Content uploaded by Tenna Riis
Author content
All content in this area was uploaded by Tenna Riis on Nov 01, 2017
Content may be subject to copyright.
EV-1
Microbial community diversity and composition varies with habitat
characteristics and biofilm function in macrophyte-rich streams
Peter S. Levi, Piotr Starnawski, Britta Poulsen, Annette Baattrup-Pedersen, Andreas Schramm
and Tenna Riis
P. S. Levi (peter.levi@drake.edu), A. Baattrup-Pedersen and T. Riis, Aquatic Biology, Dept of Bioscience, Aarhus University, Aarhus, Denmark.
Present address for PSL: Environmental Science and Policy, Drake University, Des Moines, IA, USA. – P. Starnawski, B. Poulsen and
A. Schramm, Microbiology, Dept of Bioscience, Aarhus University, Aarhus, Denmark.
Biofilms in streams play an integral role in ecosystem processes and function yet few studies have investigated the broad
diversity of these complex prokaryotic and eukaryotic microbial communities. Physical habitat characteristics can affect
the composition and abundance of microorganisms in these biofilms by creating microhabitats. Here we describe the
prokaryotic and eukaryotic microbial diversity of biofilms in sand and macrophyte habitats (i.e. epipsammon and
epiphyton, respectively) in five macrophyte-rich streams in Jutland, Denmark. e macrophyte species varied in growth
morphology, C:N stoichiometry, and preferred stream habitat, providing a range in environmental conditions for the
epiphyton. Among all habitats and streams, the prokaryotic communities were dominated by common phyla, including
Alphaproteobacteria, Bacteriodetes, and Gammaproteobacteria, while the eukaryotic communities were dominated by
Stramenopiles (i.e. diatoms). For both the prokaryotes and eukaryotes, the epipsammon were consistently the most diverse
communities and the epiphytic communities were generally similar among the four macrophyte species. However, the
communities on the least complex macrophyte, Sparganium emersum, had the lowest richness and evenness and fewest
unique OTUs, whereas the macrophyte with the most morphological complexity, Callitriche spp., had the highest number
of unique OTUs. In general, the microbial taxa were ubiquitously distributed across the relatively homogeneous Danish
landscape as determined by measuring the similarity among communities (i.e. Sørensen similarity index). Furthermore, we
found significant correlations between microbial diversity (i.e. Chao1 rarefied richness and Pielou’s evenness) and biofilm
structure and function (i.e. C:N ratio and ammonium uptake efficiency, respectively); communities with higher richness
and evenness had higher C:N ratios and lower uptake efficiency. In addition to describing the prokaryotic and eukaryotic
community composition in stream biofilms, our study indicates that 1) physical habitat characteristics influence microbial
diversity and 2) the variation in microbial diversity may dictate the structural and functional characteristics of stream
biofilm communities.
Our understanding of the role that microbes have in the
structure and function of living entities, from organisms to
ecosystems, is expanding at rapid speed. e recent devel-
opment of next-generation sequencing for prokaryotic and
eukaryotic organisms allows for more frequent and exten-
sive investigations of microbial diversity (Metzker 2010),
allowing ecologists to describe the composition of these
communities in detail and address questions on the rela-
tionships between microbial diversity and other ecosystem
attributes. In aquatic ecosystems, the relationship between
the physicochemical characteristics of the ecosystem, the
functional processes therein, and the microbial community
are often closely interwoven (Tatariw et al. 2013). Reciprocal
interactions exist among these ecosystem attributes implying
that as one changes, others do as well (Finlay et al. 1997).
For example, the diversity of microorganisms in stream bio-
films has been shown to influence hydrologic retention and,
in turn, the use of available resources (Battin et al. 2003 and
Singer et al. 2010, respectively). Furthermore, environmental
variation among different habitats within the same ecosys-
tem may create filters that determine which prokaryotes and
microbial eukaryotes can persist in certain niches (Hempel
et al. 2010, Besemer et al. 2012). ese studies demonstrate
that microbial communities can influence environmental
characteristics and, likewise, environmental characteristics
can dictate what microbes are present.
e diversity of microbial communities may vary within
and between ecosystems due to strong environmental
gradients and/or spatial variation. For instance, the diver-
sity of epilithic biofilms varied in a series of connected
lakes across an altitudinal gradient from montane to alpine
ecosystems (Bartrons et al. 2012). Similarly, spatial dis-
tance can influence diversity in disconnected ecosystems,
because certain prokaryotes and eukaryotes may be dispersal
limited (Lee et al. 2013). Much of our understanding on
the interactions of biofilms and their environment comes
© 2016 e Authors. Oikos © 2016 Nordic Society Oikos
Subject Editor: Silke Langenheder. Editor-in-Chief: Dries Bonte. Accepted 17 August 2016
Oikos 000: 001–012, 2016
doi: 10.1111/oik.03400
EV-2
from studies in marine and lentic ecosystems (reviewed by
Wahl et al. 2012), though environmental and/or spatial
variation have been shown to dictate microbial diversity in
stream ecosystems as well. For example, Acidobacteria domi-
nated the composition of microbial communities in streams
with low pH impacted by acid mine drainage, while more
neutral streams had higher abundances of alpha-, beta- and
gammaproteobacteria (Fierer et al. 2007), sub-groups of a
phylum that frequently dominates microbial communities
in aquatic ecosystems (He et al. 2014). In contrast, spatial
variation influenced diatom diversity more strongly than
habitat-specific environmental variation across a broad geo-
graphic region (Heino et al. 2010). Lotic ecosystems are
highly connected due to the constant flow of water yet also
have highly heterogeneous habitats and, therefore, provide
a strong model system to explore how spatial and environ-
mental variation may control the composition and diversity
of microbial communities.
Streams are highly connected ecosystems, receiving inputs
from upstream, riparian, and groundwater habitats, and
often have a high degree of heterogeneity among the various
habitats within a reach (e.g. pools, riffles, hyporheos). ere-
fore, microorganisms from many different source popula-
tions may be easily dispersed throughout stream networks
while habitat-specific physicochemical characteristics may
dictate the composition of microbial communities (Hempel
et al. 2010). Microbial communities grow and develop on
the myriad surfaces in stream habitats, forming biofilms held
together by water, microorganisms, polysaccharide excre-
tions, and other organic and inorganic materials (Sutherland
2001), and include epilithon on rocks, epipsammon on sand
and epiphyton on aquatic vegetation (i.e. macrophytes). e
formation and growth of aquatic biofilms are often initiated
by certain taxa that have an ability to co-aggregate, such
as certain beta- and gammaproteobacteria (Besemer et al.
2012). As the microbial biofilms continue to develop, they
can play an integral role in stream structure and function
(Finlay et al. 1997). Biologically, biofilms can assimilate and
transform water column nutrients, such as ammonium and
nitrate (Baker et al. 2009, Levi et al. 2015) and, in so doing,
alter the nutrient availability for other biota in adjacent and
downstream ecosystems. Physically, the microarchitecture of
biofilms alone can act as a zone of hydrologic storage (Battin
et al. 2003), briefly slowing the downstream transportation
of nutrients to allow greater processing and transformation
rates. e degree to which stream biofilms may influence
ecosystem structure and function depends, in part, on the
substrate on which the biofilms grow and develop.
