Vol. 78: 25–37, 2016
doi: 10.3354/ame01798
AQUATIC MICROBIAL ECOLOGY
Aquat Microb Ecol
Published online November 17
OPEN
ACCESS
Molecular investigation of protistan diversity along
an elevation transect of alpine lakes
Lars Grossmann1,*, Manfred Jensen1, Ram Vinay Pandey2, Steffen Jost1,
David Bass3, 4, Roland Psenner5, Jens Boenigk1
1
Biodiversity Department and Centre for Water and Environmental Research, University of Duisburg-Essen,
45141 Essen, Germany
2
Institute of Population Genetics, Veterinaerplatz 1, 1210 Vienna, Austria
Department of Life Sciences, Natural History Museum, London SW7 5BD, UK
4
Centre for Environment, Fisheries and Aquaculture Science (Cefas), Barrack Road, The Nothe, Weymouth,
Dorset DT4 8UB, UK
5
Institute of Ecology, University of Innsbruck, Technikerstrasse 25, 6020 Innsbruck, Austria
3
ABSTRACT: Ecological gradients are one of the main reasons for changes in biodiversity. However, whether what is known to be true for plants and animals also applies to protists in the same
way is largely unknown. Along an alpine elevation gradient, including 29 freshwater lakes, we
investigated ecological drivers of protistan diversity and community structure, as well as the
explanatory value of the species−area relationship for protistan diversity using a deep-sequencing
approach. We found that species richness does not decrease with elevation, nor is elevation a
major factor explaining the observed shifts in community composition. The observed protistan
communities differ depending on many factors, with pH and nutrient concentrations being most
important. Considering distinct subgroups, Chrysophyceae accurately reflect the pattern of the
overall protistan community. Within the investigated elevation gradient, species richness was
correlated with the area of individual lakes irrespective of their elevation. However, we found no
support for decreasing species richness with elevation, as known for plants and animals.
KEY WORDS: Eukaryotic microbial diversity · High throughput sequencing · Algae · Protozoa ·
Chrysophytes · Evenness · Dinophyta
INTRODUCTION
Biodiversity is modulated by ecological and geographical gradients (Hewitt 2000, Adams 2009,
Boenigk & Wodniok 2014). Factors such as temperature, nutrient availability or pH shape the species
inventory of both terrestrial and aquatic habitats.
This holds true for multicellular organisms in general
(metazoans, embryophytes etc.) (Gaston 2000, Frey &
Lösch 2010), as well as for protists concerning specific factors such as nutrient availability (Dziock et al.
2006). Furthermore, the species−area relationship
*Corresponding author: [email protected]
(e.g. as observed in island biogeography) predicts a
decreasing number of species with decreasing area
(Whittaker et al. 2001), i.e. as fewer habitats (and
thus fewer niches) are covered.
Environmental parameters, such as UV radiation,
local climate and land use, change with elevation
and potentially shape communities. Community
composition is known to change accordingly, as
shown for plants by Frey & Lösch (2010). In mountain
areas, habitat size generally decreases with elevation, leading to a shift in community composition and
to a lower species richness due to reduced space at
© The authors 2016. Open Access under Creative Commons by
Attribution Licence. Use, distribution and reproduction are unrestricted. Authors and original publication must be credited.
Publisher: Inter-Research · www.int-res.com
26
Aquat Microb Ecol 78: 25–37, 2016
higher elevation (Romdal & Grytnes 2007, N oguésBravo et al. 2008), as predicted by species−area-relationship theories. For higher plants and animals, the
assumption of a systematic decrease of species richness with elevation is not unquestioned (Rahbek
1995), though widely accepted. However, whether
and to what extent this relationship holds true for
protists in general or specific protistan taxonomic
groups remains unclear (Bryant et al. 2008).
Protistan community composition in freshwater
lakes is diverse and shows a strong seasonality due to
species-specific ecological optima (Häder et al. 1998,
Nolte et al. 2010), nutrient availability (Graham et al.
2009) and specific biotic interactions (Sommer et al.
2012, De Senerpont Domis et al. 2013), with the
majority of the year’s total community in dormant
stages whenever sampled. For lakes in mountainous
regions, especially high mountain lakes, mesotrophic
or oligotrophic conditions usually apply, favouring
protistan taxa better adapted to low nutrient concentrations, such as chrysophytes (Tolotti et al. 2003).
Furthermore, high mountain lakes are often comparably small, shallow and UV transparent, leading to
specific adaptations of the inhabiting organisms, e.g.
to UV-light stress (Sommaruga 2001, Sonntag et al.
2011). The remote position of these lakes may favour
high degrees of endemicity (Sommaruga 2001, Sonntag et al. 2011). However, protistan diversity along
alpine gradients (as generally protistan biogeography) is understudied, leaving the communities’ variability unrevealed (Weisse 2008). This scarceness of
cross-habitat comparisons hinders our understanding
of protistan diversity and distribution (Tringe et al.
2005, Bik et al. 2012). Deep-sequencing technologies
can create a fuller picture of protistan diversity, as
samples can be analysed both in parallel and in
depth, leading to quantitatively and qualitatively
enlarged taxa lists per sample (Triadó-Margarit &
Casamayor 2012, Bates et al. 2013). In numerous
cases, high-throughput sequencing has shown that
protistan diversity has been highly underestimated
in the past (N olte et al. 2010, Stoeck et al. 2010,
Lecroq et al. 2011). Moreover, such deeper diversity
surveys also hold the potential to reveal protistan
diversity and distribution patterns among sites and
samples that could not be seen so far (Grossmann et
al. 2016).
To analyse shifts in community composition, as
well as in species richness with elevation, we used
deep-sequencing data of 29 freshwater lakes within
an alpine transect. On the basis of such an enlarged
dataset, we intended to test whether elevation and
the reduced available area in higher elevations influ-
ence protistan community composition and species
richness as it does with plants and animals. This has
been neither verified nor falsified for protists thus far.
The species−area relationship implies that species
richness is reduced with elevation due to the smaller
total area available. However, other factors changing
with elevation imply that protistan species richness is
actually augmented: environmental gradients such
as UV irradiation are comparatively strong in high
mountain lakes and, in concert with the special climatic, geological and ecological conditions in mountain ranges, may trigger a diversification of habitats
and niches. The habitat diversification on the microscale may in turn cause a comparatively high diversity within a habitat. Furthermore, habitats at higher
elevations (in mountain ranges and on individual
peaks) are often disconnected, a fact that supports
endemicity, whereby lakes are disconnected in a
double sense in mountain areas as they already form
islands of water within land no matter where they are
situated. Diversity across habitats may, therefore,
also be augmented due to an increasing share of
endemism.
