INTERNATIONALJOURNAL OF SYSTEMATICBACTERIOLOGY, Apr. 1986, p. 257-274 Vol. 36, No. 2 0020-7713/86/020257-18$02.OO/O Copyright 0 1986, International Union of Microbiological Societies Phenotypically Based Taxonomy of Psychrotrophic Pseudomonas Isolated from Spoiled Meat, Water, and Soil GORAN MOLIN AND ANDERS TERNSTROM” Swedish Meat Research Institute, S-244 00 Kavlinge, Sweden The phenetic taxonomy of 305 strains of Pseudomonas and related organisms was numerically studied by using 215 features, including 156 assimilation tests. A total of 200 field strains were isolated from spoiling meat, and 50 strains were isolated from freshwater or soil. In addition, 55 reference strains (including 23 type strains and 4 clinical strains) were obtained. The strains clustered into 25 clusters at the 75% level when the Jaccard similarity coefficient was used. The 10 clusters that were considered significant were assigned to the Pseudomonas fragi complex (131 strains), Pseudomonas lundensis (40 strains), Pseudomonasfluorescens biovar 1 (27 strains), P.fluorescens biovar 2 (5 strains), P.fluorescens biovar 3 (6 strains), P.fluorescens biovar 4 (16 strains), Pseudomonas aureofaciens-Pseudomonas chlororaphis (3 strains), Pseudomonas aeruginosa (4 strains), Pseudomonas glathei (2 strains), and Pseudomonas mephitica (2 strains). The P. fragi complex was further divided into subclusters; the major subcluster (comprising 93 strains, including the type strain) was regarded as P. fragi sensu stricto. P. fluorescens and allied bacteria closely matched the descriptions given by Stanier et al. (J. Gen. Microbiol. 43:159-271, 1966). The characteristics for the 10 significant clusters are given. Also given are criteria which differentiate the P. fragi subclusters. The phylogenetic relationships among the meat-associated taxa were calculated, P.fluorescens biovars 2 and 3 were clearly separated from the remaining taxa. Biovar 4 is the most conservative, while biovar 3 has evolved at the highest rate. The spoilage of refrigerated meat by psychrotrophic strains and an extended battery of assimilation tests were pseudomonads results in tremendous financial losses, yet used was initiated. The major aims of this study were to very little is known about the nature and taxonomic position taxonomically clarify the phenetic relationships among P. of the spoilage organisms. Until the mid-1960s the lack of an fragi, cluster 2 of Molin and Ternstrom (20), and P. adequate classification system for pseudomonads prevented Jluorescens and to clarify the relationships between these real progress, and early investigations of the spoilage orga- groups and their external taxonomic surroundings. In addi- nisms merely added new inadequately circumscribed species tion, this study provided answers to the following two to the already vast number of almost nondescript Pseudomo- questions: (i) which taxa are dominant at a certain defined nas nomenspecies. stage of meat spoilage, and (ii) is the assimilation pattern of In 1966 Stanier et al. provided a framework for the a Pseudomonas strain reproducible when a different tech- taxonomy of Pseudomonas. The phenetic study of these nique and scoring system are used? Our results also pro- authors became the basis for a number of phenotypic and vided evidence for the phylogenetic relationships among the genotypic studies (4, 25, 27, 28, 31). However, the existence taxa. of nonfluorescent, psychrotrophic Pseudomonas strains re- sembling members of the Pseudomonas Jluorescens- MATERIALS AND METHODS Pseudomonas putida complex has been neglected. On the Isolation of meat strains. A total of 200 meat samples other hand, the occurrence of such strains in food sources (average weight, 0.5 kg) were obtained from the cutting lines has been reported frequently (5, 12, 30). Some of these of seven slaughterhouses in Sweden. The meat samples were strains have been assigned to the species Pseudomonas individually wrapped in gas-permeable polyethylene film and fragi, although quite frequently this has been done on aerobically stored at 4°C. Total aerobic plate counts were dubious grounds. The epithet of P.fragi (‘ ‘Bacteriumfragi”) determined on tryptone-glucose-yeast extract agar (Oxoid was originally used by Eichholz (9), and the organism was Ltd., London, England) (incubated at 25°C for 3 days) later considered (17) to be identical to “Pseudomonas throughout storage by using the cork borer method for fragariae” (13). In 1961 a type strain (strain ATCC 4973) was sampling (10). When the bacterial level on the meat surface proposed for the species (19). However, no adequate cir- was in the range of lo6 to 10’ CFU/cm2, one isolate was cumscription of the species was published. In 1982, two randomly picked from each tryptone-glucose-yeast extract papers concerning the phenotypic relationships