Content uploaded by Josef Hejzlar
Author content
All content in this area was uploaded by Josef Hejzlar on Nov 03, 2015
Content may be subject to copyright.
SHALLOW LAKES
The influence of nutrient loading, climate and water depth
on nitrogen and phosphorus loss in shallow lakes: a pan-
European mesocosm experiment
Jan Coppens
.
Josef Hejzlar
.
Michal S
ˇ
orf
.
Erik Jeppesen
.
S¸ eyda Erdog
˘
an
.
Ulrike Scharfenberger
.
Aldoushy Mahdy
.
Peeter No
˜
ges
.
Arvo Tuvikene
.
Didier L. Baho
.
Cristina Trigal
.
Eva Papastergiadou
.
Kostas Stefanidis
.
Saara Olsen
.
Meryem Bekliog
˘
lu
Received: 24 August 2015 / Revised: 14 September 2015 / Accepted: 18 September 2015
Ó Springer International Publishing Switzerland 2015
Abstract Losses of phosphorus (P) and nitrogen
(N) have important influences on in-lake concentra-
tions and nutrient loading to downstream ecosystems.
We performed a series of mesocosm experiments
along a latitudinal gradient from Sweden to Greece to
investigate the factors influencing N and P loss under
different climatic conditions. In six countries, a
standardised mesocosm experiment with two water
depths and two nutrient levels was conducted concur-
rently between May and November 2011. Our results
showed external nutrient loading to be of key impor-
tance for N and P loss in all countries. Almost all
dissolved inorganic nitrogen (DIN) and soluble reac-
tive phosphorus (SRP) were lost or taken up in
biomass in all mesocosms. We found no consistent
effect of temperature on DIN and SRP loss but a
significant, though weak, negat ive effect of tempera-
ture on total nitrogen (TN) and total phosphorus (TP)
loss in the deeper mesocosms, probably related to
higher organic N and P accumulation in the water in
the warmer countries. In shallow mesocosms, a
positive trend in TN and TP loss with increasing
temperature was observed, most likely related to
macrophyte growth.
Keywords Nutrient retention Nutrient budget
Shallow lake Organic matter Temperature
Guest editors: M. Bekliog
˘
lu, M. Meerhoff, T. A. Davidson,
K. A. Ger, K. E. Havens & B. Moss / Shallow Lakes in a
Fast Changing World
J. Coppens (&) S¸ . Erdog
˘
an M. Bekliog
˘
lu (&)
Limnology Laboratory, Department of Biology, Middle
East Technical University, Universiteler Mahallesi,
Dumlupinar Bulvarı, No. 1, 06800 Ankara, Turkey
e-mail: jan.coppens@metu.edu.tr
M. Bekliog
˘
lu
e-mail: meryem@metu.edu.tr
J. Hejzlar M. S
ˇ
orf
Institute of Hydrobiology, Biology Centre of the Czech
Academy of Sciences, Na Sa
´
dka
´
ch 7,
370 05 Ceske Budejovice, Czech Republic
M. S
ˇ
orf
Faculty of Science, University of South Bohemia,
Branis
ˇ
ovska
´
31, 370 05 Ceske Budejovice, Czech
Republic
E. Jeppesen S. Olsen
Department of Bioscience, Aarhus University, Vejlsøvej
25, 8600 Silkeborg, Denmark
E. Jeppesen S. Olsen
Sino-Danish Centre for Education and Research, The
University of Chinese Academy of Sciences, Beijing,
China
S¸ . Erdog
˘
an
Department of Biology, Faculty of Science and Art,
Bozok University, 66900 Yozgat, Turkey
U. Scharfenberger
Leibniz-Institute of Freshwater Ecology and Inland
Fisheries, Mu
¨
ggelseedamm 310, 12587 Berlin, Germany
123
Hydrobiologia
DOI 10.1007/s10750-015-2505-9
Introduction
Phosphorus (P) and nitrogen (N) are considered to play
key roles as limiting factors for primary production in
lakes (Likens, 1972; Schindler, 1974; Guildford &
Hecky, 2000;Lewis&Wurtsbaugh,2008). The role of
N is subject to debate (Schindler et al., 2008; Scott &
McCarthy, 2010, 2011; Paterson et al., 2011;Mossetal.,
2013), but there is no doubt that it has importance.
Human-induced processes have markedly altered the
availability of P and N (Vitousek et al., 1997) and have
increased the nutrient loading to many lakes, resulting in
eutrophication (Hasler, 1947;Correl,1998; Smith et al.,
1999) and focussing attention on nutrient relationships.
Some of the nutrients entering a lake may be
retained in the sediments or organisms (N and P) or lost
to the atmosphere (N) (Vollenweider, 1976), thereby
reducing export from the lake to downstream ecosys-
tems (Wetzel, 2001). Unfortunately, the word ‘reten-
tion’ has been used to include loss to the atmosphere
and this is potentially confusing so in this study, we use
‘nutrient loss’ or ‘nutrient change’ to refer to the
removal of any form of N and P from the water column
through a combination of sedimentation, assimilation
by organisms (e.g. algae, plants, periphyton) or loss to
the atmosphere (denitrification) (Correl, 1998; Wetzel,
2001). Such remov al may be temporary since N and P,
accumulated in the sediment, for example, can be
resuspended by sediment disturbance or released back
to the water column by diffusion or decomposition of
organic matter in the sediment (Søndergaard et al.,
2003). The permanent loss of N
2
-gas to the atmosphere
through denitrification, however, contributes substan-
tially to N loss in lakes (Seitzinger et al., 2006).
External nutrient loading, water temperature,
hydraulic residence time, lake depth, phytoplankton
and macrophyte abundance, organic matter content in
the sediment and oxygen concentration are all factors
that may determine the nutrient concentrations main-
tained in the water column (Saunders & Kalff, 2001a)
and also affect nutrient loss (Windolf et al., 1996;
Søndergaard et al., 2003; Seitzinger et al., 2006).
There is a strong, positive correlation between external
loading and nutrient loss, but the relative importance
of other factors is less well established. Hydraulic
residence time and water depth, for example, affect the
duration and extent of the contact between the water
column and the sediment so that sediment uptake
increases with residence time (Brett & Benjamin,
2008) but decreases with water depth (Windolf et al.,
1996). Low water depth may also increase plant
growth through increased light availability (Bucak
et al., 2012). Submerged macrophytes take up nutri-
ents directly from the water column and the sediment
and can influence the exchange of nutrients through
oxygen release or uptake, reduction of water move-
ment (Madsen et al., 2001) and stabilisation of the
sediment (Stephen et al., 1997). Macrophytes also
support denitrification by increasing the supply of
organic carbon to denitrifying bacteria (We isner et al.,
1994). N removal from lakes is often also higher at
high P concentrations (Finlay et al., 2013; Olsen et al.,
2015) because of increased algal production, leading
to higher N uptake (Schindler, 2012 ), increased
sedimentation of this algal biomass (Small et al.,
2014) and increased denitrification (Saunders & Kalff,
2001a; Seitzinger et al., 2006).
U. Scharfenberger
Department of Biology, Chemistry, Pharmacy, Freie
Universita
¨
t Berlin, Takustraße 3, 14195 Berlin, Germany
A. Mahdy
Department of Zoology, Faculty of Science, Al-Azhar
University (Assiut Branch), Assiut 71524, Egypt
P. No
˜
ges A. Tuvikene
Centre for Limnology, Institute of Agricultural and
Environmental Sciences, Estonian University of Life
Sciences, 61117 Rannu vald, Tartumaa, Estonia
D. L. Baho C. Trigal
Department of Aquatic Sciences and Assessment,
Swedish University of Agricultural Sciences,
P.O. Box 7050, 750 07 Uppsala, Sweden
C. Trigal
Species Information Center, Swedish University of
Agricultural Sciences, P.O. Box 7007, 750 07 Uppsala,
Sweden
E. Papastergiadou K. Stefanidis
Department of Biology, University of Patras, University
Campus, 26504 Rio, Greece
M. Bekliog
˘
lu
Kemal Kurdas¸¸ Ecological Research and Training
Stations, Lake Eymir, Middle East Technical University,
Ankara, Turkey
Hydrobiologia
123
Climate may also have a profound effect on N and P
loss in lakes (Jeppesen et al., 2011). Precipitation and
evaporation drive the water balance and can influence
water depth, hydraulic residence time and external
loading, while temperature affects biochemical reac-
tions including denitrification and other microbial
processes. N loss is assumed to increase with increas-
ing temperature because water temperature positively
influences the activity of the denitrifying bacteria
(Herrman et al., 2008; Veraart et al., 2011). Mineral-
isation of organic matter and release of inorganic P
have also been shown to increase with increasing
temperatures, potentially leading to lower net nutrient
loss (Gomez et al., 1998). Higher temperatures may
also enhance primary production and thereby increase
the uptake of N and P by macrophytes and algae,
though higher growth rates at higher temperature may
alter the nutrient content of organisms and lower their
nutrient demand (Woods et al., 2003). The nutrient
contents of phytoplankton and periphyton are also
significantly dependent on nutrient concentrations in
the water and on light availability (Sterner et al., 1997;
Danger et al., 2008).
