Periodic Colony Formation by Bacterial Species Bacillus subtilis

Jun-ichi Wakita, Hirotoshi Shimada, Hiroto Itoh,Tohey Matsuyama

1 and Mitsugu Matsushita*

Department of Physics, Chuo University, Kasuga, Bunkyo-ku, Tokyo 112-85511Department of Bacteriology, Niigata University School of Medicine, Niigata 951-8510

(Received September 25, 2000)

We have investigated the periodic colony growth of bacterial species Bacillus subtilis. Acolony grows cyclically with the interface repeating an advance (migration phase) and a rest(consolidation phase) alternately on a surface of semi-solid agar plate under appropriate envi-ronmental conditions, resulting in a concentric ring-like colony. It was found from macroscopicobservations that the characteristic quantities for the periodic growth such as the migrationtime, the consolidation time and the terrace spacing do not depend so much on nutrient concen-tration Cn, but do on agar concentration Ca. The consolidation time was a weakly increasingfunction of Ca, while the migration time and the terrace spacing were, respectively, weaklyand strongly decreasing function of Ca. Overall, the cycle (migration-plus-consolidation) timeseems to be constant, and does not depend so much on both Cn and Ca. Microscopically,bacterial cells inside the growing front of a colony keep increasing their population during bothmigration and consolidation phases. It was also confirmed that their secreting surfactant calledsurfactin does not affect their periodic growth qualitatively, i.e., mutant cells which cannot se-crete surfactin produce a concentric ring-like colony. All these suggest that theof the nutrient and the surfactin are irrelevant to their periodic growth.

KEYWORDS: pattern formation, bacterial colony, Bacillus subtilis, periodic growth, concentric ring-like pattern

results diffusion

∗E-mail: matusita@phys.chuo-u.ac.jp

911

softness are varied. We have so far found that quasi-two-dimensional colony patterns grown on agar surfaces areclassified into five types in the morphological diagram,i.e., diffusion-limited aggregation (DLA)1,7–10,23–26)-like (region A in Fig. 1), Eden2,3, 17,19,21,27)-like (B),concentric ring-like12,19,21)(C), homogeneously spread-ing disk-like14) (D) and dense-branching-morphology(DBM)18,21,25,28–30)-like (E).

The growth rates were also very different among thesefive morphologies. Growth to the size of about 5 cm re-quired about a month in the region A and a week in theregion B, while a day in the regions C and E and half aday in the region D. In fact, by microscope observationof growing zones of colonies, two distinct types of grow-ing processes were recognized. In the regions A and B, noactive movement of individual cells were observed. Onlycells located in the outermost part of a colony have accessto nutrient and proliferate by cell division, while cells inthe inner part change to spores and enter into a restingphase. On the other hand, in the regions C, D, and E,where agar plates are relatively soft, active cells exhibita random walk-like movement, which seems to drive thegrowth of colonies. Thus it seems that the morphologicalchanges of colonies crucially depend on the capability ofactive cell movement. In fact, Ohgiwari et al.9) foundthat a nonmotile mutant with no flagella showed onlyDLA-like and compact Eden-like colony patterns in allranges of agar concentration examined. In other words,the regions A and B in Fig. 1 expanded into the entireregion, and the regions C, D and E disappeared.9)

When bacterial cells were initially inoculated on thesurface of nutrient-poor and solid agar medium specified

teresting and exciting subjects, but it is, in general, verycomplicated. Understanding it is one of ultimate goalsof our study. Pattern formation by populations of simplebiological objects may be a good starting point becauseit is, under some conditions, dominated by purely phys-ical conditions. We will here pay attention to the colonyformation of bacteria,4–22) which are regarded as one ofthe simplest biological objects.

Throughout this experiment we used a bacterialspecies called Bacillus (B.) subtilis. This strain is rod-shaped (about 0.7µm in diameter and 2µm in length)and can move in water by collectively rotating flagella.However, when the environmental condition is adverse,such as on a nutrient-poor or dry substrate, bacterialcells become spores. Here we used the wild type strain(OG-01).

