315

Journal of Food Protection, Vol. 63, No. 3, 2000, Pages 315±321 Copyright ᮊ, International Association for Food Protection

Psychrobacters and Related in Freshwater Fish

CEÂ SAR J. GONZAÂ LEZ, JESUÂ S A. SANTOS, MARIÂA-LUISA GARCIÂA-LOÂ PEZ,* AND ANDREÂ S OTERO

Department of Food Hygiene and Food Technology, Veterinary Faculty, University of LeoÂn, E-24071-LeoÂn, Spain

MS 99-109:Received 27 April 1999/Accepted 7 October 1999

ABSTRACT

Three phenotypic identi®cation systems were employed to identify 106 strains of gram-negative, nonmotile, aerobic Downloaded from http://meridian.allenpress.com/jfp/article-pdf/63/3/315/1673422/0362-028x-63_3_315.pdf by guest on 29 September 2021 bacteria obtained during iced storage of wild (Salmo trutta and Esox lucius) and farmed (Oncorhynchus mykiss) freshwater ®sh. Using diagnostic tables and computer-assisted identi®cation, the isolates were Psychrobacter (64 strains), Acinetobacter (24 strains), Moraxella (6 strains), Chryseobacterium (5 strains), Myroides odoratus (2 strains), Flavobacterium (1 strain), Empedobacter (1 strain), and unidenti®ed (3 strains). Overall similarities of all strains were determined for 108 characters by numerical analysis (simple matching coef®cient of similarity [S] and clustering by unweighted pair group average linkage [UPGMA]). At the 77% similarity level, 92 strains formed nine major clusters (3 or more strains) and four small clusters (2 strains). Cluster 1 (25 isolates divided into two main subclusters) could be assigned to Psychrobacter phenylpyruvicus, clusters 2 and 3 (26 isolates) were designated as , and clusters 4 (3 isolates) and 7 (4 isolates) were identi®ed as Psychrobacter urativorans and Psychrobacter spp., respectively. Clusters 5 (®ve isolates), 6 (three isolates), and 9 (®ve isolates) were labeled as Acinetobacter spp., Acinetobacter johnsonii, and Acinetobacter lwof®i, respectively. Cluster 8 (12 isolates), with a high resemblance to Thornley's phenon 4 (a heterogeneous group of bacteria isolated from poultry and related to Acinetobacter), remained unnamed. The restriction pattern was identical for strains grouped into clusters 2 and 3 (P. immobilis) but was different for the remaining Psychrobacter isolates. A large proportion of isolates belonging to the family were closely related. Psychrobacters and A. johnsonii were present in freshly caught ®sh and river water. In the latter stages of storage, P. phenylpyruvicus and acinetobacters tended to decrease, whereas P. immobilis increased.

