Journal of General Microbiology (1984), 130, 187 1-1 882.

Printed in Great Britain 1871

Identification of Bacillus Strains Using the API System

By N . A. L O G A N ? A N D R . C . W . B E R K E L E Y *
Department of Microbiology, The Medical School, University of Bristol, Bristol BS8 I TD, UK

(Received 14 December 1983; revised 23 March 1984)

A system is described for the rapid and accurate identification of Bacillus isolates using a matrix
of results from tests in the API 20E and API 5OCHB strips and from supplementary tests. API
System tests have been shown to be more reproducible than the classical tests. A taxonomy
based upon API tests is in good agreement with those obtained by other methods. The results
matrix can also be used in computer assisted identification.

INTRODUCTION
Although diagnostic keys and tables for BaciZlus have been available for a long time (Gibson
&Topping, 1938; Smith et al., 1946; Cowan & Steel, 1965), the identification of these organisms
is still considered to be complicated (Green, 1975; King & Phillips, 1978) and in many
laboratories is taken no further than ‘aerobic spore-forming rod’, or ‘Bacillus species’. It is clear,
however, from the numerous papers published and the large number of strains received for
identification in this laboratory, that Bacillus species are of increasing importance in industry
and medicine and that diagnosis to the species level is highly desirable in many instances.
The prevailing neglect of Bacillus identification may be attributed to two factors. Firstly, the
diagnostic tests used; many of the classical tests for Bacillus described by Smith et al. (1946,
1952) and Gordon et al. (1973) require special media. These are very time consuming and
expensive to prepare and many have short shelf lives resulting in considerable wastage if their
use is infrequent. The requirement for media containing unusual ingredients increases the
familiar problems of test standardization (Sneath, 1974; Sneath & Collins, 1974); inconsistent
results may be obtained in consequence. Several tests take 14 d or more from pure culture to final
reading and this is too long for many diagnosticians to wait. Any new scheme for Bacillus
identification should therefore use widely available and standardized materials for performing a
good number of rapid tests which give reproducible results.
The second factor leading to neglect of Bacillus identification is the character of the genus.
Bacillus is an unusually wide taxon which contains most aerobic endospore-forming rods. In
terms of DNA base ratios it is the equivalent of some bacterial families (Priest, 1981). Further-
more, some species are ill-defined, existing with closely related species as complexes or spectra
in which the boundary of a particular species is difficult or impossible to identify. Even in well
established species there is considerable variation between strains. Thus classical test schemes
using few characters often do not permit identification of atypical and intermediate strains and
in spite of the excellent work of Gordon and her colleagues (Smith et al., 1952; Gordon, 1973;
Gordon et al., 1973), as well as others, it is widely agreed that there is considerable room for
improvement in the taxonomy of the genus and that a study of new isolates, particularly, is
important.
With these considerations in mind, work was started to produce a system for the ready and
rapid identification of Bacillus strains. A study of a collection of 600 strains, of which 91 were
unnamed, using 119 API and 20 supplementary tests, has been carried out (Logan & Berkeley,
1981). A development of this work (written later but published earlier), based on the 80 tests
t Present address : Department of Biological Sciences, Glasgow College of Technology, Cowcaddens Road,
Glasgow G4 OBA, UK.
0022-1287/84/0001-1661 $02.00 0 1984 SGM
1872 N . A . LOGAN AND R . C . W . B E R K E L E Y

showing greatest reproducibility and discriminatory value, was described by Logan (1980).
Despite considerable between-strain variation in many species this study showed that the
majority of taxa are quite distinct. API materials used in these trials included the 20E, 50E (both
designed for identification of members of the Enterobacteriaceae), API ZYM and four other,
non-commercially available, enzyme test strips (Logan, 1980). The 50E has now been super-
seded by the 50CH.
In this paper a development of these studies is described. It uses tests in the API 5OCHB
(Bacillus)system in place of those from the 50E, together with those used previously from the
20E, and a small number of morphological and supplementary characters. The enzyme strips
were not used as they contribute little to identification (Logan, 1980).

METHODS
Strains and growth conditions. One thousand and seventy-five strains from culture collections, industrial,
medical and veterinary specimens and from the natural environment were studied. All were checked, on inclusion
in the collection, for purity, medium requirements and growth temperatures. The strains were representative of
the whole genus, but the fastidious insect pathogens, B. larvae, B . lentimorbus and B. popilliae, and species
requiring a very high or very low pH for growth ( B . alcalophilus and B. acidocaldarius respectively) were omitted
from this study. Except for these, type strains were included for all species on the Approved Lists of Bacterial
Names (Skerman et al., 1980). The numbers of representatives of each species are shown in Table 1.
'
Strains were held at 4 "C, in the dark, on slopes of nutrient agar (Difco) containing 5 mg MnSO,. 4 H z 01- (NA
+ Mn2 ) in order to enhance sporulation (Deutsche Sammlung von Mikroorganismen, 1977) and strains were
+

