On the reproducibility of microcosm experiments – different

On the reproducibility of microcosm experiments ^ di¡erent
community composition in parallel phototrophic bio¢lm
microcosms
Guus Roeselers1, Barbara Zippel2, Marc Staal3, Mark van Loosdrecht1 & Gerard Muyzer1
1
Department of Biotechnology, Delft University of Technology, Delft, The Netherlands; 2Department of River Ecology, UFZ Centre for Environmental
Research Leipzig-Halle, Magdeburg, Germany; and 3Department of Marine Microbiology, NIOO-KNAW, Yerseke, The Netherlands
Correspondence: Gerard Muyzer,
Department of Biotechnology, Delft University
of Technology, Julianalaan 67, NL-2628 BC
Delft, The Netherlands. Tel.: 131 15
2781193; fax: 131 15 278 2355;
e-mail: [email protected]
Received 20 January 2006; revised 21 March
2006; accepted 24 March 2006.
First published online 26 June 2006.
DOI:10.1111/j.1574-6941.2006.00172.x
Editor: Michael Wagner
Keywords
biofilm; cyanobacteria; DGGE; microbial
ecology; oxygenic phototrophs.
Abstract
Phototrophic biofilms were cultivated simultaneously using the same inoculum in
three identical flow-lane microcosms located in different laboratories. The growth
rates of the biofilms were similar in the different microcosms, but denaturing
gradient gel electrophoresis (DGGE) analysis of both 16S and 18S rRNA gene
fragments showed that the communities developed differently in terms of species
richness and community composition. One microcosm was dominated by Microcoleus and Phormidium species, the second microcosm was dominated by
Synechocystis and Phormidium species, and the third microcosm was dominated
by Microcoleus- and Planktothrix-affiliated species. No clear effect of light intensity
on the cyanobacterial community composition was observed. In addition, DGGE
profiles obtained from the cultivated biofilms showed a low resemblance with the
profiles derived from the inoculum. These findings demonstrate that validation of
reproducibility is essential for the use of microcosm systems in microbial ecology
studies.
Introduction
Microcosms are constructed, simplified ecosystems that are
used to mimic natural ecosystems under controlled conditions. They provide an experimental area for ecologists to
study natural processes. Hence, microcosm studies can be
very useful to study the effects of disturbance or to determine the role of key species. These simplified systems can
still contain a high diversity of species and therefore require
self-organizing processes to reach and maintain system
stability (Odum, 1989; Kangas & Adey, 1996).
Reproducibility is indisputably essential to validate the
use of microcosm systems in microbial ecology studies.
Hence, it is surprising that reproducibility assessments are
scarcely documented in the literature (e.g. Heydorn et al.,
2000). In the present study, cultivated phototrophic biofilms
were chosen as a model to assess reproducibility.
Structure, growth dynamics and physiology of heterotrophic biofilms have been extensively studied. But until
recently phototrophic biofilms have received little attention
for this aspect. Phototrophic biofilms occur on contact
surfaces in a range of terrestrial and aquatic environments.
They can best be described as surface-attached microbial
communities driven by light as energy source. Diatoms,
FEMS Microbiol Ecol 58 (2006) 169–178
green algae and cyanobacteria are the major primary producers that generate energy and reduce carbon dioxide,
providing organic substrate and oxygen. Their oxygenic
photosynthetic activity fuels metabolic processes and conversions in the entire biofilm community, including the
heterotrophic fraction (Paerl et al., 2000). The microorganisms produce extracellular polymeric substances (EPS) that
hold the biofilm together (Wimpenny et al., 2000; Cogan &
Keener, 2004).
There is a growing interest in the application of phototrophic biofilms, e.g. bioremediation (Schumacher et al.,
2003), aquaculture (Bender & Phillips, 2004) and biohydrogen production (Prince & Kheshgi, 2005). The
study of artificial phototrophic biofilms may also increase
our understanding of the development of more complex
phototrophic biofilms, such as microbial mats and stromatolites (Des Marais, 1990). In order to enhance our understanding of the complex phototrophic biofilm physiology,
the individual community members should not be studied
separately.
Since biofilm communities in nature are often difficult
to investigate and experimental conditions are ambiguous,
a number of different laboratory-based experimental
biofilm model systems have been developed (Palmer,
2006 Federation of European Microbiological Societies
Published by Blackwell Publishing Ltd. All rights reserved
c
170
1999; Heydorn et al., 2000; Jackson et al., 2001). In most
cases these systems were used to study single-species biofilms or mixed biofilms with a predefined community
composition.
