Taxonomy of lactic acid bacteria associated with vacuum-packaged processed meat spoilage by multivariate analysis of cellular fatty acids

International Journal of Food Microbiology 28 (1995) 89-100

ELSEVIER

l-l Journal ofFocdh4ic~

Taxonomy of lactic acid bacteria associated with vacuum-packaged processed meat spoilage by multivariate analysis of cellular fatty acids G.A. Dykes a~*, T.E. Cloete b, A. von Holy a ’ Department of M?robiologv, Uniuersity of the Witwatersrand, PO WITS 2050 Johannesburg South Africa b Department of Microbiology and Plant Pathology, University of Pretoria, Pretoria 0002 South Africa Received 21 January 1994; revised 8 August 1994; accepted 18 November 1994

Abstract The taxonomy of lactic acid bacteria from vacuum-packaged processed meats is problematic, and atypical members of the leuconostocs and the Lactobacillus sake/curuatus group are often encountered. In order to resolve this problem the cellular fatty acid (CFA) content of 61 isolates from vacuum-packaged Vienna sausages and 18 reference strains was determined by gas chromatography. The relationship between strains was derived by principal component analysis of data. The CFA profiles were highly reproducible. Although no relationships could be derived using only one or two differentiating CFAs, plots of the first two principal components based on only the six most variable CFAs allowed grouping of strains. The two genera (Leuconostoc and Lactobacillus) could not be clearly separated when analysed together, but differentiation of species within each of the genera was achieved when they were analysed independently. Examination of plots for the reference strains confirmed previously established relationships between these strains. From the plot of the Lactobacilfus sake/Lactobacillus curuatus component of the study it was found that most atypical Lactobacillus sake/curoatus strains were closely related to the typical Lactobacillus sake isolates and reference strain, while the Lactobacillus curvatus strains formed an independent grouping. A small cluster of atypical strains, however, indicated that this relationship may not be true for all these strains. Among the leuconostocs only isolates of Leuconostoc mesenteroides could be clearly differentiated. Keywords:

Cellular

fatty acid; Taxonomy;

* Corresponding author. Tel.: 27-11-716-4154.

Lactic acid bacteria;

Fax: 27-11-339-7377.

0168-1605/95/$09.50 Q 1995 Elsevier Science B.V. All rights reserved SSDI 0168-3605(94)00161-8

Lactobacillus;

Leuconostoc

90

G.A. Dykes rr al. /hr.

J. Food Microbiology 28 (1995) 89-100

1. Introduction

The cellular long chain fatty acid (CFA) content of lactic acid bacteria (LAB), such as members of the genera Pediococcus (Uchida and Mogi, 19721, Leuconostoc (Schmitt et al., 1989; Tracy and Britz, 1989) and Lactobacillus (Veerkamp, 1971) are well described. In addition the use of CFA content as a reproducible tool for differentiating LAB has also been demonstrated (Uchida and Mogi, 1973a; Rizzo et al., 1987; Shaw and Harding, 1989; Decallone et al., 1991). The taxonomy of LAB responsible for the spoilage of vacuum-packaged processed meats by phenotypic methods is problematic. Many atypical isolates previously described (Reuter, 1981; Hitchener et al., 1982; Shaw and Harding, 1985) have been assigned to new genera e.g Camobacterium (Collins et al., 1987) or new species e.g. Leuconostoc gelidum and Leuconostoc camosum (Shaw and Harding, 1989). Classification of some members of the Lactobacihs sake/curvatus group, however, remains difficult since strains that are phenotypically similar to both these species, but not clearly members of either species, are often encountered (Hastings and Holzapfel, 1987; von Holy et al., 1991). Few studies on the CFA content of LAB from vacuum-packaged processed meats have been reported. Dainty et al. (1984) and Shaw and Harding (1989) differentiated members of some species of LAB from vacuum-packaged meats on the basis of CFA profiles, but did not determine the taxonomic relationship between strains derived from these profiles. The spoilage population of vacuum-packaged Vienna sausages consists of both lactobacilli and leuconostocs (von Holy et al., 1991, von Holy et al., 1992). The Lactobacihs component of the population consisted mainly of members of the Lactobacillus sake/curvatus group with many intermediate atypical strains. The Leuconostoc component of the population appeared to fall mainly in the Leuconosfoc mesenteroides sensu-stricto group (Collins et al., 1991), but are also difficult to classify on the basis of phenotype (von Holy et al., 1991). In order to clarify the taxonomy of this population, the CFA content of representative strains was examined and analysed numerically.

