Limnol. Oceanogr., 41(8), 1996, 1733-1748
0 1996, by the American Society of Limnology and Oceanography,
Organic carbon oxidation and suppression of
methane production by microbial Fe(III) oxide reduction in
vegetated and unvegetated freshwater wetland sediments
Eric E. Roden and Robert G. Wetzel
Department
of Biological
Sciences, University
of Alabama, Tuscaloosa,
35487-0206
Abstract
High concentrations (20-75 pmol cm-3) of amorphous Fe(III) oxide were observed in unvegetated surface
and Juncus eflusus rhizosphere sediments of a freshwater wetland in the southeastern United States. Incubation experiments demonstrated that microbial Fe(III) oxide reduction suppressed sulfate reduction and
methanogenesis in surface scdimcnts and mediated 240% of depth-integrated (O-10 cm) unvegetated sediment carbon metabolism, compared to I 10% for sulfate reduction. In situ CO2 and CH, flux measurements
verified that nonmethanogenic pathways accounted for - 50% of unvegetated sediment carbon metabolism.
Lower (- 1O-fold) rates of dark/anaerobic CH, flux from experimental vegetated cores relative to unvegetated
controls suggestedthat methanogenesis was inhibited in the Juncus rhizosphere, in which active Fe(III) oxide
reduction was indicated by the presence of low but readily detectable levels of dissolved and solid-phase
Fe(II). Fe(III) oxide reduction accounted for 65% of total carbon metabolism in rhizosphere sediment incubations, compared to 22% for methanogenesis. In contrast, methanogenesis dominated carbon metabolism
(72% of total) in experimental unvegetatcd sediment cores. The high Fe(III) oxide concentrations and reduction rates observed in unvegetated surface and Juncus rhizosphere sediments were perpetuated by rapid
Fe(III) regeneration via oxidation of Fe(II) compounds coupled to 0, input from the overlying water and
plant roots, respectively. The results indicate that Fe(III) oxide reduction could mediate a considerable
amount of organic carbon oxidation and significantly suppress CH, production in freshwater wetlands situated
within globally extensive iron-rich tropical and subtropical soil regimes.
Natural and agricultural wetlands generate up to 50%
of annual CH4 input to the atmosphere (Cicerone and
Orcmland 19 8 8). Understanding
factors responsible for
regional variations in wetland CH4 emission is important
for refining global atmospheric flux estimates, and hence
assessing the current and projected contribution
of CH4
to atmospheric warming (Bartlett and Harriss 1993). Recent studies indicate that a variety of factors such as wetland plant productivity (Whiting and Chanton 1993), microbial CH4 oxidation (King 1993), water table height
(Freeman et al. 1993; Moore and Dalva 1993), and temperature (Bartlett and Harriss 1993) affect rates of wetland
CH4 production and release. Another important factor is
competition among methanogenic and other anaerobic
respiratory bacteria for organic substrates in wetland sediments (Kiene 199 1). In sulfate-rich marine and brackish
environments,
sulfate-reducing bacteria effectively outcompete methanogens (Capone and Kiene 1988), and rates
of wetland CH4 production and flux are uniformly low
in such environments (Bartlett and Harriss 1993). In contrast, methanogenesis is considered to be the dominant
anaerobic carbon oxidation process in sulfate-poor, or-
Acknowledgments
We thank Trisha May for technical assistance, and R. P. Kienc
and D. R. Lovley for review of an earlier version of the manuscript.
This work was supported by grants DEB 92-20822 and DEB9407233 from the U.S. National Science Foundation.
This paper is a contribution from the Center for Freshwater
Studies, University of Alabama.
1733
ganic matter-rich freshwater sediments (Capone and Kiene
1988). Contrary to this general paradigm, however, several recent studies have shown that unidentified
nonmethanogenic organic carbon oxidation pathways equal
or exceed methanogenesis in a variety of freshwater wetland sediments (e.g. Kelly et al. 1990; Happell and Chanton 1993; Pulliam 1993). These findings raise the question
of what nonmethanogenic
pathways may be significant
in freshwater wetland sediment carbon metabolism.
It is well known that neither sulfate reduction nor
methanogenesis predominate in anaerobic sediment metabolism until microbially
reducible Fe(III) oxides are
depleted, because Fe(III)-reducing
bacteria are able to
outcompete both sulfate-reducing and methanogenic bacteria for organic substrates (Lovley 199 1). This mechanism is analogous to that documented for the competition
between sulfate-reducing and mcthanogenic bacteria. Although Fe(III) oxides are often abundant in freshwater
sediments and submerged soils, and microbial Fe(III) reduction has been studied in a variety of such environments (Lovley 199 l), when compared to sulfate reduction
and methanogenesis, the contribution of Fe(III) reduction
to in situ anaerobic sediment metabolism has not been
well documented (Westermann 1993). One difficulty in
this regard is that use of a radiotracer assay for Fe(III)
oxide reduction is not possible because rapid isotopic
exchange occurs among sedimentary Fe(III) and Fe(II)
pools (Roden and Lovley 1993).
Several lake sediment studies (see Lovley 199 1) concluded that Fe(III) reduction was of minor importance
for organic carbon decomposition. However, these studies considered only the formation of dissolved Fe(II),
1734
Roden and Wetzel
which greatly underestimates
the extent of microbial
Fe(III) reduction because most of the Fe(II) produced is
sequestered into solid-phase compounds (Lovley 199 1).
Studies in freshwater Potomac River sediments indicated
that oxidation of organic carbon coupled to the formation
of solid-phase Fe(II) in the top 1 cm of sediment could
mediate as much carbon flow as methane production in
the upper 4 cm of sediment (Lovley and Phillips 1986).
Modeling of solid-phase Fe and Mn diagenesis in the
surface sediments of Toolik Lake, Alaska, suggested that
Fe(III) and Mn(IV) reduction could account for -50% of
anaerobic carbon metabolism (Cornwell 1983). These
studies indicated that although the zone of Fe(III) oxide
reduction may be restricted to the upper few centimeters
of freshwater sediments, Fe(III) reduction may be important in carbon oxidation because this zone typically
contains the most labile organic material and supports
the highest rates of organic decomposition (Lovley 199 1).
In view of current interest in the contribution
of atmospheric CH4 to global warming and the need for quantitative models of controls on wetland CH, emission, the
role of Fe(III) oxide reduction in freshwater wetland
sediment metabolism and its potential to inhibit CH4
production and emission needs quantitative
evaluation
(Lovley 1991). In this study, we quantified Fe(III) oxide
reduction and CH4 production, along with relevant dissolved and solid-phase chemical parameters, in unvegetated surface and aquatic plant rhizosphere sediments in
a small (15 ha) riverine wetland located in the Talladega
National Forest in north-central Alabama. The soils surrounding the wetland are Fe-rich ultisols, and therefore
this site provides a model ecosystem representative of
wetlands situated in Fe-rich soil regimes, which are common to much of the southeastern U.S. as well as many
tropical regions in the southern hemisphere (Bouwman
1989). Because of their high Fe abundance, we hypothesized that these sediments would support high Fe(III)
reduction rates, and that this process would significantly
suppress sediment CH4 production and emission.
Materials
and methods
Study site-The Talladega wetland (TW) is located in
the southeastern U.S. Coastal Plain physiographic province (32”55’N, 87’26’W). The wetland was formed -50
yr ago by beaver activity that led to flooding of - 1.5 ha
of forest floor. Water temperatures range from -5OC in
February to 25-30°C throughout June-September
(M.
Ward unpubl. data). TW wetland waters are soft (alkalinity = 0.05-0.1 meq liter-l,
conductivity
= 8-50 PS
cm-l) and weakly acidic (pH 5.5-6.8) (Stanley and Ward
1996). The dominant emergent macrophyte in the TW is
the freshwater rush Juncus eflusus.
About 10 cm of unconsolidated
sediment has accumulated over the original forest soil since impoundment.
The dry-weight organic matter and Fe content of this
material are m 20 and 5%, respectively. Dissolved O2 penetration (measured with a Clark-style microelectrode; Diamond General Corp.) is limited to 5 1 mm in unvege-
tated sediments, and estimates of O2 consumption coupled to ferrous iron and CH4 oxidation in surface sediments indicate that very little of TW sediment O2 uptake
is linked to aerobic carbon respiration (Roden and Wetzel
in prep.). Furthermore, because overlying water nitrate
concentration:; are very low (< 1 PM; Stanley and Ward
1996) and Mn is only a hundredth as abundant as Fe (J.
Edmonds unpubl. data), nitrate and Mn(IV) oxide reduction are likely insignificant
in sediment carbon metabolism. Thus, Fe(III) reduction, sulfate reduction, and
methanogenesis are the major carbon mineralization
pathways in 1-W sediments. Sulfate concentrations are
low (30-40 PM) in the soft TW waters, which, as discussed
below, strongly limits the role of sulfate reduction in sediment carbon metabolism.