Many studies have quantified N and P loss in lakes
and wetlands, showing maximum capacities above
100 g m
-2
a
-1
for N (Persson & Wittgren, 2003;
Strand & Weisner, 2013) and 5 g m
-2
a
-1
for P
(Fisher & Acreman, 2004). Most of the study sites
have been in the temperate zone (e.g. Jensen et al.,
1990; Windolf et al., 1996; Kaste & Dillon, 2003;
No
˜
ges, 2005; Han et al., 2011) and only few in
Mediterranean and semi-arid regions (Romero et al.,
2002; Cook et al., 2010;O
¨
zen et al., 2010). Climate
change is expected to reinforce the consequences of
eutrophication (Jeppesen et al., 201 0 ; Moss et al.,
2011), and in Mediterranean climates high tempera-
tures and high evaporation have been shown to affect
lake eutrophication and nutrient loss (O
¨
zen et al.,
2010). Given the predicted effects of climate change,
more studies of the relationship between temper ature
and nutrient processing in warmer climate zones are
needed. In particular, knowledge of the relative
importance of climate-driven effects on N and P loss
is still very general (Olsen et al., 2015) because most
studies have been descriptive and calculated nutrient
budgets based on lake monitoring data and related
nutrient loss to environmental factors. While whole-
lake studies can present comprehensive data on
specific lakes, comparisons among lakes along a
climate gradient are difficult to make owing to
confounding local effects, lack of controlled nutrient
loading and lack of replicability.
In this study, we used a controlled experimental
environment using mesocosms with a space -for-time
approach, applying latitude as a surrogate for time, to
compare the impact of nutrient loading, nutrient
concentrations and water depth on N and P loss under
different climatic conditions. We performed concur-
rent, standardised mesocosm experiments in six
countries along a latitudinal climate gradient from
Sweden to Greece. This approach allowed us to study
the relative importance of climate compared with
nutrient loading and nutrient concentration, at differ-
ent water depths complementing the study by Olsen
et al. (2015) focusing on nitrogen loss in a collection of
independent mesocosm experiments.
Our specific hypotheses were firstly that increased
temperature and lower water volume would reduce
loss of P significantly (because of greater release from
the sediments) and that we therefore with global
warming may expect an increase in the symptoms of
eutrophication within a phosphorus-limited lake as
well as greater export downstream of phosphorus. Our
second hypothesis was that greater temperatures and
reduced water level would promote denitrification and
therefore increased loss of nitrogen. The concomitant
of these hypotheses, if supported, would be a trend
towards nitrogen as the key limit ing nutrient in
shallow lakes.
Methods
Experimental design
The experiments were conducted between May and
November 2011 in six European countries covering a
wide climate gradient and following a standardised
construction and sampling protocol (Landkildehus
et al., 2014). Mesocosms were established in Lake
Erken in Sweden, Lake Vo
˜
rtsja
¨
rv in Estonia, Lake
Mu
¨
ggelsee in Germany, Experimental Pond FROV JU
Vodn
ˇ
any in the Czech Republic, ODTU
¨
DSI 50-Yıl
dam lake in Turkey and Lake Lysimachia in Greece
(Table 1
). The setup included a pontoon bridge with
sidewalks, holding 16 closed fiberglass mesocosms
each with a diameter of 1.2 m suspended in the lakes.
The experiment was run at two contrasting depths,
Hydrobiologia
123
with a 2 9 4 factorial design with two nutrient levels
and four replicates at each depth. Since the mesocosms
were made of fiberglass, no water was exchanged with
the surrounding lake or the lake sediment.
The mesocosms were initially filled with 10 cm
sediment composed of 90% washed sand and 10%
natural sediment from a nearby oligotrophic lake in
each country. Largely artificial sediment was chosen
to obtain similar conditions in all countries in order to
minimise the effect of local sediment characteristics.
The sediment was sieved through a 10-mm mesh and
large particles were removed (e.g. plant fragments,
mussels, stones, debris). In the autumn and winter
prior to the experiment, the sediment was stored in two
tanks below a water layer of 20–50 cm. In one tank,
the concentration of the overlying water layer was
25 lgPl
-1
to prepare the sediment for the low-
nutrient mesocosms, and in the other tank, the
overlying water had a concentration of 200 lgPl
-1
.
The water was replaced once a month. To optimise
equilibration, the sediment was stirred and resus-
pended into the water column by a rake. The
equilibration process was continued until the P
concentrations of the overlying water layer remained
stable at 25 and 200 lgl
-1
. No equilibration was
performed for nitrogen because total nitrogen (TN), in
contrast to TP, responds quickly to changes in externa l
loading, suggesting low importance of internal loading
and fast equilibration of TN levels (Jeppesen et al.,
2005).
All mesocosms were initially filled with nutrient-
poor lake or tap water with a total phosphorus (TP)
concentration of \25 lgl
-1
by which the low TP
level was achieved. Soluble reactive phosphate (as
Na
2
HPO
4
) and nitrate (as Ca(NO
3
)
2
) were added to
eight of the mesocosms to establish the high-nutrient
mesocosms with 200 lgl
-1
TP and 2.0 mg l
-1
TN
levels. The dosing levels and ratios followed those
used in previous experiments (Gonza
´
lez Sagrario
et al., 2005; Jeppesen et al., 2007a).
The mesocosms for the shallow experiment were
established with an initial water level of 1 m and the
deeper mesocosms with a 2-m water depth. Water
level fluctuations were considered an integral part of
the influence of the local climate at the experimental
sites and, after establishment, the water levels in the
mesocosm were left to fluctuate following local
precipitation and evaporation. In the Czech Republic,
however, excess water due to high precipita tion had to
be removed on two occasions to prevent the meso-
cosms from overflowing. Fifty litres of water were
removed both times by filling a bucket with a mixed
Table 1 Location, altitude, location of the meteorological
stations and rainfall chemistry monitoring stations (with
distance to the experimental sites), average daily mean
temperature and total precipitation (May–November) during
the mesocosm experiments carried out across a latitudinal
gradient
Country Lake Coordinates Altitude
(m.a.s.l.)
Meteorological
station
Average
daily air T
(°C)
Total
precipitation
(mm)
Rainfall chemistry
station
Sweden Erken 59°49
0
59
00
N
18°33
0
55
00
E
11 Lake Erken (on
site)
15.1 271 Aspvreten (75 km)
Estonia Vo
˜
rtsja
¨
rv 58°12
0
17
00
N
26°06
0
16
00
E
35 Lake Vo
˜
rtsja
¨
rv
(on site)
15.6 252 Alam-Pedja, Loodi
and Otepa
¨
a
¨
(30 km)
Germany Mu
¨
ggelsee 52°26
0
0
00
N
13°39
0
0
00
E
32.4 Mu
¨
ggelsee (on
site)
16.4 424 Mu
¨
ggelsee (on site)
Czech
Republic
Vodn
ˇ
any 49°09
0
14
00
N,
14°10
0
11
00
E
395 C
ˇ
eske
´
Bude
ˇ
jovice
(30 km)
15.5 401 Vodn
ˇ
any (on site)
Turkey METU DSI
50-Yıl dam
lake
39°52
0
38
00
N
32°46
0
32
00
E
998 Ankara (15 km) 19.8 168 C¸ ubuk and C¸ amkoru
(100 km)
Greece Lysimachia 38°33
0
40
00
N
21°22
0
10
00
E
16 Agrinio
(15 km)
23.4 252 –
Hydrobiologia
123
water sample. These water removals were taken into
account as output in the nutrient budget calculations.
To establish the mesocosm food web, mixed phyto-
plankton and zooplankton inoculations from five local
lakes in each country were used, followed by the
addition of eight Myriophyllum spicatum L. plants and
finally six fish with a 1:1 male:female ratio. In all
countries three-spined sticklebacks (Gasterosteus
aculeatus L.) were stocked, except in Sweden, where
roach (Rutilus rutilus L.) and Greece, where mosquito
fish (Gambusia affinis Baird and Girard) had to be used.
Dead fish were replaced during the experiment to
maintain constant fish populations. The water in the
mesocosms was circulated by a submers ed pump
(300 l h
-1
) for the entire duration of the experiment.
