Content uploaded by Tenna Riis
Author content
All content in this area was uploaded by Tenna Riis on Feb 11, 2019
Content may be subject to copyright.
PRIMARY RESEARCH PAPER
Riparian forest modifies fuelling sources for stream food
webs but not food-chain length in lowland streams
of Denmark
I. Gonza
´lez-Bergonzoni .P. B. Kristensen .A. Baattrup-Pedersen .
E. A. Kristensen .A. B. Alnoee .T. Riis
Received: 19 November 2015 / Revised: 7 July 2017 / Accepted: 18 July 2017
ÓSpringer International Publishing AG 2017
Abstract Several studies have shown that the origin
of carbon fuelling food webs in streams depends on
riparian cover type. In forested stream sites allochtho-
nous resources fuel food webs, whereas autochthonous
resources support biomass in grassland (open-canopy)
stream sites. However, some studies suggest that
autochthonous carbon (of highest quality) is prefer-
entially assimilated regardless of riparian cover and
that the food-chain length (FCL) may be larger in
grassland than in forested sites. We used stable iso-
topes of carbon and nitrogen in adjacent grassland and
forested reaches to compare the contribution of
autochthonous vs. allochthonous resources to the
biomass of the whole macroinvertebrate assemblage
and to the most abundant taxa. Moreover, we
compared the FCL between forested and grassland
sites by estimating the trophic position of brown trout,
Salmo trutta. Autochthonous support to macroinver-
tebrate biomass was higher in grassland than in
forested sites, often changing from a dominantly
autochthonous to an allochthonous-generated biomass
from grassland to forested. This held true for the whole
macroinvertebrate assemblage and for specific spe-
cies. FCL remained similar between reach types. Our
study suggests that autochthonous resources are
assimilated to a higher extent when their availability
increases with canopy openness but allochthonous
carbon sustain macroinvertebrate biomass in forested
reaches.
Handling editor: David J. Hoeinghaus
Electronic supplementary material The online version of
this article (doi:10.1007/s10750-017-3313-1) contains supple-
mentary material, which is available to authorized users.
I. Gonza
´lez-Bergonzoni A. Baattrup-Pedersen
A. B. Alnoee
Department of Bioscience, Aarhus University, Vejlsøvej
25, 8600 Silkeborg, Denmark
I. Gonza
´lez-Bergonzoni
Laboratorio de Etologı
´a, Ecologı
´a y Evolucio
´n, Instituto
de Investigaciones Biolo
´gicas Clemente Estable,
Montevideo, Uruguay
I. Gonza
´lez-Bergonzoni (&)
Departamento de Ecologı
´a y Evolucio
´n, Facultad de
Ciencias, Universidad de la Repu
´blica, Igua
´4225,
11400 Montevideo, Uruguay
e-mail: ivg@fcien.edu.uy
P. B. Kristensen A. B. Alnoee T. Riis
Department of Bioscience, Aarhus University, Ole Worms
Alle
´, Building 1135, 8000 Aarhus, Denmark
E. A. Kristensen
EnviDan, Vejlsøvej 23, 8600 Silkeborg, Denmark
123
Hydrobiologia
DOI 10.1007/s10750-017-3313-1
Keywords Resource subsidy Allochthonous
detritus Stable isotopes Bayesian mixing models
Trophic position Carbon subsidies
Introduction
The relationship between resource availability and
energy flow suggested since the River Continuum
Concept (Vannote et al., 1980) indicates that the
energetic sources for riverine food webs shift along the
stream longitudinal gradient. Small headwater
forested stream reaches are fuelled primarily by
allochthonous sources, whereas autochthonous
sources from in-stream production, such as periphytic
algae, increase in importance as energetic sources for
food webs towards larger open-canopy reaches (Van-
note et al., 1980; Collins et al., 2016). The higher
importance of allochthonous sources in small-forested
streams may reflect the intimate contact between the
water and the riparian surroundings, representing a
high cross-ecosystem resource exchange (e.g. Gregory
et al., 1991; Moldenke & Ver Linden, 2007; Sullivan
2013). Towards larger open-canopy sites (e.g. grass-
land riparian area) this contact is reduced and an
increase in light irradiation produces an increase of in-
stream productivity, probably favouring the use of
autochthonous resources (e.g. Vannote et al., 1980;
Sullivan, 2013; Collins et al., 2016).
While the River Continuum Concept has several
limitations and it is not universal, being based on and
supported by studies conducted in north temperate
mountain forested basins (e.g. Minshall, 1967; Cum-
mins, 1974; Dekar et al., 2012), its broader ideas about
the role of riparian forest cover in determining the
energy fuelling food webs (but not only at a commu-
nity-structure level as originally stated) have been
supported by studies within the patch dynamics
concepts (e.g. Spencer et al., 2003; Thompson &
Townsend, 2003; Winemiller et al., 2010; Junker &
Cross, 2014; Collins et al., 2016). For example recent
evidence comparing small forested and larger grass-
land reaches in both tropical and temperate streams
suggest that the role of riparian forest in increasing
allochthonous contribution to biomass is consistent
across climate regions (Collins et al., 2016). The
origin of food webs has mostly been studied by
comparing stream reaches located in contrasting land
use and riparian cover types (e.g. Rounick et al., 1982;
Thompson & Townsend, 2003; Whiting et al., 2011).
However there are also many studies under semi-
experimental conditions such as before and after forest
clear-cut (e.g. Rounick et al., 1982; Noel et al., 1986;
Thompson et al., 2009) or in areas affected by
wildfires that have removed riparian forests (e.g.
Spencer et al., 2003). Whatever the riparian forest
reduction has caused, in all studies where canopy
openness increases an increase in stream primary and
secondary production has been observed (Noel et al.,
1986; Thompson & Townsend, 2003; Thompson et al.,
2009; Whiting et al., 2011). This increase in canopy
openness and productivity has been related to an
increase in the reliance of primary consumers on
autochthonous resources such as benthic algae (Rou-
nick et al., 1982; Spencer et al., 2003; Thompson &
Townsend, 2003; Junker & Cross, 2014).
However, the relationship between riparian forest
and the allochthonous subsidies to food webs has not
been supported universally. For example, stable iso-
tope analyses of carbon (C) and nitrogen (N) from a
large variety of stream types (from headwater to
lowland streams) and across biogeographic regions
(e.g. Fu
¨reder et al., 2003; Brito et al., 2006; McNeely
et al., 2006; Lau et al., 2009a,b; Jardine et al., 2013)
suggest that autochthonous resources may be the main
energetic source for food webs, also in-streams with
significant riparian forest cover in both some temper-
ate (e.g. Fu
¨reder et al., 2003; McNeely et al., 2006) and
tropical climates (e.g. Brito et al., 2006; Lau et al.,
2009a,b). Based on similar evidence for larger rivers,
the revised Riverine Productivity Model (Thorp &
Delong, 2002) postulates that most of the biomass in
all riverine food webs is fuelled by autochthonous
sources independently of the presence of riparian
forest and/or high availability of allochthonous
organic matter. This would reflect the fact that
autochthonous carbon is more easily incorporated into
the biota as it is more labile than the allochthonous
carbon (Thorp & Delong, 1994,2002 and references
therein) and more nutritious (as suggested by its lower
C:N ratio than detrital sources, Lau et al., 2009b).
The ideas in the River Continuum Concept and the
Riverine Productivity Model should not be considered
as opposite, since they target different systems and
organisation levels (whole river longitudinal gradient
and communitary changes in macroinvertebrates con-
tra lowland streams and rivers and whole metazoan
Hydrobiologia
123
biomass), but there is some debate on whether the
ideas of the Riverine Productivity Model might apply
also to headwater streams, where River Continuum
Concept predicted a dominance of allochthonous-
feeders in the community (Finlay, 2001; Brito et al.,
2006; Moulton, 2006 Lau et al., 2009b). This may
challenge the generality of the role of riparian forests
in determining the main energetic subsidy of stream
food webs (Fu
¨reder et al., 2003; McNeely et al., 2006;
Moulton, 2006; Li & Dudgeon, 2008; Lau et al.,
2009a,b).
In addition several studies have reported positive
relationships between system productivity and food-
chain length (from here on ‘‘FCL’’) (e.g. Pimm &
Kitching, 1987; Post et al., 2000; Thompson &
Townsend, 2005), among other explanatory variables
for variation in FCL such as for example the relation-
ship between hydrological variability and ecosystem
area (McHugh et al., 2010; Sabo et al., 2010; Sullivan
et al., 2015). At the reach scale, unshaded reaches are
usually more productive than forest reaches where
shading limits algal growth (Thompson et al., 2009;
Whiting et al., 2011; Junker & Cross, 2014), and Lau
et al. (2009b) revealed a longer FCL in open canopy
than in forested reaches, possibly as a consequence of
increased availability of high quality food of auto-
chthonous origin and a higher transfer efficiency to
primary consumers of autochthonous resources rela-
tive to allochthonous resources. Alongside, questions
emerge on the role of riparian forest in determining
energy pathways in stream food webs and how
forested riparian areas may regulate maximum FCL.
Here, using C and N stable isotopes we studied the
role of riparian forest in driving stream food webs in
four Danish headwater streams sampled during sum-
mer. Each of the studied streams had a forested
(closed canopy) and a grassland (open canopy) section
(c.a. 100 m) separated c.a. 500 m from each other,
allowing us to directly compare the differences in the
food webs of adjacent forest and grassland streams.