We therefore hypothesise that species richness of
protistan communities, as indicated from operational
taxonomic unit (OTU) numbers, may correlate with
individual lake size, but may not follow the classical
elevation-dependent species−area relationship, i.e. it
does not decrease with elevation but is rather shaped
by factors, or a combination of factors, independent
of elevation. We further hypothesise that community
composition of protists (applying to the complete
community, as well as to taxonomic subgroups) does
change with elevation and co-varying factors, as it
generally does along gradients of other environmental factors. We expect, however, that the community
composition at the level of higher taxonomic groups
is more stable and less affected by the elevation gradient than the composition at the OTU level, as fluctuations within taxonomic groups would not affect
the overall community composition and the environmental effects should, therefore, be damped for
higher taxonomic levels.
MATERIALS AND METHODS
Sampling and sample preparation
To compare organisms on a broad basis, we chose
a 454 deep-sequencing amplicon approach, targeting the V9 region of the 18S small subunit rRN A
(SSU rRNA). Based on the sequenced OTUs and their
Grossmann et al.: Protistan diversity in alpine lakes
reads, we compared species richness and community
structure and calculated the similarity of the different
sites sampled. We furthermore checked for discriminating environmental features that could explain the
similarity or dissimilarity of sites.
Samples were all collected in mid-August 2007
(calendar week 33−34), to ensure comparability of
community and diversity data. The 32 lakes sampled
(29 lakes analysed) form an alpine transect extending over several mountain ranges from the Salzkammergut area to the Low Tauern (Austria) and cover
an elevation gradient from 429 to 2072 m above sea
level (a.s.l.), and are of diverse water chemistry
(Table S1 in Supplement 1 at www.int-res.com/
articles/suppl/a078p025_supp1.xlsx). To assure comparability, all samples were treated equally in the different processing steps, including the sequencing
protocol. Samples were taken with a telescopic sampling device ~3 m from the shoreline of the lake, filtered on 0.2 µm polycarbonate filters (20−200 ml),
air-dried and stored at −80°C.
Water temperature, pH and conductivity were
measured in the field. Alkalinity was determined
by Gran titration with an Orion 960 pH electrode.
Total phosphorous (TP) was measured using the
molybdate method according to Vogler (1966). Dissolved organic carbon (DOC) was measured with a
Shimadzu TOC-V CPH (total organic carbon analyser) and dissolved nitrogen (DN ) was measured
with a Shimadzu TN M-1 (total nitrogen measuring
unit). The anions chloride (Cl), sulfate (SO4) and
nitrate (NO3-N) as well as the cations sodium (Na),
ammonium (N H4-N ), potassium (K), magnesium
(Mg) and calcium (Ca) were measured by means of
ion chromatography (Dionex ICS-1100). Dissolved
reactive silicon (DRSi) was measured using the
molybdate method (Smith & Milne 1981, Skalar,
SAN plus segmented flow analyser). Total phosphorous and alkalinity were measured from raw
water, and all other chemical parameters from filtrate (0.6 µm).
Genomic DNA was extracted with the DNeasy tissue kit (Qiagen). Filters were transferred to a 2 ml
tube and incubated for 1−3 h in buffer ATL of the
DN easy tissue kit supplemented with Proteinase K.
DN A was subsequently extracted following the instructions of the supplier. We used HPLC-purified
PCR primers, which carry sequences specific for the
gene of the SSU of the rRNA: forward primer 1391F
(Lane 1991, Stoeck et al. 2010) ‘ATT AGG GTT CGA
TTC CGG AGA GG’ and reverse primer Euk B
(Medlin et al. 1988, Stoeck et al. 2010) ‘CTG GAA
TTA CCG CGG STG CTG’.
27
PCR and pyrosequencing
PCR amplifications of DNA were conducted using
primers carrying a 5’-tail for the 454 sequencing
(Medinger et al. 2010). The final concentrations in all
of the PCR reactions were 1 µl of DN A template in
20 µl PCR reaction with 0.4 units of Phusion polymerase, primers at 0.25 µM final concentration, and
dNTPs at 0.2 mM final concentration, including 4 µl
Phusion buffer and 12.2 µl water. The PCR conditions
consisted of an initial denaturation at 94°C for 4 min
and 35 cycles of the following: 30 s at 95°C, annealing for 60 s at 60°C, elongation for 2 min at 72°C,
followed by a final extension step of 10 min at 72°C.
Pyrosequencing was carried out using the 454 Genome Sequencer FLX System and titanium chemistry
(454 Life Sciences).
Bioinformatics
CANGS software (Nolte et al. 2010, Pandey et al.
2010) was used for adapter and primer clipping as
well as for quality filtering. We performed a pattern
search that allowed for mutations in homopolymers
within the adapter B sequence during trimming, but
required the PCR primer sequences to be mutation
free, otherwise the read was discarded. Reads for
which an indel was identified at the transition between the read and PCR primer were removed. Quality filtering included the removal of all sequences
that did not fit the following criteria: (1) no N s; (2)
quality score > 24, when averaged across the read
after clipping adapters and primers; (3) minimum
sequence length of 200 base pairs (including PCR
primers); and (4) at least 2 copies of the read present
in the entire dataset before clipping primers. We
blasted all non-redundant sequences against the
GeneBank database (RefSeq Release 57, 14 January
2013) using BLASTn version 2.2.25+ (Altschul et al.
1990) and retrieved the taxonomic classification of
the best hit.