In many lowland streams, macrophytes provide a
dynamic surface for biofilms as the plants themselves grow
and metabolize (Hempel et al. 2010). Marcophytes vary in
the complexity of their growth forms, which can alter physi-
cochemical attributes to different degrees (e.g. water velocity;
Dodds and Biggs 2002) and create additional niches with
habitats for microorganisms (Grossart et al. 2013). Biotic
complexity and patchiness can affect the diversity of micro-
organisms within macrophyte beds (Eriksson et al. 2006).
Furthermore, the composition of microbial communities
associated with macrophytes often differs from the microbial
community suspended in the water column or in biofilms
on nearby inanimate objects (Hempel et al. 2010, He et al.
2014). However, few studies have investigated the patterns
in microbial communities across varying environmental
conditions at small scales (Lear et al. 2014), such as between
two macrophyte species with different growth morphology.
e influence of macrophytes on microbial community
composition may, in turn, affect the structure and function
of the stream biofilms.
e hypothesis that microbial diversity may influence
function is not new (Finlay et al. 1997). For example,
high diversity in algal assemblages had a positive effect on
nitrate uptake in stream ecosystems (Baker et al. 2009).
In addition, physical habitat complexity can increase both
microbial diversity and resource use in stream biofilms
(Singer et al. 2010). However, no studies have investigated
the role of both prokaryotic and eukaryotic diversity on
functional processes across a gradient in physical habitat
complexity such as that created by the growth morpholo-
gies of different macrophyte species. e goals of our study
were three-fold: 1) to describe the prokaryotic and eukary-
otic communities in epipsammic and epiphytic biofilms, 2)
to compare the composition of these communities across
habitats with varying physical substrate and environmental
conditions, and 3) to investigate whether microbial diver-
sity was related to biofilm structure and function, measured
as C:N ratio and ammonium (NH4
) uptake efficiency,
respectively.
Material and methods
Study sites, field collection, and sample preparation
We collected biofilm samples from five macrophyte-rich
streams in Jutland, Denmark, during the summer in 2012.
e streams were located in watersheds with mixed land-use
that were primarily dominated by agriculture (Table 1). We
determined the land cover of our study watersheds using
ArcGIS software and images from the geospatial database
maintained by the Geodata Board of the Danish Ministry of
Table 1. Watershed and stream characteristics of the five study streams in Jutland, Denmark. Streams are ordered by increasing macrophyte
coverage. Stream abbreviations are used throughout the manuscript.
Stream Lat. (N) Long. (E)
Watershed
area (km2)
Agricultural
land-cover (%)
Forested
land-cover (%)
Baseflow
discharge (l s–1)
Mean
width (m)
Macrophyte
coverage (%)
Fåremølle (FML) 56°27′8°15′32.3 85.4 9.4 164 2.9 5.9
Lilleå (LIL) 56°15′10°04′21.7 72.8 16.8 142 2.5 27.6
Idom (IDM) 56°20′8°28′21.5 49.5 26.9 223 3.3 37.7
Gryde (GRY) 56°19′8°32′32.6 50.4 27.3 249 4.0 40.7
Madum (MAD) 56°14′8°24′30.0 33.3 48.8 141 3.1 66.6
EV-3
the Environment (< www.gst.dk >). We selected the streams
based on their similar watershed size, land-use, and physi-
cochemical characteristics. Given that our research was an
in situ field study, the species richness of the macrophyte
communities and overall macrophyte coverage varied among
the replicate streams (2–4 species and 6 to 67%, respectively;
Table 1). e stream sediments were predominantly sand
with scant patches of gravel and fine benthic organic matter
(FBOM) in the margins.
To sample the epipsammic and epiphytic biofilms, we
collected a benthic core of sand (core diameter 28 mm,
depth 16 mm) and harvested several cuttings of the
various macrophyte species present in each study stream.
We immediately placed the samples in dark containers
with a small volume of stream water, set the containers on
ice, and transported them back to the laboratory. In the
laboratory, we carefully removed the epiphytic and epip-
sammic biofilms from the macrophytes and sediment,
respectively. For the epiphytic biofilms, we placed the mac-
rophyte samples in a plastic tray with a minimal amount of
deionized water and gently brushed the stems and leaves
of the macrophyte, creating a biofilm slurry. We removed
the macrophytes to a second tray and repeated the process.
After thoroughly rinsing and removing the macrophyte, we
combined the biofilm slurries from both trays, gently agi-
tated them, and filtered a portion of the slurry onto a Supor
polyethersulfone (PES) filter (pore size 0.2 mm). e
average wet mass of the samples was 0.03 g (range 0.01
to 0.1 g). For the epipsammic biofilms, we agitated the
sand and water by gently swirling the sample and using a
syringe to sample the suspended biofilm, fine particulate
matter and sand grains (Stevenson and Rollins 2007). We
then filtered a portion of the slurry onto a Supor PES filter
and the average wet mass was 0.1 g (range 0.02 to 0.2
g). To avoid contamination among samples, we changed
our gloves and soaked the trays and filtering apparatus in a
mild bleach solution for 10 min between processing each
sample. We sealed the filters into individual test tubes and
immediately placed them into a –80°C freezer until the
DNA extraction.
DNA extraction and sequencing
We used IonTorrent sequencing of 16S rRNA and 18S rRNA
gene amplicons to analyze the prokaryotic and eukaryotic
diversity of the epiphytic and epipsammic biofilms. We
extracted total DNA from the biofilm samples via enzymatic
and mechanical lysis with the Fast DNA Spin Kit for Soil,
a protocol optimized for biofilms (Foesel et al. 2008). To
quantify the DNA concentrations, we used a Qubit fluo-
rometer and the Qubit dsDNA HS Assay Kit. We amplified
bacterial and archaeal 16S rRNA genes using primer pair
Univ519F (5′-CAGCMGCCGCGGTAA-3′) and Univ802R
(5′-TACNVGGGTATCTAATCC-3′; Claesson et al. 2009,
Wang and Qian 2009) and amplified eukaryotic 18S rRNA
genes using primer pair Euk345f (5′-AAGGAAGGCAG
CAGGCG-3′) and Euk499r (5′-CACCAGACTTGCCCT
CYAAT-3′; Zhu et al. 2005). We conducted all the PCRs
with the KAPA HiFi PCR Kit. First, we ran 20 PCR
cycles at an annealing temperature of 49°C and 53°C for
the prokaryotes and eukaryotes, respectively; these initial
amplification reactions contained bovine serum albumin
(0.2 mg ml–1). Next, we barcoded the PCR products using
the Ion Torrent Xpress system for an additional 10 PCR
cycles, raising the annealing temperature to 61°C. en we
purified the barcoded PCR products with the Agencourt
AMPure XP Kit, quantified the DNA concentration with
the Qubit system, pooled the PCR products, and prepared
the sequencing libraries according to the Ion Torrent Manual
(Preparing Short Amplicon ( 350 basepairs) Libraries). For
quality control of the library, we used both a Qubit Fluo-
rometer and Bioanalyzer 2100 with High Sensitivity DNA
Analysis Kit. We performed an Emulsion PCR on a One
Touch Instrument with the Ion PGM Template OT2 400
Kit, and sequenced the samples on the Ion Torrent PGM
with an Ion 314 Chip and the Ion PGM Hi-Q Sequencing
Kit according to the manufacturer’s instructions.
