ArticlePDF Available

Factors shaping submerged bryophyte communities: A conceptual model for small mountain streams in Germany

Authors:
  • Landau Campus Universität Koblenz-Landau

Abstract and Figures

Several models explaining species composition of aquatic bryophytes are available for specific regions. However, a more general, conceptual model applicable to a broader range of regions is lacking.We present a conceptual model ranking environmental factors determining submerged bryophyte communities in small mountain streams. It was tested on a dataset of 54 stream sections after removing the effect of stream size and altitude. Species responses were modeled with pH as predictor variable based on 97 stream sites covering six mountain regions all over Germany. Multiple regressions revealed the importance of primary growth factors (light, Ep(CO2)) and substrate for the total submerged bryophyte coverage.The known distinction of hard- and softwater bryoflora was clearly supported. The floristic composition of headwaters was predominantly determined by the bicarbonate/ionic strength complex. Species response to pH values supported this result and thus our conceptual model. The primary growth resources light, Ep(CO2) and availability of coarse streambed material explained one third (Radjusted2 = 0.34) of total submerged bryophyte cover. Disturbances, predominantly spates, reduce biomass but do not affect the basic floristic structure.In conclusion, conceptual models and monitoring methods focusing on aquatic bryophytes need to clearly distinguish “aquatic” from “submersed by chance”. All “aquatic bryophytes” found in Germany can also occur at least temporarily at non-submerged sites. Therefore, a distinction between primary growth factors and additional resources is recommended to disentangle factors determining aquatic bryophyte communities.
Content may be subject to copyright.
Limnologica
42 (2012) 242–
250
Contents
lists
available
at
SciVerse
ScienceDirect
Limnologica
jo
u
rn
al
homepage:
www.elsevier.de/limno
Factors
shaping
submerged
bryophyte
communities:
A
conceptual
model
for
small
mountain
streams
in
Germany
Horst
Trempa,,
Dorothea
Kampmannb,
Ralf
Schulza
aInstitute
for
Environmental
Sciences,
University
of
Koblenz-Landau,
Campus
Landau,
Fortstrasse
7,
D-76829
Landau/Pfalz,
Germany
bDepartment
of
Physical
Geography,
Goethe
University,
Altenhöferallee
1,
D-60438
Frankfurt,
Main,
Germany
a
r
t
i
c
l
e
i
n
f
o
Article
history:
Received
26
July
2011
Received
in
revised
form
8
January
2012
Accepted
9
January
2012
Keywords:
Aquatic
bryophytes
Distribution
patterns
Species
response
curves
Water
Framework
Directive
a
b
s
t
r
a
c
t
Several
models
explaining
species
composition
of
aquatic
bryophytes
are
available
for
specific
regions.
However,
a
more
general,
conceptual
model
applicable
to
a
broader
range
of
regions
is
lacking.
We
present
a
conceptual
model
ranking
environmental
factors
determining
submerged
bryophyte
communities
in
small
mountain
streams.
It
was
tested
on
a
dataset
of
54
stream
sections
after
removing
the
effect
of
stream
size
and
altitude.
Species
responses
were
modeled
with
pH
as
predictor
variable
based
on
97
stream
sites
covering
six
mountain
regions
all
over
Germany.
Multiple
regressions
revealed
the
importance
of
primary
growth
factors
(light,
Ep(CO2))
and
substrate
for
the
total
submerged
bryophyte
coverage.
The
known
distinction
of
hard-
and
softwater
bryoflora
was
clearly
supported.
The
floristic
composi-
tion
of
headwaters
was
predominantly
determined
by
the
bicarbonate/ionic
strength
complex.
Species
response
to
pH
values
supported
this
result
and
thus
our
conceptual
model.
The
primary
growth
resources
light,
Ep(CO2)
and
availability
of
coarse
streambed
material
explained
one
third
(Radjusted2=
0.34)
of
total
submerged
bryophyte
cover.
Disturbances,
predominantly
spates,
reduce
biomass
but
do
not
affect
the
basic
floristic
structure.
In
conclusion,
conceptual
models
and
monitoring
methods
focusing
on
aquatic
bryophytes
need
to
clearly
distinguish
“aquatic”
from
“submersed
by
chance”.
All
“aquatic
bryophytes”
found
in
Germany
can
also
occur
at
least
temporarily
at
non-submerged
sites.
Therefore,
a
distinction
between
primary
growth
factors
and
additional
resources
is
recommended
to
disentangle
factors
determining
aquatic
bryophyte
communities.
© 2012 Elsevier GmbH. All rights reserved.
Introduction
Ecological
information
for
the
small
group
of
submerged
bryophytes
and
their
role
in
stream
ecosystems
is
sparse.
Rea-
sons
may
be
their
low
dominance
and
spatially
heterogeneous
arrangement
in
many
stream
types
(Stream
Bryophyte
Group,
1999)
or
their
reputation
as
an
exclusive
group
studied
only
by
specialists.
In
zoological
investigations
submerged
bryophytes
are
commonly
regarded
as
a
substrate
(phytal)
because
they
provide
a
unique
habitat
for
macroinvertebrates
(Butcher,
1933;
Suren,
1993;
Riis
and
Biggs,
2003).
They
also
offer
macroinvertebrates
shelter
against
physically
and
chemically
related
impacts
(e.g.
Glime,
1994;
Parker
et
al.,
2007).
Aquatic
bryophytes
have
rarely
been
used
for
classification
purposes
(e.g.
stream
typology)
or
as
bioindica-
tor
(Zechmeister
et
al.,
2003),
as
there
are
much
fewer
experts
Corresponding
author.
Tel.:
+49
7032
893717.
