Content uploaded by Lauri Arvola
Author content
All content in this area was uploaded by Lauri Arvola on Jul 01, 2016
Content may be subject to copyright.
Numerical Exploration of the
Planktonic to Benthic Primary
Production Ratios in Lakes of the
Baltic Sea Catchment
Fabien Cremona,
1
* Alo Laas,
1
Lauri Arvola,
2
Don Pierson,
3
Peeter No
˜ges,
1
and Tiina No
˜ges
1
1
Centre for Limnology, Institute of Agricultural and Environmental Sciences, Estonian University of Life Sciences, Kreutzwaldi 5,
51014 Tartu, Estonia;
2
Lammi Biological Station, Faculty of Biological and Environmental Sciences, University of Helsinki,
Pa
¨a
¨ja
¨rventie 320, 16900 Lammi, Finland;
3
Erken Laboratory, Department of Ecology and Genetics, Uppsala University, Norra
Malmava
¨gen 45, 761 73 Norrta
¨lje, Sweden
ABSTRACT
Autotrophic structure refers to the partitioning of
whole-ecosystem primary production between
benthic and planktonic primary producers. Auto-
trophic structure remains poorly understood espe-
cially because of the paucity of estimates regarding
benthic primary production. We used a conceptual
model for numerically exploring the autotrophic
structure of 13 hemiboreal lakes situated in the
Baltic Sea catchment. We also used diel variations
in primary production profiles to graphically eval-
uate levels of light and/or nutrient limitation in
lakes. The input morphometric data, light extinc-
tion coefficients and dissolved carbon parameters
were mostly obtained from in situ measurements.
Results revealed that cross- and within-lake auto-
trophic structure varied greatly: one lake was
clearly dominated by benthic production, and three
lakes by phytoplankton production. In the rest,
phytoplankton production was generally dominant
but switch to benthic dominance was possible. The
modelled primary production profiles varied
according to lake water clarity and bathymetry.
Our results clearly indicate that the relative con-
tribution of benthic primary production to whole-
lake primary production should be taken into ac-
count in studies about hemiboreal and boreal lakes.
Key words: primary production; lake bathy-
metry; chlorophyll a; dissolved organic carbon;
light extinction coefficient.
INTRODUCTION
Researching the primary production (PP) of fresh-
water ecosystems has been the cornerstone of
limnological studies for decades (Lindeman 1942).
However, the emphasis has been generally put on
planktonic PP rather than benthic PP (Vadebon-
coeur and others 2002; Liboriussen and Jeppesen
2003). Technical issues related to the difficulty of
quantifying in situ PP of benthos compared to
phytoplankton and a stronger scientific appeal of
planktonic organisms to limnologists (Sand-Jensen
and Borum 1991) may explain the situation. As a
result, benthos contribution to lake PP is still poorly
Received 8 December 2015; accepted 15 April 2016;
Author contributions FC, AL: Conceived of or designed study,
performed research, analysed data, contributed new methods or mod-
els; FC, AL, LA, DP, PN, TN: Wrote the paper.
*Corresponding author; e-mail: fabien.cremona@emu.ee
Ecosystems
DOI: 10.1007/s10021-016-0006-y
2016 Springer Science+Business Media New York
known (Vadeboncoeur and others 2002), and
consequently, autotrophic structure, that is, how
whole-lake primary production is distributed be-
tween phytoplankton and benthos remains un-
known for the majority of lacustrine systems
(Higgins and others 2014). Albeit new in its formal
description, the concept of autotrophic structure
was already intuitively suggested by Wetzel (1964)
and later by Sand-Jensen and Borum (1991).
It is important to understand the contribution of
benthic and planktonic PP for several reasons.
Firstly, autotrophic structure influences not only
energy pathways in pelagic and benthic zones but
also the food web structure of the lakes (Higgins
and others 2014). Indeed, pelagic- and benthic-
produced organic matters differ in the magnitude of
energy and pollutants they transfer to higher
trophic levels (Hecky and Hesslein 1995; Che
´telat
and Amyot 2009; Cremona and others 2010).
Secondly, autotrophic structure provides a more
complete basis for characterizing lake trophic status
and energy fluxes than benthic or planktonic pri-
mary production alone. For example, although
they might exhibit similar absolute values of ben-
thic primary production, a clear oligotrophic lake
and a eutrophic lake may possess different energy
linkages which can only be revealed by partitioning
pelagic and benthic primary production. Thirdly, it
has been demonstrated that autotrophic structure
can be a useful tool for monitoring lake trophic
status changes. Eutrophication, for example, can
lead in some deep lakes to an increase of the
planktonic PP to whole-lake PP (Vadeboncoeur and
others 2003), whereas in shallow lakes the switch
from turbid to clear phase may lead to the opposite
(Jeppesen and others 2014).
From the theoretical point of view, four factors
can be considered as the main drivers of auto-
trophic structure of lakes: availability of light and
nutrients, and bathymetric and morphometric
characteristics of the basin (Vadeboncoeur and
others 2008). In lakes that are light limited because
of turbidity or strong water colouration, rapid light
extinction along the water column should favour
phytoplankton over benthic algae, resulting in a
planktonic to benthic production ratio greater than
1 (Althouse and others 2014).
In nutrient-limited lakes with high water column
transparency nutrient additions could benefit ei-
ther periphyton or phytoplankton, depending on
morphometric and bathymetric factors that influ-
ence the relative surface of the littoral zone.
Transparent shallow lakes should be dominated by
benthos because the littoral zone would be large
relatively to lake surface area size (Sand-Jensen
and Borum 1991). In contrast, nutrient-rich lakes
can be either favourable to phytoplankton and to
macrophytes which constitute the substratum of
epiphytes. On the other hand, planktonic produc-
ers should constitute the majority of the primary
production in deep transparent lakes because of a
much greater volume relative to surface area.
Large lakes should be dominated by phyto-
plankton, whether they are deep or shallow. Deep
large lakes typically have a small fraction of lake
surface area with a well-illuminated littoral zone
(Kalff 2002), whereas shallow large lakes are
unstratified and experience frequent wind-medi-
ated sediment resuspension (Chen and others
2003). Except in lakes with a very delineated
shoreline, planktonic producers should be advan-
taged compared to benthic producers in large lakes.