Monthly additions of N and P were made to
counteract nutrient loss and to reset the nutrient
concentrations to approximately initial levels. Differ-
ent doses were used for the different treatments to
account for the differences in volume and nutrient
level: 5.1 mg P and 102 mg N were added monthly to
the four shallow/low-nutrient mesocosms (SL);
40.8 mg P and 816 mg N to the four shallow/high-
nutrient mesocosms (SH); 10.8 mg P and 216 mg N to
the four deep/low-nutrient mesocosms (DL) and
86.0 mg P and 1720 mg N to the four deep/high-
nutrient mesocosms (DH). In order to ensure a similar
nutrient loading in all countries, these quantities were
kept constant during the experiment, even when the
water volume changed in some of the countries.
The first sampling for chemical analysis was
conducted three days after the establishment of the
different nutrient treatments. Thereafter, monthly
water samples were taken and analysed for soluble
reactive phosphorus (SRP), TP, ammonium (NH
4
–N),
nitrite and nitrate (NO
2
? NO
3
–N) and TN. The water
samples were analysed in the laboratories of the
respective countries where the experiment was per-
formed. Concentrations below the detection limits of
the equipment were equated with those limits. Dis-
solved oxygen, chlorophyll-a and percent volume
inhabited by plants (PVI) were measured at monthly
intervals in all countries. Macrophyte dry weight
biomass was determined at the end of the experiment.
Nutrient loss calculations
Nutrient loss from the water column was calculated for
TN, dissolved inorganic nitrogen (DIN), TP and SRP.
While DIN and SRP represent the inorganic nutrient
pools, TN and TP encompass both inorganic and
organic nutrient components. An operationally
defined ‘organic’ portion of N and P loss was
calculated as the difference between total and inor-
ganic nutrient loss. This portion includes N and P
mainly in planktonic organisms and detritus but also in
dissolved forms other than DIN and SRP.
Nutrient loss between two sampling dates was
calculated for N and P using the mass balance model of
Messer & Brezonik (1978):
Nutrient loss ¼ input output D nutrient content;
where ‘‘Input’’ is the input of nutrients to the
mesocosms in the relevant time interval through
nutrient additions and atmospheric deposition (N
fixation is neglected); ‘‘Output’’ is loss of nutrients
through water removal and ‘‘D nutrient content’’
(= nutrient content at time t ? 1 minus nutrient
content at time t) is the change in nutrient content in
the water column between the two sampling dates.
Nutrient loss (TN
loss
,DIN
loss
,TP
loss
, SRP
loss
) was
calculated in milligrams (mg) and converted to
mg m
-2
day
-1
by dividing the nutrient loss value by
the surface area of the mesocosms and the number of
days between the two sampling dates. The difference of
TN
loss
DIN
loss
¼ OrgN
loss
and TP
loss
SRP
loss
¼
OrgP
loss
was also calculated for all mesocosms. All
calculations were first performed for each time
interval between the monthly sampling dates, after
which a weighted average was calculated for the entire
experimental period. Weighted averages were calcu-
lated by multiplying the values measured on the
sampling date by its week number and dividing the
sum of these by the sum of the week numbers.
Weighted averages were used because they reduce the
influence of the early transitional phase after the start
of the experiment (Stephen et al., 2004).
The weighted average for shallow mesocosms was
calculated based on the data until mid-September
because the shallow mesocosms in Greece had signif-
icantly decreased water levels owing to evaporation,
followed by complete drying out (after 11 September).
The average temperature in Greece was markedly
higher than in the northern countries, making the
Greek data highly relevant for analysing the relation-
ship between nutrient loss and climate. There fore, we
chose to exclude the data for October and November
Hydrobiologia
123
from all countries rather than exclude all the shallow
mesocosms in Greece. For the deep mesocosms, the
weighted average was calculated for the entire dura-
tion of the experiment (May–November).
Additionally, nutrient loss was calculated relative to
the pool of available TN, DIN, TP and SRP and
expressed as TN
%-loss
,DIN
%-loss
,TP
%-loss
and SRP
%-loss
in order to examine the actual nutrient loss compared
with maximum potential nutrient loss:
Relative nutrient loss
¼ total nutrient loss=total available nutrient pool
The total available nutrient pool for the entire
experiment duration was calculated by adding the
monthly nutrient additions, monthly nutrient loading
through precipitation and the nutrient content present
in the water column at the start of the experiment.
Meteorological data were taken from the meteoro-
logical stations listed in Table 1. N and P concentrations
of precipitation were measured near the experimental
site in Germany (TN, TP) by the Department of
Chemical Analytics of the Leibnitz Institute of Fresh-
water Ecology and Inland Fisheries (IGB) and in the
Czech Republic (DN, TP) at Pond Vodn
ˇ
any. For the
other countries, data from nearby monitoring stations
were used (Table 1). Data from No
˜
ges et al. (1998)
were used for TP concentrations of precipitation in
Estonia, as well as for Sweden, due to lack of data. TP
concentrations for Turkey were taken from Koc¸ak et al.
(2010). Monitoring data on nutrients in precipitation
were not available from Greece, so DIN and TP
concentrations from Turkey were used as substitute.
SRP loading through precipitation was ignored for the
calculation of SRP loss owing to lack of data.
Statistical analysis
SAS 9.2 Statistical Software (SAS Institute Inc., 2008,
Cary, NC) was used for statistical analysis. One-way
ANOVA and non-parametric Kruskal–Wallis tests
were used on mean values of TN and TP to test if
significantly different effects of nutrient treatments
emerged.
ANCOVA analyses using mixed linear models
(Proc Mixed in SAS) were performed separately for
the shallow and deep mes ocosms on weighted average
nutrient loss and relative nutrient loss, with average air
temperature at the experimental sites as a covariate
and nutrient treatment as a fixed factor. The shallow
and deep mesocosms were treated as two distinct
experiments because of the differences in nutrient
loading, which means that no direct test for the water
depth effect could be made. ANCOVA analysis was
performed using weighted averages because it pro-
duces consistent results with repeated measures anal-
ysis when the data do not fit normality and sphericity
assumptions (Stephen et al., 2004). Transfor mations
(natural logarithmic and square root) were used to
meet the normality and homo scedasticity assumptions
of ANCOVA.
Air temperature was used to represent the climatic
gradient in the ANCOVA analysis. Air temperature
was measured daily at each site, and water temperature
on the sampling date was significantly correlated with
average air temperature between that sampling date
and the previous sampling (P \ 0.0001, r
2
= 0.85).
Although measured temperatures differed between
countries, no significant differences in water temper-
ature among mesocosms were found within the
individual countries (tested with ANOVA). Air tem-
perature at the experimental sites could therefore be
used for all mesocosms in the statistical analysis.
In Germany, one DL mesocosm was lost during the
experiment and thus excluded from analysis. In
Germany and the Czech Republic, there were issues
with temporarily submerged mesocosms (two DH and
one SH), unaccounted water loss (one DL) and
overflowing of mesocosms (two SH, one SL), but as
the nutrient loss values of these latter mesocosms were
not affected, they were included in the statistical
analysis.
Stepwise multiple linear regress ion (Proc Reg) was
performed for the shallow and deep experiments using
weighted average nutrient loss and relative nutrient
loss as dependent variables. For every dependent
variable, one linear regression analysis was performed
with a set of multiple independent variables in order to
examine the relationship between nutrient loss and a
set of relevant physical, chemical and biological
variables measured during the experiment. Mean
temperature, water depth, mean nutrient concentra-
tions (TN, DIN, TP or SRP), TN:TP, nutrient loading
quantities (nutrient additions and atmospheric depo-
sition), macrophyte dry weight, chlorophyll-a (chl-a)
and dissolved oxygen (DO) were initially selected as
independent variables. For each dependent variable,
the coefficient of determination (R
2
) for the overall fit
Hydrobiologia
123
of the regression model and the semi-partial correla-
tion coefficients (pr
2
) for each independent variable
were determined to show their individual contribution.
Results
Mean TN, TP, DIN and SRP concentrations in the
mesocosms are shown in Fig. 1 for all countries.
ANOVA and Kruskal–Wallis tests showed that low-
and high-nutrient mesocosms had significantly different
TN and TP concentrations in all countries (P \ 0.01)
except for TN in the Czech mesocosms and the deep
Greek mesocosms. Precipitation constituted a large
source of N for the low-nutrient mesocosms. In five of
the countries, the atmospheric deposition of N ranged
from 0.84 mg DIN m
-2
day
-1
in Sweden to 2.69 mg
DIN m
-2
day
-1
in Estonia. In Germany, however, TN
deposition was as high as 10.2 mg m
-2
day
-1
.Addi-
tion of TP through precipitation ranged between
0.02 mg m
-2
day
-1
for Turkey and 0.72 mg m
-2
day
-1
for Germany and constituted a large source of
P in the low-nutrient mesocosms.