2. Methods 2.1. Culture selection and maintenance

Sixty-one strains were randomly chosen from 540 predominant isolates collected in a study of terminally spoiled, South African vacuum-packaged vienna sausages (von Holy et al., 1991). The strains were isolated on MRS Agar (Oxoid, Basingstoke, UK) modified by the addition of 0.1% cysteine monohydrochloride (Univar, Krugersdorp, SA) and 0.2% potassium sorbate (Univar, Krugersdorp, SA) to enhance the growth of lactic acid bacteria and prevent growth of other groups of microorganisms such as yeasts. Strains were subsequently cultured on MRS Agar.

G.A. Dykes et al. /ht.

J. Food Microbiology

28 (1995) 89-100

91

Table 1 Culture collection strains, their sources and sample numbers used for the multivariate analysis of fatty acid profiles from 61 lactic acid bacteria Sample number a

Strain

78 79 62 63 64 65 66 61 68 77 69 70 71 72 73 74 75 16

Leuconostoc Leuconostoc Lactobacillus Lactobacillus Lactobacillus Lactobacillus Lactobacillus Lactobacillus Lactobacillus Leuconostoc Lactobacillus Leuconostoc Lactobacillus Lactobacillus Leuconostoc Leuconostoc Lactobacillus Lactobacillus

carnosum gelidum cotynefotmis coryneformis casei sake curvatus brevis plantarum citreum confusus lactis alimentarius casei paramesenteroides mesenteroides subsp. mesenteroides plantarum curvatus

Culture collection

Number

DSM=” DSM= DSMT DSM DSMT DSMT DSM= DSM= DSMT DSM DSMT DSM= DSM= DSM DSMT DSMT KC K

5576 5578 20001 20005 20011 20017 20019 20054 20174 20188 20196 20202 20249 20258 20288 20343 0634 0875

a See Figs. l-4. b Deutsche Sammlung von Mikroorganismen (Braunschweig, Germany). ’ Federal Centre for Meat Research (Kulmbach, Germany). T Type strain.

Eighteen reference strains (Table 1) were included in the study. All working cultures were maintained in chalk litmus milk (Oxoid, Basingstoke, UK) at 4°C and sub-cultured in the same medium every 2 months. 2.2. Extraction and methylation of fatty acids All strains were grown in pantothenic acid medium (yeast extract 5 g, peptone 10 g (Oxoid, Basingstoke, UK), glucose 10 g, KzHPO, 0.5 g, KHzPO, 0.5 g, MgSO, 0.2 g, NaClO.01 g, FeSO, 0.01 g, MnSO, 0.01 g, di-ammonium citrate 3.5 g (Univar, Krugersdorp, SA), (D + ) pantothenic acid 0.01 g (Sigma, St. Louis, USA)) overnight at 25°C. One ml of this culture was then used to inoculate 500 ml of the same medium and incubated at 25°C until late log phase. Cells were harvested by centrifugation at 10000 g for 10 min, washed in sterile distilled water and lyophilized. Fatty acids were extracted from 0.12 g dried cells and converted to methyl esters by the method of Augustyn and Kock (1989). All samples were prepared and analysed in duplicate. The coefficient of variation (standard deviation/mean) x 100 (Mukwaya and Welch, 1989) was calculated as a measure of reproducibility.

92

G.A. Dykes et ul. /ht.