Spoilage of chilled raw ®sh stored in air is a result of ice, and samples from the three latter sampling sites were taken enzymatic changes (autolysis), nonenzymatic reactions at regular intervals (days 3, 6, 9, 12, and 15). Rainbow trout spec- (rancidity), and metabolic activity of bacteria, the latter be- imens were collected from two commercial ®sh farms located on  ing particularly important (7). Although available infor- river systems (Duerna and Orbigo; water temperature between 7 Њ mation on spoilage bacteria of freshwater ®sh is limited, it and 11 C) within the province of LeoÂn (northwest of Spain) dur- appears that, like in marine ®sh, Pseudomonas spp. are ing April to July 1997. Wild brown trout specimens, provided by one ¯y ®sherman, also were caught during April through July dominant (13, 15, 38). Gram-negative, nonmotile, aerobic 1997 in nine rivers (Selmo, Eria, Esla, CuruenÄo, TorõÂo, Sil, Burbia, rods and coccobacilli, identi®ed as acinetobacters, morax- Porma, and O rbigo; water temperature between 5 and 15ЊC) of ellae, psychrobacters (Moraxella-like), and ¯avobacteria, the same province. Pike was obtained from Esla and Porma rivers also occur but their relative incidence and role in spoilage (water temperature between 8 and 12ЊC). are dif®cult to establish due, in part, to their confusing clas- Water samples (200 ml) were collected simultaneously with si®cation and the consequent misidenti®cation (10). the ®sh from approximately a 50-cm depth using a sterile glass The aim of this work was to identify the genera and bottle. species of gram-negative, nonmotile, aerobic bacteria that were associated with freshwater ®sh and their surrounding Strains. Tryptone soya agar (Oxoid, Basingstoke, UK) was environment in accordance with the new classi®cations. used to count aerobic mesophilic ¯ora at 30ЊC for 2 days and The behavior of these organisms during storage and their aerobic psychrotrophic ¯ora at 7ЊC for 10 days. A Harrison disk spoilage potential are also discussed. (16) was used for random selection of 979 strains. The 106 strains (50 mesophiles and 56 psychrotrophs) included in this study were MATERIALS AND METHODS the aerobic, gram-negative, nonmotile bacilli and coccobacilli found among this population. Inocula for tests were grown in tryp- Samples. Ten lots (three specimens each) of farmed rainbow tone soya broth (Oxoid) for 24 to 36 h at 25ЊC. Control plates trout (Oncorhynchus mykiss), 10 lots of wild brown trout (Salmo with tryptone soya agar were used to ensure viability and purity trutta), and 4 lots of wild pike (Esox lucius) were studied. Freshly caught or collected ®sh were sampled at the following sites: gills, of inocula. intestines, surface of skin, muscle, and surface of the body cavity. The following reference strains were included in the numer- Surfaces of skin and the body cavity were sampled by scraping ical analysis: Myroides odoratus CECT 998 (Spanish Type Cul- and rinsing; gills, intestine, and muscle were aseptically excised ture Collection, Universidad de Valencia, Spain), Moraxella os- with a scalpel. Afterwards, the specimens were stored in melting loensis CECT 460, Acinetobacter calcoaceticus CECT 441, and Psychrobacter immobilis ATCC 43116. Type strains used as the * Author for correspondence. Tel: ϩ34-987-291119; Fax: ϩ34-987- biological controls for tests were those recommended by Lanyi 291284; E-mail: [email protected]. (24). 316 GONZA LEZ ET AL. J. Food Prot., Vol. 63, No. 3