Lakes within the Baltic Sea catchment are lo-
cated at the boundaries between light- and nutri-
ent-limited zones which is believed to be situated
around 60of north latitude (Lewis 1996;No
˜ges
and others 2011) making it unclear which param-
eter is the main driver of lake productivity and
hence autotrophic structure. It has been shown
that in small, nutrient poor lakes in Finland and
Sweden, light was often the limiting factor for
primary producers (Karlsson and others 2009), in
agreement with what has been observed in several
Estonian waterbodies (Arst and others 2008).
Strong light attenuation in these lakes irrespective
to trophic state might favour planktonic to benthic
production ratios greater than 1. However, most of
the lakes in the southern Baltic Sea catchment are
relatively shallow (z<5 m), and exhibit a large
euphotic zone relative to lake surface area which
might lead to a ratio smaller than 1. It is thus dif-
ficult to assess autotrophic structure of hemiboreal
lakes based solely on existing literature as limno-
logical parameters lead to contradictory conclu-
sions.
The increasing availability of mechanistic, statis-
tical and conceptual models and limnological
monitoring data in large, publicly available data-
bases facilitates the modelling of whole-lake me-
tabolic parameters (Solomon and others 2013;
Cremona and others 2014a,b). It is thus possible to
test the hypothesis of Sand-Jensen and Borum
(1991) that ‘‘the proportions of different pho-
totrophic communities in lakes (…) are pre-
dictable from the size, morphometry, depth and
nutrient status of the system’’. Vadeboncoeur and
others (2008) published a simple conceptual model
called lake autotrophic structure (LAS) which is
based on a 12-equation chain for assessing whole-
lake primary production in planktonic and benthic
F. Cremona and others
zones across size and nutrient gradients. LAS in-
cludes a limited set of climatic, nutrient and mor-
phometric input parameters that are often
routinely measured during lake monitoring and are
thus easily obtainable for modelling purposes.
Additionally, LAS can be customized with the aim
of modelling across small time steps and depth
increments (Higgins and others 2014). The model
has been designed for a conceptual cross-lake
comparison and for numerically exploring auto-
trophic structure. It can thus be applied along a
wide range of data gradients corresponding to most
of the limnological characteristics of lakes
(Vadeboncoeur and others 2008).
The objective of this study was to use LAS model
for assessing primary production partitioning be-
tween benthos and phytoplankton in a group of
hemiboreal and boreal lakes in Estonia, Finland
and Sweden. Our first working hypothesis was that
lake size is inversely correlated with benthic frac-
tion of PP as larger lakes tend to be phytoplankton
dominated. Our second hypothesis was that ben-
thic and planktonic primary productions of the
study lakes are light limited.
METHODS
Study Lakes
We selected a group of 11 Estonian lakes repre-
senting more than 95% Estonian inland water
volume, one Finnish lake (Vanajanselka
¨) and one
Swedish lake (Erken). Limnological parameters
(Chl a, TP, DOC, morphometric and bathymetric
data) were obtained from Estonian National Envi-
ronmental Monitoring Program (http://seire.
keskkonnainfo.ee) and grant IUT 21-2 for Esto-
nian lakes and from Finnish Environmental Insti-
tute for Lake Vanajanselka
¨. For Lake Erken,
nutrient levels and light extinction are based on a
routine monitoring programme with sampling at
approximately every 2 weeks. As no DOC mea-
surements were available for this lake, we em-
ployed the Asmala and others (2012) model which
is based on in situ values of light absorption at
420 nm. Mixing depths and thermal structure of
Erken are based on a buoy-based automated
monitoring system that provided high-resolution
temperature profiles (0.5 m and 30 min) through-
out the stratified period. Lake size of the sampled
lakes ranged across several orders of magnitude,
from huge Lake Peipsi, which with 2611 km
2
is
Europe’s fourth largest lake, to small A
¨ntu Sinija
¨rv
(2 ha). Besides Peipsi, Lake Vo
˜rtsja
¨rv (270 km
2
)
and Vanajanselka
¨(103 km
2
), the majority of the
remaining lakes were relatively small (mean sur-
face: 2.2 km
2
) and shallow (mean z: 4.35 m). The
depth ratio (DR) which is the quotient of average
by maximum depth and is an indicator of lake
shape varied between 0.75 (flat bottom lake) and
0.3 (steep-sided lake). The trophic state ranged
from oligotrophic to hypertrophic, spanning two
orders of magnitude for chlorophyll a(Chl a) and
one for total phosphorus (P, Table 1). On most of
the lakes catchment forest is the dominant land
use, followed by agriculture and wetlands. Gener-
ally, the lakes were naturally eutrophic which is a
common feature in Estonian water bodies (Ott and
Ko
˜iv 1999) and were classified as medium to high
alkalinity lakes according to the European Union
Water Framework Directive (WFD, Ministry of the
Environment 2009).
