Assessing the Bacterial Community Composition of Bivalve Mollusks Collected in Aquaculture Farms and Respective Susceptibility to Antibiotics

Aquaculture is a growing sector, providing several products for human consumption, and it is therefore important to guarantee its quality and safety. This study aimed to contribute to the knowledge of bacterial composition of Crassostrea gigas, Mytilus spp. and Ruditapes decussatus, and the antibiotic resistances/resistance genes present in aquaculture environments. Two hundred and twenty-two bacterial strains were recovered from all bivalve mollusks samples belonging to the Aeromonadaceae, Bacillaceae, Comamonadaceae, Enterobacteriaceae, Enterococcaceae, Micrococcaceae, Moraxellaceae, Morganellaceae, Pseudomonadaceae, Shewanellaceae, Staphylococcaceae, Streptococcaceae, Vibrionaceae, and Yersiniaceae families. Decreased susceptibility to oxytetracycline prevails in all bivalve species, aquaculture farms and seasons. Decreased susceptibilities to amoxicillin, amoxicillin/clavulanic acid, cefotaxime, cefoxitin, ceftazidime, chloramphenicol, florfenicol, colistin, ciprofloxacin, flumequine, nalidixic acid and trimethoprim/sulfamethoxazole were also found. This study detected six qnrA genes among Shewanella algae, ten qnrB genes among Citrobacter spp. and Escherichia coli, three oqxAB genes from Raoultella ornithinolytica and blaTEM-1 in eight E. coli strains harboring a qnrB19 gene. Our results suggest that the bacteria and antibiotic resistances/resistance genes present in bivalve mollusks depend on several factors, such as host species and respective life stage, bacterial family, farm’s location and season, and that is important to study each aquaculture farm individually to implement the most suitable measures to prevent outbreaks.


Introduction
Aquaculture is an ancient activity, practiced since the Roman Empire (140 B.C.) in Europe. It has developed over the centuries, but it was in the last three decades that it experienced its greatest growth, pressured by increased demand [1,2].
Bivalve mollusks are known to be rich in proteins, vitamin D, long-chain omega-3 fatty acids, iodine and selenium, contributing to a healthy diet [3]. These organisms represent the main aquaculture production in Portugal and, in 2012, 95.2% of the active establishments were for bivalve mollusks' production [4]. In 2015, the main species produced were

Bacterial Diversity in Clams Samples
In clams samples we identified ten different families of bacteria (Figures 1 and 2 and Table S1): Aeromonadaceae, Comamonadaceae, Enterococcaceae (only in summer), Bacillaceae, Staphylococcaceae, Streptococcaceae (only in autumn), Enterobacteriaceae, Morganellaceae, Pseudomonadaceae and Vibrionaceae (in both seasons). In summer, Morganellaceae was the most frequent bacterial family (38.1%). In fact, Morganellaceae appears more in this season in clams than in other bivalves studied (p = 0.02; Table 2). However, in autumn the results differ, with Enterobacteriaceae representing the most frequently isolated bacterial family (52.4%).   Table 2. Odds ratio (OR) and 95% confidence intervals (CI) (p ≤ 0.05) from the analyses of positive and negative associations between bivalve species and each bacterial family, bivalve species/bacterial family and season, C. gigas/bacterial family and location, bivalve species/bacterial family and nonsusceptibility to different antibiotic's class (using the results from the initial screening in selective media).    Table 2. Odds ratio (OR) and 95% confidence intervals (CI) (p ≤ 0.05) from the analyses of positive and negative associations between bivalve species and each bacterial family, bivalve species/bacterial family and season, C. gigas/bacterial family and location, bivalve species/bacterial family and nonsusceptibility to different antibiotic's class (using the results from the initial screening in selective media).

