Live Feeds Used in the Larval Culture of Red Cusk Eel, Genypterus chilensis, Carry High Levels of Antimicrobial-Resistant Bacteria and Antibiotic-Resistance Genes (ARGs)

Simple Summary The culture of the marine fish red cusk eel Genypterus chilensis is currently considered a priority for Chilean aquaculture but low larval survival rates have prompted the need for the continuous use of antibiotics, mainly florfenicol. In this study, the role of live prey (rotifers and the brine shrimp Artemia franciscana) used to feed fish larvae as a source of antibacterial-resistant bacteria in a commercial culture of G. chilensis was investigated. Samples of live feeds were collected during the larval growth period and their bacterial contents were determined. High levels of potentially opportunistic pathogens, such as Vibrio spp., as well as florfenicol-resistant bacteria, were detected. Sixty-five florfenicol-resistant isolates were recovered from these cultures and identified as Vibrio (81.5%) and Pseudoalteromonas (15.4%), which exhibited a high incidence of co-resistance to the antibiotics streptomycin, oxytetracycline, co-trimoxazole, and kanamycin. The majority of them carried the florfenicol-resistance encoding genes floR and fexA. The high prevalence of antibiotic-resistant bacteria and the associated genetic elements in live feed administered to reared fish larvae requires the prompt implementation of efficient management strategies to prevent future therapy failures in fish larval cultures and the spread of antibiotic-resistant bacteria to associated aquatic environments. Abstract The culture of red cusk eel Genypterus chilensis is currently considered a priority for Chilean aquaculture but low larval survival rates have prompted the need for the continuous use of antibacterials. The main aim of this study was to evaluate the role of live feed as a source of antibacterial-resistant bacteria in a commercial culture of G. chilensis. Samples of rotifer and Artemia cultures used as live feed were collected during the larval growth period and culturable bacterial counts were performed using a spread plate method. Rotifer and Artemia cultures exhibited high levels of resistant bacteria (8.03 × 104 to 1.79 × 107 CFU/g and 1.47 × 106 to 3.50 × 108 CFU/g, respectively). Sixty-five florfenicol-resistant isolates were identified as Vibrio (81.5%) and Pseudoalteromonas (15.4%) using 16S rRNA gene sequence analysis. A high incidence of resistance to streptomycin (93.8%), oxytetracycline (89.2%), co-trimoxazole (84.6%), and kanamycin (73.8%) was exhibited by resistant isolates. A high proportion of isolates (76.9%) carried the florfenicol-resistance encoding genes floR and fexA, as well as plasmid DNA (75.0%). The high prevalence of multiresistant bacteria in live feed increases the incidence of the resistant microbiota in reared fish larvae, thus proper monitoring and management strategies for live feed cultures appear to be a priority for preventing future therapy failures in fish larval cultures.

The use of antibiotics as prophylactic agents to control bacterial contaminants in live feed is a routine practice in Chilean larval aquaculture. Among the antibiotics currently approved for use in Chilean aquaculture, florfenicol was by far the most frequently used during 2016, accounting for 82.5% of the antibiotics used in salmon farming [44], and is currently the most used antibacterial agent in marine larval cultures. Due to the intensive and widespread use of this antibiotic, florfenicol resistance has increased in aquaculture settings [45][46][47]. Furthermore, the observation that many of the genes encoding for phenicol resistance reside on mobile genetic elements, including plasmids [46,48], which usually carry other resistance determinants [46,48], the co-selection and persistence of these resistance genes may occur through the use of other antimicrobial agents. In addition, other mechanisms of gene transfer, besides plasmid transfer, in the dissemination of antibacterial resistance among bacteria from fish farm environments have been suggested [49]. Knowledge of the prevalence of antibacterial-resistant bacteria and associated antibiotic-resistance genes (ARGs) is essential for the proper design of control strategies to improve the sanitary quality of live feed and to reduce the risk of introducing high levels of antibacterial-resistant bacteria, which could act as opportunistic pathogens for reared fish larvae. However, studies on the occurrence of antibacterial resistance among bacteria carried by live-feed cultures are still very scarce.
To our knowledge, no studies have previously been performed on the bacterial content of live feed used to feed reared marine fish larvae in Chile, including red cusk eel G. chilensis, which is currently considered a priority for Chilean aquaculture. Thus, this study will add important information on the sanitary status of live feeds and the impact of using florfenicol on the bacterial content of live-feed cultures. It could be concluded that medication of administered live feed severely increased the levels of antibiotic-resistant bacteria and related ARGs in rotifer and Artemia cultures, resulting in an important risk to reared fish larvae.
The main aim of the study was to evaluate the role of unmedicated and florfenicol-medicated live feed administered in a commercial larval culture of red cusk eel G. chilensis as an important source of antibacterial-resistant bacteria, along with the main genes encoding for florfenicol-resistance.