Sediment sampling and analysis- Undisturbed sediment cores were obtained manually from areas between
Juncus tussocks with 7.5- or lo-cm-i.d. Plexiglas tubes
with a sharpened edge at the bottom. On one occasion
(November 1993), cores were also excavated from within
two separate Juncus tussocks by removing the aboveground vegetation and inserting a core tube into the sediment while cutting external roots away from the bottom
of the tube with a knife. A few weeks before collection of
the vegetated I;ores, a pore-water equilibration
sampler
fitted with a 0.2~pm pore-sized nylon membrane was inserted into the middle of a nearby tussock, as well as into
an adjacent unvegetated area of sediment. The equilibrators were retrieved and sampled in the field a few days
before collection of vegetated and unvegetated sediment
cores. Pore-wa,ter samples (3 ml) were obtained from each
depth interval (2.5-cm spacing) with a N,-flushed plastic
syringe and transferred to 25-ml serum bottles that contained 0.5 ml of 1 N HCl. The CH, concentration in 0.1
ml of the headspace of the vials was determined with a
gas chromatog raph (Shimadzu GC- 14A) equipped with
an injection po;-t and flame ionization detector. Dissolved
Fe2+ in 0. l-O.:! ml of the acidified pore water was subsequently determined calorimetrically
with 5-10 ml of
Ferrozine (1 g per liter of 50 mM Hepes buffer, pH 7;
Lovley and Phillips 1986). Parallel 3-ml pore-water samples were fixed with 0.3 ml of 10% Zn acetate and analyzed for sulfa,te concentration by ion chromatography
(Dionex model DX- 100).
Sediment cores were returned to the laboratory and
held at room temperature (22* l°C) overnight, and in
some cases (19 95 unvegetated cores) they were used for
CH4 efllux rate measurements (see below) before being
transferred to an anaerobic chamber (Coy Products) and
sectioned at 0.25-l .O-cm intervals. Whole-sediment CH4
concentrations were determined by autoclaving 2-3-ml
samples from each depth interval in a crimp-sealed 25ml serum bottle, followed by headspace CH4 determination by gas chromatography as described below. Porewater samples (1 ml) for dissolved Fe2+ determination
were obtained by centrifugation
inside the anaerobic
chamber, acidifted with 0.1 ml of 1 N HCl, and analyzed
calorimetrically
. Amorphous Fe(III) oxide and total (sol-
1735
id plus dissolved) Fe(II) concentrations were measured
by extracting 0.1-0.5 ml of sediment with 5-10 ml of 0.5
N HCl, followed by colorimctric (Ferrozine) determination of Fe(II) and total Fe in 0.025-o. l-ml aliquots of the
extracts (Roden and Lovley 1993). Fe(III) was calculated
from the difference between total Fe and Fe(II). When
incubation experiments were conducted with material
from sectioned cores, subsamples for pore-water sulfate
concentration and 35S0 *- reduction measurements (see
below) were obtained frim two or three individual cores,
after which material from three or four replicate cores
was combined and used for measurements of Fe(III) reduction and methanogenesis as described below.
,
Microbial metabolism measurements- A preliminary
experiment was conducted with the upper 10 cm of sediment from an unvegetated core. The sediment was homogenized, passed through a l-mm sieve, and slurried
with anaerobic distilled water (1 : 1 vol/vol) under 02free N2. The slurry was incubated at room temperature
for 2 weeks to deplete Fe(III) oxides. Half of the slurry
was then oxidized by stirring and bubbling with air for
24 h, whereas the other half was maintained under anaerobic conditions. The oxidized sediments were made
anaerobic by bubbling with O,-free N2. Five-milliliter
portions of oxidized or reduced sediment were then transferred under N2 to 20-ml serum bottles. Fe(III), Fe(II),
in the
CH4, and total CO2 (see below) concentrations
headspace or dissolved and solid phases of the bottles
were followed by sacrificing duplicate bottles at l-5-d
intervals over a 3-week incubation period.
Rates of anaerobic microbial metabolism in material
from sectioned sediment cores were determined by closedvessel incubation techniques. Fc(II1) oxide reduction rates
were estimated from depletion of amorphous Fe(III) oxide or production of Fe(II) in 2-3-ml sediment samples
incubated under N2 in 15-ml serum vials over a 1-2-week
period (Lovley and Phillips 1986). Samples were sacrificed at 2-4-d intervals (two or three samples per interval).
Control experiments with heat-killed sediments verified
that Fe(III) reduction was the result of microbial activity.
On one occasion, the relative rate of dissolved vs. total
Fe(II) accumulation was examined by collecting parallel
Fe(III) resamples for pore-water Fe *+ determination.
duction rates were calculated from regressions of Fe(III)
or Fe(II) concentration vs. time (error bars indicate SE
of slope coefficient) and are expressed in C equivalents
according to the 1 : 4 ratio of CO2 production : Fe(III)
reduction observed in the “oxidized”
sediment slurry incubation experiment.
Sulfate reduction rates were measured by adding -37
kBq of 35S0,2- to 2-3 ml of sediment in 15-ml serum
vials and incubating anaerobically for 20-30 min before
terminating activity with 1 ml of 10% Zn acetate. 35Slabeled reduced inorganic sulfur (RIS) was recovered by
a single-step acidic Cr*+ extraction (Fossing and Jorgensen 1989) and trapping of the evolved H2S in 5-10 ml of
10% Zn acetate. [35S]Zn activity was quantified by mixing
2-ml portions of the traps with 10 ml of Aquasol followed
by liquid scintillation
spectrometry (Beckman model LS
5801). Sulfate reduction rates based on [35S]RIS formation were corrected for rates of 35S-labeled carbon-bonded
reduced sulfur formation in separate experiments conducted with material from the top 10 cm of an unvegetated core, according methods described by Wieder and
Lang (1988). In these experiments, formation of 35S-labeled carbon-bonded
reduced sulfur accounted for
lQ.3*2.2% (n = 10) oftotal 35S042- reduction. Reported
rates account for 2 : 1 ratio of CO2 production to SO,*reduction (Westrich and Berner 1984). Pore-water samples (1 ml) for sulfate concentration measurements were
obtained by centrifugation inside the anaerobic chamber,
fixed with 0.1 ml of 10% Zn acetate, and analyzed by ion
chromatography.
Total RIS concentrations in the sediment samples was determined by calorimetric
analysis
of the ZnS content of the Zn-acetate traps.
Rates of methanogenesis were determined from accumulation of CH,, in the headspace of the same samples
used for Fe(III) reduction. CH, concentrations in 0.1 -ml
gas samples were measured with a gas chromatograph
(Shimadzu GC- 14A) equipped with an injection port and
flame ionization detector. Rates of methanogenesis are
expressed in carbon equivalents according to the 1 : 1 ratio
of CO2 : CH4 production observed in the reduced sediment slurry incubation experiment.
Total CO2 production was determined from the combined accumulation of CO2 in the headspace (analyzed
as CH4 by a gas chromatograph with a methanizer) and
dissolved inorganic carbon (DIC) in the port water of the
same sediment samples used for determination of Fe(III)
reduction and CH4 production rates. Port-water samples
(1 ml, obtained by centrifugation) were transferred to loml serum vials, acidified with 0.5 ml of 1 N HCl to drive
DIC out of solution and into the vial headspace, and
headspace CO, concentrations analyzed as CH4 by gas
chromatography.
The temperature dependence of Fe(III) reduction, CH,
production, and sulfate reduction in TW sediments was
assessed by incubating oxidized (prepared as described
above), reduced sulfate-free, and reduced sulfate-amended (0.050 ,umol cmB3) sediments, respectively, under anaerobic conditions and following rates of microbial metabolism as described above at temperatures of 4, 10, 15,
20, 25, and 30°C. Sediment for this cxpcriment was obtained from the upper 5 cm of several unvegetated cores.
Rates were plotted according to the Arrhenius equation,
and apparent activation energies were estimated by linear
regression analysis.
Field gas flux measurements-Rates of unvegetated
sediment CO2 and CH4 efflux were determined in situ
from the increase in gas concentration measured in the
headspace of darkened floating chambers (1.6 liters covering 86.5 cm*) over a 1O-l 5-min period. Gas samples
were collected with IO-ml plastic syringes and held at
room temperature (12 d) before gas Chromatographic
analysis with a l-ml sample loop. No obvious evidence
of bubble ebullition (cf. Chanton and Whiting 1995) was
observed during these measurements, although in some
Roden and Wetzel
1736
Fe(II I) (pmol cmm3)
0
20
40
Fe(III), Fe(II) (pmol cmm3)
60
0
20
40
60
IO
0
50
100 150 200
Fe(II) (pmol cmm3)
Dissolved Fe 2+ (mM)
0
2
4
Dissolved Fe2+ (mM)
6
0
0
0
IO
IO
20
20
30
30
40
40
2
4
6
50
50
Fig. 1. Comparison of solid-phase Fe(III)-Fe(II) and dissolved Fe2+ distributions in unvegetated (A, C) vs. vegetated
(B, D) TW sediments collected in Novcmbcr 1993. Note inset
in panel D shows low but detectable dissolved Fc2+in the Juncus
rhizosphere.
cases scatter in plots of headspace CH, concentration vs.
time (r2 values ~0.9) suggested ebullition (Happell et al.
1995). In all cases the total rate of increase in headspace
CH4 concentration was used to calculate CH4 flux (Happell et al. 1995).
Field measurements of CH4 flux were corrected for CH4
oxidation at the sediment surface by comparing rates of
CH4 flux from intact sediment cores incubated at room
temperature in the laboratory under light-aerobic vs. darkanaerobic conditions (King 1990b). The cores were collected on the same date and from the same location as
the gas flux measurements and subsequently used for the
sectioning-incubation
experiments described above. Ambient laboratory light intensity was -60 pmol m-2 s-l.
