You are on page 1of 13

Water Research 37 (2003) 661–673

Mechanisms and models for anaerobic granulation in upflow


anaerobic sludge blanket reactor
Yu Liu*, Hai-Lou Xu, Shu-Fang Yang, Joo-Hwa Tay
Environmental Engineering Research Centre, School of Civil and Environmental Engineering, Nanyang Technological University,
50 Nanyang Avenue, Singapore 639798, Singapore
Received 28 December 2001; accepted 1 July 2002

Abstract

Upflow anaerobic sludge blanket (UASB) reactor has been employed in industrial and municipal wastewater
treatment for decades. However, the long start-up period required for the development of anaerobic granules seriously
limits the application of this technology. In order to develop the strategy for rapid UASB start-up, the mechanisms for
anaerobic granulation should be understood. This paper attempts to provide a up-to-date review on the existing
mechanisms and models for anaerobic granulation in the UASB reactor, which include inert nuclei model, selection
pressure model, multi-valence positive ion-bonding model, synthetic and natural polymer-bonding model, Capetown’s
model, spaghetti theory, syntrophic microcolony model, multi-layer model, secondary minimum adhesion model, local
dehydration and hydrophobic interaction model, surface tension model, proton translocation-dehydration theory,
cellular automaton model and cell-to-cell communication model. Based on those previous works, a general model for
anaerobic granulation is also proposed. It is expected that this paper would be helpful for researchers to further develop
a unified theory for anaerobic granulation and technology for expediting the formation of the UASB granules.
r 2002 Elsevier Science Ltd. All rights reserved.

Keywords: Anaerobic granulation; Granule; Mechanism; Model; UASB

1. Introduction granular sludge is the core component of an UASB


reactor. Sludge granules are dense, multi-species, micro-
The upflow anaerobic sludge blanket (UASB) reactor bial communities, and none of the individual species in
is considered desirable in high-strength organic waste- the granular ecosystem is capable of degrading complex
water treatment because of its high biomass concentra- organic wastes.
tion and rich microbial diversity [1–5]. The high biomass One major drawback of the UASB reactor is its
concentration implies that contaminant transformation extremely long start-up period, which generally requires
is rapid, and highly concentrated or large volumes of 2–8 months for the development of anaerobic granular
organic waste can be treated in compact reactors. As sludge. In order to reduce the space-time requirements
compared to other anaerobic technologies, such as of bioreactors and eventually lead to a cheaper
anaerobic filter, anaerobic sequencing batch reactor, treatment of high strength wastes, strategies for expedit-
anaerobic expanded bed and fluidized bed reactors, the ing granular formation are highly desirable. For
UASB system is highly dependent on its granulation achieving such a purpose, the mechanisms for anaerobic
process with a particular organic wastewater. Anaerobic granulation should be thoroughly understood. This
paper attempts to review the existing mechanisms and
*Corresponding author. Tel.: +65-7906913; fax: +65- models for anaerobic granulation in UASB, and also
7910676. tries to build up a more general model for microbial
E-mail address: cyliu@ntu.edu.sg (Y. Liu). granulation.

0043-1354/02/$ - see front matter r 2002 Elsevier Science Ltd. All rights reserved.
PII: S 0 0 4 3 - 1 3 5 4 ( 0 2 ) 0 0 3 5 1 - 2
662 Y. Liu et al. / Water Research 37 (2003) 661–673

2. Physico-chemical models 2.2. Selection pressure model

Microbial adhesion or self-immobilization is the The basis of anaerobic granulation had been proposed
starting point of anaerobic granulation process, and to be the continuous selection of sludge particles that
can be defined in terms of the energy involved in the occurs in the reactors, i.e. light and dispersed sludge
interaction of bacterium-to-bacterium or bacterium-to- would be washed out, while heavier components remain
solid surface. In a thermodynamic sense, when one in the system [2]. The selection pressure model suggests
bacterium approaches another, the interaction between that microbial aggregation in UASB reactor may be
them includes repulsive electrostatic force, attractive van effective protection strategies against high selection
de Waals force, and repulsive hydration interaction. pressures. In UASB reactor, selection pressure is
Based on the thermodynamics, some physico-chemical generally created by upflow flow pattern. Absence of
models for anaerobic granulation have been developed, anaerobic granulation was observed when the hydraulic
those include inert nuclei model, selection pressure selection pressure was very weak [8,9]. Fast production
model, multi-valence positive ion-bonding model, ECP of anaerobic granules was achieved through a purely
bonding model, synthetic and natural polymer-bonding physical aggregation due to the hydraulic stress applied
model, secondary minimum adhesion model, local to the anaerobic flocculant sludge [10]. Results showed
dehydration and hydrophobic interaction model and that flocculant anaerobic sludge could be converted to a
surface tension model. relatively active anaerobic granular sludge by enhancing
agglomeration with only hydraulic stress in a very short
2.1. Inert nuclei model time less than 8 h. Similarly, Arcand et al. [11] also
reported that the liquid upflow velocity had a significant
The inert nuclei model for anaerobic granulation was positive effect on mean granule size, but its effect on the
initially proposed by Lettinga et al. [1]. In the presence specific washout rate of the smaller particles was little. It
of inert microparticles in an UASB reactor, anaerobic is most likely that relatively high selection pressure in
bacteria could attach to the particle surfaces to form terms of upflow velocity seems to be in favor of fast
the initial biofilms, namely embryonic granules (Fig. 1). formation and production of anaerobic granules.
The mature granules can be further developed through
the growth of these attached bacteria under given 2.3. Multi-valence positive ion-bonding model
operation conditions. The inert nuclei model suggests
that the presence of nuclei or micro-size biocarrier for Since bacteria have negatively charged surfaces under
bacterial attachment is a first step towards anaerobic usual pH conditions, a basic idea to expedite anaerobic
granulation. The inert nuclei model was supported by granulation process is to reduce electrostatic repulsion
experimental evidence such that addition of zeolite or between negatively charged bacteria by introducing
hydro-anthracite particles with a diameter of 100 mm multi-valence positive ion, such as calcium, ferric,
into inoculated sludge seemed to be effective in aluminum or magnesium ions, into seed sludge
promoting the formation of anaerobic granules [6]. (Fig. 2). Reduced electrostatic repulsion between bacter-
Water absorbing polymer (WAP) particles were also ia would promote anaerobic granulation [12–14]. Addi-
used to enhance granulation [7]. The WAP is a tion of Ca2+ with concentrations of 80–200 mg/l, Mg2+
pulverulent resin, which swells in water and exhibits a of 12–120 mg/l or Al3+ of 300 ml/l was found to increase
complex network structure, which can provide more the rate of anaerobic granulation in UASB reactors
surfaces for microbial attachment grow than other inert [13,15,16]. However, high calcium concentration of over
particles. The results of lab-scale experiments indicated 500 mg/l [17,18] or 600 mg/l [14] was detrimental to
the availability of accelerated startup by using WAP, anaerobic granulation. High calcium concentration also
and the contact between particles and biomass could be causes serious operation problems, e.g. precipitation and
improved since the WAP has lower density than sand accumulation of calcium in anaerobic granules and
and other inert materials [7]. reduced microbial activity of granules.