among agar plate. Only one isolate was taken from each meat sample. pseudomonads isolated from meat were published (20, 29). The beef isolates were designated strains 301 through 400, Both of these studies showed that the major taxon, repre- and the pork isolates were designated strains 401 through 500. senting more than one-half of the meat strains, should be Isolation of environmental strains. Samples of fresh surface designated P. fragi because of the inclusion of the type strain water and soil were collected from 44 sites, representing an in the cluster. The second largest cluster of Molin and area of about 60,000 km2, in southern Sweden. In addition, Ternstrom (20) (cluster 2) was left unassigned, but it was three samples of tap water, two samples of adhesive bacte- suggested that this phenon might represent a new species. rial growth on stones that were fully submerged in streams However, questions concerning the homogeneity and re- (strains 529 and 548), and a sample of adhesive bacterial lationships of the two clusters still remained. Therefore, a growth on a glass surface subjected to running chlorinated new numerical, phenotypic analysis in which a new set of tap water for 1 week (strain 550) were collected. The soil samples were immediately plated onto tryptone-glucose- * Corresponding author. yeast extract agar, whereas the water samples were stored at 257 258 MOLIN AND TERNSTROM INT. J. SYST.BACTERIOL. 4°C for 10 days prior to plating. Only one isolate was taken TABLE 1. Designations and sources of strains, assimilation vigor from each sample. The selection of colonies were biased by values, and distances from the HMO and the type strain in the random picking among those colonies having diameters of P. frugi complex (cluster 1) more than 1 mm. Strains 501 through 525 and 549 were soil Distance (%) isolates, and strains 526 through 548 and 550 were water Sub- Source andlor name 2::i:!:i from: Strain cluster as receiveda isolates. The strains and sources of isolation are listed in (%) HMO Tables 1 through 4. 569T Type and reference strains. A total of 54 named reference 301 A Beef, Kavlinge 42 9 19 cultures, including 25 type strains, were used in this study. 304 A Beef, Kavlinge 43 9 16 The reference strains for the biovars of P. jluorescens were 336 A Beef, Kris tianstad 42 5 17 chosen from the strains used in a previous numerical study 429 A Pork, Kristianstad 43 7 18 (20). In addition, four clinical isolates of gram-negative, 418 A Pork, Kristianstad 44 6 15 aerobic, rod-shaped bacteria of unknown identity were ob- 399 A Beef, Kalmar 41 6 16 tained. 430 A Pork, Kristianstad 42 7 19 Three strains, “Pseudomonas bathycetes” ATCC 23597, 432 A Pork, Kristianstad 42 8 18 A Pseudomonas carboxydohydrogena DSM 1083T (T = type 334 Beef, Kris tianstad 40 7 15 Pseudomonas nautica 50418T, 350 A Beef, Savsjo 38 7 13 strain), and DSM did not 373 A Beef, Savsjo 41 7 19 grow or grew only sparsely under the experimental condi- 420 A Pork, Kristianstad 42 4 16 tions used, and the data for these strains were omitted from 473 A Pork, Varberg 43 4 16 the computations. 383 A Beef, Savsjo 40 4 16 The strains having a culture collection number were 306 A Beef, Kavlinge 42 7 14 obtained from the culture collection, except strain 573T (= 485 A Pork, Kalmar 40 9 17 CCM 3503*), which was isolated previously by Molin and 308 A Beef, Kavlinge 43 7 14 Ternstrom (20). The sources of the other reference strains 345 A Beef, Savsjo 41 8 18 are given in the footnotes to Tables through 4. 465 A Pork, Varberg 40 7 17 1 A The type and reference strains were designated strains 551 307 Beef, Kavlinge 39 11 16 437 A Pork, Tomelilla 39 11 18 through 605. The strains and culture collections from which 309 A Beef, Kavlinge 42 6 13 they were received are listed in Tables 1 through 4. 310 A Beef, Kavlinge 41 7 15 Phenotypic characterization. In all, 244 tests were per- 320 A Beef, Kavlinge 41 7 15 formed; 215 of these tests were used for the numerical 312 A Beef, Kavlinge 42 9 14 analysis. Unless indicated otherwise, the tests were per- 314 A Beef, Kavlinge 42 8 13 formed at 25°C and read after 4 days. 317 A Beef, Kavlinge 42 7 15 Cultural characteristics and growth tests. Motility was 3 19 A Beef, Kavlinge 41 9 15 examined by phase-contrast microscopy following growth in 406 A Pork fat, Kavlinge 40 7 15 nutrient broth (Oxoid) for 20 h. Apparently nonmotile strains 483 A Pork, Varberg 40 6 14 499 A Pork, Kalmar 40 7 15 were examined further after prolonged incubation, after 489 A Pork, Kalmar 42 8 18 growth at 15”C, and after growth on nutrient agar (Oxoid). 498 A Pork, Kalmar 40 7 15 Strains showing no motility were checked six times before 327 A Beef, Kris tian stad 37 9 12 being regarded as nonmotile. Spreading growth on 331 A Beef, Kristianstad 37 9 12 cytophaga agar (1) was examined after 10 days.