Ethics Statement
The study does not need ethical approval.

Live-Feed Culture Sampling
A commercial hatchery was included in the study for the culture of red cusk eel, G. chilensis, belonging to Colorado Chile S.A., located approximately 5 km south of Coquimbo at the southern end of La Herradura Bay on the Pacific coast of Chile. The experimental units deployed in the commercial hatchery consisted of two untreated tanks (rotifer B. plicatilis and brine shrimp A. franciscana nauplii), and additionally duplicated tanks of A. franciscana cultures treated once a day with florfenicol (final concentration of 20 mg/L). The length of the larval stage of the reared G. chilensis raised with live feed organisms was approximately 60 days post hatching (dph) (Figure 1). Live-feed samples were collected from duplicated culture tanks (600 L at a concentration of 200,000 rotifers per L and 10 L at a concentration of 1,000,000 naupli per L for the rotifer and Artemia cultures, respectively) immediately before feeding the reared fish larvae. Samples were taken at 6, 18, and 32 dph during the growth phase in which the larvae were fed with live feed and before feeding with artificial diets (Figure 1). For comparative purposes, rotifer and Artemia cultures were collected from massive cultures developed in the commercial hatchery. Duplicated cultures were treated with florfenicol (final concentration of 20 mg/L for 2 h), whereas other duplicated untreated cultures were used as untreated controls. After 2 h of exposure, samples from the untreated and florfenicol-treated cultures were collected, placed on ice, immediately transported to the laboratory, and processed within 1 h after collection.

Sample Processing and Culturable Bacterial Count
Live feed cultures were retained on a 65-µm mesh, aseptically concentrated and collected using sterile Falcon tubes (approximately a volume of five mL of concentrated live prey), centrifuged at 4700 × g for 10 min, and the supernatants were eliminated with a micropipette. The pellets obtained were aseptically weighed and ground by hand using a sterile glass digester with 1 mL of sterile physiological saline (0.85%) (PS) added to obtain a homogenate. Appropriate 10-fold dilutions of the homogenates in PS were prepared and 0.1-mL aliquots were inoculated in triplicate onto agar plates. Culturable counts of heterotrophic and florfenicol-resistant bacteria were determined using a spread plate method with Plate Count Agar (PCA, Becton Dickinson, Sparks, MD, USA), with 2% NaCl added, whereas Vibrio spp. counts were determined using Thiosulfate-Citrate-Bile Salts Sucrose agar (TCBS, Becton Dickinson, Sparks, MD, USA) prepared using 50% microfiltered (0.22 µm) aged seawater [50]. Plates with added florfenicol (30 µg/mL, Sigma-Aldrich ® (St. Louis, MD, USA) were used to determine the florfenicol-resistant bacteria [50,51]. All plates were incubated at 22 °C for 5 days and the bacterial numbers per gram of sample were calculated as described in Miranda and Rojas [51]. For comparative purposes, rotifer and Artemia cultures were collected from massive cultures developed in the commercial hatchery. Duplicated cultures were treated with florfenicol (final concentration of 20 mg/L for 2 h), whereas other duplicated untreated cultures were used as untreated controls. After 2 h of exposure, samples from the untreated and florfenicol-treated cultures were collected, placed on ice, immediately transported to the laboratory, and processed within 1 h after collection.

Sample Processing and Culturable Bacterial Count
Live feed cultures were retained on a 65-µm mesh, aseptically concentrated and collected using sterile Falcon tubes (approximately a volume of five mL of concentrated live prey), centrifuged at 4700× g for 10 min, and the supernatants were eliminated with a micropipette. The pellets obtained were aseptically weighed and ground by hand using a sterile glass digester with 1 mL of sterile physiological saline (0.85%) (PS) added to obtain a homogenate. Appropriate 10-fold dilutions of the homogenates in PS were prepared and 0.1-mL aliquots were inoculated in triplicate onto agar plates. Culturable counts of heterotrophic and florfenicol-resistant bacteria were determined using a spread plate method with Plate Count Agar (PCA, Becton Dickinson, Sparks, MD, USA), with 2% NaCl added, whereas Vibrio spp. counts were determined using Thiosulfate-Citrate-Bile Salts Sucrose agar (TCBS, Becton Dickinson, Sparks, MD, USA) prepared using 50% microfiltered (0.22 µm) aged seawater [50]. Plates with added florfenicol (30 µg/mL, Sigma-Aldrich ® (St. Louis, MD, USA) were used to determine the florfenicol-resistant bacteria [50,51]. All plates were incubated at 22 • C for 5 days and the bacterial numbers per gram of sample were calculated as described in Miranda and Rojas [51].