This sharp transition was caused by deforestation by
agricultural practices in the past. First, we aimed to
assess the contribution of allochthonous (i.e. coarse,
suspended, and fine benthic particulate organic mat-
ter: abbreviated as CPOM, SPOM, and FBOM,
respectively, from hereon) vs. autochthonous (i.e.
epibenthic algae biomass) energy sources to primary
consumers between forest and grassland stream
reaches considering the whole community (all
macroinvertebrates pooled) and specific species
(those common between reach types). Specifically,
based on previous works in temperate forest streams
(i.e. Junker & Cross 2014; Collins et al., 2016), we
hypothesised that the fuelling resources for food webs
depend on riparian cover type at the reach scale.
Autochthonous material would be the dominant
fuelling source for food webs (i.e. contributes [50%
to consumer biomass) in grassland sites where algal
biomass availability is greater than in forest streams,
whereas allochthonous material is the dominant
fuelling source in the forested streams. Moreover we
expect that along the entire studied stream sections the
proportion of biomass derived from allochthonous
resources increase with increasing canopy coverage.
Alternatively stream food webs could be mainly based
on autochthonous resources independently of riparian
forest presence as postulated by the Riverine Produc-
tivity Model for larger riverine systems. Second, we
aimed to assess FCL in forest and grassland sections
within the same system. FCL was estimated as the
maximum trophic position of a same top predator in
each site, and we hypothesised that FCL is longer in
grassland sites, reflecting the higher availability of
high quality autochthonous C (following the produc-
tivity-FCL relationship hypothesis, e.g. Thompson &
Townsend, 2005), resulting in a higher trophic
position of top predators. Similarily we expect that
FCL increase with increasing benthic algae biomass
and decreasing canopy coverage.
Materials and methods
Study streams and timing of study
All four streams were located in Jutland, Denmark,
and selected to represent an open and a forested reach
of approximately at least 100 m, being separated from
each other in about 500 m distance, with a sharp
transition between each reach type (Fig. 1). This study
design (eight sites in total, four forested and four
grassland, one forested and grassland site located in
each river) was chosen to try to keep other conditions
than the differences caused by the riparian settings, as
similar as possible. The direction of the flow varied
between study sites, from forest to grassland in two of
the study streams (River Alsted, a tributary to River
Gudenaa, and River Skader, a tributary to River
Hydrobiologia
123
Alling) and from grassland to forest in the other two
sites (River Granslev, tributary to River Gudenaa, and
Storkesig Brook, tributary to River Aarhus). These
different flow orientations were chosen to try to avoid
biases produced by the potential effect of flow
direction. However, the effect of river flow was not
tested explicitly due to lack of replicates (two streams
with each type of flow) Samplings for food-web
reconstruction using stable isotopes were conducted
from mid- to late June 2013.
Stream physical and chemical characteristics
A 100 m long sample reach was delineated in the
downstream end of both the open and the forested
canopy reach (Fig. 1). Stream physical conditions
were characterised according to the Danish National
Monitoring Programme for the Aquatic Environment
(Friberg et al., 2005). Measurements were performed
along 20 transects positioned perpendicular to the
stream flow every 5 m within the 100 m stream reach.
Width was measured along each transect, which was
furthermore divided into five quadrates used for depth
measurements and characterisation of substrate and
macrophyte coverage. Substrates covering C25% of
the quadrate were quantified according to six cate-
gories: rock ([60 mm), coarse gravel (10–60 mm),
fine gravel (3–10 mm), sand (0.25–3 mm), mud and
clay (\0.25 mm) and debris. Macrophyte coverage
was determined from cover estimates within each of
the quadrates applying the following five categories:
\5, 5–25, 25–50, 50–75, and[75%. This information
was used to calculate mean reach-scale macrophyte
coverage. Measurements of discharge were conducted
at the transition point between the forested and
grassland reach using the velocity-area method
according to Jensen & Frost (1992). Furthermore, we
recorded water temperature and pH in the field and
took an integrated water sample which was trans-
ported refrigerated to the lab for analysis of alkalinity
and nutrients. Stream water temperature was regis-
tered with a Hobo
Ò
Pendant temperature logger, and
pH was measured with checkkit
Ò
micro pH-WP2 0–14
pH. Total N (TN) was analysed on a Shimadzu TOC-
N, whereas NH
4
?
,NO
3
-
and PO
43-
were analysed on
a Lachat (LACHAT instruments, USA). TP was
determined based on the methodology described in
Brix & Schierup (2001). Alkalinity was estimated
from the same samples by Gran titration using
0.01 mM HCl.
Canopy cover was measured on each bank in each
transect with a LAI-2000 plant canopy analyzer (LI-
COR, Lincoln, Nebraska) determining the Leaf Area
Index (LAI) by recording light received at a horizontal
Fig. 1 Map showing the
studied sites and a sampling
diagram within each site.
The geographical
coordinates are shown for
each locality in which
adjacent forested and
grassland reaches were
sampled. On the right panel
a sampling diagram shows
the areas and distances
sampled in detail for each
site, using River Skader as
an example
Hydrobiologia
123
plane (e.g. a stream water surface) expressed as a
proportion of the light received under unshaded
conditions (e.g. no canopy). LAI characterises plant
foliage cover which is defined by the one-sided green
leaf area per unit ground surface area (Chen & Black,
1992; Monteith & Unsworth, 1973). LAI values range
from 0 to 1, 1 representing a completely unshaded
stream (Welles & Norman, 1991). Based on LAI, we
then estimated the proportion of canopy cover follow-
ing the same procedure as Kristensen et al. (2014).
Biotic measurements
Epibenthic algae biomass was estimated at reach scale
as the mean chlorophyll a (Chl-a; lgm
-2
) content of
five samples taken on each of the dominant benthic
substrate types present (i.e. [25% of substrate cover-
age) within each reach using a modified (cut through)
syringe of known area (6.74 cm
2
) on soft bottom
substratum, a kayak tube on gravel (22.23 cm
2
) and
scraping off a surface area (6.74 cm
2
) on stones. The
samples were transferred to vials and then filtrated
onto glass microfiber filters (GFC) in the laboratory. A
total of 197 samples were analysed. Samples were
extracted in ethanol (96%) and analysed in a spec-
trophotometer (Shimadzu, UV21 160/Shimadzu, UV-
1700) following standardised methods (Pedersen,
2004).
Collection of food sources and consumers
for stable isotope analysis
Sources and consumers for stable isotope analysis
were sampled to represent major basal autochthonous
and allochthonous sources, and all samples were
collected integrating samples located along the same
100 m reach used for reach characterisations. For
basal sources, we sampled filamentous algae by hand
picking and periphyton by scraping stones and wash-
ing sand. The samples were filtered through GFC
filters to remove the water and small invertebrates and
obtain epilithic and epipsammic periphyton isotopic
signatures. For CPOM, allochthonous materials such
as wood, leaves, and parts of terrestrial plants were
collected manually in the stream bottom in proportion
to its prevalence in the environment, integrating
different tree and plant species in decomposition
along all microhabitats present within the 100-m-
study reach, and an integrative water sample from the
water column was collected for SPOM (filtered on
GFC filters). FBOM was collected using an open-
ended PVC cylinder (16 cm in diameter) that was
pushed into the sediment in five locations along the
100-m reach. The sediments were lightly disturbed by
hand down to about 1-cm depth, the water in the
cylinder was mixed and a water sample taken. The
samples were stored refrigerated and transported to the
lab where they were filtered through a GFC filter and
prepared for SIA according to standard procedures
(Levin & Currin, 2012). Leaves of macrophyte species
present were also collected for SIA.
Fish were caught by single-pass electrofishing in
the study reach and macroinvertebrates by sweep
netting along transects and hand picking from rocks
and hard substratum along the 100-m reach. Despite
that it was a qualitative sampling, the effort was
similar in all stream reaches, and invertebrate taxa
present in abundances permitting stable isotope anal-
ysis (i.e. potentially representing [1 mg dry weight
after processing) were collected. Fish (a total of 43
individuals collected) were euthanized with an over-
dose of 2-phenoxy-ethanol and a portion of dorsal
muscle was extracted. All samples were refrigerated
and transported to the lab where they were immedi-
ately frozen. In the lab, invertebrates were identified
using taxonomic keys (e.g. Mey 1997; Dobson et al.,
2012) with bulk samples of macroinvertebrates
grouped per taxa (most frequently to genus level),
and source samples were identified to the lowest
possible taxonomic level and prepared for isotopic
analysis following standard procedures (including
cleaning and removal of shells and hard parts in
macroinvertebrates) (Levin & Currin, 2012). Samples
were freeze-dried, weighed (0.5–1.5 mg for animal
tissues, 2–3 mg for mosses and periphyton) and sent
for analysis at UC Davis stable isotope facility,
California, USA, using a continuous flow isotope
ratio mass spectrometer (IRMS). The natural abun-
dance of heavy and light C and N stable isotopes
(
13
C/
12
C and
15
N/
14
N), relative to the proportions of
these isotopes of a standard, Pee Dee Belemnite rock
and N of air, respectively, was determined. Propor-
tions are given as delta values (d
13
C and d
15
Nin
0
/
00
).
