The amount of excess fixed nitrogen removed from the freshwater aquatic nitrogen cycle, particularly by freshwater wetlands, through denitrification (DNF) is largely unknown. Typically, DNF rates increase within sediments that have higher organic content and a source of sufficient NO3−, in this context we measured DNF in organic-rich sediments of Lost Creek wetland on the south shore of Lake Superior, where NO3− concentrations have increased dramatically over the last century. The concentrations of N2, O2, and Ar were determined on intact water-sediment cores. Denitrification and respiration rates were determined using membrane inlet mass spectrometry and N2:Ar and O2:Ar ratios. Nitrogen flux rates measured in August 2000 and 2001 using overlying ambient wetland water, Lake Superior water, and nitrate augmented wetland water ranged from <10 to 78 μmol N m−2 h−1. These rates are low compared to those published for a variety of wetland and aquatic ecosystems. Nonetheless these are the first DNF measurements we know of to assess natural rates in the Lake Superior Basin and they help quantify a missing piece of wetland and lake nitrogen transformations.
Inputs of anthropogenic nitrogen to the Nation's aquatic ecosystems have increased substantially over the past century. Direct nitrogen loading to streams, wetlands, and lakes has also increased due to agricultural and land use practices, and industrial fixation of nitrogen for fertilizer production (Vitousek et al., 1997; Seitzinger et al., 2006). Nitrate concentrations in Lake Superior increased at a rate of about two percent per year, doubling about every 35 years (Bennett, 1986) for most of the 20th Century, but increases more recently have slowed (Sterner et al., 2007). To the coastal wetlands of Lake Superior, nitrogen rich lake water can also provide a source of NO3− to the mixing zone of these wetlands through seiche action (Trebitz et al., 2002; Morrice et al., 2004, 2009). Denitrification (DNF) plays an important role in the nitrogen cycle by NO3− reduction to gaseous dinitrogen; however, the amount of excess fixed nitrogen removed by DNF is an essential yet largely unknown piece of wetland chemistry (Davidson and Seitzinger, 2006).
The removal of nitrate from wetlands is important to the ambient water quality of wetlands and receiving waters. Wetlands can serve as a buffer region, preventing movement of NO3− from upland sources to the receiving water (Groffman, 1994; Seitzinger, 1994; Saunders and Kalff, 2001; Seitzinger et al., 2006). Lake Superior wetlands are typically low in NO3− (Morrice et al., 2004). They are however, influenced by both upland sources of NO3− through direct runoff and streamflow and by impinging lake water through lake seiche action. Denitrification rates have been measured in various aquatic ecosystem sediments for decades (Piña-Ochoa and Alvarez-Cobelas, 2006). However, earlier methods were difficult, time consuming, and often inaccurate (Kana et al., 1994; Groffman et al., 2006). The low nitrogen flux rates in sediment cores were difficult to measure, the ubiquitous nature of nitrogen compounds, and the need for considerable sample alteration and handling also limited earlier rate determinations. Although direct N2 flux measurements were made by Seitzinger et al. (1980), recent advances by Kana et al. (1994) using membrane inlet mass spectroscopy (MIMS) have allowed researchers to measure DNF rates more precisely, faster, and with considerably less sample alteration than previous techniques (Groffman et al., 2006).
The specific objective of this investigation was to estimate natural DNF rates in a coastal wetland bordering Lake Superior. This study and the data reported herein is part of a larger series of studies involving Great Lakes wetlands (Brazner et al., 2000; Trebitz, 2006; Trebitz et al., 2007) and a series of intensive studies on Lost Creek Wetland located on the south shore of Lake Superior in Bayfield County, WI, USA (Morrice et al., 2004, 2009; Trebitz et al., 2005). An estimate of the DNF rate in an environment influenced by both upland and receiving lake sources of nutrients and water is unique, and would allow us to assess the importance of DNF to the wetland nitrogen cycle and the potential influence these rates may have on the nearshore waters of Lake Superior.