Model
We used a modified version of LAS model
(Vadeboncoeur and others 2008) to assess lake
primary production and autotrophic structure. An
exhaustive description of the model is available in
the above-mentioned article. In summary, LAS is a
conceptual model based on empirical relationships
that takes into account light attenuation, nutrient
availability and morphometric parameters for cal-
culating whole-lake primary production and its
partitioning between planktonic and benthic pro-
ducers. Here we summarize the main structure of
the model and list the changes we implemented to
the model equations. The water column of each
lake was divided into 0.1-m-thick successive layers
for which plankton and benthic production were
calculated down to lake bottom. In darker lakes,
there was often no production (benthic or plank-
tonic) at all after a few metres deep, and the
modelling was stopped in these lakes as soon as we
reached thermocline. We employed a 1–4 years
average for these parameters calculated from
summer season measurements as nutrient levels
can exhibit a strong temporal variability (Berman
and Pollingher 1976). We made our model calcu-
lations with a 10-min increment for a 15-h period
from dusk to dawn. Lake-specific surface available
for periphyton colonization (A
z
) in all lakes except
Erastvere, Mullutu and Peipsi were obtained by
actual bathymetric measurements. Value for A
z
for
the three mentioned lakes was determined with
equation (1) from Vadeboncoeur and others
(2008):
Az¼A01z=zmax
ðÞ½
c;ð1Þ
Autotrophic Structure of Hemiboreal Lakes
Table 1. Limnological Lake Type and Average Values of Parameters Used for Modelling Lakes Primary Production and Autotrophic Structure with
LAS
Lake Trophic status Water framework directive type LA (km
2
)z(m) z
max
(m) DR TP (lgl
-1
)Chla(lgl
-1
)K
d
(m
-1
) DOC (mg l
-1
)pH
Antu Sinija
¨rv Oligotrophic Alkalitrophic 0.02 6 8 0.75 9 (6–17) 0.7 (0.2–2) 0.25 4.7 7.6
Erastvere Hypertrophic Light–coloured soft-water 0.16 3.5 9.7 0.36 45 (40–117) 65 (62–67) 2.96 12.3 7.7
Erken Oligo-mesotrophic Stratified with low alkalinity
1
23.7 9 20.7 0.43 24 (4.9–105) 5 (0.2–33) 0.6 3.4
2
8
Karija
¨rv Eutrophic Stratified with medium alkalinity 0.86 5.7 14.5 0.39 35 (16–77) 14 (2–60) 0.81 9.6 8.2
Mullutu Eutrophic Coastal lake 4.13 1 1.7 0.58 39 (17–60) 5.4 (3–9) 0.58 18.1 8.9
Peipsi Eutrophic – 2611 8.3 12.9 0.64 47 (20–60) 13 (8–28) 1.61 12 8.4
Saadja
¨rv Mesotrophic Stratified with medium alkalinity 7.24 8 25 0.32 20 (10–30) 5 (1–8) 0.43 9.2 7.9
Vanajanselka
¨Eutrophic Large moderately humic lake 103 7.7 24 0.32 27 (15–117) 14.5 (5–23) 2.2 9.6 8
Verevi Hypertrophic Stratified with medium alkalinity 0.13 3.6 11 0.32 59 (20–70) 21 (5–43) 1.64 11.9 8.4
Viisjaagu Mesotrophic Alkalitrophic 0.23 7.5 13 0.57 11 (10–33) 3.8 (1–7) 0.86 8.8 8.5
Vissi Mesotrophic Alkalitrophic 0.05 2.5 4.5 0.55 23 (13–34) 10 (2–18) 0.65 8.8 8.2
Vo
˜rtsja
¨rv Eutrophic – 270 2.8 6 0.56 54 (17–750) 51 (31–80) 2.76 12.5 8.5
U
¨lemiste Eutrophic Non-stratified with medium alkalinity 9.44 2.5 4.2 0.59 48 (22–55) 26 (14–42) 3.50 13.7 8.5
Numbers into brackets represent minimum and maximum summertime daily averages values, respectively.
1
No WFD type had been designated for Erken. The Estonian classification has been thus employed for assessing limnological type of this lake.
2
DOC concentration estimated from light absorption data at 420 nm.
F. Cremona and others
where A
0
is lake surface area, cis a shape factor
(c= DR/(1-DR)) and DR is the depth ratio of mean
and maximum depth (z/z
max
).
We considered that lake water column consti-
tuted the main substrate available for phytoplank-
ton colonization. Contrary to the Vadeboncoeur
and others (2008) model, we did not assess lake
volume with the shape factor as there was only a
weak relationship between lake size and depth in
our samples (F. Cremona unpubl.). We instead
calculated lake volume as a succession of truncated
cones (Carpenter 1983):
Vz¼z
3AzDzþAzþffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi
AzDzAz
p
;ð2Þ
where A
z-Dz
is the area of the upper surface and A
z
is the area of the lower surface of the stratum
whose volume is to be calculated.
Diurnal variability in surface irradiance and light
attenuation were calculated as with McBride’s
(1992) description of Beer’s law (Kirk 1994)
(equations (7) and (8) in Vadeboncoeur and others
(2008) article):
I0t¼I0max sin pt
daylen
;ð3Þ
where I
0t
is the surface light (lmol m
-2
s
-1
) at time
t,I
0max
is the surface light at solar noon (set to
1500 lmol m
-2
s
-1
), tis the number of hours after
sunrise and daylen are hours of daylight (set to
15 h). For light at depth below surface, the fol-
lowing equation was employed:
Izt ¼I0teKdzð4Þ
with K
d
being the lake-specific light attenuation
coefficient. For the majority of our lakes, this
coefficient was obtained from Paavel and others
(2006) spectrum recreation of Estonia lakes. In
lakes Erken and Vanajanselka
¨,K
d
was estimated
from the measured attenuation of PAR made using
a LICOR LI192 underwater PAR sensor. Otherwise,
instead of using an arbitrary method for assessing
K
d
we employed the modified empirical relation-
ship from Armengol and others (2003) where
K
d
= 1.448/SD with SD being Secchi depth (m).
Phytoplankton maximum productivity PPP
max
(mgCm
-3
h
-1
) was calculated from Guildford and
others (1994) using Chl aconcentration as a pre-
dictor of phytoplankton carbon fixation:
PPPmax ¼2:2Chl:ð5Þ
Depth-specific phytoplanktonic production was
determined by Chl a, solar irradiance and mor-
phometry:
PPPz¼DtX
dusk
dawn
PPPmax tanh Izt=Ik
ðÞVzVzDz
ðÞð6Þ
with Dtbeing the time increment equal to 1/6 h
(that is 10 min), I
k
being the light intensity at onset
of saturation fixed at 180 lmol m
-2
s
-1
for phy-
toplankton and D
z
the depth increment (0.1 m).
To assess whole-lake total PPP, the primary pro-
duction profile time series at different depth inter-
vals obtained from our equation (6) were summed
from surface down to the thermocline (stratified
lakes) or lake bottom (unstratified lakes) depending
on lake type.
For calculating benthic production, we used the
modified relationship from Godwin and others
(2014) where the maximum rate of periphytic
primary production is strongly correlated to light
availability, which is mostly dependent on dissolved
organic carbon concentration (DOC, mg l
-1
). In-
deed, Godwin and others (2014) observed that DOC
was the best predictor of benthic primary production
and also the main driver (66–86%) of light extinc-
tion in northern temperate lakes that exhibited
phosphorus and algal content within the ranges of
TP 6–70 lgl
-1
and chl a2–21 lgl
-1
. We made the
assumption that the same relationship was valid for
our North-European study lakes as they encompass
well these DOC, TP and chl aranges (Table 1).