We conducted the quality filtering and clustering of
the operational taxonomic units (OTUs) in Mothur, ver.
1.36.1 (Schloss et al. 2009). First, we discarded reads with
more than seven homopolymers and those with lengths
below 200 basepairs for prokaryotes and 150 basepairs
for eukaryotes. Next, we aligned and classified the data
using a combined prokaryote-eukaryote Silva reference
database, ver. 123 (Quast et al. 2013). We removed chime-
ras with the Uchime tool (Edgar 2013) and clustered the
resulting sequences into OTUs with a similarity cutoff of
97% and the furthest-neighbor algorithm. We processed
the OTU-frequency table and classification data in R and
the rarefaction curves, subsampling and diversity indices
(i.e. Shannon’s diversity index, Chao1 rarefied richness,
ACE, and Pielou’s evenness) using the Vegan package
(ver. 2.3). Our sequences are publicly available from MG-
RAST (< http://metagenomics.anl.gov >) under the IDs
4681462.3 and 4681503.3.
Stream ecosystem predictor variables
We measured various physical, chemical, and biological
characteristics at the habitat and biofilm scales, including
metrics of both structure and function. Here we briefly
describe the methods used during our present study and
include citations with more thorough descriptions of spe-
cific methods, particularly from our previous study on these
streams (Levi et al. 2015; see also Supplementary material
Appendix 1).
At the habitat scale, we quantified the biomass and
morphological complexity of the macrophyte species in
each stream. To determine macrophyte coverage and vol-
ume, we measured the median width, length, and depth
of every macrophyte bed in the study reaches. We sam-
pled the submerged macrophyte species that were found
in at least three of the five streams, which included Berula
erecta, Callitriche spp., Ranunculus peltatus and Sparganium
emersum (Table 2; Moeslund et al. 1990). We determined
the complexity of the macrophytes by measuring the perim-
eter-to-area ratio (P:A) of each species, which represented
the surface area available for epiphytic biofilm. We calcu-
lated the P:A with black-and-white images of the top 20 cm
of five replicate individuals of each macrophyte species. We
EV-4
factor and ‘stream’ as the block (Zar 2009). We conducted
a Tukey’s post hoc comparison test with all statistically
significant tests (i.e. p-value 0.05) to determine which
habitats varied from each other. To examine the variation
in diversity among habitats and streams (i.e. b-diversity),
we calculated the Sørensen index and ran separate one-way
ANOVAs with ‘habitat’ and ‘stream’ as the factors. We cre-
ated Venn diagrams to illustrate the shared OTUs among
habitats using the VennDiagram package (ver. 1.6.9) in R.
We examined the differences among the communities with
a permutational MANOVA to assess the patterns in micro-
bial composition between habitats and nested by stream
using the PERMANOVA add-on in the PRIMER soft-
ware package, ver. 6 (Clarke and Gorley 2006). Given the
significant results of the PERMANOVA, we conducted a
canonical analysis of the principal coordinates to assess the
differences among habitats (Anderson and Willis 2003).
Using Pearson’s correlation (Zar 2009), we examined
whether the richness and evenness of the microbial com-
munities were related to the measures of macrophyte com-
plexity (i.e. leaf P:A, bed density). Finally, we used both
one-way RB ANOVA and Pearson’s correlation to investi-
gate the relationship between microbial diversity and biofilm
characteristics (i.e. C:N, NH4
uptake efficiency). ough
we analyzed replicate samples for the biofilm characteris-
tics (n 5), we used the mean value in the correlations.
We used Chao1 rarefied richness and Pielou’s evenness for
these correlations rather than diversity indices like Shan-
non’s in order to separately account for species richness
and evenness. We tested all of our data for normality and
removed statistical outliers when studentized residuals
exceeded 2.5 (Zar 2009). We completed our statistical anal-
yses using SYSTAT (ver. 10) and R (ver. 3.2.0 in R-studio
< www.r-project.org >).
imported the images into ImageJ software, which measured
the perimeter and area of each individual sample (National
Institutes of Health, USA). We used the P:A and volume of
each macrophyte bed to calculate the amount of surface area
per cubic meter of macrophyte bed for each species (i.e. bed
density; m2 m–3).
At the biofilm scale, we measured metrics of both
structure and function of these communities. To calculate
the carbon-to-nitrogen ratio (C:N) of biofilms from each
stream and habitat, we determined the C and N content
using a total organic carbon analyzer. As a measure of biofilm
function, we quantified the in situ NH4
uptake efficiency
of these biofilms (Levi et al. 2015). In short, we conducted
24-h additions of 15N-NH4
in each study reach following
standard methods (Mulholland et al. 2000) within two to
four days of collecting the samples for the biofilm com-
munity analysis. We calculated the compartment-specific
NH4
uptake rate by measuring the concentration of 15N
in the epipsammon and epiphyton relative to the 15N con-
centration of the stream water (Levi et al. 2015). With these
data and the N biomass of the biofilms, we calculated the
biomass-specific NH4
uptake (i.e. uptake efficiency; mg N
mg Nbiomass–1 d–1) as follows:
uptake rate compartmentspecificNHuptake
Nbiomass
=−+
4
Our stable isotope samples were analyzed with a mass
spectrometer.
Data and statistical analyses
To examine the variation in richness and evenness among
the communities, we used a one-way randomized-block
analysis of variance (RB ANOVA) with ‘habitat’ as the
Table 2. Mean ( SE) physical and biological measurements of the different macrophyte species ordered in terms of increasing bed density.
Leaf morphology photographs represent top 20 cm of each species. Data represent mean values with standard error in parentheses.
Macrophyte species
Leaf morphology
Presence in streams*F-I-G-M F-L-I-G L-I-G-M L-I-G
Dry mass (g m–3)#50 (8) 82 (15) 195 (20) 79 (12)
Bed density (m2 m–3)#2.4 (0.6) 5.5 (1.4) 9.7 (2.3) 11.6 (3.3)
Leaf P:A 2.5 (0.2) 2.1 (0.1) 13.4 (1.3) 10.0 (0.6)
C:N 13.1 (0.8) 9.7 (0.2) 9.4 (0.3) 9.2 (0.6)
*Streams abbreviated with first letter of stream (F Fåremølle, L Lilleå, I Idom, G Gryde, M Madum).