allowed to sporulate prior to storage. For morphological studies strains were grown at 30 "C for 48 h (or 20 "C for
96 h for psychrophilic strains, 37 "C for 48 h for B. coagulans, and 55 "C for 24 h for thermophilic strains) on plates
of NA + MnZ+ . For API tests strains were grown overnight (or, for psychrophiles, 48 h) on plates of nutrient agar
(Difco) (NA) at the appropriate temperature. Certain strains would not grow satisfactorily on NA or NA + MnZ+
and special media had to be used, as follows.
Allantoin Mineral Medium (Deutsche Sammlung von Mikroorganismen, 1977) for B. fastidiosus contained
(g I - ' ) : KZHPO4, 0.8; KH2P0,, 0.2; MgS0,.7H20, 0.5; CaC1,.2H20, 0.05; FeS0,.7Hz0, 0.01;
MnS0,.4H20, 0.001 (increased to 0.005 for sporulation); allantoin (Sigma), 20; agar, 15; at pH 6.8.
'Bacillus racemilacticus' Medium (Deutsche Sammlung von Mikroorganismen, 1977) for ' B laevolacticus' and ' B .
racemilacticus' contained (g I - ' ) : glucose, 5; peptone 5; yeast extract, 5 ; CaCO,, 5 ; agar, 15; at pH 6.8. For
sporulation, 5 mg MnSO,. 4 H 2 0 1- was added prior to autoclaving.
Bacilluspasteurii medium (Gibson & Gordon, 1974) was prepared by adding 10 g NH,CI 1- I to NA or NA +
MnZ+and adjusting the medium to pH 9 prior to autoclaving.
A strain that will grow only on such special media is unlikely to be isolated accidentally and its growth require-
ments will give some indication of its identity. In such cases the API and morphological tests serve principally to
confirm or challenge this first identification.
Morphological and supplementary tests. Vegetative cell morphology observations were made on cultures grown
overnight, or for such longer time as was necessary to obtain visible growth, on NA + MnZ or another suitable
+

sporulation medium, at the appropriate temperature. Organisms were examined at 1000 x magnification, using
phase contrast microscopy, for shape of cells, presence of chains, and for a foamy or vacuolate appearance of the
cytoplasm. Widths of cells were measured using a Vickers-AEI image splitting eyepiece (Quesnel, 1971).
To determine motility, strains were grown on slopes of NA and after 6 h, or as soon as growth appeared there-
after, a loopful of the liquid at the base of the slope was examined at 1000 x magnification, by phase contrast
microscopy.
It was occasionally possible to observe spores in slides prepared from cultures grown overnight on a suitable
sporulation medium at the appropriate temperature, but most strains required incubation for 2 d, occasionally
longer, before spore and sporangial morphology could be observed. Cultures were examined for shape of spores,
their position in the sporangia, distension of the sporangia by mature spores, and for the presence of parasporal
bodies and crystals. Three categories of spore shape were used, round, ellipsoidal and cylindrical; oval spores were
recorded as both round and ellipsoidal, kidney shaped spores as ellipsoidal, and banana shaped spores as
cylindrical. Intermediate cases were recorded as positive for both of the shapes that they lay between. There were
three categories of spore position : terminal, subterminal and central or paracentral. Spores in several different
positions might be observed in one culture. Sporangial swelling was only recorded as positive if the distension was
substantial.
API tests. Tests in the API 20E strip and the API 5OCHB (Bacillus) strips (API Laboratory Products,
Basingstoke, Hants., UK) were used: 12 tests in the API 20E strip and 49 tests in the API 5OCHB strips. The latter
contain carbohydrate substrates for the detection of assimilation or acid production (according to the suspension
 Ident8cation of Bacillus struins 1873
medium used) and have 37 substrates in common with the obsolete API 50E gallery. The tests are listed in Table 1.
Strains were grown on plates of NA, or another appropriate medium, overnight or, for psychrophiles, 48 h.
Growth was harvested in 2 ml sterile normal saline and the suspension so produced was used to prepare two
further suspensions : (i) for API 20E strip in 4 ml sterile normal saline to correspond to tube no. 3 of the McFarland
(1907) series of standard opacities, which is approximately equivalent to 9 x lo8 organisms ml-' ; and (ii) for
API SOCHB strips in 10 ml of API 5OCHEB (Enterobacteriuceae/Bucillus) medium, which contains (g 1 - l )
(NH,)2S04, 2; yeast extract, 0-5; tryptone, 1 ; phenol red, 0.18; with mineral base of Cohen-Bazire et al. (1957)
10 ml; in a phosphate buffer of pH 7.5 after sterilization, to correspond to tube no. 3 of the McFarland scale. Only
the first twelve tests of the API 20E strip were inoculated, the last eight being carbohydrate tests duplicated in the
API 5OCHB strips.
Strips were incubated at 30 "C for 48 h and read at 24 and 48 h or, for psychrophiles, at 20 "C for 96 h, reading at
48 and 96 h or, for thermophiles, at 5 5 "C for 24 h reading at about 12 and 24 h. The API 5OCHB strips were tilted
at approximately 5", bases of tubes uppermost, during incubation in order to trap any gas evolved.
Results were scored according to the manufacturer's instructions. A test scoring positive at either reading time
was considered positive. Occasionally, tests appeared positive at the first reading but reverted to negative by the
final reading; in the API 50CH gallery this was due to the production of large quantities of alkali which masked
the acid production (alkaline reversion).
The tubes of the API 5OCHB strips were examined for gas bubbles.
Identification by computer. Logan (1980) and Logan & Berkeley (1981) computed similarity coefficients using
the general similarity coefficient ( S , ) of Gower (1971). Clusters were formed by unweighted pair-group average
linkage analysis (Sokal & Michener, 1958) and the results of several runs were illustrated by phenograms. These
served as a basis for breaking the set of strains into groups of manageable size for principal co-ordinate analysis.
The method of identification was that of Ross (1975) in which similarity coefficients are used. This is a
simultaneous method and represents a development of the diagnostic table. Gower (1968) developed a technique
for adding points to a principal co-ordinate analysis using the similarity coefficients between the new unit and
each of the reference units; this useful supplement to identification by similarity coefficients enables the distinc-
tion between intermediate and outlying strains when working with groups.
All these facilities are available in the CLASP program (written by G. J. S. Ross, F. B. Lauckner and D.
Hawkins, Rothamsted Experimental Station, Harpenden, Herts., UK) which was run on the ICL 4-75 computer
at the University of Bristol.
All 475 new strains and the 91 unnamed strains from the studies of Logan (1980) and Logan & Berkeley (1981)
were identified using the set of 509 named strains for reference.