Maximum benthic primary production (BPP
max
,mg
Cm
-3
h
-1
) is thus a function of DOC:
BPPmax ¼e8:1210:166½DOC
24 ð7Þ
with standard deviation of intercept within the
exponent being 0.574 and that of slope coefficient
being 0.046. Benthic primary production was cal-
culated as a function of incident light and surface
available:
BPPz¼DtX
dusk
dawn
BPPmax tanh Izt=Ik
ðÞAzDzAz
ðÞ:ð8Þ
For I
k
(light intensity at onset of saturation) va-
lue, we chose Liboriussen and Jeppesen (2006)
value of 100 lmol m
-2
s
-1
measured for periphy-
ton in Danish shallow lakes. As for planktonic
production, total benthic production was calculated
by summing depth-specific production values from
surface to bottom of lake or thermocline depending
of lake mixing regime. Benthic fraction (BF) was
calculated as the proportion (ranging from 0 to 1)
Autotrophic Structure of Hemiboreal Lakes
of total primary production carried out by benthic
producers.
Statistical Analyses
LAS output values of planktonic and benthic pro-
duction were compared with input parameters and
TP by regression analysis. Unlike some other recent
primary production models, for example, BaMM
(Holtgrieve and others 2010; Cremona and others
2014b), the LAS model does not include uncer-
tainty analysis in its core structure. Even though
the model is relying on mechanisms that are (1)
based on empirical relationships which are sup-
ported by literature, and (2) constrained within
realistic values, because we employed mostly data
from in situ measurements, we decided to imple-
ment error calculation for photosynthetic capacity
(PPP
max
and BPP
max
) of the benthic and pelagic
primary producers in order to give a better over-
view of autotrophic structure variability. This was
done with Bayesian methods for PPP
z
and BPP
z
(equations (5) and (6)) without considering vol-
ume, surface, light and time increment which are
all constants that is only errors on PPP
max
and
BPP
max
were accounted for. WinBUGS 1.4.3 (Lunn
and others 2000) statistical software was employed
for assessing error propagation. In WinBUGS,
PPP
max
and BPP
max
were designed as stochastic
functions with a normal distribution. For each lake,
upper and lower bounds for PPP
max
were selected
according to the maximum and minimum values
measured for Chl a(for PPP
max
) during monitoring
studies and calculated with Eqs. (5)and(6). For
BPP
max
, the error was calculated by assigning the
standard deviations of parameters of equation (7)
and [DOC] value corresponding to the lake. PPP
zt
and BPP
zt
were then calculated as a logical product
of PPP
max
or BPP
max
and tanh(I
zt
/I
k
). The Markov
Chain Monte-Carlo (MCMC) simulation was run
until convergence (checked graphically) was at-
tained which generally needed less than 5000
iterations. As it is generally recommended, poste-
rior distributions obtained from the simulation are
given here within 95 and 5 credible intervals
(Holtgrieve and others 2010).
RESULTS
Planktonic Primary Production
Pelagic whole-lake primary production of phyto-
plankton (PPP) ranged from 54 mg C m
-2
day
-1
in
oligotrophic A
¨ntu Sinija
¨rv to 1503 mg C m
-2
day
-1
in hypertrophic Lake Erastvere (Figure 1). Mini-
mum and maximum values of modelled PPP ranged
from 41 mg C m
-2
day
-1
in A
¨ntu Sinija
¨rv to more
than 4200 mg C m
-2
day
-1
in Karija
¨rv, with usually
one order of magnitude variation within each lake.
There were strong differences in the diurnal
isopleths of PPP among lakes, with the largest effect
apparently being related to lake transparency
(Figure 2). In the very transparent and oligotrophic
A
¨ntu Sinija
¨rv, PPP at 3 m was still half of its surface
value and in the bottom of the lake PP was only
one order of magnitude lower than on the surface.
In transparent lakes like A
¨ntu Sinija
¨rv and Mul-
lutu, PPP profiles exhibited a flat bottom curve as
nutrients were more limiting than light, meaning
that PPP reached a plateau a few hours after sunrise
and could not increase during the rest of the day.
Conversely in turbid, nutrient-rich lakes (Vo
˜rtsja
¨rv,
U
¨lemiste, Peipsi, Erastvere, Vanajanselka
¨and Ver-
evi) PPP declined from maximum in surface to a
negligible fraction after 1 m. In these lakes where
light is often more limiting than nutrients, PPP
curve is similar to a flat bowl. Indeed, light-satu-
rating conditions appeared to be met only in the
upper part of the epilimnion, even when incident
light was the strongest (that is, noon). A third
category of PPP curve could be observed in inter-
mediate lakes such as Erken, Karija
¨rv, Viisjaagu,
Vissi and Saadja
¨rv which were light and nutrient
limited but with greater transparency than the
turbid lakes. Consequently, the depth distribution
of PPP in these lakes adopted a half-circle shape
Figure 1. Biplot of simulated whole-lake benthic (BPP)
and planktonic (PPP) primary production (mg C m
-2
day
-1
) in thirteen lakes with the lake autotrophic
structure (LAS) model. Dashed line represents the
PPP:BPP ratio corresponding to 1.
F. Cremona and others
Figure 2. Dynamic profiles of instant planktonic primary production (mg C) as modelled for each lake from sunrise to
sunset. Warmer hues correspond to greater lake-specific planktonic primary production as indicated in legend (Color
figure online).
Autotrophic Structure of Hemiboreal Lakes
with PPP reaching a peak at noon and light pene-
trating deep enough to allow substantial produc-
tion below the surface.
Benthic Primary Production
Benthic production based on LAS averaged from
14 mg C m
-2
day
-1
in turbid Lake Erastvere to
900 mg C m
-2
day
-1
in clear and shallow Lake
A
¨ntu Sinija
¨rv. Eight lakes out of 13 exhibited BPP
values lower than 100, which was especially
noticeable in turbid Peipsi, Vanjanselka
¨, Verevi,
Vo
˜rtsja
¨rv and U
¨lemiste. Clear but steep-sided Lake
Viisjaagu also exhibited low benthic production as
it offered no suitable substratum for extensive
benthic growth. Minimum and maximum theo-
retical BPP values ranged from 1 mg C m
-2
day
-1
in Erastvere to close to 3000 mg C m
-2
day
-1
in
Mullutu with a variation of two orders of magni-
tude within each lake.