In the current study identical open flow-lane incubator
systems (i.e. microcosms) located at three different laboratories in Europe were inoculated with the same environmental biofilm sample at the same time and operated under
exactly the same conditions. Biomass samples were collected
from the three different incubators at the moment that the
biofilms reached their mature stage. In addition, replicate
biofilm growth experiments with the same inoculum were
carried out in small flow cell incubators located in one
laboratory.
Bacterial and eukaryotic community compositions were
compared using denaturing gradient gel electrophoresis
(DGGE) (Schäfer & Muyzer, 2001) of PCR-amplified small
subunit rRNA (SS rRNA) gene fragments in conjunction
with DNA sequencing and phylogenetic analysis.
Materials and methods
Biofilm incubator
The flow-lane incubator systems (‘large incubators’) used in
this study contained four separate flow channels through
which a volume of 4 L medium circulated over a surface
covered with 47 polycarbonate slides (76 25 1 mm). The
polycarbonate slides were used as a substratum for biofilm
adhesion. Each light chamber contained an adjustable light
source and the circulation speed of the culture medium
could be regulated precisely. The medium was refreshed
twice a week.
Biofilm growth was monitored and recorded with three
light sensors that were positioned directly under selected
polycarbonate slides. Each light sensor contained three
independent photodiodes. Decrease of subsurface light
below the substratum was used as an indicator for biomass
accumulation.
The incubator design was described in more detail by
Zippel & Neu (2005). Identically designed and produced
biofilm incubators used in this study were located at the
Department of Biotechnology of Delft University of Technology in the Netherlands, at the Department of River
Ecology of the UFZ Centre for Environmental Research in
Magdeburg (Germany), and at the Netherlands Institute of
Ecology in Yerseke (the Netherlands).
Two small flow cell-type incubators with a glass top
were used for reproducibility experiments in the laboratory at Delft University of Technology. Each flow cell
contained one polycarbonate slide. The aluminium body of
the flow cells contained a water channel for temperature
control.
2006 Federation of European Microbiological Societies
Published by Blackwell Publishing Ltd. All rights reserved
c
G. Roeselers et al.
Growth conditions and inoculum
The mineral medium used was a modification of BG11 as
described by Stanier et al., (1971). Ammonium ferric citrate
green was replaced by FeCl3. The vitamins cyanocobalamin
(40 mg L 1), thiamine HCL (40 mg L 1) and biotin (40 mg L 1)
were added. NaSiO3 9H2O (57 mg mL 1) was added to allow
the growth of freshwater diatoms (Guillard & Hargraves, 1993).
Phototrophic biofilm material was collected from an
overflow weir of the sedimentation basin of the wastewater
treatment plant (WWTP) at Fiumicino Airport (Rome)
(Albertano et al., 1999). The biomass was homogenized and
aliquoted into 50 mL tubes. In order to reduce predation
pressure, the biomass was frozen at 20 1C to kill protozoa,
metazoa and nematodes. The aliquots were shipped on dry
ice to the three laboratories. Care was taken that all tubes
remained frozen for the same period. The biofilm biomass
was used to inoculate the polycarbonate slides within the
four flow lanes in each laboratory on the same day.
Biofilms were grown at 30 1C and at a medium flow rate
of 100 L h 1. The biofilms were grown under irradiances of
60 and 120 mmol photons m 2 s 1. A diel light cycle of 16 h
light : 8 h dark was applied. The growth and development of
the biofilms was monitored for 34 days.
Biofilm cultivation experiments with the flow cell incubators were carried out at a constant temperature of 30 1C,
an irradiance of 100 mmol photons m 2 s 1 and a medium
flow rate of 0.5 L h 1.
Sample collection
Biofilms grown under different light conditions in the three
large incubators were sampled when they reached the
mature stage of their development. This stage was defined
as the moment when the average light intensity measured by
the three submerged light sensors was o10% of the applied
light. One polycarbonate slide was removed from the
incubator lanes at the exit side of the flow lane at each
sampling event.
After 10 days the polycarbonate slides were removed from
two flow cell incubators which were operated in parallel to
each other. Another two slides were removed from two flow
cell incubators that were operated for 20 days.
Slides were immediately frozen and stored at 20 1C. To
eliminate potential variability introduced by analysis at
different laboratories, samples collected from incubators in
Magdeburg and Yerseke were shipped on dry ice to the
Department of Biotechnology at Delft University of Technology for further analysis.
DNA extraction
Biofilm biomass was scraped from the polycarbonate slides
with a sterile razor blade. Genomic DNA was extracted by
FEMS Microbiol Ecol 58 (2006) 169–178
171
Reproducibility of microcosm experiments
applying c. 300 mg biomass to the UltraClean Soil DNA
Isolation KitTM (Mo Bio Laboratories, Carlsbad, CA) according to the manufacturer’s protocol. Complete cell lysis
was verified afterwards using phase-contrast microscopy.
