| HOME | HELP | FEEDBACK | SUBSCRIPTIONS | ARCHIVE | SEARCH | TABLE OF CONTENTS |
|
|
|
|||||||
|
|||||||||
a W.K. Kellogg Biological Station and Department of Zoology, Michigan State University, 3700 East Gull Lake Drive, Hickory Corners, MI 49060
b Universidad de Puerto Rico, Recinto Universitario de Mayagüez, Departamento de Agronomía y Suelos, PO Box 9030, Mayagüez, PR 00681-9030
* Corresponding author (Hamilton{at}kbs.msu.edu)
Received for publication July 2, 2004.
 |
ABSTRACT |
|---|
 TOP ABSTRACT INTRODUCTIONMATERIALS AND METHODSRESULTSDISCUSSIONREFERENCES |
|---|
Abbreviations: DNRA, dissimilatory reduction of nitrate to ammonium • DOC, dissolved organic carbon • KBS, W.K. Kellogg Biological Station
|
INTRODUCTION |
|---|
|
TOPABSTRACTINTRODUCTIONMATERIALS AND METHODSRESULTSDISCUSSIONREFERENCES |
|---|
removal by wetlands has received particular attention because of the escalating problem of NO–3 contamination of drinking water supplies, and the growing recognition that nitrogen (N) pollution of rivers causes eutrophication of marine coastal waters, leading to problems such as harmful algal blooms and oxygen depletion (Howarth et al., 1996; Mitsch et al., 2001). Most of the NO–3 removal is attributed to denitrification, a form of anaerobic bacterial respiration and a major source of atmospheric nitrous oxide (N2O), a potent greenhouse gas (Groffman et al., 2000).
The terminal steps of anaerobic microbial decomposition can occur via several alternative processes, including denitrification, sulfate 
reduction, and methanogenesis (Fenchel et al., 1998; Megonigal et al., 2004), and thermodynamic constraints on energy yields determine the competitive ability of the microbes that perform these processes. In anaerobic sediments, denitrification is the most energetically favorable form of respiration; SO2–4 reduction yields less energy and hence tends to occur when NO–3 is not available. Sulfate removal would thus be expected to occur sequentially after depletion of NO–3, unless there are other potential electron acceptors such as iron and manganese, and all of these forms of anaerobic respiration tend to preclude methanogenesis.
Anaerobic respiration in wetlands is potentially stimulated by elevated loading of NO–3 and SO2–4 as a result of air and water pollution, and the resultant removal of NO–3 and SO2–4 can help ameliorate this pollution (Kelly and Rudd, 1984; Mitsch et al., 2001; Hey, 2002). Ground water in agricultural landscapes is often enriched in NO–3 and SO2–4 due largely to fertilizer and animal waste inputs. In addition to ground water inputs, rates of loading of NO–3 and SO2–4 to wetlands via atmospheric deposition are greatly enhanced due particularly to fossil fuel combustion (Boyer et al., 2002; Mayer et al., 2002). These increased loadings represent a biogeochemical perturbation with interesting consequences for ecosystems, including greater N availability for plant growth, albeit sometimes to the point of toxicity (Fenn et al., 1998), as well as acidification of poorly buffered waters and soils and enhancement of redox transformations, all with multifarious impacts on elemental cycling. Greenhouse gas emissions are also potentially affected by the increased production of N2O, as well as inhibition of methanogenesis by the competitive superiority of denitrifiers and SO2–4 reducers (Conrad, 1996).