E-mail
address:
tremp@uni-landau.de
(H.
Tremp).
for
bryophytes
than
for
macroinvertebrates,
amphibians
or
algae
(Fritz
et
al.,
2009).
In
contrast
to
vascular
plants,
the
high
potential
for
vegetative
and
generative
(spores)
propagation
of
submerged
bryophytes
leads
to
a
high
similarity
of
its
flora
in
the
holarctic,
thus
Central
Europe
and
Scandinavia
share
many
species
with
Northern
America
and
Canada
(Frahm
and
Vitt,
1993;
Dierßen,
2001).
Terms
like
water
mosses,
stream
bryophytes
or
aquatic
bryophytes
are
difficult
to
define
in
a
rigorous
way
biologically.
All
three
terms
assume
that
the
aquatic
medium
is
either
the
only
or
the
most
favored
site
where
these
species
show
maximum
growth
and
complete
their
life
cycle
including
spore-germination,
pro-
tonema
formation,
gametophyte-
and
sporophyte
induction
and
growth
as
well
as
spore
dispersal
(Tremp,
1999).
Following
this
definition
all
aquatic
bryophytes
in
Germany
might
be
regarded
as
facultative
aquatics
as
discussed
already
decades
ago
by
Elßmann
(1923).
Some
of
them
prefer
– but
not
mandatory
–a
permanent
submerged
stage,
but
even
with
the
genus
Fontinalis
sporophyte
development
does
not
occur
in
a
long-term
fully
denudated
sit-
uation.
From
early
desiccation
experiments
(Irmscher,
1912)
it
is
0075-9511/$
see
front
matter ©
2012 Elsevier GmbH. All rights reserved.
doi:10.1016/j.limno.2012.01.003
H.
Tremp
et
al.
/
Limnologica
42 (2012) 242–
250 243
known
that
leaves
of
Fontinalis
antipyretica
die
after
14
days
but
the
stems
will
regenerate
after
four
weeks
of
drought.
Several
species
of
the
genus
Fontinalis
can
survive
up
to
one
year
in
humid
places
(Glime,
1971)
or
even
falling
dry
over
several
weeks,
as
is
commonly
found
in
ephemeral
and
periodical
karstic
streams
or
in
mountain
streams
over
wintertime.
In
such
conditions
aquatic
bryophytes
survive
on
the
dry
land,
under
cold-dry
conditions
or
even
covered
with
snow.
Compared
to
submerged
vascular
plants
“water
mosses”
seem
to
be
ecologically
unspecialized,
considering
that
the
vegetative
stages
of
most
mosses,
even
such
of
dry
habitats
as
Grimmia
pulvinata
or
Bryum
argenteum,
are
able
to
survive
completely
submerged
conditions
over
one
year
(Elßmann,
1923).
Growth
experiments
by
Zastrow
(1934)
showed
that
aquatic
and
amphibic
forms
of
aquatic
bryophytes
could
be
transferred
into
each
other
and
vice
versa.
Goebel
(1889;
cited
in
Gessner,
1955)
called
bryophytes
“halbe
Wasserpflanzen”
(semi
waterplants)
as
sub-
merged
forms
of
amphibic
or
terrestrial
bryophytes
are
often
not
only
falsely
identified
but
also
treated
as
new
species.
Summing
up
Gessner’s
(1955,
p.
270)
remark
about
the
amphibic
mode
of
life
of
some
bryopytes
.
.
.
viable
in
both
air
and
water,
but
nowhere
completely
at
home”
seems
justified.
The
search
for
specific
adaptive
species
traits
to
cope
with
the
selec-
tive
forces
of
their
habitat
is
therefore
questionable.
The
only
but
most
important
trait
shared
by
all
aquatic
bryophytes
is
their
high
regenerative
capacity,
e.g.
sprouting
from
small
pieces
of
stems
tightly
attached
with
rhizoids
and
leaves
which
are
able
to
develop
rhizoids.
It
is
stated
that
their
non-adaptive
strategy
makes
them
so
suc-
cessful
in
dealing
with
the
harsh
environment
of
the
land–water
ecotone
in
headwater
streams,
where
aquatic
vascular
plants,
adapted
well
to
the
aquatic
environment,
cannot
cope
with
such
selective
forces.
Aquatic
bryophytes
try
to
occupy
highly
disturbed
sites
of
severe
stress.
Grime
(1977)
assigned
no
viable
plant
strat-
egy
to
such
habitat
characteristics.
But
Kautsky’s
(1988)
stunted
strategy
type,
complementing
the
CSR
strategy,
matches
the
com-
paratively
small,
slow-growing,
long
living
species
with
many
various
types
of
vegetative
diaspores
well.
At
the
small
scale
in
streams,
a
vertical
bryophyte
zonation
on
boulders
and
walls
can
be
found
(Watson,
1919;
Glime,
1970;
Craw,
1976;
Glime
and
Vitt,
1987).
It
shows
an
increasing
species
richness
within
the
gradient
from
submerged
to
the
semi-aquatic,
hygropetric
or
splash
zone
(Vitt
et
al.,
1986;
Glime
and
Vitt,
1987;
Muotka
and
Virtanen,
1995).
Muotka
and
Virtanen
(1995)
described
the
shift
from
truly
aquatic
species
to
facultative
aquat-
ics
and
semi
aquatics
along
the
vertical
gradient
as
being
gradual.
This
zone
can
also
be
regarded
as
shelter
zone
for
aquatic
species
from
where
recovery
after
spates
might
occur
(Tremp
and
Kohler,
1993).
Besides
vertical
zonation
in
structurally
rich
streams,
lon-
gitudinal
changes,
classified
and
termed
upper,
middle
and
lower
zone
(Holmes
and
Whitton,
1977),
on
vegetation
occur.