Bacterial Isolates
A total of 65 florfenicol-resistant Gram-negative bacilli were recovered from live feed (27 and 38 isolates from rotifer and Artemia cultures, respectively) during the assay as well as from various cultures routinely developed in the hatchery during the period 2018-2019. Isolates were recovered using plates with Tryptic soy agar (TSA, Becton Dickinson, Sparks, MD, USA) containing florfenicol (30 µg/mL) and incubated at 22 • C for 5 days. Isolates were stored at −85 • C in CryoBank TM vials (Mast Diagnostica, Reinfeld, Germany) and were grown in Trypticase soy agar (TSA) at 22 • C for 24 h prior to use.

Bacterial Identification
The Gram staining, cell morphology, oxidase production, oxidation/fermentation (O/F) of glucose, growth on thiosulfate-citrate-bile salts sucrose (TCBS) agar, and susceptibility to the vibriostatic agent O/129 (2,4-diamino-6,7-diisopropylpteridine) (10 and 150 µg per disc) phenotypic tests were conducted according to the procedures described in Buller [52] using media supplemented with NaCl (2%). Furthermore, antimicrobial-resistant isolates were identified using bacterial 16S rRNA gene sequence analysis. For amplification of the 16S rRNA genes, isolates were cultured in Tryptic soy broth (Oxoid, Hants, UK) at 22 • C for 12-24 h and centrifuged at 9000× g for 3 min using an Eppendorf 5415D (Eppendorf AG, Hamburg, Germany) microcentrifuge to obtain a pellet. DNA extraction was carried out using the Wizard ® Genomic DNA Purification commercial kit (Promega, Madison, WI, USA) following the supplier's instructions, and the obtained DNA samples were stored at −20 • C until analysis. The amplification of the 16S ribosomal genes of the isolates was carried out using PCR, following the methodology described by Opazo et al. [53]. The resulting amplified PCR products were sequenced with Macrogen (Rockville, MD, USA) using the ABI PRISM 373 DNA Sequencer (Applied Biosystems, Foster City, CA, USA). The sequences were edited and matched to the Ribosomal Database Project [54] to identify the bacterial isolates.

Minimum Inhibitory Concentrations (MICs) of Florfenicol
Minimum inhibitory concentrations (MICs) of florfenicol against all isolates were determined using an agar dilution method, as recommended by the Clinical and Laboratory Standards Institute (CLSI) guideline M07-A10 [55]. A serial two-fold dilution pattern of the antibiotic was added into Mueller-Hinton agar (MH, Becton Dickinson, Sparks, MD, USA) supplemented with NaCl (2%) to obtain final concentrations ranging from 0.5 to 2048 µg/mL. Bacterial suspensions were prepared in sterile 0.85% saline and triplicate plates were inoculated using a Steers replicator apparatus [56], delivering approximately 10 4 colony forming units per spot, and then incubated for 48 h at 22 • C. The bacterial inocula were applied simultaneously to the agar surfaces using the inoculum-replicating apparatus capable of transferring 32 isolates onto the surface of each plate [55]. The first and last agar plates did not contain any antibiotic in order to detect possible contamination of the isolates or antibiotic carryover. The MIC was defined as the lowest concentration of the antibacterial agent that produced an absence of growth in at least two of the three plates after 48 h. A reference of strain Escherichia coli ATCC 25922, recommended by CLSI [57], was used as a quality control organism for verification of the MIC ranges on used Mueller-Hinton agar plates.

