Introduction
Sea-level rise driven by global climate change is expected to
continue for centuries and will in the near future impact about 70 % of
the global coastlines (Church et al., 2013). Rising sea level causes higher
and more frequent storm surges and leads to more incidences of floodwaters
overtopping and breaking coastal defenses (FitzGerald et al., 2008).
Reclaimed coastal areas with low elevation are especially vulnerable to
flooding. A low-cost strategy of coastal protection is “managed coastal
realignment”, whereby old coastal defenses are deliberately breached, and
new ones are constructed further inland (Cooper, 2003; French, 2008; Roman
and Burdick, 2012). The flooded areas created by managed coastal realignment
act as buffer zones, protecting populated areas or valuable assets against
flooding (Gedan et al., 2011). There are an increasing number of projects
in which coastal soils are flooded with seawater by managed costal realignment
and similar techniques (Herbert et al., 2015; Pethick, 2002; Wolters et
al., 2005).
Many studies have been performed on freshwater wetlands experiencing
salinization from seawater intrusion and less on dyked and drained
agricultural soil systems exposed to flooding (Ardon et al., 2016, 2013;
Portnoy, 1999; Portnoy and Giblin, 1997). Existing studies show that flooding
with seawater has dramatic consequences for soil biogeochemistry. Depending
on soil porosity and moisture content, soil environments can have deep oxygen
penetration (75–100 cm) (Dziejowski et al., 1997; MacDonald et
al., 1993; Neira et al., 2015), since oxygen (O2) can rapidly be
supplied from the overlying atmosphere via diffusion. Therefore, surface
soils are predominantly oxic environments, in which soil organic matter is
degraded by a wide variety of microorganisms, fungi and fauna (Boer et
al., 2005; Kalbitz et al., 2000). Aerobic degradation is catalysed by
hydrolytic enzymes and reactive oxygen radicals that can break bonds in
refractory organic compounds, such as lignin and cellulose, and facilitate
complete degradation of soil organic carbon (SOC) to CO2 (Canfield,
1994). However, when soils are flooded, O2 penetration is
dramatically reduced, since O2 solubility in water is low and
O2 diffusion in water is 104 times slower than in air (Neira et
al., 2015). O2 will therefore be depleted by microbial and abiotic
O2 consuming processes in soils flooded with seawater, and become
anoxic except for the upper few millimetres. In aquatic anoxic soils and
sediments, mutualistic consortia of microorganisms degrade organic
macromolecules into smaller moieties by the excretion of exoenzymes and
extracellular hydrolysis, which are then fermented into smaller organic
molecules, mainly acetate (Valdemarsen and Kristensen, 2010). The
fermentation products are taken up by other microorganisms and oxidized to
carbon dioxide (CO2) by the reduction of alternative electron
acceptors (e.g. nitrate, Mn oxides, Fe oxides and sulfate) (Arnosti, 2011;
Glud, 2008). Sulfate is abundant in seawater, and microbial sulfate reduction
(SR) is therefore expected to become a major mineralization pathway in soils
flooded with seawater (Sutton-Grier et al., 2011; Weston et al., 2011).
While some studies have looked at SOC mineralization pathways in different
types of soils introduced to saltwater (Ardon et al., 2016; Chambers et
al., 2013; Neubauer et al., 2013; Weston et al., 2006, 2011), a lot is still
unknown about how the dynamics between initial SOC degradation to DOC and the
terminal mineralization are affected by the introduction of saltwater
(Herbert et al., 2015). Many soils subject to managed coastal realignment
contain considerable amounts of SOC (Franzluebbers, 2010; Wolters et
al., 2005). The degradation of SOC after flooding will depend on the rate of
establishment of heterotrophic microbial communities and their ability to
degrade SOC (Schmidt et al., 2011). Labile organic carbon may be easily
degraded by marine microorganisms, while more complex organic carbon, and
especially structurally complex organic compounds, such as cellulose and
lignin, may be virtually non-degradable in anoxic environments (Kim and
Singh, 2000; Kristensen and Holmer, 2001). Flooding of coastal soils by sea
level rise and coastal realignment may therefore cause significant
preservation of the SOC contained in the soils at the time of flooding.
In this study the fate of SOC after flooding with seawater was investigated
in soils collected at Gyldensteen Strand on the northern coast of Fyn,
Denmark, an area that was designated to be flooded in a coastal realignment
project. We were especially interested in following the temporal
establishment of dominating microbial pathways and quantifying the rates and
temporal trajectories of SOC degradation in newly flooded soils. We
hypothesized that (1) total SOC degradation activity in soils after flooding
depends on SOC content and lability, and that (2) a large proportion of SOC
will be non-degradable due to the anoxic soil conditions forming after the
flooding. To test these hypotheses we performed parallel mesocosm experiments with two different
types of soils that were experimentally flooded with seawater. SOC
degradation and other biogeochemical developments in the mesocosms were
traced with high temporal and spatial resolution for the next 12 months. The
results showed how flooding with seawater impacts C degradation and soil
biogeochemistry and formed the basis for an initial evaluation of potential
feedbacks of flooding on atmospheric CO2 concentrations.
Materials and methods
Study site
This study was conducted in relation to the nature restoration project at
Gyldensteen Strand funded by the Danish Aage V. Jensen Nature Foundation. The
sampling site (55∘34′26.4′′ N 10∘08′17.0′′ E) was a
shallow intertidal habitat until 1871 (size of ∼ 600 ha), when
it was dyked and continuously drained to create new land for agriculture. The
reclaimed area was for the following 140 years mainly used for production of
different crops such as onions and grains (Stenak, 2005). As a part of the
nature restoration project, selected sections of the dykes were removed in
March 2014 and 211 ha of the area were permanently flooded with
seawater and turned into a shallow and mostly subtidal marine lagoon.