Total nitrogen loss
TN
loss
was highest in the deep and high-nutrient
mesocosms (DH [ SH [ DL [ SL) in all countries
Fig. 1 Weighted averages
with standard error for SRP,
TP, DIN and TN
concentrations in an
experimental setup with
shallow and deep
mesocosms across a
latitudinal gradient and
mean air temperature
Hydrobiologia
123
except Greece (Fig. 2a). The TN
%-loss
was highest in
the SH (85%) and DH (79%) treat ments (Fig. 3a). In
the low-nutrient mesocosms, the TN
%-loss
was 66% for
DL and 59% for SL (SH [ DH [ DL [ SL).
Nutrient treatment had a significant effect on the
TN
loss
and TN
%-loss
in both the shallow and the deep
mesocosms (Table 2a). In the deep mesocosms, mean
temperature at the experimental sites had a significant
negative effect on TN
loss
(Fig. 4a), while there was a
significant positive effect of temperature on TN
%-loss
in the shallow mesocosms (Fig. 4a).
Dissolved inorganic nitrogen loss
In all countries, DIN concentrations declined during
the experiment and reached values below 0.06 mg l
-1
Fig. 2 Mean absolute
losses of N and P from the
water columns of a
mesocosm setup across a
latitudinal gradient.
Weighted means (mg m
-2
day
-1
) are given for nutrient
treatments (L low, H high) in
shallow (S) and deep
(D) mesocosms
Hydrobiologia
123
in the SL, SH and DL treatments by the end of the
experiment. Only in the DH mesocosms, average DIN
concentrations were as high as 0.84 mg l
-1
in
November, despite high nutrient loss.
Similar to TN
loss
, DIN
loss
was highest in the high-
nutrient mesocosms and in the deep mesocosms
(DH [ SH [ DL [ SL), and the pattern was the
same in all countries (Fig. 2b). Th e overall DIN
loss was higher than the TN loss in the high
nutrient treatments but lower in the low-nutrient
mesocosms. In all treatments, the DIN
%-loss
(Fig. 3b) was higher than 90%, except in the deep
Fig. 3 Mean relative
nutrient losses as
percentages of the total
available pool in the water
columns of a mesocosm
setup across a latitudinal
gradient. S shallow, D deep,
L low nutrient loading,
H high nutrient loading
Hydrobiologia
123
mesocosms in Sweden and Germany, with
SH [ SL & DL [ DH.
Nutrient level had a significant positive effect on
the DIN
loss
and DIN
%-loss
in the shallow mesocosms
and on the DIN
loss
in the deep mesocosms (Table 3a).
No significant temperature effect was detected at the
experimental sites for either DIN
loss
or DIN
%-loss
(Fig. 4b).
The effects of temperature and nutrient level were
also determined for organic nitrogen loss (orgN
loss
)
taken as the difference between TN
loss
and DIN
loss
(not in table). Temperature had a significant and
slightly positive effect in the shallow mesocosms
(P \ 0.001), while the interaction between nutrient
level and temperature was significant for the deep
mesocosms (P \ 0.01). Figure 4c shows the negative
relationship between orgN
loss
and temperature in the
DH mesocosms.
Total phosphorus loss
TP
loss
was highest in the high-nutrient and deep
mesocosms (DH [ SH [ DL [ SL) (Fig. 2c). TP
%-
loss
was highest in the DH and SH mesocosms (86%),
followed by DL (66%) and SL (62%)
(SH & DH [ DL & SL) (Fig. 3c).
TP
loss
was significantly affected by nutrient level in
both the shallow and the deep mesocosms (Table 2 b).
The effect of temperature was significant only for the
deep mesocosms, with TP
loss
decreasing with increas-
ing temperature (Fig. 5a). Th e relationship between
temperature and TP
%-loss
depended on depth, with a
positive relationship in the shallow mesocosms but a
negative relationship in the deep mesocosms (Fig. 5a).
The initial average TN:TP ratio by weight among
treatments was 43 (SL), 47 (DL), 15 (SH) and 13
(DH). However, differences among the countries were
marked, with ratios ranging from 10 to 111 in the low
and from 7 to 25 in the high nutrient treatments.
Manual monthly nutrient additions to all treatments
were made according to a TN:TP ratio of 20, partially
causing the ratios in the different treatments to
converge during the course of the experiment. The
mean TN:TP ratios were 31 and 36 in the SL and DL
treatments, and 17 and 23 in the SH and DH
treatments, respectively.
Soluble reactive phosphorus
SRP decreased during the first month of the exper-
iment, and the average concentrations at the end of
the experiment were below 1 0 lgl
-1
in all treat-
ments except DH (34 lgl
-1
). SRP
loss
followed the
same patterns as TP and TN with DH [
SH [ DL [ SL. Differences among the countries
were small (Fig. 2d). Similar to DIN
%-loss
,SRP
%-loss
was higher than 85% in all treatments, with the high-
nutrient mesocosms showing higher relative nutrient
Table 2 Effects of nutrient
loading and temperature on
losses of phosphorus and
nitrogen pools from the
water columns of a
mesocosm experiment
carried out across a
latitudinal gradient
F-values, significance levels
ns, Not significant;
* 0.05 [ P [ 0.01;
** 0.01 [ P [ 0.001;
*** P \ 0.001 and nature
of the relationship (? or -)
of ANCOVA are shown
Depth Effect TN
loss
TN
%-loss
DIN
loss
DIN
%-loss
a
Shallow Nutrient 180***(?) 33***(?) 903***(?) 56***(?)
Temperature ns 20***(?)ns ns
Temperature 9 nutrient ns ns ns ns
Deep Nutrient 298***(?) 9**(?) 480***(?)ns
Temperature 5*(-)ns ns ns
Temperature 9 nutrient ns ns ns ns
Depth Effect TP
loss
TP
%-loss
SRP
loss
SRP
%-loss
b
Shallow Nutrient 272***(?) 32***(?) 336*** 92***(?)
Temperature ns 12**(?) ns 6*(?)
Temperature 9 nutrient ns ns 19***(-)ns
Deep Nutrient 437***(?) 20***(?) 1170***(?) 25***(?)
Temperature 16***(-) 17***(-) ns 12**(-)
Temperature 9 nutrient ns ns ns ns
Hydrobiologia
123
loss than the low-nutrient mesocosms (SH [ DH [
DL [ SL) (Fig. 3d).
Nutrient level had a significant effect on SRP
loss
in the deep mesocosms and on SRP
%-loss
in both
the shallow and the deep mes ocosms (Table 2b).
For SRP
loss
in the shallow mesocosms, a significant
interaction between temperature and nutrient
treatment was found (Fig. 5b). A significant posi-
tive effect of mean air temperature on SRP
%-loss
was recorded in the shallow mesocosms in contrast
to a significant negative effect in the deep meso-
cosms (Fig. 5b).
Tests for the effects of temperature and nutrient
level on OrgP
loss
showed no significant results for the
Fig. 4 Relationships between losses of nitrogen from the water columns of a mesocosm setup along a latitudinal gradient and mean air
temperature. S shallow, D deep, L low nutrient loading, H high nutrient loading
Hydrobiologia
123
Table 3 Regression relationships relating environmental con-
ditions to loss of nitrogen and phosphorus pools in the water
columns of a mesocosm experiment carried out across a
latitudinal gradient. Regression coefficients (b), semi-partial
correlation coefficients (pr
2
) and coefficients of determination
(R
2
) for multiple linear regressions are shown for (a) shallow
mesocosms and (b) deep mesocosms
DV Mean T° Mean NO
3
Mean SRP TN:TP Nutrient loading Macrophyte biomass Chl-a R
2
a
TN
loss
b ns – – 0.36** 0.04*** -0.11** -0.20*** 0.91
pr
2
0.02 0.82 0.00 0.06
TP
loss
b ns – – ns 0.04*** ns -0.01*** 0.92
pr
2
0.83 0.09
DIN
loss
b ns 29.34*** – ns 0.04*** ns -0.04* 0.97
pr
2
0.01 0.96 0.00
ln(SRP
loss
)
b ns – 13.72* 0.01** 0.04*** ns -0.00** 0.97
pr
2
0.00 0.00 0.95 0.01
TN
%-loss
b 2.42** – – 0.87*** 0.03*** ns -0.22** 0.62
pr
2
0.14 0.12 0.27 0.09
ln(TP
%-loss
)
b -0.11*** – – -0.04*** -0.03*** ns 0.01*** 0.72
pr
2
0.16 0.15 0.28 0.14
DIN
%-loss
b ns 8.67*** – 0.04* 0.00*** ns -0.02** 0.68
pr
2
0.14 0.04 0.43 0.07
ln(SRP
%-loss
)
b -0.07* – ns -0.02** -0.03*** ns ns 0.74
pr
2
0.06 0.06 0.61
DV Mean T8 Mean SRP TN:TP Nutrient loading Macrophyte biomass Chl-a R
2
b
TN
loss
b -2.34** – -0.24* 0.02*** ns ns 0.87
pr
2
0.02 0.02 0.83
TP
loss
b ns – ns 0.04*** ns -0.01*** 0.94
pr
2
0.92 0.02
DIN
loss
b ns – ns 0.03*** -0.20* ns 0.93
pr
2
0.92 0.01
SRP
loss
b ns -18.18*** -0.01** 0.03*** ns ns 0.99
pr
2
0.02 0.00 0.96
Hydrobiologia
123
shallow mesocosms and a significant positive effect
(P \ 0.001) of temperature in the deep mesocosms
(Fig. 5c). No significant effect of nutrient level was
found.