Phenotypic tests. Unless otherwise stated, cultures were in- cording to Sneath and Johnson (32). Separation indexes were cal- cubated at 25 Ϯ 1ЊC. The following tests were performed, as culated according to Sneath (31). previously described (9, 23): morphology; arrangement and Gram reaction; motility; colony appearance and pigmentation; pyoverdin Ampli®ed ribosomal DNA restriction analysis of Psychro- and pyocyanin production; ability to grow at 4, 35, and 42ЊC; salt bacter strains. Representative strains of the different clusters of tolerance (3 and 6% NaCl); growth at pH 4.5; growth on selective Psychrobacter (I, II, III IV, and VII; see Results) were cultured media (MacConkey agar, Simmon's citrate agar, and thiosulphate in tryptone soya broth, and DNA was extracted by a standard citrate bile salts sucrose agar); oxidase and catalase production; procedure (25). The 16S ribosomal DNA gene was ampli®ed by hemolytic activity on sheep blood; sensitivity to penicillin (1 IU polymerase chain reaction using primers pA and pH* (18) in a mlϪ1) and vibriostatic agent O/129 (150 ␮g); reduction of nitrate Mastercycler Personal thermal cycler (Eppendorf). Each ampli®- Њ and trimethylamine N-oxide (Sigma); urease production; decar- cation cycle consisted of denaturation for 30 s at 92 C, annealing for30sat55ЊC, and extension for 1 min at 72ЊC, followed by a boxylation of L-lysine and L-ornithine; arginine dihydrolase; de- Њ ␮ composition of hippurate; production of indole; esculin hydroly- ®nal extension cycle of 2 min at 72 C. Approximately 2 gof the ampli®ed products (ca. 1.5 kb in length) was digested with 10 sis; H2S production from cysteine; hydrolysis of tyrosine, DNA, Њ Downloaded from http://meridian.allenpress.com/jfp/article-pdf/63/3/315/1673422/0362-028x-63_3_315.pdf by guest on 29 September 2021 gelatin, casein, elastin, and starch; production of levan from su- UofHaeIII (Sigma) for6hat37C. The digestion products were crose; methyl red and Voges-Proskauer reactions; gluconate oxi- electrophoretically separated in a 2% agarose gel, stained with ethidium bromide, and viewed under UV light. dation; gas production from glucose; o-nitrophenyl-␤-D-galacto- pyranoside; lipolytic activity on Tween 20, Tween 80, tributyrin, RESULTS and egg yolk; acid production from carbohydrates (adonitol, L- arabinose, cellobiose, D-fructose, D-galactose, glycerol, glucose, Using diagnostic tables (1, 3, 4, 8, 21, 27, 35, 36) and meso-inositol, lactose, maltose, D-mannitol, D-mannose, melibi- the probability matrix (computer-assisted identi®cation ose, raf®nose, L-rhamnose, D-ribose, salicin, sorbitol, L-sorbose, (27)), 101 isolates were allocated to six genera: Psychro- sucrose, starch, trehalose, xylitol, and D-xylose); and assimilation bacter (64 strains), Acinetobacter (24 strains), Moraxella (6 of sole carbon sources (adonitol, D-arabinose, L-arabinose, cello- strains), Chryseobacterium (5 strains), Empedobacter (1 biose, D-fructose, D-galactose, glycerol, D-glucose, D-gluconate, strain), and Flavobacterium (1 strain). Three strains were meso-inositol, lactose, maltose, D-mannitol, D-mannose, L-rham- not identi®ed to the level, and two strains were iden- nose, salicin, sorbitol, sucrose, starch, trehalose, xylitol, n-butanol, ti®ed as Myroides odoratus. ethanol, L-arginine, L-cysteine, L-lysine, glycine, and sodium for- mate). The average probability of error (P) was 3.4%, which would not produce important distortion of the taxonomic Taxonomic schemes. Conventional tests employed for the structure. The cophenetic correlation value was 0.816. At identi®cation of the strains were mainly those recommended by the 77% similarity level (S), 92 strains (90 isolates and 2 Doern (8), Juni (21), Prieto et al. (28), Towner (36), Vandamme reference strains) formed 13 clusters (Fig. 1), 9 of them et al. (37), Bernardet et al. (1), and Bowman et al. (3, 4). containing 3 or more strains. There were four small clusters Computer-assisted identi®cation. A probability matrix for (2 strains) and 18 unclustered strains, of which 2 were ref- the identi®cation of species of gram-negative, nonmotile, aerobic erence and 16 were ®eld strains. Cluster 1 could be as- bacteria constructed by Prieto (27) was also used. This identi®- signed to Psychrobacter phenylpyruvicus. It included typi- cation matrix incorporates 35 taxa and 57 characters. No signi®- cal strains of this species (14 strains) grouped together with cant overlap was found at the 5% level, and the identi®cation P. immobilis (four strains), Moraxella spp. (four strains), score for the hypothetical median organism exceeded 0.90 for all and three oxidase-negative isolates that ®t the description taxa. of Acinetobacter (Acinetobacter johnsonii and A. calcoac- Numerical analysis. Phenotypic data identical for all strains eticus)(2). Strains within this cluster were divided into two were excluded from the numerical analysis. The remaining 108 main subclusters. One of them comprised the P. phenyl- characters were coded as negative (0), positive (1), or doubtful pyruvicus isolates (from pike and its habitats) linked at 86% (2). The simple matching coef®cient of Sokal and Michener (34) S and the other one, linked at 82% S, comprised oxidase- was used, and clustering was achieved by unweighted pair group positive (P. immobilis and Moraxella spp.) and oxidase- average linkage (33). The software employed was SPSS.PCϩ negative isolates from identical sources. Clusters 2 and 3 V3.1 (SPSS Inc., Chicago, Ill.) for clustering and BASIC com- were designated as P. immobilis. Cluster 2 contained 13 puter programs (27) for similarity coef®cients. The correlation co- isolates consisting of 11 P. immobilis isolates, 1 P. phen- ef®cient between the similarity matrix and the levels on the den- ylpyruvicus isolate, and 1 Moraxella spp. isolate. Cluster 3 drogram derived from that matrix (cophenetic correlation) was (13 isolates and the reference strain) was made up exclu- calculated with the NTSYS-PC V1.80 program (Exeter Software, Setauket, New York). Character frequency tables were generated sively of P. immobilis. Cluster 3 strains differed from clus- by using the LOTUS 123 V2.0 program (Lotus Development ter 1 strains in showing lipolytic activity on all of the tested Corp., Cambridge, Mass.) and were used to determine the most substrates. Cluster 4 contained three strains linked at 85% discriminatory characters. All tests were repeated on 10% of the S. All showed the following traits: coccobacilli, oxidase strains. The average probability of error (P) was calculated ac- positive, inability to grow on MacConkey agar, inability to

