Ecological studies on coccoid bacteria in a pine forest soil—I. Classification

Soil Eid. BiocRem. Vol. 4, pp. 459-467. Pexgamon Press 1972. Printed in Great Britain

ECOLOGICAL STUDIES ON COCCOID BACTERIA IN A PINE FOREST SOIL-I. CLASSIFICATION v/T.

E.

LOWE*

and

T. R. G. GRAY

Hartley Botanical Laboratories,

University of Liverpool

(Accepted 30 March 1972)

Summary-Two hundred and nine coccoid bacteria, most of which were yellow pigmented, were isolated from the A and C horizons of a pine forest soii. One hundred and seventy nine tests were carried out on all of these bacteria and the overall similarities between them calculated using Sneath’s similarity coefficient. Ninety per cent of the organisms could be allocated to seven principle clusters, each having high intra-group similarities of over 70 per cent. There were two Micrococcus clusters, one Staphylococcuscluster and four Arthrobucter clusters. Some of these clusters were associated with only one soil horizon while others occurred in both horizons.

Tm CLASSIFICATION and identification of organisms is an essential first step in ecological investigations. Despite this, comparatively little is known about the bacteria characteristic of different soil types, and consequently meaningful ecological studies of such organisms have been rare. Selection of strains for autecological investigations has usually been made on a haph~ard basis, without knowledge of whether such strains are typical or aberrant representatives of the species being considered. In a survey of heterotrophic bacteria found in a pine forest soil at Freshfield, Lancashire, England, Goodfellow (1966, 1969) reported the isolation of relatively large numbers of coccoid bacteria, many of which produced yellow insoluble pigments on peptone-yeast extract agar. True cocci have been isolated infrequently from soils and have generally been thought of as an insignificant fraction of the microllora (Beck and Poschenreider, 1958; Holding, Franklin and Watling, 1965; Jensen, 1963). On the other hand, Arthrobacter species and other coryneforms are commonIy isolated from many different soils and are often referred to as characteristic or indigenous soil forms (Mulder and Atheunisse, 1963; Jensen and Felumb, 1962; Casida, 1965). In the studies made by Goodfellow (1966, 1969), both true cocci and Arthrob~cter species represented between 11 and 16 per cent of the isolates taken from dilution plates and so it was decided to see whether there were a wide variety or onIy a few strains characteristic of this acid forest soil. The ability of the most characteristic strains to grow in the soils from which they were isolated and their potential to compete with one another could then be examined. Selection of typical strains can be made following the application of numerical taxonomic procedures to randomly collected strains of soil bacteria taken from isolation plates (Gray, 1969), allowing comparative ecological work to be carried out on strains with differing degrees of overall similarity. * Present address: Canada. SOIL

4M-F

Department

of Soil Science, University 459

of Saskatchewan,

Saskatoon,

Saskatchewan,

460

W. E. LOWE AND T. R. G. GRAY MATERIALS

AND METHODS

Soils

The soil from which the organisms were isolated has been described elsewhere (Goodfellow et al., 1968). It is located in a plantation of Pinus nigra ssp. Zaricio on the dunes at Freshfield, Lancashire (National Grid Reference : SD 289096). A pit was dug to expose all of the A, horizon and about 20 cm of the C horizon. Five samples were taken at various depths from the A, horizon, starting 3 -0 cm below the A,/A, horizon border and finishing 2-O cm above the A,/C horizon border. A similar set of samples were taken from the exposed part of the C horizon. They were taken back to the laboratory and sub-sampled immediately. Isolation procedure