The quantity and quality of the extracted DNA was analyzed
by spectrophotometry using the NanoDrop ND-1000TM
(NanoDrop Technologies, Delaware) and by agarose gel
electrophoresis. DNA dilutions were stored at 20 1C.
PCR amplification of rRNA gene fragments
Extreme care was taken to prevent any DNA contamination
of solutions and plastic disposables used for PCR. All heatsterilized plastic tubes were exposed to UV light for 30 min
before use. Only DNA- and RNA-free water (W4502, SigmaAldrich, St Louis, MO) was used to prepare PCR reagent
stock solutions and PCR reaction mixtures.
To amplify the bacterial 16S rRNA-encoding gene fragments, the DNA dilutions were used as template DNA in
50 mL PCR reactions using the primers 359F-GC and 907R,
and PCR conditions as described by Schäfer et al. (2001).
This PCR was carried out with a denaturation step of 5 min
at 94 1C, followed by 35 cycles of denaturation of 1 min at
94 1C, annealing of 1 min at 60 1C, and extension of 1 min at
72 1C, followed by a final extension step of 10 min at 72 1C.
To amplify the 16S rRNA-encoding gene fragments of
cyanobacteria we used the universal primer 359F-GC and an
equimolar mixture of the reverse primers 781R(a) and 781R(b),
and PCR conditions as described by Nübel et al. (1997).
To amplify eukaryotic 18S rRNA encoding-gene fragments
we used the EukA-f and Euk516-r1GC primers (Diez et al.,
2001). This PCR was carried out with a denaturation step of
15 min at 94 1C, followed by 33 cycles of denaturation of
1 min at 94 1C, annealing of 1 min at 55 1C, and extension of
3 min at 72 1C, followed by a final extension step of 10 min at
72 1C. All amplification reactions were performed in a T1
Thermocycler (Biometra, Westburg, the Netherlands).
DGGE of PCR products
DGGE was performed as described by Schäfer & Muyzer
(2001). Briefly, 1 mm thick 6% acrylamide gels with a ureaformamide (UF) gradient of 20–80% were used for bacterial
16S rRNA gene fragments. Gradients of 20–60% were used for
18S rRNA gene fragments. An acrylamide gel without UF was
cast on top of the gradient gel to obtain good loading slots.
From each PCR reaction 20 mL product, containing c. 1 mg
DNA, was mixed with 6 mL of 10 gel loading solution and
loaded onto the gel. Gels were run in 1 TAE (Tris-acetateEDTA buffer, 50 stock solution: 242 g of Tris base,
57.1 mL of glacial acetic acid, 100 mL of 0.5 M EDTA pH
8.0) for 16 h at 100 V and at a constant temperature of 60 1C.
Gels were stained in an ethidium bromide solution and
FEMS Microbiol Ecol 58 (2006) 169–178
analyzed and photographed using the GelDoc UV Transilluminator (Bio-Rad, Hercules, CA).
The dominant bands were excised from the DGGE gels
with a sterile surgical scalpel. Each small gel slice was placed
in 15 mL of sterile water for 24 h at 4 1C. Subsequently, 2 mL
of the solution was used as template DNA for re-amplification as described above. The PCR products were again
subjected to DGGE analysis to confirm their purity and
position relative to the bands from which they were originally excised. The PCR products were purified using the
QIAquick PCR Purification Kit (QIAGEN, Hilden, Germany). The purified PCR products were sequenced on an
ABI 3730 sequencer (Applied Biosystems, Foster City, CA)
by a commercial company (BaseClear, Leiden, the Netherlands). The sequencing reactions were carried out with the
appropriate specific forward primers without GC clamp.
DGGE profiles were compared visually on the basis of the
presence and relative density of bands. In addition, the
presence and absence of DGGE bands in different samples
were scored in binary matrices. The binary matrices derived
from the three DGGE gels were combined into one matrix.
This binary matrix was translated into a distance matrix
using the Jaccard coefficient. A dendrogram was then
constructed by the UPGMA (Unweighted Pair Group Method with Arithmetic mean) clustering method (Griffiths
et al., 2000). Jaccard coefficient and UPGMA calculations
were carried out with the software package Primer 6
(PRIMER-E Ltd, Plymouth, U.K.).
