Spatiotemporal aspects of silica buffering in restored tidal marshes
Sander Jacobs
a
,
*
, Eric Struyf
a
,
b
, Tom Maris
a
, Patrick Meire
a
a
Department of Biology, Ecosystem Management Research Group, University of Antwerp, Universiteitsplein 1, 2610 Wilrijk, Antwerp, Belgium
b
GeoBiosphere Science Centre, Department of Geology, Lund University, So
¨
lvegatan 12, 22362 Lund, Sweden
article info
Article history:
Received 23 January 2008
Accepted 8 July 2008
Available online 15 July 2008
Keywords:
wetlands
restoration
tidal flats
nutrient cycles
eutrophication
silica
Schelde estuary
Belgium
51
03
0
53
00
N
4
08
0
55
00
E
abstract
Losses of pelagic diatom production resulting from silica limitation have not only been blamed for toxic
algal blooms, but for the reduction in ability of coastal food webs to support higher trophic levels.
Recent research has shown the importance of advective seepage water fluxes of dissolved silica (DSi)
from freshwater marshes to pelagic waters during moments of riverine Si-limitation. In this study, we
investigated the potential impact of recently installed new tidal areas along the Schelde estuary, located
in former polder areas and characterized by so-called controlled reduced tidal regimes (CRT). Nine
mass-balance studie s were conducted in a newly cons tructed CRT in the freshwater Schelde estuary.
During complete tidal cycles both DSi and amorphous silica (ASi) concentrations were monitored at the
entrance culverts and in different habitats in the marsh. A swift DSi-delivery capacity was observed
despite the shifted spatiotemporal frame of exchange processes compared to reference marshes. As
silica-accumulating vegetation is not yet present, and difference with reference marshes’ deliveries is
surprisingly small, we indicate diatomaceous debris and phytoliths to be the main silica source.
Although further research is necessary on the driving forces of the different processes involved,
restoration of former agricultural areas under CRT-regime provide the potential to buffer silica in the
estuary.
Ó 2008 Elsevier Ltd. All rights reserved.
1. Introduction
Estuaries are biogeochemical hot-spots and are amongst the
most productive ecosystems of the world (Costanza et al., 1993). As
the interface between terrestrial and coastal waters, they support
processes that are central to the planet’s functioning (Costanza
et al., 1997). Estuaries are characterized by steep chemical gradients
and complex dynamics, resulting in major transformations in the
amount, the chemical nature and the timing of material fluxes.
Eutrophication is one of the most important problems that
confronts these systems. Eutrophication phenomena in estuaries
are related to the balance between N, P and Si in river loading, and
are thus dependent on the interactions between human activity
and natural processes in the watershed, which ultimately deter-
mine the riverine nutrient delivery into the marine environment
(Officer and Ryther, 1980; Billen and Garnier, 1997; Lancelot et al.,
1997; Cugier et al., 2005). Eutrophication can cause anoxia, extreme
turbidity and even toxicity in coastal areas and lakes, mostly
provoked by shifts in plankton community following excessive
inputs of N and P compared to Si. Decreases in the availability of
silica relative to N and P in estuaries may result in a shift in the
phytoplanktonic community from a dominance of diatoms to other
phytoplankton forms as cyanobacteria or toxic dinoflagellate,
affecting zooplankton and fisheries (see also Chı
´
charo et al., 2006;
Wolanski et al., 2006). Losses of diatom production, resulting from
silica limitation, have not only been blamed for these toxic algal
blooms, but for the reduction in ability of coastal food webs to
support higher trophic levels (Treguer et al., 1995; Cugier et al.,
2005; Kimmerer, 2005). Estuarine and marine foodwebs are based
essentially on diatoms (Irigoien et al., 2002; Kimmerer, 2005).
Dissolved silica concentrations have since long been known to
control diatom populations (Wang and Evans, 1969), diatom
blooms (Tessenow, 1966; Schelske and Stoermer, 1971; Davis et al.,
1978), and seasonal succession in plankton communities (Kilham,
1971). In fact, the availability of dissolved silica (DSi) has been
shown to control diatom silica production rates, at least seasonally,
in every natural system examined to date (Nelson and Brzezinski,
1990; Nelson and Treguer, 1992; Brzezinski and Nelson, 1996;
Nelson and Dortch, 1996; Brzezinski et al., 1998; Bidle and Azam,
2001).