Dilution plates were prepared from the separate soil samples using peptone yeast extract agar containing actidione and nystatin (PYE) (Goodfellow et al., 1968). All the yellowpigmented colonies were picked off from plates yielding 60 or fewer colonies, together with one in five randomly selected white colonies. After 7 days growth on PYE agar slopes, the isolates were streaked repeatedly on PYE agar and checked for purity by observation of Gram-stained preparations. All isolates which were Gram-positive or Gram-variable and coccoid in appearance after 24-48 h incubation at 25°C were subcultured on PYE agar slopes and stored at 5°C under oil (Rhodes, 1957). In this way, 118 isolates were obtained from the A, horizon samples and 91 from the C horizon samples. A number of reference cultures of known strains were obtained from culture collections for comparison with these organisms. The code numbers, name and culture collection numbers* of these organisms were as follows : (a) Staphylococcus roseus (NCTC 7520) ; (b) S. saprophyticus (NCTC 7666); (c) S. afermentans (NCTC 8512); (d) Staphylococcus aureus (NCTC 8532); (e) Sarcina Zutea (NCIB 611); (f) Micrococcus Zuteus (NCIB 8615); (g) M. Zysodeikticus (NCTC 2665); (h) Arthrobacter globiformis (NCIB 8907); (i) A. simplex (NCIB 8913); (j) Xunthomonus campestris (NCPPB 45) ; (k) P seu domonus synxun thu (NCIB 8 178) ; (1)P. ovalis (NCIB 8296) ; (m) P.fluorescens (NCTC 7587); (n) BuciZZuscereus (NCTC 7587); (0) B. polymyxu (NCTC 4747); (p) B. megutherium (NCTC 7581); (q) B. pantothenticus (NCTC 8122); (r) B. subtilis (NCTC 3610); (s) Escherichiu coli (NCTC 1093); (t) Agrobucterium tumefuciens (NCPPB 397); (u) A. rudiobucter (NCIB 8149). Tests applied to the organisms

The tests applied to the organisms are listed in Appendices A and described elsewhere (Goodfellow, 1968 and Goodfellow et al., 1968). taken from 24 h old PYE agar slope cultures grown at 25°C. Appendix which were carried out but which gave negative results for all the strains were not included in the taxonomic analysis.

B and have been All inocula were C lists those tests tested and which

Coding of features for data analysis and computation

Features which existed in one or two mutually exclusive states were recorded as plus or minus. Inapplicable characters were designated NC and ignored in the analysis. A nonadditive scoring method was used for the multistate characters although size of cell and Gram-reaction were scored additively (Sokal and Sneath, 1963). * NCIB-National NCPPB-National NCTC-National

Collection of Industrial Bacteria, Torrey, Aberdeen; Collection of Plant Pathogenic Bacteria, Harpenden, Collection of Type Cultures, Colindale, London.

Herts;

461

STUDIES ON COCCOID BACTERIA IN SOIL.-1

After coding, data were transferred to punch cards and analysed on the Orion computer at Rothamsted Experimental Station, Harpenden. Two methods of comparison were employed, using the similarity coefficient (sj) of Sneath (1957) and the matching coefficient (Ss) of Sokal and Michener (1958). total number of shared positive characters x 100 % Sj = total number of characters tested which varied in the sample % ss =

total number of shared positive + shared negative characters total number of characters tested which varied in the sample

x 100

In all, 179 unit characters were used in the calculation of these coefficients. After the similarities between organisms had been calculated, a second programme was used to arrange the isolates into meaningful groups using the single linkage cluster analysis technique of Sokal and Sneath (1963). Dendrograms were produced with levels reducing consecutively by 2 per cent. Similarity matrices, in which the coefficients obtained were placed into 10 per cent classes (90-99, 80-89 per cent, etc.) were also produced. Finally, clusters were represented by the results of a critical path analysis to produce minimum spanning trees. Pairs of organisms were extracted from the matrix at their highest similarities to each other; each organism could be linked to at least one other in this way. Either member of a pair might also be the nearest neighbour to a further organism so that by linking all organisms appearing most similar to each other, a chain could be produced. Where more than two organisms were linked, branches and nodes in the chain were produced. By indicating the similarity values between the organisms on the diagrams produced in this way, the relationships between individual organisms can be seen more clearly than in a dendrogram or similarity matrix. The matrices and dendrograms are large and too complicated to reproduce but they are available for consultation (Lowe, 1969); it is the minimum spanning trees which are used in this paper. Essentially similar results were obtained using the similarity and matching coefficients, although the groups produced had apparently higher internal levels of similarity (about 10 per cent higher) when the matching coefficients were employed. Because of the similarity between the analyses and because non-additive scoring is best suited to the determination of similarity coefficients (Goodfellow, 1966), the results of the analyses using Sj are discussed in more detail. CLUSTERING OF THE SOIL AND MARKER STRAINS