Materials and methods
Lost Creek wetland is located on the south shore of the western arm of Lake Superior (Figure 1, Lat/Long 46.86N, 91.14W). It is a dendritic wetland surrounded by a Laurentian mixed forest type. The total wetland area is 100 ha (Minc and Albert, 1998) with an open water area estimated at about 6.4 ha. The maximum and mean depths are 2.1 and 0.9 m, respectively. Further detailed descriptions of Lost Creek wetland can be found in Trebitz et al. (2005) and Morrice et al. (2004). The lower region of Lost Creek wetland is a dynamic environment. Lake Superior water impinges through seiche action on an average 7.9 hour cycle with amplitudes ranging from 1.3–14 cm (Trebitz et al., 2002; Morrice et al., 2004). This seiche action brings in fresh nitrogen-rich lake water and provides a well oxygenated zone in the approximate lower one third of the wetland. In periods of low water, the wetland is occasionally completely cut off from the lake by a sand bar during which the lake source of nutrients would temporarily stop. In other periods, as in late summer, lower than normal lake water levels can prevent lake water intrusion. Morrice et al. (2004) found that dissolved inorganic nitrogen (DIN) ranged from 281 μg l−1 in June of 1998 to 349 μg l−1 in April of 1999 in the nearshore waters of Lake Superior. In contrast, mid-wetland concentrations ranged from 7 μg l−1 in August 1998 to 30 μg l−1 in June 1998. The primary criteria for choosing the sediment DNF sites was the absence or presence of seiche influence at the site, the concentration of NO3−, and differences in sediment organic carbon. Site 485 in the upper wetland was far more depositional than the lower wetland site 052, which was highly influenced by Lake Superior and thus more consolidated and sandy in nature (Table 1). Briefly, the total volatile solids of the two wetland sites (485 and 052) were 21.1 and 7.5%, respectively. Grain size was considerably smaller in the upper wetland site where 47% was less than 64 μm as opposed to the lower site where 65% of the grain size was greater than 250 μm.
Intact sediment cores were taken from Lost Creek wetland at sites 485 and 052 in August 2000. The first year served as a pilot study to determine where to focus our efforts in the second year. As a result, sample site 052 was the only site sampled the second year. The DNF determination in 2000 consisted of four intact sediment cores which were collected from Lost Creek wetland along with overlying water using plexiglass core tubes (8.25 × 30 cm). The core tubes were fitted with gas-tight end caps, internal floating stir bars, and two gas-tight sampling ports. Two cores along with the overlying water were collected near the mouth of the wetland in the zone influenced by the Lake Superior seiche (site 052) and more consolidated sediments. A second set of two cores were collected in the upper region of the wetland (site 485) containing highly organic sediments and not influenced by lake water intrusion but subject to upland sources of nutrients. Additional ambient site water from the two sites was collected for core equilibration and sample make-up water in the laboratory. To collect the intact core, each core tube was pressed by hand into the sediment to half the core depth (15 cm), gently extracted, and capped under water to keep the cores as undisturbed as possible. Entrapped air bubbles were flushed out as the top cap was placed onto the cores. The intact sediment cores were then placed into an insulated container to maintain ambient field temperature during transport. At the laboratory, each core was uncapped and submerged for 24–36 hours in tanks containing wetland water in an environmental chamber set at the ambient wetland water temperature (17°C) at the time the cores were collected.
The DNF determination in August 2001 consisted of six intact sediment cores collected in the same manner from the lower wetland site only (052). The six cores were used for two separate experiments consisting of four cores each. Experiment 1 consisted of duplicate cores equilibrated and DNF rate determined using ambient wetland water overlying the sediment at the time of collection. A second set of duplicate cores were exposed to ambient wetland water augmented with NO3− added as KNO3. The amendments were to mimic NO3− intrusion with seiche but with elevated overlying NO3− levels 3–4 times present levels in Lake Superior (25–28 μmol) (Sterner et al., 2007; Kelly et al., 2011). The resulting mean NO3− concentrations during the 24 h experiment were 1.20 μmol l−1 for unaltered overlying wetland water cores 1 and 2, and 95.8 μmol l−1 for augmented cores 3 and 4. Experiment 2 consisted of cores 1 and 2 repeated after an open re-equilibration with wetland water, and cores 5 and 6 which were equilibrated using nearshore Lake Superior water, which resulted in a mean NO3− exposure concentration of 14.5 μmol l−1 during the experiment. The NO3− concentration considering all core experiments ranged from 1.20–95.8 μmol l−1.