#Volume for dry mass and bed density were calculated for 1m 1m 0.1m of the macrophyte bed. The volume was only determined to
0.1m of depth because that was the depth to which we sampled (Levi et al. 2015).
EV-5
were still increasing in species number (Supplementary
material Appendix 3). However, given the magnitude of
bacterial OTUs reported, we are confident we sampled
these communities to an appropriate depth for our analysis
(Gotelli and Colwell 2001).
e Sørensen index (e.g. similarity coefficient) demon-
strated that the prokaryotic communities had comparable
similarity among habitats and streams (Table 3; one-way
ANOVA, p 0.3 and 0.6, respectively). However, the
eukaryotes of the epiphytic biofilms on S. emersum were
more variable between the streams than the eukaryotes in the
epipsammon or epiphyton of the other three macrophytes,
which all had comparable similarity among streams (one-
way ANOVA p 0.001; Table 3). In addition, the eukary-
otic communities in Fåremølle were more similar to each
other than the eukaryotic communities were to each other in
the other four streams (p 0.04).
In general, the majority of prokaryotes in the biofilm
communities were Alphaproteobacteria (mean SE
abundance 24.6 1.9%) and Bacteriodetes (21.4 2.0%;
Fig. 2A). Two epiphytic communities on S. emersum were
dominated by cyanobacteria (abundance 34%), though
cyanobacteria only accounted for 7.2% of the OTUs among
all the biofilms. Stramenopiles (i.e. diatoms) was the domi-
nant eukaryotic phylum in most streams (47.5 4.9%;
Fig. 2B). For example, diatoms alone accounted for more
than 75% of the eukaryotic OTUs among all habitats in
Results
Prokaryotic and eukaryotic diversity across streams
and habitats
Among all biofilms we collected across the five study streams,
the prokaryotes were more diverse than the eukaryotes
and, in both taxonomic domains, richness and evenness gen-
erally varied across the stream habitats. e mean ( SE)
Chao1 rarefied richness was 4050 320 and 2270 120
for the prokaryotes and eukaryotes, respectively, while the
mean ( SE) Pielou’s evenness was 0.840 0.015 and
0.576 0.23, respectively. A similar pattern was observed
for other measures of diversity, including the inverse of
Simpson’s index, Shannon’s diversity index, and ACE
(Supplementary material Appendix 2). Across stream habi-
tats, the prokaryotes in the epipsammon and epiphyton of
Berula erecta had higher richness and evenness than the other
epiphytic communities (Fig. 1A, 1C; one-way RB ANOVA,
pHABITAT 0.001; Tukey’s PCT, p 0.05). e evenness
of the eukaryotic communities also varied by habitat (one-
way RB ANOVA, pHABITAT 0.01), where the epipsammon
and B. erecta epiphyton had higher evenness than the epi-
phyton on Sparganium emersum (Fig. 1D). e asymptotes
of the rarefaction curves for the eukaryotic communities
demonstrated that these communities were sampled deep
enough, though several of the prokaryotic communities
(A) Prokaryotes
Chao1 rarefied richness
0
2000
4000
6000
8000 (B) Eukaryotes FML
LIL
IDM
GRY
MAD
pHABITAT = 0.06
pSTREAM = 0.7
pHABITAT < 0.001
pSTREAM = 0.7
a
c
ab
bc
bc
(C) Prokaryotes
Sand
Callitrichespp.
R.peltatus
B.erecta
S.emersum
Pielou's evenness
0.0
0.2
0.4
0.6
0.8
1.0
(D) Eukaryotes
Sand
Callitrichespp.
R.peltatus
B.erecta
S.emersum
pHABITAT = 0.01
pSTREAM = 0.04
pHABITAT < 0.001
pSTREAM = 0.8
a
bc
ab
c
abc
a
b
a
ab
ab
Figure 1. Chao1 rarefied richness and Pielou’s evenness of the (A, C) prokaryotic and (B, D) eukaryotic biofilm communities. Letters
denote habitats with statistically different communities (Tukey’s PCT). Epiphytic communities listed along the x-axis in terms of decreasing
morphological complexity (i.e. macrophyte bed density; Table 2). Note: not all macrophyte species were found in all streams.
EV-6
between multiple, but not all, habitats (Fig. 3A). Among
the epiphyton communities, B. erecta had the highest per-
centage of unique prokaryotic OTUs (i.e. 10.7% of the
total OTUs) while S. emersum had the fewest unique OTUs
(2.3%). ere were fewer eukaryotic OTUs, but the pattern
among habitat was similar. More than half of the unique
OTUs were in the epiphytic biofilms alone (60.2%), 11.8%
were unique to the epipsammon, 9.7% were shared among
all habitats, and the remaining 18.3% were shared between
Fåremølle. Madum was an exception where Stramenopiles
was in much lower abundance (mean 13.4 5.3%) and
these biofilms were instead dominated by fungi or Alveolates
(32.0 40.8% and 28.5 19.2%, respectively).
e most abundant prokaryotic and eukaryotic OTUs
were present in all of the biofilms. Among the prokaryotes,
46.9% of the OTUs were unique to the epiphyton, 21.4%
were unique to epipsammon, 8.1% were present in all
habitats, and the remaining 23.5% of the OTUs were shared
Table 3. Similarity of prokaryotic and eukaryotic communities in biofilms among streams and habitats expressed as the Sørensen index
( SE). Bold values represent significantly different Sørensen indices for streams or habitats within a taxonomic domain.
Stream Prokaryotes Eukaryotes Habitat Prokaryotes Eukaryotes
FML 0.36 (0.04) 0.54 (0.02) Sand 0.36 (0.02) 0.39 (0.01)
LIL 0.40 (0.03) 0.47 (0.04) S. emersum 0.32 (0.02) 0.30 (0.03)
IDM 0.33 (0.03) 0.40 (0.03) B. erecta 0.35 (0.01) 0.41 (0.02)
GRY 0.33 (0.03) 0.38 (0.02) R. peltatus 0.31 (0.02) 0.36 (0.02)
MAD 0.32 (0.05) 0.39 (0.03) Callitriche spp. 0.33 (0.04) 0.38 (0.02)
Alphaproteobacteria
Betaproteobacteria
Deltaproteobacteria
Gammaproteobacteria
Bacteroidetes
Cyanobacteria
Verrucomicrobia
Chloroflexi
Acidobacteria
Actinobacteria
Planctomycetes
Nitrospirae
Firmicutes
Unclassified
Other
Madum
Gryde
IdomLilleåFaremølle
(A)
Sand
S. emersum
B. erectaR. peltatus
Callitriche spp.
Figure 2. Taxonomic composition of the (A) prokaryotic and (B) eukaryotic epipsammon and epiphyton at the phylum-level in the five
stream ecosystems.
