Chloramphenicol and florfenicol susceptibility of fish-pathogenic bacteria isolated in France: comparison of minimum inhibitory concentration, using recommended provisory standards for fish bacteria


  • C. Michel,

    1. Unité de Virologie et Immunologie Moléculaires, Equipe Infection et Immunité des Poissons, Institut National de Recherche Agronomique, Centre de Recherches de Jouy-en-Josas, Jouy-en-Josas cedex, France
    Search for more papers by this author
  • B. Kerouault,

    1. Unité de Virologie et Immunologie Moléculaires, Equipe Infection et Immunité des Poissons, Institut National de Recherche Agronomique, Centre de Recherches de Jouy-en-Josas, Jouy-en-Josas cedex, France
    Search for more papers by this author
  • C. Martin

    1. Unité de Virologie et Immunologie Moléculaires, Equipe Infection et Immunité des Poissons, Institut National de Recherche Agronomique, Centre de Recherches de Jouy-en-Josas, Jouy-en-Josas cedex, France
    Search for more papers by this author

Christian Michel, Unité de Virologie et Immunologie Moléculaires, INRA-CRJJ, Domaine de Vilvert 78352 Jouy-en-Josas cedex, France (e-mail:


Aim: To investigate the distribution of antimicrobial resistance to phenicols in the fish pathogenic bacteria Aeromonas salmonicida, motile Aeromonas, Yersinia ruckeri, lactic bacteria and the nutritionally fastidious Flavobacterium psychrophilum. The last species was screened on two media (diluted Mueller–Hinton and peptone-enriched Anacker and Ordal), both supplemented with horse serum.

Methods and Results: Minimal inhibitory concentration (MIC) assessment, using the agar dilution method according to proposed standards, confirmed that chloramphenicol resistance was more frequent and expressed at higher levels than florfenicol resistance. A significant resistant population, highlighted by the bimodal distribution of MICs, was detected only for chloramphenicol in A. salmonicida. No link could be found with the geographical origin of the isolates or fish species. Other cases of resistance appeared randomly distributed or related to the natural properties of the bacterial species. Although the two media used for testing F. psychrophilum resulted in comparable performances in dilution methods, Anacker and Ordal was more adapted to disc diffusion tests.

Conclusion: Despite wide use, resistance to florfenicol does not seem to occur frequently in French fish farms.

Significance and Impact of the Study: It is important to maintain a surveillance, as development of florfenicol resistance has occasionally been documented. For this purpose, and for the species studied in this work, the recently proposed standards appear generally well-adapted.


The veterinary use of antimicrobial drugs has been severely questioned in the past two decades (Gustafson and Bowen 1997), and because of the social and ecological significance of water resources, aquatic animal farming is particularly implicated. Among different reasons put forward, the potential toxicity of drugs or their metabolite residues to consumers has found an acceptable solution with the legal definition of maximal residue limits (MRL) (Gustafson and Bowen 1997). The impact of treatments on the aquatic environment is also rightly suspected, but it still has to be further documented (Guardabassi et al. 2000; Schmidt et al. 2000). Presently, the risk of selecting drug resistance in fish-pathogenic or water-associated bacteria, some of them being likely to become involved in human health, probably remains the most critical concern in fish health control practice. Medical authorities are especially vigilant in this domain, and the resulting debates have led the legislator to dictate requirements which clearly limit the veterinary use of antimicrobial drugs (Gustafson and Bowen 1997). Some products were prohibited, others were replaced with chemical species restricted to animal use.

Phenicols are illustrative of these trends. Because of its severe toxicity and its importance in the treatment of typhoid fever, chloramphenicol was strictly prohibited for animal use in most countries in the mid-1990s. At the same time, a fluorinated derivative, florfenicol, had become available. As it did not present the inconvenience of chloramphenicol and had no application in human medicine, it rapidly became popular in several animal industries, including aquaculture after Fukui et al. (1987) established its efficacy in controlling fish diseases in Japan. In contrast to chloramphenicol, little was known about the acquisition of resistant mechanisms to florfenicol by bacteria, so that the discovery of multidrug-resistant Salmonella enterica Typhimurium DT104 strains (Arcangioli et al. 1999; Bolton et al. 1999) harbouring a florfenicol resistance gene (floR) associated with a chromosomal genomic island (Briggs and Fratamico 1999; Boyd et al. 2001, 2002) triggered a series of investigations. It was found that a related gene had been formerly described in the marine fish pathogen Photobacterium damselae ssp. piscicida (Kim and Aoki 1996; White et al. 2000).

