Seasonal and topographic variations in porewaters of a southeastern USA salt marsh as revealed by voltammetric profiling†

We report electrochemical profiles from unvegetated surficial sediments of a Georgia salt marsh. In creek bank sediments, the absence of ΣH2S or FeSaq and the presence of Fe(III)–organic complexes suggest that Mn and Fe reduction dominates over at least the top ca. 5 cm of the sediment column, consistent with other recent results. In unvegetated flats, accumulation of ΣH2S indicates that SO42- reduction dominates over the same depth. A summer release of dissolved organic species from the dominant tall form Spartina alterniflora, together with elevated temperatures, appears to result in increased SO42- reduction intensity and hence high summer concentrations of ΣH2S in flat sediments. However, increased bioturbation and/or bioirrigation seem to prevent this from happening in bank sediments. Studies of biogeochemical processes in salt marshes need to take such spatial and temporal variations into account if we are to develop a good understanding of these highly productive ecosystems. Furthermore, multidimensional analyses are necessary to obtain adequate quantitative pictures of such heterogeneous sediments.


Introduction
Much of the Atlantic coast of North America is bordered by salt marshes dominated by Spartina spp. grasses. These marshes are extremely productive environments exhibiting rapid geochemical cycling. 1 Factors such as tidal regime, temperature, topography, hydrology, vegetation, infauna and microbiota [1][2][3][4] act in concert to produce a geochemically complex environment. Chemical studies of both spatial and seasonal variations are few. 1,[5][6][7] Recent studies of salt marshes [8][9][10][11][12] have employed voltammetric sensors for their high spatial resolution, ability to sample sediment porewaters with minimal physical or chemical disturbance, and ability to simultaneously measure several important dissolved analytes including O 2 , Mn 2z , Fe 2z , iron(III)-organic complexes Fe III L, soluble FeS species, and reduced sulfur species SH 2 S (~S 22 z HS 2 z H 2 S z S 0 z S x 22 ). 8,[11][12][13][14][15] In this paper, we attempt to illustrate, and account for, chemical differences between surficial sediment porewaters (upper ca. 5 cm) from two types of local environment, and changes over a period of several months.
Bioturbated marine sediments are highly heterogeneous, and sources and sinks for porewater analytes can be highly localised. Consequently, in 1D representations where lateral fluxes are neglected, it is easy to overestimate vertical transport processes such as fluxes at the sediment/water interface. 16 Onedimensional approaches generally limit our ability to account for complex features of sediments. Several studies have obtained 2D views of marine sediments by careful sectioning (e.g. ref. 17), by the diffusive gradients in thin films (DGT) technique (e.g.ref. 18) or by image analysis. 19 Using voltammetric microelectrodes, Luther et al. 20 have compiled separate 1D profiles into a 3D representation of sediment porewaters surrounding a worm burrow. Here, we employ arrays of voltammetric electrodes, lowered together into the sediment at well-defined positions and scanned sequentially, to construct temporally and spatially synchronised 3D profiles of sediment porewaters for the first time. This constitutes a further step toward adequate portrayals of these complex systems.

The study site
The Saltmarsh Ecosystem Research Facility (SERF) at the Skidaway Institute of Oceanography on Skidaway Island, Georgia, southeastern USA, was the location for this study. This is a 600' boardwalk out into intertidal, regularly inundated, tall Spartina alterniflora marsh. Two distinct unvegetated environments were sampled; flats between Spartina plots and soft, muddy creek banks. Crab, worm, and shrimp burrows were observed throughout the site, but were substantially more numerous on creek banks. Temperatures increased considerably during the study; mean air temperatures in the nearby city of Savannah varied from 11.1 uC in February and 15.0 uC in March to 26.1 uC in June and 27.8 uC in July (NOAA). Sea surface temperatures off the Savannah coast followed a very similar trend (NOAA).