→

FIGURE 1. Simpli®ed dendrogram showing the phena formed at the 77% similarity level and unweighted group average linkage clustering. J. Food Prot., Vol. 63, No. 3 PSYCHROBACTER IN FRESHWATER FISH 317 Downloaded from http://meridian.allenpress.com/jfp/article-pdf/63/3/315/1673422/0362-028x-63_3_315.pdf by guest on 29 September 2021 318 GONZAÂ LEZ ET AL. J. Food Prot., Vol. 63, No. 3

TABLE 1. Evolution of gram-negative, nonmotile, aerobic bac- ters tended to decrease in the latter stages of storage, where- teria during chill storage in air of freshwater ®sh (trout and pike) as the recovery of P. immobilis increased. Moraxella spp. Number of strains on and P. urativorans were only detected after several days of sampling day storage. During storage, the recovery of Psychrobacter, Aci- Family/genera/species 03691215 netobacter, and Moraxella was not clearly related to the Moraxellaceae sampling site. Psychrobacters (P. immobilis, P. phenylpy- Psychrobacter ruvicus, and Psychrobacter spp.) and A. johnsonii were pre- P. phenylpyruvicus 3 1 5 6 1 2 sent in freshly caught trout and pikes (skin and gills) as P. immobilis 2 6 2 3 8 12 well as in river water (Tables 1 and 2). P. urativorans 2 1 Psychrobacter spp. 1 1 DISCUSSION Moraxella spp. 1 1 3 1 Our results suggest that most of the gram-negative, Downloaded from http://meridian.allenpress.com/jfp/article-pdf/63/3/315/1673422/0362-028x-63_3_315.pdf by guest on 29 September 2021 Acinetobacter nonmotile, aerobic bacteria developing during aerobic iced A. johnsonii 536 1 storage of wild and farmed ®sh from temperate water be- A. lwof®i 1 1 1 long to the family Moraxellaceae. The data obtained from A. calcoaceticus 2 Acinetobacter spp. 1 numerical analysis demonstrated that a large proportion of Flavobacteriaceae the isolates identi®ed as psychrobacters, moraxellae, and Chryseobacterium balustinum 3 11 acinetobacters were phenotypically close. This observation Empedobacter brevis 1 is consistent with the results of other workers obtained by Flavobacterium aquatile 1 using chemotaxonomic and genetic methods. Thus, the Myroides odoratus 1 qualitative similarity in fatty acid composition was ob- Unidenti®ed 1 served as early as 1979 by Nishimura et al. (26); wax es- ters, represented by their fatty alcohols, have been found in the chromatographic pro®les of the three genera and ``false produce lipases or to oxidize carbohydrates, growth at 4 neisseria'' (5), and DNA-rRNA hybridization studies (39) but not at 35ЊC, salt tolerance, nitrate reduction, and urease have evidenced that these organisms constitute a separate activity. Eleven compounds (i.e., arginine) were used as genotypic cluster. All this information supports the proposal carbon sources. These properties suggested that strains in of Rossau et al. (29), which advocated the assignment of this cluster ®t the description of the recently proposed spe- the genera Moraxella, Psychrobacter, and Acinetobacter to cies Psychrobacter urativorans (3). Cluster 5 was regarded the family Moraxellaceae in the ␥-subclass of the class Pro- as Acinetobacter spp. It included four isolates identi®ed as teobacteria. A. johnsonii (two strains), A. calcoaceticus (one strain), and The long and complex history of these three genera (at Acinetobacter spp. and the reference strain (A. calcoaceti- that time, placed in the family Neisseriaceae) was reviewed cus CECT 441). Cluster 6 comprised three isolates that met by Juni (21). Phylogenetic studies showed that the above the description of A. johnsonii (2). The four strains in clus- cited organisms should be transferred to a separate family. ter 7 were designated as Psychrobacter spp. They resem- On the basis of DNA-rRNA hybridization data, Rossau et bled those of cluster 4 but utilized signi®cantly fewer car- al. (29) proposed the family Moraxellaceae, which contains bon sources (only 3 of the 28 compounds tested). Cluster the genus Acinetobacter and the Moraxella-Psychrobacter 8 was heterogeneous, comprising both oxidase-positive and group. The latter was divided into four subgroups: the au- oxidase-negative bacilli and coccobacilli, previously iden- thentic moraxellae, the Moraxella osloensis subgroup, the ti®ed as P. phenylpyruvicus (®ve strains) and A. johnsonii Moraxella atlantae subgroup, and the Psychrobacter-Mor- (four strains), as well as three yellow-pigmented isolates, axella phenylpyruvica subgroup. Catlin (6) proposed the two of them belonging to the emended family Flavobac- new family Branhamaceae for Moraxella and Branhamella. teriaceae (Flavobacterium aquatile and Empedobacter P. immobilis was the only species assigned to the genus brevis) and one unidenti®ed isolate. Cluster 9 consisted of Psychrobacter by Juni and Heym (23). However, strains ®ve isolates, four with high resemblance to the description obtained from proteinaceous foods (28, 30) revealed con- of Acinetobacter lwof®i and one identi®ed as Moraxella siderable variation in carbon substrate utilization. Since spp. The minor clusters (two strains) could be assigned to 1996, three new species, Psychrobacter frigidicola, Psy- P. immobilis, A. johnsonii, Chryseobacterium balustinum, chrobacter glacincola, and P. urativorans, have been de- and M. odoratus. Major features distinguishing the clusters scribed, and M. phenylpyruvica has been renamed as P. were the ability to use the following carbon sources: glu- phenylpyruvicus (3, 4). conate (separation index 18), lysine (separation index 16), In addition to P. immobilis, P. urativorans was detect- and arginine (separation index 12). The restriction pattern ed by us in rainbow trout at the end of the storage life. obtained on digestion of the ampli®ed ribosomal DNA with Note that the type strain of P. urativorans (originally named HaeIII was the same for the strains belonging to clusters 2 Microccocus cryophilus) was isolated from pork sausage. and 3 (true P. immobilis) and different for all other strains In this study, a close relationship was observed between P. of Psychrobacter. phenylpyruvicus and P. immobilis (cluster 1 to 3). P. ura- Table 1 shows that P. phenylpyruvicus and acinetobac- tivorans (cluster 4) was slightly more related to Acineto- J. Food Prot., Vol. 63, No. 3 PSYCHROBACTER IN FRESHWATER FISH 319