Attachment Growth

Inert nucleus Individual bacterium

Fig. 1. Schematic representation of the inert nuclei model.


Y. Liu et al. / Water Research 37 (2003) 661–673 663

Self-immobilization

Opposite charge
attraction

Multi-valence positive ion Negatively charged bacterium

Fig. 2. Schematic representation of the multi-valence positive ion-bonding model.

Individual bacterium

Shaped

Bridging

Polymeric chain or filamentous bacterium


Fig. 3. Schematic representation of the polymer or filament bonding model.

The multi-valence positive ion-bonding model is ing of exopolysaccharides synthesis would prevent
based on a simple electrostatic interaction between microbial aggregation [22,23]. It had been reported that
negatively charged bacteria and positive ion as illu- ECPs could change the surface negative charge of the
strated in Fig. 2. The DLVO theory shows that when the bacteria, and thereby bridge two neighbor cells physi-
two surfaces carry a charge of the same sign, there is a cally to each other as well as with other inert particulate
free energy barrier between them, which acts as a matters as illustrated in Fig. 3 [3,22,24]. Chen and Lun
repulsive force. This force could seriously prevent the [25] also observed that with increasing the organic
approach of one cell to another. The positive ion added loading rate, Methanosarcina grew in a significant way
to sludge would partially neutralize the negative charges and secreted much more ECPs to form larger clumps,
on bacterial surfaces by adsorption, thereby electrical and then Methanothrix tended to fill in the Methano-
repulsion between bacteria would be decreased in a sarcina clumps.
significant way, i.e. positive ion could initiate cell-to-cell
interaction, which is a decisive step towards granulation. 2.5. Synthetic and natural polymer-bonding model
The multi-valence positive ion could also compress the
double layer to promote cell aggregation [19]. In Synthetic polymers have been widely used in water
addition, the multi-valence positive ion may promote coagulation and flocculation processes, and can sig-
sludge granulation by bonding with extracellular poly- nificantly promote particle agglomeration. Similarly, the
mers (ECPs). The high affinity between ECPs and synthetic polymers can also be applied to expedite
calcium ion had been shown [20,21]. This in turn implies the development of UASB granules. It was found that
that calcium ion may bridge ECPs and ECPs and link the supplementation of polymer Chitosan, which has
cells and ECPs together to form an initial three a similar structure to polysaccharides, significantly
dimension structure of microbial community, in which enhanced the formation of anaerobic granules in the
bacteria can further grow. USAB-like reactors, e.g. the granulation rate in the
Chitosan-containing reactor was 2.5-fold higher than
2.4. ECP bonding model that in the control reactor without addition of the
polymer, while the specific activities of methane
The ECPs can mediate both cohesion and adhesion of production were comparable in both reactors [26]. In
cells, and play a crucial role in maintaining structural fact, it is not surprising to obtain such results since freely
integrity of microbial matrix, and the metabolic block- moving polymeric chains may form a bridge between
664 Y. Liu et al. / Water Research 37 (2003) 661–673

cells, and this would facilitate the formation of initial irreversible adhesion will occur. Theoretically, increas-
microbial nuclei, which is the first step towards ing the hydrophobicity of cell surfaces would cause a
granulation (Fig. 3). corresponding decrease in the excess Gibbs energy of the
Recently, Kalogo et al. [27] used water extract of surface, which in turn promotes cell-to-cell interaction
Moringa oleifera seeds (WEMOS) to enhance the start- and further serves as driving force for cell self-separation
up of a UASB reactor treating domestic wastewater, and from liquid phase [28,30]. There is experimental evidence
they found that the dosage of WEMOS in the feed showing that the hydrophobicity of bacterial surface
favored the aggregation of coccoid bacteria and growth plays a crucial role in initiating anaerobic granulation
of microbial nuclei, which are precursors of anaerobic [12,31–33].
granulation. WEMOS, as a kind of natural polymers, is Cell hydrophobicity can be quantified by the mea-
known to be effective in flocculating organic matter. surement of water contact angle [30,34]. According
Adsorption of WEMOS on the surface of the dispersed to the water contact angle, the hydrophobicity of
bacteria and neutralization of their surface charges microorganisms may be roughly classified into three
would be a principal mechanism to promote anaerobic categories [34]: hydrophobic surface with a contact angle
granulation [27]. It appears that both synthetic and greater than 901, medium hydrophobic surface with
natural polymers can assist anaerobic bacteria to a contact angle in between 501 and 601, and hydrophilic
aggregate together and then form granules faster. surface with a contact angle below 401. Most acidogens
exhibit hydrophilic characteristics with a water
2.6. Secondary minimum adhesion model contact angle less than 451, however acetogens and
methanogens isolated from anaerobic granules show a
The secondary minimum adhesion model is based on high surface hydrophobicity indicated by a water
the well-known DLVO theory for colloidal particles, contact angle greater than 451 [35]. These in turn
which shows that reversible adhesion takes place in the provide a physico-chemical explanation for why acido-
secondary minimum of the DLVO free energy curve. gens are most often situated in outer layer of anaerobic
The Gibbs energy of the reversible adhesion is relatively granules (Fig. 5).
small, and there is always a separation distance between
the two adhering bacteria. Thus, the reversible adhesion 2.8. Surface tension model
can change to irreversible adhesion at the primary
minimum by overcoming the energy barrier or by According to the thermodynamic theory, microbial
protruding fibrils or fimbriae, which bridge the gap granulation is a process to create a new interface,
between bacteria [28]. The secondary minimum adhesion granule–liquid by disrupting preexisting individual
model accounts for both the surface charge and the bacteria–liquid interface, and a molecular contact
surface energy or hydrophobicity, which are relevant to between the two adhering bacterial surfaces is involved.
long- and short-range forces. It appears from this model The free energy of adhesion ðDGadh Þ can be expressed as
that anaerobic granulation would start from the self- follows [28]:
immobilization of bacteria through reversible and
followed by irreversible microbial interaction. It should 1=2 1=2
DGadh ¼ 2ðrc1=2  rl Þðrl  r1=2
s Þ
be realized that the secondary minimum adhesion model
only looks into the thermodynamic aspects of bacterial in which rc is the surface free energy of bacteria, rl is
interaction, thus real meaning of this model is limited for the surface free energy of liquid and rs is the surface
a biologically defined engineering rector. free energy of inert particle. This equation shows that
if the surface free energy of bacteria is lower that that
2.7. Local dehydration and hydrophobic interaction of the liquid, the free energy of aggregation decreases
model and aggregation is favored with decreasing surface
energy of the inert carrier. The opposite trend would
Under normal culture pH conditions, the outer occur if the surface energy of bacteria is higher than
surfaces of bacteria are hydrated. Such a water layer that of the liquid. In fact, the above thermodynamic
on the surfaces of bacteria would prevent one bacterium equation is a theoretical basis of the surface tension
to approach another. It has been believed that under model. In an UASB reactor, it was found that the
normal physiological conditions strong hydration repul- aggregation of hydrophilic cells was enhanced at low
sion is the main force keeping the cells apart, thus local liquid surface tension, while the opposite was true for
dehydration of the short-distance-apart surfaces would hydrophobic cells [36]. Depending on the liquid surface
be the prerequisite for cell-to-cell aggregation. The local tension ðgÞ in the UASB reactor, bacteria may grow in
dehydration and hydrophobic interaction model as rather loose associations, in multi-layered granules (go
proposed by Wilschut and Hoekstra [29] shows that 50 mN/m), or in mixed conglomerates (g > 56 mN/m)
when bacterial surfaces are strongly hydrophobic, [36,37].
Y. Liu et al. / Water Research 37 (2003) 661–673 665