J. Food Microbiology 28 (1995) 89-100

2.3. Gas chromatography

Fatty acid methyl esters were separated by gas chromatography (Pye Unicam, Cambridge, UK) on a J and W DB wax capillary column (30 m X 0.32 mm internal diameter, coating 0.15 pm) using a flame-ionization detector (250°C) with an injector temperature of 230°C. Running conditions were: starting temperature 80°C for 1 min, then increasing at a rate of 6”C/min to a temperature of 190°C for 5 min, followed by a further increase at 20”C/min to a final temperature of 210°C for 7 min. 2.4. Data analysis Individual CFAs for each strain were expressed as relative percentages of the total fatty acids based on peak area. Fatty acids were named according to the delta system (Augustyn and Kock, 1989) with the carboxyl carbon atom designated number one. Identification of the CFAs was performed by comparing their relative retention times to a standard mixture consisting of ClO:O, C12:0, C14:0, C14:1, C16:0, C16:1, C17:0, Cl&O, Cl&l, C18:2, C18:3, C20:0, C20:1, C22:O and C22:l. Principal component analysis (Joliffe, 1986) was performed separately on four groups of strains, namely, the 79 isolates and reference strains examined in the study, the 18 reference strains used in the study, the Lactobacillus sake/Lactobacillus curvatus component of the study (29) and the Leuconostoc component of the study (31), which included Lactobacillus confusus (DSM 20196) since this species is closely related to the leuconostocs (Collins et al., 1991). In each case, analyses were performed on the six most variable CFAs (based on the coefficient of variation for individual fatty acids) for each group examined, and used to determine the relationship between strains.

3. Results and discussion 3.1. Cellular fatty acid identities and relative percentages

Reproducibility between results obtained for replicate CFA extracts was high (coefficient of variation 6.21) and duplicate samples were therefore averaged. The mean relative percentage of CFAs from all strains in this study are presented in Table 2. Thirteen CFAs (straight-chain saturated, straight-chain monosaturated, and cyclopropane), namely C12:0, C14:0, C14:1, C15:0, C16:0, C16:l C17:0, C18:0, C18:1(9), C18:1(11), C19cyc, C2O:l and C22:O were identified. Four of the CFAs, namely C16:0, C16:1, C18:1(9) and C19cyc made up more than 80% of the fatty acids in most strains, with another 4, namely, C14:0, C18:0, C18:1(11) and C22:O making a smaller (on average 2-3% each) contribution. The rest of of the CFAs contributed very small percentages (on average < 1%) to the total. All the CFAs found in this study have been previously identified in members of the LAB (Uchida and Mogi, 1973b; Dainty et al., 1984; Schmitt et al., 1989;

a

0.12 0.11 0.03 0.16 0.34 0.92 0.12 3.32 0.49 0.38 0.10 0.27 0.10 0.34 0.92 0.12 0.37 0.18 0.04 0.11 0.12 0.63 0.03 0.00

59 63 69 77 11

69 70 77 27 30 36 41 46 23

for cluster designation.

0.16 3.32 0.49 0.38 0.37 0.09 0.09

2.87 2.14 1.36 1.60 0.98 1.93 1.21 6.15 1.76 2.64 2.02 1.28 3.24 0.98 1.93 1.21 4.28 4.11 1.50 1.42 2.45 2.58 1.36 1.21 1.60 6.15 1.76 2.64 1.51 2.30 2.32

0.08

44

12 18 27 30 36 73 79 62 11 12 18 1 48 66 6 20 21

1.63 2.55

c14:o

0.18

c12:o

Fatty acid

of cellular

4

a See Figs. l-4

4c

4B

4A

3c

3B

3A

2A

1D

1c

1B

1A

Cluster

per cluster)

percentage

Strain

relative

Mean

representatives

2

Table

0.36 0.34 0.22 0.28 0.28 0.15 0.04 0.42 0.15 0.12 0.28 0.25 0.44 0.28 0.15 0.04 0.56 0.51 0.23 0.19 0.25 0.15 0.22 0.27 0.28 0.42 0.15 0.12 0.10 0.26 0.24

0.17 0.22

Cl4:l

fatty acids

30.25 34.68 32.35 27.06 28.54 26.29 59.91 26.11 30.31 23.63 31.66 35.46 35.39 34.77 26.29 59.951 26.11 31.47 32.35 26.78 26.59 34.99 36.16 27.06 28.16 28.54 30.31 23.63 31.66 38.81 36.91 39.12