Energy source contribution modelling
The potential contribution of main energetic sources to
food webs was estimated by Bayesian mixing models
Hydrobiologia
123
that use d
13
C and d
15
N of food sources and consumers
and their fractionation coefficients to estimate the
most probable proportion of the biomass generated by
each food item for an individual consumer (Parnell
et al., 2010,2013b; Phillips, 2012). We therefore used
the main food sources well known from the literature
as potential components of the diet of primary
consumer macroinvertebrates, as this prevents inclu-
sion of unlikely sources improving model accuracy
(Parnell et al., 2010). The allochthonous source input
was the mean and standard deviation of isotopic
signatures of CPOM, SPOM, and FBOM, as well as
terrestrial riparian vegetation. As autochthonous
source input, we use mean and standard deviation of
isotopic signatures of filamentous algae and periphy-
ton from stone scrapes and sand wash. This aggrega-
tion of sources into a two-source model prevents the
under-determination of mixing models and has been
suggested as a standard methodology to follow in
these cases (Fry, 2013). Macrophytes were excluded
from the analysis as they very rarely constitute an
important food source for macroinvertebrate con-
sumers as has been observed in studies of macroin-
vertebrate gut content (Cummins & Klug, 1979;
Newman, 1991), stable isotopes (e.g. Hart & Lovvorn,
2003; Belicka et al., 2012), and fatty acid composition
(e.g. Belicka et al., 2012). The mean and standard
deviations of fractionation values used in the models
were taken from a meta-analysis (Post, 2002). Addi-
tionally it should be noted that calculation results
regarding the autochthonous contribution to food webs
might be slightly underestimated as we considered
FBOM and SPOM as an allochthonous source, while
Li & Dudgeon (2008) found that a minor fraction of
this resource might, in fact, be of autochthonous
origin. On a similar way the isotopic signature of
epilithic algae may also contain traces of allochtho-
nous material deposited with the benthic biofilm
matrix. However, we chose to disregard any con-
founding effect of this, as ‘‘allochthonous’’ is doubt-
less constituted by a vast majority of allochthonous
material and ‘‘autochthonous’’ by a majority of algal
matter; furthermore, the modelling set-up was consis-
tent between sties.
We built two Bayesian mixing models per stream,
one for the grassland and one for the forested site
based on mean and SD of isotopic signatures of
autochthonous vs. allochthonous resources and sepa-
rate bulk samples for each macroinvertebrate taxon
and samples of each fish individual. The modelling
was performed for all macroinvertebrate data pooled,
followed by estimation of the autochthonous and
allochthonous contribution for each taxon sample
separately in the SIAR package (Parnell et al.,
2013a,b) in R software (R core team, 2014). This
permitted us to test for differences in the autochtho-
nous contribution for all macroinvertebrates pooled
and also for similar species present in different reach
types using all samples of the same taxa as replicates.
Trophic position estimation
To test for changes in FCL between open canopy and
forested reaches, we estimated the trophic position of
the most abundant and frequent top predator found in
each reach type, in all cases brown trout (Salmo trutta
Linnaeus, 1758). The use of fish as an indicator of
trophic position in stream reaches can be problematic
because fish may migrate locally, thereby integrating
the isotopic signatures of both reach types, which may
hamper the estimation of reach type-specific food-
chain length. However, in our study brown trout was
chosen as a model secondary consumer because it is
ubiquitous in most Danish streams, easy to collect and
exhibits a sedentary behaviour with a restricted home
range (e.g. Bachmann, 1984; Ovidio et al. 2002).
Predatory invertebrates were not used because same
taxa were not present along all our study sites. To
validate its use as model for FCL in this study, we
tested if d
15
N isotopic signatures differed between
reach types, expecting that the isotopic signatures
would differ between adjacent forested and open-
canopy reaches within the same streams. FCL was
estimated as maximum trophic position in each site,
which was calculated according to standard proce-
dures (Vander Zanden et al., 1997; Post, 2002):
Trophic position ¼d15Nconsumer d15 Nbase
=3:4
þ2;
where 3.4 is the fractionation factor representing mean
enrichment of heavy to light isotopes of 1-trophic level
in d
15
N(%) of freshwater biota (Post, 2002), and 2 is
added as it represents the theoretical primary con-
sumer level of 2. We used the average d
15
N of all
primary consumers as baseline for trophic position
estimation. This is commonly used as a representative
average baseline value for a given food web (Jardine
Hydrobiologia
123
et al., 2014). Therefore, trophic position was calcu-
lated separately with a distinct baseline of mean
primary consumers for each forested and grassland
reach. Different food-web compartments have slightly
different fractionation values (ranging from 1.6
0
/
00
enrichment in d
15
N when basal invertebrates are
assimilated by predatory invertebrates and from 2.2 to
3.9 when macroinvertebrates are assimilated by fish
(e.g. Bunn et al., 2013; Hussey et al., 2014); however,
the use of a single average fractionation value is
commonly used to estimate an ‘‘average’’ FCL and
more appropriate in this case because we do not know
the exact number of trophic steps within each food-
web compartment. We consider that the study is
unbiased in this aspect as the calculation was made
consistently along all sampling sites and similar (and
thus, comparable) to most FCL studies using stable iso-
topes (e.g. Gonza
´lez-Bergonzoni et al., 2014; Iglesias
et al., 2016; Kirstensen et al., 2016). However, we
need to note that the real FCL could be underestimated
here if more trophic steps occur within the macroin-
vertebrate assemblage (e.g. Hussey et al., 2014).
Statistical analysis
In order to identify environmental differences that
characterise forest and grassland sites, we first used all
variables known in literature to differ between grass-
land and forest stream reaches (canopy coverage, fine
substrate coverage, stream width, macrophyte cover-
age, benthic algae biomass, and total nitrogen; Teix-
eira de Mello et al., 2016, and references therein) in a
PERMANOVA test (Anderson, 2001). Following this,
and in case of finding differences in the multivariate
test we compared each physical, chemical, and
biological characteristics of the two reach types using
one-way ANOVA after checking that the assumptions
regarding homoscedasticity were met. The compar-
isons were made by grouping the recordings from all
forest and grassland reaches of all streams, as well as
separately for forest vs. grassland reaches within each
stream using each square plot for which physical
characteristics were described (i.e. in the case of
substrate coverage) or a sample was taken (i.e. in the
case of Chl-a) as a replicate. We used mixed-effects
linear regression (using stream identity as a random
factor) to test the relationship between the percentage
of riparian coverage and epibenthic algae biomass
given as Chl-a (lgm
-2
).
Because grassland and forested reaches are indeed
determined by a series of environmental characteristics
more than only canopy coverage (e.g. Teixeira de Mello
et al., 2016), we compared the autochthonous contribu-
tion to invertebrate biomass between ‘‘grassland’’ and
‘‘forest’’ reaches within each stream using one-way
mixed-effects ANOVA, using stream identity as a
random factor as the assumptions for its application
were met. For the comparison using all pooled macroin-
vertebrates, each bulk sample by taxon was used as
replicate indistinctively of the taxonomic identity.
Whenever the same species or taxon was present in
both reach types of the same stream, we tested for
intraspecific differences in autochthonous contribution
to its biomass between the grassland and forested
reaches using bulk samples of that species as replicates
([3 bulk samples in each reach type). The species tested
were Ephemera danica Mu
¨ller, 1764 and Gammarus
pulex (Linnaeus, 1758) in River Granslev and Storkesig
Brook, Simuliidae spp. in Storkesig Brook and Lim-
nephilus sp. inRiver Skader. In River Alsted, intraspeci-
fic differences in autochthonous contribution could not
be tested due to lack of replicates. To test differences
between d
15
Nandd
13
C values in sources and top
predators of forest vs. grassland, we used one-way
ANOVA applying each sample as a replicate. The tests
were conducted for each stream separately and by
pooling data from all streams to compare forested and
open-canopy reaches using all sample individuals. The
same procedure was applied to test for differences in
FCL between forest and grassland sites using the trophic
position estimates of each trout individual.
Finally, to detect which of the environmental
variables characterising forested vs grassland reaches
has more influence on determining the origin of
biomass and FCL we performed mixed-effects linear
regression models using stream identity as a random
effect factor. These models tested the relationships
between mean autochthonous support to macroinver-
tebrate biomass (using all pooled macroinvertebrate
samples) and FCL with riparian coverage, stream
width and depth, proportion of fine substrates (sand
coverage), macrophyte coverage, benthic algae bio-
mass, and nutrient concentration (TN and TP). The
accomplishments of modelling assumptions were
tested following inspection of residual patterns (Zuur
et al., 2009). The marginal r
2
value, describing the
proportion of variance explained by the fixed-effect
factor in mixed-effects regressions were estimated
Hydrobiologia
123
using a function developed by (Nakagawa & Schiel-
zeth, 2013) in R software and it is here reported as ‘‘r
2
’’ .
Results
Physical, chemical, and biological characteristics
of the study reach
PERMANOVA test showed clear differences between
forested and grassland reaches in the group of
variables tested (F=2.7; df =6, P=0.02) suggest-
ing that forested reaches are characterised by a
combination of higher canopy coverage, stream width
and dominance of fine substrates, and lower macro-
phyte coverage and biomass of benthic algae than the
grassland sections (Table 1). When each environmen-
tal factor was analysed individually we corroborated
that forested reaches had a higher riparian canopy
cover (one-way ANOVA, F=7.3; df
res
=6;
P\0.05, Table 1) and lower epibenthic algae
biomass (Chl-a) than grassland reaches (forest sites
had approximately half of the biomass of the grassland
sites), both when comparing an average for all reaches
(one-way ANOVA, F=4.4, df
res
=28; P\0.05)
and for the individual reaches (one-way ANOVA,
P\0.05 for all, Table 1). Moreover, algal biomass
decreased with increasing riparian canopy cover
(Mixed-effect linear regression: df =133, F=87.8,
P\0.0001, r
2
=0.24), suggesting that algal biomass
was, indeed, limited by light in the forested reaches.
The forested reaches also had lower macrophyte cover
than the grassland sections (one-way ANOVA,
P\0.05), except for River Skader where no signif-
icant difference in macrophyte cover between reach
types was found (one-way ANOVA, P[0.05)
(Table 1). Storkesig Brook and River Granslev were
about 1 m wider towards their forest reaches. These
are typical changes observed in Danish lowland
stream morphology associated with the presence of
riparian forests (e.g. Teixeira de Mello et al., 2016).
Otherwise, no significant differences were found
between reach types (Table 1).