Analysis of the sequence dataset
As a higher taxonomic level of analysis, protistan
metagroups (displayed and listed in Fig. 1) were chosen to fully reflect protistan biodiversity as well as to
meaningfully distinguish the different phylogenetic
lineages within the tree of life. Metazoa and embryophytes as well as Bacteria and Archaea were
excluded beforehand to diminish the bias within
Aquat Microb Ecol 78: 25–37, 2016
28
Clusters from Ward‘s-clustering
Cluster 2
LO234
OT33
LO134
EG33
WA33
IM33
WI33
FU33
WO33
IR33
MO33
AL33
HA33
Cluster 3
Cluster 1
200
150
100
LO234
LO134
SC33
GF33
NU33
MI33
SO33
AL33
IM33
AU33
BA33
WI33
WA33
MO33
PR33
KR33
OT33
EG33
FU33
OE33
IR33
WO33
OS33
HA33
BD33
US33
OB33
RI33
50
HT33
Sites ordered by clusters and species richness
Others
H1.Haptophyceae
G2.Katablepharidoph
G1.Cryptophyta
F3.Viridipl..no.Embr
F2.Rhodophyta
E8.Apusozoa
E7.Amoebozoa
E6.Glomeromycota
E5.Basidiomycota
E4.Ascomycota
E3.Microsporidia
E2.Chytridiomycota
E1.Choanoflagellida
D5.Excavata.rest
D4.Euglenozoa.rest
D3.Kinetoplastida
D2.Euglenida
D1.Heterolobosea
C2.Rhizaria.rest
C1.Cercozoa
B6.Stramenopiles.rest
B5.Bacillariophyta
B4.Synurophyceae
B3.Chrysophyceae
B2.Oomycetes
B1.Bicosoecida
A4.Alveolata.rest
A3.Apicomplexa
A2.Dinophyceae
A1.Ciliophora
OE33
NU33
MI33
KR33
BA33
PR33
SC33
AU33
GF33
SO33
OS33
RI33
US33
HT33
OB33
BD33
Cluster 3
Cluster 2
Cluster 1
Rarefacted no. of OTUs
Fig. 1. Clusters from Ward‘s clustering (top) and species richness and OTU composition of taxonomic groups (bottom) (rarefaction level = 1000). Bars (bottom) are ordered by clusters from cluster analysis and by species richness (left to right). Species
richness is shown as height of bars in rarefacted number of OTUs; community composition is reflected in the colour of bars by
identified taxonomic groups (see key). Alveolata.rest are all Alveolata not specified in the key, i.e. all Alveolata excluding Apicomplexa, Dinophyceae, and Ciliophora; likewise for the other ‘rest groups’ in the key. ‘Viridipl..no.Embr’ are all Viridiplantae
excluding embryophytes
Grossmann et al.: Protistan diversity in alpine lakes
samples and focus on eukaryotic microbial life. OTUs
affiliated with other protistan metagroups were summarised as ‘others’; sequence reads that could not be
affiliated with any of the metagroups were treated as
‘unknown’ (not shown in graphs).
Statistics software
For statistical analyses, we used the R computer
language (v 3.2.3), Rstudio (v 0.99.892), R packages
‘vegan’ (v 2.3-4; Oksanen et al. 2016), ‘ape’ (v 3.4;
Paradis et al. 2015) and ‘gclus’ (v 1.3.1; Hurley 2012).
We generally followed the recommendations of Borcard et al. (2011), and adapted R scripts provided by
these authors.
Basic statistics
Using Pearson correlations with and without ztransformed data tables, we correlated single, relevant environmental variables with species richness
and with each other to check for a possible, highly
pronounced effect of these variables. We also compared the 3 formed clusters on the basis of metagroup percentages. Here, we tested, with the Dunnett-T3 post-hoc test, for taxon percentages which
were significantly different among the 3 clusters, i.e.
p-value < 0.05.
Data tables, standardisation and
pre-transformations
Two data tables were investigated: data table Y
containing abundances of OTUs as produced by the
CANGS pipeline (Pandey et al. 2010), and an additional data table X containing chemical and other
environmental variables of the same 29 sites (lakes).
Before further analyses, the abundance data in
table Y were double standardised due to the specific
nature of the data matrix (i.e. > 90% zeros and single
counts). The data were first transformed by the function drarefy (vegan package; rarefying without loss
of information), then used to provide species richness at the rarefaction level of 1000 (Fig. 1) (species
richness being calculated by the vegan function:
rarefy(raw_sequencedata,sample=1000)). In a
further step, these data were Hellinger-transformed
(vegan package: function decostand(Y,"hel")),
producing the table Yhel. Hellinger transformation is
known to allow ordination of sites by normal princi-
29
pal component analysis (PCA) and redundancy
analysis (RDA) with abundance data (Legendre &
Gallagher 2001).
All chemical raw data of table X (besides pH) as
well as area were log-transformed by the R function
log1p. The other data (temperature, elevation, turbidity) were not log-transformed. For the ratios TP/DN
and DN/DOC, the log-pre-transformed variables DN,
TP and DOC were used.
Standardisation of all explanatory variables of X
(z-transformation) was automatically performed within
PCA calls with scale=TRUE.
Cluster analysis, PCA, RDA, NMDS and pRDA
Besides PCA and RDA, non-metric multidimensional
scaling (N MDS) was used with the same doublestandardised Yhel (function metaMDS in the vegan
package; plot not shown in the results).
Multidimensional Ward’s cluster analysis of
double-standardised table Y was performed by Ych
= vegdist(Yhel,"euc") and hclust(Ych,method
= "ward.D2") of the vegan package. Additionally,
the function hcoplot was used (Borcard et al. 2011)
to produce a dendrogram of the cluster analysis.
PCA and RDA (as well as NMDS) were completed
with the results of cluster analysis (3 clusters marked
by different colours and symbols), mainly as a control
for the impression of visible clusters within the ordination of sites.
PCA ordination of sites, either by X or by Yhel,
delivered similar results, and both were completed
with the cluster analysis results (see above).
PCA of Yhel was complemented by vegan’s
envfit approach. Although only passive and post
hoc, this approach conveys a first impression of the
influence of X variables. Non-significant X variables
can be identified via a permutation test. For this purpose, we used envfit with permutations=100 000
and p.max=0.0001.
For RDA, the model permutation test was conducted using anova(rda(Yhel,X),step =1000). To determine the most important explanatory variables,
vegan’s functions ordistep and ordiR2step were
used with input parameters: Pin=0.01 and Pout=0.1,
direction="both", permutations = how(nperm =
999). The 2 methods delivered identical results.
Partial RDAs (pRDAs) with Yhel and subsets of X
were also performed, taking into account the important variables identified by the ordistep procedures. Using pRDA, a possible influence of the less
important variables can be judged.
