Corresponding author: R. S. Stelzer, Department of Biology and Microbiology, University of Wisconsin-Oshkosh, 800 Algoma Blvd., Oshkosh, WI 54901, USA. (firstname.lastname@example.org)
 Many estimates of nitrogen removal in streams and watersheds do not include or account for nitrate removal in deep sediments, particularly in gaining streams. We developed and tested a conceptual model for nitrate removal in deep sediments in a nitrogen-rich river network. The model predicts that oxic, nitrate-rich groundwater will become depleted in nitrate as groundwater upwelling through sediments encounters a zone that contains buried particulate organic carbon, which promotes redox conditions favorable for nitrate removal. We tested the model at eight sites in upwelling reaches of lotic ecosystems in the Waupaca River Watershed that varied by three orders of magnitude in groundwater nitrate concentration. We measured denitrification potential in sediment core sections to 30 cm and developed vertical nitrate profiles to a depth of about 1 m with peepers and piezometer nests. Denitrification potential was higher on average in shallower core sections. However, core sections deeper than 5 cm accounted for 70% on average of the depth-integrated denitrification potential. Denitrification potential increased linearly with groundwater nitrate concentration up to 2 mg NO3-N/L, but the relationship broke down at higher concentrations (>5 mg NO3-N/L), a pattern that suggests nitrate saturation. At most sites groundwater nitrate declined from high concentrations at depth to much lower concentrations prior to discharge into the surface water. The profiles suggested that nitrate removal occurred at sediment depths between 20 and 40 cm. Dissolved oxygen concentrations were much higher in deep sediments than in pore water at 5 cm sediment depth at most locations. The substantial denitrification potential in deep sediments coupled with the declines in nitrate and dissolved oxygen concentrations in upwelling groundwater suggest that our conceptual model for nitrate removal in deep sediments is applicable to this river network. Our results suggest that nitrate removal rates can be high in deep sediments of upwelling stream reaches, which may have implications for efforts to understand and quantify nitrogen transport and removal at larger scales.
1.1. Conceptual Model for Nitrate Removal in Deep Sediments
 We showed in a previous study [Stelzer et al., 2011b] that nitrate removal occurred in deep sediments in Emmons Creek, a gaining stream in the Central Sand Ridge Ecoregion in Wisconsin. Based on this work and previous studies demonstrating nitrate loss along upwelling flow paths in stream sediments [Duff et al., 2008; Puckett et al., 2008; Krause et al., 2009] and redox-driven nitrogen transformations in riparian soils [Hedin et al., 1998] we developed a conceptual model for nitrate removal in deep sediments of groundwater-fed streams (Figure 1). The model illustrates nitrate-rich oxic groundwater upwelling through deep sediments before discharging to the surface water of a stream. Deposition and burial of POM (including fine and coarse particulate organic matter) in sediments [Metzler and Smock, 1990] creates a zone with favorable redox conditions (severe hypoxia or anoxia) for nitrate to serve as the terminal electron acceptor during microbial respiration, facilitating denitrification. POM and the presence of various metals [Puckett et al., 2008] serve as electron donors. As the groundwater encounters this zone, denitrification, and perhaps other nitrate removal processes, causes a decrease in nitrate concentration along the flow path. We think the depth and thickness of this zone will vary within and among streams and will be affected by vertical hydraulic gradient and sediment porosity [Triska et al., 1989]. This zone with favorable redox conditions for denitrification may occur below the depth at which surface water and groundwater mix (i.e., hyporheic zone), as shown in Figure 1, or it could encompass all or part of the hyporheic zone. This model has some similarities with previous conceptual models that have emphasized groundwater-surface water interactions in streams [Triska et al., 1989; Vervier et al., 1992; Krause et al., 2009], but is different than most previous conceptual models of hyporheic zone processes, which have focused on transport of surface water into the hyporheic zone and subsequent return to the channel [Jones and Holmes, 1996; Boulton et al., 1998; Mulholland and DeAngelis, 2000; Boulton et al., 2010]. The notion that changes in redox state can drive nitrogen transformation has also been applied in many previous studies of nitrogen loss in groundwater systems at the watershed scale [e.g., Böhlke and Denver, 1995].