Figure 1 shows a minimum spanning tree for the soil and marker strains, determined using a similarity coefficient (Sj). At this level, about 180 of the 197 soil isolates included in the analysis could be placed into or associated with clusters containing four or more organisms. Seven such clusters could be distinguished, one of which (Cluster 4) could be divided into two fairly distinct sub-clusters. These coalesced at the 74 per cent level. The soil isolates were identified using the keys and diagnostic tables from Breed et al. (1957), Baird Parker (1965) and Evans, Bradford and Niven (1955) and assigned to the genera ~~ap~yZococ~s ~~crococc~ and Art~robacter. The Ar~~robac~er spp. could be subdivided into two groups, group A isolates showing some branching of the cells and being nearly always Gram-positive while Group B types did not branch and were more variable in their Gram-reaction. The clusters were examined to see if they contained one or more of these genera or any of the marker strains. The results of this examination were as follows.

W. E. LOWE AND T. R. G. GRAY

462

2 Staphylococcus

:, rl

40

pi 0

...0 ....o.....0 .......... “>+<............................. ......... -+_.-.

I Micrococcus

Arthrobocter

Arthrabacter

B

B 5

Arthrobacter

A 3

FIG. 1. Minimum spanning tree showing the clusters of soil and marker strains using a similarity coefficient. (0) = C horizon strain; (0) = A horizon strain; lettered = marker strain; ) = SO-89% similarity; (----) = 70-79% ) = 90% similarity and above; ((similarity; (. . . . . . . .)=below70’% similarity. Solid lines surrounding groups of strains indicate that these form a cluster at the 70 per cent level of similarity on matrix diagrams.

STUDIES

ON COCCOID

BACTERIA

463

IN SOIL-I

Cluster 1. This cluster consists of 34 isolates identified as ~icrococcus one Staphylococcus and two Arthrobacter B isolates. Also associated with this cluster are ~icrococeus lysodeikticus, Staphylococcus afermentuns and Staphylococcus roseus. Cluster 2. This cluster is composed entirely of 37 Staphylococcus isolates, together with Staphylococcus saprophyticus and Staphylococcus aureus. Loosely associated with it, i.e. below the 70 per cent Sj level, are the bulk of the Gram-negative and Bacillus marker strains and two unidentified strains. Cluster 3. This cluster contains 18 Arthrobacter A isolates, one Arthrobacter B isolate and the marker strain Arthrobacter globiformis. Most of the soil isolates produced a soluble violet pigment on the rich medium of Mulder et al. (1966) and they closely resembled the description of A. polychromogenes given by Sc~p~rs-Lammertse et al. (1963). A strain ISOLATBS

TA~~~.~~RS~~HCANBEUSE~TO~E~T~~IL

Cluster Character

1

2

3

Cell length: 1 pm and over

Cell arrangement : random chains Pleomorphism Colony elevation : convex Turbidity: slight Sediment : heavy Growth at pH 5, not below No growth above 5 % NaCl Growth 7 % NaCl, not above Sole C source: gfuconate tartrate succinate alginate Growth in 0.1% phenol 1 a0% phenol 0.01% crystal violet 0.01% sodium azide 0.05 % sodium azide 0.05 % potassium teliurite Glucose oxidation fermentation Acid from Darabinose fructose mannose maltose glycerol I&S production Urease Acetyl-methyl-carbinol Aesculin hydrolysis Tributyrin hydrolysis Casein hydrolysis Starch hydrolysis Butyrous colonies Sensitive to erythromycin Strongly sensitive to erythromycin + 86-100% isolates positive. d 16-85 % isolates positive. - O-l 5 % isolates positive.

i

-

d -

+ +

: :: “I-

:: : + d + d d d -

+ d +

5

6

7

d + + d d -t-

+ +

+ -

: d -

:

+ -

: -

d d

d _

d d -t + +

d :