Antibacterial Resistance Patterns
The susceptibility of isolates to various antimicrobials was determined using a disk diffusion test according to the Clinical and Laboratory Standards Institute (CLSI) guideline VET3-A [57]. Briefly, the bacterial isolates were resuspended in phosphate-buffered saline (PBS) to obtain a turbidity corresponding to 0.5 McFarland standard (bioMerieux, Marcy-l'Etoile, France), corresponding to a concentration of (1-2) × 10 8 CFU per mL [55]. Bacterial suspensions were streaked onto plates containing Mueller-Hinton agar (MH) with 2% NaCl added, and disks (Oxoid, Basingstoke, Hampshire, Animals 2020, 10, 505 6 of 23 England) containing the following antibiotics were used: amoxicillin (AML, 25 µg), streptomycin (S, 10 µg), kanamycin (K, 30 µg), oxytetracycline (OT, 30 µg), florfenicol (FFC, 30 µg), oxolinic acid (OA, 2 µg), flumequine (UB, 30 µg), and sulfamethoxazole trimethoprim (SXT, 25 µg). Plates were incubated at 22 • C for 24 h according to CLSI guidelines [57], and isolates were considered resistant according to the criteria established by the CLSI [58] or by Miranda and Rojas [51]. Escherichia coli ATCC 25922 was used as a quality control strain, as recommended by the CLSI [59]. Several isolates (20%) were re-examined to check the reproducibility of the assay. Furthermore, the antibacterial resistance index (ARI) of untreated and florfenicol-treated larvae cultures was determined according to Hinton et al. [60] using the formula ARI = y/nx, in which y was the actual number of resistance determinants recorded in a population of size n, and x was the total number of antibacterials tested for in the sensitivity test.

Detection of Genes Encoding for Florfenicol Resistance
All isolates were assayed for the presence of genes encoding for florfenicol resistance. The presence of the floR and fexA genes, encoding for efflux pumps, were detected using the methodology described by Fernández-Alarcón et al. [45] and Higuera-Llantén et al. [61], respectively. Primers F: 5'-AATCACGGGCCACGCTGTATC-3 and R: 5 -CGCCGTCATTCTTCACCTTC-3 (floR) and F: 5'-TTTCGCTGTTCTTGTGTTCG-3 and R: 5'-ACCTTGGAAAATCCCCATTC-3 (fexA), were used. The amplification conditions were as follows: for the floR gene, denaturation at 95 • C for 3 min; 30 cycles of denaturation at 95 • C for 30 sec, annealing at 56 • C for 30 sec, and elongation at 72 • C for 30 sec; and finally, extension at 72 • C for 7 min. For the fexA gene, denaturation at 95 • C for 3 min; 45 cycles of denaturation at 95 • C for 30 sec, annealing at 56 • C for 30 sec, and elongation at 72 • C for 30 sec; and finally, extension 72 • C for 7 min. Amplicon sizes of the floR and fexA genes were 200 and 450 bp, respectively. The amplified PCR products were sequenced using Macrogen (Rockville, MD, USA), and genes were identified using a computational analysis of the Basic Local Alignment Search Tool (BLAST) sequence alignment against the gene sequences included in the GenBank nucleotide sequence database. For the detection of floR and fexA genes, strain Citrobacter freundii FB98 carrying the floR gene [45] and a sequence of gen MH747503.1 [61], flanked by primers fexAF and fexAR synthesized using Integrated DNA Technologies (IDT, Coralville, IA, USA) were used as positive controls and were included in each PCR run.

Isolation of Plasmid DNA
Bacterial isolates were screened for their plasmid content, as previously described [62]. Briefly, the geneJET Plasmid miniprep kit (Thermo-Scientific, Waltham, MA, USA) was used and the obtained plasmid DNA was run on 1.5% agarose gel electrophoresis for plasmids less than 20 kb and 0.8% agarose gel for plasmids greater than 20 kb. Gels were stained with GelRed TM (Biotium, Hayward, CA, USA) and viewed using UV transillumination. The size was estimated via comparison with known plasmid weight standards and standard molecular weight markers.

Statistical Analysis
Culturable count values were transformed to log 10 , whereas proportions of florfenicol resistance were computed for each live feed (rotifers and Artemia) culture and were arcsine-transformed before analysis. One way analysis of variance (ANOVA) [63] was performed to detect significant differences (p < 0.05) between culturable counts of heterotrophic and florfenicol-resistant bacteria from control and treated live-feed cultures. Differences between untreated and florfenicol-treated cultures were detected using Student's t-test for two independent samples [63]. In addition, all untreated and treated culture tanks were compared in a nested ANOVA analysis. When differences were statistically significant, the control and treated live feed were then compared against each other using Tukey's test [63]. The frequencies of resistance to the assayed antimicrobials of the selected isolates were compared with Pearson's chi-square test, adjusted with Bonferroni's correction, and p < 0.05 was considered to indicate statistical significance. All statistical analyses were carried out using the SPSS version 12.0 computer program [64].