RESULTS AND DISCUSSION

The results of the tests for the 1075 Bacillus strains are expressed as percentages of positive
results for each species or group of strains in Table 1.
The information provided by certain tests is, however, of no value in identification because
the results given are always negative. These tests are as follows: in the API 20E strip - lysine
decarboxylase, ornithine decarboxylase and tryptophan deaminase; and in the API SOCHB
strips - erythritol, L-xylose, and L-arabitol. In addition, D-fuCoSe and 2-ketogluconate appear to
be of little value. This is a surprisingly small number of redundant tests considering that the
systems were originally designed for the identification of Enterobacteriaceae and other Gram-
negative rods. Several of the remaining tests appear to be of only moderate value because they
merely support the separation achieved using tests of high discriminatory value; such support is,
however, valuable in many cases because of the high within-species variation encountered in
several species.
For several species, fewer than ten strains were available for study and in these instances the
results in Table 1 are of reduced value for identification. In 25 species or groups, however, large
numbers of strains were available and the results are considered to give an adequate indication
of within-species variation (Sneath, 1 9 7 8 ~Gordon,
; 1981) for each test.
The results shown in Table 1 are most conveniently discussed by considering each species or
group separately.
Bacillus cereus group
The positions of B . mycoides and B . thuringiensis, as species distinct from B . cereus, are not
clear. Logan (1 980) and Logan & Berkeley (1 98 1) were unable to separate these three species. It
can be seen from Table 1 that there are few differential tests available other than the familiar
 1874 N . A . LOGAN AND R . C. W . BERKELEY

Table 1. Percentage positive results for Bacillus species using morphological, supplementary, API 20E
and API SOCHB tests
The table shows, for each species or group, the percentage of strains studied giving positive reactions in
each test. Negative results (i.e. 0% positive) have been omitted for clarity.

!k
.-
"E
i!
3 2
E 2
u; 8
c d a d e r i c d c d cd 3
No. ot' strains ... 119 25 55' 30 37 44 27 10 12 11 2 2 47
Morphological and
supplementary tests
1. Cell width @m)$ 1.4 1.3 1.4 1.4 1.3 0.8 0.8 0.9 0.8 0.7 1.0 0.9 1.1
2. Chains of cells 96 100 100 100 100 29 70 25 50 100
3. Motility 96 100 100 100 100 100 100 100 100 100 42
4. Spores round 15 50
5 Spores ellipsoidal
. I 100 100 100 100 100 100 96 100 100 100 100 50 91
6 Spores cylindrical
. 8 22 20 29 17 50 26
7 Spores central/paracentral 70 84 6 30 75 52 100 58 100 100 62
8 Spores subterminal 100 100 100 100 100 100 100 90 100 100 100 100 86
9 Spores terminal 3 50 4
10 Sporangia swollen 51 30 100 92 91 100
11 Parasporal bodies 100
12 Crystalline inclusions 94
13. Vacuoles 83 100 100 100 100
14. Gas from carbohydrates
API 2OE tests
15. ONPG 32 23 89 42 100 100 100 2
16. ADH 60 36 87 17 2
17. LDC
18. ODC
19. Citrate (Simmons') 86 60 93 100 34 15 50 100 100 36
20. H,S 82
21. Urease 22 20 13 30 50 91 50 21
22. TDA
23. Indole 100 100
24. V-P 92 92 98 100 100 84 55 100 92 27 100 91
25. Gelatin 100 100 100 100 70 95 44 100 100 100 50 76
26. Nitrate 80 76 92 87 100 50 15 90 18 50 66
API SOCHB tests
27. Glycerol 92 96 92 70 98 55 100 100 100 100 100 55
28. Erythritol
29. D-Arabinose 4 100 4
30. L-Arabinose 4 70 100 100 57
31. Ribose 97 76 98 93 100 34 89 100 100 100 100 100 68
32. D-Xylose 4 18 100 100 15
33. L-Xylose
34. Adonitol 100 9
35. p-Met hylxyloside 33 50
36. Galactose 6 32 6 11 70 58 100 50 8
37. D-Glucose 100 100 100 100 100 100 100 100 100 100 100 100 98
38. D-Fructose 98 84 100 100 59 52 100 100 100 100 100 98
39. D-Mannose 8 41 3 23 96 100 42 100 100
40. L-Sorbose
41. Rhamnose 55 50 19
42. Dulcitol 15
43. Inositol 4 3 22 75 91 100 50 38
44. Mannitol 84 96 100 100 100 70
 Identijication of Bacillus strains 1875
Table 1 '(continued)

a d a d $ a d a d a d a d ? i ad$adadadadadadad$Fad a d a d a d + ad
2 5 3 18 6 54 3 2 131 52 81 63 33 44 15 15 3 5 4 20 38 4 32 3 18