Figure 1 shows our results9,14,17–19,21) obtained so faron the morphological diagram of bacterial colonies asa function of both the nutrient concentration Cn andthe inverse of agar concentration Ca. In general, bac-terial colonies exhibit a variety of patterns, depend-ing on both bacterial species and environmental con-ditions.4–22) We have been mainly concerned with howbacterial colonies change their shapes as environmentalconditions such as nutrient concentration and medium

§1. Introduction

Recently much attention has been paid to pattern for-mation in various fields from various viewpoints.1–3) Pat-tern formation in biological systems is one of the most in-

Journal of the Physical Society of Japan

Vol. 70, No. 3, March, 2001, pp. 911–919

912 Jun-ichi Wakita et al. (Vol. 70,

in the region A, a colony exhibited the DLA-like ram-ified pattern and had clear self-similarity with a frac-tal dimension of about 1.7. The growth process is gov-erned by the diffusion of nutrient towards the colony,7,8)

and is now known to be explained in terms of DLAmodel.1,10,23–26) As Cn increased, DLA-like branchesthickened gradually and finally fused together to forma compact pattern, similar to an Eden cluster,2,3, 9) inthe region B. The growing interface showed, however, aself-affine fractal,17,19,21,27) which was different from theone expected from the Eden model.2,3) In the region Dwhere the agar is soft and nutrient-rich, a colony grewto a homogeneous and disk-like pattern.14) Under the as-sumption that active movement of cells be regarded asBrownian movement, it was confirmed that the colonygrowth is consistent with the solution behavior of theFisher’s equation in population dynamics. In the re-gion E where the nutrient is poor and the agar soft-ness is intermediate, a colony grew radially with a dense-branching-morphology (DBM),18,21,25,28–31) and its ad-vancing envelope was smoothly rounded in contrast toDLA-like colonies. According to microscopic observa-tions there seem to be two types of bacterial cells; ac-tive and inactive cells, the former of which compose agroup named as “finger-nail” structure on the tip of eachgrowing branch and drive the tip growth.18) DBM-likecolonies with thinner branches and gaps display shorterbranches and move faster, everything depending simplyon nutrient concentration Cn.18) In the upper part ofthe region C, a colony grew by producing a concentricring-like pattern (Fig. 2(a)) that was found to be formedby alternating the region D-like colony expansion (veryfast) and the region B-like growth (very slow).12,19,21) Aconcentric ring-like pattern like this is also formed by abacterial species Proteus (P.) mirabilis.15,16,20,22) Thisspecies is famous for forming colonies of an impressively

regular concentric ring-like pattern (Fig. 2(b)) and itsbiological features are well-known.

Based on our observations described above, somereaction-diffusion type models for the population den-sity of bacterial cells and the concentration of nutrienthave been proposed.19,21,30–34) Mimura’s model19,21,34)

is one of them and its essential assumption inspired byour observation is that there exist two types of bacterialcells; active cells that move actively, grow and performcell division, and inactive ones that do nothing at all.This model is found to be able to reproduce globally allthe colony patterns seen in the experimentally obtainedmorphological diagram shown in Fig. 1, and is appar-ently quite satisfactory.

The pattern formation of bacterial colonies of speciesB. subtilis has just about to be elucidated from bothexperimental and numerical points of view. In this paperwe will especially focus on the concentric ring-like colonyin the region C, show experimental results and discusstentative attempts to elucidate the growing mechanism.

§2. Experimental Procedures

In our morphological diagram shown in Fig. 1 we var-ied only two environmental conditions; concentrationsof nutrient Cn and agar Ca of the medium agar plates.Other parameters such as temperature (35◦C) and hu-midity (90%RH) were kept constant.

Experimental procedures are very simple. A solu-tion containing 5 g of sodium chloride (NaCl), 5 g ofdipotassium hydrogenphosphate (K2HPO4) and a speci-fied amount of BACTO-PEPTONE Laboratories,Detroit, USA) as nutrient in 1 liter of distilled water wasmade, and adjusted the solution at pH = 7.1 by adding6N hydrochloric acid. The solution was then mixed witha specified amount of BACTO-AGAR (Difco). Valuesof Cn and Ca as the environmental parameters were de-termined at this stage. The solution was autoclaved at121◦C for 15 min, and then usually 20 ml of the solutionwas poured into each sterilized plastic petri dish withan inner diameter of 88 mm. But in this experiment forring-like colonies in the region C, we specially used thepetri dish with 10 ml of the solution poured, because wewanted to see more number of rings. After solidificationat room temperature for 60 min, the agar plates weredried at 50◦C for 120 min. The thickness of the agarplates thus prepared is about 1.6 mm.