A critical question for wetland management and protection is the role of these anaerobic processes in reducing N and S pollution and thereby providing improved water quality at the landscape level, but our understanding has been limited by the difficulty of measuring in situ process rates and the tendency for investigations to focus on only one of these processes. Many methods have been used to estimate the nature and rates of anaerobic decomposition in sediments, including analysis of dissolved H2 concentrations (Lovley and Goodwin, 1988; Lovley et al., 1994), molecular methods (Muyzer and Smalla, 1998), lab assays in microcosms (Groffman et al., 1996; Roden and Wetzel, 1996; D'Angelo and Reddy, 1999), and whole-ecosystem isotope tracer studies (Tobias et al., 2001a; Mulholland et al., 2004). Results drawn from these approaches can be affected by disturbance of the natural environment, contamination by oxygen, and introduction of high concentrations of substrate. Whole-ecosystem isotope tracer studies can yield spatially integrated estimates of in situ rates without experimental artifacts, but are costly and would be difficult to perform in many wetland situations. Smaller scale, in situ tracer methods can be used in diverse environments to estimate microbial process rates, and are less apt to alter natural conditions than traditional lab assays. "Push–pull" tracer experiments, in which a solution containing reactants and amended with a conservative solute tracer is injected into the sediments, then subsequently withdrawn to measure rates of disappearance of the reactant relative to the conservative solute, have proven useful in estimating microbial reaction rates from aquifers (Istok et al., 1997; McGuire et al., 2002), lake sediments (Luthy et al., 2000), and riparian wetlands (Addy et al., 2002).
The purpose of this study was to investigate rates of removal of NO–3 and SO2–4 from small wetlands of variable hydrology in the glacial landscape of southwestern Michigan. We employed push–pull tracer experiments to evaluate the removal of NO–3 and SO2–4 from ground water as it comes into contact with reduced wetland sediment. This is a measurement of the potential reaction rates insofar as the ground water is relatively enriched in nitrate and to a lesser extent sulfate compared to the sediment porewaters. The level of enrichment produced is, however, within the bounds of what wetland sediments may experience as a result of natural ground water flow or precipitation events, and thus the rates do not indicate the maximum potential under conditions of nonlimiting reactants (electron acceptors and donors). The results demonstrate the capacity of wetland sediments to rapidly deplete added NO–3 and SO2–4, and the time course of the depletion provides evidence of interactions between N and S cycling in these sediment environments, raising new questions meriting further investigation. The short time and space scales over which removal was observed underscore the potential importance of narrow riparian wetlands along headwater streams and small isolated depressions in providing this ecosystem service; protection and management of such ecosystems may therefore deserve more attention.
|
MATERIALS AND METHODS |
|---|
|
TOPABSTRACTINTRODUCTIONMATERIALS AND METHODSRESULTSDISCUSSIONREFERENCES |
|---|
Experiments were performed in five wetlands spanning a range of hydrology and vegetation. Two of the wetlands have ground water as their main water source (Loosestrife Fen and Turkey Marsh), and the other three receive most of their water from precipitation (Shaw Pond, Lux Pond 10, and LTER Kettle Pond), as confirmed using dissolved magnesium as a tracer (Stauffer, 1985; Whitmire, 2003). Wetland characteristics are summarized in Table 1, and brief descriptions of each site follow.
|
Turkey Marsh is a small, isolated depression wetland located just north of the KBS Academic Center and Gull Lake. The wetland has lower-lying areas that usually hold standing water with submersed and emergent vegetation and slightly higher areas covered with dense stands of shrubs, especially common winterberry (Ilex verticillata L.). This wetland's water levels and flooded area have fluctuated considerably over the past few years, ranging from 0 to 50 cm over the sediment surface. During the initial years of higher water levels, the vegetation was mainly composed of a mixture of submersed aquatic plants and spatterdock [Nuphar advena (Aiton) WT. Aiton], but after drying in 2000 the vegetation changed to include more emergent plants such as reed canarygrass (Phalaris arundinacea L.), hairy sedge (Carex lacustris Willd.), and rice cutgrass [Leersia oryzoides (L.) Sw.], with some scattered spikerushes (Eleocharis spp.) and broadleaf cattail (Typha latifolia L.). Agricultural fields and successional forests surround this site.
The LTER Kettle Pond lies in a small, isolated depression on the KBS Long-Term Ecological Research site. The slopes around the pond are covered by deciduous forest and conifer plantations, and agricultural row crops are grown on surrounding uplands. The pond's plant community was dominated by duckweeds (Wolffia and Lemna spp.) during wet years (1996–1999), but after drying completely in 1999 the basin filled with blunt spikerush [Eleocharis obtusa (Willd.) Schult.], rice cutgrass, and reed canarygrass. Standing water was present at the beginning of summer 2000, but the water table dropped to 5 cm below the sediment surface by the middle of the summer. Standing water has persisted since fall 2000.