Allochthonous and autochthonous food resources
in forest and grassland reaches
Overall, autochthonous and allochthonous food
sources did not overlap between stream types in either
d
13
Cord
15
N values (Supplementary Appendix,
Table 1), and the elaboration of the stable isotope
mixing models could therefore be considered optimal
(Fry, 2006) (Supplementary Appendix, Table 1;
Fig. 2).
Macroinvertebrate taxa from the most representa-
tive feeding and functional groups were collected in all
stream reaches including filter feeders, shredders,
collector-gatherers, and predatory invertebrates (Sup-
plementary Appendix, Table 2). Macroinvertebrate
samples collected in each stream are shown in the
supplementary appendix in Table 2. SIAR Bayesian
mixing models using all primary consumers pooled
within each stream (n=188 in total) showed that the
autochthonous contribution to total biomass was
always higher in the grassland than in the forested
reaches, which meets our expectations according to
our first hypothesis (Fig. 3). In three of the streams
(River Granslev, River Alsted and Storekesig Brook),
the autochthonous contribution was higher in the
grassland, and in all four streams (River Granslev,
Alsted, Storkesig Brook and River Skader) the
allochthonous contribution was higher in the forested
compared to the grassland reaches (although only
slightly for River Skader) (Fig. 3). In River Granslev,
most of the primary consumer biomass was supported
by autochthonous resources, both in forest and grass-
land reaches (means of 62 and 85% of primary
consumer biomass being generated from autochtho-
nous sources in forested and grassland reaches,
respectively, Fig. 3). In contrast, River Skader seemed
to be fuelled mostly by allochthonous sources in both
reach types as autochthonous sources never dominated
over allochthonous sources (means of 10 and 28% of
autochthonous support in forested and grassland
reaches, respectively, Fig. 3). In River Alsted (flowing
from forest to grassland) and Storkesig Brook (flowing
from grassland to forest) a strong change in main
energy support to food webs occurred, shifting from a
mainly allochthonous-supported biomass in forested
to an autochthonous-supported biomass in grassland
reaches (from mean values of 37 to 60% of
autochthonous support to biomass in River Alsted
and from 6 to 55% of autochthonous contribution to
biomass in Storkesig Brook; Fig. 3).
The pattern of increasing mean autochthonous
contribution to biomass towards grassland reaches
became significant for all study sites (mixed-effects
ANOVA, df
res
=3; F=16.2; P=0.02). This was
Hydrobiologia
123
Table 1 Physical, chemical, and biological characteristics (means with min–max range in parentheses) of closed-canopy forested and open-canopy grassland reaches measured
at base flow at the time of sampling in each stream
Grassland ?forest Forest ?Grassland
Storkesig Granslev Alsted Skader
Grassland Forest Grassland Forest Grassland Forest Grassland Forest
Median July water temperature
(°C)
17.4
(12.4–24.3)
15.5
(11.8–20.4)
13.9 (10.7–17.3) 14.5 (10.9–18.5) 11.4 (4.5–11.5) 11.2 (4.7–11.2) 16.3
(12.3–20.3)
16.2
(12.4–20.1)
Discharge (l s
-1
) 18 18 136 136 297 297 106 106
Canopy cover (%) 27 93 78.7 88.4 54.8 84 55 73
Velocity (m/s) 0.30 0.29 0.26 0.30 0.50 0.40 Nd Nd
Width (m) 1.2 (0.8–1.5) 2.3 (1.4–4.9) 2,2 (1.4–3.7) 3.1 (2.1–4) 3.0 (2.1–4.1) 4.4 (3.8–5.3) 2.1 (1.6–2.5) 3.4 (1.8–3.9)
Depth (cm) 9.0 (3–30) 8.0 (4–30) 30.0 (5–70) 30.0 (3–80) 27.1 (20.4–40) 17.3 (0–45) 22.8 (15–30) 15.0 (6–25)
Boulders (%) 12.6 (0–31) 10.9 (0–25) 6.1 (0–30) 1.5 (0–11) 13.4 (0–27.3) 16.8 (0–41.7) 2.4 (0–20) 16.2 (0–33.3)
Coarse gravel (%) 16.1 (0–33) 25.1 (7–33) 14.1 (0–36) 0.4 (0–8) 25.3 (7.7–38.5) 22.2 (6.3–40) 11 (0–26.7) 20 (0–33.3)
Fine gravel (%) 17.9 (5–29) 23.4 (13–29) 2.3 (0–30) 7.8 (0–23) 24.6 (9.1–35.7) 20.4 (6.7–31.3) 20.2
(7.7–36.4)
17.2 (6.7–25)
Sand (%) 21.6 (5–38) 24.0 (12–38) 13.7 (0–30) 36.3 (10–55) 27.2 (14–45.5) 21.3 (6.3–33.3) 32.2
(24–45.5)
22.1 (0–33.3)
Debris (%) 9.4 (0–33) 13.9 (0–38) 12.0 (0–55) 9.5 (0–33) 10.0 (0–23.1) 14.5 (0–33.3) 17.9 (0–30.8) 16.0 (0–41.7)
Macrophyte cover (%) 10.5
(0.0–26.5)
0.4 (0.0–3.0) 25.1 (4.0–48.0) 3.7 (0.0–32.5) 47.1 (16.5–98.0) 6.5 (0.0–27.5) 11.7 (0.0–31) 5.9 (0.0–
.33.0)
Benthic chl.a (mg m
-2
)43 (8–128) 32 (6–82) 34 (3–93) 12 (6–45) 34 (13–90) 8 (2–47) 28 (2–56) 8 (2–18)
TN (mg l
-1
) 2.4 (1.7–3.1) 2.7 (2.3–3) 0.8 (0.8–0.9) 0.9 (0.9–1) 2.9 (1.2–5.4) 2.9 (1.4–5.1) 4.9 (2.3–6.4) 5.8 (5.4–6.3)
NO
3
-
(mg l
-1
) 1.9 (1.4–2.4) 2.02 (1.4–2.4) 0.6 (0.5–0.7) 0.5 (0.4–0.5) 2.7 (1.1–4.9) 2.8 (1.3–4.9) 5.5 (4.8–6.4) 5.7 (4.3–6.4)
NH
4
?
(mg l
-1
) 0.06
(0.04–0.07)
0.2 (0.05–0.5) 0.05 (0.04–0.05) 0.04 (0.03–0.06) 0.02 (0.01–0.04) 0.04 (0.03–0.04) 0.04
(0.03–0.05)
0.04
(0.02–0.05)
TP (mg l
-1
) 0.19
(0.15–023)
0.21
(0.15–0.28)
0.11 (0.06–0.16) 0.11 (0.06–0.15) 0.08 (0.07–0.09) 0.08 (0.08–0.09) 0.14
(0.1–0.22)
0.13
(0.08–0.22)
PO
43-
(mg l
-1
) 0.07
(0.05–0.09)
0.08
(0.05–0.12)
0.009
(0.008–0.009)
0.008
(0.008–0.008)
0.015
(0.009–0.025)
0.015
(0.010–0.025)
0.05
(0.03–0.07)
0.06
(0.03–0.17)
Alkalinity (meqv l
-1
) 2.8 (2.6–3.1) 2.5 (2.1–3.1) 2.3 (2.2–2.3) 2.1 (2.1–2.2) 2.0 (2.0–2.1) 2.0 (2.0–2.1) 2.8 (2.6–3) 2.7 (2.1–3.2)
Temperature is given as median values, with min–max range in parentheses. Bold indicates significant difference (at 0.05 alevel) between forested and open reaches within the
same stream
Nd not determined
Hydrobiologia
123
also evident separately for River Granslev (df
res
=36;
F=8.0; P\0.01), River Alsted (df
res
=21;
F=8.0; P\0.01) and Storkesig Brook (df
res
=44;
F=37; P\0.0001), when tested with one-way
ANOVA for each stream using the results from the
estimation of the autochthonous contribution to each
taxon sample. However, the autochthonous contribu-
tion was not significantly different between grassland
and forested reaches in River Skader using this
statistical approach (df
res
=29; F=0.47; P[0.05).
At the intraspecific level, the higher autochthonous
contribution in grassland reaches remained evident for
the different taxa tested (Fig. 4). The mayfly E. danica
had a higher autochthonous contribution to its biomass
in grassland than in forest reaches of Storkesig Brook
(df
res
=6; F=50; P\0.0001), and River Granslev
(df
res
=8; F=20.7; P\0.0001) Simuliidae sp. and
G. pulex had a higher autochthonous contribution to
their biomass in grassland than in forested reaches in
Storkesig Brook (ANOVA: df
res
=8; F=162;
Fig. 2 Biplot showing
isotopic signatures of
13
C
and
15
N of main-grouped
resources (autochthonous:
benthic algae) and
allochthonous (CPOM,
SPOM, FBOM),
invertebrates and fish
(brown trout) in grassland
and forested stream reaches
of four streams. Grey
symbols indicate signature
of each bulk sample
[allochthonous (x);
autochthonous (?);
invertebrates (o); and fish
(d)] and dark circles
represent mean values with
standard deviation of
isotopic signatures of each
of these groups. For details
of invertebrates collected
see Supplementary
Appendix Tables 1 and 2
Hydrobiologia
123
P\0.0001 for Simulidae and ANOVA: df
res
=8;
F=84.9; P\0.0001 for G. pulex), and G. pulex also
had a higher autochthonous contribution in grassland
than in forested reaches in River Granslev despite a
marginal Pvalue (df
res
=6; F=4.8; P=0.07;
Fig. 4). However this did not occur for Limnephilidae
caddisflies in River Skader (df
res
=4; F=4.0;
P[0.05; Fig. 4).