Experimental design and sampling
Sampling for the mesocosm experiment was performed in November 2013, half a
year before the flooding of the site, at two different stations representing
uncultivated (UC) and cultivated (C) soils (Fig. 1). Station UC was located
in an area with low elevation, which never could be properly drained. Station
UC was therefore abandoned for agriculture and became a reed swamp that
accumulated plant material and litter. Station C, however, resembled the
majority of the re-flooded area that was farmed since the land reclamation
(fertilized, ploughed and used for monoculture; also illustrated in Fig. 1).
From each station, 15 soil cores were sampled in 30 cm long, 8 cm internal
diameter stainless steel core liners. The core liners were hammered 25 cm
down into the soil, dug up with a spade and closed at both ends with rubber
stoppers.
Map of Gyldensteen Strand with the location of the two sampling
stations for collecting uncultivated (UC) and cultivated (C) soil cores. The
dashed red line indicates the area flooded with seawater in March 2014.
In the laboratory, the headspaces of individual soil cores were gently
flooded with 22–26 salinity seawater collected from the shoreface directly
north of station UC (Fig. 1). Soil cores were then transferred to 70 L
incubation tanks filled with seawater. During the whole experiment the
flooded cores were maintained at 15 ∘C and kept in darkness. The
water in the tanks was rigorously aerated through air diffuser stones and
10–20 L of the seawater in the tanks was exchanged with fresh
seawater (also collected from the shoreface) every 14 days. Thus soil cores
were incubated under constant environmental conditions, while factors such as
diurnal temperature variations, tidal exchange, benthic primary production
and bioturbation were omitted by the experimental set-up.
The flooded soil cores were incubated for 12 months. Flux experiments were
conducted with three random soil cores from each station at various times
(weekly in the first month, biweekly for the next 3 months and monthly
thereafter). Core sectionings were performed on three randomly selected soil
cores from each station at different times during the experiment (before the
flooding, 1 week after and after 2, 4, 6 and 12 months).
Flux experiments
Fluxes of O2, dissolved organic carbon (DOC) and TCO2
(= CO32- + HCO3- + H2CO3) between
soil and overlying water were measured regularly, as described above. Cores
were equipped with stirring magnets, closed with rubber stoppers and placed
around a central magnet rotating at 60 rpm and thereafter incubated
for about 4 h in darkness. O2 was measured and water samples
were taken in the headspace of the soil cores at the beginning and end of
incubations. O2 was measured with an optical dissolved oxygen meter
(YSI ProODO). DOC samples were stored at -20 ∘C until analysis
using a Shimadzu TOC-5000 total organic carbon analyser. Samples for TCO2
analysis were kept in 3 mL gas-tight Exetainers for a maximum of 1 week
until analysis by flow injection (Hall and Aller, 1992).
Core sectioning
Core sectioning was performed by slicing each soil core into 6 depth
intervals (0–1, 1–2, 3–5, 5–10, 10–15 and 15–20 cm). Porewater
was extracted from each depth interval by centrifugation and GF/C filtration
in double centrifuge tubes (500 g, 10 min). The porewater was
sampled for various parameters: 500 µL porewater was preserved
with 30 µL saturated HgCl2 for TCO2,
250 µL porewater were preserved with 50 µL 1 M zinc
acetate (ZnAc) for total dissolved sulfide
(TH2S = H2S + HS- + S2-)
analysis, 250 µL porewater were preserved with 100 µL
0.5 M HCl for Fe2+ analysis and the remaining porewater was stored at
-20 ∘C until analysis for sulfate (SO42-) and DOC.
TCO2 and DOC samples were stored and analysed as described above.
TH2S samples were analysed using the method of Cline (1969).
Fe2+ samples were analysed using the ferrozine method (Stookey, 1970).
SO42- was analysed using liquid ion chromatography on a Dionex
ICS-2000 system.
Reactive iron, RFe, was extracted from soil subsamples from every depth
interval with 0.5 M HCl for 30 min while being
shaken (Lovley and Phillips, 1987). After centrifugation (500 g,
10 min) the supernatant was transferred to sampling vials and stored
at room temperature until analysis for reactive Fe(II) and Fe(III) – RFe(II)
and RFe(III), respectively. The supernatant was analysed for Fe2+
and RFe using the ferrozine method (Stookey, 1970) before and after
reduction with hydroxylamine (Lovley and Phillips, 1987). RFe(II) was
calculated directly, while RFe(III) was calculated from the difference
between RFe and RFe(II). An estimate of total Fe content was obtained by
boiling combusted soil subsamples in 1 M HCl for 1 h at
120 ∘C. The supernatant was stored at room temperature until
analysis by the ferrozine method.
Acid volatile sulfides (AVS) (Rickard and Morse, 2005) and chromium reducible
sulfides (CRS) were determined on soil subsamples preserved with 1 M
ZnAc and stored at -20 ∘C until analysis. AVS and CRS were
extracted by 2-step distillation as described in Fossing and Jørgensen
(1998). Sulfide concentrations in the distillates were analysed using the
method described by Cline (1969).