 We developed three hypotheses from this conceptual model: (1) Nitrate concentration of groundwater will decline along upwelling flow paths before discharging to surface water, (2) denitrification potential in deep stream sediments will substantially contribute to total (depth-integrated) denitrification potential in sediments, and (3) dissolved oxygen concentration will decline along upwelling flow paths as groundwater passes through the zone of anoxia.
 We tested these hypotheses by developing vertical nitrate profiles to about 1 m depth, by measuring denitrification potential at sediment depths to 30 cm, and by comparing dissolved oxygen (as a proxy for redox state) between deep groundwater and pore water at 5 cm sediment depth at eight sites in streams and rivers in the Waupaca River Network in Central Wisconsin. Our primary goal was to determine if our conceptual model for nitrate removal in deep stream sediments was applicable throughout a river network. Our hypotheses and conceptual model were broadly supported. In particular, we found there was substantial denitrification occurring at depth and most streams showed a loss of nitrate as groundwater upwelled through sediments. Changes in dissolved oxygen between deep groundwater and pore water at 5 cm were consistent with redox-driven nitrate loss.
2.1. Waupaca River Watershed
 The Waupaca River Watershed (WRW) is 790 km2 and is located in Central Wisconsin in the Central Sand Ridges Ecoregion, which is part of the Lake Michigan Drainage Basin. The watershed has mixed land uses, dominated by agriculture (55% of land area) and forest (35%) with a small amount of wetlands (5%) [Lin et al., 2001]. The terrain is flat to gently rolling and was influenced by glaciation. Sandy soils are widespread in the watershed as are pockets of gravel and boulders [Guldan, 2004]. The climate is humid continental with about 80 cm of annual precipitation [Holt, 1965]. Lotic systems in the WRW are base flow dominated [Arnold and Allen, 1999] and most of the streams and rivers maintain sufficiently cool temperatures throughout the year to support trout. Streams in the WRW are low gradient and fine sediment (sand and silt) is probably the dominant sediment type. Coarser sediments (gravel, cobbles, boulders) are common in riffles. Eight study sites on streams and rivers in the Waupaca River Watershed (Figure 2 and Table 1) were chosen using the following set of criteria: (1) Upwelling was the prevailing direction of water movement in the sediments and (2) fine sediments were present to facilitate the insertion of the peepers (see below). Sites ranged in stream order from 2nd to 4th. Groundwater nitrate concentrations spanned three orders of magnitude (<0.01 to 9 mg NO3-N/L), while surface nitrate concentrations were much less variable among streams (1.8 to 4.6 mg NO3-N/L) (Table 1).
Table 1. Location, Order, Mean Solute Concentrations, and Mean Vertical Hydraulic Gradient (VHG) for Study Sites in the Waupaca River Networka
Longitude, Latitude (deg)
Deep Groundwater (mg/L)
Surface Water (mg/L)
Deep groundwater was sampled from piezometers (means based on N = 18 per stream in most cases) at an average depth of 59 cm. Surface water was sampled in the midchannel from each stream (means are based on samples collected from 3 to 4 different dates, in most cases, during the study period).
Tomorrow R. I
Tomorrow R. II
2.2. Piezometer Nests, Peepers, and Solute Profiles
 Three piezometer nests, each consisting of six piezometers arranged in a semicircle with a radius of about 40 cm, were installed at each site in locations separated by several meters or more. Piezometers were constructed of CPVC (1.2 cm inner diameter) with the terminal 4.5 cm screened (3 mm holes covered with 100 μm Nitex mesh). Modified Pore Water Hesslein Samplers (peepers; 47 cm long × 10.5 cm wide, Rickly Hydrological, Columbus, Ohio, U.S.A.) were deployed within each piezometer nest at each study site. The piezometers were installed at different depths within each nest (grand mean depth to midscreen: 59 cm; mean minimum depth: 33 cm, mean maximum depth: 88 cm; depths among different nest locations were roughly equivalent) so that the nitrate and chloride concentrations in relatively deep groundwater could be characterized, while the peepers provided solute concentrations at 1.3 cm vertical intervals in the 1 to 35 cm range.