3µl of the liquid culture which was adjusted at opti-cal density (OD) of 0.5 (wavelength 600 nm) was inocu-lated at the center of each agar plate surface. The plateswere incubated in a humidified box at 35◦C and 90%RHfor designated time. Bacterial colonies grew quasi-two-dimensionally on the agar plate surface.

Macroscopic colony patterns obtained thus were pho-tographed by a digital camera C1400L (Olympus,Tokyo). The growth of colonies was recorded on videotapes by a time-laps video SVT-S5100 (SONY, Tokyo)through CCD cameras CS572S (Sankei, Tokyo) and MC-780P (Texas Instruments, USA). Microscopic growthprocesses were also recorded on video tapes through amicroscope DIAPHOT-TMD (Nikon, Tokyo).

(Difco

Fig. 1. Morphological diagram of Bacillus subtilis colonies ob-tained when varying both the concentration of nutrient Cn andsolidity of agar medium expressed as [1/Ca]. Here Ca is theagar concentration. Incubation temperature and humidity wereset at 35◦C and 90%RH, respectively. Colony patterns are clas-sified into five types, i.e., DLA-like (region A), Eden-like (re-gion B), concentric ring-like (region C), homogeneously spread-ing disk-like (region D) and DBM-like (region E) patterns.

2001) Periodic Colony Formation by Bacterial Species Bacillus subtilis 913

3.3 Microscopic observationsWe can see the colony growth microscopically by an

optical microscope. There are many bacterial cells in-side the colony and they move around actively duringmigration phases. Figures 3(b1)–3(b8) show a series ofmicroscopic snapshots of a colony during its growth inthe region C. After inoculation on an agar plate sur-face the population density of bacterial cells inside theinoculation spot increased gradually (initial lag phase:Figs. 3(b1) and 3(b2)). Then many branches consist-ing of groups of active bacterial cells came out sud-denly from the edge of the inoculation spot (begin-ning of the first migration phase: Fig. 3(b3)). Thebranches fused together at many places during the migra-tion phase (Fig. 3(b4)). The end of the migration phasewas approaching with no intimation, i.e., the growth ofbranches suddenly stopped. It was the beginning of the

colony growth, as for P. mirabilis.15,16,20,22) The lagphase represents the initial stage of colony growth frominoculation till the first migration. The colony itself doesnot grow during this period, although cell division takesplace actively and cell number increases inside the inoc-ulation spot (Figs. 3(b1) and 3(b2)). It looks as if cellsprepare their first migration (Figs. 3(a1) and 3(a2)). Thecolony then grows (first migration phase: Figs. 3(a3) and3(a4)) and rest (first consolidation phase: Figs. 3(a5) and3(a6)) alternately, forming a concentric ring-like pattern,as seen in Fig. 2(a). Each colony terrace corresponds toone migration-plus-consolidation cycle. These featuresare similar to the case of a P. mirabilis colony exceptits much more regular and smoothly rounded terraces(Fig. 2(b)).

First of all, we examined the dependences of migra-tion time, consolidation time and terrace spacing on con-centrations of nutrient Cn and agar Ca. As shown inFigs. 4(a) and 4(b), it was confirmed that the migra-tion time, the consolidation time and the terrace spac-ing were almost constant as a function of Cn over thewide range. On the other hand, as shown in Figs. 5(a)and 5(b), it was observed that the consolidation timeis a weakly increasing function of Ca, while the mi-gration time and the terrace spacing are, respectively,weakly and strongly decreasing function of Ca even insuch a narrow range. Overall, the cycle (migration-plus-consolidation) time seems to be constant, and does notdepend so much on both Cn and Ca.