Soil characteristics were also determined for every depth interval during
every core sectioning. Soil density was determined gravimetrically and soil
subsamples were dried (24 h, 105 ∘C) to determine water
content and porosity. Soil organic matter content was measured as the weight
loss of dry sediment after combustion (520 ∘C, 5 h). SOC on
selected soil samples (samples obtained after 1 week and 6 months) was also
measured by elemental analysis with a Carlo Erba CHN EA-1108 elemental analyzer
according to Kristensen and Andersen (1987).
Anoxic incubations (jar experiments)
Depth distributions of microbial TCO2 and DOC production and SR were
estimated from anoxic soil incubations (Kristensen and Hansen, 1995; Quintana
et al., 2013). The excess soil from core sectionings was pooled into four depth
intervals (0–2, 2–5, 5–10 and 15–20 cm), thoroughly homogenized
and tightly packed into 6–8 glass scintillation vials (20 mL). The
vials were closed with screw caps, buried head-down in anoxic mud and
incubated at 15 ∘C in darkness. Two jars from each jar series were
used every week for porewater extraction in the following 4 weeks. The
screw caps were changed to a perforated lid containing a GF/C filter and the
jars were centrifuged upside-down in a centrifuge tube (10 min at
500 g). The extracted porewater was sampled and analysed for
TCO2, DOC and SO42- as described above.
Data analysis
Fluxes of TCO2, DOC and O2 were calculated from the
concentration differences between start and end samples. Microbial rates in
jar experiments (DOC and TCO2 production and SR) were calculated for
0–2, 2–5, 5–10, 15–20 cm depth intervals by fitting the time-dependent
concentration changes by linear regressions after removing obvious outliers
(visual check). When the slopes were significant (p< 0.05), the volume
specific reaction rates (nmolcm-3d-1) in individual depth
layers were calculated from the regression slopes corrected for sediment
porosity. Microbial reaction rates, porewater and solid pools were depth
integrated over 0–20 cm and converted to area-specific units. Linear
data interpolation was used to correct for missing data points, e.g. for the
depth interval 10–15 cm at which microbial rates were not measured.
There was a significant linear correlation between organic matter content and
SOC for both sampling stations
[OC(%) =0.442⋅ LOI(%) + 0.178, r2=0.987, n=36].
This correlation was used to convert organic matter into SOC for the time
points at which SOC was not directly measured. A one-way ANOVA was performed on
area-specific SOC pools at the different time points to test for significant
changes in the SOC pools over time. Depth-integrated SR rates were normalized
to C units since an almost 2:1 relationship between TCO2 production
and SR (Jørgensen, 2006) was observed throughout the experiment. Errors
for soil characteristics, fluxes, porewater and solid pools were calculated
as standard errors of the mean (SEM). Errors for depth-integrated values of
microbial rates and solid pools were calculated as standard errors
propagation (SEP) of standard deviation (SD) values following ±Eq. (1):
SEP=SD0-1cm2+…+SD15--20cm2.
In a carbon budget estimating SOC degradation during the experiment, total
degradation of SOC (molm2) was calculated as the sum of the time
integrated TCO2 efflux, time-integrated DOC efflux and area-specific
TCO2 and DOC in porewater by the end of the experiment. The
percentage of the initial SOC pool degraded during the experiment was
calculated from the estimated total degradation of SOC and mean bulk SOC
pool. In a time-specific carbon degradation budget, total degradation to
TCO2 was calculated as the sum of time-integrated TCO2 efflux
and accumulated porewater TCO2 at different time points after
flooding (1 week and 2, 4, 6 and 12 months). Based on the jar experiments,
total anaerobic TCO2 production and TCO2 production by SR
(according to a 2:1 relationship between TCO2 production and SR) was
calculated by time integration at different time points after flooding
(1 week and 2, 4, 6 and 12 months). Relative contributions of SR to anaerobic
degradation to TCO2 were estimated from TCO2 production and
TCO2 production by SR measured in jar experiments.
Mean values of water content, porosity and soil organic carbon (SOC)
for all core sectionings. Error indicated as SEM (n=15).
Depth
Water
Porosity
SOC
(cm)
content
(%)
(%)
Station UC
0.5
82.9±0.7
0.82±0.04
16.2±0.8
1.5
75.5±1.6
0.97±0.02
16.1±1.2
3.5
60.5±1.8
0.79±0.01
11.0±0.8
7.5
39.3±0.9
0.60±0.01
5.2±0.2
12.5
33.0±0.7
0.54±0.01
3.5±0.2
17.5
34.5±0.8
0.56±0.01
3.5±0.2
Station C
0.5
32.0±0.6
0.58±0.02
1.4±0.0
1.5
24.8±0.5
0.53±0.01
1.1±0.0
3.5
21.6±0.3
0.40±0.01
1.0±0.0
7.5
18.9±0.4
0.35±0.01
0.8±0.1
12.5
17.9±0.3
0.34±0.00
0.9±0.0
17.5
19.8±0.4
0.37±0.01
1.0±0.0
Results
Soil characteristics
Soil at the two sampled stations had a very different appearance, as a result of
different use after land reclamation (i.e. no cultivation and
cultivation). Station UC was overgrown with mosses and grasses, and a dense
layer of roots and litter characterized the upper 5 cm of the soil,
while the deeper parts of the soil (> 10 cm depth) consisted of clay. At
station C only relatively small amounts of grass and root material were
evident in the upper 5 cm. Some of the vegetation was still alive
2 months after the flooding, as indicated by long green grass leaves seeking
light, but it slowly died out thereafter. The soil at both stations contained
partially degraded shell material from gastropods and bivalves remaining from
before 1871, when the area was a marine lagoon.