After the 24–36 h equilibration period, each core was capped under water and carefully inspected for air entrapment, a serious condition that must be avoided to determine a valid rate. Cores containing air bubbles were recapped until all micro air bubbles were eliminated. Cores were then returned to the environmental chamber and the inductive stirring apparatus started at a stir rate of about 60 rpm. Water samples for gas analysis were taken from each core by displacement using make-up water identical in composition to the initial conditions in each respective core. The make-up water was stored in separate glass-stoppered bottles and sampled for gas analysis on the same schedule as each core above. A total of 45 ml of water was displaced from each core with make-up water at each sample time. For gas analysis, 7−ml samples were taken in triplicate using glass vials with ground glass stoppers or Teflon lined screw caps at 0, 2.5, 5, 7.5, 12 and 24 h in 2000, and at 0, 4, 8, 12, and 24 hours in 2001. Each sample vial was purposely allowed to overflow during sampling to minimize the effects of the dry vial and air entrapment on the final gas concentration. The sample vials were immediately stored under water at the test temperature until they were shipped for analysis. Sample vials remained under water in a secondary glass container and were sent via overnight delivery to the analytical laboratory after 8 hours and again after 24 h. The NO3− and NH4+ were analyzed at the beginning and end of each core experimental period.
Gas analysis and calculations
The core DNF samples were analyzed for N2, O2, and Ar using membrane inlet mass spectrometer (MIMS) at the University of Maryland, Horn Point Laboratory, Cambridge, MD, according to procedures described by Kana et al. (1994). The method analytical precision is <0.05%. The DNF rate (μmol N m−2 h−1) was calculated for each core, after correcting concentrations for make-up water addition, using the formula F = (d[N2]/dT)V/A, where F = Flux rate (μmol N2 m−2 h−1), d[N2]/dT = slope of the rate plot (μmol l−1 h−1), V = core water volume (l), and A = core sediment surface area (m2). The respiration rate (μmol O2 m−2 h−1) was calculated in the same manner using the slope of the O2 rate plot. Nitrate loss rates were calculated based on the difference in NO3− concentration at 0 and 24 h, also normalized to the sediment surface area of each core as above. Core water and sediment could be contributing to the measured flux rates; although typically in these types of experiments the water's metabolic contribution to overall flux rates is minor. It was assumed for this study that most of the contribution was due to the sediment and no attempts were made to separate the possible contribution of the overlying water.
Water and sediment analysis
The sediment from each wetland site was analyzed for grain size and percent volatile solids. Bulk sediment samples were weighed, sieved through a series of stacked sieves; the contents of each sieve were removed, dried, and weighed. Fractions are expressed as percent dry weight. Extensive nutrient analysis was conducted on Lost Creek Wetland, including years before, during, and after this investigation. A thorough description of water chemistry methods and is in Morrice et al. (2004).
Year 1: Upper and lower wetland comparison (2000)
Comparing the upper (485) and lower (052) wetland sites, DNF was only detected in the lower site near the wetland mouth. The DNF rate at sites 485 and 052 ranged from below detection (approximately 10 μmol N m−2 h−1) to 78 μmol N m−2 h−1 with a mean of 60 μmol N m−2 h−1 (n = 2, site 052). The mean overlying core water NO3− concentration was significantly different (p < 0.05) during the experiment with mean concentrations at sites 485 and 052 of 0.90 and 3.93 μmol l−1, respectively. The respiration rates for these same sites were variable and ranged from 897 to 2260 μmol O2 m−2 h−1 (n = 4) with a mean respiration rate at each site of 1038 and 1860 μmol O2 m−2 h−1, respectively, and not significantly different from each other (Table 2).
Year 2: Nitrate augmentation (2001)
The second experiment was done at the lower wetland site (052) only and focused on the effect of a nitrate gradient in the core overlying water and the possible effect of the Lake Superior seiche. Three nitrate levels in the overlying core water was achieved by using (1) ambient overlying water from the wetland site, (2) nearshore Lake Superior water, and (3) ambient overlying water augmented with NO3−. The resulting mean NO3− concentrations for each overlying water source were 1.19 ± 0.41 μmol l−1 (n = 4), 14.5 ± 0.14 μmol l−1 (n = 2), and 95.8 ± 0.78 μmol l−1 (n = 2), respectively. Denitrification rates of 50 and 68 μmol N m−2 h−1 were measured on the NO3− augmented cores only. Denitrification in the sediment cores containing wetland water and Lake Superior water were both below measurable rates (<10 μmol N m−2 h−1). There was a significant difference in the NO3− concentrations (p < 0.05) between the three core water types but there was no significant difference in the corresponding rates (Table 2). The respiration rates in all the cores tested at the lower site (052) during the second year were very consistent and not significantly different from each other. They ranged from 666 to 863 μmol O2 m−2 h−1 with a mean of rate of 785 ± 72.9 (± SD, n = 8). These respiration rates were, however, significantly lower than the previous year at this same site (1860 ± 566, n = 2).