3.2 Mesoscopic observationsOne or more internal waves were sometimes observed

to advance toward the growing front from around theinner terrace during migration phase (Fig. 6(a)). Theywere seen with naked eyes, too. The internal waves con-sisted of many active bacterial cells (Fig. 6(b)), whichoriginated around the edge of the inner terrace. In thecase of a P. mirabilis colony,16,20) many internal wavesalso advance successively towards the outermost terracefront, even after the front stopped moving. Each internalwave of a B. subtilis colony consists of a superimposedmonolayer of swarming cells moving towards the growingfront.

just§3. Experimental Results

3.1 Macroscopic observationsFigure 2(a) shows a whole colony grown on a nutrient-

rich semi-solid agar medium (Ca = 6.8 g/l, Cn = 40 g/l).Such an environmental condition is located in the upperpart of the region C in Fig. 1. The growth of a concentricring-like colony as shown in Fig. 2(a) was not sensitive forhigh nutrient concentration Cn (≥ 30 g/l), but was verysensitive to agar concentration Ca since it was only seenin a narrow range of Ca from 6.5 g/l to 6.8 g/l betweenthe regions B and D. By macroscopic observations with atime-laps video system, it was found that the concentricring-like pattern was formed by alternating the region D-like colony expansion (very fast) and region B-like growth(very slow), respectively.

Figures 3(a1)–3(a8) show a series of macroscopic snap-shots of the colony growth. We can distinguish threephases (initial lag phase, the following migration andconsolidation phases that appear alternately) during the

Fig. 2. Concentric ring-like colonies of (a) Bacillus subtilis, and(b) Proteus mirabilis. The photographs were taken 2 days afterinoculation. The inner diameter of each petri-dish is 88 mm.

914 Jun-ichi Wakita et al. (Vol. 70,

2001) Periodic Colony Formation by Bacterial Species Bacillus subtilis 915

Fig. 3. (a1)-(a8) Macroscopic snapshots of the colony growth in the region C. Ca = 6.5 g/l, Cn = 46 g/l. The inner diameter of thepetri-dish is 88 mm. (a1), (a2): Initial lag phase; the initial stage of colony growth from inoculation till first migration. It takesabout 11 hours. The circular spot in the central area (about 5 mm in diameter) is the inoculation spot. (a3), (a4): First migrationphase. The colony started growing. (a5), (a6): First consolidation phase. After migration of 2–3 hours, the growth of the colonystopped, which is the beginning of the consolidation phase. (a7), (a8): Second migration phase. After consolidation of 3–4 hours,the colony starts the growth again and enter the second migration phase. (b1)-(b8) Microscopic snapshots of the colony growth inthe region C. Each one corresponds to the same number of the macroscopic one. The width of each photo is about 2 mm. (b1),(b2): Initial lag phase. After inoculation on an agar plate surface, the population density of bacterial cells increased gradually. (b3),(b4): First migration phase. Many branches consisting of groups of active bacterial cells came out suddenly from the inoculationspot, and then branches fused together at many places during the migration phase. (b5), (b6): First consolidation phase. Branchesstopped growing, but cell density continued to increase at the outermost terrace of the colony. (b7), (b8): Second migration phase.Many branches consisting of groups of active bacterial cells came out suddenly from the interface of the outermost terrace of thecolony.

§4. Discussion

In this paper we have investigated experimentally thegrowth of concentric ring-like colonies of bacterial speciesB. subtilis. Periodic patterns are also seen in otherbacterial species such as P. mirabilis as well as in na-ture and in many branches of science. Famous exam-ples are Belousov-Zhabotinsky reaction in chemical re-action and Liesegang band formation in crystal growthin gels. Although the appearance of periodic patternsin both physico-chemical systems and biological systemslook similar to each other, there are certainly some dif-

first consolidation phase (Fig. 3(b5)). During the con-solidation phase cell density continued to increase andbranches were thickening gradually. Especially the celldensity increased more and more at the edge of the outer-most terrace of the colony (Fig. 3(b6)). When the secondmigration phase started, many branches again came outfrom the outermost terrace (Fig. 3(b7)). This alternateprocess of migration and consolidation repeats again andagain afterwards.

3.4 Periodic growth by mutant cells which do not se-crete surfactant

It is known that bacterial species B. subtilis secretessurfactant called surfactin which plays an important rolefor the cell movement on the surface of an agar plate.The periodic growth of a B. subtilis colony that alter-nates a migration phase similar to the growth in theregion D and a consolidation phase similar to the onein the region B is only seen in the region C of Fig. 1.This region lies in a very narrow range of the agar con-centration Ca between the region B where bacterial cellsdo not move actively inside the colony and the region Dwhere they move actively inside it. It is, therefore, con-ceivable that the existence of their secreta surfactin hassomething to do with the appearance of the region C. Toconfirm this we examined whether the mutant cells of B.subtilis which do not secrete surfactin make a concentricring-like colony. If they still make a concentric ring-likecolony, then their secreta surfactin is not relevant to theperiodic growth.