There was very little variation in the soil characteristics between successive
core sectionings, so results were averaged for the whole experiment
(Table 1). The water content at station UC decreased with depth from 83 %
at the top to 35 % at the bottom, while water content only decreased from
32 to 20 % at station C. The same depth trend was observed for porosity.
The high water content and porosity at station UC was caused by high amounts
of plant material (e.g. roots), while the soil at station C was sandy,
homogenous and poor in organic debris.
Soil organic content varied greatly with depth at station UC, and the topsoil
was enriched with SOC (16 %) in contrast to the bottom (1 %) (Table 1).
SOC varied between 0.8 and 1.4 % at station C but there was no variation in depth. A
one-way ANOVA showed no significant difference between the SOC contents at
the different time points at either station UC or C (df = 17, F=1.9,
p=1.16 for both stations).
CO2 and DOC efflux, and O2 consumption
TCO2 effluxes in UC soil were highest at the beginning of the
experiment with a maximum of 239±30 mmolm-2d-1
measured on day 13 (Fig. 2a). Subsequently it decreased to about
130 mmolm-2d-1 31–199 days after flooding and stabilized
around 67 mmolm-2d-1 from day 220 to the end. The
TCO2 effluxes in C soil were relatively constant around an average of
29 mmolm-2d-1.
High DOC efflux was evident 1 day after flooding at station UC (108±3 mmolm-2d-1) (Fig. 2b), but it decreased to around
60 mmolm-2d-1 6–20 days after flooding and to
17 mmolm-2d-1 after approximately 2 months before the end. DOC
effluxes at station C showed a similar pattern, averaging
25 mmolm-2d-1 in the first 2 months after flooding and
decreasing to an average of 5 mmolm-2d-1 for the remaining
time of the experiment.
O2 consumption decreased almost linearly during the 1-year experiment
on both stations (Fig. 2c). At station UC initial O2 consumption was
57±3 mmolm-2d-1, 1–45 days after flooding, and then
it steadily decreased to 19±3 mmolm-2d-1 by the end.
At station C there was a less pronounced temporally decreasing trend.
O2 consumption was highest initially with about
26 mmolm-2d-1 at day 1–13 and then decreased to 9±0.6 mmolm-2d-1 by the end.
Porewater chemistry
Porewater DOC was high 1 week after flooding at both stations (on average
10.4 and 3.8 mM at stations UC and C, respectively; Fig. 3a). Over
the experiment porewater DOC decreased slightly in UC soil, while it
increased slightly in C soil.
Fluxes of total carbon dioxide (TCO2, a), dissolved
organic carbon (DOC, b) and oxygen (O2)
consumption (c) in soil cores with uncultivated (UC) and cultivated
(C) soil after flooding. Error bars indicate SEM (n=3).
Porewater TCO2 concentrations in UC soil were in the range of
5–13 mM between 1 week and 2 months after flooding, and profiles
showed a slightly increasing pattern with depth (Fig. 3b). Afterwards an
unexpected drop in TCO2 concentrations, especially in the deep soil
(> 2 cm depth), was observed. This was likely an experimental artefact;
however, it was caused by extremely high Fe2+ concentrations
> 2 mM in the porewater. During sample storage the Fe2+
was oxidized to Fe-oxyhydroxides and formed an orange-brown precipitate at
the bottom of the sample containers, probably leading to sample acidification
and TCO2 degassing (Moses et al., 1987; Hedin, 2006). Porewater
TCO2 concentrations in UC soil after 4 months were affected by this
artefact. In C soil, porewater Fe2+ did not accumulate at the same
rate as in UC soil and only exceeded 2 mM in the 10–20 cm depth
layer after 6 months. Here porewater TCO2 accumulated gradually over
time as expected (Fig. 3b). Rapid TCO2 accumulation occurred in the
first 2 months, when TCO2 increased from 3–5 to 11 mM below
3 cm depth. After 2 months, TCO2 increased further in the 2–10 cm
depth interval, while a decrease occurred below 10 cm depth, which was
probably related to Fe2+ exceeding 2 mM.
Porewater profiles for dissolved organic carbon (DOC, a),
total carbon dioxide (TCO2, b), sulfate
(SO42-) (c) and Fe2+ (d) in
uncultivated (UC) and cultivated (C) soil flooded with seawater. Error bars
indicate SEM (n=3).
Temporal and spatial variability in production of dissolved organic
carbon (DOC, a) and carbon dioxide (TCO2, b) and
sulfate reduction (SR, c) measured in jar experiments with
uncultivated (UC) and cultivated (C) soils flooded with seawater. Note the
different x axis scaling for UC and C measurements. Error bars indicate
SEM.
High concentrations of SO42- were introduced to the soil when it
was flooded with seawater. Yet the initial water infiltration and diffusion was
the only transport mechanism for dissolved SO42- in the mesocosm
set-up, and the experimental period was evidently not sufficiently long to
achieve equilibrium in SO42- in porewater concentrations down to
20 cm depth. As a result, porewater SO42- decreased steeply with
depth at both stations (Fig. 3c). By the end of the experiment in UC soil,
SO42- decreased from ∼ 17 mM at the surface to zero
below 10 cm depth. In C soil SO42- decreased linearly from
∼ 17 mM at the surface to 0–2 mM at the bottom.