 Peepers were prepared for deployment in the lab by equilibration in degassed deionized water for 24 h. After fitting nylon membranes (Biodyne A, 0.2 μM) over the deionized water-filled cells while submerged, peepers were immediately transported to the field in degassed deionized water and deployed. Each peeper was pushed as far as possible into the sediment, while usually maintaining at least a couple of cells above the sediment. Cells below the sediment were used to sample groundwater (or a mixture of groundwater and surface water where infiltration occurred) and those above the sediment were used to sample surface water. Peepers were sampled 26 to 28 days after deployment for solute concentrations with sterile 10 mL syringes fitted with 18 gauge needles. For cases in which particles were observed inside a peeper cell, the sample was discarded. Groundwater from piezometers was sampled for solute concentrations midway through each peeper incubation with 60 mL plastic syringes attached to polypropylene tubing (0.6 mm inner diameter) and filtered through Whatman GF/F filters in the field. Immediately prior to sampling, groundwater from each piezometer was purged with a volume approximately equal to the piezometer volume. Samples from peepers and piezometers were stored in polyethylene bottles and placed at −20°C until analysis of nitrate, chloride, and sulfate concentrations. Water samples collected from the piezometers and peepers were used to create high-resolution vertical profiles of nitrate and chloride concentrations in the sediments to about 75 to 95 cm.
2.3. Denitrification Potential
 Four to five sediment cores (7.6 cm diameter, 25 to 30 cm length in most cases) were collected in upwelling locations at each site for measurements of denitrification potential (see Discussion for why we considered the rates measures of denitrification potential) from late spring through early fall in 2010. In most cases cores were collected during peeper deployment periods (at Hartman Cr. cores were collected one month after the peeper incubation period). Cores were collected in or adjacent to the piezometer nests. Each core was divided into 5 cm sections and placed in Whirl-Pak bags for transport to the laboratory. Within 24 to 48 h denitrification potential was measured using the chloramphenicol-amended acetylene block method [Richardson et al., 2004; Groffman et al., 2006] in the laboratory at University of Wisconsin-Oshkosh. The 90 min incubations (and core storage prior to incubation) were carried out in a Fisher Isotemp Model 307 C incubator set to the ambient temperature of groundwater at the time of sediment core collection. Core sections were homogenized immediately prior to the denitrification incubations. Twenty-five milliliters of sediment, 20 mL of groundwater, and 5 mL of chloramphenicol solution (1 mg mL−1) were added to vessels (246 mL glass canning jars) fitted with gray butyl septa. Groundwater added to the vessels was collected from a piezometer in the vicinity of where the sediment core was collected. A sequence of evacuation by vacuum (−25 psi) and helium addition to the head space of the jars (repeated two times) was used to create anoxic conditions in the vessels. Immediately after addition of 20 mL of atomic absorption-grade acetylene (time zero), vessels were placed on a shaker (Innova Model 2000) at 175 rpm in the incubator. Head space gas was sampled with a 5 mL syringe fitted with a 21 gauge needle at 30 min intervals during the 90 min incubations and was immediately transferred to evacuated 2 mL serum vials.
 Within one week of the incubations vials were shipped to the Upper Midwest Environmental Sciences Center (UMESC) in La Crosse, WI for analysis of N2O concentration on a Hewlett Packard Model 5890 gas chromatograph fitted with a Porapak R 3.18 mm diameter 2 m stainless steel column and a 63Ni electron capture detector (ECD). Storage standards were produced by injecting certified concentrations of N2O into evacuated serum vials at the UMESC on the same days that denitrification incubations occurred at UW-Oshkosh. N2O concentration in the head space was converted to total N2O (water plus gas phase) using the equations in Groffman et al. . Denitrification potential was calculated as the rate of N2O production, based on simple linear regression models fit to the N2O time series, and reported by core section and by entire core (depth-integrated) on an areal basis. Subsamples of sediments from each core section were collected and stored at −20°C for percent organic matter analysis. Subsamples from groundwater used in the incubations were filtered (Whatman GF/F), and stored at −20°C for nitrate and dissolved organic carbon (DOC) analysis.
 Nitrate-amended denitrification incubations were run in tandem with the unamended incubations described above for two sites, Hartman Cr. and the Waupaca R., that had very low concentrations of groundwater nitrate (Table 1). Nitrate was added as sodium nitrate to the vessels, immediately after sediment addition, to achieve a final concentration of about 2 mg NO3-N/L. In all other respects the nitrate-amended and unamended incubations were treated identically.
