resistance to β-lactams, and one with resistance to phenicols. The remaining 18 isolates without resistance gene detection showed full phenotypic resistance, including seven isolates with resistance to quinolones, six isolates with resistance to aminoglycosides, four isolates with resistance to β-lactams and one isolate with resistance to sulfonamides. The detected resistance genes and the number of genes per location are summarized in Table 1. For each isolate, the resistance genotypes are shown in Figure 3.

Molecular Typing
The determination of major phylogenetic groups assigned 17 of the isolates to phylogenetic group B2, 13 isolates to group B1, eight isolates to group A, and five isolates to group D. The phylogenetic groups of each isolate are listed in Figure 3. The results for all of the isolates, as well as those for the four marine mammal species, are summarized in Table 2. Comparing multidrug resistance and assignment to a phylogenetic group showed that 11 (68.8%) of the 16 multidrug-resistant isolates belonged to group B1, while groups A and D each included two (each 12.5%) and group B2 included one (6.8) multidrugresistant isolate. For the determination of genomic relatedness, macrorestriction analyses were performed and revealed band patterns in 42 isolates. The remaining isolate (27b) was found to be non-typeable by Xbal macrorestriction. One cluster of three isolates from harbour seals from the North Sea with ≥90% similarity was identified, including two isolates from the same sampling location and sampling day and one isolate from a different sampling location and a sampling date around 4 weeks later (isolates 9, 10 and 27a). Another cluster with ≥80% similarity included three isolates from the same harbour porpoise sample from the Baltic Sea (isolates 54a, 54b and 54d), with two isolates (54b and 54d) having identical band patterns. Additionally, two isolates from three other samples had identical band patterns (isolates 39b and 39d, isolates 48a and 48b, and isolates 49a and 49b). All the other isolates showed a high degree of heterogeneity. The dendrogram of the 42 isolates is depicted in Figure 3.

Health Status and Antimicrobial Resistance
The results of the performed necropsies, as well as the general data of the animals, were used to evaluate whether different parameters have an influence on the occurrence of antimicrobial-resistant E. coli. To achieve this, the overall health status was evaluated considering all available results, classifying the necropsied animals into three health categories: good, moderate and poor. Furthermore, the two most frequent causes of disease/mortality were determined as being (broncho)pneumonia and sepsis. Other parameters of interest included body condition as well as the sex and age class of the necropsied animals. All named parameters are listed in Table 3, along with the corresponding number of animals carrying antimicrobial-resistant E. coli. No statistically significant association was found between the carriage of antimicrobial-resistant isolates and the different variables (overall health status (p = 1), (broncho)pneumonia (p = 0.7557), sepsis (p = 0.2379), body condition (p = 0.3616), age class (p = 0.9865, and sex (p = 0.7579)). The main pathological findings of the 41 necropsied animals are compiled in three tables in the Supplementary Materials, including the macroscopic findings, parasitic burden, and microbiological results of the sampled organs (Table S2), the main histopathological findings (Table S3), along with general information, diagnoses, and overall health status (Table S4). Additionally, Table S5 summarizes the main pathological findings, overall health status, state of nutrition, state of decomposition as well as sex, age class and indication of whether the animal was mercy killed or found deceased. Table 3. Antimicrobial resistance as a function of health parameters, sex and age class. Occurrence of antimicrobial resistance listed for different parameters including overall health, body condition, sex, age and the two most frequent diseases diagnosed during marine mammal necropsies (n = 41) with 20 animals harbouring antimicrobial-resistant E. coli.

Discussion
In the present study, antimicrobial-resistant E. coli isolates were detected in almost 40% of the 66 faecal and rectal swabs collected from marine mammals of the North and Baltic Seas. The occurrence of resistant E. coli was higher in the samples from the Baltic Sea (44.4%) compared to those from the North Sea (33.3%), but without statistical significance.
Although not many studies have investigated antimicrobial resistance in marine mammals thus far, resistant bacteria have been reported in various marine mammal species, including pinnipeds and cetaceans from different areas of the world [8,[83][84][85][86][87]. The occurrence of resistant E. coli isolates varied in the different studies, with reported figures of 10% [86], 26.9% [87], 47% [8], and 50% [83]. Studies in marine mammals, including different bacterial species, reported the presence of resistant isolates in 35% to 74% of the investigated samples [83,84,86,88]. No antimicrobial-resistant E. coli was obtained from the fish samples collected in this study, although resistant E. coli has been isolated from different marine fish species in other studies [40,44].