Tests
I. 0.9 1.3 0.9 0.9 0.7 1.00.9 0.9 0.8 0.8 0.8 0.7 1.5 0.8 0.7 0.9 0.6 0.7 0.7 0.8 0.8 0.7 0.9 0.8 0.6
2. 100 100 100 5 100 33 50 22 84 80 5 97 23 7 13 20 50 35 66 40 83
3. 100 100 100 100 100 100100 100 95 100 100 100 97 100 100 100 100 40 100 100 100 100 100 100 100
4. 100 100100 100 69 78
5. 100 100 100 100 20 100 100 100 61 100 100 100 100 100 100 100 100 87 100 100 100 78
6. 8 20 89 20 7 7 5 37 9
7. 50 80 100 81 40 7 49 40 40 29 40 10 75 3
8. 100 80 67 100 33 79 100 100 98 100 100 100 100 84 100 100 100 100 100 97 100 100 22
9. 50 20 100 74 67 100 15 36 12 45 33 100 100 100 55 39 53 100 100
10. 33 100 100 100 100 50 5 2 84 100 100 100 100 100 85 68 100 100 100 100
11.
12.
13. 90
14. 100 100

15. 22 7 100 100 90 67 100 100 100 100 100 100 100 60 50 70 26 12 78
16. 100 95 20
17.
18.
19. 100 80 33 44 85 100 98 92 99 89 85 4 22
20.
21. 100 100 100 65 100 100 16 7 20 25
22.
23.
24. 72 11 100 100 100 100 98 97 73 67 93 100 100 100 81 100 69 33 50
25. 100 33 4 67 100 100 100 100 98 100 38 40 93 20 30 95 100 100 100 94
26. 100 77 100 14 100 95 88 96 15 13 47 100 20 60 25 16 50

27. 50 97 96 100 98 100 86 100 100 80 100 100 100 100 97 100 100
28.
29. 23 93 7 25 10 83
30. 98 82 100 98 97 91 100 100 100 75 71 25 9
31. 16 99 100 100 100 100 98 100 100 100 20 100 80 89 25 44 100 100
32. 89 71 99 98 91 98 100 100 100 100 70 66 75
33.
34. 3 50
35. 93100 100 100 3
36. 30 44 100 100 97 100100 100 100 100 100 95 60 50 84 100 100
37. 11 100 100 100 100 100 100 100100 100 100 100 100 100 100 100 100 100 100
38. 22 7 100 100 100 100 100 100 100100 100 100 100 100 100 100 100 100 100 100
39. 94 63 99 97 15 98100 100 33 100 100 95 89 75 94 100 100
40. 2 7 5 100 75 5 16 25 62
41. 2 83 5 3 45 100 33 60 50 50 8 100
42. 2 75
43. 11 95 86 76 28 100 41 40 40 26 75 28
44. II 96 100 90 100 100 88 100 100 100 100 100 40 79 100 100 50
 1876 N . A . L O G A N AND R . C . W . B E R K E L E Y
Table 1 (continued)

6 c t i q j ~ P j Q i c t i a d P j 5 s 9 6 5s
No. of strains . . . 119 25 55 30 37 44 27 10 12 11 2 2 47
API SOCHB tests
(continued)
45. Sorbitol 7 52 50 50 21
46. u-Methyl-Dmannoside 2 15 91 50
47. u-Methyl-D-glucoside 2 12 4 3 2 52 67 100 100 100
48. N-Acetylglucosamine 99 100 100 100 100 88 92 100 100 100 100 100 47
49. Amygdalin 8 24 2 70 70 50 100 100 100 4
50. Arbutin 91 84 100 60 32 13 81 100 67 100 100 100 13
51. Aesculin 100 100 98 100 97 73 100 100 100 100 100 100 34
52. Salicin 87 80 83 13 78 100 42 100 100 100 19
53. Cellobiose 84 60 72 43 7 81 90 50 91 100 100 81
54. Maltose 98 100 100 100 100 100 100 100 100 100 100 100 62
5 5 . Lactose 8 8 2 67 25 91 100 50 2
56. Melibiose 2 78 50 100 100 50 4
57. Sucrose 47 64 55 83 100 91 100 75 100 100 100 68
58. Trehalose 98 92 100 100 100 75 89 100 58 100 100 50 89
59. Inulin 7 11 50 40
60. Melezitose 1 4 2 48 17 100 50 2
61. D-Raffinose 1 7 89 50 100 100 50 42
62. Starch 96 100 94 6 97 91 100 50 100 100 100 50 36
63. Glycogen 92 100 92 10 92 54 70 67 91 100 50 36
64. Xylitol 3 11
65. /?-Gentiobiose 18 3 4 7 63 90 83 100 100 100 15
66. D-Turanose 15 24 11 6 3 13 63 67 100 100 100 2
67. DLyxose 7 50
68. D-Tagatose 2 15 50
69. D-Fucose
70. L-Fucose 2 3 4 8 100 17
71. D-Arabitol 4 15 9 50
72. L-Arabitol
73. Gluconate 29 12 6 40 3 26 8 100 50
74. 2-Ketogluconate
75. 5-Ketogluconate 4 17 100

rhizoid growth of B. mycoides and its lack of motility, and the pathogenicity for insects of B.
thuringiensis and its production of crystalline inclusions. The results suggest that B. mycoides and
B. thuringiensis should be considered as varieties of B. cereus.
Bacillus cereus strains of serotypes 1, 3, 5 and 8 (which include strains isolated in connection
with outbreaks of emetic type food poisoning: Gilbert & Parry, 1977; Gilbert et al., 1981;Taylor
& Gilbert, 1975) gave characteristic results which enable their separation from other members
of the B. cereus group (Logan et al., 1979); they are listed as B. cereus (emetic) in Table 1.
The 37 strains of B. anthracis comprised five avirulent, six low virulence, 23 virulent and three
strains of unknown virulence; it is not possible to distinguish between these using API tests.
 Identgcation of Bacillus strains 1877
Table 1 (continued)