0.08 0.16 0.60 0.29 0.08 0.04 0.04

28.54

c16:o

lactic

0.06 0.00 0.07 0.08 1.48 1.12 2.02 0.16 0.60 0.29 0.07 0.03 0.04 1.48 1.12 2.02 0.00 0.07 0.06 0.00 0.11 0.08 0.07 0.20

0.03

CEO

of spoilgae

8.86 6.14 11.99 8.54 7.52 10.76 15.58 16.69 4.59 10.82 10.22 7.95 4.71 5.01 5.60 13.24 5.75 9.96 5.65 7.26 9.72

8.78 10.49 7.65 4.71 5.60 8.54 7.52 10.76 13.24 5.75 9.96

10.49

C16:l 0.02 0.51 4.82 3.62 3.41 2.30 1.74 2.48 0.63 0.69 1.00 0.56 0.77 0.85 2.30 1.74 2.48 0.56 1.26 1.88 0.00 0.14 0.14 3.62 2.44 3.41 0.63 0.69 1.00 0.16 0.02 0.08

0.00

c17:o

acid bacteria

1.12 1.23 1.51 0.82 3.37 3.67 1.94 0.78 5.42 4.18 1.55 2.01 1.23 3.37 3.67 1.94 2.79 0.65 2.09 1.60 1.29 2.08 1.51 1.23 0.82 0.78 5.42 4.18 4.41 1.36 4.20

1.03

2.41

38.82 34.71 39.68 24.53 51.31 50.68 49.45 21.59 18.62 8.31 55.87 36.84 27.74 11.89 28.06 13.58 19.36 49.04 38.82 37.81 19.91 35.45

27.30 20.24 11.89 13.58 51.31 50.68 49.45 19.36 49.04

21.71

3Z.A

C18:1(9) _^ -1

vacuum-packaged

c18:o

from

0.68 0.42 0.39 0.27 0.37 1.60 2.00 1.26 6.75 3.00 0.53 0.17 0.19 0.25 1.60 2.00 1.26 1.21 0.52 1.58 0.47 0.34 0.84 0.27 0.14 0.37 6.75 3.00 0.53 0.46 0.32 0.20

1.15

C18:l(ll)

and

0.57 14.74 3.09 4.08 15.12 13.04 0.25 0.33 0.34 0.57 19.40 23.17 48.99 1.01 12.94 17.79 47.03 31.56 41.48 14.74 3.09 4.08 9.04 30.91 7.93

20.77 0.39 47.03 41.48 0.33 0.34

33.67

0.00

c19cyc

Vienna sausages

0.06 0.93 1.25 2.07 1.16 0.91 0.78 0.65 1.35 1.00 1 .oo 0.47 0.50 2.15 0.91 0.78 0.65 1.22 1.51 1.10 0.31 0.21 0.23 2.07 0.81 1.16 1.35 1 .oo 1.00 0.59 0.07 0.64

0.51

C2O:l

reference

z ? 2

s

1.64 0.12 3.68 0.20 0.56 0.42 2.79 5.27 5.36 1.09 0.58 0.00

s

2.00 0.35 0.50

8

8‘9: 3.43

:

2

h

R \ $

2.31 4.29

(three

0.17 0.43 0.23

3.43 2.79 5.27 5.36

0.42 2.31 4.29

2.76

C22:o

strains

94

G.A. Dykes et ul. /ht.