EV-7
the other three macrophyte species overlapped with each
other.
e measures of macrophyte complexity (i.e. P:A, bed
density) were often not correlated with the richness or even-
ness of the microbial communities. e Pielou’s evenness of
the eukaryotes was positively correlated with bed density of
the macrophyte species (r 0.53, p 0.04). However, the
rarefied richness of both the prokaryotes and eukaryotes
as well as the Pielou’s evenness of the prokaryotes were not
correlated with either measure of macrophyte complexity.
Variation in structure and function among microbial
communities
e physical structure of the biofilm communities, expressed
as the C:N ratio, varied among habitats and was correlated
to the richness and evenness of the microbial communities.
C:N was consistently higher in the epipsammon relative
to the epiphyton (Fig. 5A; one-way RB ANOVA, pHABITAT
0.001). Among the macrophyte species, the epiphytic
C:N was lowest in S. emersum (Tukey’s PCT, p 0.01),
but did not vary among the other species. e C:N of
several habitats (Fig. 3B). Among the different macrophyte
species, Callitriche spp. had the most unique OTUs (20.0%
of the total) while S. emersum again had the fewest (2.9%).
In general, the most abundant OTUs were found across all
five streams and often shared among habitats (Supplemen-
tary material Appendix 4, 5). For instance, the most com-
mon eukaryotic OTU in the epiphyton and epipsammon
was a diatom in the order Naviculales (Supplementary mate-
rial Appendix 5).
Ecosystem- and habitat-scale predictors of biofilm
richness and evenness
e biofilm communities varied by habitat as demon-
strated using PERMANOVA and a canonical analysis of
principal coordinates (PERMANOVA, p 0.001 and
0.007, respectively). Both the prokaryotic and eukaryotic
communities in the 20 biofilms separated by habitat with
epipsammon clustering as a distinct group relative to the
epiphyton (Fig. 4). In addition, the microbial communi-
ties on Ranunculus peltatus appeared to cluster separately
for the prokaryotes and eukaryotes while the epiphyton on
Stramenopiles
Alveolata
Viridiplantae
Fungi
Haptophyceae
Cryptophyta
Other
Unclassified
Madum
Gryde
Idom
Lilleå
Faremølle
(B)
Sand
S. emersumB. erectaR. peltatus
Callitriche spp.
Figure 2. (Continued).
EV-8
NH4
mg Nbiomass–1 d
–1, respectively; Fig. 6A). While the
efficiency differed significantly between the sand and mac-
rophyte habitats (one-way RB ANOVA, pHABITAT 0.001),
uptake efficiency did not vary among the macrophyte spe-
cies. e Chao1 rarefied richness of the prokaryotic commu-
nities was negatively correlated to NH4
uptake efficiency
(r –0.66, p 0.004; Fig. 6B). e uptake efficiency of
the eukaryotes also decreased as richness increased, but the
correlation was not significant (Fig. 6D). Pielou’s evenness of
the prokaryotic and eukaryotic communities was also nega-
tively correlated with NH4
uptake efficiency (r –0.62
and –0.63, respectively; p 0.008; Fig. 6C, E). Among the
epiphyton communities alone, there were no significant
correlations between uptake efficiency and biofilm richness
or evenness.
Discussion
Microbial diversity and composition of biofilms
differed by habitat
e microbial communities in the epipsammon were
consistently more diverse than the epiphyton across all stream
biofilms in our study. e differences we observed between
the two habitats were largely driven by the rare OTUs (e.g.
abundance 0.005%) while the most abundant OTUs were
generally shared between all habitats and streams, a common
the prokaryotic communities was positively correlated
to the Chao1 rarefied richness (r 0.73, p 0.001;
Fig. 5B). Similarly, the richness of the eukaryotic com-
munities increased with C:N, but the correlation was not
significant (Fig. 5D). Pielou’s evenness was positively cor-
related with biofilm C:N of the prokaryotes and eukary-
otes (r 0.72 and 0.56, respectively; p 0.01; Fig. 5C,
E). When analyzing the epiphytic communities alone,
C:N was positively correlated with the rarefied richness of
the prokaryotes (r 0.53, p 0.04), eukaryotes (r 0.58,
p 0.02), and Pielou’s evenness of the eukaryotes (r 0.55,
p 0.03).
e biomass-specific NH4
uptake rates (i.e. uptake
efficiency) varied substantially between the microbial com-
munities from the different habitats. e uptake efficiency
of the epiphytic biofilms was nearly two orders of magnitude
higher than the efficiency of the epipsammic biofilms (mean
SE uptake efficiency 0.3 0.04 and 0.008 0.002 mg
Figure 3. Venn diagrams depicting the unique OTUs for the (A)
prokaryotic and (B) eukaryotic communities among the stream
habitats. e data represent OTUs found in 1 sample of each
habitat or combination of habitats (i.e. rare OTUs found in only
one sample are excluded).
(A) Prokaryotes
–0.3 –0.2 –0.10.0 0.10.2 0.30.4 0.5
–0.5 –0.4 –0.3 –0.2 –0.10.0 0.10.2 0.3
–0.6
–0.4
–0.2
0.0
0.2
0.4
0.6
Sand
S. emersum
B. erecta
R. pelatatus
Callitriche spp.
(B) Eukaryotes
Canonical analysis - Axis 1
Canonical analysis - Axis 2
–0.4
–0.2
0.0
0.2
0.4
0.6
Figure 4. Canonical analysis of principal coordinates of (A) prokary-
otic and (B) eukaryotic communities.
EV-9
inorganic nitrogen concentrations change rapidly in the top
0.3 mm of sand in stream ecosystems (Revsbech et al. 2005).
We sampled to a depth of 16 mm and, therefore, the strong
resource gradient allowed for the co-existence of a highly
diverse suite of microbes. In addition, the three-dimensional
matrix of the sand habitat provided the epipsammon with
more surface area for biofilm growth and development than
the surfaces of the macrophyte leaves. irdly, the high rich-
ness and evenness of the epipsammic communities may have
pattern in freshwater microbial communities (Besemer et al.
2012). Variation in the abundance and composition of
OTUs within the microbial communities was likely influ-
enced by variation in spatial and environmental conditions
as has been demonstrated in aquatic ecosystems (Heino et al.
2010, He et al. 2014, Lear et al. 2014). For one, the epip-
sammic communities experienced a higher degree of resource
heterogeneity within the microhabitats of the sand matrix
than the epiphytic communities. For example, oxygen and
(A)
Sand
Callitriche spp.
R. peltatus
B. erecta
S. emersum
C:N
0
5
10
15
20 FML
LIL
IDM
GRY
MAD
(B) Prokaryotes
1000 2000 3000 4000 5000 6000 7000
Sand
S. emersum
B. erecta
R. peltatus
Callitriche spp.