Aquat Microb Ecol 78: 25–37, 2016
30
RESULTS
Three of the sampled lakes (Ahornsee, Bad Heratingersee and Holzoestersee — all not shown in the
figures and tables) were represented by too few
sequences after bioinformatic quality filtering and
were discarded from further analyses. The remaining
29 lakes formed 3 clusters in the cluster analysis
(Ward’s clustering) on the basis of OTUs: two smaller
clusters of 2 and 6 lakes (subsequently Cluster 1 and
Cluster 2, respectively), and one larger cluster of 21
lakes (subsequently Cluster 3) (Fig. 1). These clusters
from Ward’s clustering are also shown in Figs. 2−4 in
different colours, complementing the information
specific to these figures. In Fig. 2, clusters are shown
in geographic maps, and in Figs. 3 and 4 in graphs of
combined cluster analysis and PCA. The geographical mapping of Fig. 2 reveals that the sampling sites
affiliated with the 3 clusters are mostly situated in
areas differing in geology. Lakes in Cluster 2 (6 sites)
are all situated in siliceous bedrock; with 2 exceptions, lakes in Cluster 3 (21 sites) are situated in calcareous bedrock. This geological sorting generally
corresponds with the measured pH values of the
lakes. The 2 bog ponds forming Cluster 1 had a low
pH despite being situated in a calcareous area, and
must thus be interpreted as special cases. The lakes
of the 2 smaller clusters (2 and 1) are both situated in
geographically restricted areas.
Taxon richness
Taxon richness in the sampled lakes varied between 80 and 270 different OTUs. Richness did not
correlate with elevation (Pearson correlation: 0.0078);
it was, however, positively correlated with the logarithm of lake surface area (Pearson correlation:
0.5196; p-value = 0.0038) and negatively correlated
with TP (Pearson correlation: −0.6028; p-value =
0.0005). In a log-log space, richness and lake surface
area had a Pearson correlation value of 0.475
(p-value = 0.00921). When leaving out the outlier
Schwarzensee, the correlation value changed to
0.625 (p-value = 0.00017). Lake surface area and TP
were not correlated in our dataset (Pearson correlation: −0.283), nor were elevation and TP (Pearson correlation: −0.1727). In partial correlation analysis, the
Pearson correlation value of richness (log) and area
(log) changed from 0.475 to 0.398 when TP was
eliminated as a factor, and the correlation value of
richness (log) and TP (log) changed from −0.584 to
−0.532 when area was eliminated. Furthermore, rich-
ness was strongly correlated with Pielou’s evenness
(Pearson correlation: 0.824; p-value < 0.0001).
Restricted to a subset of lakes >1000 m, evenness was
also negatively correlated with TP (Pearson correlation: −0.636; p-value = 0.0184). The 2 lakes affiliated
with Cluster 1 both had a comparatively low richness. The largest cluster (Cluster 3) showed a broad
variation of richness values (Fig. 1), i.e. especially a
broad variation of richness values in smaller lakes.
Thus, beyond the general trend of richness being
positively correlated with lake size, small lakes situated in higher elevation showed a comparatively high
richness (for richness and evenness values see Table S2
in Supplement 1 at www.int-res.com/articles/suppl/
a078p025_supp1.xlsx.
Community composition and OTU distribution
Community composition analysis revealed typical
freshwater communities of mesotrophic to oligotrophic lakes with a high relative abundance of OTUs of
Ciliophora (23.9%), Chrysophyceae (14.6%), Dinophyceae (13.5%), other Stramenopiles (Stramenopiles.rest) (9.6%), other Alveolates (Alveolata.rest)
(5.6%), Synurophyceae (5.1%) and Cryptophyta
(4.6%) for all sites. Alveolates showed generally high
read numbers due to high gene copy numbers (Fig. 1).
Generally, differences between communities (calculated on OTU basis) were high, ranging from BrayCurtis distances of 0.5 to 0.99 (usually higher than
0.8), mainly due to the high number of singleton sequences present in 1 site only. Differences between
sites within clusters were generally lower than differences between sites among clusters (Table S3 in
Supplement 1). OTU distribution was not strongly affected by elevation for those OTUs that occurred in
more than one site. Tested in a direct comparison of
OTUs that occur at more than 3 sites (1670 out of
4612 total), only 20.24% were restricted either to elevations below 700 m or to elevations above 1300 m,
whereas 62.22% of OTUs occurred in both the highest and lowest lakes sampled (Fig. S1 in Supplement 2 at www.int-res.com/articles/suppl/a078p025_
supp2.pdf).
Community composition on metagroup level
At the higher taxonomic group level, the dissimilarity between sites does not appear accordingly.
Smaller differences here include common trends of
single groups within Cluster 2 compared with the
47.2°
47.4°
47.6°
47.8°
48.0°N
13.0°E
503
OT33
505
13.2°
LO234
1302
663
484
MO33
KR33
577
13.4°
LO134
1296 WO33
538
553
WA33
FU33
IR33
514
IM33
716
SC33
624
EG33
13.6°
1699
508
HA33
WI33
604
NU33
MI33
Cluster 3
Cluster 2
Cluster 1
1514
PR33
13.8°
1639
1428
BA33
694
BD33
HT33 1157
1519
OB33
RI33
1672
1338
US33 OS33
1962 2062
779
OE33
AL33
712
AU33
1643
GF33
856
SO33
Geography (in elevation) of sites
47.2°
47.4°
47.6°
47.8°
48.0°N
13.0°E
8.22
OT33
8.6
FU33
13.4°
KR33
8.38
8.5 MO33
8.54
IR33
EG33
13.6°
7.56
8.35
HA33
WI33
7.62
8.44
SC33
LO134
LO234 5.18
5.24
WO33 8.51 NU33
8.44
8.23
13.2°
WA33
8.3
IM33
Geology of sites
7.61
AL33
7.7
OE33
13.8°
7.4
PR33
MI33
GF33
BA33 8.04
8.61
7.77
BD33
HT33
6.79
6.55
OB33
6.56
RI33
6.64
US33 OS33
6.47
6.29
8.07
SO33
AU33
8.1
Cluster 3
Cluster 2
Cluster 1
Fig. 2. Position of sites within the sampling area and cluster denomination by
Ward’s clustering (= 3 clusters). The
black box in the top panel shows the
location of the sites within Austria. Site
coordinates are plotted on an elevation
profile + elevation of individual lakes in
numbers (left), and on a geological profile + pH of individual lakes in numbers
(right). The main geological formations
shown are (from top to bottom):
molasse (pale brown), Rhenodanubian
Flysch (light green), northern limestone Alps (blue) and Austroalpine
crystalline (brown colours)
Grossmann et al.: Protistan diversity in alpine lakes
31
Aquat Microb Ecol 78: 25–37, 2016
32
MO33
large Cluster 3 (tested as cluster versus cluster with the Dunnett-T3 posthoc test); for example, a relatively low
fraction of Cryptophyta in Cluster 2
(significant p-value of < 0.001). The 2
lakes of Cluster 1 showed a comparably high fraction of Alveolates and
Stramenopiles (together 83.1%, significant p-values: Ciliophora = 0.016,
Synurophyceae < 0.001, Stramenopiles.rest < 0.001) and a correspondingly small fraction of all other taxonomic groups.