: d d -

d d -

a d

4W)

d d -

i +

d

4(l)

d d d d +

d d -

d d -

+ f +

d + -

+ d d -

d d

Ti + + i

: -

-

d d d + + d -I-

464

W. E. LOWE AND

T. R. G. GRAY

of this cluster has been deposited in the National Collection of Industrial Bacteria (NCIB 10683) as a variant of A. globiformis. Cluster 4. Sub cluster (i) and (ii) are composed of 57 Arthrobacter B isolates and 3 Micrococcus strains. These micrococci were very unreactive in the tests used and in this respect resembled the Art~rob~cter B isolates. No marker strains were found in this group, although Arthrobacter simplex was associated with one member of sub-cluster (ii) at the 60 per cent level. They most closely resemble the description of A. citreus given by Breed et al. (1957). Clusters 5 and7. These two clusters are composed entirely of Arthrobacter B strains (6 and 4 isolates respectively) but no marker strains were included in either. Cluster 6. This is a small heterogeneous cluster containing two Staphylococcus and four Micrococcus strains and the marker strains Surcina lutea and Micrococcus luteus. It coalesces with cluster 1 at the 70 per cent level of simila~ty and the organisms in it are similar to those described by Baird Parker (1963) and designated by him as Micrococcus sub-group 7. The results showed certain features were characteristic of each cluster, i.e. possessed by 85 per cent or more of the strains in one cluster and present in 15 per cent or less in at least one other cluster (Table 1). It is notable that though most of the organisms studied possessed an insoluble yellow pigment, all clusters contained both yellow and non-pigmented strains. Furthermore, although the possession of a yellow pigment was a stable characteristic of forms in clusters 1 and 4, the organisms in clusters 2 and 3 produced a large number of white mutants. Instability of pigment production in staphylococci as opposed to micrococci has also been reported by Barber (1955). DISCUSSION

Numerical taxonomic procedures have proved useful because they have provided information on the physiology of eoccoid soil bacteria and allowed the unknown Arthrobacter isolates to be grouped and distinguished from one another. We have been able to select typical strains for further experimental work which are central to the most common clusters. It is notable that about 90 per cent of the isolates could be placed into distinct clusters. This contrasts with the results of Rovira and Brisbane (1968) who examined a general collection of heterotrophic bacteria from the soil isolated on a peptone-yeast and soil extract medium. They found that only about one half of their isolates could be grouped and suggested that numerical techniques would provide better results when applied to a narrower range of micro-organisms, a view which is confirmed by the present study. Goodfellow (1969) however, has pointed out that Rovira and Brisbane’s results were based on an insufficient number of tests and that as many as 75-81 per cent of the cultures he isolated on a single medium (PYE) could be placed in distinct clusters when an adequate number of tests was employed. Although the initial identification of isolates in the present study was based on a very small number of characteristics, there is good agreement between these findings and the major clusters obtained from the assessment of overall similarity. Organisms identified as Staphylococcus and Micrococcus purely on the basis of their morphology and ability to produce acid from glucose anaerobically have been recovered, by and large, as two separate groups showing greater similarity to one another than to the Arthrobacter groups. The allocation of the two Staphylococcus marker strains, S. rafermentans and S. roseus, to the Micrococcus cluster is interesting since Baird Parker (1965) also found some strains of these species to be more closely related to the micrococci with which they shared an inability to produce acid from glucose anaerobically. On the other hand, he found S. aureus and S. saprophyiic~s were more typical staphylococci.