Benthic production profiles were strongly influ-
enced by light conditions (Figure 3). Clear-water
lakes A
¨ntu Sinija
¨rv and Mullutu hosted high PP
from the surface to the bottom. In turbid lakes,
despite more favourable substrate conditions, BPP
rapidly vanished below the surface. Lakes with
relatively steep littoral zones like Viisjaagu and
Vissi exhibited lower BPP below the surface com-
pared to deeper parts of the basin, and lower PP
than expected on the base low water turbidity.
Benthic Fraction
Benthic fraction (BF) ranged from an average 0.01
in Erastvere to 0.94 in A
¨ntu Sinija
¨rv (Table 2). Lake
Mullutu was the only lake with a balanced auto-
trophic structure. When the full range of produc-
tion values was taken into account, only a few
lakes could be categorized: A
¨ntu Sinija
¨rv with a BF
between 0.4 and 1 was clearly dominated by BPP,
and Lake Erastvere with a BF below 0.13 as well as
Vo
˜rtsja
¨rv, U
¨lemiste and, to a lesser extent, Peipsi
were dominated by PPP. The other lakes exhibited
too large variations of benthic fraction values to
determine conclusively about the resilience of their
autotrophic structure. The high uncertainty signi-
fied that these plankton-dominated lakes (as indi-
cated by their low average BF) might indeed
experience occasional or seasonal switch to peri-
phyton-dominated state in low phytoplankton
(that is, Chl a) biomass, high water transparency
periods.
According to the profiles of BF it appeared that
the fraction of BPP increased with depth (Figure 4).
This increase corresponded to larger flat surfaces in
the deeper parts of a lake for BPP relative to the
overlying water volume. The lakes with the flattest
bottoms—A
¨ntu Sinija
¨rv, Peipsi and Vissi—also
exhibited the highest benthic fraction increase with
depth and often reached a benthic fraction of 1
close to the sediment where all the production may
be carried out by the epipelic algae. However, it
should be noted that this increase in the benthic
fraction was concurrent to a very large decrease in
total production (Figures 3and 4), meaning that
absolute rates of production at the lake bottom are
several orders of magnitude lower than in the
surface, especially for turbid lakes, and thus did not
contribute substantially to total lake production.
Conversely, in steep-sided lakes like Erastvere and
Karija
¨rv benthic fraction remained consistently
marginal and well below 20% of the whole-lake
primary production.
Model Input Values and Autotrophic
Structure
Benthic fraction was negatively correlated to light-
and nutrient-connected parameters (Figure 5). The
coefficients of determination were the strongest for
the following parameters: K
d
(r
2
= 0.70), Chl a
(r
2
= 0.67), TP (r
2
= 0.23) and DOC (r
2
= 0.15).
The K
d
threshold beyond which benthic fraction
totalled less than half of the total primary produc-
tion is situated about 0.5 m
-1
. Benthic fraction
was, however, positively correlated to depth ratio
(r
2
= 0.3), meaning that the flatter lake bottom the
better it was for benthic relative contribution to
whole-lake primary production. Benthic fraction
had no relationship with the surface area of the
lake basins, or their average and maximum depths.
DISCUSSION
Strong Influence of Light- and Nutrient-
Connected Variables
The results did not support our first hypothesis that
in the study lakes, lake surface area was inversely
correlated with the benthic fraction producer of
primary production. Model simulation demon-
strated that our study lakes, irrespective of their
size, were generally phytoplankton dominated over
BPP. One half of our lakes seemed to be light lim-
ited in agreement with our second hypothesis. In
the rest, PP was light and/or nutrient limited, sug-
gesting that light and nutrients influenced strongly
the autotrophic structure of the lakes and impacted
negatively on the BPP. On the other hand, lake
shape as measured by the depth ratio was positively
F. Cremona and others
Figure 3. Dynamic profiles of instant benthic primary production (mg C) as modelled for each lake from sunrise to sunset.
Warmer hues correspond to greater lake-specific benthic primary production as indicated in legend (Color figure online).
Autotrophic Structure of Hemiboreal Lakes
correlated with the benthic fraction but in many
lakes the direct correlation with light parameter
was stronger than shape. Indeed, turbid, nutrient-
rich flat lakes like Peipsi, Vo
˜rtsja
¨rv and U
¨lemiste
exhibited very low benthic production and benthic
fraction despite favourable bathymetric conditions.
Steep-sided lakes that were relatively nutrient poor
(like Viisjaagu and Saadja
¨rv) were nevertheless
dominated by phytoplankton. Althouse and others
(2014) observed an increase in planktonic pro-
duction relative to benthic production in nutrient-
rich, turbid zones of Lake Michigan (USA). As
water nutrients have a stronger relationship with
planktonic than benthic production, light has an
opposite relationship (Hecky and others 1993;
Vadeboncoeur and others 2001). The inter-corre-
lated combination of high nutrient content and K
d
caused a theoretical dominance of phytoplanktonic
producers in most of the lakes that we modelled.
These findings are in agreement with the resource
competition theory between phytoplankton and
benthos (Ja
¨ger and Diehl, 2014) which describes
the negative feedback pelagic and benthic primary
producers have on each other’s growth. Benthic
production reduces nutrient influx from the sedi-
ments to the pelagic habitat, whereas light atten-
uation caused by phytoplankton development
limits benthic algae growth. Some lakes might have
reached the ‘‘nutrient breakpoint’’ described by
Ja
¨ger and Diehl (2014) beyond which phyto-
plankton manages to eliminate its benthic com-
petitors. The threshold is lowered in DOC-rich lakes
like the ones in the hemiboreal region because in
those lakes the compensation depth is situated well
above the mean depth. We thus suggest that phy-
toplankton-dominated autotrophic structure is a
common feature of lakes with turbid and nutrient-
rich water in the catchment of the Baltic Sea
(Scheffer and others 1997; Sand-Jensen and Staehr
2007). Despite that phytoplankton contributed the
largest share of whole-lake PP in our lakes, its
importance might not necessarily be reflected in
organic matter fluxes towards higher trophic levels.
A growing corpus of studies have been highlighting
the uncoupling between primary production and
organic matter transfer so that in lakes with
plankton-based autotrophic structure, organic
matter which was transferred to upper trophic le-
vels was mostly periphyton derived (Bertolo and
others 2005; Vander Zanden and others 2011).