Lux Pond 10 is the southeastern-most pond located on the KBS Lux Arbor Reserve. The pond lies in an isolated depression surrounded by deciduous forest. Lux Pond 10 had mainly open water with a few water lilies (spatterdock) and little submersed aquatic vegetation. The banks were lined by a mix of rice cutgrass, hairy sedge, broadleaf arrowhead (Sagittaria latifolia Willd.), and spikerushes. The water levels in Lux Pond 10 dropped during the study, exposing sediments along the edges, but the pond never dried, and the sampling points were located in areas that always had standing water. Since 1996 the water levels have varied over 1.4 m. Successional forests and fields surround this site.
Shaw Pond is an isolated depression west of Otis Lake in the Barry Game State Area, about 22 km north of KBS. This pond is surrounded by deciduous forest on sandy soils. The dominant vegetation during wet years was a diverse mix of submersed and emergent aquatic plants [particularly fragrant water lily (Nymphaea odorata Aiton), spatterdock, and pondweeds (Potamogeton spp.)]. There were also some emergent grasses and spike rushes that encircle the wetland, and these species expanded into exposed sediments when water levels were low in 2000. The pond was entirely dry during late summer 1999 through the summer of 2000, but flooded again in late 2000. During the summers of 2001 and 2002, standing waters were 30 to 75 cm deep. All sampling points remained underwater during the sampling periods of this study.
Experimental Procedures
To determine the potential of various wetlands to remove NO–3 and SO2–4 from inflowing waters via denitrification and sulfate reduction, we injected local ground water containing NO–3 and SO2–4 into wetland sediments in the field using a modification of the push–pull method previously used to study biogeochemical reaction rates in contaminated aquifers. The injections were done during the summer and thus represent a synoptic survey during the season of mild temperatures and high biological activity. The experimental injection involves a "push" and a "pull" phase (Snodgrass and Kitanidis, 1998). In the "push" phase, a solution containing a reactive solute and a conservative tracer is injected (pushed) into the saturated zone of the sediments. The injection solution is dispersed and diluted as it mixes with the ambient porewater. Concentrations of the reactant may be changed by both microbial activity and dilution, and the dilution occurs by both advection and diffusion. The "pull" phase begins immediately after the injection solution has been introduced. Consecutive porewater samples are extracted from the same well over time, and the solute concentrations are measured. The conservative tracer is used to account for dilution. Comparison of the concentrations of the reactive solute to those of the conservative solute tracer reveals the net flux of the reactive solute.
In situ push–pull experiments were conducted at three of the five wetlands in 2001 and all five wetlands in 2002. Two representative sites were chosen within the central wet area of each wetland, in areas free of plant roots. To prepare the injection solution, untreated ground water was collected from a residential well at KBS (NO–3–N = 14–16 mg L–1, SO2–4 = 53–69 mg L–1, dissolved organic carbon [DOC] = 0.6–1.0 mg L–1; analytical methods given below) and stored at 4°C until the push–pull experiment. The ground water was amended immediately before each experiment with 13 to 20 mg L–1 Br– (as NaBr) to serve as a conservative solute tracer. Higher Br– concentrations (50 mg L–1) were used for Loosestrife Fen to ensure that concentrations would remain measurable in the face of the anticipated advective loss from the higher flow at that site.
The push–pull well screens were small stainless steel mesh filters (6-mm i.d., 7.6-cm screen length; no. 4258; American Science & Surplus, Skokie, IL) attached to about 1 m of Teflon tubing (3.2-mm outside diameter). Wells were inserted vertically to a depth of 10 cm below the sediment surface (screen depth was 6–13 cm deep after installation) at least 2 d before the experiment began to ensure an anaerobic environment surrounding the screen. Upon installation, 20 mL of porewater was pulled out of the well to flush and fill the filter and tubing with porewater, thus avoiding oxygenation of the sediment. The sediments at all sites appeared to seal adequately around the well and tubing, and no packing or backfilling was required. Porewater was withdrawn to measure ambient concentrations of NO–3, SO4–2, Br–, and DOC. Sediment temperature profiles were measured at each well site using a probe that was inserted nearby.