After 7 days of flooding the Fe2+ depth distribution in porewater
was constant with depth, with on average 0.02 and 0.2 mM at stations
UC and C, respectively (Fig. 3d). Afterwards a progressive increase in
porewater Fe2+ was observed at both stations. At station UC
Fe2+ increased to up to 1.3±0.6 mM at 5–15 cm depth
after 2 months and stabilized after 6 months, when Fe2+ exceeded
4 mM below 5 cm depth. The same trend was observed at station C,
where Fe2+ accumulated to up to 3.7 mM at 15–20 cm depth
after 12 months.
Anaerobic net DOC production in jar experiments
Net DOC production after 1 week of flooding was high at the surface
0–2 cm at station UC (2666±695 nmolcm-3d-1;
Fig. 4a) and decreased exponentially with depth to 203±23 nmolcm-3d-1 at 15–20 cm depth. A gradually decreasing
net DOC production was observed in all depth layers over the experiment, and
by the end significant net DOC production
(121–172 nmolcm-3d-1) was only detected in the upper
0–5 cm. A similar pattern in net DOC production was observed at
station C, although rates were much lower than at station UC. After 1 week of
flooding, net DOC production at station C was 1155±158 nmolcm-3d-1 in the upper 0–2 cm of the soil
but only 66–83 nmolcm-3d-1 below. After 4 months it had
decreased to 135 nmolcm-3d-1 in the top 0–2 cm and
no net DOC production was detected below 5 cm depth. Very low rates
(21–25 nmolcm-3d-1) were detected in the top
0–5 cm by the end.
Depth-integrated net DOC production at station UC was initially
118–133 mmolm-2d-1 in the first 2 months after flooding
and then gradually declined to 8 mmolm-2d-1 after 12 months
(Fig. 5). Initial depth-integrated net DOC production at station C was 4-fold
lower than at station UC. Net DOC production in C soil decreased by 75 %
in the first 2 months after flooding and almost no net DOC production
occurred after 6 months.
Results from jar experiments showing area-specific net production of
dissolved organic carbon (DOC) and total carbon dioxide (TCO2), and
sulfate reduction (SR, based on SR rate measurements converted to C units) in
uncultivated (UC) and cultivated (C) soil at different times after flooding
at 1 week (1W) and 2, 4, 6 and 12 months (2M, 4M, 6M and 12M), respectively. In
columns marked with *, TCO2 production was corrected with
2⋅SR. Error bars indicate SEP (n=4).
Anaerobic TCO2 production in jar experiments
Initial depth trends in TCO2 production were generally similar to
those observed for DOC, but temporal trends were markedly different
(Fig. 4b). At station UC, TCO2 production was initially almost
1000 nmolcm-3d-1 in the top 0–2 cm and decreased
to 380 nmolcm-3d-1 at 15–20 cm depth. After 2 months,
TCO2 production had increased in the surface 0–2 cm to
6250 nmolcm-3d-1, while rates below 10 cm depth remained
relatively low. After 4 months, TCO2 production decreased to about
2500 nmolcm-3d-1 in the top 0–2 cm, while it was
not possible to determine TCO2 production rates directly for soil
deeper than 5 cm due to the problem with extremely high porewater
Fe2+ described above. As seen below, porewater SO42-
concentrations were not affected by the high porewater Fe2+
concentrations. For the affected data points TCO2 production was
calculated as a rate of SR × 2, assuming that SR was the dominating
CO2 producing process in the anoxic soil (Jørgensen, 2006). The
calculations showed that TCO2 production had decreased further after
6 and 12 months in the top 5 cm
(600–1000 nmolcm-3d-1) and was quite stable below
(0–85 nmolcm-3d-1). TCO2 production rates were
generally much lower in C soil, while relative trends for TCO2
production and their development over time were quite similar between
stations. Maximum TCO2 production rates occurred at 0–2 cm depth,
where TCO2 production varied from 400 to
780 nmolcm-3d-1 between 1 week and 2 months and then
gradually decreased to 110 nmolcm-3d-1 by the end. Similar
trends were observed in the deeper soil, where TCO2 production
decreased from 180 to 310 nmolcm-3d-1 after 7 days to
7–53 nmolcm-3d-1 after 12 months.
Area-specific TCO2 production at station UC was initially
115–200 mmolm-2d-1 in the first 2 months, and decreased to
40 mmolm-2d-1 after 6 months (Fig. 5). At station C area
specific TCO2 production was relatively stable around
44 mmolm-2d-1 for the first 4 months and decreased to 21
and 10 mmolm-2d-1 after 6 and 12 months, respectively.
Upper panels (a, b) show concentration of reactive Fe(II)
and Fe(III) in uncultivated (UC) and cultivated (C) soils before flooding
(BFF) and 12 months after flooding. Lower panels (e, f) show the
relative contributions of reactive Fe(II) and Fe(III) in the upper 20 cm at
various times after flooding at 1 week (1W) and 2, 4, 6 and 12 months (2M, 4M,
6M and 12M), respectively. Error bars indicate SEM (n=3).
Upper panels (a, b) show concentration of chromium
reducible sulfides (CRS) and acid volatile sulfides (AVS) in uncultivated
(UC) and cultivated (C) soils before flooding (BFF) and 12 months after
flooding. Lower panels (e, f) show the depth-integrated pools of AVS
and CRS in the upper 20 cm at various times after flooding at 1 week
(1W) and 2, 4, 6 and 12 months (2M, 4M, 6M and 12M), respectively. Error
bars indicate SEM (n=3).