Culturable Bacterial Counts
According to the current hatchery protocols, rotifer cultures are not treated with antibiotics, in contrast to the Artemia cultures, which are routinely treated with florfenicol or other antibacterial agents, thus no treated rotifer mass cultures at the red cusk eel hatchery were available during the sampling period (Table 1).
Florfenicol-resistant bacterial counts were high in both live-feed cultures during the larval growth period, ranging from 10 4 to 10 7 CFU/g of rotifer and from 10 6 to 10 8 CFU/g of Artemia (Table 1). When the florfenicol-resistant bacterial counts from both live-feed cultures were compared, levels of resistant bacteria from Artemia cultures were higher than those from rotifer cultures.
When the untreated and florfenicol-treated Artemia cultures administered after 18 and 32 days post hatching (dph) of the larval culture were compared, no important differences in heterotrophic bacterial counts were observed (Table 1). By contrast, at 18 and 32 dph of larval culture, the culturable counts of florfenicol-resistant bacteria and Vibrio spp. from florfenicol-treated Artemia cultures were significantly higher than those from the untreated Artemia cultures (Table 1). Furthermore, florfenicol-resistant bacterial counts from untreated and florfenicol treated Artemia cultures at 32 dph were higher than those from 18 dph.
Culturable counts of heterotrophic bacteria, florfenicol-resistant bacteria, and Vibrio spp. from rotifer cultures at 18 dph were lower than those from the cultures at 6 and 32 dph. Rotifer cultures exhibited their highest culturable counts of heterotrophic bacteria at 32 dph, whereas the highest counts of Vibrio spp. and florfenicol-resistant bacteria were observed at 6 and 32 dph. No important differences in the levels of culturable counts of Vibrio spp. between 6 and 32 dph were observed.
When rotifer and Artemia cultures were treated with florfenicol, bacterial culturable counts remained at high levels (10 5 to 10 7 CFU/g), indicating that florfenicol did not reduce the bacterial content in either of the live feeds, as displayed in Figure 2.

Bacterial Isolates
The identification of 65 florfenicol-resistant isolates recovered from rotifer and Artemia cultures (27 and 38 isolates, respectively) used in the rearing of G. chilensis larvae showed a high predominance of Vibrio species in both rotifer and Artemia cultures (85.2% and 78.9%, respectively). In contrast, a low frequency of isolates belonging to the Pseudoalteromonas genus was found in both live-feed cultures (14.8% and 15.8%, respectively) ( Table 2; Table 3).
The isolates identified as Vibrio spp. via the analysis of their 16S rRNA gene sequences shared the phenotypic properties of the genus Vibrio, including Gram-negative, glucose fermenter, oxidase-positive, susceptible to the vibriostatic agent O-129, and able to grow in TCBS agar.
The nucleotide sequences of the isolates were deposited in GenBank under the accession numbers displayed in Tables 2 and 3.

Minimum Inhibitory Concentrations (MICs)
High levels of florfenicol resistance were observed for the isolates, with MIC values ranging from 32 to 512 µg/mL for isolates recovered from rotifer cultures (

Antimicrobial Resistance Patterns
Similar antimicrobial resistance patterns of the recovered strains were not related to specific bacterial species or the live feed sampled. Furthermore, resistant isolates from untreated and treated cultures exhibited similar antibacterial resistance patterns, as shown in Tables 4 and 5. Furthermore, similar antimicrobial indexes (ARI) were observed among the florfenicol-resistant bacteria isolated from both live-feed cultures, exhibiting ARI values of 0.79 and 0.71 for isolates recovered from rotifer and Artemia cultures, respectively. The majority of isolates from the rotifer and Artemia cultures exhibited multiresistance or simultaneous resistance to at least three classes of antimicrobials (24 out of 27 and 36 out of 38 isolates recovered from rotifer and Artemia cultures, respectively) ( Table 4; Table 5). High levels of multiresistance among florfenicol-resistant isolates from both cultures were detected, with 70.4% of isolates from rotifer cultures exhibiting simultaneous resistance to 7-8 antimicrobials, whereas a high proportion of isolates from Artemia cultures were resistant to 5-6 antimicrobials (76.3%). The reference strain Escherichia coli ATCC 25922 used for quality control exhibited a zone diameter of inhibition of 23 mm, which agrees with the values recommended by CLSI [57].