Before injection, the ground water solution was sparged with ultrapure He for at least 1 h to remove dissolved oxygen. Five hundred milliliters of the solution was transported to the field in a set of 60-mL syringes. The anoxic solutions were pushed by syringe through the wells and into the sediment over the course of 10 to 15 min (33–50 mL min–1).
Samples were immediately withdrawn by syringe after the push phase and were periodically withdrawn over time for up to 48 h after the initial injection. For each sampling, the filter and tubing were first flushed by removing 5 mL of the solution, then 20 mL of the porewater was collected for analysis. The samples were filtered in the field through 0.2-µm membrane (sterile Millex-gs; Millipore, Billerica, MA) syringe filters and cooled to 4°C until analysis. Samples were analyzed for Br–, NO–3, and SO2–4 using membrane-suppression ion chromatography. Selected samples were also analyzed for DOC by high-temperature platinum-catalyzed combustion to CO2 followed by infrared gas analysis.
For each experiment, we calculated NO–3 and SO2–4 removal rates based on Br–, NO–3, and SO2–4 concentrations (Snodgrass and Kitanidis, 1998). Both zero- and first-order rate models were fit to the data and the model with the best fit was selected to determine potential rates; in most cases the choice was obvious. Linear regressions were performed in SYSTAT 9.0 software (Systat, 1998), and the regression slopes were always significant at P < 0.05. For the case of these enzymatically catalyzed reactions, an apparent zero-order fit would represent a reaction that would be first-order at lower concentrations of NO–3 or SO2–4, but has become saturated (i.e., limited by some other factor, such as labile organic matter) at higher concentrations of NO–3 or SO2–4.
To fit the model for zero-order reactions, the concentrations of the reactive solute (NO–3 or SO2–4) were transformed to remove the effect of dilution as follows:
 | [1] |
reactant(t) (i.e., the removal rate) is described by the zero-order decay equation:
 | [2] |
The plot of reactant(t) versus time will fall on a straight line, with the linear regression slope equal to the reaction rate, if the solute disappears under zero-order kinetics (Snodgrass and Kitanidis, 1998).
In wetlands where nitrate and sulfate removal were better fit to the first-order reaction model, the change in concentration of the reactant Creactant(t) can be modeled using an exponential function:
 | [3] |
The slope of a linear regression line fit to the plot of ln [Creactant(t)/Ctracer(t)] versus time will give an estimate of the first-order reaction rate k. Since k is based on the ratio of Creactant/Ctracer, complete tracer mass recovery is not necessary to obtain accurate estimates (McGuire et al., 2002).
The analyses described above assume that: (i) the solutes are injected simultaneously in a well-mixed slug; (ii) the time required to inject the solution is short compared to the overall length of the experiment; (iii) the dominant processes are advection, dispersion, and constant-coefficient zero- or first-order reactions; (iv) the background concentrations of SO2–4, Br–, and NO–3 are negligible or are subtracted from concentrations measured during the experiment; and (v) if flow is heterogeneous then samples collected earlier are more representative of in situ processes than samples collected later. The injected solution was well-mixed before injection and the injection time was short compared to the total length of the experiment. Background concentrations of NO–3 were always negligible compared to the ground water, and SO2–4 concentrations were corrected for background levels when they were measurable. Heterogeneous flow is unlikely to be a significant factor over the small spatial scale and short time course of these experiments.
To be sure that there were no significant reactions of the conservative tracer Br– with wetland sediments, we added Br– at concentrations similar to those used in the push–pull experiments to sediments from several representative sites in sealed 1-L glass jars. Following addition of the Br– and thorough mixing, porewaters were subsampled immediately, then after 24 h and 1 wk.
|
RESULTS |
|---|
|
TOPABSTRACTINTRODUCTIONMATERIALS AND METHODSRESULTSDISCUSSIONREFERENCES |
|---|
|
|
|
|
Background concentrations of surface water and pore-water SO2–4 (Table 1) show that porewaters had markedly lower SO2–4 concentrations than surface waters, even at sites that were close to 100% ground water–fed. Nitrate was below the ion chromatography detection limit of 10 to 15 µg N L–1 in all background samples, and below or close to this limit in surface waters of these wetlands. An influx of surface water into the samples withdrawn during the experiments would not explain the observed increase in SO2–4 concentrations during the period of NO–3 removal because concentrations in overlying water were so low. Also, the appearance of the excess SO2–4 does not correspond with the most frequent sampling, which would cause the greatest potential for surface water to be drawn into the sediments.