SR in jar experiments
Significant SR was measured in the top 0–5 cm
(470 mmolm-2d-1) in UC soil 1 week after flooding, while no
SR was detected below (Fig. 4c). After 2 months, high SR was only measured in
the top 0–2 cm (3128±190 mmolm-2d-1). After
4 months SR was still highest in the topsoil (1217±147 mmolm-2d-1), while significant SR was detected down to
10 cm depth. From 4 months to the end of the experiment, SR gradually decreased at all depths
to 338±147 and 43±6 mmolm-2d-1 at 0–2 and
5–10 cm depth, respectively. Since SO42- did not reach the bottom
(15–20 cm) during the experiment at station UC, no SR occurred here.
In C soil SR occurred at considerably lower rates than in UC soil. After
1 week SR was 177±25 mmolm-2d-1 at 0–2 cm depth and
decreased exponentially with depth to zero at 15–20 cm depth. By months 2
and 4, SR occurred at all depths (20–159 mmolm-2d-1).
Afterwards SR decreased in the upper 15 cm while no SR was detected in the
15–20 cm depth interval.
Depth-integrated SR at station UC increased from 24 to
63 mmolm-2d-1 between week 1 and month 2, corresponding to
48 and 126 mmolm-2d-1 carbon mineralization, respectively
(Fig. 5). SR had decreased to 27.7 mmolm-2d-1 after
12 months. SR increased during the first 4 months in C soil
(6–12 mmolm-2d-1) and then decreased to
4 mmolm-2d-1 after 12 months.
Solid pools of Fe and S
Before flooding, RFe(II) in UC soil increased with depth from
4 µmolcm-3 at 0–1 cm depth to 13 µmolcm-3
at 15–20 cm depth, while a corresponding increase in RFe(III) occurred from
19 to 44 µmolcm-3 (Fig. 6). The RFe pools at station C were
relatively constant with depth, on average 2.5 and
23 µmolcm-3 for RFe(II) and RFe(III), respectively. Twelve
months after flooding, RFe(II) in UC soil had increased to
34–59 µmolcm-3, while RFe(III) had accumulated to 134.5±85 µmolcm-3 in the top and decreased to an average of
4 µmolcm-3 below. A similar trend was obtained in C soil
with RFe(III) accumulating to 51.9±1.4 µmolcm-3 on the
surface. In UC and C soil, total RFe initially consisted of 78 and 92 %
Fe(III), respectively, while it was reduced to 19 and 10 % by the end.
Clearly, RFe(III) became reduced to RFe(II) during the experiment due to the
anoxic conditions created by flooding.
The RFe content was quite heterogeneous at the study sites and there were
large variations between soil cores. Based on all the depth profiles obtained
over the experiment, average total Fe content in UC and C soil was 19.3±2.8 and 26.7±1.8 molm-2, respectively.
Although the jar experiments suggested high SR in both soil types, dissolved
sulfide (TH2S) was never detected in the porewater. Instead, a large
fraction of the sulfide produced during SR accumulated as AVS and CRS in both
soil types (Fig. 7). One week after flooding, AVS and CRS in UC soil were low
(0.2–2.7 µmolcm-3), except at 2–5 cm depth, where AVS
content was slightly elevated. Twelve months after flooding, AVS and CRS had
increased to 25±10 and 41±11 µmolcm-3 at 2–5 cm
depth, respectively, while no accumulation occurred below 10 cm depth. A
similar pattern was observed in C soil, in which AVS and CRS were initially
constant with depth averaging 0.1 and 3.5 µmolcm-3,
respectively, and accumulated to 6.4±1 and 8.4±0.7 µmolcm-3 after 12 months of flooding, respectively.
Over the whole experiment total sulfide accumulated as AVS and CRS gradually
increased, from 0.5 molm-2 before flooding to
4.7 molm-2 after 12 months in UC soil, and from 0.63 to
2 molm-2 in C soil.
Budgets for SOC degradation
Area-specific SOC pools were 710.9±54 and 232.5±22 molm-2 (n=18) in UC and C soil, respectively (Table 2).
Total SOC degradation estimated as the sum of TCO2 and DOC effluxes,
and porewater accumulation over the 1-year experiment was 49.6 and
14.8 molm-2 at stations UC and C, respectively, corresponding to
7 and 6 % of the SOC pools.
Total SOC mineralization to TCO2 was estimated as the sum of
TCO2 efflux and porewater accumulation during the whole experiment
(Table 3), which was 40.0 and 12.0 molm-2 at stations UC and C
respectively. The importance of anaerobic SOC degradation for total
TCO2 mineralization could be calculated from jar experiments, and a
total of 32.6 and 10.8 molm-2 SOC was converted to TCO2
anaerobically, corresponding to 82 and 90 % of flux-based total
TCO2 production at stations UC and C, respectively. The SR measured in
jar experiments corresponded to 25.3 and 4.3 molm-2 CO2
production at stations UC and C during the experiment. Thus 63 and 36 % of
the flux-based total TCO2 production was driven by SR in UC and C
soil, respectively, starting at 30–40 % after 1 week and gradually
increasing up to 100 % by the end of the experiment. This means that the
remaining 19 and 54 % of the flux-based total TCO2 production was
produced by other anaerobic processes than SR in UC and C soil, respectively
(e.g. nitrate or Fe reduction).
Carbon budget table showing mean soil organic carbon (SOC) ± SEP
(n=18) in uncultivated (UC) and cultivated (C) soil. Total time
integrated efflux and accumulation of total carbon dioxide (TCO2) and
dissolved organic carbon (DOC) in porewater are also shown.