Low DNF rates were detected on the lower wetland site (052) in the first and second years (the second year only with NO3− augmentation). These are the first natural DNF rates reported for this kind of Great Lakes environment that we know of. At the lower site, the region of seiche influence with a more consolidated type of sediment containing lower volatile organic carbon, the overlying NO3− was higher. The DNF rate in the first set of cores in 2000 at site 052 was 78 and 42 μmol N m−2 h−1 with moderate NO3− levels of 3.9 μmol l−1. An overall ONF-NO3− relationship was not evident; in the second experiment where DNF remained undetected at NO3− concentrations of 14.5 μmol l−1 and was only measurable at NO3− levels far greater (95 μmol l−1). Johnston at al. (2001) studied two freshwater riverine wetlands of the St. Louis River, a main tributary of Lake Superior, and found DNF was consistently NO3− limited. Potential rates determined using sediment slurries were below detection in unamended samples, however, NO3− amended samples greatly stimulated DNF. Amended DNF rates were significantly higher in the summer samples, when temperatures are highest, than samples collected in the spring. Several other variables are likely influencing the rate such as frequency and magnitude of NO3− exposure, active nitrifying organisms, organic matter, interstitial carbon, oxygen, plants (Groffman, 1994; Nowicki et al., 1997; van Luijn et al., 1999; Cornwell et al.,1999; Wallenstein et al., 2006), and other competing NO3− removal pathways such as anaerobic ammonium oxidation (anammox) and iron-driven DNF (Burgin and Hamilton, 2007) which were beyond the scope of this investigation.
There is much interest in DNF in recent literature as confidence in the various methods and their ability to measure low rates has improved (Groffman et al., 2006). Interest has also grown tremendously in understanding the capacity of natural systems to remove increasing inputs of reactive N introduced by anthropogenic activities (Vitousek et al., 1997; Hansson et al., 2005; Sierszen et al., accepted). We used an accepted, sensitive method and had difficultly detecting significant rates even at summer temperatures, when rates are usually at maxima in most systems (Nowicki et al., 1997; Johnston et al., 2001; Golterman, 2004; Piña-Ochoa and Álvarez-Cobelas, 2006). The method itself is labor and equipment intensive and requires significant care with sampling and analysis, as with most accurate DNF rate measurements. Though few, the rates measured are worth reporting here since we have no previous data for the environment we are considering. We have sufficient confidence in our analytical methodology employed to estimate some initial bounds on the role of DNF in this type of habitat.
Comparison to other environments and aquatic systems
The rates we quantified were in the low range reported in the literature for most other environments (Figure 2). The rates we were able to detect were in instances using wetland water with generally higher NO3− concentrations, some amended. Interestingly, though, use of higher NO3− water taken directly from Lake Superior did not stimulate a detectable rate.
It is perhaps not surprising that rates are low, since we studied a relatively undeveloped, unenriched coastal wetland system adjacent to one of the more oligotrophic lakes in the world. Our data compare favorably to that of open oceans which are also oligotrophic systems (Figure 2). The watershed setting for this particular coastal wetland is not unusual for most in the Lake Superior basin (Trebitz et al., 2002; Morrice et al., 2004, 2009). Summer dissolved oxygen levels in the wetland can be low in backwater pockets (Trebitz et al., 2005), but in general the setting is one in which most other factors known to promote higher DNF would not appear optimal. Examples of suboptimal factors at Lost Creek include; short water residence time (Nixon et al., 1996; Kelly, 1997; Nowicki et al., 1997; Morrice et al., 2004), summer maximum temperatures are not especially high and do not persist long, sediment organic content probably is not very labile and originates from vascular/macrophytic decay rather than fresh phytoplankton, NO3− is lower than adjacent Lake Superior (Morrice et al., 2004), and aquatic bottom sediments do not support active bioturbating benthic communities which are known to enhance nitrification/denitrification through introduction of oxygen in burrows (Golterman, 2004; Cornwell et al., 1999). Macrophyte metabolism may also pump oxygen into sediments, but the size of our sediment core experiments restricted us to the more open water areas of the wetland and did not allow for measurements in submerged or emergent macrophyte beds.