Riparian coverage was the main environmental
driver of changes in the energetic support to food
webs, being the only factor strongly and negatively
correlated to the mean autochthonous support to
macroinvertebrate biomass (Mixed-effects linear
regression df =3, F=39.1, P=0.01 r
2
=0.23,
Table 2). Additionally we found a tendency (with
marginal Pvalues) to decrease autochthonous support
to biomass with increasing stream width (df =3,
F=7.7, P=0.06 r
2
=0.52) and nitrogen concen-
trations (df =3, F=9.7, P=0.05 r
2
=0.26,
Table 2). The proportion of autochthonous generated
Fig. 3 Results of SIAR
Bayesian mixing models for
the estimation of
autochthonous and
allochthonous contribution
to biomass (95, 75 and 50%
credibility intervals, in
increasing colour tone
respectively) of primary
consumer invertebrates in
forested and grassland
reaches of four streams. Left
panels forested reaches;
right panels grassland
reaches for each stream.
From above to below, the
streams studied: River
Granslev, River Alsted,
Storkesig Brook, and River
Skader. The autochthonous
contribution is higher at
grassland reaches within the
same stream (although in
River Skader 95%
credibility intervals of
autochthonous contribution
overlapped greatly between
the forested and open reach)
Hydrobiologia
123
biomass was not related to macrophyte cover, benthic
algae biomass, stream depth, substrate cover, or total
phosphorous concentrations in our dataset (Table 2).
Food-chain length
Brown trout d
15
N from each reach type was signifi-
cantly different (one-way ANOVA, Pvalues
always \0.001, Table 3), being lower in forest than
in grassland reaches, probably reflecting the
differences in d
15
N of basal resources (Figs. 2,5;
Table 3). As expected, we thus validate the use of this
species in FCL estimation. However, opposite to our
expectations and despite the different contribution of
autochthonous and allochthonous resources in forest
and grassland reaches food webs, the FCL did not
differ between forest and grassland reaches neither
when all sites were analysed together (mixed-effects
ANOVA, df
res
=3; F=0.02; P=0.8) nor when
differences were compared within each stream (one-
Fig. 4 The autochthonous contribution to the biomass of the
same species present in forested and grassland reaches of each
stream based on SIAR Bayesian mixing model estimations for
each individual. The autochthonous contribution was highest in
the grassland compared to the forested reach for E. danica
(df
res
=6; F=50; P\0.0001), G. pulex (df
res
=8;
F=84.9; P\0.0001), and Simuliidae sp. (df
res
=8;
F=162; P\0.0001) in Storkesig Brook, E. danica (df
res
=8;
F=20.7; P\0.0001) and marginally for G. pulex in River
Granslev (df
res
=6; F=4.8; P=0.07), but not significant for
Limnephilidae sp. (df
res
=4; F=4.0; P[0.05) in River
Skader. Significant differences between forested and grassland
reaches are indicated by asterisk
Hydrobiologia
123
way ANOVA, P\0.05) (Table 3; Fig. 5). The only
environmental parameter found as a potential deter-
minant of FCL in our dataset was the proportion of fine
sediments in substrate, significantly decreasing FCL
(Mixed-effects linear regression df =3, F=33.4,
P=0.01 r
2
=0.36, Table 2).
Discussion
The autochthonous contribution to the biomass of
macroinvertebrates was higher in grassland than in
forest sections of the same stream as suggested in our
first hypothesis, and it was also the dominant fuelling
source in the grassland reach of three of the four
streams (generating slightly [50% of the biomass of
macroinvertebrates). Furthermore, the proportion of
autochthonous contribution to biomass of macroin-
vertebrates decreased with increasing canopy cover-
age. These findings contrast with the Riverine
Productivity Model postulates about metazoan food
webs being based on autochthonous sources indepen-
dent of its availability or canopy coverage, and
supports the previous theories and empirical studies
evidencing the patchy nature of carbon assimilation
dynamics in streams (Spencer et al., 2003; Junker &
Cross, 2014; Collins et al., 2016). In this sense, the
presence of riparian forest at the reach level promotes
an enhanced use of allochthonous resources by
primary consumers (Vannote et al., 1980; Spencer
et al., 2003; Junker & Cross, 2014; Collins et al.,
2016). Our findings also match with those of Leberfin-
ger et al. (2011) who compared the contribution of
autochthonous and allochthonous food resources with
Table 2 Changes in autochthonous contribution to macroinvertebrate biomass and food-chain length with varying environmental
parameters in grassland and forested stream reaches
Parameters Mixed-effects linear regression test parameters
Autochthonous contribution Mean trophic position
F;Pvalue; df r
2
F; df; Pvalue r
2
Canopy cover (%) 39.1; 0.01; 3 0.23 (-) 0.07; 0.79; 3
Width (m) 9.7; 0.05; 3 0.26 (-) 3.7; 0.14; 3
Depth (cm) 4.6; 0.12; 3 0.2; 0.6; 3
Sand (%) 0.2; 0.68; 3 33.4; 0.01; 3 0.36 (-)
Macrophyte cover (%) 3.1; 0.15; 3 0.6; 0.5; 3
Benthic chl.a (mg m
-2
) 4.5; 0.12; 3 0.6; 0.47; 3
TN (mg l
-1
)7.7; 0.06; 3 0.52 (-) 0.5; 0.54; 3
TP (mg l
-1
) 1.2; 0.35; 3 2.6; 0.2; 3
Relationships are tested as mixed-effects linear models including stream identity as a random factor, significant relationships are
marked in bold and marginal Pvalues remarked in italics, the coefficient of determination (r
2
) is given for significant relationships
Table 3 Comparison of isotopic signatures and trophic position of brown trout (Salmo trutta)
Stream nd
13
Cd
15
N Trophic position
Alsted 10 0.01; 8; [0.1 17.9; 8; <0.01 1.5; 8; [0.1
Granslev 7 20.2; 5; <0.01 11.3; 5; <0.01 3.1; 5; [0.1
Skader 9 1.8; 7; [0.1 6.9; 7; <0.05 3.8; 7; [0.1
Storkesig 10 17.1; 8; <0.01 51.2; 8; <0.0001 0; 8; [0.1
All streams 36 2.6; 34; [0.1 22.5; 34; <0.0001 0.3; 34; [0.1
Details on statistical test parameters are shown as follows: number of individuals analysed (n), ANOVA F, degrees of freedom of
residuals and Pvalue (F;df
residuals
;Pvalue) for tests within d
13
C, d
15
N, and trophic position in each stream when all forested and
grassland reaches are pooled. Significant differences are given in bold. For more information see Fig. 5
Hydrobiologia
123
shredder macroinvertebrate biomass in closed- and
open-canopy reaches in Sweden. These authors found
that although allochthonous resources were always the
most important food resource for macroinvertebrates
in both closed and open-canopy sections, the auto-
chthonous contribution to shredder biomass was
higher in open than in closed-canopy conditions
(Leberfinger et al., 2011).
The period before sampling (spring) was typical in
its temperature and precipitation being within the
average range of most previous years (1961–1990)
(Cappelen, 2011); however. we need to remark that
this was a one-season and one-year study, so the
generalisation about its application to other seasons
and regions should be avoided without further evi-
dence. Our results are representative of the time period
reflected by stable isotopes in consumer tissues, and
for many invertebrates this may be the period
2–3 weeks before sampling and about one month for
fish during summer in Danish streams (at least for
d
15
N) (Riis et al., 2012). Therefore, these results are
valid for the early summer season and not the whole
year. Based on a previous seasonal study of a forested
temperate stream in North America, we expect a
predominantly autochthonous-based primary con-
sumer biomass towards summer with rising temper-
atures and allochthonous predominance during
autumn, winter, and spring (Junker & Cross, 2014).
Despite the autochthonous contribution to inverte-
brates probably represented the maximum annual
level in our study, our results together with that of
related studies (e.g. Leberfinger et al., 2011; Junker &
A
B
C
Fig. 5 Variations in
isotopic signatures and
trophic position of brown
trout (S. trutta) in forested
and grassland reaches of the
studied streams. Forested
stream reaches are marked
in grey and grassland
reaches in white, significant
differences (ANOVA
P\0.05) are marked with
asterisk.AMean and
standard deviation of d
13
C
values of trout individuals;
d
13
C values are lower in
grassland than in forested
reaches in River Granslev
and Storkesig Brook.
BMean and standard
deviation of d
15
N values of
trout individuals; d
15
Nis
always higher in grassland
than in forested reaches.
CMean and standard
deviations of estimations of
trophic position of trout;
trophic positions do not
differ between forested and
grassland reaches. When all
forested and grassland
reaches are pooled in the
analysis, d
15
N is higher in
grassland than in forested
reaches. Details on statistics
can be found in Table 3
Hydrobiologia
123
Cross, 2014; Collins et al., 2016) do not support that
the hypothesis based on Riverine Productivity Model
(i.e. the largest fraction of riverine biomass being of
autochthonous origin) also apply to small-forested
streams (e.g. Brito et al., 2006; Mc Neely et al., 2006;
Li & Dudgeon, 2008; Lau et al., 2009b). Allochtho-
nous resources are frequently found to be prevalent
over autochthonous foods (e.g. in half of the scenarios
in our study), at least for macroinvertebrate assem-
blages in temperate forested streams, which contrasts
with the hypotheses of the Riverine Productivity
Model made for larger riverine systems (Thorp &
Delong, 2002).