Then, it became of interest to check whether similar gene or multidrug-resistance could be acquired by other fish bacteria, namely in areas where florfenicol was frequently used. Partial data were already available from general surveys conducted in other countries and addressing the main drugs used in fish treatments (Rangdale et al. 1997; Bruun et al. 2000). Farming traditions and the panel of licensed drugs may vary according to the country, and it appeared critical to obtain specific information from France, where chloramphenicol and florfenicol had been subsequently used to combat the rainbow trout fry syndrome (RTFS) caused by Flavobacterium psychrophilum. Considering that a selective effect of antimicrobial use on the emergence of resistant fish bacteria has been documented in several reports (Aoki et al. 1981; Kim and Aoki 1993), a further objective was to track the possible evolution of bacterial resistance in fish farms, focusing on the comparison of chloramphenicol and florfenicol resistance in selected fish-pathogenic species. Finally, in fastidious-growing F. psychrophilum, standards recently adapted for fish bacteria susceptibility testing (Alderman and Smith 2001) recommend the use of diluted Mueller–Hinton (MH) medium in accordance with a report of Hawke and Thune (1992). These standards are provisional, and as improving media for producing viable cultures has also been recently suggested (Michel et al. 1999), there was an opportunity to assess a possible influence of the medium on the results, both in minimum inhibitory concentration (MIC) assay and in the classical disc diffusion method.

Materials and methods

Bacterial strains

Five species or groups of species (Table 1) frequently associated with fish were studied: Aeromonas salmonicida, Yersinia ruckeri and F. psychrophilum are major pathogens worldwide. Although more often opportunistic, motile aeromonads (mainly A. hydrophila and A. sobria) may be isolated from almost all cases of fish diseases. The occurrence of Gram-positive lactic organisms with clinical manifestations has been increasing in the last 15 years, and Carnobacterium piscicola, Vagococcus salmoninarum, and the recently appeared Lactococcus garvieae are now a common source of concern in French fish farms. Most of the strains were collected from clinical cases during the last 25 years and preserved by freeze-drying or freezing at −70°C in the Jouy-en-Josas Microbial Collection. Others were actively prospected during the course of the survey, with the help of field operators and diagnostic laboratories. Two reference strains, Escherichia coli CIP 7624 and Staphylococcus aureus CIP 7625 were also included as controls in antimicrobial testing studies.

Table 1.  Numbers and origin of the bacterial species and isolates included in the study. Motile aeromonads are presented as a single complex because of very close properties, while lactic bacteria, in which several fish pathogenic species are well documented, are presented with more detail
SpeciesIsolatesGeographical originFish species
Aeromonas salmonicida52FranceRainbow trout (16), Atlantic salmon (3), brown trout (11), brook trout (4), hybrid ‘brook trout ×  rainbow trout’ (3), arctic char (1), umber (2), pike (1), turbot (5), unspecified (6)
Yersinia ruckeri50France (45), Italy (2), Belgium (1), Denmark (1), Idaho (1)Rainbow trout (34), pike (1), fathead minnow (2), rudd (1), sturgeon (2), turbot (9), unspecified (1)
Motile aeromonads49  
A. caviae (1), A. eucrenophila (1), Aeromonas spp. (4), A. sobria (14), A. hydrophila (29) France (42), Denmark (3), Canada (1), Thailand (2), Japan (1)Rainbow trout (13), brown trout (5), brook trout (1), cyprinids (11), ornamental fish (10), tilapia (1), catfish (1), Ophicephalus (2), cod (3), unspecified (2)
Flavobacterium psychrophilum44France (30), Denmark (5), Germany (2), Switzerland (1), Spain (2), Chile (1), USA (1), unspecified (2)Rainbow trout (38), brown trout (1), coho salmon (1), tench (1), unspecified (3)
Lactic Gram-positive bacteria40  
Carnobacterium piscicola (23) France (18), Belgium (1), Germany (4)Rainbow trout (4), brown trout (4), arctic char (5), pike (1), goldfish (2), ornemental fish (2), unspecified (5)
Lactococcus garvieae (4) France (2), Italy (2)Rainbow trout
L. lactis (1) FranceTilapia
Lactococcus spp. (5) France (1), Spain (4)Rainbow trout (4), discus (1)
Streptococcus iniae (2) IsraelRainbow trout (1), unspecified (1)
Vagococus salmoninarum (5) FranceRainbow trout

Strains were revived in tryptic soya broth (TSB, bioMérieux, Marcy-l'Etoile, France), except lactic bacteria and F. psychrophilum. The former was revived in brain–heart infusion (BHI, bioMérieux), the latter in Anacker and Ordal enriched to 10% tryptone (EAO) supplemented according to Michel et al. (1999) with 5% defibrinated horse serum (Difco) and trace element solution 0·02%. Incubation temperature was 22°C, with an exception for F. psychrophilum, which was grown at 18°C. Stock cultures were stored on agar medium at 4°C until experimental use.