STUDIES ON COCCOID BACTERIA IN SOIL-I

465

The marker strains used in this study had all been maintained in culture collections for some time and yet they showed high affinity with the fresh isolates. This situation differs from that reported by Goodfellow (1969) and Rovira and Brisbane (1968). Rovira and Brisbane suggested that soil isolates and laboratory isolates form two ecologically distinct groups and could not be expected to show close affinity. It seems more likely that the choice of marker strains was inappropriate in their studies. This choice is bound to be difficult when general collections of heterotrophs are examined since so few of the named cultures from national collections originate from soil and because soil contains such a wide variety of bacteria. The arrangement of Arthrobacter isolates in two large clusters and several smaller ones coalescing at a similarity level of 70 per cent supports the view that the arthrobacters form a heterogeneous group and that allocation of organisms to it purely on the grounds of their pleomorphism is not entirely satisfactory. The two soil horizons from which the organisms were isolated are very different from one another. The A1 horizon is acid (pH 4 -3) and contains many roots and the leached products of leaf and root decomposition occurring in the A0 horizon. The C horizon is alkaline (‘pH 8 - 1) and contains fewer roots and comparatively few products of leaf decomposition. The fungal populations of these two horizons are different (Parkinson and Balasooriya 1967; Gray and Baxby, 1968) and so it was thought worthwhile to see if the bacterial clusters were associated with one horizon or the other. Table 2 shows the percentage of the A, and C horizon isolates in the clusters defined at the 70 per cent similarity level. Clusters which were predominantly associated with the A, horizon were 3 (Arthrobacter A), 4(i) (Arthrobucter B) and 5 (Arthrobacter B). Those associated predominantly with the C horizon included clusters 4(ii) (Arthrobucter B), 6 (Micrococcus) and 7 (Arthrobacter B). The large Staphylococcus and Micrococcus clusters (1 and 2) were well represented in both horizons although the isolates from the A1 horizon formed distinct sub-clusters at high levels of similarity within these groups, i.e. at about 86 per cent. The approximate numbers of bacteria per gram of soil from each of these clusters are also given in Table 2, emphasizing that these bacteria are probably indigenous forms and are unlikely to be aerial contaminants which are washed into the soil.

TABLE 2. OCCURRENCEOF THEMAJORCLUSTERS(DEFINEDAT THE70 PERCENTLEVEL)IN THEAI AND C HORIZON SOIL.3

Cluster Cluster

1 2 3 4 (0 4m 5 6 7 Not clustered

Name of cluster

Number of strains in cluster

Micrococcus Staphylococcus Arthrobacter A Arthrobacter B Arthrobacter B Arthrobacter B Micrococcus Arthrobacter B ?

37 37 19 29 31 6 8 4 26

in

Al

and C soil (%)

70:30 60:40 79:21 76~24 16:84 loo:0 20:80 0:lOo 67:33

Calculated

Calculated

number of cluster/g of A1 soil

number of cluster/g of C soil

1 x 106

8 8 3 1.5 4 2

x x x x x x 0 -

105 lo5 105 105 105 lo4

4 5 1 1 12.5

x x x x x 0 1 x 1 x -

105 105 105 105 lo5 105 105

466

W. E. LOWE AND T. R. G. GRAY

The organisms central to the largest clusters in the minimum spanning tree (Fig. 1) were as follows: 1, A34; 2, A81 ; 3, A21; 4(i), A49; 4(ii), C39. These strains were selected for studies on patterns of growth and competitive interactions in soil (Lowe and Gray, 1973a, b). AcknowZedgemenfs-We wish to acknowledge the helpful comments made by Dr M. GOODFELLOW this work. One of us (W.E.L) carried out this work while holding an S.R.C. studentship