The mutant cells used here that cannot secrete sur-factin was obtained from the wild type strain (OG-01)by nitrosoguanidine-mutagenesis. Figure 7 clearly showsthe periodic growth of the mutant cells. The result thatthe mutant cells were also able to form a concentric ring-like colony clearly implies that the surfactin is not essen-tial for the emergence of periodic growth.

Fig. 4. (a) Migration time (open circles), consolidation time(open squares) and cycle (migration-plus-consolidation) time(open diamonds) are shown as a function of nutrient concen-tration Cn. It was found that they are almost constant for Cn,and the averaged values of the migration, consolidation and cycletimes are given, respectively, by 2.1, 5.1 and 7.1 hr. (b) Terracespacing is shown as a function of nutrient concentration Cn. Itwas found to be almost constant for Cn and the avaraged valueof the terrace spacing was given by 5.3 mm. Averages were takenover at least 10 measurements. We omitted the first migrationand consolidation times and the first terrace spacing from av-eraging since they were very scattered and difficult to measure.Agar concentration was fixed at Ca = .8 g/l.6

916 Jun-ichi Wakita et al. (Vol. 70,

seems to drive the colony growth. The cell populationdensity around the colony interface in the consolidationphase continued to increase. Furthermore, even if thenutrient concentration Cn is increased very much in theregion C in Fig. 1, the periodic growth can be observedeven more clearly. These results suggest that the diffu-sion of the nutrient is irrelevant for their periodic growth.

B. subtilis cells secrete surfactant called surfactinwhich plays an important role for their movement onthe surface of an agar plate. The surfactin certainly en-hances the cell motility. As shown in §3.4, however, themutant cells which do not secrete surfactin were also ableto form a concentric ring-like colony. This means thattheir secreta surfactin is irrelevant to the occurrence oftheir periodic growth. It is, therefore, implied that theimprovement of cell motility does not necessarily giverise to the periodic growth.

From our macroscopic observations for the periodicgrowth shown in §3.1, it was found that the spatio-temporal characteristics of a concentric ring-like colony,such as the migration time, the consolidation time andthe terrace spacing were almost constant for nutrientconcentration Cn, but not constant for agar concentra-

ferent features. Nevertheless, we still hope to understandto some extent the growth mechanisms in biological sys-tems from a physics viewpoint. There may exist someuniversal mechanisms even for biological pattern forma-tion in common with physico-chemical systems.

When we meet such a regular concentric ring-like pat-tern made by some microorganisms, we tend to thinkthat it may be caused by their genes. However, we stillhope that we can understand the periodic growth mech-anism in B. subtilis colonies from a macroscopic view-point. In §3.3 we showed the periodic growth process byan optical microscope. It turned out that the internalwaves consist of many active bacterial cells and its flow

Fig. 5. (a) Migration time (open circles), consolidation time(open squares) and cycle (migration-plus-consolidation) time(open diamonds) are shown as a function of agar concentrationCa. It was found that the migration time is a weakly decreas-ing function and the consolidation time is a weakly increasingfunction. (b) Terrace spacing is shown as a function of agar con-centration Ca. It was found to be a strongly decreasing functionfor Ca. Averages were taken over at least 10 measurements. Weused here only second migration time, second consolidation timeand second terrace spacing, since the other ones were difficult tomeasure. Nutrient concentration was fixed at Cn = g/l.46

Fig. 6. Snapshots of an internal wave. (a) Macroscopic observa-tion. The internal wave is seen as a ring on the first terrace andadvances towards the growing front from inside. The inner di-ameter of the petri-dish is 88 mm. (b) Microscopic observation.The internal wave consists of active bacterial cells. Width of thephoto is about 2 mm.

2001) Periodic Colony Formation by Bacterial Species Bacillus subtilis 917

is normally flagellated and short (about 1.5–2.0µm inlength) and a swarmer cell which is hyperflagellated andlong (about 10–80µm in length) when it is cultivatedon the surface of an agar plate. The swarmer cells havestrong motility thanks to their hyperflagellation and they

tion Ca. The more the agar concentration increasedwithin the narrow range of Ca in the region C (whichmeans that the harder the agar substrate became), thenarrower the terrace spacing became. This is rather triv-ial. The more interesting point is that the migrationtime was shortened and the consolidation time extended.Overall, the cycle (migration-plus-consolidation) timeseems to be constant, and does not depend so much onboth Cn and Ca. The insensitivity of the spatio-temporalcharacteristics on Cn suggests again that the diffusion ofnutrient is irrelevant for the periodic growth seen in theregion C.