2.9. Consideration on the physico-chemical models 3. Structural models

It appears from the above discussion that each Anaerobic granulation is a complex process, in which
physico-chemical model only accounts for the contribu- biological and microbiological factors are involved other
tion of one or two factors to the initial granulation than physico-chemical forces. In the past two decades
process in UASB reactor. Since these factors exert their tremendous research progress had been made in under-
influences under specific environmental conditions and standing microbiological characteristics of UASB gran-
in specific steps during the whole granulation process, ules and interaction among different species in the
the physico-chemical models just provide some simpli- granules. In such a circumstance a series of structural
fied description on anaerobic granulation. models for anaerobic granulation has been developed to
The inert nuclei model can be easily understood, and interpret the observed phenomena.
shows that the formation of UASB granules is favored
by the presence of inert particles in the reactor. 3.1. Capetown’s model
However, there is evidence that anaerobic granules
could be developed even without the addition of any Similar to the polymer-bonding model as discussed
inert materials [18]. It should be realized that not only earlier, the Capetown model suggests that ECPs are
attachment on a solid surface, but also self-immobiliza- produced by Methanobacterium strain AZ, a hydrogen-
tion of bacteria can lead to the formation of microbial utilizing methanogen [39,40]. Under the conditions of
aggregates. As to the multi-valence positive ion-bonding high hydrogen partial pressure and limited cysteine, the
model, some studies showed that calcium ion did not amino acids (except cysteine) would be over-secreted.
make contribution to sludge granulation [17] and that a The overproduced amino acids could induce the ECPs
high concentration of magnesium ion would cause formation, and then Methanobacterium strain AZ and
disaggregation of granules [13], on the other hand, this other bacteria would be enmeshed in the ECPs matrix,
model does not account for any biological effect on thus leading to the initiation of anaerobic granulation
anaerobic granulation. Research on membrane fusion (Fig. 3). In the Capetown’s model, the overproduction
indeed indicated that Ca2+ might cause conformational of ECPs is considered a key step for initiating anaerobic
changes of some surface proteins or polypeptide groups granulation.
that could interact with two surfaces and bridge them
together [38]. It had been proposed that the beneficial 3.2. Spaghetti model
effect of calcium addition on anaerobic granulation was
probably due to the calcium-induced dehydration and Based on the microstructure of UASB granules
fusion of bacterial surfaces [15]. The calcium-induced observed by scanning electron microscope, Wiegant
cell fusion might initiate the formation of cell cluster, [41] proposed a spaghetti model for anaerobic granula-
which acts as microbial nuclei of anaerobic granulation. tion. This model shows that in the initial development of
In the secondary minimum adhesion model, the the UASB granules, filamentous Methanosaeta firstly
DLVO theory cannot make predictions at short distance attach on precursors, and further form a three dimen-
due to breakdown of the computation of electrical sional network through a branched-growth process
interactions, and it also neglects the forces which are (Fig. 3). Other bacteria, such as Methanosarcina, could
important at short distances, such as hydrogen bonding be easily entrapped in this network [42,43]. The
and other effects involved in solution and hydrophobic structured aggregates would further grow due to cellular
bonding [28], while in the local dehydration and surface multiplication of the entrapped bacteria, and becomes
tension models, bacterial granulation is oversimplified to denser and spherically shaped by the hydrodynamic
a purely thermodynamic process, and such a simple shear force caused by the upflow liquid and biogas. It
description is usually not adequate because microbial must be emphasized that in the spaghetti model, the
aggregation is a very complex biological phenomenon formation of the structured aggregate is a decisive step
and many unidentified components are involved. It in the overall granulation process.
would be impossible to build up a pure thermodynamic
model with confidence. Bacteria indeed have no rigidly 3.3. Syntrophic microcolony model
definite surface boundary, simple geometry, or uniform
molecular surface composition, i.e. bacteria cannot be Biodegradation reaction of organic wastes by anae-
treated as physically defined dead colloidal particles. robic bacteria is a complex process, in which many
The physico-chemical forces alone could not thoroughly different species are involved (Fig. 4), and those species
explain the entire microbial granulation process. Thus, it have to live in a close synergistic relationship, where
is suggested that the physico-chemical phenomena products such as hydrogen and other intermediates can
involved in microbial granulation must be related to be efficiently transferred among the respective bacterial
the biological triggers that control them. groups. The syntrophic microcolony model suggests that
666 Y. Liu et al. / Water Research 37 (2003) 661–673

the above requisites would eventually lead to the instead of random aggregation of suspended microbes’’.
formation of stable microcolonies or consortia, i.e. In order to maintain high metabolic efficiency, the
initial granules [44]. Anaerobic granule indeed can granule-associated cells would present in an organized
be regarded as the gathering together of cells to form structure, and signaling mechanisms to organize the
fairly stable, contiguous, multicellular association under syntrophic species can be predicted [46]. Therefore, it
physiological conditions in a defined biological system. appears from the syntrophic microcolony model that the
The close packing of bacteria in granule architecture driving force for sludge granulation should be a result of
definitely facilitates the exchange of metabolites. the needs of bacterial survival/balance and optimal
In the UASB granules, different groups of bacteria combination of different biochemical functions of
carry out sequential metabolic processes (Fig. 4), and multiple species under the culture conditions.
interspecific syntrophic reactions would be energetically
beneficial. Because of the need for such close proximity,
3.4. Multi-layer model
random cell-cell association in the UASB granules
would lower metabolic efficiency. As pointed out by
Based on the microscopic observations, a multi-layer
Fang [45], ‘‘biogranules are developed through evolution
model for anaerobic granulation was initially proposed
by MacLeod et al. and Guiot [47,48]. According to this
model, the microbiological composition of granules is
Complex organic wastes different in each layer. The inner layer mainly consists of
methanogens that may act as nucleation centers
necessary for the initiation of granule development.
Hydrolytic bacteria H2-producing and H2-utilizing bacteria are dominant
species in the middle layer, and a mixed species including
Monomers rods, cocci and filamentous bacteria takes predominant
(e.g. glucose, amino acids, fatty acids) position in the outermost layer (Fig. 5). To convert a
target organic to methane, the spatial organizations of
Fermentative acidogenic bacteria methanogens and other species in UASB granules are
essential.
The layered structure of UASB granules is supported
Organic acids, alcohols, ketones by the works of Arching et al. and Lens et al. [49,50]
with immunological and histologic methods, [11] with a
dynamic model, [51] with microelectrodes, [52–54] with
Acetogenic bacteria
fluorescence in situ hybridization using 16S rRNA-
targeted oligonucleotides. A distinct layered structure
Acetate, CO 2 and H2 was also found in the methanogenic–sulfidogenic
aggregates, with sulfate-reducing bacteria in the outer
Methanogens 50–100 mm, methanogens in the inner part [51]. Unlike
the initial multi-layer model proposed by MacLeod et al.
[47], recent research showed that UASB granules had
Methane gas large dark non-staining centers, in which neither
archaeal nor bacterial signals could be found [55]. In
Fig. 4. Microbial groups involved in anaerobic biodegradation fact, the non-staining center in the UASB granules
process (from Bitton [77]). might be formed as a result of the accumulation of