Model Sensitivity
In our model simulations, autotrophic structure
was mostly constrained by lake-specific bathymetry
and light availability during a typical sunny day.
However, the variation of natural conditions makes
it difficult to predict the autotrophic structure based
only on these factors. A simple sensitivity analysis
of two important input variables (maximum pri-
mary production for plankton and periphyton)
proved that only four to five lakes could decisively
considered being phytoplankton or periphyton
dominated. It means that the applicability of the
model is lower on lakes that are very heteroge-
neous on some other important model variables
(substrate availability, light extinction), which is
often the case for larger lakes. Although using long-
term averages as input variables to the model
equations resulted in a relatively consistent general
Table 2. Modelled Planktonic (PPP) and Benthic (BPP) Primary Production (mg C m
-2
day
-1
) and Benthic
Fraction (BF) of the Sample Lakes
Lake PPP BPP BF
A
¨ntu 54 42–147 898 116–2597 0.94 0.43–0.98
Erastvere 1503 1459–1554 14 1–224 0.01 0–0.13
Erken 544 198–3465 259 38–1095 0.32 0–0.84
Karija
¨rv 1065 330–4132 148 15–1293 0.12 0–0.79
Mullutu 106 57–174 92 3–2959 0.46 0.01–0.98
Peipsi 536 336–1101 64 4–692 0.1 0–0.67
Saadja
¨rv 634 128–983 228 21–1876 0.26 0.02–0.93
Vanajanselka
¨442 151–679 57 5–454 0.11 0.01–0.75
Verevi 821 212–1620 74 5–1061 0.08 0–0.82
Viisjaagu 301 90–471 46 4–354 0.13 0.01–0.79
Vissi 585 135–1029 337 31–2866 0.36 0.03–0.95
Vo
˜rtsja
¨rv 1156 721–1782 56 3–656 0.04 0–0.47
U
¨lemiste 453 282–790 55 1–245 0.1 0–0.46
Each set of two columns corresponds to average value followed by 5 and 95 credible intervals, respectively.
F. Cremona and others
picture of a lake’s autotrophic structure, it is pos-
sible that a lake might undergo occasional or even
regular shifts of autotrophic structure caused by
variability in forcing or anthropogenic pressures.
Episodes of water mixing, increased turbidity and
nutrient resuspension might favour the develop-
Figure 4. Benthic production as a proportion of whole-lake production (benthic fraction). Contour graph of benthic
fraction dynamic profiles were modelled for each lake from sunrise to sunset, down to thermocline for stratified lakes and
to lake bottom for unstratified lakes. Warmer hues correspond to larger benthic fraction as indicated in legend (Color
figure online).
Autotrophic Structure of Hemiboreal Lakes
ment of phytoplankton at the expense of periphy-
ton, whereas windless, calm days might have
opposite effects: increase of water transparency
would favour periphyton growth in the short term
but might trigger also a cyanobacterial bloom in
nutrient-rich lakes like Vanajanselka
¨. Cyanobacte-
ria scum would then reduce light penetration into
the water column (Chen and others 2003). Our
application of the LAS model is realistic because the
range of possible input values and model parame-
ters is already well known and often based on local
measurements. Besides light and nutrient param-
eters, lake morphometry and bathymetry are also
critically important as they define the ratio of
planktonic volume to littoral surface area, that is,
respectively, phytoplankton and periphyton habitat
sizes. We encourage users of LAS to employ mea-
sured lake bathymetric data instead of assessing
lake shape with a shape factor proxy, especially for
groups of lakes that differ little in this aspect.
A Tool for Exploring Theoretical
Autotrophic Structure
Because of the lack of reliable information on the
BPP in lakes, modelling is usually the only way to
Figure 5. Biplots of benthic fraction (BF) values of thirteen hemiboreal lakes calculated with LAS model versus seven
input parameters and TP (mg l
-1
). Equations are given for statistically significant (P<0.05) relationships only. SE
a
standard error of slope, SE
b
standard error of intercept.
F. Cremona and others
assess the autotrophic structure of lakes
(Vadeboncoeur and others 2008). It also enables an
assessment of the limiting factors of PP in lakes. To
date, LAS has been a useful tool for this purpose,
driving primary production research from a one-
dimensional plankton-based approach to a two-
dimensional approach that includes benthic pro-
duction. The fact that we modified some of the
input parameters (mainly I
k
and BPP
max
) according
to the light and primary production relationships
that are prevailing in the Baltic region improved
the applicability of LAS. However, it should be
specified that the method that we employed for
calculating BPP
max
is valid essentially for Northern
temperate lakes that have naturally high DOC and
TP and Chl aranging 2–20 and 6–70 lgl
-1
,
respectively (Godwin and others 2014). In lower
latitude lakes that have often longer growing
periods for phytoplankton and where light extinc-
tion might be essentially caused by particles (Chen
and others 2003), the applicability of the BPP
max
formula is more questionable, however, has some
drawbacks as well. The first one is that in natural
conditions the peak of primary productivity is often
reached well below the surface because the mixed
layer depth is situated below the lower limit of the
euphotic zone and also considering the harmful
effects of UV radiation on cells in the surface layer
(Thornton and others 1990). However, in the LAS
model neither UV attenuation by water or water
currents nor algae sinking is taken into account.
Secondly, this model does not separate between
different types of benthic microalgae, like epipelon,
epipsammon, epilithon which can display very
different growth rates and requirements. It should,
however, be noted that LAS was developed for a
conceptual analysis of the importance of benthic
primary production in lakes and its primary pur-
pose is that of a comparative tool for cross-LAS
rather than a lake-specific descriptive model.
Numerically exploring theoretical primary produc-
tion partition in a lake is conceptually much dif-
ferent from predicting primary production
dynamics based a mechanistic approach. Thirdly,
LAS would benefit greatly from additional equa-
tions describing primary production of the third
contributor to autotrophy, that is, macrophyte-
specific productivity. Macrophyte production is
included indirectly only as one of the factors that
affect areal periphyton production (Vadeboncoeur
and others 2008). Macrophytes can constitute an
important component in energy fluxes to higher
trophic levels, comparable to phytoplankton and
periphyton (Kalff 2002; Cremona and others 2009),
and underestimation of their contribution to total
primary production could be problematic from a
mass balance perspective. However, the dual pri-
mary producer paradigm, which the model rests
on, falls short of accurately describing a lake type
that is shallow, soft bottomed and relatively poor in
Chl a. This type of lake is often dominated by free
floating (for example, Lemna sp.) or canopy-form-
ing macrophytes such as water lily (Nuphar sp.).