Comparative sequence analysis
Partial 16S and 18S rRNA gene sequences with lengths of
between 400 and 500 bp were first compared to the sequences stored in the Genbank nucleotide database using
the BLAST algorithm (Altschul et al., 1990) in order to obtain
a tentative identification of the biofilm community members. Subsequently, the sequences (including closest BLAST
hits) were imported into the ARB SS rRNA database
(available at http://www.arb-home.de) (Ludwig et al., 2004)
and aligned based on the secondary structure of the SS
rRNA. The dissimilarity values were used to calculate
distance matrices. Distance matrix trees were generated by
the Neighbour-Joining (NJ) method with the Felsenstein
correction as implemented in the PAUP 4.0B software
(Sinauer, Sunderland, MA). The NJ calculation was subjected to bootstrap analysis (1000 replicates).
Nucleotide sequence accession numbers
The sequences were deposited in the GenBank Nucleotide
database and assigned accession numbers DQ366036–
DQ366083.
2006 Federation of European Microbiological Societies
Published by Blackwell Publishing Ltd. All rights reserved
c
172
Results
Biofilm growth rate
The biofilm growth rate was monitored by the decrease of
subsurface light. The lag-time after inoculation was 4 days,
and visible growth of the phototrophic biofilms started
between 4 and 8 days after inoculation. It was found
that the growth rates were in each incubator higher at
120 photons mmol m 2 s 1. Phototrophic biofilms growing
at 120 mmol photons m 2 s 1 reached their mature stage
c. 5 days earlier than the biofilms incubated at lower
light intensities. The biofilm growing at 120 mmol
photons m 2 s 1 in Magdeburg grew fastest and reached its
mature stage within 12 days of inoculation (Fig. 1).
G. Roeselers et al.
third band (Fig. 3a, bands 2 and 3) were affiliated to
Phormidium-like cyanobacteria.
The DGGE profiles from the Delft biofilms contained five
dominant bands more than the inoculum profile, suggesting
a higher biodiversity (Fig. 3a). Both profiles from the Delft
incubator were very similar although the 120 mmol m 2 s 1
profile contained two bands (Fig. 3a, bands 5 and 8) that
were absent in the 60 mmol m 2 s 1 profile. Band 5 was
DGGE and phylogenetic analysis of Bacteria
The bacterial 16S rRNA gene-DGGE profiles from the
cultivation experiments in the two flow-cell incubators were
nearly identical. The profiles from the 10-day-old biofilms
from both flow cells consisted of one dominant band and
two faint bands. Both profiles from the 20-day-old biofilm
showed one dominant band and several very faint bands
(Fig. 2).
The DGGE profiles obtained from the inoculum and the
cultivation experiments with the large incubator setups
located in three different laboratories (Delft, Magdeburg
and Yerseke) were also compared. The bacterial 16S rRNA
gene sequence DGGE profiles (Fig. 3a) revealed little similarity between the inoculum and the cultivated biofilms. The
inoculum profile contained only three dominant bands. The
top band in this profile (Fig. 3a, band 1) was affiliated to
chloroplasts of a Scenedesmus-like alga. The second and the
Fig. 2. DGGE patterns of 16S rRNA gene fragments obtained after
enzymatic amplification using general bacterial primers and genomic
DNA samples from 10- and 20-day-old phototrophic biofilms cultivated
in two flow cell incubators (A and B).
Fig. 1. Growth curves of phototrophic biofilms growing in three incubators at two incident light intensities (60 and 120 mmol photons m
development is indicated as the increasing light absorbance derived from the decrease of subsurface light.
2006 Federation of European Microbiological Societies
Published by Blackwell Publishing Ltd. All rights reserved
c
2
s 1). Biofilm
FEMS Microbiol Ecol 58 (2006) 169–178
173
Reproducibility of microcosm experiments
Fig. 3. DGGE patterns of SS rRNA gene fragments obtained after enzymatic amplification using general bacterial primers (a), primers specific for
cyanobacteria (b), and general eukaryotic 18S rRNA primers (c), and genomic DNA samples from phototrophic biofilms growing in three incubators
(located in Delft, Yerseke, and Magdeburg) at two incident light intensities (60 and 120 mmol photons m 2 s 1). Numbers at the left of each lane
correspond to bands that were excised, PCR amplified, and sequenced.
affiliated to Cytophaga-like bacteria and band 8 was affiliated
to a deep-branching unidentified bacterium (Fig. 5).
The bacterial 16S rRNA gene-DGGE profiles from the
Yerseke biofilms were highly similar for both light intensities. The high light profile contained one band that was
absent in the low light profile. Two other dominant bands,
which were present in both profiles, were affiliated to
Cyanobacterium stanieri species (Fig. 3a; band 11) and
Erythrobacter longus species, aerobic bacteria that contain
bacteriochlorophyll a (Fig. 3a; band 12, and Fig. 5).