KR33
NO3N
WO33
area
HA33
DN.DOC
IR33 pH NU33
SO4
FU33
BA33 DRSi AL33
WI33
Cond Ca
HCO3
Mg
Cl Na
WA33
RI33
OE33
DN
MI33
PR33
K
0.0
IM33
Temp
NH4N
elevation
OT33
AU33
EG33
BD33
GF33
DOC turbidity
PC2
HT33
OB33
OS33
US33
TP
TP:DN SC33
0.5
SO33
Multivariate statistics
1.0
LO234
LO134
0.5
0.0
–0.5
PC1
–1.0
pH
0.5
KR33
Cond
WO33
MO33
IR33
FU33
HA33 OE33
BA33
NU33 AL33
WI33
HT33
RI33
US33
MI33
OS33
SC33
0.0
WA33
0
PR33
OT33
BD33
GF33
AU33
IM33
Temp
RDA2
OB33
SO33
EG33
–0.5
TP:DN
–1
–1.0
LO234
LO134
–1.0
–0.5
0.0
0.5
1.0
RDA1
Fig. 3. Principal component analysis (top) and redundancy analysis (bottom)
showing relevant environmental variables for cluster sorting (colouring of
clusters by Ward’s clustering, see Fig. 1). PCA on the basis of OTUs shows significant (blue) and non-significant (turquoise) environmental factors by environmental fitting (env.fit). In forward selection of environmental variables in
RDA, significant factors sufficient to describe the sorting of samples can be
reduced to 4 factors: conductivity (Cond), pH, TP:DN (total phosphate/
dissolved nitrogen) and temperature (Temp)
In the PCA plots on the basis of
OTUs, the 3 clusters were clearly separated, Cluster 3 (the largest cluster)
being scattered. Most of the measured environmental variables were
significant (at a significance level of
0.001) as determined by vector fitting
(Fig. 3). However, in forward selection
of variables in RDA (which reduces
variables to a necessary minimum)
conductivity, TP (or TP/DN ), pH and
temperature remained as explanatory
factors, together explaining 88.4% of
the clustering of sites (Fig. 3). In particular, pH (and covariant factors, e.g.
HCO3, N a, Ca) and TP were independent of each other in the PCA,
and therefore have a high explanatory value in combination. Cluster 1 is
characterised by low pH (around 5.2)
and low calcium levels, but high
phosphate and DOC concentrations.
Cluster 2 is likewise characterised by
lower pH (6.3−6.8) and low calcium
concentrations (around 3 µg l−1), but
also by low values of phosphate and
DOC. In contrast, Cluster 3 covers a
wide range of values, but with consistently higher pH values (7.4−8.6) and
calcium levels (14−44 µg l−1). Temperature only weakly co-varied with
elevation, and could thus not be substituted by elevation in the RDA (forward selection) calculation. N either
elevation nor lake size (applied as
log(lake surface area)) had a high
Grossmann et al.: Protistan diversity in alpine lakes
All OTUs (rlevel: 1000)
MO33
OTUs of Chrysophyceae + Synurophyceae (rlevel: 100)
MO33
FU33
IR33 NU33
HA33
AL33
BA33
WI33
OE33
RI33
MI33
HT33
OB33
0.1
OS33
WI33
GF33
US33
0.0
IM33
OT33
PC2
AU33
OB33
AU33
US33
BD33
WA33
GF33
EG33
–0.1
SC33
–0.1
HT33
RI33
OS33
IM33
BD33
EG33
MI33
IR33
BA33
PR33
WA33
0.0
FU33
OE33
WO33
AL33
KR33
HA33
NU33
PR33
KR33
WO33
0.1
33
OT33
SO33
SO33
–0.2
–0.2
–0.3
Cluster 1
Cluster 2
Cluster 3
–0.3
LO234
Cluster 1
Cluster 2
Cluster 3
LO234
LO134
LO134
–0.2
–0.1
0.0
0.1
0.2
–0.2
0.3
–0.1
Environmental variables
0.0
OB33
WI33
RI33
HT33
US33
MO33
OS33
MO33
0.0
KR33
KR33
OS33
OB33
EG33
FU33
OE33
GF33
OT33
BD33
BD33
OE33
AL33
IR33
SC33
GF33
NU33
AU33
–0.1
OT33
PR33
MI33
PR33
SO33
AU33
IM33
–0.5
MI33
BA33
0.0
WA33
PC2
RI33
IM33
US33
HT33
IR33
AL33
HA33
BA33 FU33
0.3
HA33
NU33
WO33
0.2
OTUs of Dinophyta (rlevel: 100)
0.1
0.5
0.1
SO33
SC33
–0.2
–1.0
EG33
–1.0
LO234
–0.3
Cluster 1
Cluster 2
Cluster 3
–0.5
LO134
0.0
0.5
1.0
Cluster 1
Cluster 2
Cluster 3
–0.2
LO234
LO134
–0.1
0.0
0.1
0.2
0.3
PC1
PC1
Fig. 4. Comparative principal component analyses and Ward’s clustering showing clusters of sampling sites. Clusters differ on
the basis of Ward’s clustering (indicated by colours of sites) and PCA (position of sites in the graph), the basis being as follows:
all OTUs (top left) alike in Figs. 1−3, environmental variables (bottom left), OTUs of Chrysophyceae + Synurophyceae (top
right) and OTUs of Dinophyta (bottom right). Differing clusters 2 and enlarged Cluster 1 are encircled for comparison.