TABLE 2. Distribution of gram-negative, nonmotile, aerobic bacteria according to the sampling site Sampling site

Fish species/bacterial species Skin Abdominal cavity Fish muscle Water Gills Intestines

Pike P. phenylpyruvicus 5 1 5 1 1 P. immobilis 8 5 3 2 Moraxella spp. 1 1 2 A. johnsonii 1 1 2 A. calcoaceticus 1 Brown trout P. phenylpyruvicus 1

P. immobilis 1 1 1 1 Downloaded from http://meridian.allenpress.com/jfp/article-pdf/63/3/315/1673422/0362-028x-63_3_315.pdf by guest on 29 September 2021 Moraxella spp. 1 A. johnsonii 1 1 A. calcoaceticus 1 Chryseobacterium balustinum 1 1 Rainbow trout P. phenylpyruvicus 4 2 P. immobilis 5 5 5 P. urativorans 2 1 Psychrobacter spp. 1 1 2 Moraxella spp. 1 A. johnsonii 7 4 1 A. twof®i 1 1 1 Acinetobacter spp. 1 Chryseobacterium balustinum 3 Empedobacter brevis 1 Flavobacterium aquatile 1 Myroides odoratus 1 1 Unidenti®ed 1 11 bacter (cluster 6) than to P. immobilis (Fig. 1), whereas schemes and computer-assisted identi®cation) performed Psychrobacter spp. (cluster 7) joined the unnamed cluster well with strains obtained from freshwater ®sh; however, 8 at 76% S. Strains in this latter cluster showed a high numerical analysis demonstrated the high similarity of most resemblance to the description of phenon 4 of Thornley, acinetobacters (especially A. johnsonii) and the oxidase- which is a heterogeneous group of gram-negative bacteria positive strains. This con®rms previous works (21, 22) that isolated from poultry and related to acinetobacters (35). emphasized the dif®culties in delineating the of Some strains in Thornley's phenon 4 were later identi®ed gram-negative, nonmotile, aerobic coccobacilli. Cluster 9 as psychrobacters by Juni and Heym (22). The ampli®ed (A. lwof®i) deserves mentioning because it was poorly re- ribosomal DNA restriction analysis of strains of Psychro- lated (63.8% S) to the remaining clusters obtained. Using bacter suggests that the phenotypic differences observed DNA hybridization, Bouvet and Grimont (2) found that between the strains of P. immobilis included in clusters 2 species 8 and 9 (A. lwof®i) were distantly related to all other and 3 might not re¯ect an evolutionary divergence. Our Acinetobacter species and Acinetobacter strains studied by analysis, and the limited taxonomic information on psy- them. chrobacters, reinforce the view that further studies are re- There is considerable confusion over the role of Mor- quired to explore the possibility of a more detailed speci- axellaceae in the spoilage of proteinaceous foods. Their ation of this genus. Furthermore, much still needs to be relative incidence and low spoilage potential suggest that learned about its natural ecology. these bacteria are not important spoilers (11, 14), although The genus Acinetobacter is biochemically and geneti- more detailed studies (i.e., spoilage activity in foods and/ cally heterogeneous. At least 19 genospecies have been rec- or their components) are required before de®nite conclu- ognized, 11 of which can be differentiated by phenotypic sions can be reached. In this study, they constituted 9.6% properties (19). Since Acinetobacter is the only oxidase- of the total ¯ora, and the majority failed to produce typical negative genus within the family Moraxellaceae, identi®- ®sh spoilage compounds (trimethylamine and H2S, etc.). cation of unknown isolates to genus level does not appear The lipolytic activity of a large proportion of our isolates dif®cult. In contrast, phenotypic identi®cation at the geno- does not appear to be of signi®cance in fresh ®sh deterio- mic species level is generally regarded as complex, time ration since changes in the lipid fraction are mainly due to consuming, and in some cases impossible (19, 36). Two of chemical reactions (17). Most studies report that Moraxel- the identi®cation systems used by us (published taxonomic laceae decreases as spoilage progresses (10, 28). However, 320 GONZAÂ LEZ ET AL. J. Food Prot., Vol. 63, No. 3