2.4. Dissolved Oxygen
 Dissolved oxygen was measured in the deep groundwater and in the pore water at 5 cm sediment depth when groundwater samples were collected from piezometers. Dissolved oxygen was measured in the deep groundwater by pumping water from one piezometer from each nest (the deepest piezometer in most cases) with a peristaltic pump (Global Water) and sending it through a flow cell in which a combined dissolved oxygen and temperature probe (YSI 85) was inserted. Dissolved oxygen in the pore water was measured at 5 cm sediment depth in each piezometer nest with a Microelectrodes dissolved oxygen electrode that was pushed into the sediment. The YSI oxygen probe was air-calibrated in the field and the Microelectrodes probe was calibrated with 0 and 100% O2-saturated water in the lab at the temperatures of ambient sediment. Pore water temperature at 5 cm sediment depth was measured with a handheld thermometer.
2.5. Vertical Hydraulic Gradient
 Vertical hydraulic gradient (VHG) was measured twice at each piezometer; once at peeper deployment or a few days prior and a second time when the piezometers were sampled for solutes (VHG was measured on only one occasion in Emmons Creek). The static head was measured with a Solinst level tape and vertical hydraulic gradient was calculated as the difference between the static head height and the stream water height divided by the distance from the top of the sediments to the midscreen of the piezometer [Dahm et al., 2006].
2.6. Solute and Particulate Organic Matter Analysis
 Nitrate, chloride, and sulfate concentrations were measured using a Dionex ICS-1000 ion chromatograph equipped with an IonPac AS14A column. Dissolved organic carbon was analyzed on an O.I. Analytical Model 1030 W TOC analyzer. The percent organic matter of sediments (as dry weight) was calculated based on mass loss after combusting samples, previously dried at 60°C, at 500°C.
2.7. Statistical Analysis
 Nitrate and chloride concentration profiles were used to calculate NO3-N/Cl−ratios and unpaired one-tailedt tests were used to compare NO3-N/Cl− ratios between deep groundwater (from 6 piezometers) and shallower groundwater (from the 6 deepest peeper cells) at each piezometer nest. We compared NO3-N/Cl− ratios instead of NO3-N concentrations in attempt to account for hydrological causes (e.g., dilution) of differences in NO3-N concentrations between deep groundwater and shallower groundwater. Nested two-way ANOVA was used to determine if mean denitrification potential and % organic matter of sediment differed among the eight sites and among core sections (nested within site). Nested two-way ANOVA was used to assess if mean dissolved oxygen differed among sites and between deep and shallow sediment depths (depth was nested within site). Least squares linear regression was used to determine how core section depth affected denitrification potential at each site and across all sites and percent organic matter across all sites. Pairedttests were used to determine if denitrification potential differed between the nitrate-amended and unamended sediments. One-way ANOVA was used to determine if VHG and DOC concentration differed among sites. Statistical analyses were performed using Systat v. 13.
 The vertical hydraulic gradient data and the chloride profiles both suggest that upwelling was the prevailing direction of water movement through the sediments. Mean VHG ranged from 0 to 0.202 and was consistently positive at 7 of the 8 sites (Table 1). VHG differed among sites (ANOVA P < 0.001) and was highest at Radley Creek. In most cases the chloride concentrations of deep and shallower groundwater were similar and both were distinct from surface water chloride concentrations (Figure 3). This pattern is expected if the dominant direction of water movement through the sediments is upwelling as chloride tends to behave conservatively. In most streams, the chloride data suggested that surface water did not penetrate the sediments beyond about 5 to 10 cm (Figures 3a, 3b, 3d, 3f–3i, 3l, 3m, 3q, and 3r). However, there were several cases in which the chloride concentrations of the shallower groundwater were in between the concentrations of surface water and deep groundwater and the data suggested classic two-end-member mixing (Figures 3j and 3t–3x). In Hartman Cr. and the Waupaca River, in particular, the chloride profiles suggested that there was surface water influence at sediment depths of 20 to 40 cm. VHG was positive but relatively low at these two sites (Table 1). There were a couple of locations at which there was too much variation in the chloride profiles to assess the extent of surface water and groundwater mixing (Figures 3o and 3p).