Fig. 5. Schematic representation of the multi-layer model (from Guiot and Pauss [48]).
Y. Liu et al. / Water Research 37 (2003) 661–673 667

metabolically inactive, decaying biomass and inorganic granules strongly depends on the degradation thermo-
materials [53]. dynamics and kinetics of individual substrate. There-
fore, it appears that different dominating catabolic
3.5. Consideration on the structural models pathways may give rise to different structural granules.
So far, any individual structural model cannot explain
According to the Capetown’s model, anaerobic a spontaneous and sudden washout of the established
granulation would not occur in UASB reactors treated granular sludge bed as a result of a change in wastewater
with acetate, propionate, or butyrate because of composition, which is a common problem encountered
inadequate hydrogen partial pressure. However, there in the operation of UASB systems. If a factor that is
is experimental evidence that anaerobic granules formed independent of the wastewater composition can initiate
in UASB systems fed with those above-mentioned the formation of UASB granules, a change in the
organics [49,56]. On the other hand, for the granule- wastewater composition should not lead to the washout
associated bacteria, high hydrogen partial pressure is of the entire granular sludge bed. Thus, it is a reasonable
not desirable because the partial pressure of hydrogen speculation that there should be a substrate composi-
must be kept as low to ensure efficient fermentation of tion-associated factor that highly contributes to the
the volatile fatty acids (Fig. 5). These may imply that the formation of UASB granules, but is not yet included in
Capetown’s model would be available only under some the present structural models.
specific conditions. The importance of ECP in anaerobic
granulation has been evidenced [3,22]. It seems that ECP
may play an important role in building spatial structure 4. Proton translocation-dehydration theory
and maintaining the stability of UASB granules, but
its contribution to the initiation of anaerobic granula- 4.1. Theory development
tion is still debatable. In addition, a high amount of ECP
seems unnecessary for making up active granules, while Based on the consideration of proton translocating
too much ECP would cause deterioration in floc activity on bacterial membrane surfaces, a new proton
formation [3]. translocation-dehydration theory for anaerobic granula-
In the syntrophic microcolony model, a close syner- tion was proposed [15,32]. The overall granulation
gistic relationship among different microbial groups is process starts from dehydration of bacterial surfaces,
essential for efficiently breaking down the complex and followed by embryonic granule formation, granule
organic wastes (Fig. 4). In fact, the syntrophic micro- maturation and post-maturation.
colonies provide the kinetic and thermodynamic re-
quirements for intermediate transference and therefore 4.1.1. Dehydration of bacterial surfaces
efficient substrate conversion [57]. It seems certain that During the hydrolysis/acidification and degradation
the synergistic requirements would drive bacteria to of complex organic compounds, the electron transport
form granules, in which different species function in a on the respiration chain of bacterial surfaces is
synergistic way and can easily survive. activated. This couples with the activation of proton
Contrary to the multi-layer model, anaerobic granules pumps on the membranes of these bacteria. Proton
with non-layered structure were also reported [4,5,58]. translocation can establish a proton gradient across the
There is evidence that a layered and non-layered bacterial cell surface and subsequently cause the surface
microstructure of the UASB granules would be devel- protonation. The energized bacterial surfaces may result
oped with carbohydrates and the substrates having a in the breaking of hydrogen bonds between negatively
rate-limiting hydrolytic or fermentative step (e.g. pro- charged groups and water molecules as well as partial
teins), respectively [4,45]. This is probably due to neutralization of the negative charges on bacterial
different initial steps in the carbohydrate and protein surfaces. This might induce the dehydration of the
degradation. The carbohydrate degradation to small bacterial surfaces.
molecules is faster than its subsequent degradation of
the intermediates, whereas the initial step in the protein 4.1.2. Embryonic granule formation
degradation is a rate-limiting step. Results from The proton pumps on the cell surfaces of acetogens
fluorescence in situ hybridization combined with con- and methanogens are activated, and their surface
focal scanning laser microscopy clearly showed that dehydration occurs due to the presence of high-energy
protein-fed granules possesses non-layered structure protons. By the action of external hydrodynamic shear
with a random distribution of Methanosaeta concilii forces, these relatively neutral and hydrophobic acido-
[55]. However, different types of granules may also form gens, acetogens, and methanogens may adhere to each
on the same substrate [3,35]. Based on microscopic other due to the weakened hydration repulsion. Their
examination of the UASB granules, recently Fang [45] adhesion is strengthened by further dehydration of the
proposed that the microbial distribution of the UASB bacterial surfaces because the formation of initial
668 Y. Liu et al. / Water Research 37 (2003) 661–673