Often
the
upper
zone
is
dominated
by
bryophytes.
The
upper
zone
in
silicate
streams
can
be
divided
floristically
further
when
alkalinity
and
pH
rise
with
distance
from
the
source
(Demars
and
Thiébaut,
2008)
and
can
be
distinct
when
a
stable
acidity
gradient
of
physiological
relevance
i.e.
pH
4–7
is
developed
(Tremp
and
Kohler,
1993;
Tremp,
1999).
Numerous
publications
in
relation
with
the
European
Water
Framework
Directive
(EU,
2000;
Hering
et
al.,
2006;
Szoszkiewicz
et
al.,
2006)
stimulated
scientific
research
in
this
field
and
gave
proposals
for
monitoring.
However
the
application
(Staniszewski
et
al.,
2006)
and
applicability
(Demars
and
Edwards,
2009)
of
macrophytes,
and
even
more
bryophytes
in
freshwater
monitor-
ing
is
still
limited
and
sometimes
questionable
due
to
lack
of
sound
data.
Hence,
the
present
paper
has
the
following
three
objec-
tives:
Fig.
1.
Conceptual
model
of
the
abiotic
environment
and
strictly
submerged
bryophytes
in
streams.
The
bryophyte
community
firstly
differs
between
hard-
water
and
softwater
type.
The
three
corners
of
the
triangle
indicate
site
factors
which
reduce
submerged
bryophytes
directly:
mechanical
stress
due
to
sub-
strate
instability/current
velocity
and
the
subsequent
grinding
of
plant
material.
Carbonate-incrustation
and
high
acidity
reduces
aquatic
species
occurrence
dramat-
ically.
Apart
from
these
extremes
the
primary
growth
factors
(inner
circle)
shape
the
bryophyte
community.
(i) we
propose
an
integrative
conceptual
model
for
submersed
bryophyte
composition
and
structure;
(ii) we
then
test
some
of
its
predictions
using
data
collected
across
Germany;
(iii) finally
we
compare
our
findings
with
existing
conceptual
mod-
els
from
Northern
America,
New
Zealand
and
Finland.
A
conceptual
model
of
aquatic
bryophyte
occurrence
Several
conceptual
models
in
aquatic
bryophyte
ecology
can
be
found,
for
regions
of
different
relief
energy,
i.e.
for
alpine
streams
(Suren,
1996;
Suren
and
Ormerod,
1998;
Suren
and
Duncan,
1999)
or
lower
mountainous
streams
of
the
boreal
zone
(Muotka
and
Virtanen,
1995),
and
a
general
model
for
aquatic
macrophytes
(Riis
and
Biggs,
2001).
A
conceptual
model,
applicable
to
a
broader
range
of
regions,
however,
is
lacking
(Fig.
1).
The
ranking
of
the
impact
of
environmental
variables
on
species
composition
depends
primarily
on
the
specific
range
of
the
values
of
variables
considered,
secondly
on
the
regions
investigated,
and
thirdly
on
the
bryophyte
map-
ping
method.
Many
investigations,
however,
cover
only
a
restricted
range
of
environmental
parameters
(many
sampling
points
in
the
same
stream).
For
example,
the
effect
on
the
floristic
composition
only
becomes
evident
when
a
wide
range
of
substrates
is
covered.
Moreover,
all
complexes
of
environmental
variables
can
be
over-
ridden
by
the
influence
of
the
relief
energy
(see
Table
1).
Fig.
1
shows
the
factors
and
factor
complexes
which
are
postulated
as
primary
for
structuring
aquatic
bryophyte
com-
munities
in
headwater
streams.
The
model
highlights
first
the
softwater–hardwater
gradient,
which
differentiates
the
com-
munity
structure.
Secondly,
it
depicts
the
productivity
factors
(=primary
growth
factors),
which
enable
growth
of
permanent
sub-
merged
bryophytes,
and
thirdly
the
disturbance
regime
due
to
transported
solids
(bed
instability,
grinding
effect),
which
modifies
the
aquatic
bryophyte
communities
and
in
its
extremes
prevents
the
development
of
true
aquatic
macrophyte
vegetation.
This
view
is
obtained
from
streams
where
movement
of
bed
material
is
com-
mon,
destroying
vegetation
almost
completely.
Nevertheless,
some
bryophytes
can
be
found
at
sheltered
sites
as
in
the
lee
of
large
boul-
ders
above
the
middle
water
layer.
The
conceptual
model
(Fig.
1;
244 H.
Tremp
et
al.
/
Limnologica
42 (2012) 242–
250
Table
1
Environmental
factors
predominantly
differentiating
stream
bryophyte
vegetation.
The
primary
growth
factors
water
and
temperature
are
not
taken
into
account.
Light,
turbidity
Ylla
et
al.
(2007)
CO2and
turbulence
to
enhance
its
acquisition
Jenkins
and
Proctor
(1985),
Proctor
(1990)
Nutrients
Steinman
(1994),
Vanderpoorten
and
Palm
(1998),
Vanderpoorten
and
Durwael
(1999)
Altitude
Ormerod
et
al.
(1994),
Suren
and
Ormerod
(1998)
Current
velocity,
water
level
fluctuation
Vitt
and
Glime
(1984),
Glime
and
Vitt
(1987)
Flow
variability,
disturbance,
streambed
stability
Englund
(1991),
Muotka
and
Virtanen
(1995),
Suren
and
Ormerod
(1998),
Riis
and
Biggs
(2003)
Saprobity,
organic
pollution Kolkwitz
(1950),
Szoszkiewicz
et
al.