Other environmental variables
Average values for depth, DO, PVI, macrophyte biomass
and chl-a are shown in Fig. 6. Average depth clearly
decreased towards the warmer countries. PVI and
macrophyte biomass were highest in the warmer coun-
tries with PVI being highest in the shallow mesocosms,
while chl-a was highest in deep mesocosms. Highest PVI
in the shallow mesocosms was generally measured near
the end of the experiment in September. In the deep
mesocosms, while PVI decreased slightly in November
in some mesocosms, macrophytes mostly remained until
the end of the experiment. DO in the mesocosms was
markedly lower in the warmer countries.
Depth and DO concentrations showed significant
negative correlations with temperature ( r
2
\ -0.60;
P \ 0.001), PVI had a significant positive correlation
with temperature (r
2
[ 0.60; P \ 0.001) in the shal-
low mesocosms, while mean TP and mean TN were
significantly correlated with P loading and N loading,
respectively (r
2
[ 0.60, P \ 0.001). Therefore, these
variables had to be excluded from the regression
analysis. Multiple linear regression analysis (Table 3)
showed a significant positive influence of nutrient
loading on TN
loss
, DIN
loss
,TP
loss
and SRP
loss
, while
other factors had only minor influence. For relative
nutrient loss, nutrient loading was still significant but
contributed less to the overall fit of the regression
models. Mean temperature, TN:TP and mean nutrient
concentration were the other significant factors for the
relative nutrient loss.
Chl-a had a significant, mostly negative, relation-
ship with weighted average and relative N and P loss in
the shallow mesocosms, but was only significant for
TP loss in the deep mesocosms. Macrophyte biomas s
at the end of the experiment was negatively related to
TN loss in the shallow mesocosms and DIN
loss
in the
deep and was not related to relative nutrie nt loss.
TN:TP demonstrated a significant positive relation-
ship with TN and SRP loss in the shallow mesocosms,
but a negative relationship in the deep mesocosms.
Discussion
Our results show that external loading had a large
influence on nutrient loss along the climate gradient
together with a comparatively modest influence of
temperature, depth, and algal and macrophyte bio-
mass. In all six countries, TN
loss
, DIN
loss
,TP
loss
and
SRP
loss
were determined by the availability of nutri-
ents through nutrient loading, with the highest nutrient
loss (weighted mean) occurring in high nutrient
treatments in both the shallow and the deep meso-
cosms. Higher nutrient loading means that more
Table 3 continued
DV Mean T8 Mean SRP TN:TP Nutrient loading Macrophyte biomass Chl-a R
2
ln(TN
%-loss
)
b 0.09* – 0.01* -0.00* ns ns 0.28
pr
2
0.07 0.07 0.14
ln(TP
%-loss
)
b 0.08* – -0.02* -0.00*** ns ns 0.60
pr
2
0.11 0.20 0.29
ln(DIN
%-loss
)
b ns – 0.01* ns ns ns 0.09
pr
2
0.09
ln(SRP
%-loss
)
b ns 24.82*** -0.01*** ns ns 0.79
pr
2
0.40 0.39
– Not included in the model for that dependent variable (DV), ns not significant
* 0.05 [ P [ 0.01; ** 0.01 [ P [ 0.001; *** P \ 0.001
Hydrobiologia
123
nutrients are available in the system for plant and algal
uptake, thus creating a larger potential for nutrient loss
from the water column through uptake. The results of
the multiple linear regression clearly demonstrated the
importance of external loading for nutrient loss, with
nutrient loading constituting the largest component of
the coefficients of determination. This is in agreement
with the findings of previous studies (Windolf et al.,
1996; Saunders & Kalff, 2001a, b; Søndergaard et al.,
2003; Brett and Benjamin, 2008). The relative nutrient
loss as a proportion of the available nutrient pool
(TN
%-loss
,DIN
%-loss
,TP
%-loss
and SRP
%-loss
) was also
Fig. 5 Relationships between losses of phosphorus from the water columns of a mesocosm setup along a latitudinal gradient and mean
air temperature. S shallow, D deep, L low nutrient loading, H high nutrient loading
Hydrobiologia
123
significantly higher in the high-nutrient mesocosms.
The relative loss of DIN and SRP was close to 100% in
all treatments and countries, showing that the uptake
capacity of the mesocosm communities was not
saturated. Within the nutrient loading range used in
our experiments, nearly all available inorganic nutri-
ents were lost in our mesocosms, either through
biomass uptake or loss to the sediment or atmosphere.
The mesocosms acted as a near complete sink for
inorganic nutrients in all countries on the climate
gradient.
We also found a significant, though weak, decrease
in TN and TP loss in the deep mesocosms when
moving from the colder northern to the warmer
southern countries, but no effect on DIN and SRP
loss, implying a lower organic N and P loss (or higher
organic N and P accumulation in the water column) in
the warmer south. In the shallow mesocosms, how-
ever, we found a tendency for TP and TN loss to
increase with temperature, though only significantly
so for relative nutrient loss. Our hypothesis that
warming would decrease phosphorus loss whilst
increasing nitrogen loss was partly supported, though
only for P in the water columns of the deeper
mesocosms and for N in the shallower mesocosms.
The decreasing TN
loss
and TP
loss
with increasing
temperature in the deep mesocosms can mainly be
attributed to lower nutrient loss in Turkey and Greece.
This may reflect the markedly higher average temper-
ature in Turkey (20°C) and Greece (24°C) during the
experiment, compared with 15–16°C in Sweden,
Estonia, Germany and the Czech Republic.
Previous studies in micro cosms and field measure-
ments have shown increased d enitrification rates with
increasing temperature (Pinay et al., 2007; Herrman
et al., 2008; Veraart et al., 2011), in part caused by a
decrease in DO concentrations with rising temperature
due to lower solubility of oxygen in water and higher
respiration rates compared with photosynthesis (Ver-
aart et al., 2011; Scharfenberger et al., unpublished
data). We also found a strong negative correlation
between DO and temperature, with DO concentrations
being lowest in Greece and Turkey, potentially
enhancing denitrification (Pearson et al., 2012; Small
et al., 2014). Howeve r, DIN
loss
was generally higher in
Estonia and the Czech Republic, indicating that DO
was not of key importance for the increase in N loss in
these fully mixed mesocosms.
Moreover, DIN
loss
showed no relationship with
average temperature in either the shallow or the deep
mesocosms. This finding concurs with Kosten et al.
(2009) who found no indication of lower nitrate
concentrations in the warm lakes along a latitudinal
gradient (5°–55°S), suggesting that other factors
limited denitrification and offset the potential increas-
ing effects of temperature. A comparative study by
Olsen et al. (2015) of various mesocosm experiments
with considerably higher average target nitrate con-
centrations (10 mg l
-1
in high N treatments) than in
our study (2 mg l
-1
in high N treatments) revealed
accumulation of nitrate in the water column in warmer
climates, while nitrate was rapidly removed in exper-
iments conducted in the temperate zone (Gonza
´
lez
Sagrario et al., 2005; Jeppesen et al., 2007b; Feucht-
mayr et al., 2009). Accumulation of nitrate in the
warm mesocosms was ascribed to limitation of
denitrification due to low organic matter availability
as no correlation with DO was found (Olsen et al.,
2015).
In our experiment, both nitrate and organic matter
availability may be potential candidates limiting N
loss (Davidsson & Leonardson, 1998; Pin
˜
a-Ochoa &
A
´
lvarez-Cobelas, 2006; Taylor & Townsend, 2010).
Nitrate declined throughout the experiment in the low-
nutrient mesocosms to average values of 30–40 lgl
-1
for all countries. Both the decline in nitrate concen-
trations in the mesocosms and the high DIN
%-loss
values ([95%) indicate that almost all available DIN
was taken up or lost to the sediment from the
mesocosms in all countries. Further N loss through
denitrification under warm climate conditions may
therefore have been limited by nitrat e availability,
preventing higher DIN loss in the southern countries.