Anaerobic conditions were established by flushing the
hcadspace of the cores with N2 for 1 h prior to conducting
CH4 flux measurements.
Experimental cores-A series of experimental vegetated and unvegetated cores (1 O-cm i.d.) were constructed
with reduced, homogenized TW sediment collected with
an Ekman grab sampler and incubated in a glasshouse
under conditions of continuous flooding and natural lighttemperature levels for a period of 120 d (SeptemberDecember 1995) before sampling as described below.
Above- and belowground plant biomass in the vegetated
cores were 334&2 1 (SD, n = 5) and 730f 170 (range, n
= 2) g dry wt rn2, respectively, at the time of sampling.
Rates of CH4 flux from vegetated (n = 5) and unvegetated (n = 4) sediment cores were measured under darkanaerobic conditions to suppress CH4 oxidation. CH4 accumulation was followed in 1O-cm (i.d.) x loo-cm Plexiglas cylinders with built-in mixing fans placed over the
sediment cores. Subsequent to the CH4 flux measurements, rates of Fe(III) reduction, sulfate reduction, methanogenesis, and associated dissolved/solid-phase
chemical distributions were determined in a single unvegetated
core and duplicate vegetated cores. Cores were sectioned
in the anaerobic chamber after removal of aboveground
biomass (if present). Fe(III) reduction and CH4 production were determined as described above by incubating
5-ml portions of sediment in loo-ml serum bottles over
a 6-d incubation period. Triplicate bottles from five depth
intervals in each of two cores were sacrificed at 0, 3, and
6 d. Sulfate reduction rates were determined by injecting
- 37 kBq of 35S042- into - 5 ml of sediment contained
in a de-tipped 5-ml plastic syringe sealed with a butyl
rubber stopper. The samples (duplicates from five depth
intervals in each of two cores) were incubated for 0.5-2
h and activity was stopped by placing the syringes in a
0°C icebath for 5 min before transferring the syringes to
a - 20°C freezer. A parallel time-zero blank was included
for each depth interval to correct for reduction activity
during the termination period, which averaged 25 * 19%
(n = 20) of tctal reduced 35S (RIS and carbon-bonded
reduced S; determined as described above) recovery in
the live samples. Contemporaneously
with these rate
measurements, the rate of CH4 production in subsamples
(20 ml; n = 5) of the homogenized sediment used to
construct the experimental cores was determined by following CH, ajxumulation
in the headspace of loo-ml
serum bottles over a 2-3-d period. These bottles were
prepared at the same time that the experimental cores
were constructed and were flushed periodically with 02free N2 to reduce the background CH4 concentration in
the headspace. Rates of CH, production in these samples
were extrapolated to an areal basis according to the volume of sediment (1 liter) used to construct the experimental cores.
Results
Sediment 1-e distributions -A preliminary analysis of
Fe distributions in TW sediments (November 1993) revealed high concentrations (20-50 pmol cm-3) of amorphous Fe(III) oxide in the upper 3 cm of unvegetated
sediments (Fig. IA). Sampling of unvegetated sediments
1737
Wetland Fe(III) reduction
A
Fe(N) (pmol cms3)
0
0
25
1
F
2
E4
r"
z.
$ 6
50
75 100
0
25 50 75 100
0
25
50
75 100
8
IO
I
0
I
I
0I
50 -loo 150 200 0
100
200 0
100
200
Fe(II), TRIS (pmol cm9)
J
I
I
0
25
I
1
I
I
50 0
I
25
I
I
1
50
Sulfate (FM)
B
Organic C oxidation
( pmol C cmw3 d-j)
Fig. 2. Chemical distributions (A) and rates of anaerobic microbial metabolism (B) in
unvegetated TW sediments collected in January (i), May (ii), and September 1995 (iii). Fe(III)Fe(II) distributions represent concentrations at time-zero in the sediment incubation experiments. Error bars for Fe(III) reduction and methanogenesis rates represent the standard error
of the slope coefficient for regressions of Fe(III) or CH, concentration vs. time. Error bars for
sulfate reduction rates represent range of duplicate (May 1995) or standard deviation of
triplicate (September 1995) samples.
on several occasions over the next 2 yr showed that comparable depth-integrated inventories of Fe(III) oxide (0.G
1.6 mol mB2) were present in surface sediments throughout an annual cycle (Figs. 2A, 3). Fe(III) oxides were
progressively reduced to Fc(I1) compounds with depth
(Figs. 1A, 2A). Millimolar
levels of dissolved Fe2+ were
present in unvegetated sediment porewaters (Fig. 1C) but
constituted ~5% of total Fe(II). RIS concentrations in-
Roden and Wetzel
1738
I
I
I
I
I
I
I
I
l
l
SOd2- (PM)
l
0
J
M
M
J
S
10
20
30
CH4m-w
40
0
0
0
IO
10
20
20
30
30
40
40
50
50
0.5
1
1.5
2
N
Month
Fig. 3. Depth-integrated (O-3 cm) Fe(III) oxide abundance
measured on several occasions in unvegetated TW sediments.
Numbers above symbols indicate year of sampling; error bar
for September 1995 = SD, n = 3; other values are results from
three or four pooled cores (January, May 1995) or from a single
core (November 1993, June 1994).
creased with depth (Fig. 2A) but accounted for < 1% of
total Fe(II) (assuming as a first approximation
that all
RIS was in the form of 0.5 N HCl-soluble sulfide, e.g.
FeS).
High amorphous Fe(III) oxide concentrations were observed throughout the rhizosphere of Juncus-inhabited
sediments (Fig. lB), presumably from Fe(II) oxidation
driven by release of O2 from plant roots (Armstrong 1967;
Green and Etherington 1977). Measurements of O2 distributions in vegetated experimental cores made with a
needle mini-electrode (Diamond General Corp.) showed
O2 penetration to depths of l-2 cm, lo-20 times deeper
than in unvegetated sediments (data not shown). Such a
deepening of O2 penetration has been noted previously
in flooded rice microcosms (Frenzel et al. 1992) and was
likely caused by O2 release from a dense layer of fine
adventitious roots in the top -2 cm of vegetated sediment. No 0, was detected below this depth in the vegetated cores. However, the kinetics of Fe(II) oxidation in
TW sediments are very fast (Roden and Wetzel in prep.),
so that rapid Fe(II) oxidation and Fe(III) oxide regeneration may occur in rhizosphere sediments even though
dissolved O2 concentrations are very low or undetectable
(~5 PM with needle electrode). Consistent with this assertion, levels of dissolved and solid-phase Fe(II) (0.0050.0 1 mM and 5-l 0 hmol cmm3, respectively) in the Juncus
rhizosphere were relatively low compared to the unvegetated sediments (Fig. 1). In contrast, below the rhizosphere, pore-water Fe2+ concentrations increased to millimolar levels comparable to those present in unvegetated
sediments.
Sulfate and methane distributions- Sulfate concentrations in unvegetated sediments determined with the porewater equilibration
sampler decreased from -30 PM in
Fig. 4. Distrikttion of SO, 2- (A) and CH, (B) in vegetated
(Juncus @USUS)vs. unvegetated TW sediment pore waters collected with pore-water equilibration samplers deployed in November 1993.
the overlying water to 5-10 PM within the upper few
centimeters of sediment, below which no further decrease
in sulfate concentration
was observed (Fig. 4A). CH4
reached saturating (- 1 mM) concentrations below the
top few centimeters of unvegetated sediments (Fig. 4B).
Pore-water sulfate concentrations in the Juncus rhizosphere were not depleted relative to the overlying water
concentration,
and CH4 concentrations were very low
compared to those at comparable depths in unvegetated
sediments (Fig. 4). Beneath the rhizosphere, sulfate was
depleted to, and CH4 increased to, concentrations comparable to those observed in unvegetated sediments.
Pore-water sul fate concentrations in unvegetated cores
sectioned under anaerobic conditions
decreased only
slightly with depth and varied bctwcen 20 and 40 PM
below the top 2 cm (Fig. 2A). Comparable time-invariant
pore-water sulfa*;e concentrations
(-20 PM) were observed during long-term anaerobic incubation of reduced
sediments (data not shown). Frequent checks with anaerobic indicator strips (Difco Lab.) verified that the atmosphere in the anaerobic chamber was consistently 02free, which ruled out the possibility that small quantities
of sulfate were formed by oxidation of RIS during core
sectioning and(or) sediment centrifugation. The origin of
these substantial background sulfate concentrations, and
the lower but significant sulfate levels observed at depth
in the pore-water equilibration
samplers, is unclear. Similar background sulfate concentrations
have been observed previously in other freshwater sediments (see Roden and Tuttle 1993), and as discussed previously (Roden
and Tuttle 1993) these probably represent some form of
microbially unavailable sulfate pool (e.g. associated with
dissolved organic matter) rather than a sulfate uptake
threshold concentration because rapid reduction of carrier-free 35S0 2- was always observed (see below) in such
sediments. It4is possible that more of this unavailable
pool is recoverec, in pore waters obtained by centrifu-
1739
Wetland Fe(III) reduction
I I
s
6. 80
y=3.946x+3.245
r2 = 0.892
60
’ 60
a
E
2 40
=
S
2 20
”
=
2 0
‘E
0
fl
0
0
,+T+4 E
O35
IO
5
15
y =0.906x - 1.06
r2 = 0.909
52-
g1
0
5
10
15
20
25
Time (d)
1
2
3
4
5
CO2 (pm01 cmm3)
Fig. 5. Fe(III) reduction, CO, production, and CH4 production in oxidized (A, B) and
reduced (C, D) TW sediment slurries. Results represent average of duplicate determinations.
gation than by collection in passive diffusion samplers,
which would explain the higher background sulfate concentrations in pore waters collected by the former procedure. Fortunately,
several independent lines of evidence (discussed below) suggest that the presence of such
unavailable sulfate pools does not pose a major problem
for assessing the contribution
of sulfate reduction to integrated TW sediment carbon metabolism.