(2006)
Ionic
strength
and
buffering
incorporating
electrical
conductivity,
total
hardness,
temporary
hardness,
pH,
aluminum
Watson
(1919),
Butcher
(1933),
Sørensen
(1948),
Vitt
et
al.
(1986),
Frahm
(1992),
Tremp
and
Kohler
(1993),
Tremp
(1999),
Vanderpoorten
et
al.
(2000)
inner
circle)
for
aquatic
bryophytes
includes
a
view
expressed
in
Slack
and
Glime
(1985)
that
community
structure
of
aquatic
bryophytes
is
primarily
controlled
by
non-equilibrium
processes
to
which
the
species
respond
as
opportunists.
This
includes
the
fact
that
aquatic
bryophytes
are
highly
adaptive
to
changing
ambi-
ent
light
and
nutrient
conditions
as
well
as
temperature
(Maberly,
1985).
Their
ability
to
adapt
has
to
be
taken
into
account
not
only
over
a
short
period
of
time
and
quick
physiological
responses
but
also
due
to
their
evergreen
status
over
the
entire
annual
growth
period
and
even
over
their
whole
lifetime
existence.
Thus,
they
encompass
stages
where
they
might
exist
only
in
a
cryptic
stage,
e.g.
as
rhizoid
fragments.
The
environmental
factors
are
assigned
different
weightings
by
the
individual
authors
(Table
1).
Materials
and
methods
Study
sites
The
core
dataset
consists
of
54
sections
of
100
m
each,
investi-
gated
in
52
mountain
streams
in
the
state
of
Baden-Württemberg
(SW-Germany,
see
Fig.
2).
The
width
of
the
streambed
is
rela-
tively
small
in
proportion
to
its
roughness.
Thus,
the
dragging
power
of
stream
velocity,
without
taking
transported
solids
into
account,
seldom
exceeds
the
tolerance
of
aquatic
cryptogams.
Due
to
the
unique
heterogeneous
geology
of
Baden-Württemberg
(Embleton,
1983)
streams
originated
from
Gneiss
and
Granite
base
rock
formation,
Triassic
red
sandstone,
limestone
and
the
highly
variable
Keuper
formation,
Jurassic
limestone
as
well
as
Pleistocene
moraine
material.
The
field
work
was
conducted
in
1996
and
1997.
A
dataset
of
96
stream
sections
in
69
streams
situated
in
mountain
regions
in
the
Southern,
Western
and
Northern
part
of
Germany
as
the
Odenwald,
Pfälzer
Wald,
Solling
and
the
Harz
mountains
was
used
for
comparison.
The
field
work
here
was
con-
ducted
between
1992
and
1994.
In
both
datasets
spatially
induced
floristic
autocorrelation
is
reduced,
due
to
sampling
the
same
stream
manifold,
which
weak-
ens
the
explaining
effect
of
environmental
variables
(Borcard
et
al.,
1992;
Isaak
and
Hubert,
2001;
Heino
and
Virtanen,
2006).
Most
mountain
streams
were
situated
in
forested
catchments,
thus
the
plant
species
were
not
exposed
to
excessive
nutrient
inputs.
With
only
a
few
exceptions
the
streams
are
summercold;
therefore,
extreme
temperature
ranges
are
not
found
and
a
pronounced
sea-
sonality
effect
on
submerged
species
is
of
minor
relevance.
Species
sampling
For
the
generation
of
both
datasets,
the
submerged
vegetation,
bryophytes,
vascular
plants
and
algae
of
genus
Batrachosper-
mum
and
Hildenbrandia
were
mapped
from
June
to
September
by
wading
through
the
shallow
brooks.
Species
which
could
not
be
identified
on
the
spot
were
collected
and
determined
in
the
laboratory.
Species
known
to
occur
submerged
only
occasion-
ally
(i.e.
due
to
higher
water
table)
like
Racomitrium
aciculare
(Hedw.)
Brid.,
Dichodontium
pellucidum
(Hedw.)
Schimp.
and
Thamnobryum
alopecurum
(Hedw.)
Gangulee
were
not
assessed.
Only
permanently
submerged
plants
of
the
frequently
occurring
amphiphytes
were
included,
e.g.
Berula
erecta
(Huds.)
Cov-
ille.
For
the
core
dataset
cover
estimates
for
species
were
recorded,
taking
only
the
wetted
area
into
account.
In
shallow
mountain
brooks
species
individuals
seldom
overlap.
This
facilitates
species
cover
estimation
in
percentages.
Prerequisites
for
this
mapping
procedure
are
good
visibility,
low
numbers
of
true
submerged
species,
and
a
stream
width
of
less
than
3
m.
These
conditions
were
encountered
in
almost
all
cases.
In
the
additional
dataset
truly
aquatic
species
were
mapped
by
presence–absence.
Nomenclature
of
higher
plants
follows
the
German
standard
list
of
Wisskirchen
and
Haeupler
(1998),
for
bryophytes
the
bryophyte
flora
of
Baden-Württemberg
(Nebel
and
Philippi,
2000–2005)
was
used.
Fig.
2.
Map
showing
the
54
sampled
reaches
in
52
different
1st
and
2nd
(3rd)
order
streams
Baden-Württemberg
(SW-Germany).
In
the
western
part,
silicatic
mountain
ranges
(Black
Forest,
Southern
Odenwald)
strech
from
SSW–NNE
direc-
tion
(quadrats).
The
other
regions
with
underlying
Triassic
limestone,
the
Keuper
formation
and
Jurassic
limestone
are
situated
in
the
center.
Moraine
material
covers
parts
of
SE
of
Baden-Württemberg.
H.
Tremp
et
al.