* Bacillus cereus strains isolated in connection with outbreaks of emetic-type food poisoning and strains of serotypes
1 , 3, 5 and 8 commonly associated with such outbreaks.
f Groups 1 to 3 of Walker & Wolf (1971).
$ Mean cell width for strains studied.

Bacillus firmus and B. lentus

A spectrum-like arrangement of strains of B . jirmus, B. lentus and pigmented isolates (SM
strains) from salt marsh and sea water (Turner & Jervis, 1968) was observed by Gordon et al.
(1977) and discussed by Logan & Berkeley (1981). The species results in Table 1 incorporate
results of tests on the SM strains; these were allocated to the species to which they showed
greatest similarity. The great within-species and between-species variation of B. Jirmus and B.
lentus and the extent to which they have common characters is evident in the results shown, but
1878 N . A . LOGAN AND R . C . W . B E R K E L E Y

differentiation is possible because B. lentus strains produce acid from a wider range of carbo-
hydrates.
Bacillus laterosporus
This is a species with highly characteristic biochemical and morphological features and the
strains examined showed little variation.
Bacillus alvei and ‘B. thiaminolyticus’
Although the results shown in Table 1 suggest some similarity between the two groups and
only a few strains of each have been studied, it is clear that they may be separated easily using
API tests. As Gibson & Gordon (1974) point out, more strains of these organisms need to be
studied.
‘Bacillus carotarum’
Considerable between-strain variation is seen in this group and its homogeneity is in doubt.
Logan (1980) and Logan & Berkeley (1981) found ‘B.carotarum’ to exist as a series or spectrum of
strains and Gibson (1935) observed considerable morphological variation in strains allocated to
this species. Several of the strains studied were received as B. megaterium or B. cereus-B.
megaterium intermediates and came from the N. R. Smith collection. Smith et al. (1946, 1952)
regarded ‘B. carotarum’ as a synonym of B. megaterium but this was rejected by Gibson &
Gordon (1974), whose decision was supported by Bonde (1973, 1975), Hunger & Claus (1978,
1981), Logan (1980) and Logan & Berkeley (1981). None of the strains studied in the present
work showed appreciable similarity to B . cereus or B. megaterium.
Mesophilic species showing little or no acid production from carbohydrates
The species B. badius, B.fastidiosus, ‘B.freudenreichii’, B. brevis, B. pasteurii and B. sphaericus
all produce little or no acid from carbohydrates and the few positive results gained from tests in
the API 5OCHB strips are of little importance in differentiation. The patterns of results given in
the API 20E strip, the characteristic vegetative cell and sporangial morphologies, and the
special media requirements of certain species ( B .fastidiosus and B. pasteurii) make separation
simple.
Psychrophilic species
Of the five psychrophilic groups considered (‘B. psychrosaccharolyticus’, B. insolitus, B.
globisporus, ‘B. psychrophilus’ and B. macguariensis), none is represented by more than three
strains; the results shown are, therefore, of limited value in identification. Within-species
variation is low, except in the case of B . insolitus, and the patterns of results are individual
enough to support the status of each species.
The Bacillus subtilis group and B. megaterium
The B. subtilis spectrum or group as defined by Gibson (1944) and Gordon et al. (1973)
comprises B. subtilis, B. pumilus and B. lichenformis. To this may be added ‘B.arnyloligzaefaciens’
which was regarded by Gordon et al. (1973) as a synonym of B. subtilis, a position formalized in
the Approved Lists (Skerman et al., 1980).Logan (1980) and Logan & Berkeley (198 1) found that
B. megaterium clustered with the B. subtilis group and so it too is considered here.
O’Donnell et al. (1980) studied eight strains of each of the four species comprising the B.
subtilis group and found that they could separate B. subtilis, B. pumilus, B. lichenformis and ‘B.
amyloliquefaciens’ using API tests, DNA reassociation studies and pyrolysis gas liquid
chromatography but that ‘B.amyloliquefaciens’ could not be separated from B . subtilis when the
classical tests of Gordon et al. (1973) were used. Study of a larger number of strains by Logan
(1980) and Logan & Berkeley (1981) revealed the presence of intermediate strains which
obscured the distinction so that the species cannot now be clearly separated using the API tests.
Only two tests in Table 1, acid production from inulin, and chains of cells, are of value in
separating the two species.
 IdentiJicationof Bacillus strains 1879
The separation of the other species presents no difficulty, despite appreciable within-species
variation.
Bacillus circulans, B . macerans and B . polymyxa
Gibson & Topping (1938) found B. circulans to be a species complex and later workers have
endorsed this description (Logan & Berkeley, 1981). The results summarized in Table 1 for the
44 strains of this species studied indicate great between-strain variation. The pattern of
reactions is, however, individual enough to enable B . circulans to be separated from the two
closely related species B. macerans and B. polymyxa. Logan (1980) and Logan & Berkeley (198 1)
found B. macerans to be so closely related to B. circulans and B. polymyxa that it lay between and
almost joined them in principal co-ordinate plots.
Bacillus coagulans, ‘B . laevolacticus’ and ‘B. racemilacticus’
The species B. coagulans forms a very diffuse group showing great variation between strains so
that a species pattern of reactions is difficult to define. Other authors have also found the species
to be heterogeneous (Logan & Berkeley, 1981). ‘Bacillus laevolacticus’ and ‘B. racemilacticus’
were originally separated mainly on the basis of the isomer, or isomers, of lactic acid produced
by glucose fermentation (Gibson & Gordon, 1974); their similarity is clear from the results,
albeit of few strains, shown in Table 1, but separation is possible. The close relationship of these
two species to B. coagulans is also clear but, again, differentiation is possible.
Bacillus stearothermophilus and the ‘B . caldolyticus’ group
The results of Logan (1980) and Logan & Berkeley (1981) supported the division of B.
stearothermophilus into the three groups proposed by Walker & Wolf (1971); the results given for
B. stearothermophilus in Table 1 are divided into the three groups accordingly. Group 1 is
represented by only four strains but appears fairly homogeneous. Although groups 1 and 3 are
heterogeneous (Walker & Wolf subdivided these groups) and there are few tests available for
differentiation each has a characteristic pattern of results. It is possible that further study will
show that the members of these groups represent a discontinuous series or spectrum.
The species ‘B.caldolyticus’, ‘B.caldotenax’ and ‘B.caldovelox’ (Heinen & Heinen, 1972), each
represented by a single strain, show great similarity and a close relationship to B. stearothermo-
philus group 3.
Bacillus pan tothenticus
In the analyses of Logan (1980) and Logan & Berkeley (198 1) this species showed phenotypic
relatedness to B. coagulans and B. stearothermophilus; it is, however, a homogeneous species with
characteristic morphological and biochemical properties that allow it to be differentiated from
these two species with ease.
Other species
The foregoing covers 26 of the 3 1 Bacillus species included in the Approved Lists of Bacterial
Names (Skerman et al., 1980). The remaining five species are B. larvae, B. lentimorbus, B.
popilliae, B. alcalophilus and B. acidocaldarius; the first three of these are fastidious and require
special growth media and conditions that are more easily supplied in the classical test
procedures. Bacillus alcalophilus and B . acidocaldarius require very high and very low pH,
respectively, for growth and are not, therefore, compatible with the narrow pH ranges of the
media in API strips. These organisms may, however, be identified satisfactorily by the nature of
their requirements for growth.
A taxon-radius model type of identification system, for use with a computer, is easily
constructed from a table of per cent positive results (Sneath, 1978b). The method assumes, how-
ever, that the taxa are roughly hyperspherical and that intermediate forms are relatively
uncommon; the model breaks down if there is considerable overlap of hyperspheres (Sneath &
Sokal, 1973) but this would indicate the need for, and may be remedied by, reclassification.
Clearly, the spectra and species complexes occurring in Bacillus ( B .firmus, B. lentus and B.
1880 N . A. LOGAN A N D R . C. W . BERKELEY