Bacterial Diversity in Mussels samples
Ten different families of bacteria were found among mussels samples: Aeromonadaceae (only in summer), Enterococcaceae (only in autumn), Bacillaceae, Enterobacteriaceae, Micrococcaceae, Morganellaceae, Pseudomonadaceae, Shewanellaceae, Staphylococcaceae and Vibrionaceae (in both seasons). In aquaculture farm 2, the most frequently found bacterial family in summer was Shewanellaceae (33.3%), whereas in autumn the most frequently found was Vibrionaceae (33.3%). In aquaculture farm 3, Enterobacteriaceae Antibiotics 2021, 10, 1135 5 of 15 predominated in both seasons (with 60.0% in summer and 44.4% in autumn). In aquaculture farm 4, the most frequently found bacterial family in summer was Vibrionaceae (29.4%) and in the autumn it was Enterobacteriaceae (93.1%) (Figures 1 and 2 and Table S1). Overall, we verified that Enterobacteriaceae appears more in autumn in mussels than in other bivalves (p ≤ 0.01) and was most frequent in autumn than in summer among mussels samples (p ≤ 0.01). Our statistical analyses showed that Morganellaceae were not usually associated with mussels samples (protective association; p = 0.02). In this bivalve species, Shewanellaceae and Vibrionaceae were most frequently found in summer than in autumn (p ≤ 0.01 and p = 0.02, respectively) ( Table 2). Table 2. Odds ratio (OR) and 95% confidence intervals (CI) (p ≤ 0.05) from the analyses of positive and negative associations between bivalve species and each bacterial family, bivalve species/bacterial family and season, C. gigas/bacterial family and location, bivalve species/bacterial family and nonsusceptibility to different antibiotic's class (using the results from the initial screening in selective media).

Bacterial Diversity in Japanese Oysters Samples
In Japanese oysters samples we identified nine different bacterial families: Bacillaceae, Shewanellaceae, Yersiniaceae (only in summer), Pseudomonadaceae (only in autumn), Enterobacteriaceae, Enterococcaceae, Moraxellaceae, Morganellaceae and Vibrionaceae (in both seasons). In this group of samples, in aquaculture farm 1 Enterobacteriaceae was the most frequently found bacterial family in summer (64.7%; Figures 1 and 2 and Table S1). Indeed, this family appears more in summer in Japanese oysters than in the other bivalves studied (p ≤ 0.01; Table 2). In autumn, Morganellaceae was the most common family (37.5%) in aquaculture farm 1. In fact, Morganellaceae is more frequently found in autumn than in summer on all the studied aquaculture farms (p ≤ 0.01). In farm 5 (only samples collected in summer), Vibrionaceae was the most frequently found bacterial family (43.8%), while in farm 6 (only samples collected in autumn), Enterobacteriaceae predominated (41.4%).

Initial Evaluation of Decreased Susceptibilities
Initial screening with selective media containing antibiotics allowed the identification of decreased susceptibilities. Figure 3 ( Figure S1 and Table S2) presents the results of this screening (eliminating known intrinsic resistances for the analysis). Decreased susceptibility to oxytetracycline prevails in all bivalve species, aquaculture farms and seasons. bivalve species (protective association) when compared to other species of bivalve analyzed in this study (p ≤ 0.01; Table 2). Furthermore, this decreased susceptibility appears more associated with summer than autumn in these bivalves (p = 0.02; Table 2). Decreased susceptibility to amoxicillin and colistin were only found in samples from region B (farms 5 and 6), whereas decreased susceptibility to cefotaxime was only recovered in samples from region A (farm 1). Decreased susceptibility to nalidixic acid was only observed in samples collected in autumn (31.3% in farm 6 and 10.0% in farm 1).

Antibiotic Susceptibility of β-Lactamase-and Plasmid-Mediated Quinolone Resistance (PMQR)-Producing Strains
Investigation of resistance genes by PCR revealed six qnrA genes among S. algae, ten qnrB genes among C. braakii, C. freundii and E. coli, and three oqxAB genes from R. ornithinolytica strains (Table 3). qnrA genes were found in region A and B aquaculture farms, in both seasons, although they predominated in summer. However, qnrB and oqxAB genes were only found in region A aquaculture farms in autumn. Three PMQR-producing strains (C. braakii, C. freundii and S. algae) revealed a susceptibility profile to all quinolones tested, with zone diameters ranging from 31 to 33 mm (disk diffusion) and concentrations of <0.015 to 0.125 mg/L (MIC) for ciprofloxacin, concentrations of 0.5 to 2 mg/L for flumequine, and a zone diameter of 27 mm for levofloxacin. The remaining sixteen PMQRproducing strains revealed a decreased susceptibility to at least one quinolone tested, with concentrations of 0.5 to >16 mg/L for ciprofloxacin and 2 to 64 mg/L for flumequine. Nonsusceptibilities to amoxicillin, colistin and oxytetracycline (MIC concentrations of 8 to 64 In clams, no decreased susceptibility to colistin was found and decreased susceptibility to chloramphenicol was only identified in samples collected in summer (15.4%).
In mussels, no reduced susceptibility to cefotaxime was found. In contrast, amoxicillin and oxytetracycline reduced susceptibility were present in all farms and seasons. Autumn was the only season where reduced susceptibility to chloramphenicol was found (16.7% in farm 3 and 18.5% in farm 4). However, reduced susceptibility to colistin was only found in a sample from farm 2, collected in summer (16.7%). Reduced susceptibility to nalidixic acid was not found in farm 3.
Although, in Japanese oysters, only decreased susceptibility to oxytetracycline was found in all farms and seasons, this decreased susceptibility was not so frequent in these bivalve species (protective association) when compared to other species of bivalve analyzed in this study (p ≤ 0.01; Table 2). Furthermore, this decreased susceptibility appears more associated with summer than autumn in these bivalves (p = 0.02; Table 2). Decreased susceptibility to amoxicillin and colistin were only found in samples from region B (farms 5 and 6), whereas decreased susceptibility to cefotaxime was only recovered in samples from region A (farm 1). Decreased susceptibility to nalidixic acid was only observed in samples collected in autumn (31.3% in farm 6 and 10.0% in farm 1).