bacterial aggregate facilitates the effective metabolite generated by proton extrusion catalyzed by membrane
transference among the bacteria and accelerates the ATPases (H+/ATPases) at the expense of ATP, and the
metabolism rate. These initial bacterial aggregates metabolic end-product efflux is an additional mechan-
are embryonic granules. ism for proton extrusion from Streptococci and other
bacterial cells that result in the generation of proton
4.1.3. Granule maturation gradients. Protons are disposed as acid to regulate their
With continuous growth and multiplication of the cytoplasmic pH conditions. These in turn can cause
bacteria in an embryonic granule, some dispersed protonation and dehydration on the bacterial surfaces.
bacteria in the medium may also adhere to the The fundamentals of energy metabolism show that the
embryonic granule and be integrated into the bacterial proton translocation across cellular membrane exists in
consortia. In embryonic granules with on-going metha- both aerobic and anaerobic respiration processes.
nogenic series metabolism, the distribution of each Anaerobic respiration is not as efficient in ATP synthesis
group of bacteria depends on the orientation of as aerobic respiration because the alternate electron
intermediate metabolites transference, which is the most acceptors, such as nitrate, sulfate or carbon dioxide,
efficient way for anaerobes to transfer their intermedi- have less positive reduction potentials than oxygen [59].
ates. This will result in the formation of well-organized This would imply that less energy is available to make
bacterial consortia as mature granules. ATP in anaerobic respiration, that is, the proton
translocation activity across cellular membrane in
4.1.4. Post-maturation anaerobic respiration would be much lower than that
The bacterial proton translocating activity in mature in aerobic respiration process. According to the proton
granules keeps the bacterial surfaces at a relatively translocation-induced dehydration theory as presented
hydrophobic state. The maintenance of the structure of earlier, microbial granulation could be observed in any
mature granules is governed by the mechanism of aerobic or anaerobic system, and is independent of the
proton translocation-dehydration. On the other hand, types of substrate, bioreactors and operation conditions.
an ECP layer outside of a granule causes the hydration However, microbial granulation has never been reported
of the granule surface and protects the granule against in conventional activated sludge process in the last 100
attachment to gas bubbles and shear stress in the UASB year operation practice, and nearly 100% of anaerobic
reactor. granules form only in UASB reactors. The feasibility
and efficiency of other types of anaerobic bioreactors in
4.2. Consideration on the proton translocation- development of anaerobic granules has not been
dehydration theory sufficiently demonstrated yet. The proton transloca-
tion-dehydration theory provides useful information for
In the proton translocation-dehydration theory, the understanding how anaerobic granules form in a
proton translocation-induced dehydration of bacterial molecular level, but it is most likely that this theory
surface is considered a key element of the whole model. does not account for those operation conditions-
According to the chemiosmotic mechanism, for most of associated metabolic changes/requirements of micro-
aerobic bacteria, ATP is generated by oxidative phos- organisms, which would highly contribute to the
phorylation, in which process electrons are transported formation of UASB granules.
through the electron transport system (ETS) from an
electron donor (substrate) to a final electron acceptor
(O2). The molecules directly using the H+ gradient built 5. Cellular automaton model
up by electron transport can be considered H+-ATP as
pumps. In methanogens, ATP synthesis is linked with Cellular automaton model has been used to describe
methanogenesis by electron transport, proton pumping the formation of microcolonies and biofilms [60–62].
and a chemiosmotic mechanism [59]. Similar to aerobic The cellular automaton model is defined as spatially and
respiration, anaerobic respiration is useful because it is temporally discrete system where the state of an
more efficient than fermentation and allows ATP automaton is determined by a set of rules that act
synthesis by electron transport and oxidative phosphor- locally but apply globally [61]. In the model, cellular
ylation in the absence of oxygen. Thus, it appears that automata form a class of systems composed of
the proton translocation-driven phosphorylation would individual units (cells), each with a defined state, and
be a common mechanism for energy generation in both each cell can change its state following the transition
aerobic and anaerobic respirations. It must be pointed rules, which are influenced by its own state and those of
out that some bacteria, e.g. Streptococcus, have no other cells [61]. This model aims to reproduce a
respiration chain and can produce ATP only via microbial structure under substrate-transfer-limited con-
substrate-level phosphorylation. In this case, the proton ditions. Substrate gradients created by local consump-
gradients across those bacterial surfaces are often tion of substrate allow the bacteria situated on
Y. Liu et al. / Water Research 37 (2003) 661–673 669

‘‘mounds’’ to have more substrate available than those optimal way of interaction with the environment. The
situated in ‘‘valley’’ [63]. Thus, the structure of micro- cellular automaton model shows that biofilm structure is
colony or biofilm is related to the availability of determined by localized substrate concentration [61],
resource. Details on the automaton model can be found however it is revealed that a cell indeed can read its
in the excellent review paper by Wimpenny and position in a concentration gradient of an extracellular
Colasanti [61]. signal factor, and to determine its developmental fate
It had been reported that a simple and practical way accordingly [71]. Based on recent research on cell-to-cell
towards the rapid anaerobic granulation was to increase communication [72–74], it can be predicted that cell-to-
the organic loading rate based on an 80% reduction of cell signaling mechanisms would be effective in devel-
biodegradable chemical oxygen demand with supple- oping anaerobic granules and organizing the spatial
mentary monitoring of effluent suspended solid washout structure of granule-associated bacteria in response to
[64–66]. These are consistent with the prediction of the environmental stresses. In fact, larger-scale organization
cellular automaton model. Cellular automaton model had been observed in the distribution of distinct species
can simulate a dynamic development of a microcolony and of distinct metabolic processes within the UASB
or biofilm under varying environmental conditions. The granules [46].
model actually can produce a large variety of distinct As Fig. 4 shows, a number of different groups of
morphologies in response to changes in growth condi- bacteria carry out sequential metabolic processes in the
tions [60,61]. However, as pointed out by Tolker-Nielsen UASB granules. In order to efficiently utilize a target
and Molin [63], the cellular automaton model does not waste, those bacteria have to be spatially organized
account for cell mobility toward resource and the role of (Fig. 5). As summarized by Shapiro [46], the benefits of
cell-to-cell communication in the development of spatial an organized microbial structure include more efficient
organization of microcolony or biofilm. proliferation; access to resource and niches that cannot
Recently, based on the cellular automaton theory, a be utilized by isolated cells; collective defense against
series of multi-dimensional biofilm models with hetero- antagonists than eliminate isolated cells, and optimiza-
geneous biomass and substrate distribution in two or tion of population survival by differentiation into
three dimensions have been developed [62,67–70]. In the distinct cell types. These are strongly supported by
multi-dimensional biofilm models, it is generally as- experimental evidence that the UASB granules are much
sumed that biofilm growth is due to the processes of more resistant than suspended sludge to the toxicity of
diffusion, reaction and growth including biomass hydrogen sulfide, heavy metals and aromatic pollutants
growth, division and spreading. Many studies suggested in wastewater [32,33,45,75].
that the structure of granules would be similar to the In an UASB reactor, it has been generally observed
structure of biofilms [3,47,63], thus the multi-dimen- that a change in wastewater composition would result in
sional models proposed to explain the spatial organiza- a washout of the established granular sludge bed within
tion of bacteria in biofilms could be useful in a short time period. This phenomenon can be reason-
interpretation and simulation of anaerobic granulation. ably explained by the cell-to-cell communication me-
Thus, future study needs to look into the applicability of chanism. As pointed out earlier, the bacteria in an
the multi-dimensional biofilm models in anaerobic UASB granule are not randomly distributed but rather
granulation process. It should be pointed out that as organized to best meet the needs of each for a defined
models are being more and more complex, model organic substrate, i.e., the spatial organization of the
calibration becomes a difficult task. In fact, without an UASB granules is developed to cope with the constraints
adequate calibration, quantitative results generated imposed by the substrate fed and corresponding meta-
from the model might be meaningless. bolic processes. When the composition of wastewater is
changed, the UASB granule-associated bacteria have to
re-organize microbial spatial distribution and structure
6. Cell-to-cell communication model in order to adapt to new metabolic processes required
for the oxidation of present organic substrate. In fact,
Although many mechanisms and models for anaero- structure changes induced by a substrate shift have been
bic granulation are currently available in the literature, reported in biofilm culture processes [63,76]. Such a
none of them could provide a complete description for substrate change-induced structural re-organization
anaerobic granulation process. So far, intercellular would result in a partial or complete breakup of the
communication and multicellular coordination have UASB granules grown on the former substrate. The
been known as an effective way for bacteria to achieve observed washout of sludge blanket from the UASB
an organized spatial structure. It has been shown that would result from the substrate change-caused granule
quorum sensing is a prominent example of social breakup. It appears from the cell-to-cell communication
behavior in bacteria, as signal exchange among indivi- model that organized bacterial community, such as
dual cells allows the entire population to choose an biofilms and granules, are not simply scaled-up version
670 Y. Liu et al. / Water Research 37 (2003) 661–673