circulans for example) cannot be regarded as hyperspherical and appreciable overlap of taxa
occurs. Furthermore, a minimum of ten strains should be used for each taxon in such systems
(Sneath, 19783) and several of the taxa in the present work do not satisfy this requirement.
Discriminant analysis is an extension of the taxon-radius model method and gives reliable
identification with groups that show partial overlap in their properties; it achieves this by
weighted transformation of character axes so that average within-taxon scatter is standardized,
and by altering angles between character axes to make clusters as hyperspherical as possible.
This advantage over the taxon-radius model, however, is counteracted by the method requiring
the taxa to be few in number, the clusters to be of similar size, shape and orientation, and the
unknown to belong to one of the taxa included (Sneath & Sokal, 1973; Sneath, 1978b). The first
and last of these problems are substantially overcome by the use of the generalized distance
function D 2(Mahalanobis, 1936; Sneath & Sokal, 1973) which is used in canonical variates
analysis. Such methods, however, have limited application ; large numbers of individuals are
required in each taxon but the gain in discrimination over simpler methods may be slight
(Sneath & Sokal, 1973).
A large number of taxa are considered in the present work; this, taken with the inadequate
representation of some of these taxa, argues against the use of discriminant analysis. On the
other hand, the difficulties of separating several of the groups and the high degree of phenetic
overlapping that occurs with such polythetic taxa make sequential methods, such as dicho-
tomous keys and polyclaves, inappropriate (Sneath & Sokal, 1973).
It may be concluded, therefore, that simultaneous methods such as the use of diagnostic
tables, identification by similarity coefficients, and, for the clearly defined and adequately
represented taxa, taxon-radius models are most suited to the task of identifying members of the
genus Bacillus.
Other identification schemes for Bacillus such as those of Gibson & Topping (1938), Smith et
al. (1946, 1952), Wolf & Barker (1968), Gordon (1973), and Gordon et al. (1973), based upon
dichotomous keys, and those of Cowan & Steel (1965, 1974) based upon tables, do not allow the
identification of so many species as the system described here and usually do not permit the
identification of atypical and intermediate strains. The keys of Gordon et al., (1973) and Bonde
(1978, and personal communication) for example, use 27 and 22 characters and recognize only
21 and 10 taxa respectively. The computer-assisted identification system described by Willemse-
Collinet et al. (1980), and based upon studies of 30 strains from culture collections, used 24 tests,
taking up to 7 d for the identification of only 18 species, and was unable to achieve a satisfactory
diagnosis in several instances, although misidentification did not occur. Our results enable the
recognition of 38 taxa and the availability of results for a large number of tests (Table 1) allows
the identification of atypical and intermediate strains. Establishment of the validities of some of
the taxa recognized awaits the results from studies on more strains and of further information
from other sources. The system does, however, make possible the diagnosis of 26 of the 31
Bacillus species included in the Approved Lists of Bacterial Names (Skerman et al., 1980).