Methods
On four occasions-February, March, June and July-8 cm in diameter 6 at least 20 cm deep cores were taken from within 30' of the boardwalk during falling tides. Sediments were only cored while submerged, and the overlying water was retained throughout sampling and analysis.
High resolution voltammetric profiling, using Au|Hg microelectrode probes, was carried out essentially as described previously. 8,21 Profiles were always obtained some centimetres from obvious burrow openings. For the March cores and the June flat core, four or five working electrodes were bundled together and scanned sequentially at each depth using a DLK-MUX-1 electrode multiplexer (Analytical Instrument Systems, Inc.: AIS); electrode tips were no closer than 6 mm in the horizontal plane, and no further apart than 42 mm. For the February and March cores, a combination pH microelectrode (Diamond General Development Corp.) was similarly positioned next to the working electrode(s). For the July cores, a DLK-MAN-1 micromanipulator (minimal depth increment 0.1 mm) and controller (AIS), were substituted for the manual micromanipulator (Narishige) used otherwise. The working electrode used in the June flat core, and two of those used in the July flat core ( Fig. 7 and 8), were steel-housed probes constructed from surgical stainless steel capillary tubing and Teflon-coated 75 mm in diameter gold wire (A-M Systems); total tip diameters were approximately 0.5 mm. Allowing for the smaller electrode diameter, sensitivities for these electrodes were similar to those for glass-bodied electrodes, and voltammograms for the two types of electrodes were indistinguishable.
Voltammetric peak and wave heights and potentials were extracted from scan files using a linear background subtraction procedure. Where necessary, overlapping peaks were deconvoluted into two Gaussian curves, using Peakfit v. 4 (Jandel Scientific Software). The electrodes were calibrated against O 2 and Mn 2z standards before use, and the pilot ion method 8 was used to quantify Fe 2z and SH 2 S. Sensitivities are not yet available for Fe III L and FeS aq , and hence these analytes are reported as peak currents rather than concentrations.
Results from bundled electrodes are represented as arrays of coloured polygons. Each polygon shows analyte concentrations at a given depth below the sediment/water interface (SWI). The polygons are arranged in order of increasing depth, reading from the left to the right of the figure, and then from top to bottom. The vertices indicate the coordinates of the working electrodes, and the position of the pH electrode (when present) is marked with an 6. Spectral hue indicates concentration of the analyte; colouring within the polygon is calculated by linear interpolation of the concentrations observed at each electrode at that depth.

Results
With the onset of summer the marsh changed markedly. Extensive new growth of Spartina was observed, and thin coatings of brown algae formed on creek banks. Fiddler crab population and activity were observed to increase dramatically, leaving creek banks riddled with burrows. Throughout the sampling period, bank sediments were mostly brownish, while flat sediments were dark, low-saturation colours.

Bank sediments
In the February bank core profile ( Fig. 1), O 2 was detected down to 0.5 mm below the SWI. The pH decreased from the SWI down to a shallow minimum of 6.47 at ca. 5 mm. Mn 2z appeared at 9 mm and slowly increased to ca. 30 mm at a depth of 51 mm. It was joined by Fe 2z at 16 mm, also steadily increasing to a maximum of 108 mm. A peak assigned to Fe III L, E p # 20.4 V, appeared at 45 mm. Below 51 mm, Mn 2z and Fe 2z tailed off, while the Fe III L signal increased to ca. 75 nA.
In March, only Fe 2z was observed (Fig. 2), with considerable local heterogeneity. This core was taken from a point at which the creek bank sloped steeply, so the SWI and the Fe 2z polygons were sharply inclined. (In Fig. 2, the slope is roughly parallel to the plotted y-axis, so the designated origin is also the lowest point in each polygon.) (1) The largest Fe 2z source is near the surface around the designated origin, seen as a broad profile beginning immediately below the SWI, reaching 318 mm at 6 mm and disappearing around 20 mm. At the other electrodes, respectively clockwise; (2) ca. 50 mm between 20 and 30 mm below SWI, then (3) only ca. 30 mm between 10 and 13 mm below SWI, and (4) a narrow feature, maximum 144 mm, between 6 and 9 mm below SWI, and a second, maximum 38 mm, between 36 and 42 mm. The initial decrease in pH was marked, from ca. 7.6 in the overlying waters to 7.22 at the SWI to ca. 6.7 below 8 mm. There was no subsequent increase with depth in this case.
In June, O 2 was only detected above the SWI (Fig. 3). Porewater Mn 2z appeared immediately at the SWI, and by 7 mm had reached a stable concentration of about 450 mm. Below 34 mm it was abruptly replaced by Fe 2z , which reached a maximum of 870 mm at 37 mm, then disappeared by 51 mm. It was replaced in turn by Fe III L, first seen at 44 mm, reaching a maximum of 55 nA at 58 mm, and tailing off at the bottom of the profile, 74 mm.
In the July bank core (Fig. 4), Mn 2z was observed out into the overlying water, taking a maximum of 111 mm at 10 mm below SWI, and persisting underneath the Fe 2z maximum of 336 mm at 35 mm. As the profile ends at 39 mm, it is not possible to see whether Fe III L is still present below ca. 4 cm, as in the June core (Fig. 3).