of individual bacteria. Further research is required to Biochemical forces:


investigate the cell-to-cell communication-based me-
chanism for anaerobic granulation.
* Cellular surface dehydration.
* Cellular membrane fusion.
* Signaling and collective action in bacterial commu-
7. A general model for anaerobic granulation nity.

For bacteria in an anaerobic culture to form granules As described by the proton translocation-dehydration
a number of conditions have to be fulfilled, and the theory [15,32], cellular surface dehydration and mem-
contributions of physical, chemical and biological forces brane fusion may lead to the initiation of anaerobic
to granulation process could not be considered sepa- granulation, while cooperative self-organization of
rately. So far, any individual model as presented earlier bacteria will assist to form an organized spatial structure
seems not to be able to reasonably depict the entire [46] [74].
anaerobic granulation process. Based on the existing
mechanisms for the formation of anaerobic granules, it Step 3: Microbial forces to make cell aggregation
is encouraged to propose a general four-step model for mature:
anaerobic granulation as follows. * Production of extracellular polymer by bacteria, such
Step 1: Physical movement to initiate bacterium-to- as exopolysaccharides etc.
bacterium contact or bacterial attachment onto nuclei.
* Growth of cellular cluster.
The forces involved in this step are:
* Metabolic change and genetic competence induced
by environment, which facilitate the cell-cell interac-
* Hydrodynamic force. tion, and results in a highly organized microbial
* Diffusion force. structure.
* Gravity force.
* Thermodynamic forces, e.g. Brownian movement. Step 4: Steady state three-dimensional structure of
* Cell mobility. Cells can move by means of flagella, microbial aggregate shaped by hydrodynamic shear
cilia and pseudopods, while cell movement may also forces. The microbial aggregates would be finally shaped
be directed by a signaling mechanism. by hydrodynamic shear force to form a certain
structured community. The outer shape and size of
Step 2: Initial attractive forces to keep stable multi- microbial aggregates are determined by the interactive
cellular contacts. Those attractive forces are: strength/pattern between aggregates and of hydrody-
namic shear force, microbial species and substrate
Physical forces: loading rate.
* Van der Waals forces. This four-step model for anaerobic granulation
* Opposite charge attraction. attempts to cover the current understanding of granula-
* Thermodynamic forces including free energy of tion as much as possible. It must be realized that
surface; surface tension. identification of gross engineering events in relation
* Hydrophobicity. to anaerobic granulation would be relatively easy, but
* Filamentous bacteria that can serve bridge to link or the events at molecular or genetic level still require a
grasp individual cells together. more profound understanding of the mechanisms
responsible for anaerobic granulation. As Tolker-Niel-
It should be emphasized that the hydrophobicity of sen and Molin [63] noted, ‘‘it probably does not make
bacterial surface plays a crucial role in the initiation of sense to make firm decisions about one or the other
biofilms and anaerobic granules [12,15,30,32]. Accord- explanation as the rule for community development.’’
ing to the thermodynamics theory, increasing the
hydrophobicity of cellular surfaces would cause a
corresponding decrease in the excess Gibbs energy of References
the surface, which in turn promotes cell-to-cell interac-
tion and further serves as a driving force for bacteria to [1] Lettinga G, van Velsen AFM, Hobma SW, de Zeeuw W,
self-aggregate out of liquid phase (hydrophilic phase). Klapwijk A. Use of the upflow sludge blanket (USB)
reactor concept for biological waste water treatment
Chemical forces: especially for anaerobic treatment. Biotechnol Bioeng
1980;22:699–734.
* Hydrogen liaison. [2] Hulshoff Pol LW, Heijnekamp K, Lettinga G. The
* Formation of ionic pairs. selection pressure as a driving force behind the granulation
* Formation of ionic triplet. of anaerobic sludge. In: Lettinga G, Zehnder AJB,
* Interparticulate bridge and so on. Grotenhuis JTC, Hulshoff Pol LW, editors. Granular
Y. Liu et al. / Water Research 37 (2003) 661–673 671