Multidrug resistance, defined as resistance to at least one antimicrobial of three or more antimicrobial classes [89], occurred in the present study in 37.2% of the resistant E. coli isolates, which is in accordance with an Irish study on harbour and grey seals that found multidrug resistance in 43.5% of the E. coli isolates [90]. Another study investigating Enterobacterales observed a higher prevalence, with 71% of the isolates exhibiting multidrug resistance [85]. In the present study, harbour porpoises were found to carry significantly higher numbers of multidrug-resistant E. coli isolates compared to the investigated seal species, while there was no significant difference between species regarding the carriage of E. coli isolates with resistance to at least one antimicrobial drug. One other study investigating antimicrobial resistance in harbour porpoises and harbour seals found that harbour porpoises had a significantly greater risk of carrying antimicrobial-resistant bacteria compared to seals, although the same study found that all E. coli isolates from harbour porpoises were susceptible to the tested antimicrobials [86]. The sample size of the present study needs to be increased to verify its statistical conclusions. In addition, the comparison of species and areas might be biased due to the difference in sample sizes of the individual species for the two areas. Nevertheless, harbour porpoise samples were retrieved mainly from the Baltic Sea and harbour seal samples mainly from the North Sea, which might indicate a species difference rather than a spatial one. Furthermore, due to the delay in sample processing, it is possible that the number of E. coli was underestimated, although it has been shown that samples collected at larger distances and longer processing periods allow the isolation and characterization of Gram-negative bacteria [91][92][93]. Nevertheless, it cannot be excluded that the delay in sample processing could also be the reason for the observed lack of resistant E. coli in the sampled fishes. The reasons for the difference in the occurrence of multidrug resistance between the marine mammal species detected in the present study are unknown. The sampled harbour porpoises, harbour seals, and grey seals inhabited the same areas. Furthermore, the three seal species as well as the harbour porpoises have similar broad prey spectra consisting mainly of fish and including most of the sampled fish species [94][95][96][97]. However, one difference is that seals haul out on land in larger groups [98], while harbour porpoises predominantly occur as solitary individuals or in small groups [99]. Nevertheless, close social contacts would suggest a higher occurrence of antimicrobial resistance in seals rather than in porpoises.
Considering the tested antimicrobial classes, most studies on E. coli from marine mammals found the highest resistance rates against β-lactams [8,90]. In other bacteria, the most frequently reported resistances were also against β-lactams, as well as against fluoroquinolones [84,100,101]. To our knowledge, only one study on marine mammals reported high rates of resistance against aminoglycosides together with fluoroquinolones in E. coli [102], consistent with the finding of the present study. Here, the highest frequencies were seen for resistance to aminoglycosides (72.1%), closely followed by resistance to non-extended spectrum β-lactams (60.5%). The high rates of resistance to aminoglycosides are unexpected if comparing this result with antimicrobial drug sales in Germany, from where most of the samples were obtained. Between 2011 and 2014, the amounts of antimicrobial drugs sold to German-based veterinarians that belonged to the antimicrobial classes tested in this study were in the following order: penicillins, tetracyclines, sulfonamides, aminoglycosides, folate synthesis inhibitors, fluoroquinolones, cephalosporins, and phenicols [103]. In the same period, β-lactams and fluoroquinolones were most frequently prescribed in hospitals, while for outpatients, β-lactams and tetracyclines had the highest prescription rates [103]. This indicates that factors other than just the consumption of antimicrobial drugs in the human and animal health sectors may influence antimicrobial resistance patterns in the environment. Of specific concern is the high rate of resistance seen to streptomycin, as this is one of the watch group antibiotics in the WHO Access, Watch, Reserve (AWaRe) classification of antibiotics for the evaluation and monitoring of use [104]. On the other hand, none of the isolates showed extended-spectrum β-lactamase activity, and only one isolate was identified as being an AmpC β-lactamase producer, whereas both antimicrobial resistances were frequently detected in Enterobacterales from marine mammals of the US west coast [85,87].