/
Limnologica
42 (2012) 242–
250 245
Environmental
variables
For
the
core
dataset
(n
=
54)
the
following
environmental
vari-
ables
were
recorded:
Substratum.
The
percentage
cover
of
substratum
was
estimated
according
to
the
classes
loam
(special
type
of
bedrock
mate-
rial),
mud
(sedimented
organic
and
anorganic
fine
material),
sand
(<0.2
cm),
gravel/pebble
(0.2–6.3
cm),
stones/cobble
(6.3–20
cm),
blocks
(>20
cm),
bedrock
(underlying
bedrock
and
sinter
forma-
tion)
and
wood.
To
focus
on
the
question
of
streambed
stability,
the
sediment
fractions
were
differentiated
only
in
finer
(gravel,
sand,
mud)
more
instable
material
and
coarser
bed
material
(block,
stone).
Physico-chemistry.
Water
temperature,
pH
and
electrical
con-
ductivity
were
measured
once
per
mapping
section
with
WTW
(Wissenschaftlich-Technische
Werkstätten
Ltd.)
devices.
Acid
neutralizing
capacity
(ANC4.3)
was
measured
titrating
100
ml
of
stream
water
sample
with
hydrochloric
acid
of
a
normality
of
either
0.01
N
or
0.1
N
with
potentiometric
endpoint
detection
on
pH
4.3.
To
give
a
rough
estimate
of
excess
carbon
dioxide
partial
pressures
Ep(CO2)
from
pH
and
ANC4.3 measurements
cor-
rected
for
temperature,
we
applied
the
formula
given
in
Neal
et
al.
(1998a,
second
equation
on
p.
173).
This
equation
allows
for
vari-
ation
in
temperature
and
average
ionic
strength.
Ep(CO2)
is
the
ratio
of
CO2to
what
would
normally
be
dissolved
in
water
of
the
same
temperature
at
equilibrium.
The
intention
was
to
get
a
mea-
sure
which
combines
pH
and
ANC4.3 and
thus
allows
an
estimate
of
the
CO2supply
for
aquatic
bryophytes
in
a
multiple
regression
model.
Stream
flow
characteristics.
The
discharge
was
estimated
by
mul-
tiplying
the
cross-section
of
suitable
sites
(i.e.
approximately
rectangular)
by
current
velocity
estimated
with
small
pieces
of
wood
drifting
over
a
distance
of
between
5
and
10
m.
Watercover
was
estimated
in
percentage
with
respect
to
emerged
structures
within
the
flowing
water.
Turbulence
was
estimated
on
a
five
point
ordinal
scale
ranging
from
no
flow
to
strong
turbulent
flow.
Insolation.
The
insolation
was
expressed
in
hours
of
potential
direct
sunshine
on
the
water
surface.
In
each
mapping
section
it
was
measured
at
meters
0,
50
and
100
with
a
horizontoscope
developed
by
F.
Tonne
(Schütz
and
Brang,
1995).
This
is
an
acrylic
hemisphere
with
an
integrated
compass
and
sun
path
chart
of
here
49
northern
latitude.
The
sun
tracks
give
the
astronomic
sunshine
duration
at
the
15th
of
every
month.
This
value
was
multiplied
by
30
and
the
results
for
all
months
summed
up,
which
resulted
in
potential
sunshine
hours
per
year.
A
mean
value
of
the
three
estimates
was
used
for
analysis.
Altitude.
Altitude
in
meters
above
sea
level
was
taken
directly
from
maps
(1:
25,000).
For
the
additional
dataset
(n
=
96)
only
data
on
pH
and
ANC4.3
and
altitude
were
available.
Data
analyses
Species
composition–environment
interaction
was
analyzed
for
the
core
dataset
(n
=
54)
by
means
of
partial
canonical
corre-
spondence
analysis
(pCCA).
The
unimodal
ordination
method
was
chosen
because
the
gradient
length
of
the
first
axis
of
the
DCA
detrended
by
segments
was
larger
than
four
(Lepˇ
s
and ˇ
Smilauer,
2003).
Scaling
is
focussed
on
inter-species
distances.
The
analy-
ses
should
focus
on
reach
scale
variables,
consequently
altitude
and
the
variable
discharge,
which
approximates
stream
size,
were
entered
as
covariables,
their
influence
thus
factored
out.
Since
the
covariables
were
not
strongly
correlated
to
the
remaining
variables,
the
explaining
power
of
the
latter
was
kept.
Statistical
analyses
were
performed
with
Canoco
4.5
(Ter
Braak
and ˇ
Smilauer,
2002).
This
ordination
method
may
be
applied
to
non-normal
data,
how-
ever,
ecological
structures
emerge
more
clearly
when
the
data
do
not
show
a
strong
asymmetry
(Legendre
and
Legendre,
1998).
We
applied
logarithmic
and
arcsine-squareroot
data-transformation
to
reduce
impacts
of
outliers
and
approximate
normal
distribution.
Graphically
expressed
frequency
distributions
and
the
asymptotic
significance
(P)
of
a
Kolmogorov–Smirnov
test
showed
high
values,
so
normal
distribution
could
be
assumed
with
exception
for
the
chemistry
data.
As
we
sampled
different
geological
areas,
pH,
con-
ductivity
and
acid
neutralizing
capacity
were
distributed
bimodal,
which
cannot
be
treated
with
usual
transformations.
Subsequently
all
abiotic
variables
were
standardized
for
analysis.
Percentage
val-
ues
of
species
cover
estimates
were
log(y
+
1)
transformed
and
rare
species
downweighted
(Lepˇ
s
and ˇ
Smilauer,
2003).