'Flat' sediments
The February flat core (Fig. 5) closely resembled the February bank core, although concentration maxima were somewhat greater and shallower throughout. The pH was 6.59 at the SWI, decreased to a minimum of 6.51 at ca. 4 mm, then steadily underwent a marked increase with depth, to a final value of 6.87. O 2 was not observed below the SWI. Mn 2z and Fe 2z were first seen just below the SWI, increasing gradually with depth to maxima of 135 mm for Mn 2z at 36 mm, 741 mm for Fe 2z at 40 mm. Fe III L appeared at 20 mm, and again currents became more intense with the disappearance of Mn, reaching a record 150 nA at the bottom of the profile.
SH 2 S dominated the March flat core (Fig. 6), though traces of other analytes were occasionally visible. It was first observed around 12 mm below the SWI on all five electrodes, and reached concentrations near the bottom of the profile of 560, 16, 114, 506, and 460 mm, respectively, clockwise from the designated origin, indicating a 'front' of sulfide diffusing both upward and in the y-direction. O 2 was only detected above the SWI. The pH was 6.81 at the SWI, decreasing to a minimum of 6.20 at 20 mm, then recovering to 6.3 by 45 mm, the bottom of the profile.
In June, again O 2 was not observed in the sediment (Fig. 7), and again SH 2 S dominated the core at depth; it was first detected 9 mm below the SWI, and reached 1680 mm at the bottom of the profile. However, large signals were also evident for other species. Fe III L appeared at 3 mm depth, Fe 2z at 4 mm, Mn 2z and FeS aq at 9 mm. Fe III L, Fe 2z and Mn 2z shared a maximum at 11 mm below SWI, of 50 nA, 1630 mm and 1310 mm, respectively. Mn 2z and Fe III L abruptly disappeared at 15 mm, but Fe 2z and FeS aq persisted down to 24 mm, with a second Fe 2z maximum at 19 mm. SH 2 S in July (Fig. 8) began at ca. 8 mm below the SWI on all four electrodes. Again, there is an apparent 'front', moving upward and laterally from the peak value of 2290 mm at the bottom of the profile, in the upper left corner of the polygon. FeS aq was often observed below ca. 5 mm on all electrodes, reaching current intensities around 20 nA, and disappearing as SH 2 S increased over ca. 500 mm. Fe 2z was briefly observed just below the SWI at the designated origin, reaching 110 mm at 4 mm depth before abruptly disappearing. O 2 did not penetrate into the sediment.      (Table 1). This includes the reduction of MnO 2 to Mn 2z , FeOOH to Fe 2z , and the reduction of Fe III L (Eq. 3), which is particularly rapid. 15 Precipitation with Fe 2z seems to occur via a detectable intermediate, FeS aq (eqn. (4)). Abiotic reoxidation by O 2 regenerates MnO 2 , FeOOH, and SO 4 22 , completing the redox cycles (Table 2). pH profiles in the February and March bank cores ( Fig. 1  and 2), and in the February flat core (Fig. 5), are typical of organic matter-rich coastal sediments. 9,30 The pH decrease in the first ca. 5 mm is ascribed to H z -producing O 2 and NO 3 2 oxidation of organic carbon. Subsequent increase with depth is ascribed to H z -consuming reduction of Mn and Fe oxides, confirmed here by the appearance of Mn 2z and then Fe 2z . The thermodynamic sequence 22 of O 2 disappearance followed by the appearance of Mn 2z , Fe 2z was observed in the other bank cores (Fig. 2-4), at successively shallower depth-indeed, in July Mn 2z had diffused out into the overlying water.

Discussion
Heterotrophic SO 4 2reduction initially yields H 2 S; but neither H 2 S, nor the partially reoxidised forms S x 22 , S 0 , or S 2 O 3 22 , were ever detected in the first 40-80 mm of any of the bank cores ( Fig. 1-4). Nor was FeS aq , which would be generated in the presence of such high levels of Fe 2z . 14 Also, sulfides rapidly reduce Fe III L (eqn. (3)) yet Fe III L was seen in the February and June bank cores ( Fig. 1 and 3). These observations suggest that SO 4 22 reduction is not an important mechanism in these surficial sediments. Thus, sulfide reduction of MnO 2 and FeOOH (eqn. (1) and eqn. (2)) is unlikely to be significant over the depth profiled, and the principal reduction mechanisms for these minerals appear to be biological.
While this suggestion runs counter to the prevailing view that SO 4 22 reduction is the principal electron-accepting process in salt marshes, at least in vegetated sediments, 31,32 similar results have recently been obtained in the nearby Sapelo Island salt marsh. Lowe et al. 4 found that iron-reducing bacteria were most abundant in the top 6 cm of a core from an unvegetated creek bank, whereas SO 4 22 -reducing bacteria were most abundant in deeper sediments. Porewater SO 4 22 did not decrease significantly from SWI levels until around 6 cm depth. Both observations suggest that Fe and Mn reduction dominates in surficial sediments. Kostka et al. 33 reported Fe reduction rates approximately 4 times greater than SO 4 22 reduction rates in the upper 5 cm of unvegetated creek banks; although they suggested that abiotic reaction with sulfide species accounted for some of the Fe reduction, on the basis that the sum of Fe and SO 4 22 rates substantially exceeded their measured carbon mineralisation rate. In a subsequent paper, 34 they used geochemical parameters, rate measurements and bacterial counts to conclude that Fe reduction was the predominant microbial respiration process in either bioturbated or vegetated sediments at their study site.