Preparation of inocula

Preliminary tests had shown that most of the species, except F. psychrophilum, were able to grow in Mueller–Hinton (MH, Difco) medium. Working cultures were obtained in liquid MH medium after 24 or 48 h incubation at 22°C, depending on the growth rate of the strains. In the case of F. psychrophilum, EAO broth supplemented with horse serum and trace element solution was seeded and incubated at 18°C. The bacterial culture absorbance was measured on a spectrophotomer Camspec M330 (Camspec, Cambridge, UK) at 625 nm, just before adjusting the inocula in MH, both for disc diffusion tests and MIC assessment.

Disc diffusion tests

Freshly yielded cultures were adjusted to a 0·08 absorbance at 625 nm, as indicated in the standards, before inoculating MH (or alternate media when necessary) dishes using swabs. Commercial discs loaded with ampicillin (10 μg), streptomycin (10 μg), neomycin (30 μg), gentamicin (15 μg), oxytetracycline (30 μg), chloramphenicol (30 μg), erythromycin (15 μg), polymyxin B (50 μg), trimethoprim + sulphonamide (1·25 + 23·75 μg), furane (300 μg) and flumequine (30 μg) were purchased from Bio-Rad France (Marnes-la-Coquette, France). Florfenicol discs (30 μg) were supplied by Shering Plough Vétérinaire (Segré, France). Incubation was generally for 48 h, and reading was performed twice, at 24 and 48 h, using a ruler for measuring the inhibition diameters.

Antimicrobial drugs and MIC test determination

MICs were determined by the agar dilution method, as described by Alderman and Smith (2001). Chloramphenicol was purchased from Sigma (St Quentin-Fallavier, France) and florfenicol was supplied by Shering Plough Vétérinaire. Stock solutions at 2 mg ml−1 were prepared in ethanol 100%, filtered through 0·20 μm mesh Minisart filters (Sartorius, Göttingen, Germany), and volumes of 2 ml were aliquoted and frozen at −20°C for future use. After thawing, the drugs were serially diluted in the appropriate medium (MH or EAO agar only supplemented with 5% horse serum) previously melted and held at 45°C. Twofold step concentrations, ranging from 0·025 to 256 μg ml−1 of active product, were prepared as 25 ml samples and immediately poured into Petri dishes.

The dishes were inoculated using a ‘home-made’ detachable multi-stem inoculator, which consists of 26 calibrated steel stems sliding freely in a metal plate fit with a drill support. Just before use, the inoculator was sterilized 10 min under UV light (1·48 W cm−2). Ready to use bacterial suspensions (about 105 bacteria per millilitre) were distributed in the 26 wells of a previously UV-irradiated teflon plate. The arrangement of the wells was such that the steel stems could come into contact with the suspensions and take off equal volumes of 4 μl which were subsequently deposited onto the agar surface. Series of dishes corresponding to the different dilutions of the same antibiotic were inoculated with a set of 26 bacterial spots corresponding to different strains. A dish without any antibiotic was added as a control in every series of tests. Inoculated dishes were incubated at 22°C (18°C for F. psychrophilum) and read after 24, 48 h, and when necessary 72 h. The MIC was considered to correspond to the first drug dilution at which no growth of a bacterial strain was detectable.

Comparison of media for antimicrobial susceptibility of Flavobacterium psychrophilum

MICs were assessed in parallel in MH diluted to 1/7 (Alderman and Smith 2001) and in the modified EAO (both media containing horse serum 5%), using a set of 21 F. psychrophilum strains, and respecting all the other recommendations of the standards. To complete the observations, EAO medium without horse serum enrichment was also tested. Five strains were also tested using the disc method, on horse serum-supplemented EAO and diluted MH media. Cultures were grown in modified EAO broth. Suspensions were adjusted to a 0·08 absorbance at 625 nm, as with other bacteria, but incubation was 72 h at 18°C, and readings were carried out at 48 and 72 h.