Rapid light extinction below the surface in these
lakes caused by a macrophyte canopy can signifi-
cantly impede phytoplankton and periphyton pro-
duction during the growing season (Sand-Jensen
and Borum 1991). A LAS model user wishing to
assess macrophyte-specific production thus faces a
dilemma. If light values above the macrophyte ca-
nopy are entered as an input value, the modeller
will obtain an overestimated periphyton produc-
tion which might be used as a proxy for macro-
phyte and periphyton production combined, but is
still structurally different. On the other hand, if
below-canopy light values are employed, the
model would give a false picture of lake primary
production as macrophyte productivity is com-
pletely ignored. This issue needs consideration
regarding shallow lakes, especially those that adopt
self-stabilizing steady states dominated alterna-
tively by phytoplankton and macrophyte assem-
blages with periphyton playing a more marginal
role (Liboriussen and Jeppesen 2003). It could be
necessary to incorporate factors into the LAS model
that are crucial requirements for macrophytes such
as substratum structure, shoreline length and
exposure to wind.
CONCLUSION
The numerical exploration of LAS revealed phyto-
plankton-dominated conditions in the majority of
the study lakes caused by strong light extinction,
and nutrient content, irrespective of lake mor-
phometry. Consequently, lakes from the Baltic Sea
catchment that exhibit the same turbid, nutrient-
rich properties should be commonly dominated by
phytoplankton primary producers as well. As a rule
of thumb, an average light extinction coefficient of
0.5 m
-1
can be considered as the threshold beyond
which a hemiboreal lake primary production will
be phytoplankton dominated. Ideally, the model
should be updated with additional equations so
that an autotrophic structure concept based on
three end members, phytoplankton, periphyton
Autotrophic Structure of Hemiboreal Lakes
and macrophytes, as it was envisioned by many
limnologists, might become quantifiable.
ACKNOWLEDGEMENTS
The authors are grateful to Toomas Ko
˜iv, Ingmar
Ott and Pille Meinson for assistance in data ana-
lysing. Sean C. Godwin (Simon Fraser University)
provided great help for the benthic production
calculations. This research was supported by Start-
Up Personal Research Grant PUT 777 to FC and IUT
21-2 of the Estonian Ministry of Education and
Research, Estonian Science Foundation grant
ETF9102, the EU through the European Regional
Development Fund, program ‘‘Research Interna-
tionalisation’’ project LIMNO, the Swiss Grant
‘‘Enhancing public environmental monitoring
capacities’’ and MARS project (Managing Aquatic
ecosystems and water Resources under multiple
Stress) funded under the 7th EU Framework Pro-
gramme, Theme 6 (Environment including Climate
Change), Contract No.: 603378 (http://www.mars-
project.eu).
REFERENCES
Althouse B, Higgins S, Vander Zanden J. 2014. Benthic and
planktonic primary production along a nutrient gradient in
Green Bay, Lake Michigan, USA. Freshw Sci 33:487–98.
Armengol J, Caputo L, Comerma M, Feijoo
´C, Garcı´a JC, Marce
´
R, Navarro E, Ordon
˜ez J. 2003. Sau reservoir’s light climate:
relationships between Secchi depth and light extinction
coefficient. Limnetica 22:195–210.
Arst H, Erm A, Herlevi A, Kutser T, Leppa
¨ranta M, Reinart A,
Virta J. 2008. Optical properties of Boreal lake waters in
Finland and Estonia. Boreal Environ Res 13:133–58.
Asmala E, Stedmon CA, Thomas DN. 2012. Linking CDOM
spectral absorption to dissolved organic carbon concentrations
and loadings in boreal estuaries. Estuar Coast Shelf Sci
111:107–17.
Bertolo A, Carignan R, Magnan P, Pinel-Alloul B, Planas D,
Garcia E. 2005. Decoupling of pelagic and littoral food webs in
oligotrophic Canadian Shield lakes. Oikos 111:534–46.
Berman T, Pollingher U. 1976. Annual and seasonal variations of
phytoplankton, chlorophyll, and photosynthesis in Lake
Kinneret. Limnol Oceanogr 19:31–55.
Carpenter S. 1983. Lake geometry: implications for production
and sediment accretion rates. J Theor Biol 105:273–86.
Chen Y, Qin B, Teubner K, Dokulil MT. 2003. Long-term
dynamics of phytoplankton assemblages: microcystis-domi-
nation in Lake Taihu, a large shallow lake in China. J
Plankton Res 25:445–53.
Che
´telat J, Amyot M. 2009. Elevated methylmercury in High
Arctic Daphnia and the role of productivity in controlling their
distribution. Glob Change Biol 15:706–18.
Cremona F, Hamelin S, Planas D, Lucotte M. 2009. Sources of
organic matter and methylmercury in littoral macroinverte-
brates: a stable isotope approach. Biogeochemistry 94:81–94.
Cremona F, Planas D, Lucotte M. 2010. Influence of functional
feeding groups and spatiotemporal variables on the d
15
N sig-
nature of littoral macroinvertebrates. Hydrobiologia 647:51–
61.
Cremona F, Ko
˜iv T, Kisand V, Laas A, Zingel P, Agasild H,
Feldmann T, Ja
¨rvalt A, No
˜ges P, No
˜ges T. 2014a. From bacteria
to piscivorous fish: estimates of whole-lake and component-
specific metabolism with an ecosystem approach. PLoS One
9:e101845.
Cremona F, Laas A, No
˜ges P, No
˜ges T. 2014b. High-frequency
data within a modeling framework: on the benefit of assessing
uncertainties of lake metabolism. Ecol Model 294:27–35.
Godwin SC, Jones SE, Weidel BC, Solomon CT. 2014. Dissolved
organic carbon concentration controls benthic primary pro-
duction: results from in situ chambers in north-temperate
lakes. Limnol Oceanogr 59:2112–20.
Guildford SJ, Hendzel LL, Kling HJ, Fee EJ. 1994. Effects of lake
size on phytoplankton nutrient status. Can J Fish Aquat Sci
51:2769–83.