REFERENCES
BONDE,G . J. (1973). The Genus Bacillus. An Examina- COWAN,S . T. & STEEL,K. J. (1965). Identification o j
tion of 460 Mesophilic Strains. Research Council Medical Bacteria. Cambridge : Cambridge Univer-
series no. 2. Copenhagen : International Service sity Press.
System. COWAN,S . T. & STEEL,K. J. (1974). Identification of
BONDE, G. J. (1975). The genus Bacillus. An experi- Medical Bacteria, 2nd edn. Cambridge: Cambridge
ment with cluster analysis. Danish Medical Bulletin University Press.
22, 41-61. DEUTSCHE SAMMLUNG VON MIKROORGANISMEN (1977).
BONDE, G. J. (1978). Current status of Bacillus Catalogue of Strains, 2nd edn, p. 276. Edited by D.
classification. Abstracts of the XI1 International Claus & C. Schaab-Engels. Miinchen : Gesellschaft
Congress of Microbiology, p. 86. Miinchen : Interna- fur Strahlen und Umweltforschung mbH.
tional Association of Microbiology Societies. GIBSON,T. (1935). The urea-decomposing microflora
COHEN-BAZIRE, G., SISTROM, W. R. & STANIER, R. Y . of soils I. Description and classification of the
(1 957). Kinetic studies of pigment synthesis by non- organisms. Zentralblatt f i r Bakteriologie, Parasiten-
sulfur purple bacteria. Journal of Cellular and kunde, Infektionskrankheitenund Hygiene, Abt. II 92,
Comparative Physiology 49, 25-68. 364-3 80.
 IdentiJicationof Bacillus strains 1881
GIBSON,T. (1944). A study of Bacillus subtilis and LOGAN,N. A. (1980). Studies on the taxonomy and
related organisms. Journal of Dairy Research 13,248- identification of members of the genus Bacillus: appli-
260. cation of miniaturized, standardized, commercially
GIBSON,T. & GORDON,R. E. (1974). Bacillus. In available test systems. PhD thesis, University of
Bergey's Manual of Determinative Bacteriology, 8th Bristol.
edn, pp. 529-550. Edited by R. E. Buchanan & N. E. LOGAN, N . A. & BERKELEY,R. C. W. (1981).
Gibbons. Baltimore : Williams & Wilkins. Classification and identification of members of the
GIBSON,T. & TOPPING,L. E. (1938). Further studies genus Bacillus using API tests. In The Aerobic
of the aerobic sporeforming bacilli. Proceedings, Endospore-jorming Bacteria : Classification and Iden-
Society of Agricultural Bacteriologists, 43-44. tification, pp. 105-140. Edited by R. C. W. Berkeley
GILBERT,R. J. & PARRY,J. M. (1977). Serotypes of & M. Goodfellow. London: Academic Press.
Bacillus cereus from outbreaks of food poisoning and LOGAN,N. A., CAPEL,B. J., MELLING, J. & BERKELEY,
from routine foods. Journal of Hygiene 78, 69-74. R. C. W. (1979). Distinction between emetic and
GILBERT,R. J., TURNBULL, P. C. B., PARRY,J. M. & other strains of Bacillus cereus using the API system
KRAMER,J. M. (1981). Food poisoning and other and numerical methods. FEMS Microbiology Letters
clinical infections associated with Bacillus species 5, 373-375.
with particular reference to B. cereus. In The Aerobic MCFARLAND, J. (1907). The nephelometer. Journal of
Endospore-jorming Bacteria : CIassification and the American Medical Association 49, 1 176-1 178.
Identification, pp. 297-314. Edited by R. C. W. MAHALANOBIS, P. C. (1936). On the generalized
Berkeley & M. Goodfellow. London: Academic distance in statistics. Proceedings of the National
Press. Institute of Sciences of India 2, 49-55.
GORDON,R. E. (1973). The genus Bacillus. In Hand- O'DONNELL, A. G., NORRIS, J. R., BERKELEY,
book of Microbiology, vol. 1, Organismic Microbi- R. C. W., CLAUS,D., KANEKO, T., LOGAN,N. .4. &
ology, pp. 71-88. Edited by A. I. Laskin & H. A. NOZAKI,R. (1980). Characterization of Bacillus
Lechevalier. Cleveland : The Chemical Rubber Co. subtilis, Bacillus pumilus, Bacillus licheniformis and
Press. Bacillus amyloliguefaciens by pyrolysis gas-liquid
GORDON,R. E. (1981). One hundred and seven years chromatography, deoxyribonucleic acid-deoxyribo-
with the genus Bacillus. In The Aerobic Endospore- nucleic acid hybridization, biochemical tests and
forming Bacteria :Classificationand Identification, pp. AP I systems . Internat ional Journal of Systematic
1-15. Edited by R. C. W. Berkeley & M. Good- Bacteriology 30, 448-459.
fellow. London : Academic Press. PRIEST,F. G. (1981). DNA homology in the genus
GORDON,R. E., HAYNES,W. C. & PANG,C. H.-N. Bacillus. In The Aerobic Endospore-jorming Bacteria :
(1973). The Genus Bacillus. Washington, DC : United Classificationand Identification, pp. 33-57. Edited by
States Department of Agriculture. R. C. W. Berkeley & M. Goodfellow. London:
GORDON,R. E., HYDE,J. L. & MOORE,J. A. (1977). Academic Press.
Bacillus firmus-Bacillus lentus: a series or one QUESNEL, L. B. (1971). Microscopy and micrometry.
species? International Journal of Systematic Bacteri- Methods in Microbiology 5A, 1-103.
ology 27, 256262. Ross, G. J. S. (1975). Rapid techniques for automatic
GOWER,J. C. (1968). Adding a point to vector identification. In Biological Identification with Com-
diagrams in multivariate analysis. Biometrika 55, puters, pp. 93-102. Edited by R. J. Pankhurst.
582-58 5 . London & New York: Academic Press.
GOWER, J. C. (1971). A general coefficient of similarity SKERMAN, V. B. D., MCGOWAN, V. &SNEATH, P. H. A.
and some of its properties. Biometrics 27, 857-874. (1980). Approved lists of bacterial names. Interna-
GREEN,D. M. (1975). Anthrax bacillus. In Medical tional Journal of Systematic Bacteriology 30, 225-
Microbiology, 12th edn, vol. 2, pp. 449453. Edited 420.
by R. Cruickshank, J. P. Duguid, B. P. Marmion & SMITH,N. R., GORDON,R. E. & CLARK,F. E. (1946).
R. H. A. Swain. Edinburgh, London & New York: Aerobic Mesophilic Sporeforming Bacteria. Miscella-
Churchill Livingstone. neous Publication no. 559. Washington, DC : United
HEINEN,U. J. & HEINEN,W. (1972). Characteristics States Department of Agriculture.
and properties of a caldoactive bacterium producing SMITH,N. R., GORDON,R. E. & CLARK,F. E. (1952).
extracellular enzymes and two related strains. Archiv Aerobic Sporeforming Bacteria. Monograph no. 16.
fur Mikrobiologie 82, 1-23. Washington, DC: United States Department of
HUNGER,W. & CLAUS,D. (1978). Bacillus megaterium : Agriculture.
A species complex. Abstracts of the XII International SNEATH,P. H. A. (1974). Test reproducibility in
Congress of Microbiology, p. 87. Munchen : Interna- relation to identification. International Journal of
tional Association of Microbiology Societies. Systematic Bacteriology 24, 508-523.
HUNGER, W. & CLAUS,D. (1981). Taxonomic studies SNEATH,P. H. A. (1978a). Classification of micro-
on Bacillus megaterium and on agarolytic Bacillus organisms. In Essays in Microbiology, pp. 9/1-9131,
strains. In The Aerobic Endospore-forming Bacteria : Edited by J. R. Norris & M. H. Richmond.
Classification and Identification, pp. 217-239. Edited Chichester: John Wiley.
by R. C. W. Berkeley & M. Goodfellow. London: SNEATH,P. H. A. (19786). Identification of micro-
Academic Press. organisms. In Essays in Microbiology, pp. 10/1-
KING,A. & PHILLIPS,I. (1978). The identification of 10/32. Edited by J. R. Norris & M. H. Richmond.
pseudomonads and related bacteria in a clinical Chichester : John Wiley.
laboratory. Journal of Medical Microbiology 11, 165- SNEATH, P. H. A. & COLLINS, V. G. (1974). A study in
176. test reproducibility between laboratories : report of a
1882 N. A. LOGAN A N D R . C. W . BERKELEY