The distributions of MIC values of chloramphenicol and florfenicol for the different kinds of bacteria used in this study are presented in Fig. 1. In 230 of 235 cases, florfenicol MICs generally fell within the limits of 0·25–8 μg ml−1. Only five isolates of motile aeromonads or Gram-positive bacteria were found to require higher doses to be inhibited. In all other cases, even when the range of inhibiting values appeared high enough to question a practical application of the drug to fish treatment, as was the case for Y. ruckeri strains and lactic bacteria, the distributions followed a unimodal pattern. The situation was almost comparable for chloramphenicol, although higher MIC values were generally recorded, and a good correspondence was observed between the results of individual strains (not shown). Those displaying low MICs to chloramphenicol most often displayed low MICs to florfenicol, and conversely. A noticeable exception was Aeromonas salmonicida, in which two populations differing in their mean choramphenicol MICs (100 and 1·5 μg ml−1) could be clearly recognized.

Figure 1.

Distribution of minimum inhibitory concentration (MIC) values of chloramphenicol and florfenicol in five series of bacterial strains isolated from clinically or covertly infected fish (see Table 1 for more details on motile aeromonads and lactic bacteria strains)

The susceptibility of F. psychrophilum did not differ from the general case. No resistant strain was detected, and the recorded MICs were confined within low limits, never exceeding 2 μg ml−1 of chloramphenicol and 1 μg ml−1 of florfenicol (Fig. 1). The nature of the culture medium did not seem to affect the MIC assessment significantly. Keeping in mind the magnitude of possible technical variations, generally accepted to occur within the limits of the next dilutions, Fig. 2 shows that little difference was observed among the different media. It may only be suspected that diluted MH produces slightly more optimistic values than modified AO. In diffusion tests, conversely (Fig. 3), the bacterial growth appeared to be altered on diluted MH, to such a degree that the development of confluent colonies could not always be detected, preventing any possibility of reading.

Figure 2.

Comparison of chloramphenicol and florfenicol MIC, assessed in a sample of 21 Flavobacterium psychrophilum strains using three different media (MH: Mueller–Hinton diluted to 1/7; AO: Anacker and Ordal enriched to 10 g l−1 tryptone; MH+ and AO+ were supplemented with 5% horse serum)

Figure 3.

Results of agar diffusion test performed in similar conditions, using two Flavobacterium psychrophilum strains and two different culture media: Mueller–Hinton diluted to 1/7 (MH), Anacker and Ordal enriched to 10 g l−1 tryptone (AOAE). Both were supplemented with 5% horse serum


Florfenicol was adopted by French fish farmers only in 1994, after chloramphenicol, which had been considered one of the best ways of combating the RTFS caused by F. psychrophilum (Baudin Laurencin et al. 1988), was definitively prohibited to comply with the directives of the European Community. It could therefore be expected that florfenicol resistance would be less frequent than chloramphenicol resistance, and indeed, in this survey, the bacterial species considered to represent the most serious threat for fish health (A. salmonicida, F. psychrophilum and Y. ruckeri) almost always remained susceptible to florfenicol.

Resistant strains were only observed in lactic bacteria, in which frequent intermediate levels of resistance are considered to occur naturally (Aoki et al. 1990), and in motile Aeromonas, which are generally more ubiquist and whose origins, taxonomic status and genetic background are probably more diversified. The fact that no clear-cut population of resistant Aeromonas spp. strains could really be identified and that resistant strains seemed to be distributed randomly (Fig. 1) is more evocative of the intrinsic properties of strains than of the consequence of selective pressure after massive use of the drug. A single isolate of Y. ruckeri produced florfenicol-resistant colonies in the inhibition zone in the disc diffusion test. After further investigations, no floR gene was detected in this isolate or in a highly resistant Aeromonas sobria strain (Cloeckaert, A., 1992, personal communication).

This contrasts strongly with the chloramphenicol MICs recorded in A. salmonicida strains, which displayed a bimodal distribution and revealed the existence of a large and well-delineated resistant population. In this case, a correspondence with former fish farming practices is beyond doubt, inasmuch as convincing data about the selection of chloramphenicol resistance through repeated treatments of aquatic species have been collected on several occasions (Aoki et al. 1981; Kim and Aoki 1993). The resistance mechanism, nevertheless, is not so firmly established. Although transfer experiments of A. salmonicida chloramphenicol resistance were reported (Popoff and Davaine 1971), several subsequent studies failed to demonstrate a clear correlation with the presence of plasmids (Toranzo et al. 1983; Livesley et al. 1997).