Table 1 reveals that the behavior of these organisms de- 8. Doern, G. V. 1992. The Moraxella and Branhamella subgenera of pends on the genus and even on the species. the genus Moraxella, p. 3276±3280. In A. Balows, H. G. TruÈper, M. Dworkin, W. Harder, and K. H. Schleifer (ed.), The prokaryotes, 2nd The majority of our yellow-pigmented isolates and the ed. Springer-Verlag, New York. type strain M. odoratus formed minor clusters related to the 9. GarcõÂa-Armesto, M. R., M. Prieto, C. Alonso, M. L. GarcõÂa-LoÂpez, Moraxellaceae; however, two strains grouped together with M. C. GarcõÂa-FernaÂndez, and A. Otero. 1993. Numerical taxonomy acinetobacters and psychrobacters in cluster 8. Certain phe- of psychrotrophic bacteria isolated from raw ewes milk. J. Dairy notypic relationships between Moraxella and bacteria of the Res. 60:371±383. former Flavobacterium-Cytophaga group isolated from 10. Gennari, M., M. Alaqua, F. Ferri, and M. Serio. 1989. Characteriza- tion by conventional methods and genetic transformation of Neis- dairy products have been previously reported (20). The tax- seriaceae (Psychrobacter and Acinetobacter) isolated of fresh and onomy of yellow-pigmented, nonfermentative, gram-nega- spoiled sardines. Food Microbiol. 6:199±210. tive, nonmotile bacteria (formerly named ¯avobacteria) has 11. Gillespie, N. C., and N. M. Macrae. 1975. The bacterial ¯ora of also undergone marked changes. At present, the emended some Queensland ®sh and its ability to cause spoilage. J. Appl. Bac- family Flavobacteriaceae includes, among others, three teriol. 39:91±100. Downloaded from http://meridian.allenpress.com/jfp/article-pdf/63/3/315/1673422/0362-028x-63_3_315.pdf by guest on 29 September 2021 yellow-pigmented genera (the emended genus Flavobacte- 12. GonzaÂlez-Serrano, C. J. 1996. Asociaciones bacterianas alterantes y agentes de infecciones alimentarias en peces de agua dulce alma- rium, Chryseobacterium, and Empedobacter) that are wide- cenados a refrigeracioÂn. Ph.D. dissertation. University of LeoÂn, ly distributed in nature and are especially common in aquat- Spain. ic habitats (1, 37). The low percentages of yellow-pig- 13. Gram, L., and H. H. Huss. 1996. Microbiological spoilage of ®sh mented bacteria among the whole population studied and ®sh products. Int. J. Food Microbiol. 33:121±137. (Ͻ1%) and their metabolic characteristics indicate that 14. Gram, L., G. Trolle, and H. H. Huss. 1987. Detection of speci®c spoilage bacteria from ®sh stored at low (0ЊC) and high (20ЊC) tem- these bacteria are not important contributors to the spoilage peratures. Int. J. Food Microbiol. 4:65±72. of freshwater ®sh. 15. Gram, L., C. Wedell-Neergard, and H. H. Huss. 1990. The bacteri- The low number of Moraxellaceae (only three strains) ology of fresh and spoiling Lake Victoria nile perch (Lates niloticus). detected during the shelf life of brown trout is noteworthy, Int. J. Food Microbiol. 10:303±316. though members of this family were initially present on 16. Harrigan, W. F., and M. E. McCance. 1976. Laboratory methods in skin, gills, and intestines and in water. This fact could be food and dairy microbiology. Academic Press, Inc., London. 17. Huss, H. H. 1995. Quality and quality changes in fresh ®sh. Food related to the high levels of carnobacteria and other lactic and Agricultural Organization, Rome. acid bacteria developing in this species of ®sh (12). Yellow- 18. Hutson, R. A., D. E. Thompson, and M. D. Collins. 1993. Genetic pigmented bacteria appear to be more prevalent in farmed interrelationships of saccharolytic Clostridium botulinum types B, E, ®sh and pond water. and F and related clostridia as revealed by small-subunit rRNA gene sequences. FEMS Microbiol. Lett. 108:103±110. ACKNOWLEDGMENTS 19. Janssen, P., K. Maquelin, R. Coopman, I. Tjernberg, P. Bouvet, K. Kersters, and L. Dijkshoorn. 