Denitrification rates in an ecosystem context
The reported low rates here should be viewed in two contexts, the wetland scale and the larger basin wide scale of Lake Superior. At the wetland scale, Morrice et al. (2004) observed NO3− uptake where seiche water mixes with and brings higher NO3− waters from Lake Superior (the lower wetland area of this study, Figure 1). This phenomenon had initially stimulated these studies. Denitrification, at rates we measured, apparently plays a small but perhaps significant role in this particular process (perhaps 5–20% of the wetland's retention in that dynamic area; Morrice et al., 2004).
At a more basin-wide scale of Lake Superior, where nitrate concentrations have long been increasing (Bennett, 1986; Sterner et al., 2007), the role of DNF is unknown. We can make some coarse but interesting calculations with respect to the potential role of coastal wetlands. If we apply our “maximum” mean rate of 60 μmol N m−2 h−1 to the open water area of all wetlands in the Lake Superior basin (Brazner et al., 2000) for three warm months, and assume only about 10% of that is open water area (Lost Creek is ∼6%), coastal wetlands could denitrify 3.34 × 109 mmol N. We would not expect the cold months to contribute much more in a year (Nowicki et al., 1997; Johnston et al., 2001; Piña-Ochoa and Álvarez-Cobelas, 2006). Compared to an estimated lower bound for annual N input to Lake Superior (Sterner et al., 2007), one calculates that open waters in coastal wetlands during the warm months remove, through DNF in submerged sediments, an amount <0.1% of the annual N input from watershed delivery and even less when compared to all inputs (watersheds and air) to Lake Superior. The suggested coastal wetland DNF role in removal and retention is very small.
There is a second lake-basin perspective. Lake Superior's nitrate concentration has been increasing (Bennett, 1986) and although the rate of increase has recently slowed, the increase from 1973 to 2005-2006 is about ∼5–6 μmol as a lake wide average (Sterner et al., 2007; Kelly et al., 2011). If we assume that about 10% of the open water of all coastal wetland has a mixing zone which receives and removes NO3− at the summer rate calculated above (likely a significant overestimate), the removal of NO3− from the lake by this process would represent <0.02% of the annual rate of lake wide NO3− increase. Thus, there is virtually no capacity to moderate lake NO3−trends through interaction with fringing coastal wetlands. Sediments in the body of the lake, by virtue of their vastly larger area, probably have more impact, but we have no measurements reported for them and considering they are regularly at much colder temperatures (most of the lake bottom is 4°C year-round), we suspect rates are even more difficult to detect and quantify.
The DNF rates we measured were low, and to our knowledge, the first measurement of natural rates in wetlands within the Lake Superior Basin. Even employing sensitive methodology, rates were often below detection. The DNF rates measured, however, are consistent with potential rates determined using sediment slurry techniques on a similar riverine system within the western arm of Lake Superior. This study provides greater insight in the difficulty of determining DNF rates and in the magnitude of DNF in NO3− limited systems such as Lost Creek. Rates were measurable only in the lower region of the wetland near the outlet to Lake Superior which is occasionally supplied with higher NO3− concentrations. Whether this supply of NO3− stimulated greater DNF rates is inconclusive and requires further investigation. Augmentation of core water with nitrate to levels above normal wetland concentrations did stimulate DNF but when all the core samples were considered, there was no apparent relationship between NO3− concentration and DNF rates. We suspect that there are a considerable number of variables influencing both the occurrence and the magnitude of DNF in Lost Creek wetland.
When considering the greater whole of the Lake Superior NO3− balance, inferences can be made on a much broader scale. Given the low rates, small wetland area, and an even smaller wetland area influenced by lake water through seiche action, calculations indicate the basin wide impact would be minimal. More extensive studies of coastal wetland DNF would be needed to determine if periodic inputs of NO3− are enough to stimulate DNF and thus become a viable pathway for removing NO3− from the system. It is unlikely that DNF would play a significant role in mediating nitrate levels in the adjacent nearshore waters of Lake Superior.
We thank Todd Kana and staff at Horn Point Laboratory, University of Maryland, for advice, instruction, and analysis of core water samples for dissolved gasses; Anne Cotter for the nutrient analysis of core water and wetland samples; and Corlis West for the sediment characterization. Thanks to Brian Hill and MaryAnn Starus for reviewing and commenting on this manuscript. This work was wholly funded by the U.S. Environmental Protection Agency and has been approved for publication after review by EPA's National Health and Environmental Effects Research laboratory. The contents do not necessarily reflect the views of the agency, nor does mention of commercial products constitute endorsement or recommendation for use.
This article not subject to United States copyright law.