3.2. Nitrate Profiles
 For most of the study sites nitrate concentration was higher in deep groundwater than in shallower groundwater (Figure 3). At the Tomorrow River Sites I and II, Bear Cr., Emmons Cr. and the Crystal River nitrate concentration tended to decline to very low concentrations (at or below the detection limit) as groundwater moved from deeper to shallower sediments (e.g., Figures 3a–3g, 3i, 3l, 3m, 3o, and 3q). The profiles suggest that at most locations nitrate decline occurred at relatively deep depths in the sediment (20 to 40 cm), below the hyporheic zone based on the chloride profiles. At two piezometer nest locations (Emmons Cr. Nest 2 and Radley Cr. Nest 3) groundwater nitrate concentration remained high as water moved from deeper to shallower sediments (Figures 3h and 3r). At one nest location at Tomorrow River Site II (Figure 3n) chloride and nitrate concentrations were both higher in the deep groundwater than in the shallow groundwater. This may reflect that different groundwater flow paths were sampled by the peepers and piezometers at this location. Finally, all piezometer nest locations at Hartman Cr. and the Waupaca R. revealed nitrate concentrations at or below the detection limit for both deep and shallower groundwater. The ratio of NO3-N/Cl−was lower in shallower groundwater than in deep groundwater at 14 of 18 of the locations in which the nitrate concentration in the deep groundwater was above the detection limit (one-tailed unpairedt tests, P < 0.05).
 Mean denitrification potential differed among core sections (ANOVA P < 0.01) and decreased with increasing core section depth (linear regression P < 0.01) (Table 2). However, core sections deeper than 5 cm accounted for 70%, on average, of depth-integrated denitrification potential. The magnitude of denitrification potential differed strongly among sites (ANOVA P < 0.001,Figure 4). There was high variation in denitrification potential among cores at most sites (note standard deviations in Figure 4) and denitrification decreased significantly with increasing core section depth only at two sites (Tomorrow R. I and II, linear regression P < 0.05, Figure 4). Denitrification potential tended to be much higher on average at locations with high concentrations of groundwater nitrate such as Bear Cr., Tomorrow River Site II, and Radley Creek (Figures 4b, 4e, and 4f and Table 1). Denitrification potential increased linearly with groundwater nitrate concentration at low concentrations (<2 mg NO3-N/L) but denitrification potential varied considerably at high groundwater nitrate concentrations (>5 mg NO3-N/L), a pattern that suggests nitrate saturation (Figure 5). Denitrification potential was much higher for the nitrate-amended incubations than the unamended incubations for Hartman Cr. (12.2 versus 0.2μg N2O-N/cm2/h) and the Waupaca R. (16.4 versus <0.1 μg N2O-N/cm2/h) (paired t tests P < 0.001).
Table 2. Denitrification Potential and Sediment Percent Organic Matter (Mean, Standard Deviation, N) by Core Section for the Study Sites in the Waupaca River Network
Core Section (cm)
Denitrification Potential (μg N2O-N/cm2/h)
Percent Organic Matter
3.4. Dissolved Oxygen
 Dissolved oxygen concentration was higher in the deep groundwater than in the pore water at 5 cm depth (ANOVA P < 0.001, Table 3). At most sites (Tomorrow R. I, Bear Cr., Emmons Cr., Crystal R., Tomorrow R. II, and Radley Cr.) deep groundwater was oxic (53% of O2 saturation on average, equivalent to 5.6 mg O2/L on average at these sites) and approached anoxia (4% O2 and 0.5 mg O2/L on average and <1% O2 in most cases) at 5 cm (Table 3). Piezometer nest 2 at Emmons Cr. and nest 3 at Radley Cr. were exceptions, in which pore water at 5 cm remained oxic (>17% O2 and >1.9 mg O2/L). Alternatively, Hartman Cr. and the Waupaca R. had very low oxygen concentrations in the deep groundwater (1.5% O2 and 0.2 mg O2/L on average) and in the pore water at 5 cm (0.4% O2 and <0.1 mg O2/L on average).