Analysis of the metabolic processes during our
mesocosm experiment by Scharfenberger et al. (un-
published data) showed an increase in both gross
primary production (GPP) and ecosystem respiration
(ER) with increasing temperature. Because ER
increased more steeply than GPP, a shift from
autotrophy to heterotrophy occurred with increasing
temperatures. Therefore, net primary production can
be expected to be lowest, on average, in the warmer
countries. In the warmer mesocosms, more carbon
fixed by primary production was released again as
CO
2
, leaving less organic carbon available for deni-
trification. Organic matter availability was therefore
Hydrobiologia
123
likely another factor that limited DIN
loss
in the warmer
south.
In the deep mesocosms, the decreasing trend in
TN
loss
and TP
loss
but not in DIN
loss
or SRP
loss
with
increasing temperature also indicates a key role of
processes related to organic N and P. TN and TP loss,
besides denitrification, includes settling of sestonic
particles and removal from the water column through
uptake by non-sestonic organisms (such as macro-
phytes, fish, large invertebrates and periphyton). But
inorganic N and P loss also encompasses conversion of
inorganic nutrients to organic forms assim ilated in the
phytoplankton biomass or released into the water
column. Average inorganic N and P loss was higher
than total N and P loss in the deep, high-nutrient
mesocosms, indicating organic N and P production
and accumulation, as also evidenced by the negative
relationship between temperature and both orgN
loss
and orgP
loss
in the DH mesocosms. This finding is in
accordance with the higher biomass of phytoplankton
recorded in the deep, high-nutrient mesocosms in
these two countries, shown in Fig. 6.
In the shallow mesocosms, phytoplankton biomass
was not higher in the Greek and Turkish mesocosms,
but instead macrophyte development was high. In
these countries, water levels in the shallow mesocosms
fell from 100 cm to 20–50 cm over the course of the
experiment, which likely enhanced macrophyte devel-
opment through more favourable light conditions
(Bucak et al., 2012). Macrophytes can play an
important role in nutrient loss from the water column,
either through direct nutrient uptake or by creating
conditions favourable for denitrification (Kankaala
et al., 2002;Desmet et al., 2011). Higher macrophyte
development and higher nutrient uptake by plan ts can
therefore explain why TP and TN loss tended to be
higher in the warmer countries and why a positive
relationship with temperature was found for TN
%-loss
and TP
%-loss
. However, we found a negative relation-
ship between macrophyte biomass and TN
loss
and
DIN
loss
in the multiple linear regression, but the
contribution of macrophyte biomass to the fit of the
regression model was low and macrophyte biomass
was only measured at the end of the experiment and
may thus not reflect the macrophyte nutrient uptake
throughout the experiment.
The effect of macrophyte nutrient uptake was
further quantified by multiplying the macrophyte
biomass by estimates of macrophyte N and P content
taken from Duarte (1992) who compiled data on C, P
and N content of a series of freshwater angiosperms
based on literature data. Lower and upper estimates
from this reference show that in countries with high
macrophyte growth in the shallow mesocosms (Tur-
key, Greece, Estonia), between 30 and 100% of the
available N-pool and between 10 and 80% of the
available P-pool were potentially locked up in the
macrophyte biomass. Although this analysis only
provides a crude estimate, it suggests that macrophyte
uptake potentially played an important role for the
nutrient loss in the shallow mesocosms.
The observed differences in nutrient loss among
countries could potentially be due to inaccurate esti-
mations of the atmospheric deposition in certain
countries for which deposition measurement data were
missing. However, reducing atmospheric deposition to
zero or doubling the estimate used for atmospheric
deposition in Sweden and Greece, where deposition
was estimated based on data from Estonia and Turkey
respectively, would lead only to a 20–30 change % in
calculated TN and DIN loss in the low-nutrient
mesocosms and would be negligible in the high-nutrient
mesocosms. For TP loss, the differences would be only
about 10%, except in the DL mesocosms in Greece
where atmospheric deposition is an important determi-
nant for the nutrient input (90% change in retention
estimate). However, such differences do not affect the
overall results or the climate pattern observed because
the differences in nutrient loss between the treatments
and countries are sufficiently large. The main conclu-
sions remain the same even if reduced or increased
values of atmospheric deposition are used.
Nutrient loss pathways through sedimentation,
uptake by organisms (in cluding fish) or denitrification
were not quantified separately in our experiment. The
relationships between the trends in macrophyte and
algal biomass and the trends in N and P loss suggest,
however, that algal and plant uptake played an
important role in nutrient loss. All available pools of
SRP and DIN were then used by macrophytes or
phytoplankton and locked up in the biomass.
c
Fig. 6 Differences in environmental factors in the water
columns of a mesocosm setup along a latitudinal gradient and
mean air temperature. Weighted averages with standard error
are shown for water depth, percent volume inhabited by plants
(PVI), chlorophyll-a (chl-a), dissolved oxygen (DO) and
macrophyte dry weight biomass (measured at the end of the
experiment) for shallow and deep mesocosms
Hydrobiologia
123
Hydrobiologia
123
Conclusion
Our study confirmed the importance of nutrient
loading in determining N and P loss in the water
column of shallow lake ecosystems. Further, we found
consistently high levels of DIN and SRP loss in all
countries along the climate gradient, demonstrating
the high nutrient uptake capacity of shallow lake
ecosystems. In deep mesocosms, we found a decrease
in TN and TP loss with increasing temperature but no
decrease in DIN or SRP loss, indicating higher algal
production of organic N and P in warmer systems. In
shallow mesocosms, we found an increasing trend of
TN and TP loss with increasing temperature, related to
high macrophyte growth in the warmer countries.
Increasing N loss through denitrification under
warmer conditions due to lower availability of DO
seemed to have played a minor role. Other factors,
such as nitrate (at low N loading) and organic matter
availability, therefore likely determined DIN
loss
.
While uptake by plant and algal biomass seemed to
play an important role, the ultimate fate of the
nutrients cannot be ascertained. Furt her studies
involving detailed analysis of sediment uptake, algal
and plant uptake and denitrification under different
climate conditions would be valuable. Our results,
based on a space-for-time mesocosm experiment , with
space used as a proxy for climate, should be
interpreted with care given the small size of the
mesocosms. Mixing conditions in mesocosms are
different from natural mixing in lakes, and sediment
characteristics can play an important role in nutrient
loss processes. However, since our experiment was
repeated under similar conditions in each country and
given the control of nutrients and water depth, we
believe that our results provide a valuable contribution
to further our understanding of nutrient processing in
lakes under different climate conditions.
Acknowledgments This study was supported by FP7/ENV-
2009-1 under grant agreement 244121 (REFRESH project), the
Middle East Technical University, METU-BAP programme of
Turkey (BAP-07-02-2009), TU
¨
BI
˙
TAK (Project no: 105Y332
and 110Y125), the MARS project (Managing Aquatic
ecosystems and water Resources under multiple Stress)
funded under the 7th EU Framework Programme, Theme 6
(Environment including Climate Change), Contract No.:
603378 (http://www.mars-project.eu) and the Danish projects
CIRCE, CRES and CLEAR (a Villum Kann Rasmussen Centre
of Excellence project). JC was supported by TU
¨
BI
˙
TAK 2215
Scholarship Programme. We acknowledge Ms. Christiane
Herzog (Department of Chemical Analytics and Biogeochem-
istry of IGB) and Kurt Pettersson (Erken laboratory) for mete-
orological data collection. We are thankful to Anne Mette
Poulsen for proofreading the manuscript.
References
Brett, M. T. & M. M. Benjamin, 2008. A review and reassess-
ment of lake phosphorus retention and the nutrient loading
concept. Freshwater Biology 53: 194–211.
Bucak, T., E. Saraog
˘
lu, E. E. Levi, U
¨
. N. Tavs¸anog
˘
lu, A. I.
C¸akırog
˘
lu, E. Jeppesen & M. Bekliog
˘
lu, 2012. The influ-
ence of water level on macrophyte growth and trophic
interactions in eutrophic Mediterranean shallow lakes: a
mesocosm experiment with and without fish. Freshwater
Biology 57: 1613–1642.
Cook, P. L. M., K. T. Aldrige, S. Lamontage & J. D. Brookes,
2010. Retention of nitrogen, phosphorus and silicon in a
large semi-arid riverine lake ecosystem. Biogeochemistry
99: 49–63.
Correl, D. L., 1998. The role of phosphorus in the eutrophication
of receiving waters: a review. Journal of Environmental
Quality 27: 261–266.
Danger, M., G. Lacroix, C. Oumarou, D. Benest & J. Me
´
riguet,
2008. Effects of food-web structure on periphyton stoi-
chiometry in eutrophic lakes: a mesocosm study. Fresh-
water Biology 53: 2089–2100.
Davidsson, T. E. & L. Leonardson, 1998. Seasonal dynamics of
denitrification activity in two water meadows. Hydrobi-
ologia 364: 189–198.