Within the estuarine ecosystem, fringing tidal marshes act as
a biogeochemical filter, removing inorganic and organic substances
from the floodwaters and changing substance speciation (e.g.
Gribsholt et al., 2005). The interaction between tidal marshes and
estuaries or coastal zones received much attention through
*
Corresponding author.
E-mail addresses: sander.jacobs@ua.ac.be (S. Jacobs), eric.struyf@geol.lu.se
(E. Struyf).
Contents lists available at ScienceDirect
Estuarine, Coastal and Shelf Science
journal homepage: www.elsevier.com/locate/ecss
0272-7714/$ – see front matter Ó 2008 Elsevier Ltd. All rights reserved.
doi:10.1016/j.ecss.2008.07.003
Estuarine, Coastal and Shelf Science 80 (20 08) 42–52
numerous exchange studies (e.g. Valiela et al., 2007; Spurrier and
Kjerfve, 1988; Whiting et al., 1989; Childers et al., 1993), with
emphasis on C, P and N. Dominant questions were whether
marshes were importing or exporting N, P, C or particulate matter,
often testing the ‘‘outwelling’’ hypothesis (e.g. Dame et al., 1986),
which states that a large part of the organic matter produced in the
intertidal marshes is not used in internal trophic chains but is
transported into the adjacent sea areas and increases their
productivity. Only a limited number of mass balance studies have
targeted freshwater tidal marshes (e.g. Childers and Day, 1988;
Gribsholt et al., 2005; Struyf et al., 200 6). The freshwater systems
are characterized by botanical properties resembling inland fresh-
water wetlands and by more direct contact with human-impacted
river water. These characteristics make freshwater tidal marshes
potentially important process interfaces. Struyf et al. (2006) have
shown the importance of advective seepage water fluxes of
dissolved silica (DSi) from freshwater marshes to pelagic waters
during moments of riverine Si-limitation. Tidal freshwater marshes
contain large amorphous silica stocks in marsh soils, built up
through sedimentation of diatom shells and incorporation of silica
in marsh vegetation (Struyf et al., 2005). Export is the result of
consequent dissolution of this amorphous silica (ASi) in marsh pore
water from litter and sediments, and advective export of marsh
pore- and puddle water between tidal flooding events (Struyf et al.,
2007a,b). Silica limitation of diatoms (Conley et al., 1993; Smayda
1997) and the consequent negative effects on food web structure
may be avoided. However, data are available only from few tidal
freshwater wetlands and conclusions are presently only applicable
on a local scale. Furthermore, a recent review stresses the need for
more research on silica cycling in wetlands, as it rivals their impact
on other biogeochemical cycles and, to date, this topic has not
received sufficient attention (Struyf and Conley, in press).
In this study, we investigated the potential impact of recently
installed new tidal areas along the Schelde estuary, located in
former polder areas and characterized by so-called controlled
reduced tidal regimes (CRTs) (Cox et al., 2006; Maris et al., 2007).
Along the Schelde estuary, more than 50% of marsh area will
eventually be located in such areas, and may result in international
application. This article focuses on the silica biogeochemistry
within these new systems and aims to explore spatiotemporal
patterns of deposition and dissolution in recently flooded formerly
agricultural polder areas. In the first implemented CRT, an intensive
spatiotemporal sampling scheme was carried out during the first
16 months of development. This research expands the growing
awareness that ecosystems and associated biogenically fixed
amorphous Si rather than geological weathering control silica
availability in the aquatic environment on a shorter, biological
timescale (Conley, 2002; Humborg et al., 2004; Derry et al., 2005).
2. Materials and methods
Nine mass-balance studies were conducted in a newly
constructed CRT in the freshwater Schelde estuary: on May 16, July
3, September 10 and 11 and October 10, 2006, and on March 20 and
21 and June 4 and 5, 2007. During nine complete tidal cycles both
DSi and ASi concentrations were monitored at the entrance culverts
as well as in different habitats in the marsh.