Stoichiometry of CO2 production coupled to Fe(III) reduction and methanogenesis-An experiment was conducted to define the stoichiometry
of CO2 production
coupled to Fe(III) reduction and CO2 production during
methanogenesis in TW sediments. A parallel accumulation of CO2 and Fe(II) occurred during the first 2 weeks
of incubation of oxidized sediment slurries (Fig. 5A). The
ratio of Fe(II) production to CO2 production (3.9; Fig.
5B) was very close to that expected (4.0) for organic carbon oxidation coupled to Fe(III) oxide reduction. Rates
of CH4 production were low during the period of Fe(III)
reduction in oxidized sediments, but increased as Fe(III)
oxides were depleted to background levels (Fig. 5A). Rates
of CO2 and CH, production were about equal during
incubation of reduced sediments (Fig. 5C, D). Based on
these results, we used a 1 : 1 ratio of CO2 production to
CH4 production to calculate total rates of carbon Aowthrough methanogenesis in other experiments.
Evaluation of closed-vesselincubation technique-Fe(III)
oxide-rich (N 50 pmol cm-3) sediment from the upper 1
. cm of an unvegetated TW core was used to evaluate the
closed-vessel incubation technique as a means of partitioning organic carbon oxidation among the different an-
aerobic respiratory processes. Rates of C02, CH4, and
Fe(II) production were linear over a 3-d incubation period
(Fig. 6A-C). Much shorter incubation times were required to measure sulfate reduction rates (Fig. 6D) because of the very low concentrations and rapid turnover
of sulfate in TW sediments. Production of [35S]RIS was
essentially linear for the 3.5-h time-course incubation
shown in Fig. 6D, which suggested that reduced 35S oxidation was not a major problem on the time scale of the
experiment. Nevertheless, the standard incubation time
for sulfate reduction assays was kept very brief (0.5 h) to
minimize potential reduced 35S oxidation artifacts.
The sum of organic carbon mineralization
coupled to
Fe(III) reduction, sulfate reduction, and methanogenesis
was equal to the combined rate of CO2 and CH4 production during the surface sediment incubations (Table I),
which suggested that measurements of individual
metabolic processes by closed-vessel incubation together gave
an accurate picture of overall organic carbon metabolism.
Fe(III) reduction accounted for 67% of total carbon mineralization, compared to 4% for sulfate reduction and 29%
for methanogenesis (Table 1).
Anaerobic metabolism in unvegetated sediments - Finescale (2.5-mm intervals) sediment core sectioning-incubation experiments conducted on three different occasions showed high rates of Fe(III) reduction in unvegetated TW surface sediments (Fig. 2B). No significant rates
of Fe(III) consumption
or Fc(I1) production were observed in sediment from > 3.5-cm depth. A comparison
of dissolved Fe2+ vs. total 0.5 N HCl-extractable
Fe(II)
accumulation
in incubation experiments conducted in
September 199 5 showed that former accounted for only
1740
Roden and Wetzel
1
2
3
0
Time (d)
c
;
50
5
45
z
’
T)
il-/”
2
3
Time (d)
60
m; 55
1
40 -
T’
35 ?sE
0
t
1
0
40
:
LL 35
I
I
I
I
0
1
2
3
R
;r; 0.2
4
2 0.15
3o
2
25 a
&
0” 0.1
cn
20 g
% 0.05
15 LL
T
7
[SOJ~-] = 0.025 pmol cti3
s
‘Y=
0 h-0
tica
0
1
2
3
4
Time (h)
Time (d)
Fig. 6. CO, production (A), CH, production (B), Fc(II1) reLluction (C), and 35S0,2- reduction (D) in anaerobically incubated TW surface (O-l cm) !;ediments. Results represent
average k range of duplicate samples sacrificed at each time point.
2.8*0.5% (n = 10) of total Fe(III) reduction, which demonstrated the importance of considering solid-phase Fe(II)
formation during measurements of microbial Fe(III) oxide reduction rates in sediments (cf. Lovley 1991).
Rates of sulfate reduction and methanogenesis were low
compared to Fe(III) reduction (all expressed in common
C equivalent units) in the top few centimeters of sediment
and increased with depth as Fe(III) oxides were depleted.
With regard to methanogenesis, this effect was very distinct in January and May 1995, but less so in September
1995. It is difficult to reconcile the high rates of sulfate
reduction measured at depth in unvegetated TW sediments with the low and depth-invariant
sulfate concentrations (i.e. absence of a concentration gradient) below
3 cm, unless reoxidation of reduced sulfur was very rapid.
This process was unlikely because the only significant
potential oxidant for reduced sulfur oxidation was Fe(III)
Table 1. Rates of anaerobic microbial metabolism (pm01 C
cm-3 d-l) in TW surface sediment (O-l cm) incubation experiment.
Process
Fe(III) reduction
Sulfate reduction
Methanogenesis
Total C metabolism
CO, + CH, production
Rate
0.963 kO.220
0.054+0.004
0.412kO.03O”f
1.43fO.127Jf
1.38kO.032
% total
C metabolism*
67.4
3.8
28.8
96.5
* Process rate/total C metabolism.
j’ Accounting for observed (Fig. SD) - 1 : 1 ratio of CO, to CH4
production during methanogenesis in TW sediments.
oxide, concentrations of which were relatively low beneath the top f&w centimeters of sediment (Figs. 1A, 2A).
Additionally,
several studies have demonstrated that
complete oxid,ation of reduced sulfur back to sulfate does
not readily occur with Fe(III) oxide as an electron acceptor, even when the Fe(III) is added as a freshly precipitated, highly reactive amorphous gel (Aller and Rude 1988;
King 1990a).
Although it possible for sulfate regeneration to occur
in anaerobic sediments via thiosulfate or clemental sulfur
disproportionation
coupled to reaction of H2S with Fe(III)
or Fe(II) compounds (Jorgensen 1990; Thamdrup et al.
1993), this prol;ess would ultimately lead to the formation
of substantial quantities of iron-sulfide minerals (e.g. FeS,
Fe&), which i;; not supported by the low abundance of
these compounds at depth in TW sediments (Fig. 2). Ironsulfide concentrations manifold in excess of those observed in situ would be generated in only a few weeks at
the measured r#atesof sulfate reduction. Finally, high rates
of carbon oxidation coupled to sulfate reduction between
3- and lo-cm depth in unvegetated TW sediments are
inconsistent with the ratio of CH, production to total
CO2 production observed during May 1995 incubation
of sediments from this depth interval (1.22+0.19, n = 8;
data not shown), which indicated that methanogenesis
was the dominant pathway for anaerobic carbon metabolism. CH, to total CO2 production ratios of 10.3 would
have resulted if the measured sulfate reduction rates were
correct, Taken together, these lines of evidence argue
strongly against acceptance of the measured rates of sulfate reduction beneath the top few centimeters of unvegetated TW sediments. Thus, these data were omitted
from calculations that partitioned depth-integrated
organic carbon mineralization
among Fe(III) reduction, sul-
Wetland Fe(III) reduction
1741
Table 2. Depth-integrated carbon mineralization rates (mmol C m-2 d-l) in unvegetated
TW sediments. Rates of methanogenesis account for the observed (Fig. 5D) - 1 : 1 ratio of
CO2 to CH, production during methanogenesis in TW sediments. Error terms for Fe(III)
reduction, sulfate reduction, and mcthanogcncsis represent the depth-wcightcd sum of the
error terms for individual depths (c.g. bars shown in Fig. 2B); error terms for total C metabolism
represent the square root of the sum of squares of the error terms for the different processes
(cf. Barry 1978).
1995
17 Jan
3 May
22 Sep
Avg
Process
Fe(III) reduction
Methanogenesis
Fe(III) reduction
Sulfate reduction
Methanogenesis
Total C metabolism
Fe(III) reduction
Sulfate reduction
Methanogenesis
Total C metabolism
Fe(III) reduction
Sulfate reduction
Methanogenesis
Depth
interval
b-0
o-3
o-3
o-3
o-3
O-10
O-10 .
o-3
o-3
o-7
o-7
*
i$
O-3”
3-7$
7-105
O-1011
O-10
Rate
50.9f 12.9
4.2kO.9
63.8k21.1
8.7k 12.3
44.4+ 5.0
117k24.9
22.4k4.9
6.Ok3.3
30.1 k 1.9
58.5k6.2
31.1k29.3
7.1 k4.4
8.5k5.6
20.3+ 10.1
12.7k2.4
41.5k 11.8
79.6k31.9
% total C
mineralization
54.6
7.4
38.0
38.2
10.3
51.4
39.0
8.9
52.1
Total C metabolism
* Average -t SD of January, May, and Scptcmber data (n = 3).
t Average of May (two cores) and Scptcmbcr (three cores) data; error term represents SD of
depth-integrated rate measured in the five individual cores.