Species
occur-
ring
in
less
than
three
mapping
sections
were
excluded.
Species
response
curves
were
thus
fitted
for
species
with
at
least
25
occurrences
in
the
extended
dataset.
These
were
Scapania
undu-
lata
(L.)
Dumort.,
Chiloscyphus
polyanthos
(L.)
Corda,
Brachythecium
rivulare
Schimp.,
Rhynchostegium
riparioides
(Hedw.)
Cardot,
Amblystegium
tenax
(Hedw.)
C.E.O.
Jensen,
and
Fissidens
crassipes
Wilson
ex
Bruch
and
Schimp.
Segments
showing
none
of
these
species
were
eliminated,
yielding
a
sample
size
of
n
=
145.
Species
occurrence
was
expressed
as
presence–absence
therefore
their
probability
of
occurrence
was
modeled
as
a
parameter
of
the
bino-
mial
distribution
(Lepˇ
s
and ˇ
Smilauer,
2003).
The
model
was
fitted
using
the
logit
link
function
(Ter
Braak
and
Looman,
1987).
The
model
complexity
of
the
generalized
linear
models
(GLMs)
simple
linear
(linear)
or
second
order
polynomial
(quadratic)
was
chosen
by
using
the
Akaike
Information
Criterion
(AIC).
Here
smaller
values
indicate
better,
more
parsimonious,
models
(Akaike,
1978).
Signif-
icance
of
the
model
was
determined
by
a
deviance
based
F-test
comparing
the
fitted
models
with
the
0-model
f(y)
=
const.
(Lepˇ
s
and ˇ
Smilauer,
2003).
Relationship
between
submerged
vegetation
cover
and
environ-
mental
variables
was
analyzed
with
a
multiple
regression
model
using
Systat
10.2.
Results
Ordination
results
At
54
sampling
sites
of
the
core
dataset
39
true
submerged
species
26
bryophytes,
3
macrophytic
algae,
1
macrophytic
lichen
and
9
vascular
macrophytes
were
found.
Species–environment
interaction
was
analyzed
by
means
of
partial
canonical
correspondence
analysis
(pCCA)
to
test
if
the
con-
ceptual
model
(Fig.
1)
is
supported
by
real
data.
The
overall
variance
in
species
dispersion
(total
inertia)
was
4.19,
including
the
covari-
ables
altitude
and
discharge
(3.66).
The
eigenvalue
of
the
highly
significant
first
canonical
axis
(EV
=
0.533;
P
=
0.002)
explains
14.6%
of
the
variability
of
species
data.
Variability
in
species
composition
was
explained
to
22.5%
by
the
two
first
ordination
axes
(Fig.
3)
and
to
31.0%
by
all.
Besides
variables
with
unique
information
indicated
by
a
low
variance
inflating
factor
in
the
analysis,
highly
redundant
variables
(electrical
conductivity,
ANC4.3,
bed
material,
watercover)
occurred
(Table
2).
The
ordination
diagram
(Fig.
3)
shows
the
community
structure
of
aquatic
bryophytes
indicating
a
strong
dependency
on
physico-chemistry:
ANC4.3,
pH
and
electri-
cal
conductivity.
Two
vascular
amphiphytes
with
a
low
frequency
(Veronica
anagallis-aquatica
L.,
Nasturtium
officinalis
L.)
showed
a
different
pattern.
The
species
structure
deducted
from
the
diagram
(Fig.
3)
is
con-
current
with
field
observations
for
mountain
streams
over
a
broad
range
of
geological
entities.
In
Table
2
the
environmental
variables
246 H.
Tremp
et
al.
/
Limnologica
42 (2012) 242–
250
0.80.60.40.20.0-0.2-0.4-0.6-0.8-1.0
-0.6
-0.4
-0.2
0.0
0.2
0.4
0.6
0.8
1.0
Sca und
Mar ema
Hyo arm
Hyg och
Rhy rip Bra riv
Bra plu
Fis cra
Chi pol
Amb ten
Amb rip
Fon ant
Cra com Cra fil
Bat mon
Hil riv
Agr sto
Gly flu
Pha aru
Ver a na
Nas off
Ver b ec
Pot. hours
of sunshine
Watercovered area
pH
ANC4.3
El. con-
ductivity
*
**
Block+
stone
Turbu lent+
strong turb.
**Gravel+sand+mud
*Temperature
Fig.
3.
Biplot
of
species
and
directly
measured
environmental
variables
along
the
first
two
axes
of
pCCA.
Because
altitude
and
discharge
were
entered
as
co-variables
they
are
not
shown
in
the
diagram.
are
ranked
following
their
variance
explaining
effect
for
submerged
macrophyte
distribution.
Due
to
the
strong
correlation
among
the
chemical
variables,
which
can
be
expected
from
their
causal
relationships,
their
conditional
value
(Lambda-A)
decreases,
inde-
pendent
of
which
variable
was
chosen
first.
Bed
material
variables
provided
a
low
significant
contribution
to
species
differentiation
(Table
2).
The
influence
of
turbulence
and
direct
sunlight
was
not
significant.
Species
response
curves
The
species
response
curves
illustrate
the
functional
relation-
ship
between
the
pH
gradient
and
the
probability
of
species
occurrence.
Species
response
(Fig.
4)
to
a
single
physic-chemical
variable
(Table
2)
based
on
the
complete
dataset,
supports
the
already
indicated
differences
of
species
reactions
to
the
water
chemistry
complex
(Fig.
3)
more
precisely.
All
regressions
were
significant
(P
<
0.002).
A
linear
model
was
fitted
for
three
species,
for
the
other
species
a
second
order
polynomial
model.