Table 3. Dissolved Oxygen (Percent of Saturation Concentration Based on Temperature) in Pore Water at 5 cm Sediment Depth and in Deep Groundwater (Mean Sediment Depth Was 71 cm) at Each Piezometer Nest (1–3) at the Eight Study Sites in the Waupaca River Network
Shallow Pore Water
Tomorrow R. I
Tomorrow R. II
3.5. Organic Carbon
 Percent organic matter of sediment decreased on average with increasing sediment depth (Table 2, linear regression P < 0.01). Mean organic matter ranged from 17.6% (of dry mass), on average, at 0 to 5 cm to 7.5% at 25 to 30 cm. At some sites including Emmons Cr., Hartman Cr., and the Tomorrow River Site I sediment organic matter was frequently as high as 30 to 50% in shallower core sections (0 to 15 cm). Some sediment cores collected from Hartman Cr., Tomorrow River Site II and Waupaca R. had % organic matter as high as 15 to 45% at depths to 25 cm. Some cores from the Tomorrow R. Site I and II, Radley Creek, and Bear Cr. contained higher percent organic matter in deeper sections than in shallower sections suggesting that particulate organic matter had become buried by sandy sediments. Mean percent organic matter of sediment differed by as much as an order of magnitude among sites (Table 4, P < 0.001). Dissolved organic carbon concentration in deep groundwater ranged from 4 to 9 mg DOC/L among sites (Table 4) but means were not statistically different (ANOVA P = 0.33). There was no relationship between % organic matter and DOC concentration (Table 4).
Table 4. Mean and Standard Deviation Sediment Percent Organic Matter and Dissolved Organic Carbon (DOC) in Deep Groundwater at the Study Sites in the Waupaca River Networka
Sediment Percent Organic Matter
Deep Groundwater DOC (mg/L)
The percent organic matter is from the core sections (N = 21 to 24 per site) and the DOC is from the deep groundwater (N = 2 to 3 per site) both collected for the denitrification incubations.
Tomorrow R. I
Tomorrow R. II
4.1. Conceptual Model
 Our conceptual model for nitrate processing in deep sediments of groundwater fed streams was largely supported and we think it is applicable to the Waupaca River Network, and probably to other N-rich networks of gaining streams. Our hypotheses about changes in nitrate concentration, denitrification potential, and dissolved oxygen concentration with sediment depth were supported. First, for the sites in which nitrate was detectable in deep groundwater, NO3-N/Cl−ratios declined in most cases (14 out of 18 nest locations) as groundwater moved upward through the sediments before discharging to surface water. The profiles suggested that most nitrate decline occurred at relatively deep sediment depths (20 to 40 cm). Second, denitrification, while at higher rates in shallower sediments on average, occurred at the deepest depths at which it was measured (20–25 or 25–30 cm in most cases) at every site and core sections deeper than 5 cm accounted for about 70% of the depth-integrated denitrification potential. Third, dissolved oxygen concentrations were much lower on average in pore water at 5 cm sediment depth than in deep groundwater, suggesting oxygen concentration declined as groundwater passed through sediments with high oxygen demand (i.e., those rich in particulate organic carbon). The decline in oxygen concentrations from deep groundwater to pore water at 5 cm (where it approached anoxia) suggests that a redox driven process such as denitrification, which is performed by facultatively anaerobic bacteria [Tiedje, 1982], contributed to the decline in nitrate concentrations as groundwater moved upward through the sediments.
4.2. Nitrate Profiles
 As described above redox changes appeared to shape the nitrate profiles and probably contributed to low nitrate concentrations in the shallower sediments. At the two locations in which nitrate concentration remained high as water moved from the deep to shallower sediments (Emmons Cr. Nest 2 (Figure 3h) and Radley Cr. Nest 3 (Figure 3r)), the pore water at 5 cm was oxic (Table 3). This result suggests that the oxic conditions prevented nitrate from serving as the terminal electron acceptor during microbial respiration at these two locations which lead to no or little net change in nitrate concentration along the flow paths. At all other locations the pore water at 5 cm approached anoxia (<1% O2 and <0.1 mg/L), sufficiently bereft in oxygen for denitrification to occur [Tiedje, 1988; Tesoriero et al., 2005], and predictably nitrate concentrations declined from deep to shallower groundwater at most of these locations or remained low. However, at two locations in which anoxia was present in the pore water at 5 cm, Tomorrow River II Nest 2 and Radley Creek Nest 1 (Figures 3n and 3p and Table 3), the nitrate concentrations in shallower groundwater (0 to 30 cm) were relatively high and changed in unusual ways along the profiles, which suggests that nitrate loss was limited by some other factor than redox status, that the peeper and piezometers intersected multiple flow paths with different concentrations of nitrate, or that nitrate saturation occurred (see below). At both Hartman Cr. and the Waupaca R. deep groundwater was consistently at or below the detection limit for nitrate. The very low oxygen concentrations in the deep groundwater at these sites (<3% O2) suggests that conditions were favorable for denitrification at sediment depths approaching 1 m or greater. We think that that the zone where redox conditions are favorable for denitrification extends to deeper sediment depths at these two locations than at the other sites (see Figure 1) and that the low nitrate concentrations in the deep groundwater reflect previous nitrate removal [Tesoriero et al., 2007; Browne et al., 2008]. Saad  reported very high nitrate concentrations (>10 mg NO3-N/L) in deep groundwater (to 30 m depth) in the Waupaca River Watershed. The relatively high chloride concentrations in deep groundwater at Hartman Cr. and the Waupaca R. (Table 1) are also characteristic of groundwater that recharged in agricultural areas [Browne et al., 2008] and that would be expected to have high nitrate concentrations. Collectively, these results suggest that nitrate was removed from groundwater associated with Hartman Cr. and the Waupaca R. at deeper depths than those captured by our piezometers or in other locations (e.g., riparian zones) where the groundwater encountered favorable conditions for nitrate removal.