during

REFERENCES BAIRD PARKERA. C. (1963) A classification of micrococci and staphylococci based on physiological and biochemical tests. J. gen. Microbial. 30, 409-427. BAIRD PORKERA. C. (1965) The classification of staphylococci and micrococci from world wide sources. J. gen. Microbial. 36, 363-387. BARBERM. (1955) Pigment production by staphylococci. J. gen. Microbial. 13, 338-345. BECK T. and POSCHENREIDER H. (1958) Uber der artenmassige Zusammensetzung der Mikroflora eines sehr sauren Waldmoorprofiles. Zbl. Bukt. Purasitkde. Abt ZZ,111, 672-683. BREEDR. S., MURRAY E. D. G. and SMITHN. R. (1957) Bergey’s Manual of Determinative Bacteriology. 7th edn Williams & Wilkins, Baltimore. CASIDAL. E. (1965) Abundant microorganisms in soil. Appl. Microbial. 13, 327-334. EVANSJ. B., BRADFORDW. L. and NIVEN C. F. (1955) Comments concerning the taxonomy of the genera Micrococcus and Staphylococcus. Int. Bull. bact. Nomen. Tax. 5, 61-66. G~I~DFELLOWM. (1966) The Classification of Bacteria in a Pine Forest Soil. Ph.D. Thesis, University of Liverpool. G~~DFELL~WM. (1968) Pronerties and comoosition of the bacterial flora of a uine forest soil. J. Soil Sci. 19, 154167. _ ’ G~~DFELLOWM. (1969) Numerical taxonomy of some heterotrophic bacteria isolated from a pine forest soil. In The Soil Ecosvstem (J. G. Sheals. Ed.) DD. 83-105. Svstematics Association, London. G~~DFELLOWM. and GRA; T. R. G. (1966) ‘A multipoint inoculation method for performing biochemical tests on bacteria. In Methods for the Identification of Bacteria. A (F. Skinner and B. E. Gibbs, Eds) pp. 117-123, Academic Press, London. G~~DFELLOWM., HILL I. R. and GRAY T. R. G. (1968) Bacteria in a pine forest soil. In Z7reEcology of SoiI Bacteria (T. R. G. Gray and D. Parkinson, Eds) pp. 500-515, Liverpool University Press. GRAY T. R. G. (1969) Identification of soil bacteria. In The Soil Ecosystem (J. G. Sheals, Ed.) pp. 73-81. Systematics Association, London. GRAY T. R. G. and BAXBYP. (1968) Chitin decomposition in soil-II. The ecology of chitinoclastic microorganisms in forest soil. Trans. Br. mycoI. Sot. 51, 293-309. HOLDINGA. J., FRANKLIND. A. and WATLINGR. (1965) The microflora of peat-podzol transitions. J. Soil Sci. 16, 44-59. JENSENV. (1963) Studies on the microflora of Danish beech forest soils-III. Properties and composition of the bacterial flora. Zbl. Bakt. ParasitKde. Abt II. 116, 594-611. JENSENV. and FELUMBG. (1962) Description of some coryneform bacteria isolated from forest soils. K. VetHojsk. Aarsskr., 195-210. LOWE W. E. (1969). An Ecological Study of Cocoid Bacteria in Soil. Ph.D. Thesis, Liverpool University. LOWE W. E. and GRAY T. R. G. (1973a) Ecological studies on coccoid bacteria in a pine forest soil-II. Growth of bacteria introduced into soil. Soil Biol. Biochem. 5, to be published. LOWE W. E. and GRAY T. R. G. (1973b) Ecological studies on coccoid bacteria in a pine forest soil-III. Competitive interactions between bacterial strains in soil. Soil. Biol Biochem. 5, to be published. MULDER E. G., ADAMSE,A. D., ATHENEUISSE J., DEINEMAK. H., WOLDENDORPJ. W. and ZEVENHUIZEN L. P. T. M. (1966) The relationship between Brevibacterium linens and bacteria of the genus Arthrobacter. J. appl. Bact. 29, 44-71. MULDER E. G. and ATHENEUI~SE J. (1963) Morphologie, physiologie et ecologic des Arthrobacter. An&. Inst. Pasteur, Paris 105, 46-74. PARKINSOND. and BALASOOR~YA I. A. (1967) Studies on fungi in a pine wood soil-I. Rev. ecol. Biol. Sol. 4, 463-478. RHODESM. E. (1957) The preservation of Pseudomonas under mineral oil. J. appZ. Bact. 20, 108-118. ROV~RAA. D. and BRISBANEP. G. (1968) Numerical taxonomy and soil bacteria. In The Ecology of Soil Bacteria (T. R. G. Grav and D. Parkinson. Eds) __ pp. 337-350. Liverpool University Press. SCHIPPERS-LA~MERTSE A. F:, MULJSERSA. 0. and KLATSER-OEDEKERK B: (1963). Arthrobacter polychromogenes nov. spec., its pigments and a bacteriophage of this species. Antonie van Leeuwenhoek J. 29, l-15. SNEATHP. H. A. (1957) The application of computers to taxonomy. J. gen. Microbial. 17, 201-226. SOKAL R. R. and MICHENERC. D. (1958) A statistical method for evaluating systematic relationships. Kansas Univ. Sci. Bull. 38, 1409-1438. SOKALR. R. and SNEATHP. H. A. (1963) Principles of Numerical Taxonomy. Freeman, San Fransisco.