(D) Eukaryotes
Chao1 rarefied richness
1000 1500 2000 2500 3000 3500
C:N
0
5
10
15
20
r = 0.73
p < 0.001
r = 0.58
p = 0.02
pHABITAT < 0.001
pSTREAM < 0.001
a
c
bbb
(C) Prokaryotes
0.70 0.75 0.80 0.85 0.90 0.95 1.00
r = 0.72
p < 0.001
(E) Eukaryotes
Pielou's evenness
0.3 0.4 0.5 0.6 0.7 0.8
r = 0.56
p = 0.01
Figure 5. (A) C:N of epipsammon and epiphyton among the habitats and streams. Letters denote habitats with statistically different C:N
(Tukey’s PCT). Biofilm C:N of the prokaryotic and eukaryotic communities were correlated with Chao1 rarefied richness (B and D,
respectively) and Pielou’s evenness (C and E, respectively).
(A)
Sand
Callitriche spp.
R. peltatus
B. erecta
S. emersum
Uptake efficiency
(mg NH4
+ mg Nbiomass
-1 d-1)
0.0
0.1
0.2
0.3
0.4
0.5
0.6
LIL
IDM
GRY
MAD
(B) Prokaryotes
1000 2000 3000 4000 5000 6000 7000
Sand
S. emersum
B. erecta
R. peltatus
Callitriche spp.
(D) Eukaryotes
Chao1 rarefied richness
1000 1500 2000 2500 3000 3500
Uptake efficiency
(mg NH4
+ mg Nbiomass
-1 d-1)
0.0
0.1
0.2
0.3
0.4
0.5
0.6
r = –0.66
p = 0.004
r = –0.44
p = 0.08
pHABITAT < 0.001
pSTREAM < 0.001
a
bbb
b
(C) Prokaryotes
0.70 0.75 0.80 0.85 0.90 0.95 1.00
r = –0.62
p = 0.008
(E) Eukaryotes
Pielou's evenness
0.3 0.4 0.5 0.6 0.7 0.8
r = –0.63
p = 0.007
Figure 6. (A) Biomass-specific NH4
uptake (i.e. uptake efficiency) of epipsammon and epiphyton among the habitats and streams. Letters
denote habitats with statistically different uptake efficiency (Tukey’s PCT). Uptake efficiency of the prokaryotic and eukaryotic
communities were correlated with Chao1 rarefied richness (B and D, respectively) and Pielou’s evenness (C and E, respectively).
EV-10
eukaryotes on S. emersum further demonstrates that the
stoichiometry, morphological complexity, and/or high shear
stress of the macrophyte species likely limited the richness of
these communities.
Environmental characteristics at the stream scale likely
influenced the composition of the biofilm communities as
well. e most common eukaryotes in our study were dia-
toms (i.e. Stramenopiles) with the exception of the biofilms
in Madum, where heterotrophic Alveolata and Fungi were
the most dominant phyla. e macrophyte R. peltatus densely
covered two-thirds of the total stream area in Madum, which
may have led to shading of the epiphytic biofilms and higher
organic matter retention (Eriksson et al. 2006) while streams
with less dense macrophyte coverage and, therefore, higher
light availability, were dominated by autotrophic eukaryotes
(e.g. Stramenopiles). Among habitats, the comparably high
richness of epiphyton on B. erecta and epipsammon may be
due to the preferred habitat of B. erecta in the stream chan-
nel relative to the other three macrophyte species. Berula
erecta is most prominent in the stream margin, whereas R.
peltatus and Callitriche spp., which must remain submerged,
and are most prevalent in the thalweg (Riis et al. 2001).
erefore, the communities with high richness on B. erecta
may be subjected to less shear stress, but greater shading
and sediment deposition on the leaves (Piggott et al. 2012);
physical conditions that can affect the microbial community
composition in those epiphytic biofilms.
Previous studies have demonstrated that variation in
environmental characteristics across streams can strongly
dictate the composition of microbial communities therein.
However, differences among communities may not be appar-
ent unless the environmental gradients are drastic (e.g. pH
range from 4.0 to 6.3; Fierer et al. 2007). e streams in our
study were similar lowland, macrophyte-rich streams in a
relatively homogeneous landscape. Furthermore, each reach
was connected to upstream habitats and the surrounding
catchment by the unidirectional flow of water in the streams.
erefore, the high similarity among the streams was likely
the result of limited dispersal constraints on the microbes
in stream networks across a landscape with similar environ-
mental characteristics (Heino et al. 2010). Among streams,
the Sørensen similarity index indicated that the eukaryotic
communities across habitats in Fåremølle were more similar
to each other than communities across habitats within other
streams. Fåremølle was the stream most influenced by human
impacts with a highly straightened channel, low macrophyte
coverage, and high agricultural land-use, which likely acted
as environmental filters on the microbial composition of the
biofilms (Tatariw et al. 2013). Taken together, our results
demonstrate that both habitat and environmental charac-
teristics influence the richness and evenness of microbial
communities.
Is microbial richness and evenness related to biofilm
structure and function?
An objective of our study was to examine the relationship
between metrics of microbial diversity and metrics of biofilm
structure and function. e relationship between biodiversity
and ecosystem function is complex, with results indicating
both positive and negative relationships or, in some cases,
been related to more frequent disturbance at the biofilm-
water interface (e.g. bed movement) relative to the biofilm
communities on the macrophytes.
e richness and evenness of the microbial communities
in the epiphyton varied between species, with some com-
munities having comparable measurements of diversity to
the epipsammon and other communities having much lower
measurements (e.g. Berula erecta and Sparganium emersum,
respectively). e generally lower richness and evenness in
the epiphyton relative to the epipsammon suggests that
intense competition and species sorting dictated the com-
position of the microbial communities on the macrophytes
(Besemer et al. 2012, Lee et al. 2013), which may be related
to a higher degree of substrate stability (i.e. macrophyte leaf
versus sand grains; Eriksson et al. 2006). Furthermore, the
high proportion of unique OTUs in the epiphytic commu-
nities for both prokaryotes and eukaryotes (47 and 60%,
respectively) suggests that the epipsammon did not act as
a source population for the epiphyton. Rather, the OTUs
unique to the epiphyton likely came from upstream or
riparian habitats.
e variation we observed between epiphytic communi-
ties on the different macrophyte species may be related to
macrophyte morphology, chemical composition, habitat
conditions, or a combination of these factors. Among the
four macrophyte species, leaf P:A and bed density varied
by nearly an order of magnitude with complex macrophyte
species providing more microhabitats and niches for biofilm
growth and development (e.g. Callitriche spp. and R. pelata-
tus; Eriksson et al. 2006). For example, macrophytes with a
high degree of complexity decrease water velocity and shear
stress at the leaf–water interface (Dodds and Biggs 2002),
which may affect the richness and evenness of the microbial
communities (Grossart et al. 2013). We observed that Cal-
litriche spp., the most complex macrophyte, had the most
unique eukaryotic OTUs, while S. emersum, the species with
the lowest morphological complexity, had the fewest OTUs.