 The decrease in denitrification potential with increasing sediment depth at those sites with high groundwater nitrate concentration may have been due to a decline in the % organic matter content in sediments at increasing depth (Table 2). However, denitrification occurred in the deepest sediment sections at all sites which suggests that there were organic carbon sources, either in particulate or dissolved form, for denitrifying bacteria at depth at these locations. At several locations where sediment cores were collected for denitrification measurements, the 20–25 cm sediment sections had substantial quantities of organic matter (15 to 45%). We also frequently observed dark, organic-rich layers at depth when cores were collected. In some cases the organic-rich sediments extended from the top to the bottom of the core, which suggested continuous organic matter deposition. In other cases organic-rich sediments were covered by several layers of sand, which suggested that particulate organic matter burial had occurred. These observations are consistent with our conceptual model (Figure 1). In addition, the moderate to high concentrations of DOC in deep groundwater suggests that dissolved sources of carbon were available to denitrifying bacteria at depth.
 Our results suggest that measurements of denitrification that are restricted to surficial sediments (e.g., 5 cm or less) may underestimate the total amount of denitrification that is occurring throughout the sediment column. Our previous work at Emmons Creek [Stelzer et al., 2011b] showed that denitrification potential decreased with sediment depth, but sediment sections below 5 cm accounted for 68% of the depth-integrated denitrification rate, similar to our findings throughout the Waupaca River Network. However, the denitrification rates that we measured in the past and current study are probably best characterized as potential rates, as we have done throughout this paper. This is because we used deep groundwater for all of the incubations, which typically had higher nitrate concentrations, on average, than shallower groundwater, which is what the sediments collected for denitrification are exposed to in nature. Thus, shallow sediments used in the incubations were probably exposed to higher nitrate concentrations than what they encounter in nature, which may have led to overestimates of denitrification rate. In addition, denitrification was measured after anoxia was induced in the laboratory. Although most sediments at 5 cm depth approached anoxia in nature (Table 3) we do not have in situ dissolved oxygen data from the intermediate depths (10 to 30 cm) at which the sediments were collected for denitrification. Therefore, is possible that the induced anoxia per se caused elevated denitrification rates compared to what occurs in situ. These potential biases may have caused our measured denitrification rates to be higher than what occurs in nature but they would not necessarily have led to an overestimate of the relative contributions of deep sediments to depth-integrated denitrification. Other investigators who have measured how denitrification changes with sediment depth in rivers [Storey et al., 2004; Fischer et al., 2005; Inwood et al., 2007; Lansdown et al., 2012] showed that denitrification can occur in sediments to depths of 15 to 100 cm. At some locations denitrification rates measured by Storey et al.  and Fischer et al. in deep sediments (20 cm and deeper in this case) could account for about half or more of the depth-integrated denitrification rate in sediments, whereas at other locations denitrification rates in deep sediments were much lower.