$ Average + range of May and September data (n = 2).
9 May data.
)(Error term is the square root of the sum of squares of the O-3-, 3-7-, and 7-lo-cm error
terms.
fate reduction, and methanogenesis in unvegetated TW
sediments.
Average depth-integrated Fe(III) reduction, sulfate reduction, and methanogenesis rates measured in unvegetated TW cores collected on three different dates (Table
2) indicated that Fe(III) reduction could account for - 40%
and sulfate reduction - 10% of total carbon metabolism
in the upper 10 cm of unvegetated TW sediments. Total
rates of organic carbon mineralization
decreased rapidly
within the upper 10 cm of unvegetated sediments (Fig.
7), and rates of microbial metabolism in the forest soil
present before wetland impoundment (below 1O-cm depth)
were found to be only about a hundredth of those above
(data not shown). Thus, our incubation experiments with
material from the top 10 cm accounted for essentially all
the metabolic activity in TW sediments.
Temperature dependence of metabolic pathways-The
unvegetated sediment incubation experiments were conducted at 22& 1°C whereas surface sediment (l-2-cm
depth) temperatures varied between 7 and 21°C on the
dates the cores were collected. Therefore the relative temperature dependence of the major anaerobic microbial
processes was assessed to evaluate whether routine 22°C
incubations gave an accurate representation of their relative contribution to in situ sediment carbon metabolism.
The temperature response experiment showed that Fe(III)
reduction, sulfate reduction, and methanogenesis had
similar temperature dependence (E, values of 43.6k4.3,
5 1.8 L 7.2, and 49.8 + 5.6 kJ mol-l, respectively; data not
shown; error terms represent standard error of the slope
of regression of In rate vs. l/Y). Thus, the 22°C incubations accurately reflected the relative importance of these
pathways to sediment carbon metabolism at in situ temperatures.
Rhizosphere eflects on sediment chemistry and microbial metabolism-High
concentrations of Fe(III) oxide
were formed in Juncus-inhabited
experimental cores during the course of the 4-month incubation period (Fig. 8A).
In contrast, only the upper 0.5 cm of the unvegetated
cores contained significant quantities of Fc(II1) (Fig. 8C).
Similar to observations in cores obtained from the field
(Fig. l), relatively low (compared to unvegetated cores)
but easily detectable concentrations of solid-phase and
dissolved Fe(II) (5-10 pmol cmm3 and 0.02-0.1 mM, respectively) were present in the vegetated cores (Fig. 8B,
D). As mentioned above, the presence of low Fe(II) concentrations and high Fe(III) oxide concentrations in rhizosphere sediments was likely due to Fe(II) oxidation
coupled to 0, release by plant roots.
Elevated levels of SOd2- (compared to those measured
Roden and Wetzel
1742
Organic C oxidation ( pmol C cmm3 d-l)
0
g
r
-5.
B
1
2
3
4
Fe2+ (mM)
Fe(III), Fe(II) (pmol cmm3)
0
25
50 75 100
0
R(x) = 2.4exp(-0.23x)
20
0
6
8
1 2 3 4
S0,a2- (mM)
0
5
Fe(l II), Fel’ll) (pmol cmm3)
0
I
I
25
50
6
,/O
0.03 0.06
B
2
4
CH4 mw
6
Fe2+(mM)
75 100
0
2
4
6
0
I
w
Fig. 7. Depth distribution of total organic carbon oxidation
rate in unvegetated TW sediments measured in January (M),
May (A), and September (0) 1995. Total carbon oxidation rates
were calculated from the sum of measured rates of Fe(III) oxide
reduction, sulfate reduction (O-3 cm only; see text), and methanogenesis (Fig, 2).
in situ) accumulated in the closed-system experimental
cores because of evapotranspiration
(Fig. 8A, C). There
was no evidence of downcore sulfate depletion in the
vegetated sediments; rather, sulfate concentrations
increased with depth. In contrast, sulfate was depleted to
background levels within the upper 5 cm of the unvegetated cores.
The concentration of dissolved CHQ in the vegetated
cores was - loo-fold lower than that in unvegetated sediments (Fig. 8B, D). Consistent with this major contrast,
the rate of dark-anaerobic CH4 flux from vegetated cores
was a tenth the rate of CH4 flux from parallel unvegetated
cores (Fig. 9A). The vegetated core CH4 flux rates were
also - 1O-fold lower than the rate of CH4 production in
the homogenized sediment used to construct the experimental cores [extrapolated to an areal basis according to
the volume of sediment (1 liter) used in the cores]. The
latter CH4 production rates showed good agreement with
rates of unvegetated core CH, flux (Fig. 9A).
Subsequent to the CH4 flux measurements, duplicate
vegetated cores were sacrificed for incubation experiments with unhomogenized
subsamples of rhizosphere
sediment from different depth intervals. Overall rates of
microbial metabolism were accelerated - 7-fold relative
to unvegetated sediments as a result of including fresh
root material in the incubations. Fe(III) oxide reduction
occurred without a lag and accounted for 65% of total
carbon metabolism (Fig. 9B). Sulfate reduction accounted
4
I-/
4
IO
2
5
$m
-IO
t
a0
08
O
e 0
0
OCO
00
15
0
0.1 0.2 0.3 0.4
S042- (mM)
20 t-0
0
0
cl
2
4
6
CH4 m-w
Fig. 8. Distriautions of Fe(III) oxide (N), total Fe(II) (Cl),
dissolved Fe2+ ((D), CH, (0), and SOd2- (*) in experimental
vegetated (A, B) and unvegetatcd (C, D) TW sediment cores
after incubation i n a glasshousc under conditions of continuous
flooding and natural light and temperature levels for 4 months.
Multiple data points at each depth interval represent results
from triplicate vegetated and duplicate unvegetated cores, except for SOd2-, for which results from duplicate vegetated and
a single unvegetated core are shown.
for only 13% of total carbon mineralization
compared to
22% for methanogenesis. In contrast, methanogenesis accounted for 72% of total metabolism in the unvegetated
sediment core compared to 23 and 5% for Fe(III) reduction and sulfate reduction, respectively (Fig. 9B).
Discussion
Fe diagenesis and organic carbon oxidation coupled to
Fe(III) oxide reduction in unvegetated sediments-Down-
1743
Wetland Fe(III) reduction
core conversion of Fe(III) oxides to Fe(II) compounds
suggested that Fe(III) reduction was occurring in unvegetated TW sediments (Figs. 1A, 2A). A similar conversion
of Fe(III) to Fe(II) with depth has been shown previously
for riverine sediments (Lovley and Phillips 1986; Wallmann et al. 1993). The finding that only a minor portion
of the reduced Fe present in TW sediments was in the
form of iron-sulfide compounds, together with the absence of abiotic Fe(III) oxide reduction in heat-killed sediments (data not shown), indicated that direct microbial
reduction coupled to organic carbon oxidation was the
dominant mechanism of Fe(III) reduction in unvegetated
TW sediments. That direct microbial activity rather than
interaction with sulfide dominated Fe(III) reduction was
not unexpected because Fe(III) oxides provided -2OOfold greater oxidizing capacity [5-l 5 pmol C cmA3, which
equals 20-60 pmol Fe(III) cmw3 x 0.251 compared to SOd2(0.04-0.08 pmol C cme3, which equals 20-40 PM SOd2x 2/ 1,000, ignoring correction for sediment porosities of
90-95%) in the top few centimeters of sediment (Fig. 2A),
and because Fe(III)-reducing
bacteria are able to competitively inhibit sulfate-reducing bacteria in Fe(III)-rich
freshwater sediments (Lovley 199 1). Wallmann et al.
(1993) also found that iron-sulfide minerals accounted
for only a minor fraction of in situ Fe(II) pools and for
only a minor fraction of Fe(II) produced during anaerobic
sediment incubation. In contrast to these results in freshwater sediments, recent investigations of Fe speciation in
shallow-water (< 20 m) coastal marine (Thamdrup et al.
1994) and salt-marsh sediments (Kostka and Luther 1994)
indicated that sulfide-linked
Fe(III) reduction predominated over direct microbial reduction, although substantial quantities of nonsulfide-associated
reduced iron were
present, which suggested that direct microbial reduction
contributed at least to some extent to sediment Fe diagenesis.
The unvegetated core sectioning-incubation
results
(Table 2) indicated that microbial Fe(III) reduction mediated nearly as much anaerobic carbon mineralization
as did methanogenesis in unvegetated TW sediments, i.e.
the process diverted a significant amount of sediment
carbon metabolism away from methanogenesis. Similar
to findings for freshwater Potomac River sediments (Lovley and Phillips 1986), rates of carbon oxidation coupled
to Fe(III) reduction in the upper few centimeters of sediment, where rates of organic matter decay were maximal
(Fig. 7), were comparable to depth-integrated
rates of
methanogenic metabolism deeper in the sediment column, where overall rates of metabolism were lower. This
conclusion based on bottle incubation studies, which may
or may not accurately reflect in situ conditions (Lovley
and Phillips 1986), is supported by in situ gas flux measurements, which showed that CH, emission (corrected
for CH4 oxidation at the sediment surface) accounted for
< 50% of total inorganic carbon flux from TW sediments
(Table 3). Taken together, these results strongly suggest
that nonmethanogenic
processes, principally
microbial
Fe(III) oxide reduction, play a quantitatively
significant
role in unvegetated TW sediment carbon metabolism.