These dependencies on Cn and Ca were also seen forP. mirabilis colonies (Fig. 2(b)) which grow smoothlyrounded concentric ring-like colonies in more wide rangeof Ca than B. subtilis colonies and its cycle time is ap-parently constant independently of environmental condi-tions such as Cn and Ca.16) In our previous paper20) itwas found that P. mirabilis swarmer cells move collec-tively and there seem to be two threshold values of cellpopulation density for their capability of activity. That isto say, when the cell population density at the migrationfront decreases less than some lower threshold, then thefront stops, while when the cell density at the outermostconsolidation terrace increases more than some higherthreshold, then the following migration starts. All theexperimental results described in this paper also suggestthat the periodic growth of B. subtilis may be explainedby the same mechanism as that of P. mirabilis. If otherbacterial species also form concentric ring-like colonies,then it is considered to be a universal pattern. We haveconfirmed experimentally that the bacterial species Ser-ratia marcescens forms a concentric ring-like colony, too.P. mirabilis cannot secrete surfactant such as B. subtilisdoes, but they exhibit the cyclic process of cell differenti-ation and dedifferentiation between a swimmer cell which

h

e

h

e

Fig. 7. Concentric ring-like colony of mutant cells of B. subtiliswhich do not secrete surfactant called surfactin. Ca = 7 g/l,Cn = g/l.15

918 Jun-ichi Wakita et al. (Vol. 70,

1) T. Vicsek: Fractal Growth Phenomena, 2nd ed. (World Sci-entific, Singapore, 1992).

2) Dynamics of Fractal Surfaces; ed. F. Family and T. Vicsek(World Scientific, Singapore, 1991).

3) A.-L. Barabasi and H. E. Stanley: Fractal Concepts in SurfaceGrowth (Cambridge University Press, Cambridge, 1995).

4) P. Singleton: Introduction to Bacteria, 5th ed. (Wiley, NewYork, 1999).

y

Prof. Masayasu Mimura on his 60th birthday.

5) T. Matsuyama, M. Sogawa and Y. Nakagawa: FEMS Micro-biol. Lett. 61 (1989) 243.

6) H. Fujikawa and M. Matsushita: J. Phys. Soc. Jpn. 58 (1989)3875.

7) M. Matsushita and H. Fujikawa: Physica A 168 (1990) 498.8) H. Fujikawa and M. Matsushita: J. Phys. Soc. Jpn. 60 (1991)

88.

migrate in group by making rafts together to expandtheir territory. They can, therefore, move on an harderagar surface.

Some numerical models have been proposed, whicare considered as one of other important approaches toundesrtand the growth mechanism.19,21,30–34) Mimura’smodel19,21,34) is one of those models, and its essentialassumption is that there exist two types of bacterialcells; active cells that move actively, grow and performcell division, and inactive ones that do nothing at all.This model is found to be able to reproduce globallyall the colony patterns seen in the experimentally ob-tained morphological diagram shown in Fig. 1, and ap-parently quite satisfactory. However, the model explainsthe periodic growth by the competition of the cell move-ment and nutrient depletion and diffusion, which seemsto be irrelevant to our case. Cell division was also ob-served to take place rather actively during consolidationphases (§3.3), in contradiction to the model. Lacasta etal. also proposed two coupled reaction-diffusion equa-tions for bacteria and nutrient concentrations, where thebacterial diffusion coefficient can adopt two different ex-pressions, corresponding to two possible mechanisms ofmotion between short bacteria and long bacteria. Thismodel was also found to be able to reproduce the fivdifferent colony patterns. However, it is not clear thatthe assumption of the bacterial diffusion coefficient whicis based on two possible mechanisms of cell motion forshort bacteria and long bacteria is true in the case of B.subtilis. Making plausible and consistent models is stillan interesting future problem.