2.1. Study area
The Schelde estuary is one of the last European estuaries with
a complete fresh- to saltwater tidal gradient, located in the Neth-
erlands and Belgium. Maps and extensive descriptions of hydrology
and ecology can be found in several recent papers (Temmerman
et al., 2003; Meire et al., 2005; Van Damme et al., 2005; Soetaert
et al., 2004). The studied CRT area is a newly constructed inunda-
tion area, the ‘‘Lippenbroek’’ (surface approximately 80,000 m
2
),
situated at Moerzeke (51
03
0
53
00
N; 4
08
0
55
00
E). Maximal tidal
amplitude in the Schelde at this point is approximately 6 m. The
area was mostly used as cropland (rotation system with Zea mays
and Solanum tuberosum; the lower parts were planted with Populus
sp. trees or over-grown with Salix sp. trees (Fig. 1A). The rotation
system was abandoned in 2003.
During the two-year construction phase (2003–2005), crops
were replaced by pioneer vegetation (mainly Epilobium hirsutum
and Urtica dioica)(Fig. 1B). Part of the polder was devegetated due
to building construction work (Fig. 1B). Tidal inundation was initi-
ated in March 2006. Since the first inundation, vegetation has been
progressively replaced by flood-tolerant species (mainly Lythrum
salicaria, Lycopus europaeus and Phragmites australis)(Fig. 2).
Because site elevation is several metres under mean high water
level, reconstruction of spring-neap tide flooding variation required
the construction of separate inlet culvert and outlet culvert (Maris
et al., 2007). At the riverside, an inlet culvert permits flooding from
4.80 m TAW and higher, whilst a valved outlet culvert guarantees
one-way emptying from 1.5 m TAW and lower (TAW is the Belgian
Ordnance Level, which is approx. 2.3 m below mean sea level at the
Belgian coast). Consequently, only the top of the tidal cycle is
permitted to flood the polder surface. This results in a controlled
reduced tidal area (CRT) with unique tidal features, such as
a pronounced spring–neap variation and a prolonged stagnant
phase (Fig. 3, for details see Cox et al., 2006; Maris et al., 2007). The
marsh is surrounded by a dike at 8 m TAW. Because of the deep
artificial dike bases and thick riverine clay deposit in the CRT,
ground water fluxes were assumed to be small compared to
observed tidal surface water fluxes.
: drainage structures
: Salix wood
: Populus plantage
: bare ground
: pioneer vegetation
: sampling point
3
1
5
8
6
2
3
4
10
11
12
15
16
Schelde
N 50m
: drainage structures
: Salix wood
: Populus plantage
: agricultural fields
: pre-building contours
Schelde
N
ABC
50m
: drainage structures
: Salix wood
: Populu
s plantage
: bare ground
: pioneer vegetation
Schelde
N50m
9
7
13
14
Fig. 1. Schematic overview of study site before (A) and after building works (B). (C) Sampling locations in the CRT. Sampling intensity at locations is given in Tables 1 and 2.
S. Jacobs et al. / Estuarine, Coastal and Shelf Science 80 (2008) 42–52 43
2.2. Sampling
A total of 796 data points were obtained during the nine mass
balance studies. Surface water samples were collected at the
entrance and outlet culvert (1 in Fig. 1C) and in selected habitats
throughout the marsh (2–16 in Fig. 1C). Sampling covered the full
13 h of the tidal cycle for May, July and October 2006 campaigns,
and double cycles of one night (‘‘a’’ in text) plus day (‘‘b’’ in text) of
26 h for September 200 6 and March and June 2007 campaigns.
Sampling intensity was highest during the first campaign
(Tables 1 and 2). This intensity was necessary to explore the spatial
patterns in the marsh; however this exhaustive scheme was not
entirely repeated during all campaigns. The selection of habitats
during subsequent campaigns was based on maximal cover of
different habitat features. A selection of samples was analysed for
ASi (Table 2), also covering different habitat features.
Samples were taken approximately 10 cm below the surface,
and stored in dark incubators at 5
C for a maximum of 24 h.
Fig. 2. Vegetation development in devegetated zones. A and B (upper) show overview; C and D (lower) detail. A and C are taken in spring 2006 (1 month after first inundations),
B in summer 2006, and D in summer 2007.
0 100 200 300 400 500 600 700
0
5000
10000
15000
20000
Time
(
minutes
)
Volume (m
3
)
Instream Stagnant Bulk outstream Seepage outstream
Fig. 3. Water mass balance during a typical tide in Lippenbroek. Inset illustrates tidal curves at typical neap (–), mean ($$$) and spring (–) tide outside and inside Lippenbroek CRT
(from Cox et al., 2006). Grey lines indicate (0) start instream, (110) stop instream ¼ start stagnant phase, (220) start outstream ¼ stop stagnant phase and (580) stop outstream.