Sulfide accumulation
While the February flat core (Fig. 5) most closely resembled the February bank core (Fig. 1), extensive porewater SH 2 S accumulation, and hence SO 4 22 reduction, were observed in the March flat core (Fig. 6). The deeper pH minimum, around 20 mm below the SWI, may be due to H z -producing reoxidation of H 2 S or FeS. 30 The June flat core (Fig. 7) was also mostly sulfidic, with SH 2 S reaching 2 mM. High concentrations of Fe 2z led to formation of FeS aq and presumably precipitation of FeS s (eqn. (4)); between 15 and 22 mm, FeS s was theoretically supersaturated unless the pH was ¡ 5. Fe III L was observed above 15 mm, and since it rapidly reacts with H 2 S to form Fe 2z and S 0 (eqn. (3)), SH 2 S was most probably dominated by S 0 above 15 mm. Nonetheless, H 2 S was rapidly supplied to the upper core by diffusion along the steep SH 2 S concentration gradient from the bottom of the profile, allowing ready abiotic reduction of reactive Fe and Mn (eqn. (1)-(3)), and possibly accounting for much of the observed Fe 2z and Mn 2z .
Observed SH 2 S concentrations in 'flat' sediments increased dramatically during the study period, and cores became sulfidic at a shallower and shallower depth (Fig. 5-8). The same trend has previously been reported for New Hampshire salt marshes, and ascribed to the release of significant amounts of dissolved organic carbon by tall form Spartina alterniflora during the summer growth period. 1,35 This release, combined with elevated summer temperatures, greatly increases SO 4 22 reduction rates, as SO 4 22 reduction in marine sediments is organic matter-limited and temperature-dependent. 36,37 Here, H 2 S accumulation appeared to begin in March (Fig. 6), somewhat earlier than in the more temperate New Hampshire climate. 1

The role of bioturbation
We propose that the dichotomy between largely suboxic bank sediments and largely sulfidic flat sediments is due to the more extensive bioturbation inferred in creek banks (cf.ref. 33,34,38). This ensures a strong supply of O 2 , and freshly formed MnO 2 and FeOOH, all probably more reducible than SO 4 22 . 22 Not only does this disfavour H 2 S production, but these oxidants will also reactively consume any H 2 S that is formed (eqn. (1)- (3) and (8)).
There are several distinct supply mechanisms: bioturbation by crabs and other infauna mixes well-oxidised material down the sediment column, 2,39 providing a deep reservoir of Fe and Mn oxides, and of Fe-organic complexes. Furthermore, in well-worked sediments, MnO 2 and FeOOH contents are broadly dependent on particle surface area. 40 Consequently, the finer bank sediments can be expected to contain more reactive Fe and Mn, and hence have an inherently greater poising capacity. Conversely, burrowing and deposit feeding bring reduced sediments and associated porewaters up to the surface, directly exposing them to well oxygenated water or to air. 39,41 This substantially enhances O 2 reoxidation of subsurface reduced species (eqn. (5)-(8)), particularly Fe 2z for which oxidation is rapid. 25 Bioirrigation also supplies O 2 to the deep sediment, both through the actions of occupying macrofauna 42,43 and passively by flow of oxygenated water through burrows. [44][45][46] Further, on the sloping banks, burrows can drain fully, and subsequent drying allows air into burrows and cracks.
Since fiddler crab population (cf. ref. 47) and activity (cf. ref. 48) appeared to increase substantially during the sampling period, bioturbation and bioirrigation should have increased substantially, too. 2 Polychaete worms may also make a significant contribution to sediment oxidation, even when burrowing crab population is high 19 and polychaete density would also be expected to increase in late spring. 49 Thus, even in summer conditions favouring SO 4 22 reduction, bioturbation and bioirrigation were able to provide sufficient O 2 and reactive Mn and Fe oxides to keep bank sediments suboxic to the 40-80 mm limit of profiling depth.