We also found that biofilms on S. emersum often had lower
richness and evenness than the other epiphytic communi-
ties. Similarly, other studies have reported that different
macrophyte species in marine and lentic ecosystems exhibit
distinct microbial communities (Burke et al. 2010) as a result
of environmental variation or the chemical composition of
the host species (Hempel et al. 2010). e simplified mor-
phology and/or chemical composition of S. emersum, which
had the highest C:N of the four species, may have limited
the eukaryotic richness and evenness. e chemical compo-
sition of a macrophyte species effects the exudates released
(e.g. N, phenolic compounds), which, in turn, can influ-
ence the composition of the microbial community (Hempel
et al. 2010, Kofoed et al. 2012). e stoichiometry and
complexity of B. erecta, Ranunculus peltatus and Callitriche
spp. were more similar, suggesting that variation in richness
and evenness among the epiphyton on these macrophtyes
was due to other characteristics, such as physicochemical
variation among the habitats. Additionally, the Sørensen
similarity index, a measure of the similarity in the compo-
sition of OTUs between communities, demonstrated that
the eukaryotic communities on S. emersum were less similar
to each other than epiphytic and epipsammic communities
were among the other habitats. e low similarity among
EV-11
Conclusions
Our study provides an assessment of the diversity and
composition of epipsammic and epiphytic biofilms in
macrophyte-rich streams, including the eukaryotic commu-
nity. e epipsammon had consistently higher diversity than
the epiphyton, though biofilms in both habitats had few
abundant OTUs and many rare ones. We observed consistent
habitat-scale differences in the composition and diversity of
the biofilm communities, as well as in their NH4
uptake
efficiency, across the five streams. Furthermore, variation
of growth morphology and C:N of the macrophyte species
influenced microbial richness and evenness, supporting our
hypothesis that the physical characteristics of freshwater
habitats determine microbial community composition by
creating microhabitats and niche heterogeneity (Eriksson
et al. 2006, Singer et al. 2010). e low diversity and number
of unique OTUs in the epiphyton of S. emersum, the mac-
rophyte species with the least complex growth morphology,
illustrates the effect morphological complexity can have
on biofilm communities. Our study provides a glimpse at
the intimate ecological linkages between microbial diver-
sity and ecosystem ecology. Further advances in sequenc-
ing technology and libraries of microbial taxa will continue
to illuminate the myriad of connections across vast spatial
scales in ecology.
Acknowledgements – We are grateful to Anne Stentebjerg, Andrea
Torti, Xihan Chen, Anette Baisner Alnøe, Kamilla Maetzke and
Christoffer Bruus Pedersen for technical, laboratory, and field
assistance. We thank Dr. Nanci J. Ross for her statistical expertise.
We appreciate the comments and suggestions from Dr. Silke
Langenheder which improved the manuscript.
Funding – We thank the Carlsberg Foundation (no. 2013010258;
TR), Danish Research Council (FNU, no. 272-09-0012; TR), EU
MARS project (no. 603378; ABP), and the Danish National
Research Foundation (DNRF104) for financial support.
References
Anderson, M. J. and Willis, T. J. 2003. Canonical analysis of
principal coordinats: a useful method of constrained ordination
for ecology. – Ecology 84: 511–525.
Baker, M. A. et al. 2009. Differences in nitrate uptake among
benthic algal assemblages in a mountain stream. – J. N. Am.
Benthol. Soc. 28: 24–33.
Bartrons, M. et al. 2012. High bacterial diversity in epilithic
biofilms of oligotrophic mountain lakes. – Microb. Ecol. 64:
860–869.
Battin, T. J. et al. 2003. Effects of current velocity on the nascent
architecture of stream microbial biofilms. – Appl. Environ.
Microbiol. 69: 5443–5452.
Besemer, K. et al. 2012. Unraveling assembly of stream biofilm
communities. – ISME J. 12: 1–10.
Burke, C. et al. 2010. Composition, uniqueness and variability of
the epiphytic bacterial community of the green alga Ulva
australis. – ISME J. 5: 590–600.
Claesson, M. J. et al. 2009. Comparative analysis of pyrosequenc-
ing and a phylogenetic microarray for exploring microbial
community structures in the human distal intestine. – PLoS
ONE 4: e6669.
Clarke, K. R. and Gorley, R. N. 2006. PRIMER ver. 6: user
manual/tutorial: 192.
no relationship at all (reviewed by Hooper et al. 2005). e
inconsistent results demonstrate that these relationships are
dependent on the metrics used to estimate both diversity and
function. In our study, we quantified the stoichiometry of
the biofilms, as expressed by the ratio of carbon-to-nitrogen
(C:N), and the uptake efficiency of ammonium (NH4
)
using a short-term stable isotope release in each stream
(15N-NH4
; Levi et al. 2015). We acknowledge that bio-
film C:N and NH4
uptake efficiency are only two potential
metrics that may be related to microbial diversity given that
microbial taxa perform many ecological functions within
a biofilm community (Peter et al. 2011). However, stud-
ies have shown that the structure and diversity of biofilm
communities can have a positive effect on stream nitrogen
cycling (Baker et al. 2009, Tatariw et al. 2013). Furthermore,
the stoichiometry of organisms often varies among taxa and
trophic state (e.g. autotroph versus heterotroph) and can
influence nutrient cycling within aquatic ecosystems (Sterner
et al. 1992).
e richness and evenness of the microbial communi-
ties were positively related to biofilm C:N ratio and nega-
tively with NH4
uptake efficiency. In previous studies,
microbial diversity has been linked with functional met-
rics in aquatic ecosystems when a targeted analysis of the
microbial community has been used. For example, the
microhabitat conditions created by macrophytes had a
direct effect on the structure and function of nitrifying and
denitrifying bacteria (Kofoed et al. 2012). Also, the diver-
sity of the methanogen community was positively related
to CH4 production in peatlands (Yavitt et al. 2011). In
our study, we analyzed the full suite of microbes because
all autotrophic and heterotrophic organisms rely on nitro-
gen, whether attained through direct assimilation from their
environment or ingestion (Sutherland 2001). erefore,
we could not limit our analysis to one specific phylum or
class of prokaryotes or eukaryotes. We did observe lower
NH4
uptake efficiency in the epiphyton on S. emersum,
which were dominated by Cyanobacteria. e high C:N
of S. emersum suggests that this macrophyte exuded less N
than the species with lower C:N ratios (Sterner et al. 1992),
thereby creating a microhabitat that favored Cyanobacteria
given their ability to fix nitrogen. A similar negative rela-
tionship between diversity and resource use efficiency was
observed in phytoplankton communities, which included
some Cyanobacteria-dominated assemblages (Filstrup et al.
2014). We measured low uptake rates in the epipsammon
as well, which may be due to alternative NH4
sources
below the water–sediment interface. For example, NH4
availability increases rapidly with depth in stream sedi-
ments (Revsbech et al. 2005), which epipsammic microbes
likely exploit more readily rather than surface water NH4
.