Outstream consists of a bulk outstream (overmarsh tidal frame) and a seepage phase (here at approx. 340 min. Phase lines are indicated in relevant figures throughout the MS.
S. Jacobs et al. / Estuarine, Coastal and Shelf Science 80 (2008) 42–5244
Dissolved silica (DSi) was analysed on a Thermo IRIS ICP (Induc-
tively Coupled Plasmaspectrophotometer) (Iris
Ò
). For each of the
samples analysed for ASi (Table 2), three sub samples of 25 ml each
were filtered over 0.45-
m
m filters, from a well-mixed total sample
of 150 ml. After drying at 20
C, ASi was extracted from the filters in
a0.1MNa
2
CO
3
solution at 80
C in a shaker bath. Sub samples were
taken at 60, 120 and 180 min. Blank extractions revealed insignifi-
cant DSi release from filters, recipients or chemicals. [ASi] (mg l
1
)
was then calculated by extrapolating the linear line through the
three extraction points in a time-extracted silica plot (DeMaster,
1991). This approach corrects for the additional release of Si from
mineral silicates. The ASi wet-alkaline extraction is prone to addi-
tional release of DSi from amorphous mineral silicates. Despite its
flaws, ASi wet-alkaline extraction is for the moment still the most
representative method to analyse for ASi (Saccone et al., 2007).
2.3. Water and silica mass balances
All calculations and statistical analyses were performed in R (R,
2006). Inlet and outlet culverts are the only exchange points with
the river. Their dimensions are exactly known. Flow velocity was
measured acoustically (Sontek ‘‘Argonaut’’). Water mass balances
were calculated with an averaged discharge value throughout the
water column for every 2 min, assuring accurate volume-weighing
of concentration values during all tidal phases. Measurements,
calibration and operation of the flowmeters were performed by
Flanders hydraulics research laboratory (W&L) experts. Concen-
tration profiles as well as nutrient discharges were calculated as
6004002000
3.0
3.5
4.0
May 06
600400
200
0
0
1
2
3
4
5
6
Jul 06
6004002000
4.0
4.5
5.0
5.5
6.0
Sep 06a
6004002000
4.0
4.5
5.0
Sep 06b
6004002000
3.8
4.0
4.2
4.4
4.6
4.8
5.0
Oct 06
6004002000
6.3
6.4
6.5
6.6
Mar 07a
6004002000
6.2
6.3
6.4
6.5
6.6
6.7
Mar 07b
600400
2000
1.0
1.5
2.0
2.5
3.0
3.5
4.0
Jun 07a
6004002000
1.0
1.5
2.0
2.5
3.0
3.5
Jun 07b
Time
(
min
)
DSi-concentration (g.L
-1
)
Fig. 4. DSi concentration profiles at in- and outstream location for all campaigns (tidal phases as in Fig. 3). Campaign month is indicated above each subpanel (dots are measured
concentrations, lines represent linear extrapolation).
Table 1
Sampling intensity of DSi samples at different locations (see Fig. 1C) and during
different campaigns
Location May
06
Jul
06
Sep
06a
Sep
06b
Oct
06
Mar
07a
Mar
07b
Jun
07a
Jun
07b
Total
1 53 8 15 17 24 16 1 1 15 15 174
21623571112 55129
8 7 12 12 12 43
61010 12 5542
5108 12 30
388 12 28
1 7 21 21
4109 19
15 16 16
16 14 14
12 12 12
98 8
11 7 7
10 5 5
14 5 5
13 3 3
73 3
Total 179 95 84 40 84 16 11 25 25 559
Table 2
Selection of samples analysed for ASi at different locations (see Fig. 1C) and during
different campaigns
Location May
06
Jul
06
Sep
06a
Sep
06b
Oct
06
Mar
07a
Mar
07b
Jun
07a
Jun
07b
Total
11216451286151593
297712 5545
665 12 5533
84612 22
33 12 15
1 7 12 12
56 6
46 6
11 5 5
Total 35 45 15 18 60 8 6 25 25 237
S. Jacobs et al. / Estuarine, Coastal and Shelf Science 80 (2008) 42–52 45
linear interpolations (Becker et al., 1988). ASi concentrations were
measured on average 10 times (range 4–16) and DSi concentrations
19 times (8–53) along each tidal cycle at the culverts. Interpolation
provided 700 values/tidal cycle, for discharge (D)aswellas
concentrations. These values were used to calculate absolute mass
balance by cumulative summing of (D (m
3
) [Si] (mg l
1
)) along
the instream and outstream phase separately. Stagnant phase and
volumes entering through small leaks in the outstream culvert
were not taken into consideration.