1. Food Microbiology 25 (1995) 59-100

Decallone et al., 1991). While the relative percentage composition of the CFAs for the LAB as a group (i.e Cl&O, C16:1, C18:1(9) and C19cyc comprising > 80% of the total) was similar to these studies, the relative percentages of individual fatty acids for a given reference strain were often different. For example, Schmitt et al. (1989) found that Leuconostoc mesenteruides (DSM 20343) contained lower amounts of C18:1(9) (0.05) and higher amounts of C19cyc (23.8) and C14:O (11.5) while in this study the same strain was found to contain higher amounts of C18:1(9) (36.98) and lower amounts of C19cyc (9.79) and C14:O (1.5). Since medium composition and culture conditions may affect the CFAs of LAB, the differences are explainable in terms of the different media and conditions used (Dainty et al., 1984; Schmitt et al., 1989). Examination of the reference strains indicated no consistent differences between members of the leuconostocs and the lactobacilli in this study. Individual strains within these genera could, however, be differentiated based on the relative percentages of C16:1, C18:l and C19cyc and the presence or absence of C15:O and C17:O. For example, Lactobacillus sake (DSM 20017) was clearly differentiated from Lactobacillus curuatus (DSM 20019) by the presence of a higher relative percentage C16:O and C18:1, and a lower relative percentage of C17:O and C18:2. It was also noted that high amounts of C18:l correlated with low amounts of C19cyc and uice-versa in many strains. Since C18:1(9) is a precursor of C19cyc, the correlation between the relative percentage of these two CFAs was expected (Veerkamp, 1971). Satisfactory grouping of strains to species level on the basis of differences in a few CFAs was not apparent. This is in contrast to the studies of Dainty et al. (1984) and Shaw and Harding (1989) who were able to differentiate species groups on the basis of the relative percentage of 2 CFAs.

4. Muttivariate

analysis

of data

Numerous methods have been applied to numerically analyse CFA data for taxonomic purposes (Eerola and Lehtonen, 1988). Most of these use coefficients [such as the overlap coefficient of Bousfield et al. (1983)] requiring extensive calculations, usually on the whole data set. Although a more rapid method for the analysis of CFA data for the LAB based on ranking of 6 CFAs, in order of relative percentages has been described (Decallone et al., 1991), specialized calculations of similarity are still required. The use of principal component analysis which is rapid and easy to perform has been applied to the CFA data of genera such as Bucieroides, Wolinella and Campylobacter (Brondz and Olsen, 1991) but not to the LAB. In this study, principal component analysis of only the six most variable CFAs was found to be most suitable for determining the taxonomic relationship between LAB from spoiled vacuum-packaged Vienna sausages. Use of all CFAs or only the CFAs occurring in highest proportions did not allow optimal discrimination of independent groups. Since CFAs occurring in small relative percentages are often ignored in numerical analyses (Decallone et al., 1991) this study included CFAs that were selected on the basis of variability independent of their relative

G.A. Dykes et al. / ht. 1. Food Microbiokgy 28 (1995) 89-100

95

Cl Fig. 1. Scatterplot of the first two principal components (Cl and C2) based on the six most variable fatty acids for 61 isolates from spoiled vacuum-packaged Vienna sausages and 18 reference strains (Cluster 1A: all Leuconostoc and Lactobacifhs strains not included in other clusters; Cluster IB, IC and ID: see Fiis. 2-4).

percentage so as to achieve maximal separation of strains. Since in all cases projections of more than the first two principal components provided no additional info~ation on the relationship between strains only plots of the first two principal components are presented. The first two principal components based on six CFAs (C12:0, C15:0, C17:0, C18:1(11), Clgcyc, C22:O) for all the strains examined in this study (Fig. l), described 60.6% (37.1% + 23.5%) of the variation in the CFA matrix between them. Most strains were grouped in a single cluster (1A) which consisted of both the ~ctobaci~l~ and Leucorzostoc components of the population. Cluster 1B consisted of four reference strains (Leuconostoc citreum (DSM 20288), Leuconostot lactis (DSM 20202), Lactobacillus con$su.s (DSM 20196) and Lactobacihs corynifomti (DSM 20001)), cluster 1C of 4 lactobacilh (all atypical members of the Lactobac~Z~ sa~e/curva~ group) and cluster 1D of seven Leuconostoc mesenteroides isolates. Clear differentiation between all members of the two genera was not possible when CFA data for all strains was analysed together (Fig. 1). A number of independent clusters (lB, C and D) were, however, observed in this projection. The relationship between these strains was maintained in subsequent analyses which are discussed below. The first two principal components based on 6 CFAs (C12:0, C15:0, C17:0, Cl&l(ll), Clgcyc, C22:O) of the 18 reference strains examined in this study described 61.6% (42.5% + 19.1%) of the variation in the CFA matrix between them. Strains were widely scattered in this projection (Fig. 2). Only one cluster (2A) was apparent. This cluster contained four Leuconostoc reference strains (~euconost~ gels (DSM 5578), ~uco~ost~ camoszm (DSM 5576), Leuconos-

96

G.A. Dykes et al. /ht.