Given the physical and biological complexity of microbial
biofilms (Singer et al. 2010), other potential mechanisms
may also affect nutrient use within these communities, such
as internal nutrient cycling. Further studies on the diversity
of specific taxonomic groups may provide clearer evidence
for a link between microbial diversity and biofilm function,
while studies on the overall diversity of biofilms may be
confounded by other dynamics, such as the multifunction-
ality of microbial communities or the dominance of inactive
taxa (Jiang 2007, Peter et al. 2011).
EV-12
Metzker, M. L. 2010. Sequencing technologies – the next genera-
tion. – Nat. Rev. Genet. 11: 31–46.
Moeslund, B. et al. 1990. Danske vandplanter. – Danish Environ-
mental Protection Agency (DMU).
Mulholland, P. J. et al. 2000. Nitrogen cycling in a forest stream
determined by a N-15 tracer addition. – Ecol. Monogr. 70:
471–493.
Peter, H. et al. 2011. Multifunctionality and diversity in bacterial
biofilms. – PLoS ONE 6: e23225.
Piggott, J. J. et al. 2012. Multiple stressors in agricultural streams:
a mesocosm study of interactions among raised water tem-
perature, sediment addition and nutrient enrichment. – PLoS
ONE 7: e49873.
Quast, C. et al. 2013. e SILVA ribosomal RNA gene database
project: improved data processing and web-based tools.
– Nucleic Acids Res. 41: D590–596.
Revsbech, N. P. et al. 2005. Nitrogen transformations in microen-
vironments of river beds and riparian zones. – Ecol. Eng. 24:
447–455.
Riis, T. et al 2001. Plant distribution and abundance in relation to
physical conditions and location within Danish stream sys-
tems. – Hydrobiologia 448: 217–228.
Schloss, P. D. et al. 2009. Introducing mothur: open–source,
platform–independent, community–supported software for
describing and comparing microbial communities. – Appl.
Environ. Microbiol. 75: 7537–41.
Singer, G. et al. 2010. Physical heterogeneity increases biofilm
resource use and its molecular diversity in stream mesocosms.
– PLoS One 5: e9988.
Sterner, R. et al. 1992. Stoichiometric relationships among
producers, consumers and nutrient cycling in pelagic ecosys-
tems. – Biogeochemistry 17: 49–67.
Stevenson, R. J. and Rollins, S. L. 2007. Ecological assessments
with benthic algae. – In: Hauer, F. R. and Lamberti, G. A.
(eds), Methods in stream ecology, 2nd edn. Elsevier, pp. 895.
Sutherland, I. 2001. e biofilm matrix – an immobilized but
dynamic microbial environment. – Trends Microbiol. 9:
222–227.
Tatariw, C. et al. 2013. Denitrification in a large river: considera-
tion of geomorphic controls on microbial activity and
community structure. – Ecology 94: 2249–2262.
Wahl, M. et al. 2012. e second skin: ecological role of epibiotic
biofilms on marine organisms. – Front. Microbiol. 3: 292.
Wang, Y. and Qian, P.–Y. 2009. Conservative fragments in
bacterial 16S rRNA genes and primer design for 16S ribosomal
DNA amplicons in metagenomic studies. – PLoS ONE 4:
e7401.
Yavitt, J. B. et al. 2011. Methanogen diversity and community
composition in peatlands of the central to northern Appala-
chian Mountain region, North America. – Biogeochemistry
109: 117–131.
Zar, J. H. 2009. Biostatistical analysis. – Prentice-Hall.
Zhu, F. et al. 2005. Mapping of picoeucaryotes in marine ecosystems
with quantitative PCR of the 18S rRNA gene. – FEMS Micro-
biol. Ecol. 52: 79–92.
Dodds, W. K. and Biggs, B. J. F. 2002. Water velocity attenua-
tion by stream periphyton and macrophytes in relation to
growth form and architecture. – J. N. Am. Benthol. Soc. 21:
2–15.
Edgar, R. C. 2013. UPARSE: highly accurate OTU sequences from
microbial amplicon reads. – Nat. Meth. 10: 996–998.
Eriksson, B. K. et al. 2006. Biotic habitat complexity controls
species diversity and nutrient effects on net biomass production.
– Ecology 87: 246–254.
Fierer, N. et al. 2007. Environmental controls on the landscape-
scale biogeography of stream bacterial communities. – Ecology
88: 2162–2173.
Filstrup, C. T. et al. 2014. Cyanobacteria dominance influences
resource use efficiency and community turnover in
phytoplankton and zooplankton communities. – Ecol. Lett.
17: 464–474.
Finlay, B. J. et al. 1997. Microbial diversity and ecosystem function.
– Oikos 80: 209–213.
Foesel, B. U. et al. 2008. Nitrosomonas Nm143–like ammonia
oxidizers and Nitrospira marina–like nitrite oxidizers dominate
the nitrifier community in a marine aquaculture biofilm.
– FEMS Microbiol. Ecol. 63: 192–204.
Gotelli, N. J. and Colwell, R. K. 2001. Quantifying biodiversity:
procedures and pitfalls in the measurement and comparison of
species richness. – Ecol. Lett. 4: 379–391.
Grossart, H.-P. et al. 2013. Molecular and functional ecology of
aquatic microbial symbionts. – Front. Microbiol. 4: 59.
He, D. et al. 2014. Contrasting diversity of epibiotic bacteria and
surrounding bacterioplankton of a common submerged
macrophyte, Potamogeton crispus, in freshwater lakes. – FEMS
Microbiol. Ecol. 90: 551–62.
Heino, J. et al. 2010. Geographical patterns of micro-organismal
community structure: are diatoms ubiquitously distributed
across boreal streams? – Oikos 119: 129–137.
Hempel, M. et al. 2010. Community composition of bacterial
biofilms on two submerged macrophytes and an artificial
substrate in a pre-alpine Lake. – Aquat. Microb. Ecol. 58:
79–94.
Hooper, D. U. et al. 2005. Effects of biodiversity on ecosystem
functioning: a consensus of current knowledge. – Ecol.
Monogr. 75: 3–35.
Jiang, L. 2007. Negative selection effecets suppress relationships
between bacterial diversity and ecosystem functioning.
– Ecology 88: 1075–1085.
Kofoed, M. V. W. et al. 2012. Higher nitrate-reducer diversity in
macrophyte-colonized compared to unvegetated freshwater
sediment. – Syst. Appl. Microbiol. 35: 465–72.
Lear, G. et al. 2014. Fine-scale spatial patterns in bacterial
community composition and function within freshwater
ponds. – ISME J. 8: 1715–26.
Lee, J. E. et al. 2013. Both species sorting and neutral processes
drive assembly of bacterial communities in aquatic micro-
cosms. – FEMS Microbiol. Ecol. 86: 288–302.
Levi, P. S. et al. 2015. Macrophyte complexity controls nutrient
uptake in lowland streams. – Ecosystems 18: 914–931.
Supplementary material (available online as Appendix oik-
03400 at < www.oikosjournal.org/appendix/oik-03400 >).
Appendix 1–5.