Changes in energy assimilation between forested
and grassland sections occurred even for the same
species, as observed for G. pulex, E. danica and
Simuliidae sp. Therefore, the increase in the auto-
chthonous contribution to primary consumers with
increasing canopy openness does not necessarily arise
from a change in assemblage-specific composition (as
studied in the River Continuum Concept) but from
dietary changes within the same species. Our findings
of a plastic dietary assimilation of these invertebrate
species are in agreement with those of previous dietary
(e.g. Lo
´pez-Rodrı
´guez et al., 2009) and stable isotope
studies (e.g. Collins et al., 2016). Particularly, dietary
studies have shown that G. pulex is a well-known
generalist herbivore that is able to consume allochtho-
nous carbon from fungi and bacteria and also
autochthonous carbon sources, such as periphytic
algae, depending on food type availability (Moore,
1975; Grac¸a et al., 1993). The same applies to the filter
feeder E. danica (e.g. Austin & Baker, 1988;Lo
´pez-
Rodrı
´guez et al., 2009) and Simulidae (e.g. Burton,
1973; Moore, 1977) that filter particulate organic
matter from both allochthonous and autochthonous
sources located upstream present in the water column
(Moore, 1977; Wallace & Merritt, 1980).
The patterns of autochthonous and allochthonous
carbon assimilation in the biota of riverine systems
probably reflect a combination of changes in food
availability and quality. Thus, whenever the epiben-
thic algae biomass increased (towards the open-
canopy grassland sites) the proportion assimilated by
the invertebrates may have increased, probably due to
their higher nutritional quality compared to terrestrial
sources (Thorp & Delong, 2002; Lau et al., 2009b;
Marcarelli et al., 2011). We base this argument upon
the fact that terrestrial sources were readily available
at all the stream reaches regardless of type (reflected in
the similar % of debris in the stream bottom, Table 1),
but that epibenthic algae biomass was higher in
grassland reaches where the autochthonous consump-
tion increased with decreasing canopy coverage. This
interpretation of our findings also agrees with exper-
imental evidence demonstrating a preferential assim-
ilation of autochthonous C at the top of a food web
when both allochthonous and autochthonous C are
equally available (Lau et al., 2009c).
Food type availability driven by forest canopy
cover can explain a large part of the variability
observed regarding autochthonous or allochthonous
dominance of stream food webs (Finlay, 2001; Bunn
et al., 2003; Brito et al., 2006; Lau et al., 2009b).
However, despite that within the same river grassland
reaches have higher autochthonous contribution than
forest reaches, this relationship does not seem to be as
straightforward across all the study sites. For example,
the open section of River Granslev has [70% of
canopy coverage and the macroinvertebrate biomass is
mostly autochthonous fuelled, while the open-canopy
section of River Skader has \60% canopy coverage
and its biomass is largely allochthonous fuelled. This
suggests that the carbon dynamics in streams can be
highly patchy, and other factors than the presence of
riparian forest (e.g. stream reach morphology) may
drive the carbon assimilation pathways at a larger
scale (Sullivan, 2013). In this sense it is worth noting
that within our dataset, despite that canopy coverage
was the strongest predictor of the autochthonous
support to food webs, we also found weak tendencies
to decrease autochthonous contribution to inverte-
brates with increasing stream width and nitrogen
concentration. This could be attributed to the role of
local reach characteristics affecting this pattern. For
example the dominance of pool and/or riffle stream
habitats may determine whether allochthonous mate-
rial can be deposited being available for consumers or
if it is only being transported downstream (Sullivan,
2013). In this sense, the upstream characteristics can
also influence locally, and even though we tried to
control for this in our study, we could still have an
effect of stream flow direction in our results. For
example, the only forest reach with dominant auto-
chthonous contribution is River Granslev, where the
stream flows from the grassland to forest and high
availability of autochthonous matter drifting to the
forest section could explain this. The opposite is true
Hydrobiologia
123
for River Skader, where the streamflow comes from
forest to grassland direction and both grassland and
forest are mostly allochthonous dominated. Given this
high cross-system variability we argue that large scale
factors such as stream morphology and upstream
riparian patch size affect locally at a lower scale on the
physical environment (i.e. light irradiation and pres-
ence of different micro habitats) and food availability,
which may ultimately determine the patchy patterns of
carbon assimilation in small temperate streams.
Trophic position of brown trout was found to be
around three, corresponding well to an invertivorous
species and matching previous results for this species
in streams (e.g. Cucherousset et al., 2007; Kristensen
et al., 2016). Contrary to our expectations, the
contrasting isotopic signature of trout between forest
and grassland reaches did not represent a change in
FCL between stream reach types, and our second
hypothesis was therefore rejected. It needs to be
remarked, however, that the contrasting isotopic
signatures in brown trout of same-stream forested
and grassland sections support their sedentary beha-
viour having a home range of probably few-hundred
metres with the streams (at least in the last month
before sampling: time reflected by the isotopic
analysis).
Some of the main determinants of stream FCL are
known to be ecosystem size, productivity and distur-
bance regime (Pimm & Kitching, 1987; Post et al.,
2000; Thompson & Townsend, 2005; McHugh et al.,
2010; Sabo et al., 2010). Data from streams reviewed
in Warfe et al. (2013) show that a positive relationship
between FCL and ecosystem size is found very often
(observed in eight out of ten reviewed studies), and a
negative relationship between disturbance regime and
FCL is also common (found in seven out of eleven
study cases), but positive relationships between pro-
ductivity and FCL are less frequently found (in seven
out of sixteen studies reviewed there). This also
matches the observed for rivers, where the increase in
FCL with ecosystem area depends on decreasing
disturbance intensity with increasing area, but a
relationship with food resource availability is not
evident (Sabo et al., 2010). In our study (comparing
streams with similar ecosystem area and disturbance
regimes), the availability of detritus was similar in
both reach types, while the availability of benthic
algae was greater in grassland than in the forested
reaches (probably reflecting higher all together local
productivity in grassland streams), but this did not
translate into a difference in FCL between reach types.
A lack of a relationship between productivity and
FCL was reported several times before and sometimes
discussed as a consequence of the presence of mobile
species that translocate energy form the most produc-
tive patch to the less productive ones (e.g. Warfe et al.,
2013). This could apply to our study, especially
considering that the spatial extent of the study was of
few-hundred metres, and most of the studied species
could move such distances (in a longer time-scale than
reflected in stable isotope analysis), distributing
energy across the ecosystem patches.
Our result contrasts the findings of Lau et al.
(2009a) that tropical forest streams were at least one
trophic position shorter than similar open-canopy
streams. However, two main differences between the
methods applied to estimate trophic positions in Lau
et al. (2009a) and our study might partly account for
these differences. Firstly, they only used the organism
with the lowest d
15
N value regardless of its taxonomic
identity in each stream reach as base for estimating
trophic position, whereas we used the average of
primary consumers in each site. Secondly, Lau et al.
(2009a) collected and used different top predators
between stream reach types, while we always found
and used the same species (S. trutta). Thus, the
enlarged FCL revealed in Lau et al. (2009a) may be the
result of the inclusion of species consuming fish or
feeding selectively on predatory invertebrates in open-
canopy reaches and not necessarily a change in trophic
strategy of the same species, as tested here.
In our study site, independently of riparian
cover, FCL was found related to substrate cover-
age, decreasing as the sand coverage in the stream
bottom increased. Sand-substrates could be consid-
ered as more homogeneous than coarser substrates,
which create more micro-scale heterogeneity of
flow conditions and represent high refuge and food
resource availability increasing diversity of
macroinvertebrates (e.g. Kovalenko et al., 2012).
Moreover, a higher dominance of this substrate
type implies a lower diversity of substrate types (a
well-known proxy for habitat heterogeneity in
streams, e.g. Palmer et al., 2009; Kovalenko
et al., 2012). Higher habitat heterogeneity usually
promotes a higher functional and taxonomic diver-
sity creating more trophic niches, partly due to
attenuation of flow disturbances in streams (e.g.
Hydrobiologia
123
Warfe et al., 2008; Kovalenko et al., 2012). Thus,
it seems reasonable that an increasing dominance of
sand in the stream-bottom decrease habitat hetero-
geneity producing shorter FCL because more
exposure to disturbances and less trophic diversity
usually means simpler (and more likely shorter)
food-chain lengths (Kovalenko et al., 2012).
In our study, we used paired grassland and forest
stream reaches to demonstrate that closed-canopy
coverage in forest reaches can alter energy pathways
in streams. We have shown that these changes in
subsidies to the whole macroinvertebrate biomass
occur also at intraspecific levels (within same species),
suggesting that the change in energetic sources for
macroinvertebrate biomass can be independent of a
change in species composition. The study also high-
lighted the need for continued field and experimental
research into mechanisms driving resource subsidies
in streams to unveil ecosystem level consequences of
changed subsidies at the food-web base. This becomes
particularly important in streams where riparian forest
planting is used as a measure to mitigate agricultural
and climate-induced impacts on the ecosystems (e.g.
Broadmeadow et al., 2011) or when testing the effect
of clear-cutting or wildfire removing riparian areas.
Our study shows that riparian forest may not only
change stream community structure, as reported
before in previous studies (e.g. Gregory et al., 1991;
Moldenke & Ver Linden, 2007; Teixeira-de Mello
et al., 2016), but may also affect the resource subsidies
and food-web structure with potential ecosystem
consequences. A greater understanding of this subject
will improve the implementation of riparian buffer
corridors, thus avoiding adverse consequences for
stream biodiversity and functioning. For example,
even though invertebrates can exploit a wide diversity
of autochthonous and allochthonous foods, autochtho-
nous-derived carbon seems essential during the crit-
ical life stages of many invertebrates in spring and
summer (Junker & Cross, 2014).
Acknowledgements The authors are grateful for financial
support from The Danish Natural Science Research Council (T.
Riis Grant #272-09-0012), the Carlsberg Foundation (T. Riis
Grant #2013_01_0258) and the European Union 7th Framework
projects REFRESH under Contract No. 244121 and MARS
under Contract No. 603378 (A. Baattrup-Pedersen).