OTU data for cluster analysis and PCA are Hellinger-transformed
explanatory value for cluster sorting (Fig. 3 and
Table S4 in Supplement 1 at www.int-res.com/
articles/suppl/a078p025_supp1.xlsx).
PCA (at the OTU level) supports the observed
trend from cluster analysis shown by the dissimilarity
and separated positioning of sites in the graph (Fig. 3).
In contrast, PCA at the metagroup level is not in
accordance with the cluster analysis results.
Combined cluster analysis and PCA based on environmental variables on the one hand and OTUs on
the other hand revealed almost identical PCA plots
(Fig. 4). Therefore, the applied environmental variables co-varied with the community composition; pH
(plus conductivity), TP and also temperature seem to
be good proxies for predicting protistan community
similarity among lakes in an elevation gradient. This
Aquat Microb Ecol 78: 25–37, 2016
34
also held true when the analysis was restricted to distinct taxonomic groups, namely Chrysophyceae and
Ciliophora (Ciliophora not shown in Fig. 4). The PCA
based exclusively on chrysophycean OTUs even
showed higher cluster consistency, revealing the
potential of Chrysophyceae as an indicator for the
complete protistan community in an alpine context.
In contrast, clusters were indistinct when the analysis
was restricted to Dinophyta (Fig. 4).
DISCUSSION
Ecological and biogeographical theories are largely
derived from metazoans and embryophytes; the applicability to other organisms, specifically microbial
organisms such as protists, remains unclear. Species−area-relationship theories predict lower species
numbers in smaller habitats. In an elevation gradient,
land area as well as the number of bodies of fresh
water and their mean sizes decrease gradually when
ascending to mountain peaks. Likewise, other factors, such as temperature, generally change alongside an elevation gradient. Here, we investigated
alpine freshwater lakes covering an elevation gradient of about 1500 m (429 m to 2072 m a.s.l.) and differing in ecological parameters for patterns of protistan biodiversity. This is a gradient covering lakes up
to the level of alpine mountain lakes and allowing for
a comparison of lowland and alpine lakes including
all intermediate elevations.
Species richness and elevation
Based on our analyses, we can reject a general reduction of diversity (in terms of species richness) of
freshwater protists with increasing elevation. On the
contrary, individual lakes at higher elevation showed
a comparatively high diversity even though these
lakes are generally small and were thus expected to
harbour a low diversity.
In accordance with our findings, Wang et al. (2011)
reported an increase in bacterial diversity with
elevation. However, the protists analysed in the latter
study, i.e. only diatoms, showed a decreasing richness with higher elevations. The deviating pattern
between the study of Wang et al. (2011) and our
results is presumably due to the different habitat
types, i.e. lotic waters in the study of Wang et al.
(2011) versus lentic systems in our study. In the continuum of a mountain stream, protist species richness
may accumulate during downstream travel of the
water resulting in the observed correlation. In contrast, in lentic waters, species richness is presumably
mostly dependent on local habitat characteristics
rather than the exchange or transport between sites.
The species−area relationship can be confirmed
with respect to individual lake size. In contrast, we
cannot confirm a species−area relationship with respect to overall habitat availability: despite the decreasing number and the decreasing mean size of
lakes with elevation, i.e. a decreasing overall lake
area (and volume) with elevation, protistan species
richness did not decrease with elevation.
Our results, therefore, demonstrate that individual
habitat size matters for protist species richness as
predicted from species−area-relationship theories.
Protists seem to follow the general rule of higher species numbers in larger bodies of water. However,
considering lakes of comparable size, species richness in relatively small mountain lakes of higher elevation is comparatively high, and must, therefore, be
upheld by other factors than habitat size. The special
climatic, geological and ecological conditions in
mountain ranges might produce diversification of
habitats and niches, at least for single-celled organisms. Thus, in terms of the scaling ratio of organism
and habitat size, a small mountain lake can hold
hardly any niches for larger organisms but many for
single-celled protists. The habitat diversification on
the microscale in combination with the special habitat characteristics of high mountain lakes and their
relative remoteness may explain the comparatively
high diversity and at the same time a high degree of
endemism (Sommaruga 2001, Sonntag et al. 2011).
Furthermore, species richness was negatively correlated with lake trophy in our study, as was species
evenness for lakes higher than 1000 m (only smaller
lakes in the dataset). N utrient scarceness has been
shown to favour high species evenness (Hillebrand et
al. 2007, Soares et al. 2013) and is likely to play a role
in the observed evenness values in lakes of higher
elevation. Species-richness values that deviate from
the general correlation of richness with lake size, as
in the case of the observed smaller lakes at higher
elevations, have to be kept up by factors such as UV
radiation or endemism (Sommaruga 2001, Sonntag et
al. 2011), which have not yet been investigated sufficiently. So, whereas species richness of animals and
plants decreases with elevation, protistan species
richness does not follow this rule (as shown in our
study), but is rather influenced by multiple factors.
The reasons for this deviation from the elevation patterns known for plants and animals have to be investigated in more detail in subsequent studies.
Grossmann et al.: Protistan diversity in alpine lakes
Community composition and elevation
In our study, community composition was dependent on multiple factors: elevation and co-varying
factors explained some, even though a minor part, of
the variance. Only 20% of the OTUs occurring at 3 or
more sites were restricted to either high or low
elevation. The observed changes in community composition were thus mainly independent of elevation.
Also, cluster analysis resulted in clusters of lakes not
separated by elevation levels, but strongly deviating
in, for example, pH and nutrient levels. However,
elevation generally co-varies with nutrient concentrations, i.e. lakes at a higher elevation contain lower
nutrient concentrations. Community shifts associated
with nutrient concentration have also been demonstrated at higher taxonomic levels: Jeppesen et al.
(2005), for instance, reported a community shift
towards diatoms, chrysophyceans, cryptophytes and
dinophytes with decreasing nutrient load in freshwater lakes. We did not observe such decisive
changes in protistan community structure at the
metagroup level, probably due to comparably low
nutrient concentrations (oligotrophic to mesotrophic)
in the sampled lakes, and due to a weak correlation
of elevation and nutrients. At the level of OTUs, differences in nutrient level were, however, large
enough to result in clear changes in community
composition.
Besides nutrient level, pH and conductivity, temperature was also significant in accounting for variation in protistan community composition. However,
it was not directly connected to the elevation gradient. This is in contradiction to what Wang et al.