Incubation of representative wetland sediments in sealed jars with Br– added to similar concentrations as used in the push–pull experiments showed no evidence for nonconservative behavior (Table 4). There were no statistically significant differences between sampling time points. The variation was probably due to imperfectly mixed porewaters in the jars and analytical error associated with the ion chromatography.
|
|
DISCUSSION |
|---|
|
TOPABSTRACTINTRODUCTIONMATERIALS AND METHODSRESULTSDISCUSSIONREFERENCES |
|---|
Processes of Nitrate Removal
Nitrate and SO2–4 removal appeared to take place in accordance with thermodynamic theory, where terminal electron acceptors are consumed by anaerobic respiration in the order of their free-energy yield as microorganisms compete for labile products of organic matter decomposition. Since denitrification yields more energy in the process of anaerobic respiration, denitrifiers should have a competitive advantage and SO2–4 reduction should be limited until NO–3 has been depleted. All of the push–pull experiments clearly exhibited this response, with SO2–4 removal only commencing after NO–3 had been depleted (e.g., Fig. 2).
Microbial transformation is likely to be the principal process responsible for NO–3 removal, as opposed to assimilation by microbes or plant roots. Sediment porewaters in these sites were generally very rich in ammonium (often between 200 and 1000 µg N L–1), a preferred nitrogen source for assimilative uptake by microbes, and activity of plant roots was likely limited at most sites because to the extent possible we avoided sampling close to plants. While this was difficult in some sites (e.g., Turkey Marsh, where spatterdock rhizomes were abundant), other sites were devoid of nearby vegetation and yet the pattern of NO–3 removal was similar across all sites. Furthermore, we have found similar uptake rates in intact, root-free sediment cores brought to the laboratory (A.J. Burgin and S.K. Hamilton, unpublished data).
The microbial transformation responsible for NO–3 removal in wetlands and other saturated sediments is often assumed to be denitrification to N2 (and N2O as a byproduct) via anaerobic respiration (respiratory denitrification), in which NO–3 serves as an alternate electron acceptor for the oxidation of organic matter, and either NO–3 or labile carbon can be limiting factors. The apparent production of SO2–4 observed in about half of our experiments as NO–3 was removed suggests that at least part of the NO–3 removal was somehow linked to SO2–4 production or consumption. This observation is worth further consideration because it could indicate that the NO–3 removal was not only due to respiratory denitrification, and if at least part of the NO–3 is converted to ammonium, the removal may not be a permanent N sink.
Two alternative hypotheses could explain the coincidence of NO–3 removal and SO2–4 production: (i) denitrifiers inhibited SO2–4 reducers by competing for substrate (i.e., H2), allowing SO2–4 that was continuously produced by oxidation of sulfide to accumulate as long as NO–3 was available; or (ii) sulfur-oxidizing bacteria utilized the NO–3 instead of O2 to oxidize sulfide to elemental S and SO2–4, thereby producing either ammonium or N2. The plausibility of the first hypothesis is supported by lab studies demonstrating that SO2–4 reduction can be inhibited by the superior affinity of denitrifiers for H2 (Klüber and Conrad, 1998). If that were the case, the SO2–4 production we observed would have to reflect reduced SO2–4 consumption in the face of natural rates of SO2–4 influx, perhaps from overlying oxic zones, or in situ production. However, SO2–4 concentrations in overlying waters were low compared to the added ground water, so the SO2–4 may be more likely to have come from sulfide oxidation in the surficial sediment layers. Additional experiments using S isotope tracers would be required to test this hypothesis.