1. Food Microbiology 28 (1995) 89-100

63

66 69

77

64

c2

71

68

I

76

67

Cl (Cl and C2) based on the six most variable fatty Fig. 2. Scatterplot of the first two principal components acids for 18 lactic acid bacteria reference strains (Cluster 2A: four Leuconostoc and two Lactobacillus reference strains).

tot paramesenteroides (DSM 202881, Leuconostoc mesenteroides (DSM 20343)) and two Lactobacillus references strains (Lactobaciflus sake (DSM 20017) and Lactobacillus coryneformis (DSM 20001)). Clearly, the Leuconostoc reference strains of

this study (many of which were present in a single cluster) appeared more homogeneous than the Lactobacillus reference strains, which were widely scattered, on the basis of CFA content. Leuconostoc citreum (DSM 20188) and Leuconostoc factis (DSM 20202) differed widely on the basis of CFA profiles from all other Leuconostoc reference strains, but appeared closely related to each other. Since these two Leuconostoc species form an independent phylogenetic branch in the Leuconostoc sensu strict0 group of Collins et al. (1991), this result was not unexpected. Although Shaw and Harding (1989) differentiated between Leuconostot gelidum, Leuconostoc carnosum and Leuconostoc mesenteroides on the basis of CFAs, this division was not apparent in the present study. The results of Collins et al. (1991) which indicated that Leuconostoc paramesenteroides forms an independent Paramesenteroides group with some lactobacilli such as Lactobacilhs confhsus was not confirmed by our CFA data. Leuconostoc paramesenteroides (DSM 20288) appeared closely related to the majority of Leucorwstoc strains, while Lactobacillus confusus (DSM 20196) was more closely related to Leuconostoc lactz3 (DSM 20202) and Leuconostoc citreum (DSM 20188) on the basis of CFA content. Of particular note was the separation of Lactobacillus curvatus strains (DSM 20019, KO875) from the Lactobacilhs sake (DSM 20017) strain in this projection. The first two principal components based on six CFAs (C12:0, C14:1, C15:0, C17:0, C19cyc, C2O:l) for the Lactobacillus sake/Lactobacilhs curvatw component of this study (Fig. 3), described 61.7% (32.9% + 28.8%) of the variation in the CFA matrix between them. Three clusters were apparent in this projection.

G.A. Dykes et al. /ht.

J. Food Microbiology 28 (1995) 89-100

97

Cl Fig. 3. Scatterplot of the first two principal components (Cl and CL?)based on the six most variable fatty acids for 29 Lactobacilhs sake/Lactobacilhs curvatus strains (Cluster 3A: four atypical Luctobacillus sake/curvatus strains; Cluster 3B: five typical Lactobacihs curuatus, two typical Lactobacihs sake and one atypical Lactobacillus sake/curvatus strains; Cluster 3C: all remaining strains in this analysis).

Cluster 3A contained four strains all of which were atypical members of the Lactobacillus suke/curvatus group. Cluster 3B consisted of the reference strain of Lactobacillus curvatus (DSM 20019), four typical Lactobacillus curvatus strains, two typical Lactobacillus sake strains and one atypical Lactobacillus sake/curvatus strain. All remaining strains which included Lactobacillus sake (DSM 20017), a single typical Lactobacillus curvatus strain as well as typical Lactobacillus sake and atypical Lactobacillus suke/curvatus strains were found in cluster 3C. Based on CFA content the majority of strains showed a high degree of homology with the type strain of Lactobacillus sake (DSM 20017) (cluster C). Since all but 1 of these strains were either typical Lactobacillus sake or atypical Lactobacillus sake/curvatus strains, it is suggested that all strains in this cluster are Lactobacillus sake strains despite phenotypic differences (van Holy et al., 1991) between them. The separate Lactobacilhs curuatus cluster (3B) containing most typical Lactobucilhs curvatus strains confirmed this. The presence of hvo typical Lactobacillus sake strains in this cluster, however, indicated that relationships derived from CFA content may not always be clearly defined and that caution should be exercised when attempting to identify single isolates of these species by comparison to a CFA data base. The four strains in cluster A of this projection (Fig. 3) differed markedly from all other strains on the basis of CFA content. This feature was noteworthy and the exact relationship of these strains to the reference strain and to each other should be investigated further by other means, such as DNADNA hybridization. The first two principal components based on six CFAs (C12:0, C15:0, C17:0, C18:1(11), C20:1, C22:O) for all the leuconostocs examined in this study (Fig. 41,