(2012) found for protists (only diatoms in this case)
in a mountain stream along an elevation gradient.
In their study, elevation was the main explanatory
factor for species turnover, mainly because diatom
diversity decreases with lower nutrient levels at
higher elevations and other protistan groups dominate such nutrient poor habitats. The influence of
temperature in our dataset is not in accordance with
studies on embryophytes and metazoa, for which
temperature clearly correlates with the elevation
gradient and distribution patterns in mountainous
regions on a large scale (Livingstone & Dokulil
2001). The multifactorial drivers of protistan community composition in our study most likely reflect a
combination of respective local habitat characteristics of the different sites. This is in accordance with
other studies of protistan mountain lake communities (Tolotti et al. 2003, Triadó-Margarit & Casamayor 2012) as well as with studies on bacterial dis-
35
tribution patterns in lakes at high elevations (Sommaruga & Casamayor 2009).
Organismic level of investigation
The comparative, combined cluster analysis and
PCA (on the basis of environmental variables as well
as on the basis of OTUs), as carried out with our dataset, further underlines the connection of multifactorial ecological conditions and changing community
composition at the level of OTUs. A resolution at the
metagroup level can reflect overarching ecological
trends among habitats, as, for example, in comparisons between soil and freshwater ecosystems
(Grossmann et al. 2016). For a more precise and profound resolution of similar habitats, an analysis at the
OTU level was necessary. A result of this analysis
was that Chrysophyceae proved to be promising
candidates for most precisely reflecting the total
community within an elevation gradient and qualifying as bio-indicators in an alpine context. Owing to
their sensitivity to environmental changes, such as
climate, pH, nutrient concentrations, seasonality
(Lotter et al. 1997, Pla & Catalan 2005) and stressors
such as heavy-metal pollution (Kamenik et al. 2005),
Chrysophyceae are an established tool in palaeolimnological studies (Facher & Schmidt 1996). Based on
our findings regarding the diversity of alpine freshwater protists, we propose to draw on Chrysophyceae as a model taxon in studies of current diversity as well.
CONCLUSIONS
As our investigation of protistan communities in an
alpine elevation gradient implies, protistan species
richness does not decrease with increasing elevation,
nor are elevation and co-varying factors the main
drivers of protistan community composition in an
alpine context. Protistan diversity patterns might
thus be quite different from those of higher organisms, and allow for the testing of theories derived
from the distribution and species diversity of plants
and animals. We anticipate that such comparisons
could possibly reveal patterns yet unknown, but
decisive for single-celled organisms.
Acknowledgements. We thank Anneliese Wiedlroither, Karin
Pfandl, Barabara Findenig and Peter Stadler for their help in
sampling, and we thank the DFG for financial support (DFG
project BO 3245/3-1 and BO 3245/2-1).
Aquat Microb Ecol 78: 25–37, 2016
36
LITERATURE CITED
➤
➤
➤
➤
➤
➤
➤
➤
➤
➤
➤
➤
➤
➤
➤
Adams J (2009) Species richness — patterns in the diversity
of life. Springer, Berlin and Heidelberg
Altschul SF, Gish W, Miller W, Myers EW, Lipman DJ (1990)
Basic local alignment search tool. J Mol Biol 215:403−410
Bates ST, Clemente JC, Flores GE, Walters WA, Wegener
Parfrey L, Knight R, Fierer N (2013) Global biogeography
of highly diverse protistan communities in soil. ISME J 7:
652−659
Bik HM, Sung W, De Ley P, Baldwin JG, Sharma J, RochaOlivares A, Thomas WK (2012) Metagenetic community
analysis of microbial eukaryotes illuminates biogeographic patterns in deep-sea and shallow water sediments. Mol Ecol 21:1048−1059
Boenigk J, Wodniok S (2014) Biodiversität und Erdgeschichte. Springer Spektrum, Berlin
Borcard D, Gillet F, Legendre P (2011) N umerical ecology
with R. Springer, New York, NY
Bryant JA, Lamanna C, Morlon H, Kerkhoff A, Enquist BJ,
Green JL (2008) Microbes on mountainsides: contrasting
elevational patterns of bacterial and plant diversity. Proc
Natl Acad Sci USA 105:11505−11511
De Senerpont Domis LN , Elser JJ, Gsell AS, Huszar VLM
and others (2013) Plankton dynamics under different
climatic conditions in space and time. Freshw Biol 58:
463−482
Dziock F, Henle K, Foeckler F, Follner K, Scholz M (2006)
Biological indicator systems in floodplains — a review.
Int Rev Hydrobiol 91:271−291
Facher E, Schmidt R (1996) A siliceous chrysophycean cystbased pH transfer function for Central European lakes.
J Paleolimnol 16:275−321
Frey W, Lösch R (2010) Geobotanik — Pflanze und Vegetation in Raum und Zeit. 3. Auflage. Spektrum, Heidelberg
Gaston KJ (2000) Global patterns in biodiversity. N ature
405:220−227
Graham LE, Graham JM, Wilcox LW (2009) Algae, 2nd edn.
Pearson, San Francisco, CA
Grossmann L, Jensen M, Heider D, Jost S and others (2016)
Protistan community analysis: key findings of a largescale molecular sampling. ISME J 10:2269−2279
Häder DP, Kumar HD, Smith RC, Worrest RC (1998) Effects
on aquatic ecosystems. J Photochem Photobiol B 46:
53−68
Hewitt G (2000) The genetic legacy of the quaternary ice
ages. Nature 405:907−913
Hillebrand H, Gruner DS, Borer ET, Bracken MES and
others (2007) Consumer versus resource control of producer diversity depends on ecosystem type and producer
community structure. Proc N atl Acad Sci USA 104:
10904−10909
Hurley C (2012) Package ‘gclus’: clustering graphics. https://
cran.r-project.org/web/packages/gclus/gclus.pdf
Jeppesen E, Søndergaard M, Jensen JP, Havens KE and
others (2005) Lake responses to reduced nutrient loading — an analysis of contemporary long-term data from
35 case studies. Freshw Biol 50:1747−1771
Kamenik C, Koinig KA, Schmidt R (2005) Potential effects of
pre-industrial lead pollution on algal assemblages from
an Alpine lake. Verh Internat Verein Limnol 29:535−538
Lane D (1991) Nucleic acid techniques in bacterial systematics, 1st edn. Wiley, Chichester
Lecroq B, Lejzerowicz F, Bacher D, Christen R and others
(2011) Ultra-deep sequencing of foraminiferal micro-
➤
➤
➤
➤
➤
➤
➤
➤
➤
➤
➤
➤
➤
➤
➤
➤
barcodes unveils hidden richnes of early monothalamous
lineages in deep-sea sediments. Proc Natl Acad Sci USA
108:13177−13182
Legendre P, Gallagher ED (2001) Ecologically meaningful
transformations for ordination of species data. Oecologia
129:271−280
Livingstone DM, Dokulil MT (2001) Eighty years of spatially
coherent Austrian lake surface temperatures and their
relationship to regional air temperature and the N orth
Atlantic Oscillation. Limnol Oceanogr 46:1220−1227
Lotter AF, Birks HJB, Hofmann W, Marchetto A (1997) Modern diatom, cladocery, chironomid, and chrysophyte cyst
assemblages as quantitative indicators for the reconstruction of past environmental conditions in the Alps. I.