STUDIES

ON COCCOID

BACTERIA

IN SOIL-I

467

APPENDIX Section A: Coded morphological features used in the computer analysis (1) Cell length: < 0.8 pm, 0.8-l *Opm, > 1 .O pm; (2) cell width: < 0.8 pm, > 0.8 pm; (3) cell arrangement: random, chains, packets, irregular clumps; (4) pleomorphism; (5) cystite production (Mulder and Atheneuisse, 1963); (6) Gram stain: +ve, -ve, variable; (7) endospore production; (8) endospore position: central, terminal; (9) sporangium shape : swollen, non-swollen ; (10) motility; (11) colony elevation : raised, convex, umbonate; (12) colony shape: punctiform, circular, lobed; (13) colony pigmentation: (Rembrandt colour chart, Talens, Apeldoorn, Holland) insoluble; white, reddish, yellow; shade of yellow pigment; soluble; violet; (14) growth in nutrient broth: pellicle or ring, turbidity heavy or slight; (15) growth on Mulder’s rich medium (Mulder and Atheneuisse, 1963) (12 h): Butyrous, opaque. Section B: Coded physiological and biochemical features used in the computer analysis (1) Growth limits on PYE agar*: not below pH 7.0,6.0, 5.0,4.0; not above pH 7.0,8.0, 9.0; not below 15”C, 10°C; not above 3o”C, 35”C, 4o”C, 45°C. (2) Growth limit on nutrient agar* : not above 5 %, 7 %, 10 %, 15 %, (w/v) NaCl. (3) Growth on PYE agar containing*: 1 .O%, 0.1% (w/v) phenol; 0.01 ‘A, 0.001% (w/v) crystal violet; 0.05 %, 0.01% (w/v) sodium azide; 0.05% (w/v) potassium tellurite. (4)Growth on Oxoid sensitivity agar + discs containing: 50 units polymixin, 5 pg aureomycin, 2 units bacitracin, 2 units penicillin, 12 rgerythromycin, 5 pg novobiocin, 2 pg streptomycin, 5 pg chloromycetin (zone < 2 mm, 2-10 mm > 10 mm). (5) Sole nitrogen sources (w/v): 0.1% ammonium dihydrogen phosphate, 0.1% aspartic acid. (6) Sole carbon sources (w/v): 1 .O”k gluconate, 1 .O % lactate, 1 .O% succinate, 1.0 % tartrate, 1 .O”A citrate, 0.1% oxalate, 2.0% alginate (all as sodium salts), 0.05 % p-hydroxybenzoic acid. (7) Biochemical reactions: Catalase (strong, medium, weak)*, oxidase, indole production, HIS production (7 days, 7-14 days), urease, acetyl-methyl-carbinol production (7 days, 7-14 days), nitrate reduction*, hydrolysis of gelatin*, aesculin*, xylan, laminarin, tributyrin*, pectin, starch*, and casein*, arginine decarboxylase. Acid production from glucose: oxidative, fermentative (1 day, 4 days, 7-14 days). Acid production from?: arabinose, fructose, galactose, rhamnose, mannose, cellobiose, lactose, maltose, sucrose, trehalose, melizitose, ralhnose, glycogen, inulin, amygdalin, salicin, glycerol, mannitol, inositol (strong, weak). Incubation: multipoint inoculation tests 3-4 days unless otherwise stated, test tube tests, 7 days unless otherwise stated. Section C: Tests giving negative results with the strains used Decarboxylation of lysine and ornithine. Hydrolysis of cellulose, chitin, lignin and araban. Clearing of humic acid. Utilization of sole carbon sources: acetate, formate, vanillin, vanillic acid, coumarin, glucosamine. Utilization of sole nitrogen sources: cysteine, glutamic acid. Production of fluorescent pigments. * Multipoint inoculation test (Goodfellow and Gray, 1966). t Multipoint inoculation and tube tests.