I. Gonza
´lez-Bergonzoni received support from SNI (Agencia
Nacional de Investigacio
´n e Innovacio
´n, ANII, Uruguay).
References
Anderson, M. J., 2001. A new method for non-parametric
multivariate analysis of variance. Austral Ecology 26:
32–46.
Austin, D. A. & J. H. Baker, 1988. Fate of bacteria ingested by
larvae of the freshwater mayfly, Ephemera danica.
Microbial Ecology 15: 323–332.
Bachmann, R. A., 1984. Foraging behavior of free-ranging wild
and hatchery brown trout in a stream. Transactions of the
American Fisheries Society 113: 1–32.
Belicka, L., E. Sokol, J. M. Hoch, R. Jaffe
´& J. Trexler, 2012. A
molecular and stable isotopic approach to investigate algal
and detrital energy pathways in a freshwater marsh. Wet-
lands 32: 531–542.
Brito, E. F., T. P. Moulton, M. L. De Souza & S. E. Bunn, 2006.
Stable isotope analysis indicates microalgae as the pre-
dominant food source of fauna in a coastal forest stream,
south-east Brazil. Austral Ecology 31: 623–633.
Brix, H. & H. H. Schierup, 2001. Limnologi, Analysefor-
eskrifter (in Danish). Aarhus University, Aarhus,
Denmark.
Broadmeadow, S. B., J. G. Jones, T. E. L. Langford, P. J. Shaw
& T. R. Nisbet, 2011. The influence of riparian shade on
lowland stream water temperatures in southern England
and their viability for brown trout. River Research and
Applications 27: 226–237.
Bunn, S. E., P. M. Davies & M. Winning, 2003. Sources of
organic carbon supporting the food web of an arid zone
floodplain river. Freshwater Biology 48: 619–635.
Bunn, S. E., C. Leigh & T. D. Jardine, 2013. Diet-tissue frac-
tionation of d
15
N by consumers from streams and rivers.
Limnology and Oceanography 58: 765–773.
Burton, G. J., 1973. Feeding of Simulium hargreavesi gibbins
larvae on Oedegonium algal filaments in Ghana. Journal of
Medical Entomology 10(1): 101–106.
Cappelen, J., 2011. Hvordan var det nu det var vejret i 2010?
Vejret 127: 1–19. (In Danish).
Chen, J. M. & T. A. Black, 1992. Defining leaf area index for
non-flat leaves. Plant, Cell and Environment 15: 421–429.
Collins, S. M., T. J. Kohler, S. A. Thomas, W. W. Fetzer & A.
S. Flecker, 2016. The importance of terrestrial subsidies in
stream food webs varies along a stream size gradient. Oikos
125(5): 674–685.
Cucherousset, J., J. C. Aymes, F. Santoul & R. Ce
´re
´ghino, 2007.
Stable isotope evidence of trophic interactions between
introduced brook trout Salvelinus fontinalis and native
brown trout Salmo trutta in a mountain stream of south-
west France. Journal of Fish Biology 71: 210–223.
Cummins, K. W., 1974. Structure and function of stream
ecosystems. BioScience 24: 631–641.
Cummins, K. W. & M. J. Klug, 1979. Feeding ecology of stream
invertebrates. Annual Review of Ecology and Systematics
10: 147–172.
Dekar, M. P., R. S. King, J. A. Back, D. F. Whigham & C.
M. Walker, 2012. Allochthonous inputs from grass-domi-
nated wetlands support juvenile salmonids in headwater
streams: evidence from stable isotopes of carbon, hydro-
gen, and nitrogen. Freshwater Science 31: 121–132.
Hydrobiologia
123
Dobson, M., S. Pawley, M. Fletcher & A. Powell, 2012. Guide to
freshwater invertebrates. Freshwater Biological
Association.
Finlay, J. C., 2001. Stable-carbon-isotope ratios of river biota:
implications for energy flow in lotic food webs. Ecology
82: 1052–1064.
Friberg, N., A. Baattrup-Pedersen, M. Pedersen & J. Skriver,
2005. The new Danish stream monitoring programme
(NOVANA) – preparing monitoring activities for the water
framework directive era. Environmental Monitoring
Assessment 111: 27–42.
Fry, B., 2006. Stable isotope ecology. Springer, New York,
USA.
Fry, B., 2013. Alternative approaches for solving underdeter-
mined isotope mixing problems. Marine Ecology Progress
Series 472: 1–13.
Fu
¨reder, L., C. Welter & J. K. Jackson, 2003. Dietary and
stable isotope (d
13
C, d
15
N) analyses in alpine stream
insects. International Review of Hydrobiology 88:
314–331.
Gonza
´lez-Bergonzoni, I., F. Landkildehus, M. Meerhoff, T.
L. Lauridsen, K. O
¨zkan, T. A. Davidson, N. Mazzeo & E.
Jeppesen, 2014. Fish determine macroinvertebrate food
webs and assemblage structure in Greenland subarctic
streams. Freshwater Biology 59: 1830–1842.
Grac¸a, M. A. S., L. Maltby & P. Calow, 1993. Importance of
fungi in the diet of Gammarus pulex and Asellus aquaticus.
Oecologia 96: 304–309.
Gregory, S. V., F. J. Swanson, W. A. McKee & K. W. Cummins,
1991. An ecosystem perspective of riparian zones. BioS-
cience 41: 540–551.
Hart, E. A. & J. R. Lovvorn, 2003. Algal vs. macrophyte inputs
to food webs of inland saline wetlands. Ecology 84:
3317–3326.
Hussey, N. E., M. A. MacNeil, B. C. McMeans, J. A. Olin, S.
F. J. Dudley, G. Cliff, S. P. Wintner, S. T. Fennessy & A.
T. Fisk, 2014. Rescaling the trophic structure of marine
food webs. Ecology Letters 17(2): 239–250.
Iglesias, C., M. Meerhoff, L. Johansson, I. Gonzalez-Ber-
gonzoni, N. Mazzeo, J. P. Pacheco, F. Teixeira de Mello, G.
Goyenola, T. Lauridsen, M. Søndergaard, T. A. Davidson
& E. Jeppesen, 2016. Stable isotope analysis confirms
substantial differences between subtropical and temperate
shallow lake food webs. Hydrobiologia 784(1): 111–123.
Jardine, T. D., W. L. Hadwen, S. K. Hamilton, S. Hladyz, S.
M. Mitrovic, K. A. Kidd, W. Y. Tsoi, M. Spears, D.
P. Westhorpe, V. M. Fry, F. Sheldon & S. E. Bunn, 2014.
Understanding and overcoming baseline isotopic variabil-
ity in running waters. River Research and Applications
30(2): 155–165.
Jardine, T. D., R. J. Hunt, S. J. Faggotter, D. Valdez, M.
A. Burford & S. E. Bunn, 2013. Carbon from periphyton
supports fish biomass in waterholes of a wet–dry tropical
river. River Research and Applications 29: 560–573.
Jensen, J. L. & K. Frost, 1992. Fagdatacenter for hydrometriske
data, hedeselskabet. Hydrometrisk feltarbejde 10: 52. (in
Danish).
Junker, J. R. & W. F. Cross, 2014. Seasonality in the trophic
basis of a temperate stream invertebrate assemblage:
importance of temperature and food quality. Limnology
and Oceanography 59: 507–518.
Kovalenko, K. A., S. M. Thomas & D. M. Warfe, 2012. Habitat
complexity: approaches and future directions. Hydrobi-
ologia 685: 1–17.
Kristensen, P. B., E. A. Kristensen, T. Riis, A. J. Baisner, S.
E. Larsen, P. F. M. Verdonschot & A. Baattrup-Pedersen,
2014. Riparian forest as a management tool for moderating
future thermal conditions of lowland temperate streams.
Inland Waters 5: 27–38.
Kristensen, P. B., T. Riis, P. B. Dylmer, E. A. Kristensen, M.
Meerhoff, B. Olesen, F. Teixeira de Mello, A. Baattrup-
Pedersen, G. Cavalli & E. Jeppesen, 2016. Baseline iden-
tification in stable-isotope studies of temperate lotic sys-
tems and implications for calculated trophic positions.
Freshwater Science 35(3): 909–921.
Lau, D. C. P., K. M. Y. Leung & D. Dudgeon, 2009a. What does
stable isotope analysis reveal about trophic relationships
and the relative importance of allochthonous and auto-
chthonous resources in tropical streams? A synthetic study
from Hong Kong. Freshwater Biology 54: 127–141.
Lau, D. C. P., K. M. Y. Leung & D. Dudgeon, 2009b. Are
autochthonous foods more important than allochthonous
resources to benthic consumers in tropical headwater
streams? Journal of the North American Benthological
Society 28: 426–439.
Lau, D. P., K. Y. Leung & D. Dudgeon, 2009c. Evidence of
rapid shifts in the trophic base of lotic predators using
experimental dietary manipulations and assimilation-based
analyses. Oecologia 159: 767–776.
Leberfinger, K., I. Bohman & J. A. N. Herrmann, 2011. The
importance of terrestrial resource subsidies for shredders in
open-canopy streams revealed by stable isotope analysis.
Freshwater Biology 56: 470–480.
Levin, L. A. & C. Currin, 2012. Stable isotope protocols: sam-
pling and sample procesing Scripps Institution of
Oceanography Technical Report. eScholarship, University
of California.
Li, A. O. Y. & D. Dudgeon, 2008. Food resources of shredders
and other benthic macroinvertebrates in relation to shading
conditions in tropical Hong Kong streams. Freshwater
Biology 53: 2011–2025.