Formation of iron (III)-organic complexes
Soluble iron(III)-organic complexes have been detected electrochemically 8,11 or by other techniques 6,50 in salt marsh sediments on a few previous occasions. Fe III L was a significant feature of four of the eight core profiles given here (Fig. 1, 3, 5,  7). Taillefert et al. 21 have collated several mechanisms for formation of Fe III L in marine sediments (Table 3): In this study, Fe III L only occurred at depth, so O 2 or 2FeL n 3 2 2n z H 2 S z nH z A 2Fe 2z z S 0 z nHL 2b (4) Fe 2z z H 2 S FeS aq A FeS s a e.g.ref. 14, 15, 22-24. b L is assumed to be a bidentate oxygen donor ligand, and n is the number of such ligands in the complex.
e.g.ref. 25-29. bacterially mediated NO 3 2 (ref. 51) oxidation of Fe II L complexes (eqn. (9) and (10)) seem unlikely formation mechanisms. The Fe III L current intensity was generally inversely correlated with Mn 2z , indicating that Mn oxidation of Fe II L complexes (eqn. (11)) was not a significant formation mechanism in this case, either. Furthermore, oxygen-donor ligand-mediated electron transfer from FeOOH (eqn. (12)) is slow above pH 6, 52,53 while the pH was always above 6 in the February and March cores, and presumably also in the equally suboxic June bank core (Fig. 3). Thus nonreductive dissolution of FeOOH (eqn. (13)) was most likely the principal mechanism of Fe III L formation in these cores. However, it is not clear why the Fe III L current intensity maxima were also deeper than the Fe 2z concentration maxima (Fig. 1, 3 and 5), i.e. at a depth where a significant proportion of FeOOH had already been consumed. We hypothesize that Fe III L was still produced above that depth, but efficiently removed by metal-reducing bacteria which were able to utilise it as an electron acceptor.

Sediment heterogeneity
We note that concentrations in plane differ widely from electrode to electrode in the 3D profiles (Fig. 2, 6 and 8) as might be expected in a heterogeneous sediment. The SH 2 S profiles of the March and July flat cores ( Fig. 6 and 8) suggest a 'front' diffusing obliquely upward from deeper sediments. However, the March bank Fe 2z (Fig. 2) and July flat FeS aq profiles (Fig. 8) show patchier profiles which may reflect microbial or chemical niches. Nonetheless, in each array similar vertical trends are seen at each electrode, and therefore we believe that single profiles are still qualitatively representative of their immediate environment. Thus, while it may be possible to identify the processes occurring in a sediment on the basis of a 1D profile, a quantitative understanding appears to require 3D representation. We strongly recommend increased use of multidimensional techniques, especially in bioturbated sediments such as those studied here, which exhibit significant heterogeneity.

Conclusion
We report electrochemical profiles from unvegetated surficial sediments of a Georgia salt marsh. In creek bank sediments, the absence of SH 2 S or FeS aq and the presence of Fe(III)-organic complexes indicate that Mn and Fe reduction dominates over at least the top ca. 5 cm of the sediment column, consistent with other recent results. In unvegetated flats, accumulation of SH 2 S indicates that SO 4 22 reduction dominates over the same depth, as the product SH 2 S is observed. A summer release of dissolved organic species from the dominant tall form Spartina alterniflora, together with elevated temperatures, appears to result in increased SO 4 22 reduction intensity and hence high summer concentrations of SH 2 S in flat sediments. However, increased bioturbation and/or bioirrigation seem to prevent this from happening in bank sediments. Studies of biogeochemical processes in salt marshes need to take such spatial and temporal variations into account if we are to develop a good understanding of these highly productive ecosystems. Furthermore, multidimensional analyses are necessary to obtain adequate quantitative pictures of such heterogeneous sediments.
This study is part of ongoing attempts to characterise the chemistry of southeastern USA salt marshes. Future studies will examine the solid phase speciation and size distribution, the microbial population, the nature and extent of bioturbation, the biogeochemistry around sediment features such as burrows, and modelling early diagenesis, in these sediments and in other local environments of the salt marsh. Table 3 Fe III L formation processes (9) 4FeL n 2 2 2n z O 2 z 4H z A 4FeL n 3 2 2n z 2H 2 O (10) 10Fe 2z z 2NO 3 2 10FeL n 3 2 2n z N 2 (11) 2Fe 2z z MnO 2 2FeL n 3 2 2n z Mn 2z (12) FeOOH z Fe*L n 2 2 2n A Fe 2z z Fe*L n 3 2 2na (13) FeOOH z nHL 2 z (4 2 n)H z A FeL n 3 2 2n z 2H 2 O a The star label is added to track the reactive pathway of that iron atom.