Desmet, N. J. S., S. Van Belleghem, P. Seuntjens, T. J. Bouma,
K. Buis & P. Meire, 2011. Quantification of the impact of
macrophytes on oxygen dynamics and nitrogen retention in
a vegetated lowland river. Physics and Chemistry of the
Earth 36: 479–489.
Duarte, C. M., 1992. Nutrient concentration of aquatic plants -
patterns across species. Limnology and Oceanography 37:
882–889.
Feuchtmayr, H., R. Moran, K. Hatton, L. Connor, T. Heyes, B.
Moss, I. Harvey & D. Atkinson, 2009. Global warming and
eutrophication: effects on water chemistry and autotrophic
communities in experimental hypertrophic shallow lake
mesocosm. Journal of Applied Ecolology 46: 713–723.
Gomez, E., M. A. Fillit, M. C. Ximenes & B. Picot, 1998.
Phosphate mobility at the sediment–water interface of a
Mediterranean lagoon (etang du Mejean), seasonal phos-
phate variation. Hydrobiologia 374: 203–216.
Gonza
´
lez Sagrario, M. A., E. Jeppesen, J. Goma
´
, M. Sønder-
gaard, J. P. Jensen, T. L. Lauridsen & F. Landkildehus,
2005. Does high nitrogen loading prevent clear-water
conditions in shallow lakes at moderately high phosphorus
concentrations? Freshwater Biology 50: 27–41.
Guildford, S. J. & R. E. Hecky, 2000. Total nitrogen, total
phosphorus, and nutrient limitation in lakes and oceans: is
there a common relationship? Limnology and Oceanogra-
phy 45: 1213–1223.
Finlay, J. C., G. E. Small & R. W. Sterner, 2013. Human
influences on nitrogen removal in lakes. Science 342:
247–250.
Hydrobiologia
123
Fisher, J. & M. C. Acreman, 2004. Wetland nutrient removal: a
review of the evidence. Hydrology and Earth System Sci-
ences 8: 673–685.
Han, H., N. Bosch & J. D. Allan, 2011. Spatial and temporal
variation in phosphorus budgets for 24 watersheds in the
Lake Erie and Lake Michigan basins. Biogeochemistry
102: 45–58.
Hasler, A. D., 1947. Eutrophication of lakes by domestic drai-
nage. Ecology 28: 383–395.
Herrman, K. S., V. Bouchard & R. H. Moore, 2008. Factors
affecting denitrification in agricultural headwater streams
in Northeast Ohio, USA. Hydrobiologia 598: 305–314.
Jensen, J. P., P. Kristensen & E. Jeppesen, 1990. Relationships
between nitrogen loading and in-lake concentrations in
shallow Danish lakes. Verhandlungen der Internationalen
Vereinigung fu
¨
r Theoretische und Angewandte Limnolo-
gie 24: 201–204.
Jeppesen, E., M. Søndergaard, J. P. Jensen, K. E. Havens, O.
Anneville, L. Carvalho, M. F. Coveney, R. Deneke, M.
T. Dokulil, B. Foy, D. Gerdeaux, S. E. Hampton, S. Hilt, K.
Kangur, J. Ko
¨
hler, E. H. R. R. Lammens, T. L. Lauridsen,
M. Manca, M. R. Miracle, B. Moss, P. No
˜
ges, G. Persson,
G. Phillips, R. Portielje, S. Romo, C. L. Schelske, D.
Straile, I. Tatrai, E. Willen & M. Winder, 2005. Lake
responses to reduced nutrient loading – an analysis of
contemporary long-term data from 35 case-studies.
Freshwater Biology 50: 1747–1771.
Jeppesen, E., M. Søndergaard, M. Meerhoff, T. L. Lauridsen &
J. P. Jensen, 2007a. Shallow lake restoration by nutrient
loading reduction - some recent findings and challenges
ahead. Hydrobiologia 584: 239–252.
Jeppesen, E., M. Søndergaard, A. R. Pedersen, K. Jurgens, A.
Strzelczak, T. L. Lauridsen & L. S. Johansson, 2007b.
Salinity induced regime shift in shallow brackish lagoons.
Ecosystems 10: 47–57.
Jeppesen, E., B. Moss, H. Bennion, L. Carvalho, L. De Meester,
H. Feuchtmayr, N. Friberg, M. O. Gessner, M. Hefting, T.
L. Lauridsen, L. Liboriussen, H. Malmquist, L. May, M.
Meerhoff, J. O. Olafson, M. Soons, J. T. A. Verhoeven, M.
Kernan & R. W. Battarbee, 2010. Interaction of climate
change and eutrophication. In Kernan, M., R. W. Battarbee
& B. Moss (eds), Climate Change Impacts on Freshwater
Ecosystems. Wiley, Chichester: 119–215.
Jeppesen, E., B. Kronvang, J. E. Olesen, J. Audet, M. Sønder-
gaard, C. C. Hoffmann, H. E. Andersen, T. L. Lauridsen, L.
Liboriussen, S. E. Larsen, M. Beklioglu, M. Meerhoff, A.
O
¨
zen & K. O
¨
zkan, 2011. Climate change effects on
nitrogen loading from cultivated catchments in Europe:
implications for nitrogen retention, ecological state of
lakes and adaptation. Hydrobiologia 663: 1–21.
Kankaala, P., A. Ojala, T. Tulonen & L. Arvola, 2002. Changes
in nutrient retention capacity of boreal aquatic ecosystems
under climatic warming: a simulation study. Hydrobiologia
469: 67–76.
Kaste, O. & P. J. Dillon, 2003. Inorganic nitrogen retention in
acid-sensitive lakes in southern Norway and southern
Ontario, Canada—a comparison of mass balance data with
an empirical N retention model. Hydrological Processes
17: 2393–2407.
Koc¸ak, M., N. Kubilay, S. Tug
˘
rul & N. Mihalopoulos, 2010.
Atmospheric nutrient inputs to the northern Levantine
basin from a long-term observation: sources and compar-
ison with riverine inputs. Biogeosciences 7: 4037–4050.
Kosten, S., V. M. L. Huszar, N. Mazzeo, M. Scheffer, L. D. S.
L. Sternberg & E. Jeppesen, 2009. Lake and watershed
characteristics rather than climate influence nutrient limi-
tation. Ecological Applications 19: 1791–1804.
Landkildehus, F., M. Søndergaard, M. Beklioglu, R. Adrian, D.
G. Angeler, J. Hejzlar, E. Papastergiadou, P. Zingel, A. I.
C¸ akirog
˘
lu, U. Scharfenberger, S. Drakare, T. No
˜
ges, M.
S
ˇ
orf, K. Stefanidis, N. Tavs¸anog
˘
lu, C. Trigal, A. Mahdy, C.
Papadaki, L. Tuvikene, S. E. Larsen, M. Kernan & E.
Jeppesen, 2014. Climate change effects on shallow lakes:
design and preliminary results of a cross-European climate
gradient mesocosm experiment. Estonian Journal of
Ecology 63: 71–89.
Lewis, W. M. & W. A. Wurtsbaugh, 2008. Control of lacustrine
phytoplankton by nutrients: erosion of the phosphorus
paradigm. International Review of Hydrobiology 93:
446–465.
Likens, G., 1972. Nutrients and Eutrophication, American
Society of Limnology Oceanography Special Symposium
1. American Society of Limnology Oceanography,
Lawrence.
Madsen, J. D., P. A. Chambers, W. F. James, E. W. Koch & D.
F. Westlake, 2001. The interaction between water move-
ment, sediment dynamics and submersed macrophytes.
Hydrobiologia 444: 71–84.
Messer, J. J. & P. L. Brezonik, 1978. Denitrification in the
sediments of Lake Okeechobee, Florida. Verhandlungen
der Internationale Vereinigung der Limnologie 20:
2207–2216.
Moss, B., S. Kosten, M. Meerhoff, R. W. Battarbee, E. Jeppesen,
N. Mazzeo, K. Havens, G. Lacerot, Z. Liu, L. De Meester,
H. Paerl & M. Scheffer, 2011. Allied attack: climate
change and eutrophication. Inland Waters 1: 101–105.
Moss, B., E. Jeppesen, M. Søndergaard, T. L. Lauridsen & Z.
Liu, 2013. Nitrogen, macrophytes, shallow lakes and
nutrient limitation: resolution of a current controversy?
Hydrobiologia 710: 3–21.
No
˜
ges, P., 2005. Water and nutrient mass balance of the partly
meromictic temperate Lake Verevi. Hydrobiologia 547:
21–31.
No
˜
ges, P., J. Arvo, L. Tuvikene & T. No
˜
ges, 1998. The budgets
of nitrogen and phosphorus in shallow eutrophic Lake
Vortsja
¨
rv (Estonia). Hydrobiologia 363: 219–227.