Lo
´pez-Rodrı
´guez, M. J., J. M. Tierno de Figueroa & J. Alba-
Tercedor, 2009. Life history of two burrowing aquatic
insects in southern Europe: leuctra geniculata (Insecta:
Plecoptera) and Ephemera danica (Insecta: Ephe-
meroptera). Aquatic Insects 31: 99–110.
Marcarelli, A. M., C. V. Baxter, M. M. Mineau & R. O. Hall,
2011. Quantity and quality: unifying food web and
ecosystem perspectives on the role of resource subsidies in
freshwaters. Ecology 92: 1215–1225.
McHugh, P. A., A. R. McIntosh & P. G. Jellyman, 2010. Dual
influences of ecosystem size and disturbance on food chain
length in streams. Ecology Letters 13: 881–890.
McNeely, C., S. M. Clinton & J. M. Erbe, 2006. Landscape
variation in C sources of scraping primary consumers in
streams. Journal of the North American Benthological
Society 25: 787–799.
Mey, W., 1997. Nilsson, A. (ed), Aquatic Insects of North
Europe. A taxonomic handbook. Vol. 1: Ephemeroptera.
Plecoptera, Heteroptera, Megaloptera, Neuroptera,
Coleoptera, Trichoptera and Lepidoptera. 1996. Hard-
bound, Apollo Books, Denmark.
Hydrobiologia
123
Minshall, G. W., 1967. Role of allochthonous detritus in the
trophic structure of a woodland springbrook community.
Ecology 48: 139–149.
Moldenke, A. R. & C. Ver Linden, 2007. Effects of clearcutting
and riparian buffers on the yield of adult aquatic
macroinvertebrates from headwater streams. Forest Sci-
ence 53: 308–319.
Monteith, J. L. & M. H. Unsworth, 1973. Principles of envi-
ronmental physics, 2nd ed. Edward Arnold, London, UK.
Moore, J. W., 1975. The role of algae in the diet of Asellus
aquaticus L. and Gammarus pulex L. Journal of Animal
Ecology 44: 719–730.
Moore, J. W., 1977. Some factors effecting algal consumption in
subarctic ephemeroptera, plecoptera and simuliidae.
Oecologia 27: 261–273.
Moulton, T. P., 2006. Why the world is green, the waters are
blue and food webs in small streams in the Atlantic rain-
forest are predominantly driven by microalgae? Oecologia
Australis 10: 78–89.
Nakagawa, S. & H. Schielzeth, 2013. A general and simple
method for obtaining R2 from generalized linear mixed-
effects models. Methods in Ecology and Evolution 4(2):
133–142.
Newman, R. M., 1991. Herbivory and detritivory on freshwater
macrophytes by invertebrates: a review. Journal of the
North American Benthological Society 10: 89–114.
Noel, D., C. W. Martin & C. A. Federer, 1986. Effects of forest
clearcutting in New England on stream macroinvertebrates
and periphyton. Environmental Management 10: 661–670.
Ovidio, M., E. Baras, D. Goffaux, F. Giroux & J. C. Philippart,
2002. Seasonal variations of activity pattern of brown trout
(Salmo trutta) in a small stream, as determined by radio-
telemetry. Hydrobiologia 470(1): 195–202.
Palmer, M. A., H. L. Menninger & E. Bernhardt, 2009. River
restoration, habitat heterogeneity and biodiversity: a failure
of theory or practice? Freshwater Biology 55(1): 1–18.
Parnell, A., R. Inger & S. Bearhop, 2010. Source partitioning
using stable isotopes: coping with too much variation.
PLoS ONE 5(3): e9672.
Parnell, A., R. Inger, S. Bearhop & A. L. Jackson, 2013a. SIAR:
Stable isotope analysis in R.
Parnell, A. C., D. L. Phillips, S. Bearhop, B. X. Semmens, E.
J. Ward, J. W. Moore, A. L. Jackson, J. Grey, D. J. Kelly &
R. Inger, 2013b. Bayesian stable isotope mixing models.
Environmetrics 24: 387–399.
Pedersen, B., 2004. NOVANA, Teknisk anvisning for marin
overva
˚gning, 2.3 klorofyl a. I Tekniske anvisninger for
marin overva
˚gning. Miljøministeriet, Danmarks Miljøun-
dersøgelser (In Danish).
Phillips, D. L., 2012. Converting isotope values to diet com-
position: the use of mixing models. Journal of Mammalogy
93: 342–352.
Pimm, S. L. & R. L. Kitching, 1987. The determinants of food
chain lengths. Oikos 50: 302–307.
Post, D. M., 2002. The long and short of food-chain length.
Trends in Ecology & Evolution 17: 269–277.
Post, D. M., M. L. Pace & N. G. Hairston, 2000. Ecosystem size
determines food-chain length in lakes. Nature 405:
1047–1049.
R Core Team (2014) R: a language and environment for sta-
tistical computing. R foundation for statistical computing,
Vienna, Austria. ISBN 3-900051-07-0, http://www.R-
project.org/.
Riis, T., K. Dodds, P. B. Kristensen & A. J. Baisner, 2012.
Nitrogen cycling and dynamics in a macrophyte-rich
stream as determined by a
15
N-NH
?4
release. Freshwater
Biology 57: 1579–1591.
Rounick, J. S., M. J. Winterbourn & G. L. Lyon, 1982. Differ-
ential utilization of allochthonous and autochthonous
inputs by aquatic invertebrates in some New Zealand
Streams: a stable carbon isotope study. Oikos 39: 191–198.
Sabo, J. L., J. C. Finlay, T. Kennedy & D. M. Post, 2010. The
role of discharge Variation in scaling of drainage area and
food chain length in rivers. Science 330(6006): 965–967.
Sullivan, S. M. P., 2013. Stream foodweb d
13
C and geomor-
phology are tightly coupled in mountain drainages of
northern Idaho. Freshwater Science 32(2): 606–621.
Sullivan, S. M. P., K. Hossler & C. M. Cianfrani, 2015.
Ecosystem structure emerges as a strong determinant of
food-chain length in linked stream–riparian ecosystems.
Ecosystems 18(8): 1356–1372.
Spencer, C. N., K. O. Gabel & F. R. Hauer, 2003. Wildfire
effects on stream food webs and nutrient dynamics in
Glacier National Park, USA. Forest Ecology and Man-
agement 178: 141–153.
Teixeira de Mello, F., M. Meerhoff, I. Gonza
´lez-Bergonzoni, E.
A. Kristensen, A. Baattrup-Pedersen & E. Jeppesen, 2016.
Influence of riparian forests on fish assemblages in tem-
perate. Environmental Biology of Fishes 99(1): 133–144.
Thompson, R. M. & C. R. Townsend, 2003. Impacts on stream
food webs of native and exotic forest: an intercontinental
comparison. Ecology 84: 145–161.
Thompson, R. M. & C. R. Townsend, 2005. Energy availability,
spatial heterogeneity and ecosystem size predict food-web
structure in streams. Oikos 108: 137–148.
Thompson, R. M., N. R. Phillips & C. R. Townsend, 2009.
Biological consequences of clear-cut logging around
streams – moderating effects of management. Forest
Ecology and Management 257: 931–940.
Thorp, J. H. & M. D. Delong, 1994. The riverine productivity
model: an heuristic view of carbon sources and organic
processing in large river ecosystems, vol 70. Blackwell,
Oxford, ROYAUME-UNI.
Thorp, J. H. & M. D. Delong, 2002. Dominance of auto-
chthonous autotrophic carbon in food webs of hetero-
trophic rivers. Oikos 96: 543–550.
Vander Zanden, M. J., G. Cabana & J. B. Rasmussen, 1997.
Comparing trophic position of freshwater fish calculated
using stable nitrogen isotope ratios (d
15
N) and literature
dietary data. Canadian Journal of Fisheries and Aquatic
Sciences 54: 1142–1158.
Vannote, R. L., G. W. Minshall, K. W. Cummins, J. R. Sedell &
C. E. Cushing, 1980. The river continuum concept. Cana-
dian Journal of Fisheries and Aquatic Sciences 37:
130–137.
Wallace, J. B. & R. W. Merritt, 1980. Filter-feeding ecology of
aquatic insects. Annual Review of Entomology 25:
103–132.
Warfe, D. M., L. A. Barmuta & S. Wotherspoon, 2008. Quan-
tifying habitat structure: surface convolution and living
space for species in complex environments. Oikos 117:
1764–1773.
Hydrobiologia
123
Warfe, D. M., T. D. Jardine, N. E. Pettit, S. K. Hamilton, B.
J. Pusey, S. E. Bunn, P. M. Davies & M. M. Douglas, 2013.
Productivity, disturbance and ecosystem size have no
influence on food chain length in seasonally connected
rivers. PLoS ONE 8(6): e66240.
Welles, J. M. & J. M. Norman, 1991. Instrument for indirect
measurement of canopy architecture. Agronomy Journal
83(5): 818–825.
Winemiller, K. O., A. S. Flecker & D. J. Hoeinghaus, 2010.
Patch dynamics and environmental heterogeneity in lotic
ecosystems. Journal of the North American Benthological
Society 29(1): 84–99.
Whiting, D. P., M. R. Whiles & M. L. Stone, 2011. Patterns of
macroinvertebrate production, trophic structure, and
energy flow along a tallgrass prairie stream continuum.
Limnology and Oceanography 56: 887–898.
Zuur, A. F., E. N. Ieno, N. J. Walker, G. M. Smith & A.
A. Saveliev, 2009. Mixed effects models and extensions in
ecology with R. Springer, New York.
Hydrobiologia
123
A preview of this full-text is provided by Springer Nature.
Content available from Hydrobiologia
This content is subject to copyright. Terms and conditions apply.