anaerobic sludge: microbiology and technology. The activated sludge system. Appl Environ Microbiol 1994;
Netherlands: Wageningen, 1988. p. 153–61. 60:3041–8.
[3] Schmidt JE, Ahring BK. Granular sludge formation in [20] Forster CF, Lewin DC. Polymer interaction at activated
upflow anaerobic sludge blanket (UASB) reactors. Bio- sludge surfaces. Effl Water Treat J 1972;12:520–5.
technol Bioeng 1996;49:229–46. [21] Rudd T, Sterritt RM, Lester JN. Complexation of heavy
[4] Fang HHP, Chui HK, Li YY. Effect of degradation metals by extracellular polymers in the activated sludge
kinetics on the microstructure of anaerobic biogranules. process. J Water Pollut Control Fed 1984;56:1260–8.
Water Sci Technol 1995;32:165–72. [22] Schmidt JE, Ahring BK. Extracellular polymers in
[5] Wu JH, Liu WT, Tseng IC, Cheng SS. Characterization of granular sludge from different upflow anaerobic sludge
microbial consortia in a terephthalate-degrading anaerobic blanket (UASB) reactors. Appl Microbiol Biotechnol
granular sludge system. Microbiol 2001;147:373–82. 1994;42:457–62.
[6] Hulshoff Pol LW. The phenomenon of granulation of [23] Cammarota MC, Sant’Anna Jr. GL. Metabolic blocking
anaerobic sludge. Ph.D. Thesis, Agricultural University of of exopolysaccharides synthesis: effects on microbial
Wageningen, The Netherlands: Wageningen, 1989. adhesion and biofilm accumulation. Biotechnol Lett 1998;
[7] Imai T. Advanced start up of UASB reactors by adding 20:1–4.
of water absorbing polymer. Water Sci Technol 1997;36: [24] Shen CF, Kosaric N, Blaszczyk R. The effect of selected
399–406. heavy metals (Ni, Co and Fe) on anerobic granules and
[8] Alphenaar PA, Visser A, Lettinga G. The effect of their extracellular polymeric substance (EPS). Water Res
liquid upflow velocity and hydraulic retention time on 1993;27:25–33.
granulation in UASB reactors treating wastewater [25] Chen J, Lun SY. Study on mechanism of anaerobic sludge
with a high-sulphate content. Bioresour Technol 1993;43: granulation in UASB reactors. Water Sci Technol 1993;
249–58. 28:171–8.
[9] O’Flaherty V, Lens PN, de Beer D, Colleran E. Effect of [26] El-Mamouni R, Leduc R, Guiot SR. Influence of synthetic
feed composition and upflow velocity on aggregate and natural polymers on the anaerobic granulation
characteristics in anaerobic upflow reactors. Appl Micro- process. Water Sci Technol 1998;38:341–7.
biol Biotechnol 1997;47:102–7. [27] Kalogo Y, Seka AM, Verstraete W. Enhancing the start-
[10] Noyola A, Mereno G. Granulation production from raw up of a UASB reactor treating domestic wastewater by
waste activated sludge. Water Sci Technol 1994;30:339–46. adding a water extract of Moringa oleifera seeds. Appl
[11] Arcand Y, Guitot SR, Desrochers M, Chavarie C. Impact Microbiol Biotechnol 2001;55:651–64.
of the reactor hydrodynamics and organic loading on the [28] Rouxhet PG, Mozes N. Physical chemistry of the
size and activity of anaerobic granules. Chem Eng J interaction between attached microorganisms and their
Biochem Eng J 1994;56:23–35. support. Water Sci Technol 1990;22:1–16.
[12] Mahoney EM, Varangu LK, Cairns WL, Kosaric N, [29] Wilschut J, Hoekstra D. Membrane fusion: from liposome
Murray RGE. The effect of calcium on microbial to biological membrane. Trend Biochem Sci 1984;9:
aggregation during UASB reactor start-up. Water Sci 479–83.
Technol 1987;19:249–60. [30] van Loosdrecht MCM, Lyklema J, Norde W, Schraa G,
[13] Schmidt JE, Ahring BK. Effects of magnesium on Zehnder AJB. Electrophoretic mobility and hydro-
thermophilic acetate-degrading granules in upflow anae- phobicity as a measure to predict the initial steps of
robic sludge blanket (UASB) reactor. Enzyme Microbial bacterial adhesion. Appl Environ Microbiol 1987;53:
Technol 1993;15:304–10. 1898–901.
[14] Yu HQ, Tay JH. Fang HHP The role of calcium in sludge [31] Wu WM, Kickey RF, Zeikus JG. Characterization of
granulation during UASB reactor start-up. Water Res metabolic performance of methanogenic granules treating
2001;35:1052–60. brewery wastewater: role of sulfate-reducing bacteria.
[15] Teo KC, Xu HL, Tay JH. Molecular mechanism of Appl Environ Microbiol 1991;57:3438–49.
granulation—II: proton translocating activity. J Environ [32] Tay JH, Xu HL, Teo KC. Molecular mechanism of
Eng 2000;126:411–8. granulation. I:H+ translocation-dehydration theory. J
[16] Yu HQ, Fang HHP, Tay JH. Enhanced sludge granulation Environ Eng 2000;126:403–10.
in upflow anaerobic sludge blanket (UASB) reactors by [33] Tay JH, He YX, Yan YG. Anaerobic biogranulation using
aluminum chloride. Chemosphere 2001;44:31–6. phenol as the sole carbon source. Water Environ. Res
[17] Guiot SR, Gorur SS, Bourque D, Samson R. Metal effect 2000;72:189–94.
on microbial aggregation during upflow anaerobic sludge [34] Mozes N, Rouxhet PG. Methods for measuring hydro-
bed-filter (UBF) reactor start-up. In: Lettinga G, Zehnder phobicity of microorganisms. J Microbiol Methods 1987;
AJB, Grotenhuis JTC, Hulshoff Pol LW, editors. Granular 6:99–112.
anaerobic sludge: microbiology and technology. The [35] Daffonchio D, Thavessri J, Verstraete W. Contact angle
Netherlands: Wageningen, 1988. p. 187–94. measurement and cell hydrohpobicity of granular sludge
[18] Thiele JH, Wu WM, Jain MK, Zeikus JG. Ecoengineering from upflow anaerobic sludge bed reactors. Appl Environ
high rate biomathanation system: design of improved Microbiol 1995;61:3676–80.
syntrophic biomathanation catalysis. Biotechnol Bioeng [36] Thaveesri J, Daffonchio D, Lessens B, Vandermeren P,
1990;35:990–9. Verstraete W. Granulation and sludge bed stability in
[19] Zita A, Hermansson M. Effects of ionic strength on upflow anaerobic sludge bed reactors in relation to surface
bacterial adhesion and stability of flocs in a wastewater thermodynamics. Appl Environ Microbiol 1995;61:3681–6.
672 Y. Liu et al. / Water Research 37 (2003) 661–673