Total mass balances were first calculated as percentages
((
P
(out)
P
(in))/
P
(in)) in order to compare between different
tidal volumes, and then as absolute masses (
P
(out)
P
(in)). In
a conservative mass balance, it is assumed that there is no net
import or export of water. However, due to inter-tide variations,
stocking or surplus release of water volumes takes place. To
compare between tides, this conservative correction was calculated
as a percentage for each campaign, recalculated on the final mass
balance and shown as a range. However, general patterns were not
seriously influenced by this effect (Figs. 5 and 6).
Additional tidal features were measured in order to compare
between tides: average flooding height was calculated from total
volume of each entering tide and total surface of the study area, water
temperature was continuously monitored at culverts, and DSi and ASi
concentrations were monitored in adjacent river water. General
relationships between silica delivery and these tidal features were
explored through PCA and ANOVA analysis (Chevenet et al., 1994).
3. Results
3.1. Concentration profiles
DSi concentration profiles show different seasonal patterns
(Fig. 4). Instream phases (see Fig. 3) are marked by steep
concentration changes, whilst the fluctuations during stagnant
phase do not exceed 0.2 mg l
1
. Outstream concentration profiles
are highly variable and show increases, decreases or both: at
starting concentrations below 2 mg DSi l
1
, concentrations increase
with 125.0% and 126.6% (June 2007 in Fig. 4) or even with a factor 17
(July 2006 in Fig. 4) at final concentration. For instream concen-
trations higher than 2 mg l
1
, profiles show slight increases (10.5%
to 31.9%, May, September and October 2006 in Fig. 4). When
instream concentrations become higher than 6 mg l
1
, profiles
show a status quo or slight decrease (6%, March 2007 in Fig. 4)
towards final concentrations.
Concentration profiles of ASi present a more variable pattern
over a smaller concentration range (not shown). Although for
September 2006 and March 2007 only a limited number of samples
was analysed for ASi, and differences are generally lower for ASi
compared to DSi, there is a general evolution of increase or status
quo (May, July, September 2006 and October 2006, not shown)
towards strong ASi decrease in ASi- concentration profiles during
later tidal cycle (March and June 2007b, not shown), with the
exception of June 2007a.
3.2. Mass balances
Calculated DSi mass balances indicate enrichment of exported
water in summer months (July 2006 and June 2007, Fig. 5, upper
left), but also in late autumn (October 2006) and during one spring
campaign (March 2007). Although July 2006 shows spectacular
enrichment in percentage, absolute numbers (Fig. 5 upper right)
are lower due to small water mass at neap tide, while the opposite
is true for the 2007 campaigns. ASi mass balances confirm the
transition from slight ASi delivery or status quo towards ASi caption
by the marsh (Fig. 5, lower graphs).
May 06
Jul 06
Sep 06a
Sep 06b
Oct 06
Mar 07a
Mar 07b
Jun 07a
Jun 07b
relative ASi delivery (%)
-150
-100
-50
0
50
100
150
May 06
Jul 06
Sep 06a
Sep 06b
Oct 06
Mar 07a
Mar 07b
Jun 07a
Jun 07b
absolute ASi delivery (kg)
-200
-100
0
100
200
May 06
Jul 06
Sep 06a
Sep 06b
Oct 06
Mar 07a
Mar 07b
Jun 07a
Jun 07b
relative DSi delivery (%)
-100
-50
0
50
100
May 06
Jul 06
Sep 06a
Sep 06b
Oct 06
Mar 07a
Mar 07b
Jun 07a
Jun 07b
absolute DSi delivery (kg)
-150
-100
-50
0
50
100
150
Fig. 5. DSi (upper graphs) and ASi (lower graphs) mass balance of all campaigns. Balance is represented as percent, (out(g) in(g))/in(g)) (left graphs); and in absolute numbers (kg
delivered/retained) (right graphs). Error bars represent deviation from conservative mass balance, represented as percent in both directions.
S. Jacobs et al. / Estuarine, Coastal and Shelf Science 80 (2008) 42–5246