Climate. J Paleolimnol 18:395−420
Medinger R, Nolte V, Pandey RV, Jost S and others (2010)
Diversity in a hidden world: potential and limitation of
next generation sequencing for surveys of molecular
diversity of eukaryotic microorganisms. Mol Ecol 19: 32–40
Medlin L, Elwood HJ, Stickel S, Sogin ML (1988) The
characterization of enzymatically amplified eukaryotic
16S-like rRNA-coding regions. Gene 71:491−499
N ogués-Bravo D, Araújo MB, Romdal T, Rahbek C (2008)
Scale effects and human impact on the elevational
species richness gradients. Nature 453:216−219
N olte V, Pandey RV, Jost S, Medinger R, Ottenwälder B,
Boenigk J, Schlötterer C (2010) Contrasting seasonal
niche separation between rare and abundant taxa conceals the extent of protist diversity. Mol Ecol 19:
2908−2915
Oksanen J, Blanchet FG, Kindt R, Legendre P and others
(2016) ‘vegan’ 2.3.4.4— community ecology package.
http://CRAN.R-project.org/package=vegan
(accessed
March 2016)
Pandey RV, N olte V, Schlötterer C (2010) CAN GS: a userfriendly utility for processing and analyzing 454 GS-FLX
data in biodiversity studies. BMC Res Notes 3:3
Paradis E, Blomberg S, Bolker B, Claude J and others (2015)
Package ‘ape’: analysis of phylogenetics and evolution.
http: //mirror-60ox.hmdc.harvard.edu/CRAN/web/
packages/ape/ape.pdf
Pla S, Catalan J (2005) Chrysophyte cysts from lake sediments reveal the submillennial winter/spring climate
variability in the northwestern Mediterranean region
throughout the Holocene. Clim Dyn 24:263−278
Rahbek C (1995) The elevational gradient of species richness: a uniform pattern? Ecography 18:200−205
Romdal TS, Grytnes JA (2007) An indirect area effect on
elevational species richness patterns. Ecography 30:
440−448
Smith JD, Milne PJ (1981) Spectrophotometric determination of silicate in natural waters by formation of αmolybdosilicic acid and reduction with a tin(IV)−ascorbic
acid−oxalic mixture. Anal Chim Acta 123:263−270
Soares EM, Figueredo CC, Gücker B, Boëchat IG (2013)
Effects of growth condition on succession patterns in
tropical phytoplankton assemblages subjected to experimental eutrophication. J Plankton Res 35:1141−1153
Sommaruga R (2001) The role of solar UV radiation in the
ecology of alpine lakes. J Photochem Photobiol B 62:
35−42
Sommaruga R, Casamayor EO (2009) Bacterial ‘cosmopolitanism’ and importance of local environmental factors for
community composition in remote high-altitude lakes.
Freshw Biol 54:994−1005
Grossmann et al.: Protistan diversity in alpine lakes
➤ Sommer U, Adrian R, De Senerpont Domis L, Elser JJ and
➤
➤
➤
➤
others (2012) Beyond the plankton ecology group (PEG)
model: mechanisms driving plankton succession. Annu
Rev Ecol Evol Syst 43:429−448
Sonntag B, Summerer M, Sommaruga R (2011) Factors
involved in the distribution pattern of ciliates in the
water column of a transparent alpine lake. J Plankton
Res 33:541−546
Stoeck T, Bass D, N ebel M, Christen R and others (2010)
Multiple marker parallel tag environmental DN A sequencing reveals a highly complex eukaryotic community in marine anoxic water. Mol Ecol 19:21−31
Tolotti M, Thies H, Cantonati M, Hansen CME, Thaler B
(2003) Flagellate algae (Chrysophyceae, Dinophyceae,
Cryptophyceae) in 48 high mountain lakes of the northern and southern slope of the Eastern Alps: biodiversity,
taxa distribution and their driving variables. Hydrobiologia 502:331−348
Triadó-Margarit X, Casamayor EO (2012) Genetic diversity of planktonic eukaryotes in high mountain lakes
Editorial responsibility: Karel Šimek,
České Budějovice, Czech Republic
➤
➤
➤
➤
➤
37
(central Pyrenees, Spain). Environ Microbiol 14:
2445−2456
Tringe SG, von Mering C, Kobayashi A, Salamov AA and
others (2005) Comparative metagenomics of microbial
communities. Science 308:554−557
Vogler P (1966) Zur Analytik der Phosphorverbindungen in
Gewässern. Limnologica 4:437−444
Wang J, Soininen J, Zhang Y, Wang B, Yang X, Shen J
(2011) Contrasting patterns in elevational diversity
between microorganisms and macroorganisms. J Biogeogr 38:595−603
Wang J, Soininen J, Zhang Y, Wang B, Yang X, Shen J
(2012) Patterns of elevational beta diversity in micro- and
macroorganisms. Glob Ecol Biogeogr 21:743−750
Weisse T (2008) Distribution and diversity of aquatic protists: an evolutionary and ecological perspective. Biodivers Conserv 17:243−259
Whittaker RJ, Willis KJ, Field R (2001) Scale and species
richness: towards a general, hierarchical theory of species diversity. J Biogeogr 28:453−470
Submitted: January 13, 2016; Accepted: July 26, 2016
Proofs received from author(s): November 8, 2016