For
the
latter
values
of
species
optimum
±
standard
error
and
tolerance
along
the
pH-gradient
were
obtained:
S.
undulata
(pH
4.32
±
0.37;
tolerance
=
1.12),
B.
rivulare
(pH
6.78
±
0.16;
tolerance
=
0.85)
and
C.
polyanthos
(pH
6.86
±
0.82;
tolerance
=
0.82).
Multiple
linear
regression
model
The
fraction
of
variance
of
the
submerged
vegetation
cover
accounted
for
by
the
multiple
linear
regression
model
(Radjusted2)
is
0.34
(Table
3).
Slopes
of
all
the
resource
variables
are
positive
and
Table
2
Forward
selection
of
directly
measured
variables
of
a
partial
canonical
corre-
spondence
analysis
(pCCA).
The
variables
altitude
and
discharge
were
treated
as
covariables.
P
refers
to
the
significance
level
obtained
with
a
Monte
Carlo
permuta-
tion
test
(499
permutations).
Variable
Marginal
Conditional
effects
Lambda-1
Lambda-A
P
F
aAcid
neutralizing
capacity4.3 0.48
0.48
0.002
7.63
apH
0.26
0.13
0.060
2.00
Water
temperature
0.19
0.12
0.076
1.91
Water
covered
area 0.36
0.11
0.042
1.95
Blocks
+
stones
0.27
0.13
0.028
2.15
Gravel
+
sand
+
mud
0.21
0.12
0.016
2.14
Turbulent
+
strong
turbulent
0.24
0.06
0.358
1.51
aElectrical
conductivity
0.48
0.09
0.160
1.11
Potential
hours
of
sunshine 0.12
0.06
0.360
1.09
aIndependent
(marginal)
effects
of
chemistry
variables
are
high
(Lambda-1).
9.08.07.06.05.04.0
pH
0.0
0.2
0.4
0.6
0.8
1.0
Probability of occurrence
Amblystegium tenax
Brachythecium rivulare
Chiloscyphus polyanthos
Fissidens crassipes
Rhynchostegium riparioides
Scapania undulata
Fig.
4.
Species
response
curves
fitted
with
a
generalized
linear
model
(GLM).
For
occurrence
of
Brachythecium
rivulare,
Chiloscyphus
polyanthos
and
Scapania
undulata
a
second
order
polynomial
predictor
was
appropriate.
For
the
other
species
a
linear
predictor
fitted
the
species
data
better.
significant,
so
it
can
be
concluded
that
there
is
a
positive
relation-
ship
to
the
response
variable.
Considering
the
standard
coefficients
it
can
be
concluded
that
every
single
resource
contributes
to
a
same
amount
to
submerged
vegetation
cover,
which
is
predominantly
bryophyte
cover.
The
Durbin–Watson
coefficient
(2.058)
provides
no
hint
to
strongly
autocorrelated
residuals,
which
would
have
inflated
the
analysis
result.
Beside
large
scale
variables
(hard–softwater
type)
taxonomic
richness
becomes
locally
modified
by
structural
variables
(Table
4).
Variance
of
the
taxonomic
richness
accounted
for
by
the
multiple
linear
regression
model
(Radjusted2)
is
37%
(Durbin–Watson
coeffi-
cient
2.080).
Discussion
To
verify
our
proposed
model
of
environmental
factors
deter-
mining
aquatic
bryophyte
community
structure,
an
extensive
dataset
of
truly
aquatic
bryophytes,
investigated
over
a
broad
range
of
environmental
situations,
was
used.
In
contrast
to
other
habitat
templates,
where
substrate
stability
and
disturbance
were
the
main
focus
(Suren,
1996;
Suren
and
Ormerod,
1998;
Suren
and
Duncan,
1999;
Muotka
and
Virtanen,
1995),
our
findings
support
the
soft-
and
hardwater
phenomenon
as
the
most
discriminating
cause
of
bryophyte
species
composition.
Primary
growth
factors,
bed
struc-
ture,
and
disturbance
can
be
addressed
as
secondary
factors.
These
might
only
modify
the
communities,
impoverishing
or
enhancing
species
richness.
Implemented
environmental
variables
Sunshine
duration
estimates
would
have
been
more
precise
if
measured
in
wintertime,
too,
when
there
is
no
dense
foliage.
Nev-
ertheless,
the
estimates
are
more
satisfying
compared
to
the
often
used
percentage
of
tree
shading
at
zenith
estimates
(foliage
cover).
This
is
not
sufficient
because
light
not
from
above,
but
shining
from
the
sides
often
accounts
for
a
considerable
part
of
direct
sun-
light
reaching
the
water
surface.
Contrastingly,
with
the
here
used
horizontoscope
method,
bank
height
is
considered
adequately.
In
macrophyte
rich
streams
estimates
of
bed
material
coverage
in
winter
and
in
summer
would
have
been
reasonable
as
temporary
sedimentation
occurs
during
the
vegetation
period.
Average
cur-
rent
velocity,
often
used
as
surrogate
for
CO2supply,
can
be
very
similar
in
a
turbulent
upland
stream
and
in
apparently
low
flowing
lowland
rivers
(Proctor,
1990).
Therefore,
Ep(CO2),
even
as
a
rough
H.
Tremp
et
al.
/
Limnologica
42 (2012) 242–
250 247
Table
3
ANOVA
table
and
parameter
estimates
for
the
multiple
linear
regression
model
linking
the
resource
variables
light
(potential
sunshine
duration)
and
excess
carbon
dioxide
Ep(CO2)
and
stable
substrate
(blocks
and
stones)
with
submerged
vegetation
cover.
Small
differences
are
due
to
rounding
errors.