[37] Grootaerd H, Liessens B, Verstraete W. Effects of directly rRNA-targeted oligonucleotides reveals localization of
soluble and fibrous rapidly acidifying chemical oxygen methanogenes and selected uncultured bacteria in meso-
demand and reactor liquid surface tension on granulation philic and thermophilic sludge granules. Appl Environ
and sludge-bed stability in upflow anaerobic sludge Microbiol 1999;65:1280–8.
blanket reactors. Appl Microbiol Biotechnol 1997;48: [54] Tagawa T, Syutsubo K, Sekiguchil Y, Ohashi A, Harada
304–10. H. Quantification of methanogen cell density in anaerobic
[38] Papahadjopoulos D, Nir S, Duzgunes N. Molecular granular sludge consortia by fluorescence in-situ hybridi-
mechanisms of calcium-induced membrane fusion. zation. Water Sci Technol 2000;42:77–82.
J Bioenerg Biomemb 1990;22:157–75. [55] Rocheleau S, Greer CW, Lawrence JR, Cantin C,
[39] Palns SS, Loewenthal RE, Dold PL, Marais GR. Laramee L, Guiot SR. Differentiation of Methanosaeta
Hypothesis for pelletisation in upflow anaerobic sludge concilii and Methanosarcina barkeri in anaerobic meso-
blanket reactor. Water SA 1987;13:69–80. philic granular sludge by in situ hybridization and confocal
[40] Sam-Soon PA, Looewenthal RE, Dold PL, Marais DVR. scanning laser microscopy. Appl Environ Microbiol
Pelletization in upflow anaerobic sludge bed reactors. In: 1999;65:2222–9.
Hall ER, Hobson PN, editors. Anaerobic digestion. [56] van Lier JB, Sanx Martin JL, Lettinga G. Effect of
Oxford, UK: Pergamon Press, 1988. p. 55–60. temperature on the anaerobic thermophilic conversion of
[41] Wiegant WM. The Spaghetti theory on anaerobic granular volatile fatty acids by dispersed and granular sludge. Water
sludge fermentation, or the inevitability of granulation. Res 1995;30:199–207.
Proceeding of the Granular Anaerobic Sludge. Wagenin- [57] Schink B, Thauer, R. Energetics of syntrophic methane
gen, 1998. p. 146–52. formation and the influence of aggregation In: Proceedings
[42] Sanchez JM, Arijo S, Munoz MA, Morinigo MA, Borrego of the Granular Anaerobic Sludge. Wageningen, 1988.
JJ. Microbial colonization of different support materials p. 5–17.
used to enhance the methanogenic process. Appl Microbiol [58] Grotenhuis JTC, van Lier JB, Plugge CM, Stams AJM,
Biotechnol 1994;41:480–6. Zehnder AJB. Effect of ethylene glycol-bis(b-aminoethy-
[43] Wu WM, Jain MK, Zeikus JG. Formation of fatty acid- lether)-N,N-tetraacetic acid (EGTA) on stability and
degrading anaerobic granules by defined species. Appl activity of methanogenic granular sludge. Appl Microbiol
Environ Microbiol 1996;62:2037–44. Biotechnol 1991;36:109–14.
[44] Hirsh R. Microcolony formation and consortia. In: [59] Prescott LM, Harley JP, Klein DA. Microbiology. New
Marshall KC, editor. Microbial adhesion and aggregation. York: McGraw-Hill, 1999.
Berlin: Springer, 1984. p. 373–93. [60] Ben-Jacob E, Schochet O, Tenenbaum A, Cohen I, Czirok
[45] Fang HHP. Microbial distribution in UASB granules and A, Tamas V. Generic modeling of cooperative growth
its resulting effects. Water Sci Technol 2000;42:201–8. patterns in bacterial colonies. Nature 1994;368:46–9.
[46] Shapiro JA. Thing about bacterial populations as multi- [61] Wimpenny JWT, Colasanti R. A unifying hypothesis
cellular organisms. Annu Rev Microbiol 1998;52:81–104. for the structure of microbial biofilms based on cell-
[47] MacLeod FA, Guiot SR, Costerton JW. Layered structure ular automaton models. FEMS Microbiol Ecol 1997;22:
of bacterial aggregates produced in an upflow anaerobic 1–16.
sludge bed and filter reactor. Appl Environ Microbiol [62] Kreft JU, Picioreanu C, Wimpenny JWT, van Loosdrecht
1990;56:1598–607. MCM. Individual-based modeling of biofilms. Microbiol
[48] Guiot SR, Pauss A, Costerton JW. A structured model of 2001;147:2897–912.
the anaerobic granules consortium. Water Sci Technol [63] Tolker-Nielsen T, Molin S. Spatial organization of
1992;25:1–10. microbial biofilm communities. Microbial Ecol 2000;
[49] Arching BK, Schmidt JE, Winther-Nielsen M, Macario 40:75–84.
AJL, de Macario EC. Effect of medium composition and [64] De Zeeuw WJ. Granular sludge in UASB-reactors. In:
sludge removal on the production, composition and Lettinga G, Zehnder AJB, Grotenhuis JTC, Hulshoff Pol
architecture of thermophilic (551C) acetate-utilizing gran- LW, editors. Granular ananerobic sludge: microbiology
ules from an upflow anaerobic sludge blanket reactor. and technology. The Netherlands: Wageningen, 1988.
Appl Environ Microbiol 1993;59:2538–45. p. 132–45.
[50] Lens P, de Beer D, Cronenberg C, Ottengraf S, Verstraete [65] Fang HHP, Chui HK. Maximum COD loading capacity in
W. The use of microsensors to determine distributions in UASB reactors at 371C. J Environ Eng 1993;119:103–19.
UASB aggregates. Water Sci Technol 1995;31:273–80. [66] Tay JH, Yan YG. Influence of substrate concentration on
[51] Santegoeds CM, Damagaad LR, Hesselink C, Zopfi J, microbial selection and granulation during start-up of
Lens P, Muyzer G, De Beer D. Distribution of sulfate- upflow anaerobic sludge blanket reactors. Water Environ
reducing and methanogenic bacteria in anaerobic aggre- Res 1996;68:1140–50.
gates determined by microsensor and molecular analysis. [67] Hermanowicz SW. A model of two-dimensional biofilm
Appl Environ Microbiol 1999;65:4618–29. morphology. Water Sci Technol 1997;37:219–22.
[52] Sekiguchi Y, Kamagata Y, Syutsubo K, Ohashi A, Harada [68] Noguera DR, Pizarro G, Stahl DA, Rittmann BE.
H, Nakamura K. Diversity of mesophilic and thermophilic Simulations of multispecies biofilm development in three
granular sludge determined by 16S rRNA gene analysis. dimensions. Water Sci Technol 1999;39:123–30.
Microbiol 1998;22:2655–65. [69] Picioreanu C, van Loodrecht MCM, Heijnen JJ. Discrete-
[53] Sekiguchi Y, Kamagata Y, Nakamura K, Ohashi A, differential modelling of biofilm structure. Water Sci
Harada H. Fluorescence in situ hybridization using 16S Technol 1999;39:115–22.
Y. Liu et al. / Water Research 37 (2003) 661–673 673

[70] Picioreanu C, van Loodrecht MCM, Heijnen JJ. Two- [74] Ben-Jacob E, Cohen I, Levine H. Cooperative self-
dimensional model of biofilm detachment caused by organization of microorganisms. Adv Phys 2000;49:395–554.
internal stress from liquid flow. Biotechnol Bioeng [75] Bae JW, Rhee SK, Hyun SH, Kim IS, Lee ST. Layered
2001;72:205–18. structure of granules in upflow anaerobic sludge blanket
[71] Gurdon JB, Bourillot PY. Morphogen gradient interpreta- reactor gives microbial populations resistance to metal
tion. Nature 2001;413:797–803. ions. Biotechnol Lett 2000;22:1935–40.
[72] Davies DG, Parsek MR, Pearson JP, Iglewski BH, [76] Wolfaardt GM, Lawrence JR, Robarts RD, Caldwell SJ,
Costerton JW, Greenberg EP. The involvement of cell- Caldwell DE. Multicellular organization in degradative
to-cell signals in the development of a bacterial biofilm. biofilm community. Appl Environ Microbiol 1994;60:
Science 1998;280:295–8. 434–46.
[73] Pratt LA, Kolter R. Genetic analysis of bacterial biofilm [77] Bitton G. Wastewater microbiology. New York:
formation. Curr Opin Microbiol 1999;2:598–603. Wiley-Liss, 1999.

You might also like