Phenicol Resistance Genes
A high occurrence of the floR gene was detected among the florfenicol-resistant isolates recovered from rotifer and Artemia cultures when compared with the occurrence of the fexA gene. A high proportion of the resistant isolates (21 out of 27 strains (77.8%)) recovered from the rotifer cultures carried the floR gene, and among them, three isolates also carried the fexA gene (Table 2). On the other hand, only one isolate carried only the fexA gene, whereas in five resistant isolates, none of the assayed genes were detected. Among the isolates recovered from the Artemia cultures, 23 out of 38 isolates (60.5%) harbored the floR gene, and among them, only one isolate (2.6%) also carried the fexA gene (2.6%). In contrast, the fexA gene was detected in only five isolates (13.2%), whereas ten isolates (26.3%) from Artemia cultures were negative for the assayed genes (Table 3). Resistant

Phenicol Resistance Genes
A high occurrence of the floR gene was detected among the floRfenicol-resistant isolates recovered from rotifer and Artemia cultures when compared with the occurrence of the fexA gene. A high proportion of the resistant isolates (21 out of 27 strains (77.8%)) recovered from the rotifer cultures carried the floR gene, and among them, three isolates also carried the fexA gene (Table 2). On the other hand, only one isolate carried only the fexA gene, whereas in five resistant isolates, none of the assayed genes were detected. Among the isolates recovered from the Artemia cultures, 23 out of 38 isolates (60.5%) harbored the floR gene, and among them, only one isolate (2.6%) also carried the fexA gene (2.6%). In contrast, the fexA gene was detected in only five isolates (13.2%), whereas ten isolates (26.3%) from Artemia cultures were negative for the assayed genes (Table 3). Resistant isolates carrying neither the floR nor fexA genes (5 and 10 isolates from rotifer and Artemia cultures, respectively) exhibited MIC values of 32 to 128 µg/mL for isolates from rotifer cultures, and MIC values of 32 to 1024 µg/mL for isolates from Artemia cultures.

Plasmid Content
Plasmid DNA was detected in 42 out of 59 florfenicol-resistant isolates examined, and from these, 14 and 28 isolates were recovered from rotifer and Artemia cultures, respectively ( Table 2; Table 3). Interestingly, a higher number of rotifer isolates were negative for plasmid content (13 out of 27 isolates), compared to the isolates recovered from Artemia cultures (4 out of 32 assayed isolates). Among the plasmid-carrying isolates, 1 and 11 isolates from rotifer and Artemia cultures, respectively, contained two plasmid bands. The number and size of plasmid bands found are shown in Tables 2 and 3. The majority of resistant isolates from rotifer and Artemia cultures contained plasmids with molecular weights ranging from 35 to 50 kb (12 and 29 isolates, respectively). The most commonly found plasmid bands had molecular weights of 40 and 35 kb (16 and 18 isolates) ( Table 4; Table 5).