Carbon budget (molm-2)
Station UC
Station C
Initial SOC pool
710.9±54
232.5±22
TCO2 efflux
39.9
11.2
DOC efflux
8.9
2.4
TCO2 porewater accumulation
0.1
0.8
DOC porewater accumulation
0.7
0.5
Total SOC degradation
49.6
14.8
Percentage of SOC pool degraded
7 %
6 %
Budget table showing cumulated time-integrated total degradation to
carbon dioxide (TCO2) in flooded uncultivated (UC) and cultivated (C)
soil, based on TCO2 fluxes and total anaerobic TCO2
production based on jar experiments. Estimated partitioning between aerobic
respiration, sulfate reduction and other anaerobic respiration processes is
also shown. Different times after flooding are indicated by 1W (1 week) and
2M, 4M, 6M and 12M (2, 4, 6 and 12 months, respectively).
Station UC
Station C
1W
2M
4M
6M
12M
1W
2M
4M
6M
12M
Degradation to TCO2 (molm-2)
2.07
10.4
18.8
27.4
40.0
0.5
2.7
5.0
6.6
12.0
Anaerobic degradation to TCO2 (molm-2)
0.8
8.7
19.9
24.2
32.6
0.3
2.5
6.0
8.0
10.8
Aerobic respiration (% of total)
61
16
0
12
18
40
7
0
0
10
Sulfate reduction (% of total)
15
45
65
62
62
20
30
37
39
36
Other anaerobic respiration processes (% of total)
24
39
35
26
20
40
63
63
61
54
Discussion
Temporal trends in SOC degradation
The UC and C soil had very different organic contents. UC soil had not been
used for agriculture and organic matter consisting of dead and alive plant
matter had accumulated in the topsoil (Table 1), while lower organic matter
content was evident in C soil due to lower plant cover and regular mechanical
soil reworking during agricultural cultivation (Benbi et al., 2015; Six et
al., 1998). Consequently, the bulk SOC pool was three times higher in UC soil
than in C soil. The source of soil organic matter at both stations was
terrestrial and wetland plants, such as grasses, reed and herbs rich in
cellulose and lignified tissues (Arndt et al., 2013; Sullivan, 1955). Such
organic matter is refractory towards degradation in anaerobic marine
sediments (Kristensen, 1990, 1994) compared to structurally simple
phytoplankton, microphytobenthos and macroalgae, which are common organic
carbon sources in coastal marine sediments (Dubois et al., 2012; Fry et
al., 1977). It was therefore uncertain as to what extent the SOC at Gyldensteen
Strand could serve as a substrate for developing microbial communities after
the flooding with seawater. Nevertheless, we observed high heterotrophic
activity (e.g. O2 uptake and TCO2 production) right after the
flooding, indicating that at least part of the SOC in both soil types was
readily available for microbial degradation.
Cleavage of particulate organic carbon to DOC by extracellular enzymes is the
primary degradation step in waterlogged anoxic soils and sediments (Arnosti,
2011; Weiss et al., 1991). The produced DOC is thereafter converted into short
chain fatty acids and acetate by microbially mediated fermentation and
hydrolysis, which then are terminally oxidized to CO2, e.g. by SR
(Canfield et al., 2005; Valdemarsen and Kristensen, 2010). DOC production can
therefore generally be considered the rate-limiting step for organic carbon
degradation. However, a small proportion of produced DOC is recalcitrant and
may accumulate in soil porewater over time in an experimental set-up without
advective porewater transport. In this experiment we observed high DOC
concentrations in porewater and the highest DOC production in jar experiments
as soon as 7 days after flooding with seawater (Figs. 3a, 5). Part of this DOC
may have leached to the porewater, e.g. as a result of cell lysis due to
flooding (Kalbitz et al., 2000), while the rest was produced by microbial
degradation of particulate SOC (Kim and Singh, 2000). Microbial degradation
of soil organic matter to DOC was initiated immediately after flooding
irrespective of the shift to anoxic conditions. Differences in DOC production
rates indicated that the availability of degradable SOC was clearly higher
in UC soil compared to C soil following the overall difference in total SOC
content. However, total DOC production ceased rapidly in both soil types and
was close to zero after 1 year. Valdemarsen et al. (2014) similarly observed
gradually decreasing DOC production over 2 years in eight different sediment
types from Odense Fjord, indicating gradual depletion of degradable organic
matter despite high sediment organic content and abundance of energetically
favourable electron acceptors. It therefore appears that only a minor portion
of SOC (6–7 %; Table 2) is available for microbial degradation under the
present conditions (flooded with seawater and anoxic conditions). The low
degradability of SOC after flooding probably reflects limitations of the
anaerobic microbial communities on degradation of complex organic matter of
terrestrial origin (Fors et al., 2008; Yucel et al., 2013).
Heterotrophic DOC oxidizing microbes were also active immediately after
flooding as shown by initial TCO2 effluxes and high TCO2
production in the jar experiments 7 days after flooding (Figs. 2a, 5). Rapid
microbial CO2 production has previously been observed in experiments
with experimentally flooded soils (Chambers et al., 2011; Neubauer et
al., 2013; Weston et al., 2011). In both soil types, TCO2 production
in the surface soil increased over the first 2 months, peaked, and then
decreased gradually towards the end. These temporal dynamics were out of
phase with DOC availability, indicating that microbes oxidizing DOC to
CO2 adapt slower to flooded conditions than fermenting and
hydrolysing microbes. Similar cases of initial DOC production due to leaching
and/or substrate hydrolysis outpacing fermentation and SR has been observed
before (Arnosti et al., 1994), maybe due to lag response in the microbial
community (Bruchert and Arnosti, 2003). Nevertheless, the majority
(∼ 80 %; Table 2) of produced DOC over the whole experiment was
oxidized completely to TCO2, while the rest effluxed to the overlying
water (∼ 19 %) or accumulated in porewater (∼ 1 %).