The possibility that sulfur-oxidizing bacteria used NO–3 to oxidize sulfide or elemental S to SO2–4 is supported by recent research, mostly in marine ecosystems (Fossing et al., 1995; Martin and Brigmon, 1994; Bonin, 1996; Phillipot and Højberg, 1999; Zopfi et al., 2001) but also in fresh waters (Megonigal et al., 2004). It remains uncertain whether S-oxidizing bacteria involved in NO–3 uptake are denitrifiers (i.e., they reduce NO–3 to N2), or produce ammonium in a form of dissimilatory reduction of nitrate to ammonium (DNRA), and there is evidence that they may be able to switch between these pathways (Brunet and Garcia-Gil, 1996; Dannenberg et al., 1992; Otte et al., 1999).
If the S-driven NO–3 uptake is a form of denitrification, the initial oxidation step may proceed with the following stoichiometry (Fossing et al., 1995):
 |
The resultant elemental S may be stored in the cells before being oxidized on to SO2–4 in a separate step. Further oxidation of the elemental S to SO2–4 could occur by this reaction (Fossing et al., 1995):
 |
If these two reactions occurred sequentially, the molar ratio of NO–3 consumed to SO2–4 produced would be 8:5 (=1.6) as in this combined reaction:
 |
However, over the short time course of our experiments, it is possible that at least some of the sulfide could have been oxidized only to elemental S.
If sulfur-oxidizing bacteria used NO–3 to carry out DNRA, the N would remain in biologically available form and thus the NO–3 removal would not be permanent, in contrast to the denitrification reactions given above. Dannenberg et al. (1992) observed molar ratios of SO2–4 production to NO–3 consumption that were close to unity in cultures of Desulfovibrio desulfuricans that produced ammonium from NO–3 in conjunction with sulfide oxidation. DNRA is generally thought to be significant in highly reducing environments capable of maintaining sustained anaerobic metabolism (Tiedje, 1988), which might include the sediments of biologically productive wetlands. DNRA could be in direct competition with denitrification for NO–3, especially in anoxic saturated sediments (Nijburg et al., 1997). Even though the conditions for DNRA are similar to those for denitrification (reduced environment, available NO–3, and labile organic substrates), DNRA is thought to be favored in nitrate-limited, carbon-rich environments while denitrification is favored when carbon is limited in availability (Kelso et al., 1997; Silver et al., 2001).
Few studies have examined S-driven DNRA in freshwater systems. Freshwater wetlands are low in S compared to marine systems, but can contain enough to support significant S transformations, as has been demonstrated near our study sites (e.g., Lovley and Klug, 1983). Brunet and Garcia-Gil (1996) found that in the water column of a stratified lake, NH+4 production coincided with NO–3 and H2S depletion. Dannenberg et al. (1992) showed that freshwater strains of SO2–4 reducing bacteria were better able to perform S-driven DNRA than their marine counterparts. The recent profusion of 15N isotope studies in diverse ecosystems has piqued interest in DNRA as an important ecosystem process because 15N tracer flow from NO–3 to NH+4 has been observed (e.g., Bonin, 1996; Tobias et al., 2001b; Silver et al., 2001; An and Gardner, 2002), although these studies have not elucidated whether the apparent DNRA was linked to S transformations. In contrast, Yin et al. (2002) found that DNRA could not account for most of the NO–3 removal in freshwater sediments from rice paddies.
The difference between observed concentrations and those expected based on Br– concentrations indicates the stoichiometry of SO2–4 production compared to NO–3 removal (Fig. 3) . Expected concentrations were calculated as the product of the observed Br– concentration and the ratio of NO–3 or SO2–4 to Br– in the injection solution. Comparison of the mean differences between observed and expected concentrations of NO–3 and SO2–4 during the sampling period in which NO–3 was detectable in the porewater indicates the potential importance of S oxidation coupled to NO–3 removal (Table 5). Assuming the 1.6:1 stoichiometry for NO–3 removal to SO2–4 production in a denitrifying reaction, S oxidizers could account for 50 and 13% of the total NO–3 removal in the two experiments, respectively. If on the other hand the reaction is a form of DNRA with 1:1 stoichiometry, S oxidizers could account for 31 and 8% of the total NO–3 removal. These could be minimum estimates because of the possibility that some of the S oxidation does not proceed completely and appear as SO2–4 in the porewater. Further evaluation of NO–3 removal linked to S oxidation across a broader range of freshwaters is underway in our laboratory.