Antibiotic Susceptibility of β-Lactamase-and Plasmid-Mediated Quinolone Resistance (PMQR)-Producing Strains
Investigation of resistance genes by PCR revealed six qnrA genes among S. algae, ten qnrB genes among C. braakii, C. freundii and E. coli, and three oqxAB genes from R. ornithinolytica strains (Table 3). qnrA genes were found in region A and B aquaculture farms, in both seasons, although they predominated in summer. However, qnrB and oqxAB genes were only found in region A aquaculture farms in autumn. Three PMQRproducing strains (C. braakii, C. freundii and S. algae) revealed a susceptibility profile to all quinolones tested, with zone diameters ranging from 31 to 33 mm (disk diffusion) and concentrations of <0.015 to 0.125 mg/L (MIC) for ciprofloxacin, concentrations of 0.5 to 2 mg/L for flumequine, and a zone diameter of 27 mm for levofloxacin. The remaining sixteen PMQR-producing strains revealed a decreased susceptibility to at least one quinolone tested, with concentrations of 0.5 to >16 mg/L for ciprofloxacin and 2 to 64 mg/L for flumequine. Non-susceptibilities to amoxicillin, colistin and oxytetracycline (MIC concentrations of 8 to 64 mg/L for the last one) were also found among S. algae strains. All R. ornithinolytica harboring an oqxAB gene were also resistant to oxytetracycline, with concentrations ranging from 16 to 64 mg/L. Decreased susceptibility to β-lactam antibiotics among Citrobacter spp. was mostly intrinsic. This genus also revealed decreased susceptibility to phenicols and oxytetracycline, with a concentration of 32 mg/L for chloramphenicol (C. braakii), 8 to 16 mg/L for florfenicol and 8 mg/L for oxytetracycline. E. coli harboring qnrB19 and bla TEM-1 genes revealed non-susceptibility to β-lactam antibiotics as well, in addition to non-susceptibilities to chloramphenicol, florfenicol, ciprofloxacin, flumequine, oxytetracycline and trimethoprim/sulfamethoxazole. Our study detected 11 multidrug resistance strains: one C. braakii recovered from clams, one C. freundii recovered from Japanese oysters, and eight E. coli and one S. algae recovered from mussels.