In the present study, the resistance gene most frequently detected with regard to total numbers was bla TEM . Frequent occurrence of strA, strB, aadA1, sul2, sul1, tet(A), and tet(B) was also seen. Few instances bla SHV , bla OXA-like-1 , bla CMY-2 , catA1, catA2, floR, tet(D), dfrA1, dfrA5, and qnrS were detected. Most of these genes have already been reported in other studies of resistant bacteria from marine mammals [85,87,90]. In the present study, a corresponding resistance gene was not detectable for all resistant isolates. Eighteen phenotypically resistant and 14 intermediate resistant isolates did not carry any of the tested resistance genes. On the one hand, this can be explained by the fact that not all known resistance genes were tested, which is especially true for aminoglycosides, wherein 16 resistant isolates lacked a corresponding resistance gene [105]. On the other hand, resistances can be caused by mutations, which were also not tested here, and which are known to particularly occur in quinolone resistance, with seven of the quinolone-resistant isolates lacking a corresponding resistance gene [106][107][108]. Additionally, other resistance mechanisms such as chromosomally encoded efflux pumps may be responsible for the lack of detected resistance genes [109,110].
Resistant isolates were further assigned to the four major phylogenetic groups known for E. coli [81]. This assignment is primarily performed to classify isolates with regard to their phylogenesis [111]. Additionally, in E. coli isolates of human origin, a certain affiliation was registered regarding phylogenetic group and commensalism or pathogenicity. In this regard, isolates of phylogenetic group A are considered to be mainly commensals; isolates of group B1 are mainly commensals, but may include pathogens more often, while isolates of group D and especially group B2 are likely to be human extraintestinal pathogens [112][113][114][115]. Considering all species in the present study together, almost 40% of the isolates were assigned to the phylogenetic group B2 and then in declining frequency to groups B1 (30.2%), A (18.6%) and D (11.6%). Thus, half of the isolates are more likely to be pathogenic than commensal. Considering the individual species, harbour porpoise isolates mainly belonged to group B1, while harbour seal and ringed seal isolates most frequently belonged to group B2. The three grey seal isolates were distributed evenly between the groups A, B1 and B2. A high prevalence of E coli from pinnipeds in phylogenetic group B2 was also found in two other studies [116,117], while in one of these studies, another pinniped colony exhibited a higher occurrence of group D isolates [117]. This suggests a higher proportion of antimicrobial-resistant E. coli with pathogenic potential in seal species compared with harbour porpoises in the present study and/or hints at a different phylogenesis. In this regard, it was reported that the phylogenetic group assignment of E. coli isolates clearly differed in prevalence between birds (D/B1), nonhuman mammals (A/B1), and humans (A/B2) [118]. Another study assigned human isolates mainly to phylogenetic group A, followed by D, B2, and B1, although high variance was seen [119]. In Australia, the phylogenetic group B2 is dominant in E. coli isolates from humans and omnivorous terrestrial mammals, while carnivorous species mainly carry isolates of group B1 [118,120]. These classifications apply to the results presented here on harbour porpoise as non-human mammalian and piscivorous (suggesting this is grouped within carnivores) species but not for the seals, which are also non-human mammals feeding on fish. Potentially, seals are indeed carriers of group B2 isolates phylogenetically, or they may carry a high degree of human E. coli isolates, referring to the studies assigning human isolates predominantly to group B2. However, further research is needed to address this question. Concerning multidrug resistance, the highest frequency was found for isolates assigned to phylogenetic group B1. However, due to the difference in phylogenetic group abundance of the different species, as well as the fact that harbour porpoises carried most of the multidrug-resistant isolates, this result should be judged with caution.
In the present study, necropsy results did not reveal any statistically significant parameter that might have had an influence on the occurrence of antimicrobial-resistant E. coli in the investigated marine mammals. This is not surprising as there are no known negative impacts of antimicrobial-resistant bacteria on their carrier unless they are pathogenic and cause diseases that need antimicrobial treatment. Moreover, as the majority of diseased wildlife will not receive antimicrobial treatment, a treatment-associated increase in the amount of antimicrobial-resistant bacteria is unlikely to occur in diseased animals. Although our results did not show any correlation between the occurrence of antimicrobial resistance and the overall health status of wild marine mammals, elsewhere, a higher prevalence of antimicrobial resistance was reported for in stranded or bycaught marine mammals than in live marine mammals [83]. The higher prevalence of antimicrobial resistance was postulated to be potentially caused by the poorer overall health status of the stranded or by-caught animals, but could also be caused by the different sampling methods of live versus dead animals (rectal swabs versus organ samples) [83].