1997. Discrimination of Acinetobacter We thank Mr Carlos BayoÂn, one of the best ¯y ®shermen in our genomic species by AFLP ®ngerprinting. Int. J. Syst. Bacteriol. 47: region, for kindly providing us with the brown trout specimens and the Department of Animal Biology (University of LeoÂn, Spain) for providing 1179±1187. us with pikes. This work was supported by grants from the Spanish Co- 20. Jooste, P. J., T. J. Britz, and J. Haast. 1985. A numerical taxonomic misioÂn Interministerial de Ciencia y TecnologõÂa (projects ALI94-0079 and study of Flavobacterium-Cytophaga strains from dairy sources. J. ALI97-0601). Dr. GonzaÂlez-Serrano was bene®ciary of a fellowship from Appl. Bacteriol. 59:311±323. the Spanish Ministerio de EducacioÂn y Ciencia. 21. Juni, E. 1992. The genus Psychrobacter, p. 3241±3246. In A. Bal- ows, H. G. TruÈper, M. Dworkin, W. Harder, and K. H. Schleifer (ed.), REFERENCES The prokaryotes, 2nd ed. Springer-Verlag, New York. 22. Juni, E., and G. Heym. 1980. Transformation assay for identi®cation 1. Bernardet, J. F., P. Segers, M. Vancanneyt, F. Berthe, K. Kersters, of psychrotrophic Achromobacter. Appl. Environ. Microbiol. 40: and P. Vandamme. 1996. Cutting a Gordian knot: emended classi®- 1106±1114. cation and description of the genus Flavobacterium, emended de- 23. Juni, E., and G. Heym. 1986. Psychrobacter immobilis gen. nov., scription of the family Flavobacteriaceae, and proposal and Flavo- sp. nov.: genospecies composed of gram negative, aerobic, oxidase- bacterium hydatis nom. nov. (basonym, Cytophaga aquatilis Strohl positive coccobacilli. Int. J. Syst. Bacteriol. 36:388±391. and Tait 1978). Int. J. Syst. Bacteriol. 46:128±148. 24. Lanyi, B. 1987. Classical and rapid identi®cation methods for med- 2. Bouvet, P. J. M., and P. A. D. Grimont. 1986. Taxonomy of the genus ically important bacteria, p. 1±65. In R. R. Colwell and R. Grigorov Acinetobacter with the recognition of A. baumanni sp nov., A. hae- (ed.), Methods in microbiology, vol. 19. Academic Press, Inc., Lon- molyticus sp nov., A. johnsonii sp nov. and A. juni sp nov. and don. emended descriptions of A. calcoaceticus and A. lwof®i. Int. J. Syst. Bacteriol. 36:228±240. 25. Lawson, P., S. E. Gharbia, H. N. Shah, and D. R. Clark. 1989. Rec- 3. Bowman, J. P., J. Cavanagh, J. J. Austin, and K. Sanderson. 1996. ognition of Fusobacterium nucleatum subgroups Fn-1, Fn-2 and Fn- Novel Psychrobacter species from antarctic ornithogenic soils. Int. 3 by ribosomal RNA gene restriction patterns. FEMS Microbiol. J. Syst. Bacteriol. 46:841±848. Lett. 53:41±45. 4. Bowman, J. P., D. S. Nichols, and T. A. McMeekin. 1997. Psychro- 26. Nishimura, Y., H. Yamamoto, and H. Iizuka. 1979. Taxonomical bacter glacincola sp. nov., a halotolerant, psychrophilic bacterium studies of Acinetobacter. Z. Alleg. Mikrobiol. 19:307±308. isolated from antarctic sea ice. Syst. Appl. Microbiol. 20:209±215. 27. Prieto, M. 1994. Integrated software for probabilistic identi®cation 5. Bryn, K., E. Jantzen, and K. Bovre. 1997. Occurrence and pattern of microorganisms. Comp. Appl. Biosci. 10:71±73. of waxes in Neisseriaceae. J. Gen. Microbiol. 102:33±43. 28. Prieto, M., M. R. GarcõÂa, M. L. GarcõÂa, A. Otero, and B. Moreno. 6. Catlin, B. W. 1991. Branhamaceae fam. nov., a proposed family to 1992. Numerical taxonomy of gram-negative, nonmotile, nonfer- accomodate the genera Branhamella and Moraxella. Int. J. Syst. mentative bacteria isolated during chilled storage of lamb carcasses. Bacteriol. 41:320±323. Appl. Environ. Microbiol. 58:2245±2249. 7. Connell, J. J. 1990. Control of ®sh quality. Blackwell Fishing New 29. Rossau, P., A. van Landschoot, M. Gillis, and J. Deley. 1991. Tax- Books, Oxford. onomy of Moraxellaceae fam. nov., a new bacterial family to ac- J. Food Prot., Vol. 63, No. 3 PSYCHROBACTER IN FRESHWATER FISH 321