Discussion
As we observed in our study, bivalve mollusks usually concentrate a great diversity of bacterial species/families, which makes them susceptible to various diseases and may represent a risk to human health, since some of these bivalves are eaten raw (e.g., oysters). Indeed, statistically significant differences in bacterial composition between bivalve species from the same aquaculture farm and season (clams and Japanese oysters from farm 1, region A) were detected. Within the same bivalve species, we also observed variations between farms from the same region (mussels in region A) and different regions (Japanese oysters in region A and B), although these variations were not statistically significant. Fernández et al. detected variations in bacterial composition of post larvae and adult stages of Crassostrea corteziensis, Crassostrea gigas and Crassostrea sikamea at different cultivation sites [13]. They concluded that the most frequent phyla were Proteobacteria, Bacteroidetes, Actinobacteria and Firmicutes (in that order), using a high-throughput sequencing approach (pyrosequencing). In a different study, Pierce and Ward evaluated the gut microbiome from Crassostrea virginica and Mytilus edulis and confirmed that these species had similar (but not identical) gut microbiomes that vary with the seasons [11]. The most abundant phyla were Proteobacteria, Tenericutes, Verrucomicrobia, Bacteroidetes, Cyanobacteria, Planctomycetes, Actinobacteria, Firmicutes, and Fusobacteria. In our study, the most frequently identified phylum was also Proteobacteria, which comprised the following families: Aeromonadaceae, Comamonadaceae, Enterobacteriaceae, Moraxellaceae, Pseudomonadaceae, Shewanellaceae, Vibrionaceae, and Yersiniaceae. These findings agree with the hypothesis previously proposed that bacterial composition in bivalve mollusks is influenced by host species and respective life stage, diet, rearing conditions, bacterial composition of the aquatic habitat, salinity, and temperature [9,21,22].
The bacteria found in this study belonging to the genera Acinetobacter, Aeromonas, Bacillus, Micrococcus, Photobacterium, Pseudomonas, and Vibrio are ubiquitous in the water environments and commensal microbiota of bivalves [9]. Some of these bacteria, especially those belonging to Proteobacteria phylum, are important for bivalve mollusks' metabolism, since they are able to fix nitrogen in the gastrointestinal tract of these organisms and degrade cellulose and agar (the main elements of the food ingested by bivalve mollusks) [21]. Species belonging to the genera Aeromonas, Pseudomonas and Vibrio are also responsible for diseases in bivalve larvae [9]. Information about the frequency and pathogenicity of other genera found in this study in bivalve mollusks is scarce. However, there are several studies reporting human infections caused by all seven genera described above, namely wound infections, foodborne diseases (by ingestion of raw seafood), myonecrosis, septicemia, necrotizing fasciitis, empyema, bacteremia, endocarditis, severe respiratory, urinary and biliary tract infections, meningitis, and keratitis [9,[18][19][20][22][23][24][25][26][27].
Non-commensal bacteria of bivalve mollusks such as Enterobacter spp., Enterococcus faecalis, Enterococcus faecium, E. coli, Klebsiella aerogenes, K. pneumoniae, M. osloensis, M. morganii, and Providencia rettgeri could be indicators of fecal contamination, since these bacteria are widely found in the gastrointestinal tract of humans and several other animals [13,25,32,34,35,40,46,48,51]. These and other bacteria can enter aquaculture farms through runoff from land (especially during periods of high precipitation), sewage, maritime traffic and birds or marine mammals. Fecal material from land and sewage can concentrate a high bacterial diversity, as well as several heavy metals, antibiotics, and organic substances, promoting a selective pressure on bacteria normally present in an aquatic environment [13].
The initial screening with selective media containing antibiotics and, subsequently, the MIC results revealed the prevalence of decreased susceptibility to oxytetracycline in all bivalve species, aquaculture farms and seasons. This prevalence could be explained by the high frequency of prescription of this antibiotic in aquaculture, due to broad spectrum of activity, low cost, and potency [7]. With one exception, we could not establish an association between an antibiotic and a specific bivalve species, location, or season. The exception was the decreased susceptibility to oxytetracycline that was statistically associated (p = 0.02) with summer in Japanese oysters.
Decreased susceptibility to β-lactams was also found in the present study. This antibiotic class is frequently used in aquaculture in several countries [25] and high resistance rates are usually observed in bivalve mollusks (especially regarding amoxicillin), often associated with bla TEM and bla CTX-M [13]. In our study, we also detected the bla TEM-1 gene in eight multidrug resistant E. coli strains with decreased susceptibility to amoxicillin, amoxicillin/clavulanic acid and/or ceftazidime. Noteworthy, strains with non-susceptibilities to β-lactams might have other resistance mechanisms not studied here (e.g., efflux pumps).