Discussion
Considering the high importance of florfenicol in treating and preventing bacterial infections in Chilean aquaculture, the occurrence of bacterial resistance to this compound was investigated in live feeds used in larval cultures of red cusk eel, G. chilensis, which is considered a priority for Chilean aquaculture. In addition, the presence of the main determinants conferring resistance to florfenicol and the plasmid content of the selected resistant bacteria was investigated.
To our knowledge, this is the first report on the occurrence of antimicrobial-resistant bacteria in the live feeds that are most frequently used in marine fish larvae rearing, as most of the previous studies have attempted to reduce bacterial load in these cultures [19,[25][26][27][28][29] or characterize the bacterial species associated with the live feeds [7][8][9][10][11]13,15,16,18,19,43]. Our findings have demonstrated that live-feed cultures used in a Chilean commercial hatchery of red cusk eel have high levels of resistant bacteria in both untreated and treated live feeds, suggesting that the usage of florfenicol is not a necessary causal condition for the development and maintenance of elevated frequencies of florfenicol resistance. That is to say, mass-produced live feeds in the commercial hatchery never exposed to antibacterial therapy exhibited high levels of florfenicol-resistant bacteria. Thus, the well-accepted paradigm that a selective pressure is needed for the selection and persistence of antibacterial resistance in marine fish aquaculture is not supported. Previous findings obtained from a Chilean scallop hatchery, in which high levels of antibacterial resistant bacteria were detected in untreated scallop larvae [66], did support the paradigm.
In commercial hatchery operations, the practice of rotifer batch cultivation characterized by high rotifer densities and high loads of organic matter favors the proliferation of copiotrophic bacteria, such as Vibrio spp. Antibacterial treatments were shown to be ineffective at reducing the bacterial load in live feed, and they may even increase the occurrence of vibrios exhibiting antibacterial resistance. In this study, treatments of live-feed cultures with florfenicol were not able to reduce the bacterial load, but most probably led to the emergence of antibiotic-resistant bacteria, thus impeding the establishment of normal non-pathogenic bacteria.
The high levels of florfenicol-resistant bacteria in both live feeds used in marine fish larval culture indicates a widespread antibacterial resistance within the microbiota associated with red cusk eel farming. As was observed, live-feed cultures treated with florfenicol exhibited no significant reductions in their bacterial content and resistant bacteria, and most of the culturable levels observed during the fish larval rearing period remained high (>10 6 CFU/g). In a previous study, no significant differences (p < 0.05) in the culturable counts of heterotrophic bacteria and Vibrio spp. between untreated and oxolinic-acid-treated rotifer cultures were observed, resulting in the conclusion that antibiotic-treated rotifers used to feed reared larvae of blue-fin seabream are not needed to improve larval survival [67]. By contrast, the bacterial loads on both rotifers and Artemia were reduced with antibiotics [10], but until now, effective bacterial reduction in mass-produced rotifers remains an unsolved issue. Having previously reported that the use of rotifers with reduced bacterial load improved the survival of turbot larvae during first feeding [10], it appears to be highly important to reduce the bacterial load of the rotifer culture prior to use in red cusk eel culture.
However, it is important to consider that variations in several aspects of florfenicol administration in live feeds, such as the purity of antibiotic used, administered concentration by volume, and routes and frequency of administration of florfenicol could preclude the efficacy of antibiotic treatment to significantly reduce bacterial loads in live feeds. To date, there exists only one study of the use of florfenicol that reduced the bacterial load in Artemia culture, using a dose of 300 mg/L of florfenicol and resulting in an important reduction of Vibrio counts. In this study, when florfenicol was administered, Vibrio counts were reduced from 1.20 × 10 9 ± 9.00 × 10 8 CFU/g to 2.00 × 10 5 ± 7.00 × 10 4 CFU/g and from 4.50 × 10 8 ± 2.40 × 10 8 CFU/g to 3.10 × 10 7 ± 7.00 × 10 6 CFU/g of Artemia nauplii (capsulated and decapsulated cysts, respectively) [68]. However, the potentially pathogenic species, V. alginolyticus and V. parahaemolyticus remained in the treated Artemia naupli [68]. Vibrio counts of treated Artemia observed in this study were similar to those observed in Artemia nauplii (decapsulated cysts), but lower than those observed in capsulated cysts in the previous study; however, it must be noted that in this study, a much lower concentration of florfenicol was administered (20 mg/L versus 300 mg/L).
Despite the previous observations, antibiotic treatment can reduce bacterial load in live feeds but cannot prevent the live feeds from being recolonized within a short period [27], thus suggesting the need to find other therapeutic alternatives. In addition, well-documented toxic effects of florfenicol on fish and crustacean larvae [69][70][71][72] indicate the urgent need for considering the use of alternative strategies to control pathogenic bacteria in live feeds administered to marine fish aquaculture, including bacteriophages, probiotics, and natural products [42,73,74]. In a recent study, probiotics, organic acids, and essential oils were demonstrated to be effective at controlling the pathogenic activity of vibrios on A. franciscana nauplii [75]. Thus, using live feeds as carriers for non-antibiotic therapeutic applications in marine fish cultures appears to be a very attractive option.
It has been argued that live feeds carry a low diversity of associated antibiotic-resistant microbiota, composed mostly of vibrios, which are most probably transferred to the fish larval culture, causing detrimental effects on the reared larvae [10,11]. Although no information on florfenicol-resistant species associated with live feeds used in marine fish larval culture is currently available, the dominance of Vibrio observed in this study are in accordance with other studies showing Vibrio species as the dominant component of the microbiota in the routine production of the rotifer Brachionus plicatilis [8,76]. Furthermore, in a previous study, Vibrio spp. were reported as the main constituents of the bacterial microbiota in the gut of the turbot larvae, and the authors concluded they were probably introduced by the rotifers [11].