Sulfate reduction mediated only a minor fraction of
n
Tb
9
25
20
E
0
E
!5
5
Y-d4
15
IO
5
0
Unveg
Orig Sed
75
.-6
B
f
50
3
l-5
25
8
0
FeRed
S04’
Red CH4 Prod
Fig. 9. Rates of CH4 flux (A) and relative rates of anaerobic
microbial metabolism (B) measured in expcrimcntal vegetated
(Juncus eflusus) and unvegetated TW sediment cores after incubation in a glasshouse under conditions of continuous flooding and natural light-temperature levels for 4 months.
anaerobic carbon mineralization
in unvegetated TW sediments (Table 2). This result is not unexpected because
of the very low sulfate concentrations and high organic
matter content of these sediments, and because of the
abundance of Fe(III) oxides. In contrast to these results,
recent studies (Wieder et al. 1990; Nedwell and Watson
1995) have indicated that sulfate reduction can be an
important sediment carbon oxidation process in mineralpoor (i.e. Fe-poor) freshwater peatlands with enhanced
sulfate inputs. Sulfate reduction apparently dominates
anaerobic carbon metabolism in organic-rich, shallowwater coastal marine and salt-marsh sediments, even those
that contain significant quantities of Fe (Thamdrup et al.
1994; Kostka and Luther 1995). This result may be explained by the strong potential for rapid, spontaneous
chemical reduction of Fe(III) oxides in surface sediments
by upward-diffusing
dissolved sulfide produced by sulfate
reduction deeper in the sediment. Thus, intense competition between sulfide-linked and microbial Fe(III) reduction is likely in organic-rich marine sediments, a phe-
1744
Roden and Wetzel
Table 3. In situ unvegetated TW sediment CH4 and CO, Rux rates (mmol C m-2 d-l).
Water depths ranged from 2 to 10 cm. Error terms represent the standard error of the slope
coefficient (n = 6-8).
CH, llux
CorrcctMeasured
cd-l12.5-+ 1.7
42.7
12.7k6.4
43.4
5.7-t0.6
19.5
2.5kO.5
8.5
2.2-t-0.5
5.9
2.3kO.3
6.1
1.2kO.7
2.4
4.020.2
8.2
2.4kO.9
4.8
0.58kO.04
1.2
11.6k2.2
64.0
16.2kll.O
89.5
4.5kO.7
25.0
4.OkO.6
22.0
Methanogenic C
flux/total*
Date”
co, flux
Tota!
v-4
25 Aug 94
187k5.7
200
42.8
133k2.5
146
59.5
7 Sep 94
147k20.2
153
25.5
168k74.3
171
10.0
17 Jan 95
lO.lk1.4
12.:a
95.6
47.5k8.4
49.E’
24.5
3 May 95
14.5k3.2
15.7
31.3
19.7zk5.0
23.7
69.0
80.3k7.8
82.17
11.7
21.5k3.1
22.1
10.7
22 Sep 95
203k20.2
214
59.8
164k70.5
181
99.1
129f16.6
133
37.5
126k12.1
130
33.7
Mean
43.6 (n=14)
SD
29.3
* Surface sediment temperatures were 20°C in August 1994, 2 1°C in September 1994, 7°C in
January 1995, 14°C in May 1995, and 20°C in September 19’35.
t Corrected for CH, oxidation at the sediment surface according to relative rates of CH, flux
from intact sediment cores under light-aerobic vs. dark-anaerobic conditions (seetext). Lightaerobic : dark-anaerobic CH, flux ratios of 0.38&O. 16 (n = 2), 0.49+0.03 (n = 3), and
0.1810.06 (n = 3) were dctermincd in January, May, and September 1995; an average lightaerobic: dark-anaerobic CH, flux ratio of 0.348 was assumed for August and September
1994.
$ Accounting for observed (Fig. 5D) - 1 : 1 ratio of CO, to CH, production during methanogenesis in TW sediments.
nomenon of far less significance in freshwater environments where sulfate is typically depleted to limiting concentrations within the top few centimeters of sediment.
An exception to this generalization is apparent in some
relatively organic-poor, deeper water (L 190 m) continental margin sediments in which microbial Fe(III) reduction
was estimated at 30-50% of organic carbon oxidationcomparable to that mediated by sulfate reduction (Canfield et al. 1993).
The high concentrations of Fe(III) oxide and high rates
of organic carbon oxidation coupled to Fe(III) reduction
in unvegetated TW surface sediments could not be perpetuated without rapid renewal of Fc(II1) via oxidation
of dissolved and solid-phase Fe(II) compounds coupled
to O2 input from the overlying water. We have found that
the kinetics of this process arc very fast, - lo-100 times
faster than those of Fe(III) oxide reduction (Roden and
Wetzel in prep.). Rapid Fe(II) reoxidation, together with
the assumption of substantial particle mixing rates caused
by gas bubble ebullition and(or) macrofaunal activity,
provides an explanation for how high concentrations of
Fe(III) oxide can persist in TW surface sediments despite
rapid rates of Fe(III) reduction. These findings indicate
that models that incorporate Fe cycling and the suppression of methanogenesis by Fe(III) reduction are required
to simulate CH4 production in and emission from Fe-
rich wetland sediments, in contrast to previous models
(e.g. James’s I 993 model for Fe-poor, calcarcous Florida
everglades sediments) that only considered competition
between meth anogenic and sulfate-reducing bacteria.
Fe cycling and suppression of CIJI, production in vegeof dissolved and solidtated sedimelzts-Comparison
phase chemical distributions
in vegetated and unvegetated sedimerts demonstrated that Juncus had a major
influence on sediment chemistry and suggested that organic matter Idecomposition pathways differed in vegetated vs. unve:;etated TW sediments. Dissolved CH4 concentrations were very low within the Juncus rhizosphere
but increased to concentrations approaching those in unvegetated sediments below it (Fig. 4A). Several alternative but not mutually exclusive explanations may account
for this observation: the plants acted as conduits that
promoted CH4 release from the sediments to the atmosphere (Chanton and Dacey 199 1); introduction of oxygen
into rhizosphere sediments via plant roots stimulated CH4
oxidation in this zone (King 1994); and CH4 production
within the rhizosphere was suppressed by competition
among methanogens and other anaerobic respiratory bacteria, which utilized external electron acceptors [e.g.
Fe(III)], to oxidize organic substrates that would have
otherwise fueled CH4 production. Although Juncus roots
1745
Wetland Fe(III) reduction
may well have served as a conduit for CH4 release and a
site for CH4 oxidation in TW sediments, the finding that
rates of CH4 flux from the experimental Juncus-inhabited
cores under dark-anaerobic
conditions (which circumvented CH4 oxidation) were 10 times lower than those
from unvegetated cores suggested that something other
than, or in addition to, these two processes affected CH4
production-flux
in vegetated sediments. This interpretation is consistent with the fact that vegetated core darkanaerobic CH4 flux rates were also an order of magnitude
lower, on an area-normalized
basis, than rates of CH4
production in the sediment used to construct the experimental cores (Fig. 9A).
In support of the third explanation above, Fe distributions measured in field cores (Fig. 1) and experimental
vegetated sediment cores (Fig. 8) indicated that Fe(II)
oxidation coupled to plant root O2 release maintained a
substantial pool (20-50 pmol cmm3) of Fe(III) oxides in
rhizosphere sediments. These high Fe(III) concentrations
suggest that microbial Fe(III) reduction could have been
a dominant pathway of anaerobic carbon metabolism in
rhizosphere sediments, and thereby significantly
suppressed rates of CH4 production. Although formation of
Fe(III) and Mn(IV) oxides in vegetated lake sediments
(Tessenow and Baynes 1978) and formation of Fe(III)
oxide coatings on the surface of aquatic plant roots (e.g.
Green and Etherington
1977; St-Cyr et al. 1993) have
been demonstrated previously, the potential for inhibition of methanogenesis by Fe(III) oxides in the rhizosphere of aquatic plants has not been evaluated previously.
Methanogenesis should have been inhibited in the Juncus rhizosphere, because the concentrations of Fe(III) oxide present were sufficient (based on results shown in Figs.
2B and 6, and summarized in Tables 1 and 2) to allow
Fe(III) reducers to compete effectively with methanogens.
The presence of measurable levels of dissolved and solidphase Fe(II) (Figs. lB, D, 8B) indicated active Fe(III)
reduction in these sediments. These results are consistent
with recent Mossbauer spectroscopic analysts of Fe associated with Phragmites australis roots, which indicated
that Fe(II) accounted for a significant proportion (- 30%)
of total Fe (Wang and Peverly 1996). Active Fe(III) reduction and suppression of CH4 production in the Juncus
rhizosphere were demonstrated directly by the vegetated
core sediment incubation experiments, in which Fe(III)
reduction commenced without a lag and mediated -3fold higher rates of organic carbon metabolism relative
to methanogenesis (Fig. 9B).
Although the overall, long-term relationship between
vegetation cover and methane release seems to be positive, as revealed by strong correlations between plant production and(or) biomass and methane emission (Whiting
and Chanton 1993), the influence of plants on CH4 production and release in different types of wetland sediments at varying time scales is at present unknown. In
particular, the extent to which plants enhance sediment
CH4 generation through release of organic exudates or
production of root-derived organic detritus (e.g. Holzapfel-Pschorn et al. 1986) vs. the extent to which they at-
tenuate CH, emission by facilitating methane oxidation
in the rhizosphere or by suppressing CH4 production
through stimulation
of competing anaerobic microbial
respiratory processes [e.g. Fe(III) reduction] is unknown.