The other decreased susceptibility detected in this study was to chloramphenicol, although this antibiotic has already been banned for use in food-producing animals in Europe since the 1990s. This decreased susceptibility can persist in the environment due to coselection with other antibiotics (especially florfenicol, which is widely used in aquaculture, and decreased susceptibility to this antibiotic was also found in this study) and/or heavy metals. Moreover, there are soil bacteria that are capable of producing this substance [25]. Other studies also confirmed the presence of this antibiotic resistance in bivalve mollusks [13,59].
Worryingly, our study detected decreased susceptibility to colistin, an antibiotic of last resort against multidrug resistant Gram-negative bacteria in human medicine. This antibiotic is also used in veterinary medicine, including aquaculture, the latter hypothesized by some authors as the source of certain colistin resistance genes [60,61]. In our study, fourteen strains revealed decreased susceptibility to this antibiotic. Of these, seven, identified as Shewanella algae and Photobacterium damselae, had non-intrinsic resistance and were isolated from mussels collected in aquaculture farm 2 and Japanese oysters collected in aquaculture farm 5 (in the region A and B, respectively). These results may reflect a selective pressure in these regions that facilitates the dissemination of strains with decreased susceptibility to colistin. No plasmid-mediated colistin resistance-encoding genes were detected, which suggested that other resistance mechanisms not studied here are responsible for the decreased susceptibility to this antibiotic (other mcr-variant genes; efflux pumps; or pmrC, pmrE, mgrB genes, among others genes and operons that play a role in lipopolysaccharide modification and consequent decreased susceptibility to colistin) [62]. Previous studies had already detected decreased susceptibility to colistin and mcr-1 gene in clams and mussels, respectively [61,63].
Our study revealed a low prevalence of decreased susceptibility to nalidixic acid (13%) in all bivalve samples analyzed. This antibiotic belongs to a class commonly used in aquaculture, quinolones [25]. Former studies in bivalve mollusks, also detected low levels of decreased susceptibility to quinolones [13,14]. Of the twenty strains with decreased susceptibility to nalidixic acid, only three strains of R. ornithinolytica revealed a positive result for the oqxAB gene. These three strains also revealed decreased susceptibility to other quinolones, such as flumequine and ciprofloxacin. Investigation of quinolones resistance genes by PCR in all Gram-negative strains, regardless their phenotype, revealed the presence of six qnrA-type genes among S. algae and ten qnrB-type genes among C. braakii, C. freundii and E. coli. Interestingly, not all strains harboring a qnr gene revealed a resistance phenotype to quinolones (one C. braakii with a qnrB-type, one C. freundii with a qnrB44 and one S. algae with a qnrA12). This may be caused by non-functional proteins, such as in the case of C. braakii with a deletion in qnrB gene that originated a premature stop codon, or a low expression of these genes, difficult to detect by phenotypic methods [64]. These results highlight the importance of using both phenotypic and genotypic methods in research of antibiotic resistances/resistance genes, since there is not always phenotypic and/or genotypic expression. Although qnrA, qnrB and oqxAB genes are frequently found in aquaculture environments [25,65,66], little information is known about their frequency in bivalve mollusks, thus this study can contribute to better knowledge in this field.
All E. coli strains that harbor a qnrB19 gene had a decreased susceptibility to quinolones (ciprofloxacin and/or flumequine) and also presented resistance to the combination trimethoprim/sulfamethoxazole (also used in aquaculture [25]). This resistance was previously reported in bivalve mollusks [67] and could be associated with the acquisition of genes sulI and sulII (for sulfamethoxazole resistance) and dhfrI and dhfrII (for trimethoprim resistance), causing the alteration of the antibiotic target [68].
In this study, we observed low resistance rates/few resistance genes to the antibiotics tested (except for oxytetracycline). However, it is important to implement surveillance plans in aquaculture farms, since this environment can be a reservoir of antibiotic resistance and/or antibiotic resistance genes. The implementation of measures that help to prevent outbreaks is also crucial, because fighting an outbreak is more difficult and expensive. Examples of such measures are limiting stock movements, avoiding exposure to elevated temperatures and high or low salinity, strict hygiene measures, and decreasing stock densities, among others. In the presence of an outbreak, it is important to identify the pathogen responsible and, if possible, to test its susceptibility to antibiotics, so that veterinarians can use a narrow-spectrum antibiotic at the correct concentration. Whenever possible, antibiotic administration by bath or feed should be avoided, giving preference to more individual methods to prevent the exposure of healthy individuals and the aquatic environment to a selective pressure. Investment in alternatives to antibiotics should be made, such as antimicrobial peptides (produced by several species of bivalve mollusks), bacteriophages, probiotics, and vaccines, always considering animal welfare and the product's safety for human consumption [9,69].