Finally, it should be noted that the marine mammals of the North and Baltic Seas perform extended home ranges [59,61,64,65], giving them the opportunity to disseminate antimicrobial-resistant bacteria over longer distances. The detected rates of antimicrobial resistance indicate that dissemination of the resistant bacteria into the sea as well as onto beaches may pose a direct health risk for humans [36,56], especially as humans, wildlife and the environment share similar E. coli strains [55,121]. This highlights the importance of monitoring sentinel species from a One Health perspective.

Conclusions
The results presented herein underscore the potential of marine mammals as sentinels for aquatic ecosystem health and human health by indicating the prevalence of antimicrobial resistance circulating in the marine environment. In addition, it is shown that marine mammals of the North and Baltic Seas serve as reservoirs and vectors for antimicrobial-resistant bacteria, thus participating in the circulation of resistant bacteria and resistance genes. The occurrence of multidrug-resistant isolates is of specific concern for human and domestic animal health. To combat antimicrobial resistance, the surveillance of different ecosystems by monitoring important reservoir and vector species is essential. Till date, there is insufficient data regarding the role of wildlife and the environment in the complex mechanism of antimicrobial resistance. Thus, it is necessary to unravel the processes underlying its emergence, preservation, and dissemination in environmental settings to ensure the effectiveness of mitigation strategies against this global health threat.
Supplementary Materials: The following are available online at https://www.mdpi.com/article/ 10.3390/antibiotics11091248/s1, Figure S1: The map depicts the samples of fishes and marine mammals from the North and Baltic Sea coasts of Schleswig-Holstein, Germany; Figure S2: The figure shows the dendrogram including the pulsed-field gel electrophoresis band patterns of the 42 isolates that were typeable by Xbal macrorestriction; Table S1: The table lists information for each sample including sample ID, species, sample location and sea (NS = North Sea, BS = Baltic Sea), sample method, age class, sex (m = male, f = female) and if a necropsy was performed; Table S2: The table shows all marine mammal and fish sample numbers and which sample grew on which antibiotic containing plate as well if E. coli was isolated from at least one of the plates. No E. coli isolation was performed in table cells with x because of missing bacterial growth on all antibiotic containing agar plates. All samples showed bacterial growth either on Blood agar, Gassner agar or in the Mossel bouillon indicating bacterial presence in the samples; Table S3: Necropsy findings. The table lists all necropsied marine mammals, their sample IDs, the species (Pv = Phoca vitulina (harbour seal), Hg = Halichoerus grypus (grey seal), Ph = Phusa hispida, Pp = Phocoena phocoena (harbour porpoise)), if antimicrobial-resistant E. coli were isolated from the sample, date of necropsy, freezing prior to necropsy, main macroscopic findings, parasitic burdens, and results of microbiology testing of tissues; Table S4: Histopathological findings. The table lists the sample IDs, if antimicrobial-resistant E. coli were isolated from the sample, and the main histopathological findings for each investigated organ; Table S5: Main diagnoses and general information. The table lists the sample IDs, if antimicrobialresistant E. coli were isolated from the sample, the main diagnoses, a classification of the overall health status per animal, the state of nutrition, the age, the sex, if the animal was mercy killed or died on its own, and the state of decomposition.  Institutional Review Board Statement: Sampling procedures that included catching and handling of seals in German waters were reviewed by one of the animal welfare officers of the University of Veterinary Medicine Hanover, Foundation, and approved by the responsible governmental ethical committees. These procedures were conducted under animal experiment ethical permit 70-6/07 of the Ministry of Energy Transition, Climate Protection, the Environment, and Nature of Schleswig-Holstein (MEKUN SH). Two samples of alive harbour porpoise were taken in Danish waters before auditory threshold measurements were performed, conducted under permission issued to Jonas Teilmann, Aarhus University by the Danish Nature Agency (Danish Ministry of Environment, NST-3446-0016) and the Animal Experiments Inspectorate (Danish Ministry of Food, Agriculture and Fisheries, 2010/561-1801). All samples of Swedish ringed seals were obtained during regular hunts. Seal hunting in Sweden is regulated by the Hunting Act (1987:259) and the Hunting ordinance (1987:259). The Swedish Environmental Protection Agency decides yearly on a quota of ringed seals that can be shot for protective measures. Ethical approval is not needed for collecting samples from the hunt. No animal was harmed just for the purpose of this study.
Informed Consent Statement: Not applicable.

Data Availability Statement:
The data presented in this study are available in this published article and its Supplementary Materials.