|
|
Further studies are needed to reach a conclusion about the role of bacterial sulfur oxidation in NO–3 removal, perhaps utilizing S and N isotopic tracers, but the temporal correspondence between NO–3 removal and SO2–4 production suggests that S bacteria could be significant in NO–3 uptake. If so, then the proximate controlling factors could be rather different than if respiratory denitrification were the principal process of NO–3 removal, and paradoxically, increased SO2–4 loading to wetlands could prove to be linked to larger populations of sulfur-transforming bacteria that would contribute to rapid removal of episodic NO–3 inputs. If, on the other hand, the observed SO2–4 production proves to be caused by competitive interactions between denitrifiers and SO2–4 reducers, then the increased denitrification caused by NO–3 pollution may indirectly increase SO2–4 pollution. Thus in either case there would be interesting linkages between N and S cycling that bear on water quality.
Variation in Nitrate Removal Rates among Wetlands
The precipitation- and ground water–fed wetlands had similar potential rates of NO–3 removal. This is unexpected since concentrations of NO–3 and SO2–4 are higher in local ground water than in precipitation, and two of these wetlands are primarily ground water–fed. However, NO–3 concentrations in surface-water and porewater samples were rarely above the analytical detection limit in the wetlands studied. An explanation for the consistently low ambient NO–3 concentrations is that NO–3 is efficiently removed from inflowing waters upstream of the sampling locations. Many studies in riparian areas have demonstrated that ground water entering the riparian zone can lose NO–3 by plant and microbial uptake over short spatial scales (Seitzinger, 1988; Simmons et al., 1992; Hill, 1996; Hedin et al., 1998). In one study of a coastal marsh system, NO–3 removal was 90% complete within the first 50 cm of the marsh near the upland boundary (Tobias et al., 2001b). Results from another study showed that a narrow riparian zone of a few meters width along a small stream was the most important location for NO–3 removal (Hedin et al., 1998). Thus, NO–3 might have been removed before water reached the more central sampling locations in this study. Episodic inputs of NO–3 via precipitation falling directly onto the wetland might also be subject to rapid removal, especially in wetlands with shallow water columns where contact with the sediments is greater. The finding that potential NO–3 removal rates are high where NO–3 concentrations are evidently low most or all of the time suggests that NO–3 is either produced and consumed in a tightly coupled way, that episodic inputs keep denitrifiers poised to utilize NO–3, or that sulfur-transforming bacteria can opportunistically switch to NO–3 utilization (as discussed above).
Processes and Rates of Sulfate Removal
Sulfate removal always followed NO–3 depletion, but the reaction kinetics differed depending on the water source for the wetland. Sulfate reduction via anaerobic respiration is the most likely process responsible for this removal, considering its delay relative to NO–3 and the fact that plant uptake has little impact on SO2–4 concentrations. Ground water–fed wetlands exhibited zero-order SO2–4 removal rates, indicating that rates were independent of SO2–4 concentrations, perhaps due to labile substrate limitation. Turkey Marsh had the highest rates of SO2–4 removal, although based on measurements during a previous year using 35SO2–4, this wetland did not have the highest ambient rates of SO2–4 reduction (Whitmire, 2003). Even though Turkey Marsh had higher rates, SO2–4 removal began about 10 h later there than it did in Loosestrife Fen (16.3 ± 6.1 h vs. 6.4 ± 1.8 h, P = 0.020). This is probably because NO–3 removal was faster in Loosestrife Fen, and SO2–4 removal only commenced after NO–3 was depleted.
There were no significant differences in potential SO2–4 removal rates among the precipitation-fed wetlands, with an average SO2–4 removal rate of 0.074 ± 0.04 h–1 (Table 3). This shows that SO2–4 reduction can take place in response to SO2–4 additions to precipitation-fed sites, even though ambient rates of SO2–4 reduction were not measurable using 35SO2–4 in these wetlands (Whitmire, 2003). Microbial sulfate-reducing populations could be maintained in these systems in several ways. First, the in situ mineralization of organic sulfur compounds could allow SO2–4 production and consumption to be tightly coupled (Fenchel et al., 1998). Second, SO2–4 reduction potentials may be high in the precipitation-fed wetlands due to episodic inputs by atmospheric deposition of SO2–4. Sulfate deposition is relatively high in southwestern Michigan; SO2–4 concentrations in precipitation at KBS average 2.7 mg L–1 (National Atmospheric Deposition Program/National Trends Network, 1997). Average SO2–4 concentrations in the surface waters of these wetlands vary but are generally higher (Table 1), and hence surface water could provide a source of SO2–4 to the sediments. Porewaters were always depleted in SO2–4 relative to the overlying surface waters, suggesting consumption in the sediments. Over longer periods, diffusion between overlying water and shallow subsurface sediments could be an important source of SO2–4 to the porewater environment, as has been found in shallow lakes (Kelly and Rudd, 1984).