The top band in both profiles from the Magdeburg
biofilms had an identical position to the top bands in both
Delft profiles. These bands (Fig. 3a; band 4, band 9 and band
13) showed affiliation to the chloroplasts of Scenedesmus
(Fig. 5). The high light profile from Magdeburg contained
one dominant band that was absent in the low light profile
(Fig. 3a; band 14) and one dominant band that was only
faintly visible in the low light profile (Fig. 3a; band 15).
Furthermore, one dominant band that was present in the
low light profile was absent in the high light profile (Fig. 3a;
band 18).
The UPGMA dendogram that was constructed from the
DGGE profiles obtained from the three large incubators
shows the highest similarity between samples derived from
the same lab. It is remarkable that the DGGE profiles and the
UPGMA dendogram (Fig. 4) show that the community
composition and biodiversity of biofilms obtained from
one incubator but cultured at different light intensities were
more similar than the biofilms grown at the same light
conditions but in different laboratories.
FEMS Microbiol Ecol 58 (2006) 169–178
Fig. 4. UPGMA dendrogram showing the combined clustering analyses
of the digitized DGGE profiles (Figs 3a–c) using the unweighted pairwise
grouping method with mathematical averages (Jaccard coefficient of
similarity). The analysis is based on the presence or absence of bands at
certain positions in each lane of each gel.
DGGE and phylogenetic analysis of
cyanobacteria and chloroplasts
The cyanobacterial 16S rRNA gene DGGE profile (Fig. 3b)
from the Delft biofilms showed high similarities for both
growth irradiances. Several bands are present in both
profiles. Band number 5 (Fig. 3b), visible in the low light
profile, is also faintly visible in the high light profile. Bands 1
and 6 are present under both light conditions, and are both
affiliated to Synechocystis sp. (Fig. 6). A band with the same
position is present in the inoculum, but its sequence was not
2006 Federation of European Microbiological Societies
Published by Blackwell Publishing Ltd. All rights reserved
c
174
G. Roeselers et al.
Fig. 5. Evolutionary tree showing the phylogenetic affiliations as revealed by comparative analysis of the bacterial 16S rRNA gene sequences derived
from the DGGE bands shown in Fig. 3a. The sequences obtained in this study are printed bold. Escherichia coli (AJ567606) was used as an out-group,
but was pruned from the tree. Accession numbers of sequences are noted behind the taxon names. Scale bar indicates 10% estimated sequence
divergence. Numbers on the branches are bootstrap values; only values higher than 50% are given.
obtained. Band couple 2 and 7 as well as couple 3 and 8 are
also present under both light conditions and show the same
phylogenetic affiliation (Fig. 6). Band number 4 is only
present at the high light profile.
The high and low light profiles from Yerseke are almost
identical. A dominant band at the top of both profiles (Fig.
3b, band 9) is affiliated to the same Scenedesmus chloroplast
as band 5 from the Delft low light profile. Another band
present at both light conditions in Yerseke shares its position
with bands 3 and 8 in the Delft profiles. These bands
correspond with sequences affiliated to Phormidium tenue
(Figs 3b and 7). Identically positioned bands 11 and 12 are
both affiliated to Microcoleus vaginatus.
The Magdeburg biofilm profiles show a very thick dominant band at the high light intensity that is very faintly
2006 Federation of European Microbiological Societies
Published by Blackwell Publishing Ltd. All rights reserved
c
visible in the low light intensity profile (Fig. 3b; band 13),
which is affiliated to Planktothrix sp. The thick dominant
band that is present in the low light profile (Fig. 3b; band
14), which was only faintly visible in the high light profile, is
affiliated to a Microcoleus vaginatus strain (Fig. 6).
DGGE and phylogenetic analysis of eukaryotes
The 18S rRNA gene DGGE profiles (Fig. 3c) show a band
that was dominantly present in the biofilms cultivated at
both light intensities in the three different labs. This band is
faintly visible in the inoculum. The sequences derived from
these bands (Fig. 3c; bands 3, 5, 8, 11, 14 and 15) are
identical and are closely affiliated to the unicellular alga
Chlorella fusca (Fig. 7). The profile derived from the
FEMS Microbiol Ecol 58 (2006) 169–178
175
Reproducibility of microcosm experiments
Fig. 6. Evolutionary tree showing the phylogenetic affiliations as revealed by comparative analysis of the cyanobacterial 16S rRNA gene sequences
derived from the DGGE bands shown in Fig. 3b. The sequences obtained in this study are printed bold. Escherichia coli (AJ567606) was used as an outgroup sequence, but was pruned from the tree. Accession numbers of sequences are noted behind the taxon names. Scale bar indicates 10% estimated
sequence divergence. Numbers on the branches are bootstrap values; only values higher than 50% are given.
inoculum contains only two dominant bands. Band 1 is
affiliated to Scenedesmus communis and band 2 is affiliated
to Coelastrella multistriata (Fig. 3c; bands 1 and 2).