The present results are generally in good accordance with those of independent surveys conducted in other countries, although quantitative comparisons are difficult to undertake and may only be meaningful when comparable methods, referring to generally accepted standards, are applied to similar bacterial species. This was the case in some recent studies mainly relevant to F. psychrophilum (Rangdale et al. 1997; Bruun et al. 2000; Schmidt et al. 2000). Unfortunately, this agent is unable to grow on classical MH, so that several kinds of media have been proposed to meet the requirements of the bacterium and improve its growth. As other Flavobacteriaceae are suspected to express a large variability in antimicrobial tests (Jones et al. 1986), we performed our tests on several media, in order to appraise the possible influence of the medium composition. It is clear from Figs 1 and 2 that all the tested isolates were susceptible to both chloramphenicol and florfenicol. Moreover, the recorded florfenicol MIC values are close to average values reported by other workers: 1 μg ml−1 (Rangdale et al. 1997), 0·6 μg ml−1 (Bruun et al. 2000), 0·45–0·65 μg ml−1 in the present case, according to the medium used. Until now, florfenicol resistance of F. psychrophilum seems very exceptional in Europe, and only some strains with unusually high MIC values were occasionally observed by Rangdale et al. (1997) in the UK.

A medium interference was detected, but it was limited and did not change the test results significantly (Fig. 2). The diluted MH appeared to produce lower MIC values in all cases. Although physiological mechanisms in bacteria maintained in adverse conditions may be quite complex and their effects difficult to anticipate (Gilbert et al. 1990), the possibility that insufficient covering of bacterial requirements could result in a slightly increased susceptibility to drugs was apparently supported by the results of diffusion tests (Fig. 3). Compared with EAO, diluted MH clearly appeared less suitable. The different performances observed between the dilution and diffusion methods were probably related to the different sizes of inocula. Such observations may be useful in the application of the tentative standards established in 1998 (Alderman and Smith 2001). The provisionally recommended diluted MH may certainly be used without concern in dilution methods, but in the case of the disc diffusion method EAO should be preferred.

Further analysis of the recorded MIC values, referring to such factors as fish species, geographical origin, sector of activity or isolation date, did not provide very clear associations with the resistance to phenicols. Indeed, the few resistant strains observed in motile Aeromonads or lactic bacteria originated from various environments, including ornamental fish as well as farmed trout. Once again, A. salmonicida seemed to be an exception, although larger samples would probably have allowed stronger conclusions. Considering the fish species for which sufficient information was available, chloramphenicol resistance was more frequent (about 80%) in strains isolated from species highly susceptible to furunculosis (brown trout, Atlantic salmon, brook trout and its hybrids) than in strains originating from intensively reared rainbow trout (about 60%). This may simply reflect the lower susceptibility of rainbow trout and the greater need for treatments in the other species.

In conclusion, the possibility that transferable or selected resistance to florfenicol could occur following treatments appears much less significant than after chloramphenicol use. This is in accordance with the paucity of documented cases of florfenicol resistance reported in fresh water environments in Europe, although regular relapses in RTFS oblige farmers to repeatedly treat young fry for several weeks after hatchery. It is less consistent with the first description of a floR-like gene by Kim and Aoki (1996) in P. damselae ssp. piscicida. Florfenicol was then intensively used by Japanese farmers to treat yellowtail (Carangidae) reared in marine cages. We did not incorporate marine bacteria in the present work, but up to now, diffusion disc tests performed with some available Vibrio anguillarum (20) and P. damselae ssp. piscicida (five) isolates did not reveal any chloramphenicol or florfenicol resistance. It is thus likely that transferable resistance to florfenicol may occur in fish pathogens as in terrestrial species, but apparently it does not propagate as effectively as plasmidic resistance to many other drugs. Indeed, we need more information about the ability of aquatic bacterial species to exchange compatible genetic material. Although we have no strong reason to endorse alarmist attitudes, our current lack of knowledge is a strong incitement for maintaining a strict surveillance of florfenicol use and its potential consequences in aquatic animal farming.


This work was conducted as part of our participation in a programme (AQS 99/13) granted by the Direction Générale de l'Alimentation, Ministère de l'Agriculture et de la Pêche, France. The authors wish to acknowledge Carmen Alonso (Malaga University, Spain) for her precious contribution to the very first attempts at MIC determination with F. psychrophilum, Drs J-C. Raymond (SAVU, Saint-Jean-de-Védas, France), P. Rault (SYSAAF, Nouzilly, France), and G. Blanc (ENV Nantes, France), who were actively associated in the field isolate collect, Dr J-C. Abric (Schering-Plough Vétérinaire, France) who kindly supplied florfenicol products, and our colleagues E. Chaslus-Dancla and A. Cloeckaert (INRA, Nouzilly, France), for floR gene investigations.