SOC degradation pathways
SO42- was an important electron acceptor in both soils and SR
accounted for 63 and 36 % of the total TCO2 production during the
experiment in UC and C soil, respectively (Table 3). One week after flooding,
active SR corresponding to 30–40 % of anaerobic TCO2 production
was detected in the jar experiment. The relative importance of SR increased
gradually over the experiment and by the end accounted for up to 100 % of
the anaerobic TCO2 production in both soil types. This is in
accordance with Weston et al. (2006), who measured SR in freshwater marsh soil
exposed to saltwater in anoxic flow through reactors, and found that the
relative importance of SR for total TCO2 production increased from
18 % initially to > 95 % after 4 weeks. The delay in SR probably
reflects a lag phase for the community of SO42- reducing microbes
to respond to elevated SO42- levels. The delay in SR could also
reflect initial competition with other TCO2 producing pathways (e.g.
NO3- and Fe reduction) in the time right after flooding when
NO3- and oxidized Fe might have been abundant. However, as the soil
became reduced due to increased SOC degradation activity and limited
O2 supply, electron acceptors other than SO42- were
rapidly depleted and SR became the dominant respiration pathway.
By combining results from flux and jar experiments, it was possible to confine
the relative importance of different microbial respiration pathways in
flooded soils. The difference between TCO2 effluxes (aerobic and
anaerobic processes) and TCO2 production in jar experiments
(anaerobic processes) suggested that aerobic respiration only played a minor
role in the flooded soils (18 and 10 % in UC and C soil, respectively).
On the other hand, SR was quantitatively a very important pathway,
constituting 63 and 36 % of total C-mineralization to TCO2 in UC
and C soil, respectively. Hence 19 (UC) to 54 % (C) of TCO2
production occurred by respiration processes that are not directly accounted for.
Weston et al. (2006) found that Fe reduction was responsible for about
60 % of CO2 production in the first 4 days after saltwater
intrusion in coastal soils. When considering the high initial concentrations
and the rapid decrease in soil RFe(III) in our experiment (Fig. 6),
respiratory Fe-reduction was probably an important respiration process
initially. However, based on this experiment it was not possible to
distinguish between biological and chemical Fe-reduction.
Fate of SOC
In this study we observed that only 6–7 % of the total SOC pools in
coastal soils were degraded by microbial processes in the first year after
flooding with seawater. The low final SOC degradation rates, and especially
the very low final DOC production in both soil types, suggest that the
majority of SOC present in soils at the time of flooding will be permanently
buried due to the limited ability of anaerobic microbial communities to
degrade complex organic matter of terrestrial origin (Burdige, 2007;
Canfield, 1994; Hedges and Keil, 1995). For comparison Neubauer et al. (2013)
similarly found long-term reduction of degradation rates and lability of SOC
pools in a tidal freshwater marsh experiencing saltwater intrusion, which
also supports preservation of SOC. Hence flooding of coastal soils due to sea-level
rise or intentional flooding by managed realignment may lead to
significant C-preservation. At Gyldensteen Strand SOC burial will be in the
order of 48±6×103 kgSOCha-1 (average ± SEM, n=30) when considering a detailed investigation of the soil characteristics
down to 20 cm depth (T. Valdemarsen, unpublished results). However, this
C-preservation does not constitute a permanent C-sink as it only relates to
the SOC buried in the soils at the time of flooding.
Efficient Fe-driven sulfide buffering in flooded soils
Accumulation of free H2S is often seen in metabolically active
organic enriched marine sediments, where it has toxic effects on benthic
fauna (Hargrave et al., 2008; Valdemarsen et al., 2010). It was therefore a
concern whether free H2S would accumulate in the soils from
Gyldensteen after flooding, since this could hamper the succession of benthic
fauna as well as overall ecological developments. However, despite the
extremely high initial SR rates in the flooded soils, comparable to SR
measured beneath fish farms (Bannister et al., 2014; Holmer et al., 2003), no
accumulation of free H2S occurred in any of the soil types. Dent
(1986), Portnoy and Giblin (1997), and Weston et al. (2011) observed a
similar lack of H2S accumulation in soils introduced to saltwater,
suggesting that newly flooded soils have a high capacity to buffer
H2S. Budget considerations suggest that most of the produced
H2S was immediately re-oxidized, e.g. with O2 in the surface
soils, while a significant proportion (37 and 93 % in UC and C soil,
respectively) precipitated as different Fe–S compounds, for instance
FeS and Fe3S4 in AVS and FeS2 and S0 in CRS
(Reddy and DeLaune, 2008; Rickard and Morse, 2005; Valdemarsen et al., 2010).
The depth profiles of solid Fe and S showed that sulfide precipitation
occurred at the same depths at which active SR was measured, i.e. in the
upper 10 cm in UC soil and down to 20 cm depth in C soil. The
decreasing microbial activity and increasing Fe(II) over time will create a
long-term sulfide buffering capacity in the soil (Schoepfer et al., 2014).