The profiles from both light intensities from Delft are
similar. At both intensities, the biofilms from Delft and
Yerseke show bands that are affiliated to the rotifer Brachionus plicatilis (Fig. 3c; bands 4, 6, and 9, and Fig. 7). A
Spaeromonas-like fungus appears to be present only in the
high light biofilms (120 mmol photons m 2 s 1) from Yerseke
and Magdeburg (Fig. 3c; band 7 and band 13, and Fig. 7).
Discussion
In summary, the cultivated biofilms were inhabited by a
phylogenetically diverse array of prokaryotes including unicellular and filamentous cyanobacteria, as well as bacteria
belonging to the Bacteroidetes (formerly known as the
Cytophaga–Bacteroides–Flavobacteria group), the Alphaproteobacteria and the Betaproteobacteria. The biofilms also
included eukaryotes such as green algae, fungi and protozoa.
This shows that phototrophic biofilms, although depending
on the primary production of oxygenic phototrophs, can
develop as a small ecosystem with many different functional
groups of organisms, perhaps reflecting the presence of a
variety of ecological niches due to spatial heterogeneity.
FEMS Microbiol Ecol 58 (2006) 169–178
The community composition of the large microcosm
biofilms was in all cases very different from the initial
inoculum. This suggests that the conditions within the
incubator do not resemble the environmental conditions at
the overflow weir of the sedimentation basin from which the
inoculum was collected. In addition, the treatment of the
inoculum may have had a large effect on the community
development in the incubators.
Although we see that the growth characteristics and the
community composition of the phototrophic biofilms in the
large incubators are influenced by the light intensities it is
remarkable that the effect of cultivation in different incubators in different laboratories is more prevalent than the
effect of light conditions. In general, all DGGE profiles (Figs
3a–c) derived from the same incubator setup shared dominant bands despite their cultivation under different light
regimes, while the profiles from the same light regimes were
profoundly different for each incubator setup.
This is confirmed by the UPGMA dendrogram (Fig. 4)
describing the relatedness of the three DGGE profiles. The
dendrogram shows a clear separation between the different
incubators and no similarity within the corresponding light
intensities. These results indicate that, despite efforts to
operate the incubators in each laboratory under identical
2006 Federation of European Microbiological Societies
Published by Blackwell Publishing Ltd. All rights reserved
c
176
G. Roeselers et al.
Fig. 7. Evolutionary tree showing the phylogenetic affiliations as revealed by comparative analysis of the eukaryotic SS rRNA gene sequences derived
from the DGGE bands shown in Fig. 3c. The sequences obtained in this study are printed bold. Trypanosoma cruzi (AF245380) was used as an out-group
sequence, but was pruned from the tree. Accession numbers of sequences are noted behind the taxon names. Scale bar indicates 10% estimated
sequence divergence. Numbers on the branches are bootstrap values; only values higher than 50% are given.
conditions, the species composition of the mature biofilm
communities was highly variable.
These differences suggest a poor reproducibility in species
composition of microcosm experiments with benthic microbial communities, such as phototrophic biofilms. The
literature describes experiments with heterotrophic biofilms
which exhibited a high degree of reproducibility (Heydorn
et al., 2000; Jackson et al., 2001; Lewandowski et al., 2004).
However, these studies focused mainly on the structural
development of heterotrophic biofilms with a defined
number of species. Although the structural development
was not compared in detail in this study, we observed that
the diversity in community composition is not reflected in
the growth curves of the biofilms (Fig. 2).
It has been reported that DGGE profiles of replicate
pelagic marine mesocosms showed a high resemblance
(Schäfer et al., 2001; Lindstrom et al., 2004). Pelagic
environments are traditionally conceptualized as chemically
and physically more homogeneous, and hence biologically
more homogenous, than benthic habitats such as biofilms.
The spatial heterogeneity creates a manifold of microenvironments within multi-species biofilms, and it can be anticipated that this will stimulate a high species diversity and
2006 Federation of European Microbiological Societies
Published by Blackwell Publishing Ltd. All rights reserved
c
structural complexity. This will result in poor reproducibility of model ecosystem experiments.
The duplicate cultivation experiments with the flow cell
incubators showed high DGGE profile similarity. Although
the flow cells are small in size and are completely closed
systems, which were sterilized before inoculation (unlike the
large microcosms), these results indicate that reproducible
cultivation of phototrophic biofilms is possible.