Hecky RE, Campbell P, Hendzel LL. 1993. The stoichiometry of
carbon, nitrogen, and phosphorus in particulate matter in
lakes and oceans. Limnol Oceanogr 38:709–24.
Hecky RE, Hesslein RH. 1995. Contributions of benthic algae to
lake food webs as revealed by stable isotope analysis. J N-Am
Benthol Soc 14:631–53.
Higgins SN, Althouse B, Devlin SP, Vadeboncoeur Y, Vander
Zanden MJ. 2014. Potential for large-bodied zooplankton and
dreissenids to alter the productivity and autotrophic structure
of lakes. Ecology 95:2257–67.
Holtgrieve GW, Schindler DE, Branch TA, A’mar Z. 2010.
Simultaneous quantification of aquatic ecosystem metabolism
and reaeration using a Bayesian statistical model of oxygen
dynamics. Limnol Oceanogr 55:1047–63.
Ja
¨ger CG, Diehl S. 2014. Resource competition across habitat
boundaries: asymmetric interactions between benthic and
pelagic producers. Ecol Monogr 84:287–302.
Jeppesen E, Meerhoff M, Davidson TA, Trolle D, Søndergaard M,
Lauridsen TL, Beklioglu M, Brucet S, Volta P, Gonzalez-Ber-
gonzoni I, Nielsen A. 2014. Climate change impacts on lakes: an
integrated ecological perspective based on a multi-faceted ap-
proach, with special focus on shallow lakes. J Limnol 73:88–111.
Kalff J. 2002. Limnology. New Jersey: Prentice Hall.
Karlsson J, Bystro
¨m P, Ask J, Ask P, Persson L, Jansson M. 2009.
Light limitation of nutrient-poor lake ecosystems. Nature
460:506–9.
Kirk JTO. 1994. Light and photosynthesis in aquatic ecosystems.
2nd edn. Cambridge: Cambridge University Press.
Lewis WM Jr. 1996. Tropical lakes: how latitude makes a differ-
ence p. In: Schiemer F, Boland KT, Eds. Perspectives in tropical
limnology. Amsterdam: SPB Academic Publishing. p 43–64.
Liboriussen L, Jeppesen J. 2003. Temporal dynamics in epipelic,
pelagic and epiphytic algal production in a clear and a turbid
shallow lake. Freshw Ecol 48:418–31.
Liboriussen L, Jeppesen E. 2006. Structure, biomass, production
and depth distribution of periphyton on artificial substratum
in shallow lakes with contrasting nutrient concentrations.
Freshw Biol 51:95–109.
Lindeman RL. 1942. The trophic-dynamic aspect of ecology.
Ecology 23:399–418.
McBride GB. 1992. Simple calculation of daily photosynthesis by
means of five photosynthesis-light equations. Limnol Ocea-
nogr 37:1796–808.
F. Cremona and others
Ministry of the Environment. 2009. Procedure for the estab-
lishment of bodies of surface water and a list of the bodies of
surface water the State of which is to be established, classes of
the states and the values of quality indicators corresponding to
these state classes, and the procedure for the establishment of
the classes of state (RTL 2009, 64, 941). Order from
28.07.2009 No. 44. Ministry of the Environment, 11 pp.
(www.riigiteataja.ee/akt/13210253). Estonian.
No
˜ges T, Arst H, Laas A, Kauer T, No
˜ges P, Toming K. 2011.
Reconstructed long-term time series of phytoplankton pri-
mary production of a large shallow temperate lake: the basis
to assess the carbon balance and its climate sensitivity.
Hydrobiologia 667:205–22.
Ott I, Ko
˜iv T. 1999. Estonian small lakes: special features and
changes. Tallinn: Est Environ Inf Centre.
Paavel B, Arst H, Reinart A, Herlevi A. 2006. Model calculations
of diffuse attenuation coefficient spectra in lake waters. Proc
Est Acad Sci 55:61–81.
Sand-Jensen K, Borum J. 1991. Interactions among phyto-
plankton, periphyton, and macrophytes in temperate fresh-
waters and estuaries. Aquat Bot 41:137–75.
Sand-Jensen K, Staehr P. 2007. Scaling of pelagic metabolism to
size, trophy and forest cover in small Danish lakes. Ecosystems
10:128–42.
Scheffer M, Rinaldi S, Gragnani A, Mur LR, van Nes EH. 1997.
On the dominance of filamentous cyanobacteria in shallow,
turbid lakes. Ecology 78:272–82.
Solomon CT, Bruesewitz DA, Richardson DC, Rose KC, Van de
Bogert MC, Hanson PC, Kratz T, Larget B et al. 2013.
Ecosystem respiration: drivers of daily variability and back-
ground respiration in lakes around the globe. Limnol Ocea-
nogr 58:849–66.
Thornton KW, Kimmel BL, Payne FE. 1990. Reservoir limnol-
ogy: ecological perspectives. New York: Wiley.
Vadeboncoeur Y, Lodge DM, Carpenter SR. 2001. Whole-lake
fertilization effects on distribution of primary production be-
tween benthic and pelagic habitats. Ecology 82:1065–77.
Vadeboncoeur Y, Vander Zanden MJ, Lodge DM. 2002. Putting
the lake back together: reintegrating benthic pathways into
lake food web models. Bioscience 52:44–54.
Vadeboncoeur Y, Jeppesen E, Vander Zanden MJ, Schierup HH,
Christoffersen K, Lodge DM. 2003. From Greenland to green
lakes: cultural eutrophication and the loss of benthic path-
ways in lakes. Limnol Oceanogr 48:1408–18.
Vadeboncoeur Y, Peterson G, Vander Zanden MJ, Kalff J. 2008.
Benthic algal production across lake size gradients: interac-
tions among morphometry, nutrients, and light. Ecology
89:2542–52.
Vander Zanden MJ, Vadeboncoeur Y, Chandra S. 2011. Fish
reliance on littoral-benthic resources and the distribution of
primary production in lakes. Ecosystems 14:894–903.
Wetzel RG. 1964. A comparative study of the primary produc-
tion of higher aquatic plants, periphyton, and phytoplankton
in a large, shallow lake. Int Rev Hydrobiol 49:1–61.
Autotrophic Structure of Hemiboreal Lakes
A preview of this full-text is provided by Springer Nature.
Content available from Ecosystems
This content is subject to copyright. Terms and conditions apply.