Olsen, S., E. Jeppesen, B. Moss, K. O
¨
zkan, M. Bekliog
˘
lu, H.
Feuchtmayr, M. Gonza
´
lez-Sagrario, L. Wei, S. Larsen &
M. Søndergaard, 2015. Effect of nitrogen and phosphorus
loading, salinity, temperature and water level on the
nitrogen retention capacity in lakes: an experimental
approach. Freshwater Biology 60: 646–662.
O
¨
zen, A., B. Karapinar, I. Kucuk, E. Jeppesen & M. Beklioglu,
2010. Drought-induced changes in nutrient concentrations
and retention in two shallow Mediterranean lakes subjected
to different degrees of management. Hydrobiologia 646:
61–72.
Paterson, M. J., D. W. Schindler, R. E. Hecky, D. L. Findlay &
K. J. Rondeau, 2011. Comment: lake 227 shows clearly
that controlling inputs of nitrogen will not reduce or pre-
vent eutrophication of lakes. Limnology and Oceanogra-
phy 56: 1545–1547.
Hydrobiologia
123
Pearson, L. K., C. H. Hendy, D. P. Hamilton & W. B. Silvester,
2012. Nitrogen-15 isotope enrichment in benthic boundary
layer gases of a stratified eutrophic iron and manganese
rich lake. Aquatic Geochemistry 18: 1–19.
Persson, J. & H. B. Wittgren, 2003. How hydrological and
hydraulic conditions affect performance of ponds. Eco-
logical Engineering 21: 259–269.
Pin
˜
a-Ochoa, E. & M. A
´
lvarez-Cobelas, 2006. Denitrification in
aquatic environments: a cross-system analysis. Biogeo-
chemistry 81: 111–130.
Pinay, G., B. Gumiero, E. Tabacchi, O. Gimenez, A.
M. Tabacchi-Planty, M. M. Hefting, T. P. Burt, V.
A. Black, C. Nilsson, V. Iordach, F. Bureau, L. Vought, G.
E. Petts & H. De
´
camps, 2007. Patterns of denitrification
rates in European alluvial soils under various hydrological
regimes. Freshwater Biology 52: 252–266.
Romero, J. R., I. Kagalou, J. Imberger, D. Hela, M. Kotti, A.
Bartzokas, T. Albanis, N. Evmirides, S. Karkabounas, J.
Papagiannis & A. Bithava, 2002. Seasonal water quality of
shallow and eutrophic Lake Pamvotis, Greece: implica-
tions for restoration. Hydrobiologia 474: 91–105.
SAS Institute Inc., 2008. SAS/STAT
Ò
9.2 User’s Guide. SAS
Institute Inc., NC.
Saunders, D. L. & J. Kalff, 2001a. Nitrogen retention in wet-
lands, lakes and rivers. Hydrobiologia 443: 205–212.
Saunders, D. L. & J. Kalff, 2001b. Denitrification rates in the
sediments of Lake Memphremagog, Canada-USA. Water
Research 35: 1897–1904.
Scharfenberger, U., E. Jeppesen, M. Bekliog
˘
lu, M. Søndergaard,
D. G. Angeler, A. I. C¸akırog
˘
lu, S. Drakare, J. Hejzlar, A.
Mahdy, E. Papastergiadou, M. S
ˇ
orf, K. Stefanidis, A.
Tuvikene, P. Zingel, R. Adrian (unpublished data). Effects
of trophic status, water level and temperature on shallow
lakes metabolism, metabolic balance and CO
2
-flux: a
comprehensive, standardised pan-European mesocosm
experiment.
Schindler, D. W., 1974. Eutrophication and recovery in exper-
imental lakes: implications for lake management. Science
184: 897–899.
Schindler, D. W., 2012. The dilemma of controlling cultural
eutrophication of lakes. Proceedings of the Royal Society
B: Biological Sciences 279: 4322–4333.
Schindler, D. W., R. E. Hecky, D. L. Findlay, M. P. Stainton, B.
R. Parker, M. J. Paterson, K. G. Beaty, M. Lyng & S. E. M.
Kasian, 2008. Eutrophication of lakes cannot be controlled
by reducing nitrogen input: results of a 37-year whole-
ecosystem experiment. Proceedings of the National
Academy of Sciences of the United States of America 105:
11254–11258.
Scott, J. T. & M. J. McCarthy, 2010. Nitrogen fixation may not
balance the nitrogen pool in lakes over timescales relevant
to eutrophication management. Limnology and Oceanog-
raphy 55: 1265–1270.
Scott, J. T. & M. J. McCarthy, 2011. Response to comment:
nitrogen fixation has not offset declines in the Lake 227
nitrogen pool and shows that nitrogen control deserves
consideration in aquatic ecosystems. Limnology and
Oceanography 56: 1548–1550.
Seitzinger, S., J. A. Harrison, J. K. Bo
¨
hlke, A. F. Bouwman, R.
Lowrance, B. Peterson, C. Tobias & G. Van Drecht, 2006.
Denitrification across landscapes and waterscapes: a syn-
thesis. Ecological Applications 16: 2064–2090.
Small, G. E., J. B. Cotner, J. C. Finlay, R. A. Stark & R.
W. Sterner, 2014. Nitrogen transformations at the sedi-
ment–water interface across redox gradients in the Lau-
rentian Great Lakes. Hydrobiologia 731: 95–108.
Smith, V. H., G. D. Tilman & J. C. Nekola, 1999. Eutrophica-
tion: impacts of excess nutrient inputs on freshwater,
marine, and terrestrial ecosystems. Environmental Pollu-
tion 100: 179–196.
Stephen, D., B. Moss & G. Phillips, 1997. Do rooted macro-
phytes increase sediment phosphorus release? Hydrobi-
ologia 342–343: 27–34.
Stephen, D., D. M. Balayla, E. Becares, S. E. Collings, C. Fer-
nandez-Alaez, M. Fernandez-Alaez, C. Ferriol, P. Garcia,
J. Goma, M. Gyllstro
¨
m, L.-A. Hansson, J. Hietala, T.
Kairesalo, M. R. Miracle, S. Romo, J. Rueda, A. Stahl-
Delbanco, M. Svensson, K. Vakkilainen, M. Valentin, W.
J. Van de Bund, E. Van Donk, E. Vicente, M. J. Villena &
B. Moss, 2004. Continental-scale patterns of nutrient and
fish effects on shallow lakes: introduction to a pan-Euro-
pean mesocosm experiment. Freshwater Biology 49:
1517–1524.
Sterner, R. W., J. J. Elser, E. J. Fee, S. J. Guildford & T.
H. Chrzanowski, 1997. The light:nutrient ratio in lakes: the
balance of energy and materials affects ecosystem structure
and process. American Naturalist 150: 663–684.
Strand, J. A. & S. E. B. Weisner, 2013. Effects of wetland
construction on nitrogen transport and species richness in
the agricultural landscape – experiences from Sweden.
Ecological Engineering 56: 14–25.
Søndergaard, M., J. P. Jensen & E. Jeppesen, 2003. Role of the
sediment and internal loading of phosphorus in lakes.
Hydrobiologia 506–509: 135–145.
Taylor, P. G. & A. R. Townsend, 2010. Stoichiometric control of
organic carbon-nitrate relationships from soils to the sea.
Nature 464: 1178–1181.
Veraart, A. J., J. J. M. de Klein & J. J. M. M. Scheffer, 2011.
Warming can boost denitrification disproportionately due
to altered oxygen dynamics. PLoS One 6: e18508.
Vitousek, T. M., J. D. Aber, R. W. Howarth, G. E. Likens, P.
A. Matson, D. W. Schindler, W. H. Schlesinger & D.
G. Tilman, 1997. Human alteration of the global nitrogen
cycle: sources and consequences. Ecological Applications
7: 737–750.
Vollenweider, R. A., 1976. Advances in defining critical loading
levels for phosphorus in lake eutrophication. Memoire
dell’Instituto Italiano di Idrobiologia 33: 53–83.
Weisner, S. E. B., P. G. Erikson, W. Graneli & L. Leonardson,
1994. Influence of macrophytes on nitrogen removal in
wetlands. Ambio 23: 363–366.
Wetzel, R., 2001. Limnology: Lake and River Ecosystems.
Academic Press, New York.
Windolf, J., E. Jeppesen, J. P. Jensen & P. Kristensen, 1996.
Modelling of seasonal variation in nitrogen retention and
in-lake concentration: a four-year mass balance study in 16
shallow Danish lakes. Biogeochemistry 33: 25–44.
Woods, H., W. Makino & J. Cotner, 2003. Temperature and the
chemical composition of poikilothermic organisms.
Functional Ecology 17: 237–245.
Hydrobiologia
123