Sample Characterization
The Portuguese Institute for the Sea and Atmosphere (IPMA) provided the bivalve samples used in this study. In summer of 2019, were collected one sample of clams (R. decussatus) and one sample of Japanese oyster (C. gigas) in aquaculture farm 1 from region A (south of Portugal); three samples of mussels (Mytilus spp.) from aquaculture farms 2, 3 and 4, also in region A; and one sample of Japanese oyster collected from aquaculture farm 5 in region B (central region of Portugal). The aquaculture farms from region A present in this study are distributed along its coastline.
In autumn of 2019, the sampling previously described was repeated for both regions, except for the sample of Japanese oyster in region B, which was collected in a different aquaculture (farm 6).
All samples were frozen and transported on ice to the National Institute of Health Dr. Ricardo Jorge, where they were analyzed. In this study, one sample corresponds to 3 to 10 individuals, depending on the species (minimum 50 g, for each sample).

Bacterial Isolation and Identification
Fifty grams of each sample were homogenized in peptone water (Stomacher 80 Biomaster ® , Seward, UK), making a 1:10 dilution, and incubated for 12 to 18 h at 37 • C. Each dilution was plated in selective media, containing specific concentrations of different antibiotics (allowing an initial screening of decreased susceptibilities), and incubated for 18 to 20 h at 37 • C. Aeromonas agar, MacConkey agar and Thiosulfate Citrate Bile Salts Sucrose Agar contained the following standard antibiotic concentrations to select antibiotic resistant strains: 100 mg/L of amoxicillin, 2 mg/L of cefotaxime, 20 mg/L of chloramphenicol, 0.5 mg/L of colistin, 50 mg/L of nalidixic acid and 8 mg/L of oxytetracycline.
Mannitol salt agar and UriSelect™4 chromogenic agar contained 8 mg/L of oxytetracycline. Plates with and without antibiotic were used as controls. Colonies with different morphology (to avoid duplications) were selected and DNA extracted, according to manufacturer's instructions (MagNA Pure 96 Instrument, Roche, Manheim, Germany). Strains were identified by MALDI-TOF and amplification of the 16S rRNA gene, as previously described [70].

Statistical Analyses of Results
Statistical analyses were performed to detect positive or negative associations between bivalve species and each bacterial family, bivalve species/bacterial family and season, C. gigas/bacterial family and location, bivalve species/bacterial family and nonsusceptibility to different classes of antibiotics (using the results from the initial screening in selective media). Only factors identified as statistically significant are shown. Fisher's exact test was used to assess differences in bacterial families/season/location/nonsusceptibility to different classes of antibiotics between bivalve species and one-tailed p values of ≤0.05 were considered to be statistically significant. Associations were established by calculation of odds ratios with 95% confidence intervals. The null hypothesis was rejected for p values of ≤0.05. All statistical analyses were calculated using OpenEpi software, version 3.01 [71].
All Staphylococcus spp. were tested for the presence of mecA, mecC, vanA, vanB and vanD genes [25], whereas all Enterococcus spp. were studied for the presence of vanA, vanB and vanD genes.

Antibiotic Susceptibility Testing of Strains with Resistance Genes
Antibiotic susceptibility was studied by disk diffusion (Bio-Rad, Marnes-la-Coquette, France) and minimum inhibitory concentration (MIC) by in-house broth microdilution for nineteen strains that revealed the presence of resistance genes. Antibiotics tested and respective concentrations and breakpoints are listed in Table 4. The antibiogram was completed with disc combination test (DCT), double disc synergy test (DDST), faropenem (10 µg) and temocillin (30 µg) to search for extended-spectrum β-lactamase (ESBL), metallo-βlactamase (MBL), AmpC cephalosporinases and carbapenemases, as already reported [25]. The strains were considered multidrug resistant if they presented resistance to three or more structurally unrelated antibiotics. EUCAST species-specific intrinsic resistances were considered (https://www.eucast.org/expert_rules_and_intrinsic_resistance/) (accessed on 13 April 2021).  Table 4. Cont. were not available, therefore breakpoints from EUCAST and CLSI M100 for Pseudomonas spp. were used, as reported elsewhere [72]. 5 Breakpoints for CHL were used for FLO as well; breakpoints for CIP were used for FMQ; and breakpoints for tetracycline were used for OTC.

Conclusions
In recent years, there has been an increase in studies on microbiota and antibiotic resistances/resistance genes present in aquaculture, mainly on fish. This study presents an important contribution to fill the gaps in the knowledge of bacterial diversity and antibiotic resistance mechanisms in bivalve mollusks. We could observe a great variety of bacterial species and antibiotic resistances among clams, mussels and Japanese oysters, seasons, and locations. This fact highlights the need to study and adapt the surveillance plans and measures to prevent the spread of antibiotic resistance to each specific location and animal species. Therefore, bivalve mollusks can play an important role in monitoring these aquaculture environments, since their filter feeding habits make them excellent indicators of environmental pollution [13].