Caveats Concerning the Bromide Tracer
Caveats concerning the use of Br– as a conservative tracer in push–pull experiments include its potential inhibitory effect on microbial processes (Groffman et al., 1995) and the possibility of sediment or plant uptake, which may confound results (Kung, 1990; Whitmer et al., 2000). Some inhibitory effects of Br– on microbial processes have been noted at high concentrations of Br– (100 mg L–1), but in the experiments reported here Br– was added at lower concentrations (<20 mg L–1 in most cases). Also, the observation that NO–3 depletion commenced immediately and that rates did not increase as the Br– was diluted suggests that Br– at the concentrations used in this study did not significantly inhibit microbial activity.
Sediment uptake of Br– was not observed when we incubated saturated sediments from several wetlands in sealed jars with added Br– (Table 4), and thus was unlikely to occur during the short-term experiments. Although the possibility of plant uptake of Br– (and NO–3 and SO2–4) cannot be excluded for all of the sites in this study, it is unlikely to be important. In all wetlands except Loosestrife Fen, about 50% of the Br– was recovered, and these sites span a wide range of plant densities, including sites with essentially no roots in the sediments. We selected areas that were relatively free of plants for the injection sites. Since background porewater concentrations of Br– were very low, it is more likely that dispersion (mixing with ambient porewaters) and diffusion away from the injection point explain the partial recovery. Advective loss was especially likely in Loosestrife Fen, where surface water flow was visible. Also, the potential NO–3 and SO2–4 removal rates observed in these wetlands were similar to those seen in contaminated aquifers (McGuire et al., 2002), where plant roots are not present and bacteria are the sole potential cause of biotic uptake or transformation.
Implications for Wetland Management
The rapid rates of NO–3 and SO2–4 removal demonstrate how very small areas of wetland sediment are capable of improving water quality, and such areas often occur at critical points of water flow between surface and ground water reservoirs. The ground water–fed wetlands have the potential to remove NO–3 in ground water discharge before it enters streams and lakes. Precipitation-fed wetlands are more isolated and have less influence on surface water quality, but they have the potential to process water entering as runoff or precipitation before it eventually infiltrates to ground water systems, where excessive NO–3 concentrations are a growing problem. Thus small isolated wetlands and riparian zones of headwater streams appear capable of significant water quality improvement, and if one goal of wetland protection is to preserve and enhance this ecosystem service, protection may need to be extended to these smaller wetland units.
This research underscores the need for further research on the microbial processes responsible for the NO–3 removal. The ultimate fate of the NO–3 that is removed remains uncertain because of the possibility that some is converted to ammonium rather than being denitrified, which would retain reactive N in the ecosystem. Seasonal variation in loading of NO–3 and SO2–4 as well as episodic inputs of these ions to wetlands may affect the nature and efficacy of microbial processes that remove them. In addition, the possible linkages of NO–3 removal with sulfur transformations may prove important as a mechanism to oxidize reduced sulfur compounds to SO2–4.
|
ACKNOWLEDGMENTS |
|---|
|
REFERENCES |
|---|
|
TOPABSTRACTINTRODUCTIONMATERIALS AND METHODSRESULTSDISCUSSIONREFERENCES |
|---|
Related articles in JEQ:
| |||||||||||||
and conservative (Br–) tracer concentrations in two push–pull experiments representing wetlands of contrasting hydrology: (A) ground water–fed Turkey Marsh and (B) precipitation-fed Lux Pond 10.
values, whereas first-order rates are expressed as k values. Values are means (standard deviations) for two sites in each wetland (total of 16 experiments).