It could be argued that small and unidentified differences
in the operating conditions of the large incubators induced
the divergence in species composition. It must be noted that
extreme care was taken to keep all conditions identical. Since
the large incubators are open systems, another explanation
could be that the starting community was not very resistant
to invaders.
Succession is usually defined as an ordered unidirectional
process of species replacement leading to a stable climax
community. However, it has also been proposed that regular
disturbances could prevent communities from reaching a
stable climax stage (Massol-Deya et al., 1997). Our biofilm
communities diverged to different compositions from identical starting communities and conditions. Therefore, it
seems reasonable to hypothesize that a rather unstable
FEMS Microbiol Ecol 58 (2006) 169–178
177
Reproducibility of microcosm experiments
community was cultivated (Grimm et al., 1992) or that, for
instance, fluctuating nutrient concentrations disturbed the
succession towards a stable reproducible climax community.
At the moment of sampling, we identified the biofilms as
mature. However, the end of exponential growth does not
necessarily mean that a stable climax community has
established. Therefore, we cannot exclude that the biofilms
were still in a transient state, developing slowly towards a
final convergence.
There is often an implicit, and untested, assumption that
when a model ecosystem becomes more complex in terms of
species diversity and environmental parameters, then it
becomes more difficult to maintain conditions identical
and stable between replicate experiments (Kangas & Adey,
1996; Wynn & Paradise, 2001). It has been postulated that
chaotic dynamics and other nonlinear phenomena can play
a role in community ecology (Allen et al., 1993; Vandermeer
et al., 2002). Although empirical evidence of chaos, or
complex behaviour, in ecosystems is scarce (Clodong &
Blasius, 2004), it is possible that the observed variation
results from intrinsic complex and even chaotic behaviour
of microbial communities (Becks et al., 2005).
Ecological studies have shown that there are trade-offs of
microcosm size with predictability and experimental reproducibility (Kangas & Adey, 1996). These scale effects could
be relevant for the observed differences in reproducibility
between the large incubators and the small flow cell incubators.
Our findings demonstrate that secure experimental validation of reproducibility is essential for the use of microcosm systems in microbial ecology studies, especially for
conclusions concerning differences in community composition and biodiversity in benthic systems.
Future experiments should focus on the relationship
between the observed differences in community composition and other biofilm parameters such as chlorophyll A
content per gram dry mass, oxygen profiles, and EPS
fractions. In addition, the temporal and spatial aspects of
phototrophic biofilm community compositions in response
to environmental disturbances remain an interesting subject
for further investigations.
Acknowledgements
This research was supported by the European Union (PHOBIA project, contract QLK3-CT-2002-01938). We thank
Patrizia Albertano and coworkers (University of Rome ‘Tor
Vergata’, Italy) for providing the inoculum. We thank Emel
Sahan for assistance in the UPGMA analysis. We thank Jan
Rijstenbil (NIOO-KNAW, Netherlands Institute of Ecology)
for coordination of the PHOBIA project. This is publication
no. 3824 of NIOO-KNAW.
FEMS Microbiol Ecol 58 (2006) 169–178
References
Albertano P, Congestri R & Shubert LE (1999) Cyanobacterial
biofilms in sewage treatment plants along the Thyrrenian coast
(Mediterranean Sea), Italy. Arch Hydrobiol: Algol Stud 94
(Suppl): 13–24.
Allen JC, Schaffer WM & Rosko D (1993) Chaos reduces species
extinction by amplifying local population noise. Nature 364:
229–232.
Altschul SF, Gish W, Miller W, Myers EW & Lipman DJ (1990)
Basic local alignment search tool. J Mol Biol 215: 403–410.
Becks L, Hilker FM, Malchow H, Jurgens K & Arndt H (2005)
Experimental demonstration of chaos in a microbial food web.
Nature 435: 1226–1229.
Bender J & Phillips P (2004) Microbial mats for multiple
applications in aquaculture and bioremediation. Bioresour
Technol 94: 229–238.
Clodong S & Blasius B (2004) Chaos in a periodically forced
chemostat with algal mortality. Proc Biol Sci 271: 1617–1624.
Cogan NG & Keener JP (2004) The role of the biofilm matrix in
structural development. Math Med Biol 21: 147–166.
Des Marais DJ (1990) Microbial mats and the early evolution of
life. Trends Ecol Evol 5: 140–144.
Diez B, Pedros-Alio C, Marsh TL & Massana R (2001)
Application of denaturing gradient gel electrophoresis
(DGGE) to study the diversity of marine picoeukaryotic
assemblages and comparison of DGGE with other molecular
techniques. Appl Environ Microbiol 67: 2942–2951.