Although it is very difficult to understand the growthmechanism of concentric ring-like patterns in various sys-tems, we still believe that there may exist simple and uni-versal mechanisms which are independent of individualcomponents and common to many systems. We hopthat the experimental results presented here will con-

Acknowledgments

We are grateful to K. Watanabe, T. Arai, N.Kobayashi, I. Rafols, M. Itoh and O. Moriyama for manstimulating discussions. This work was supported in partby Grants-in-Aid for Scientific Reserch from the Min-istry of education, Science and Culture of Japan (Nos.09640471 and 11214205). This paper is dedicated to

tribute to elucidate them.

9) M. Ohgiwari, M. Matsushita and T. Matsuyama: J. Phys. Soc.Jpn. 61 (1992) 816.

10) T. Matsuyama and M. Matsushita: Appl. Environ. Microbiol.58 (1992) 1227.

11) E. Ben-Jacob, H. Shmueli, O. Shochet and A. Tenenbaum:Physica A 187 (1992) 378.

12) H. Fujikawa: Physica A 189 (1992) 15.13) T. Matsuyama and M. Matsushita: Crit. Rev. Microbiology

19 (1993) 117.14) J. Wakita, K. Komatsu, A. Nakahara, T. Matsuyama and M.

Matsushita: J. Phys. Soc. Jpn. 63 (1994) 1205.15) A. Nakahara, Y. Shimada, J. Wakita, M. Matsushita and T.

Matsuyama: J. Phys. Soc. Jpn. 65 (1996) 2700.16) O. Rauprich, M. Matsushita, C. J. Weijer, F. Siegert, S. E.

Esipov and J. A. Shapiro: J. Bacteriol. 178 (1996) 6525.17) J.Wakita, H. Itoh, T. Matsuyama and M.Matsushita: J.Phys.

Soc. Jpn. 66 (1997) 67.18) J. Wakita, I. Rafols, H. Itoh, T. Matsuyama and M.

Matsushita: J. Phys. Soc. Jpn. 67 (1998) 3630.19) M. Matsushita, J. Wakita, H. Itoh, I. Rafols, T. Matsuyama,

H. Sakaguchi and M. Mimura: Physica A 249 (1998) 517.20) H. Itoh, J.Wakita, T. Matsuyama and M.Matsushita: J.Phys.

Soc. Jpn. 68 (1999) 1436.21) M. Matsushita, J. Wakita, H. Itoh, K. Watanabe, T. Arai, T.

Matsuyama, H. Sakaguchi and M. Mimura: Physica A 274(1999) 190.

22) T. Matsuyama, Y. Takagi, Y. Nakagawa, H. Itoh, J. Wakitaand M. Matsushita: J. Bacteriol. 182 (2000) 385.

23) T. A. Witten and L. M. Sander: Phys. Rev. Lett. 47 (1981)1400.

24) The Fractal Approach to Heterogeneous Chemistry; ed. D.Avnir (Wiley, Chichester, 1989).

25) E. Ben-Jacob and P. Garik: Nature 343 (1990) 523.26) P. Meakin: Phase Transitions and Critical Phenomena,

Vol. 12, ed. C. Domb and J. L. Lebowitz (Academic Press,New York, 1988) p. 335.

27) T. Vicsek, M. Cserzo and V. K. Horvath: Physica A 167(1990) 315.

28) Y. Sawada, A. Dougherty and J. P. Gollub: Phys. Rev. Lett.56 (1986) 1260.

29) D. Grier, E. Ben-Jacob, R. Clarke and L. M. Sander: Phys.Rev. Lett. 56 (1986) 1264.

30) K. Kawasaki, A. Mochizuki, M. Matsushita, T. Umeda and N.Shigesada: J. Theor. Biol. 188 (1997) 177.

31) S. Kitsunezaki: J. Phys. Soc. Jpn. 66 (1997) 1544.32) Y. Kozlovsky, I. Cohen, I. Golding and E. Ben-Jacob: Phys.

Rev. E 59 (1999) 7025.33) A. M. Lacasta, I. R. Cantalapiedra, C. E. Auguet, A.

Penaranda and L. Ramırez-Piscina: Phys. Rev. E 59 (1999)7036.

34) M. Mimura, H. Sakaguchi and M. Matsushita: Physica A 282(2000) 283.

2001) Periodic Colony Formation by Bacterial Species Bacillus subtilis 919