Although phylogenetic analysis based on 16S rRNA gene sequences confirmed that 53 out of 65 isolates belonged to the genus Vibrio, 16S rRNA sequence analysis is unable to resolve closely related species, such as those clustered in the Vibrio group [77]. In this study, most of the studied isolates identified as Vibrio spp. (41 out of 53 isolates) exhibited low percentages of identity (≤99.8%). Thus, resistant isolates were identified only to the genus level, and further analysis, such as multilocus sequence analysis (MLSA) using various housekeeping genes or whole-genome sequencing (WGS), is required to accurately define the taxonomic affiliations of Vibrio isolates, especially at the species level [78,79].
This study demonstrated that live feed is a main source of antibiotic-resistant Vibrio in the red cusk eel culture, which is a highly significant risk to the health status of reared fish larvae fed with rotifer and Artemia considering that previous studies have demonstrated the pathogenic role of various Vibrio species in the experimental culture of this species in Chile [3][4][5]. However, further studies are required to elucidate the impact on the microbiota of the larvae reared using these live feeds.
To our knowledge, no studies of resistant microbiota and the associated antibiotic-resistance genes (ARGs) of reared marine fish in Chile have been undertaken despite the intensive use of antibiotics in this industry, which are mainly used in the larval culture and the role of live feeds as carriers of bacteria to reared fish larvae. However, the prevalence and concentration of ARGs, as well as their sources and spreading pathways in marine aquaculture, are little understood. Thus, the incidence of antibacterial resistance and the characterization of related genes is an urgent necessity for this industry, but information on the occurrence of resistance genes in farmed marine fish species in Chile is still very scarce.
As was previously described [46], bacterial resistance to florfenicol is primarily mediated by the floR gene, which is a specific drug exporter that confers resistance to florfenicol and chloramphenicol [80], and is widely disseminated among Gram-negative bacteria from animal agricultural sources [81][82][83]. The presence of floR has previously been reported in bacteria isolated from mucus and intestinal contents of farmed salmon, as well as lake water and sediments associated with salmon farming in Chile [45,61]. However, this is the first report of the presence of floR and fexA genes in non-salmonid marine fish aquaculture in Chile. It is interesting to note the high incidence of the floR gene among isolates recovered from both live-feed cultures (67.7%) is in accordance with isolates recovered from freshwater and marine environments impacted by salmon farms in Chile [45,61]. Furthermore, this is the first report of floRand fexA-carrying Vibrio spp. isolated from live feed used in marine fish larval culture, prompting the necessity of implementing continuous monitoring and sanitary control of live feeds to avoid the spread of these resistance genes in the hatchery.
Furthermore, in a previous study, the fexA gene was detected in a high number of bacteria, which were mainly identified as Pseudomonas and were recovered from the fecal content of salmon reared in a Chilean marine farm that made intensive use of antibacterials [61], contrasting with the low occurrence of fexA-carrying isolates observed in this study. These authors concluded that the gut microbiota of medicated farmed salmon could serve as a perfect reservoir for genes encoding for resistance to florfenicol, such as floR and fexA [61], but that these genes had not thus far been reported in non-salmonid farmed species.
From this study, a high number of isolates exhibited a co-resistance to oxytetracycline, the second-most frequently used antibacterial in Chilean aquaculture. This concurs with previous studies that have reported an association of the floR gene with tetracycline resistance. Dang et al. [84] detected the floR gene associated with several tetracycline-resistant isolates from aquaculture in China, whereas Gordon et al. [83] found that the floR gene was linked to a tetracycline resistance gene in an Aeromonas bestiarum isolate recovered from a freshwater environment. This fact might preclude the efficacy of the therapeutic use of oxytetracycline in marine fish larval culture.
Considering that none of the assayed resistance genes were detected in 39.5% of the studied florfenicol-resistant isolates, we could conclude that other mechanisms, such as enzymes or other exporter systems, could be responsible for the exhibited resistance; however, further research is required to elucidate the mechanisms conferring the florfenicol resistance detected in these isolates.
Finally, florfenicol-resistance-encoding genes, mainly floR, have frequently been associated with plasmids carried by human and animal pathogenic bacteria [85][86][87][88][89], thus the co-presence of this gene and plasmid elements in a high number of the studied isolates (71.2%) provides very favorable conditions for the genetic exchange among bacteria comprising the live feed and fish larval gut microbiota. This increases the feasibility of the spread of florfenicol-encoding genes to bacterial pathogens. However, the ability of detected plasmids and florfenicol-encoding genes to be transferred horizontally was not addressed in this study and remains to be investigated to properly determine the potential risk of transfer of these elements to the fish larvae microbiota.
Thus, it is critical for the marine fish aquaculture industry in Chile to monitor the emergence and spread of resistance to florfenicol and the associated main components of the resistome and mobilome to avoid future therapy failures in marine fish larval cultures.

Conclusions
The results of this study have demonstrated that live feed administered to reared larvae played an important role as a major source of antimicrobial-resistant bacteria and associated determinants in the commercial culture of red cusk eel larvae. In addition, the study revealed that florfenicol-resistant microbiota from live feed administered to reared fish larvae mainly belonged to the Vibrio and Pseudoalteromonas genera and that they exhibited co-resistance to oxytetracycline, streptomycin, and co-trimoxazole. Furthermore, the high prevalence of plasmids among floR-carrying bacteria suggests that these bacteria may contribute to the increased spread of bacterial resistance to florfenicol and other antibacterials in environments associated with these cultures.