98

G.A. Dykes ef al. /1n1. _I. Food Microbiology 28 (1995) 89-100

c2

Cl Fig. 4. Scatterplot of the first two principal acids for 31 Leuconostoc strains (Cluster Leuconostoc mesenteroides strains; Cluster

components (Cl and CL?) based on the six most variable fatty 4A: three Leuconostoc reference strains; Cluster 4B: nine 4C: all remaining strains in this analysis).

described 69.3% (46.1% + 23.2%) of the variation in the CFA matrix between them and three clusters were apparent. Cluster 4A consisted of three reference strains Leuconostoc luctis (DSM 202021, Leuconostoc citreum (DSM 20188) and Lactobacih confusus (DSM 20196). Cluster 4B consisted of nine isolates all of which were phenotypically Leuconostoc mesenteroides strains (Garvie, 1986X Cluster 4C consisted of all other reference strains and isolates included in this analysis. The close relationship between all Leuconostoc strains examined in this study, except Leuconostoc citreum (DSM 201881, Leuconostoc lactis (DSM 20202) and Lactobucillus confusu.s (DSM 201961, was confirmed from Fig. 2. In addition, cluster 4B indicated the ability of CFA analysis to differentiate strains of Leuconostoc mesenteroides associated with the spoiled meat environment from other Leuconostoc species. The fact that Leuconostoc mesenteroides (DSM 20343) was not included in this cluster, however, indicated that strains of this species from other environments may differ on the basis of CFA content. Care should thus be taken when comparing the CFAs of strains from diverse environments. Since all remaining strains in this projection clustered together (4C) differentiation of any further Leuconostoc groups was not possible on the basis of CFA content. The taxonomic status of strains in the Lactobacillus sake/curvatus group associated with meat spoilage was partially resolved by the present study which indicated that most atypical species were probably Lactobacillus sake. The presence of an independent group of atypical strains, however, indicated that further examination of this relationship was necessary. The taxonomic status of the leuconostocs associated with meat spoilage was not clarified by this method, and further work by other taxonomic methods is required.

G.A. Dykes et al. /ht.