Our vegetated sediment CH4 flux and anaerobic incubation experiments provide the first evidence that microbial Fe(III) oxide reduction and attendant Fe cycling
can cause significant suppression of methanogenesis in
the rhizosphere of Fe-rich wetland sediments.
An additional factor that may complicate interpretation of the controls on CH4 production-flux
in Fe-rich
vegetated sediments is that, in addition to acting as a
mechanism for suppression of CH, production, rapid Fe
cycling in the rhizosphere could indirectly enhance vegetated sediment CH, flux, because consumption of O2
during Fe(II) oxidation may limit the activity of CH4oxidizing bacteria, whose metabolism in rhizosphere sediments is likely limited by O2 availability
(King 1994).
Unfortunately,
we did not determine rhizosphere CH4
oxidation rates (e.g. as a fraction of total CH4 production),
so we cannot evaluate the extent to which this phenomenon may have influenced the efficiency of CH4 oxidation
in TW compared to other wetland sediments. The relative
importance of these potentially competing effects of rhizosphere Fe cycling on CH4 production vs. oxidation requires further investigation.
Suppression of sulfate reduction in the rhizosphereRhizosphere sulfate concentrations measured in the porewater equilibrator were not depleted relative to the overlying water concentration (Fig. 4B), whereas sulfate was
rapidly depleted to background levels below the root zone.
Although the possibility that the rhizosphere sulfate pool
was being rapidly replenished by sulfide oxidation processes cannot be eliminated, an alternative explanation
is that sulfate reduction was relatively inactive within the
rhizosphere. As in unvegetated surface sediments, the
concentration of electron-accepting
equivalents associated with Fe(III) oxide (Fig. 1B) in rhizosphere sediments
was -200-fold
greater than that associated with sulfate,
and this large difference together with the ability of Fe(III)reducing bacteria to outcompcte sulfate-reducing bacteria
in the presence of abundant amorphous Fe(III) oxide
strongly suggests that sulfate reduction should not have
been a particularly active carbon mineralization
process
in the Juncus rhizosphere. This suggestion is supported
by results of the experimental vegetated core experiments,
in which sulfate was not depleted with depth (Fig. 8B; cf.
8C), and in which sulfate reduction, determined by shortterm incubations in the absence of the oxidizing influence
of plants roots, mediated only a minor fraction (- 10%)
of total carbon metabolism (Fig. 9B).
Conclusions
and implications
Collectively our results provide quantitative
support
for previous suggestions (Lovley 199 1; Westermann 1993)
that microbial Fe(III) reduction could significantly suppress both CH4 production and sulfate reduction in Fe-
1746
Roden and Wetzel
rich freshwater wetland sediments. Although in no instance was methanogenesis completely inhibited by Fe(III)
reduction even when Fe(III) oxides were present in concentrations of up to 50 pmol cm-3 in unvegetated surface
and Juncus rhizosphere sediments, on an areal basis Fe(III)
reduction was found to mediate an equal or greater amount
of carbon metabolism relative to methanogenesis. These
findings may provide at least a partial explanation for
recent reports that some unspecified process(es) other than
methanogenesis was responsible for ~50% of integrated
sediment carbon metabolism in southeastern U.S. wetlands (e.g. Kelly et al. 1990; Happell and Chanton 1993;
Pulliam 1993). Additionally,
our findings indicate that
Fe(III) oxide reduction could markedly limit rates of sediment CH4 production in the vast freshwater wetlands of
the southern hemisphere, where Fe concentrations are
substantial (2-6% dry wt; Irion and Forstner 1975) as a
result of the high Fe content of the surrounding highly
weathered tropical and subtropical Oxisol and Ultisol soil
groups (Bouwman 1989). This suggestion is consistent
with the fact that average rates of CH4 flux from flooded
tropical wetland sediments are comparable to those from
mineral-poor boreal and arctic peatlands even though the
temperature, which is universally a strong regulator of
CH4 production rates in aquatic systems, is 15-25°C higher in the former systems (Bartlett and Harriss 1993). Although many factors other than temperature may influence rates of CH4 flux from these systems, our results
demonstrate that Fe(III) oxide suppression of CH4 production in Fe-rich systems needs to be considered.
Several recent studies have pointed to the potential for
sulfate reduction to significantly limit wetland sediment
CH4 production-flux
even in relatively low-sulfate freshwater systems (Wieder et al. 1990; Freeman ct al. 1994;
Nedwell and Watson 1995). In these studies, enhanced
sulfate input attributable to acid precipitation
or sulfide
oxidation during sediment drainage and oxidation may
have been responsible to some extent for the observed
suppression of CH4 production.
Another recent study
showed that sediment CH4 abundance and release rates
were - 1O-fold lower in northern Belize wetlands that
were on gypsum (CaSO,)-rich limestone marls vs. those
that were on sulfate-poor
alluvial
sands (Post and
Rejmankova unpubl.). The potential role of Fe(III) reduction in carbon mineralization
was not evaluated in
these studies but was likely minor since they were all
conducted in relatively mineral-poor peatland sediments.
An important implication of these studies together with
our findings in mineral-rich
wetland sediments is that
other factors such as temperature, organic matter abundance, and plant productivity
being equal, it would likely
be inappropriate to extrapolate CH, emission rates measured in freshwater wetlands with relatively low sediment
Fe or sulfate abundance to those with high Fe(III) oxide
or sulfate abundance.
Studies using methodology analogous to that described
here need to be conducted in representative wetlands
within geographical regions that provide a spectrum of
soil-sediment chemistry to assess the extent to which
competing anaerobic microbial processes can regulate
freshwater we,tland CH4 production. At the very least, a
modest number of sediment samples for Fe(III) oxide and
sulfate quantification
could be collected from environments when CH4 emission rates are measured. If sufficient
information elf this type can be obtained, it may be possible to include sediment Fe(III) or sulfate abundance (e.g.
along with primary production; cf. Whiting and Chanton
1993) in simple predictive models of freshwater wetland
CH4 release. T’his approach may be particularly important
for developing accurate predictive models for rates of CH4
emission from wetlands situated within Fe-rich soil-sediment regimes in tropical regions, which are estimated to
account for - 40% of CH4 release to the atmosphere from
natural wetlands (Crill 1996).
Recent studies (Wang et al. 1993; Yagi and Minami
1993) have revealed a strong negative relationship between sediment redox potential and CH4 production-flux
in rice paddy soils that was related to the intensity of
sediment oxidation processes. It is likely that this pattern
was caused in part by formation of alternate electron
acceptors such as Fe(III) oxide and(or) sulfate during sediment oxidation. In support of this assertion, a recent
investigation of competition among anaerobic respiratory
bacteria in rice paddy soils demonstrated that Fe(III) reduction media ted 60-70% of organic carbon metabolism
during the first 2 weeks of anaerobic incubation, compared to lo-20% for sulfate reduction and 15-30% for
methanogenesis (Achtnich et al. 1995). These results are
comparable tc older studies (see Lovley 199 1) that indicated that F’e(II1) oxide reduction accounted for 3565% of organic carbon oxidation during the period of
Fe(III) reduction following rice soil flooding. In view of
these findings and the quantitative suppression of surface
sediment CH4 production by Fe(III) oxides documented
here, it is likely that periodic regeneration ofFe(II1) oxides
coupled to sediment oxidation during periodic wetland
drainage could be an important mechanism limiting CH4
production and efflux during periods of flooding. Additionally, although release of 0, by rice roots has been
considered mainly in relation to bacterial CH4 oxidation
(Frenzel et al. 1992; King 1994), our results with Juncus
suggest that this process could potentially maintain a pool
of Fe(III) oxide sufficient to significantly suppress methanogenesis in :-ice soils with a relatively high Fe abundance.
References
ACHTNICH, C., F. BAK, AND R. CONRAD. 1995. Competition
for electron donors among nitrate reducers, ferric iron reducers, sulfate reducers, and methanogens in anoxic paddy
soil. Biol. Fert. Soils 19: 65-72.
ALLER, R. C., AND P. D. RUDE. 1988. Complete oxidation of
solid phase sulfides by manganese and bacteria in anoxic
marine sediments. Geochim. Cosmochim. Acta 52: 751765.
ARMSTRONG,W. 1967. The oxidizing activity of roots in waterlogged soils. Physiol. Plant. 20: 920-926.
BARRY,B. A. 19?8. Errors in practical measurement in science,
engineering, and technology. Wiley.
BARTLETT,K. B., AND R. C. HARRISS. 1993. Review and as-
Wetland Fe(III) reduction
sessment of methane emissions from wetlands. Chemosphere 26: 261-320.
BOUWMAN,A. F. 1989. Global distribution of the major soils
and land cover types, p. 33-59. In A. F. Bouwman [ed.],
Soils and the greenhouse effect. Wiley.
CANFIELD,D. E., AND OTHERS. 1993. Pathways of organic carbon oxidation in three continental margin sediments. Mar.
Geol. 113: 27-40.