comodate the genera Moraxella, Acinetobacter and Psychrobacter 35. Thornley, M. J. 1967. A taxonomic study of Acinetobacter and re- and related organisms. Int. J. Syst. Bacteriol. 41:310±319. lated genera. J. Gen. Microbiol. 49:211±257. 30. Shaw, B. G., and J. B. Latty. 1988. A numerical taxonomy study of 36. Towner, K. J. 1992. The genus Acinetobacter, p. 3137±3143. In A. non-motile non-fermentative gram negative bacteria from foods. J. Balows, H. G. TruÈper, M. Dworkin, W. Harder, and K. H. Schleifer Appl. Bacteriol. 65:7±21. (ed.), The prokaryotes, 2nd ed. Springer-Verlag, New York. 31. Sneath, P. H. A. 1978. Identi®cation of microorganisms, p. 1±31. In 37. Vandamme, P., J. F. Bernardet, P. Segers, K. Kersters, and B. Holmes. J. R. Norris and M. H. Richmond (ed.), Essays in microbiology. John 1994. New perspectives in the classi®cation of the Flavobacteria: Wiley and Sons, Inc., New York. description of Chryseobacterium gen. nov., Bergeyella gen. nov., and 32. Sneath, P. H. A., and R. Johnson. 1972. The in¯uence of numerical Empedobacter nom. rev. Int. J. Syst. Bacteriol. 44:827±831. taxonomy similarities of errors in microbiological tests. J. Gen. Mi- 38. Venugopal, V. 1990. Extracellular proteases of contaminant bacteria crobiol. 72:377±392. 33. Sneath, P. H. A., and R. R. Sokal. 1973. Numerical taxonomy. The in ®sh spoilage: a review. J. Food Prot. 53:341±350. principles and practice of numerical classi®cation. W. H. Freeman 39. Veron, M., A. LenvoiseÂ-Furet, C. CousteÂre, C. Ged, and F. Grimont. and Co., San Francisco. 1993. Relatedness of three species of ``False Neisseriae'', Neisseria 34. Sokal, R. R., and C. D. Michener. 1958. A statistical method for caviae, Neisseria cuniculi, and Neisseria ovis by DNA-DNA hy- evaluating systematic relationships. Kansas Univ. Bull. 38:1409± bridizations and fatty acid analysis. Int. J. Syst. Bacteriol. 43:210± 1438. 220. Downloaded from http://meridian.allenpress.com/jfp/article-pdf/63/3/315/1673422/0362-028x-63_3_315.pdf by guest on 29 September 2021