J. Food Microbiology 28 (1995) 89-100

99

References Augustyn, O.P.H. and Kock, J.L.F. (1989) Differentiation of yeast species and strains within a species, by cellular fatty acid analysis. 1. Application of an adapted technique to differentiate between strains of Succharomyces cereutiiae. J. Microbiological Methods 10, 9-23. Bpusfield, I.J., Smith, G.L., Dando, T.R. and Hobbs, G. (1983) Numerical analysis of total fatty acid profiles in the identification of coryneform, nocardioform and some other bacteria. J. Gen. Microbial. 129, 375-394. Brondz, I. and Olsen, I. (1991) Multivariate analysis of cellular fatty acids in Bacteroides, Preuotella, Porphyromonas, Wolinella, and Campylobucter spp. J. Clin. Microbial 29, 183-189. Collins, M.D., Farrow, J. A. E., Phillips, B. A., Ferusu S. and Jones D. (1987). Classification of Lactobacih dioergens, Lactobacillus piscicola and some catalase-negative, asporogenous, rod-shaped bacteria from poultry in a new genus, Cumobacterium. Int. J. System. Bacterial. 37, 310-316. Collins, M.D., Rodrigues, U., Ash, C., Aguirre, M., Farrow, J.A.E., Martinez-Murcia, A., Phillips, B.A., Williams, A.M. and Wallbanks, S. (1991) Phylogenetic analysis of the genus Lactobacilh and related lactic acid bacteria as determined by reverse transcriptase sequencing of 16s rRNA. FEMS Microbial. Letts 7, 5-12. Dainty, R.H., Hibbard, C.M. and Edwards, R.A. (1984) Cellular fatty acids of streptobacteria isolated from vacuum packaged meats. System. Appl. Microbial. 5, 233-240. Decallone, J., Delmee, M., Wauthoz, P., El Lioui, M. and Lambert, R. (1991) A rapid procedure for the identification of lactic acid bacteria based on the gas chromatography analysis of the cellular fatty acids. J. Food Protect. 54, 217-224. Eerola, E. and Lehtonen, O-P. (1988) Optimal data processing procedure for automatic bacterial identification by gas-liquid chromatography of cellular fatty acids. J. Clin. Microbial. 26, 1745-1753. Garvie, E.I. (1986) The genus Leuconostoc. In: P.H.A. Sneath, N.S. Mair, M.E. Sharpe and J.G. Holt (Eds), Bergey’s manual of systematic bacteriology Vol. 2., Williams and Wilkins, Baltimore, pp. 1071-1075. Hastings, J.W. and Holzapfel, W.H. (1987) Numerical taxonomy of lactobacilli surviving radurization of meat. Int. J. Food Microbial. 4,33-49. Hitchener, B.J., Egan, A.F. and Rogers, P.J. (1982) Characteristics of lactic acid bacteria isolated from vacuum packaged beef. J. Appl. Bacterial. 52, 331-337. Joliffe, LT. (1986) Principal component analysis, Springer-Verlag, New York. Mukwaya, GM. and Welch, D.F. (1989) Subgrouping of Pseudomonas cepacia by cellular fatty acid composition. J. Clin. Microbial. 27, 2646-2649. Reuter, G. (1981) Psychrotrophic lactobacilli in meat products. In: T.A. Roberts, G. Hobbs, J.H.B. Christian, and N. Skovgaard (Ed& Psychrotrophic microorganisms in spoilage and pathogenicity, Academic Press, London, pp. 253-258. Rizzo, A.F., Korkeala, H. and Mononen, I. (1987) Gas chromatography analysis of cellular fatty acids and neutral monosacharides in the identification of lactobacilli. Appl. Env. Microbial. 53,2883-2888. Schmitt, P., Mathot, A.G. and Dies, C. (1989) Fatty acid composition of the genus Leuconostoc. Milchwissenschaft 44.556-559. Shaw, B.G. and Harding, C.D. (1985) Atypical lactobacilli from vacuum-packaged meats: Comparison by DNA hybridization, cell composition and biochemical tests with a description of L.uctobac&r carnis sp. nov. System. Appl. Microbial. 6, 291-297. Shaw, B.G. and Harding, C.D. (1989) Leuconostoc gelidum sp. nov. and Leuconostoc camosum sp. nw. from chill stored meats. Int. J. System. Bacterial. 39, 217-223. Tracy, R.P. and Britz, T.J. (1989) Cellular fatty acid composition of Leuconostoc oenos. J. Appl. Bacterial. 66, 445-456. Uchida, K. and Mogi, K. (1972) Cellular fatty acid spectra of Pediococcus species in relation to their taxonomy. J. Gen. Appl. Microbial. 18, 109-129. Uchida, K and Mogi, K. (1973a) Cellular fatty acid spectra of Sporolactobucillus and some other Bacillus-Lactobacillus intermediates as a guide to their taxonomy. J. Gen. Appl. Microbial. 19, 129-140.

100

G.A. Dykes et al. /ht.

J. Food Microbiology 28 (1995) 89-100

Uchida, K. and Mogi, K. (1973b) Cellular fatty acid spectra of hiochi bacteria, alchohol tolerant lactobacilli, and their group separation. J. Gen. Appl. Microbial. 19, 233-249. Veerkamp, J.H. (1971) Fatty acid composition of Bi’dobacterium and Lactobacilh strains. J. Bacteriol. 108, 861-867. von Holy, A., Cloete, T.E. and Holzapfel, W.H. (1991) Quantification and characterization of microbial populations associated with spoiled, vacuum-packed Vienna sausages. Food Microbial. 8, 95-104. von Holy, A., Holzapfel, W.H. and Dykes, G.A. (1992) Bacterial populations associated with Vienna sausage packaging. Food Microbial. 9, 45-53.