Griffiths RI, Whiteley AS, O’Donnell AG & Bailey MJ (2000)
Rapid method for coextraction of DNA and RNA from natural
environments for analysis of ribosomal DNA- and rRNAbased microbial community composition. Appl Environ
Microbiol 66: 5488–5491.
Grimm V, Schmidt E & Wissel C (1992) On the application of
stability concepts in ecology. Ecol Model 63: 143–161.
Guillard RRL & Hargraves PE (1993) Stichochrysis immobilis is a
diatom, not a chrysophyte. Phycologia 32: 234–236.
Heydorn A, Ersboll BK, Hentzer M, Parsek MR, Givskov M &
Molin S (2000) Experimental reproducibility in flow-chamber
biofilms. Microbiology 146: 2409–2415.
Jackson G, Beyenal H, Rees WM & Lewandowski Z (2001)
Growing reproducible biofilms with respect to structure and
viable cell counts. J Microbiol Methods 47: 1–10.
Kangas P & Adey W (1996) Mesocosms and ecological
engineering. Ecol Eng 6: 1–5.
Lewandowski Z, Beyenal H & Stookey D (2004) Reproducibility
of biofilm processes and the meaning of steady state in biofilm
reactors. Water Sci Technol 49: 359–364.
Lindstrom ES, Vrede K & Leskinen E (2004) Response of a
member of the Verrucomicrobia, among the dominating
bacteria in a hypolimnion, to increased phosphorus
availability. J Plankton Res 26: 241–246.
Ludwig W, Strunk O, Westram R, et al. (2004) ARB: a software
environment for sequence data. Nucleic Acids Res 32:
1363–1371.
2006 Federation of European Microbiological Societies
Published by Blackwell Publishing Ltd. All rights reserved
c
178
Massol-Deya A, Weller R, Rios-Hernandez L, Zhou JZ, Hickey RF
& Tiedje JM (1997) Succession and convergence of biofilm
communities in fixed-film reactors treating aromatic
hydrocarbons in groundwater. Appl Environ Microbiol 63:
270–276.
Nübel U, Garcia-Pichel F & Muyzer G (1997) PCR primers to
amplify 16S rRNA genes from cyanobacteria. Appl Environ
Microbiol 63: 3327–3332.
Odum HT (1989) Ecological engineering and self-organization.
Ecological Engineering: An Introduction to Ecotechnology
(Mitsch WJ & Jrgensen SE, eds), pp. 79–101. John Wiley and
Sons, New York.
Paerl HW, Pinckney JL & Steppe TF (2000) Cyanobacterialbacterial mat consortia: examining the functional unit of
microbial survival and growth in extreme environments.
Environ Microbiol 2: 11–26.
Palmer RJ Jr (1999) Microscopy flowcells: perfusion chambers for
real-time study of biofilms. Methods Enzymol 310: 160–166.
Prince RC & Kheshgi HS (2005) The photobiological production
of hydrogen: potential efficiency and effectiveness as a
renewable fuel. Crit Rev Microbiol 31: 19–31.
Schäfer H & Muyzer G (2001) Denaturing gradient gel
electrophoresis in marine microbial ecology. Methods in
2006 Federation of European Microbiological Societies
Published by Blackwell Publishing Ltd. All rights reserved
c
G. Roeselers et al.
Microbiology, Marine Microbiology, Vol. 30 (Paul JH, ed), pp.
425–468. Academic Press, New York.
Schäfer H, Bernard L, Courties C, et al. (2001) Microbial
community dynamics in Mediterranean nutrient-enriched
seawater mesocosms: changes in the genetic diversity of
bacterial populations. FEMS Microbiol Ecol 34: 243–253.
Schumacher G, Blume T & Sekoulov I (2003) Bacteria reduction
and nutrient removal in small wastewater treatment plants by
an algal biofilm. Water Sci Technol 47: 195–202.
Stanier RY, Kunisawa R, Mandel M & Cohen-Bazire G (1971)
Purification and properties of unicellular blue-green algae
(order Chroococcales). Bacteriol Rev 35: 171–205.
Vandermeer J, Evans MA, Foster P, Hook T, Reiskind M & Wund
M (2002) Increased competition may promote species
coexistence. Proc Natl Acad Sci USA 99: 8731–8736.
Wimpenny J, Manz W & Szewzyk U (2000) Heterogeneity in
biofilms. FEMS Microbiol Rev 24: 661–671.
Wynn G & Paradise CJ (2001) Effects of microcosm scaling and
food resources on growth and survival of larval Culex pipiens.
BMC Ecol 1: 3.
Zippel B & Neu TR (2005) Growth and structure of phototrophic
biofilms under controlled light conditions. Water Sci Technol
52: 203–209.
FEMS Microbiol Ecol 58 (2006) 169–178