Pseudomonas working party. Antonie van Leeuwen- WALKER,P. D. & WOLF, J. (1971). Taxonomy of
hoek 40,481-527. Bacillus stearothermophilus. In Spore Research 1971,
SNEATH,P. H. A. & SOKAL,R. R. (1973). Numerical pp. 247-262. Edited by A. N. Barker, G. W. Gould
Taxonomy. San Francisco: W. H. Freeman. & J. Wolf. London & New York: Academic Press.
SOUL, R. R. & MICHENER, C. D. (1958). A statistical WILLEMSE-COLLINET, M. F., TROMP,TH. F. J. &
method for evaluating systematic relationships. HUIZINGA, T. (1980). A simple and rapid computer-
University of Kansas Science Bulletin 44, 467-507. assisted technique for the identification of some
TAYLOR, A. J. & GILBERT, R. J. (1975). Bacillus cereus selected Bacillus species using biochemical tests.
food poisoning : a provisional serotyping scheme. Journal of Applied Bacteriology 49, 385-393.
Journal of Medical Microbiology 8, 543-550. WOLF,J . & BARKER, A. N. (1968). The genus Bacillus:
TURNER, M. & JERVIS,D. I. (1968). The distribution of aids to the identification of its species. In Identijica-
pigmented Bacillus species in saltmarsh and other tion Methods for Microbiologists, part B, pp. 93-109.
saline and non-saline soils. Nova Hedwigia 16, 293- Edited by B. M. Gibbs & D. A. Shapton. London:
297. Academic Press.