CAPONE,D. G., AND R. P. KIENE. 1988. Comparison of microbial dynamics in freshwater and marine environments:
Contrasts in anaerobic carbon catabolism. Limnol. Oceanogr. 33: 725-749.
CHANTON,J. P., AND J. W. H. DACEY. 199 1. Effects of vegetation on methane flux, reservoirs, and carbon isotopic
composition, p. 65-92. In T. D. Sharkey et al. [eds.], Trace
gas emissions by plants. Academic.
-,
ANDG. J. WHITING. 1995. Tract gas exchange in freshwater and coastal marine environments: ebullition and
transport by plants, p. 98-l 25. In P. A. Matson and R. C.
Harriss [eds.], Biogenic trace gases: Measuring emissions
from soil and water. Blackwell.
CICERONE,R. J., AND R. S. OREMLAND. 1988. Biogeochemical
aspects of atmospheric methane. Global Biogeochem. Cycles 2: 299-327.
CORNWELL,J. C. 1983. The geochemistry of manganese, iron
and phosphorus in an Alaskan arctic lake. Ph.D. thesis,
Univ. Alaska. 238 p.
CRILL, P. M. 1996. Latitudinal differences in methane fluxes
from natural wetlands. Int. Vcr. Theor. Angew. Limnol.
Verh. 25: 163-171.
FOSSING,H., AND B. B. JBRGENSEN.1989. Measurement of
bacterial sulfate reduction in sediments: evaluation of a
single-step chromium reduction method. Biogeochemistry
8: 205-222.
FREEMAN,C., J. HUDSON,M. A. LOCK, B. REYNOLDS,AND C.
SWANSON. 1994. A possible role of sulphate in the suppression of wetland methane fluxes following drought. Soil
Biol. Biochem. 26: 1439-1442.
-,
M.A. LOCK,AND B. REYNOLDS. 1993. FluxesofCO,,
CH4, and N,O from a Welsh peatland following simulation
of water table draw-down: Potential feedback to climatic
change. Biogeochemistry 19: 5 l-60.
FRENZEL,P., F. ROTHFUSS,AND R. CONRAD. 1992. Oxygen
profiles and methane turnover in a flooded rice microcosm.
Biol. Fert. Soils 14: 84-89.
GREEN,M. S., AND J. R. ETHERINGTON. 1977. Oxidation of
ferrous iron by rice (Oryza sativa L.) roots: A mechanism
for waterlogging tolerance? J. Exp. Bot. 28: 678-690.
HAPPELL,J. D., AND J. P. CHANTON. 1993. Carbon remineralization in a north Florida swamp forest: Effects of water
level on the pathways and rates of soil organic matter decomposition. Global Biogeochem. Cycles 7: 475-490.
-AND W. J. SHOWERS.1995. Methane transfer
acioss the ‘water-air interface in stagnant wooded swamps
of Florida: Evaluation of mass-transfer cocfhcients and isotopic fractionation. Limnol. Oceanogr. 40: 290-298.
HOLZAPFEL-PSCHORN,
A., R. CONRAD,AND W. SEILER. 1986.
Effects of vegetation on the emission of methane from submerged paddy soil, Plant Soil 92: 223-233.
IRION, G., AND U. FORSTNER. 1975. Chemical and mineral
conditions of Amazonian lake clays. Naturwisscnschaften
62: 179.
JAMES,R. T. 1993. Sensitivity analysis of a simulation model
of methane flux from the Florida Everglades. Ecol. Model.
68: 119-146.
1747
JBRGENSEN,
B. B. 1990. A thiosulfate shunt in the sulfur cycle
of marine sediments. Science 249: 152-154.
KELLY,~. A.,C.S. MARTENS,ANDJ.P. CHANTON. 1990. Variations in sedimentary carbon remineralization rates in the
White Oak River estuary, North Carolina. Limnol. Oceanogr. 35: 372-383.
KIENE, R. P. 199 1. Production and consumption of methane
in aquatic systems, p. 111-146. In J. E. Rogers and W. B.
Whitman [eds.], Microbial production and consumption of
greenhouse gases. Am. Sot. Microbial.
KING, G. M. 1990a. Effects of added manganic and ferric
oxides on sulfate reduction and sulfide oxidation in intertidal sediments. FEMS (Fed. Eur. Microbial. Sot.) Microbiol. Ecol. 73: 13 l-l 38.
1990b. Regulation by light of methane emission from
a wetland. Nature 345: 5 13-5 15.
. 1993. Ecological aspects of methane oxidation, a key
determinant of global methane dynamics. Adv. Microb.
Ecol. 12: 3-40.
. 1994. Associations of methanotrophs with the roots
and rhizomes of aquatic vegetation. Appl. Environ. Microbiol. 60: 3220-3227.
KOSTKA, J. E., AND G. W. LUTHER. 1994. Partitioning and
speciation of solid phase iron in saltmarsh sediments. Geochim. Cosmochim. Acta 58: 1701-17 10.
-,AND1995. Seasonal cycling of Fe in saltmarsh
sediments. Biogeochemistry 29: 159-l 8 1.
LOVLEY, D. R. 199 1. Dissimilatory Fe(III) and Mn(IV) reduction. Microbial. Rev. 55: 259-287.
-,
AND E. J. P. PHILLIPS. 1986. Availability of ferric iron
for microbial reduction in bottom sediments of the freshwater tidal Potomac River. Appl. Environ. Microbial. 52:
75 l-757.
MOORE,T. R., AND M. DALVA. 1993. The influence of tcmperature and water table position on carbon dioxide and
methane emissions from laboratory columns of peatland
soils. J. Soil Sci. 44: 651-664.
NEDWELL,D. B., AND A. WATSON. 1995. CH, production,
oxidation and emission in a UK ombrotrophic peat bog:
Influence of SOd2- from acid rain. Soil Biol. Biochem. 27:
893-903.
PULLIAM, W. M. 1993. Carbon dioxide and methane exports
from a southeastern floodpain swamp. Ecol. Monogr. 63:
29-53.
RODEN,E. E., AND D. R. LOVLEY. 1993. Evaluation of 55Feas
a tracer of Fe(III) reduction in aquatic sediments. Geomicrobial. J. 11: 49-56.
AND J. H. TUTTLE. 1993. Inorganic sulfur turnover in
oligohaline estuarine sediments. Biogeochemistry 22: 8 l105.
ST-C&R, L., D. FORTIN, AND P. G. C. CAMPBELL. 1993. Microscopic observations of the iron plaque of a submerged
aquatic plant (Vallisneria americana Michx). Aquat. Bot.
46: 155-l 67.
STANLEY,E. H., AND A. K. WARD. 1996. Inorganic nitrogen
and phosphorus regimes in surface and interstitial waters
of an Alabama wetland. J. N. Am. Bcnthol. Sot. In press.
TESSENOW,
U., AND Y. BAYNES. 1978. Experimental effects of
Isoetes lacustris L on the distribution of E,, pH, Fe and
Mn in lake sediment. Int. Ver. Theor. Angew. Limnol. Verh.
20: 2358-2362.
THAMDRUP,B., K. FINSTER,J. W. HANSEN,AND F. BAK. 1993.
Bacterial disproportionation of elemental sulfur coupled to
chemical reduction of iron or manganese. Appl. Environ.
Microbial. 59: 101-108.
3H. FOSSING,AND B. B. JORGENSEN.1994. Manganese,
1748
Roden and Wetzel
iron, and sulfur cycling in a coastal marine sediment, Aarhus Bay, Denmark. Geochim. Cosmochim. Acta 58: 5 1155129.
WALLMANN, K., K. HENNIES,I. KONIG, W. PETERSEN,
AND H.
D. KNAUTH. 1993. New procedure for determining reactive Fe(III) and Fe(II) minerals in sediments. Limnol.
Oceanogr. 38: 1803-l 8 12.
WANG, T. G., AND J. H. PEVERLY. 1996. Oxidation states and
fractionation of plaque iron on roots of common reeds. Soil
Sci. Sot. Am. J. 60: 323-329.
WANG, Z. P., R. D. DELAUNE, P. H. MASSCHELEYN,
AND J. W.
H. PATRICK. 1993. Soil redox and pH effects on methane
production in flooded rice soil. Soil Sci. Sot. Am. J. 57:
382-385.
WESTERMANN,
P. 1993. Wetland and swamp microbiology, p.
295-302. In T. E. Ford [ed.], Aquatic microbiology: An
ecological approach. Blackwell.
WESTRICH,J. T., AND R. A. BERNER. 1984. The role of sedi-
mentary organic matter in bacterial sulfate reduction: The
G model tested. Limnol Oceanogr. 29: 226-239.
WHITING, G. J., ANDJ. P. CHANTON. 1993. Primary production
control of methane emission from wetlands. Nature 364:
794-795.
WIEDER,R. K., AND G. E. LANG. 1988. Cycling of inorganic
and organic sulfur in peat from Big Run Bog, West Virginia.
Biogeochemistry 5: 22 l-242.
J. B. YAVITT, AND G. E. LANG. 1990. Methane production and sulfate reduction in two Appalachian peatlands. Biogeochemistry 10: 8 l-104.
YAGI, K., AND K. MINAMI. 1993. Spatial and temporal variations of methane flux from rice paddy fields, p. 353-368.
In R. S. Oremland [ed.], Biogeochemistry of global change.
Chapman &. Hall.
Submitted: 2 April 1996
Accepted: 24 July 1996
Amended: 13 August 1996