Sum
of
squares
df
Mean
square
F-ratio
P
Regression 5.647
3 1.882
9.281
<0.001
Residual 9.128
45
0.203
Coefficients
Std.
error
Std.
coefficients
Tolerance
t-Value
P
Constant 1.671
0.530
3.158
0.003
Pot.
sunshine
duration
0.544
0.201
0.339
0.871
2.702
0.010
Ep(CO2)
0.531
0.178
0.355
0.963
2.978
0.005
Blocks
and
stones
0.011
0.004
0.367
0.883
2.939
0.005
estimation,
might
be
better
suited
(Demars
and
Trémolières,
2009)
for
quantifying
the
supply
of
carbon
dioxide
especially
on
aquatic
bryophytes,
for
which
it
is
the
only
carbon
source.
The
range
of
0.3
up
to
44
times
saturation
matches
the
reported
values
from
Neal
et
al.
(1998b)
and
Demars
and
Trémolières
(2009)
well.
Mapping
procedure
At
54
sampling
sites
(52
streams)
of
the
core
dataset
39
true
submerged
species
– 26
bryophytes,
3
macrophytic
algae,
1
macrophytic
lichen
and
9
vascular
macrophytes
were
found.
In
these
small
streams
aquatic
macrophytes
of
genus
Ranuncu-
lus
or
Potamogeton,
which
usually
occur
in
mid-size
and
larger
streams,
were
missing.
The
mapping
procedure
is
comparable
to
many
macrophyte
investigations
which
allows
for
compar-
isons
(e.g.
Hering
et
al.,
2006).
However,
it
does
not
match
the
phytosociological
bryophyte
mapping
where
quadrat
size
rarely
exceeds
0.1
m2.
But
with
respect
to
the
reach
scale
(Frissell
et
al.,
1986),
variability
caused
by
different
plot
sizes
due
to
different
streambed
widths
is
small
compared
to
variability
in
species
rich-
ness.
All
aquatic
and
many
terrestrial
bryophytes
are
able
to
occur
in
the
intermediate
spray
zone,
but
only
a
few
species
are
able
to
occur
at
permanently
inundated
sites.
Species
of
the
semi-aquatic
zone
are
mostly
affected
by
variables
other
than
water
quality
(Vanderpoorten
and
Palm,
1998).
Defining
instream-bryophytes
“between
two
banks”
(Scarlett
and
O’Hare,
2006)
results
in
high
numbers
of
species
but
it
incorporates
numerous
semi-aquatic
and
terrestrial
species,
which
can
be
found
elsewhere,
too.
Good
com-
parisons
are
possible
when
a
clear
distinction
is
made,
as
in
Heino
and
Virtanen
(2006).
With
the
exception
of
Amblystegium
(Lep-
todictyum)
riparium,
known
as
a
species
which
is
able
to
occur
permanently
submerged,
our
findings
are
in
concordance
with
those
named
“obligatory
aquatic”
in
Dierßen
(2001)
and
also
those
listed
in
Appendix
1
of
Heino
and
Virtanen
(2006).
Including
such
knowledge
species,
which
are
inundated
only
by
chance
due
to
higher
water
table,
could
be
classified
accordingly.
Often
terres-
trial
bryophyte
species
are
incorporated
in
species
lists
of
aquatic
habitats
(Schaumburg
et
al.,
2004;
Meilinger
et
al.,
2005),
which
seems
unavoidable
in
times
of
higher
discharge,
but
will
weaken
the
interpretative
ecological
strength
in
investigations
and
assess-
ment
approaches.
Differentiating
aquatic
bryophytes
sensu
Vitt
and
Glime
(1984)
in
“obligatory”
(i.e.
obligate
aquatics
occur
only
sub-
merged)
and
“facultative”
(i.e.
often
submerged
but
able
to
tolerate
periods
of
desiccation)
seems
inappropriate.
Even
species
named
“obligate
aquatics”
are
able
to
live
outside
water
over
prolonged
periods
of
time.
The
permanently
moist,
but
not
submerged
spray
zone
provides
suitable
conditions
for
all
aquatic
bryophytes.
Because
of
the
strict
definition
of
bryophytes
in
this
study,
i.e.
occurring
submerged
over
a
prolonged
period
of
time,
50%
fewer
species
were
incorporated
in
the
analysis
compared
to
other
inves-
tigations.
Recently
published
European
datasets
(Holmes
et
al.,
1998;
Szoszkiewicz
et
al.,
2006;
Demars
and
Edwards,
2009)
are
available,
allowing
comparisons
of
species
demands
and
for
defin-
ing
species
as
true
aquatics
and
thus
suitable
for
“in
stream
monitoring”.
If
focussing
on
true
aquatic
species
the
thalweg
map-
ping
procedure
applied
by
Fritz
et
al.
(2009)
seems
very
promising.
Ordination
results
We
avoided
problems
of
measuring
environmental
variables
with
too
narrow
ranges
by
the
number
of
sampled
regions
and
there
found
different
geologies.
The
chosen
variables
match
the
pro-
posed
model
(Fig.
1)
fully.
Despite
focusing
only
on
small
mountain
streams,
we
observed
a
complete
species
turnover
due
to
the
strong
discriminating
factor
geology,
which
is
closely
related
to
the
vari-
ables
ANC4.3,
electrical
conductivity
and
pH.
These
have
to
be
seen
as
a
complex,
which
can
be
assigned
to
soft-
and
hardwaters.
Values
of
electric
conductivity
between
200
and
300
!S/cm
were
com-
pletely
lacking
in
the
Baden-Württemberg
core-dataset.
Floristic
comparison
shows
that
this
bimodality
acts
as
simple
discrimina-
tor