 We showed that denitrification potential, integrated throughout the entire core, increased linearly with increasing groundwater nitrate concentration up to about 2 mg NO3-N/L and became saturated at higher nitrate concentrations (>5 mg NO3-N/L,Figure 5). The strong linearity below 2 mg NO3-N/L and the strong stimulatory response of denitrification to nitrate amendments in sediments from Hartman Cr. and the Waupaca R. suggests that denitrification was limited by nitrate at concentrations below 2 mg NO3-N/L. Many other investigators have shown that nitrate addition increases denitrification rates in laboratory assays [Kemp and Dodds, 2002; Richardson et al., 2004; Silvennoinen et al., 2008] and reach-scale measurements of denitrification have been shown to be positively correlated with surface water nitrate concentrations in streams [Mulholland et al., 2009]. There was considerable variation in denitrification potential among those sites in the Waupaca River Network that had high groundwater nitrate concentration. At some of the locations with high groundwater nitrate concentration, denitrification potential was 20 μg N2O-N/cm2/h or higher, while the rates were much lower at other locations. The locations with the two highest depth-integrated denitrification potential (Nest 1 at Bear Cr and Nest 3 at Tomorrow River Site II, 28.6 and 33.9μg N2O-N/cm2/h respectively, Figure 5) had nitrate profiles that were consistent with nitrate loss (Figures 3d and 3o). The location with the highest groundwater nitrate concentration (Radley Creek Nest 3 at 15 mg NO3-N/L) had very low depth-integrated denitrification potential (0.4 and 0.7μg N2O-N/cm2/h, data points overlap in Figure 5). The nitrate profile for Nest 3 at Radley Creek (Figure 3r) was consistent with the denitrification results, exhibiting high groundwater nitrate concentration and little change in nitrate concentration with depth. As described previously, the oxic conditions in both the pore water at 5 cm and the deep groundwater at this location may have contributed to the lack of nitrate loss. The evidence of nitrate saturation of denitrification from our study is consistent with results from other investigators that have shown that denitrification efficiency [Mulholland et al., 2009; Kreiling et al., 2011] or nitrate uptake [Bernot et al., 2006; Mulholland et al., 2008] in lotic ecosystems can become saturated at high nitrate concentrations. Our results suggest that while stream sediments provide an opportunity for nitrate removal as groundwater moves from recharge to discharge points, the capacity of stream sediments to remove groundwater nitrogen may be overwhelmed in the many regions of the world that experience chronically high groundwater nitrate concentrations [Tesoriero et al., 2007; Rupert, 2008; Schilling and Jacobson, 2008; Howden et al., 2011].
4.4. Implications for Nitrogen Budgets
 Many studies of nitrogen uptake and removal in streams, including those based on whole-stream injections [Mulholland et al., 2008], sediment cores [Herrman et al., 2008], and nitrogen retention at the network scale [Wollheim et al., 2008] have not considered nitrogen processing in deep sediments. However, as our results have shown deep stream sediments, like riparian zones [Hedin et al., 1998], can be locations where oxic nitrate-rich groundwater from upland areas encounters favorable redox conditions for reduction. Not including nitrogen removal and transformation in deep sediments could lead to underestimates of nitrate loss in watersheds and other errors associated with closing nutrient budgets, particularly in cases where the rate of groundwater flux through deep sediments is high. We think that our conceptual model for nitrate removal in deep sediments will be most applicable in situations where oxic, nitrate-rich groundwater [Pretty et al., 2006; Burow et al., 2010; Hansen et al., 2011] discharges to streams through fine sediment. Further research is needed to determine if nitrate loss occurs as commonly in deep, coarser sediments. Accounting for nitrate removal in deep sediments will probably be most important in watersheds with sandy soils that have moderate to high rates of nitrogen inputs, where oxic, nitrate-rich groundwater discharge to streams is likely to occur.
 We showed that our conceptual model of nitrate removal in deep sediments was largely supported when this model was tested at eight locations on streams and rivers in a river network. Denitrification potential was higher in shallow sediments on average but denitrification occurred at depths to 30 cm and sediments deeper than 5 cm accounted for 70% of the depth-integrated denitrification potential. In most cases nitrate profiles suggested nitrate loss occurred during upward movement of groundwater through sediments, although there was considerable variation among and within study sites. The depth at which nitrate loss occurred in stream sediments tended to vary and upgradient processes such as denitrification capacity in upland soils and riparian zones likely influences how much nitrate loss occurs in deep stream sediments. We think that organic matter deposition and burial, through their influence on redox state and availability of electron donors, play a key role in nitrate removal in deep sediments. Our results suggest that deep stream sediments should receive more consideration by biogeochemists interested in quantifying nutrient transformation and removal in rivers and watersheds.
 We are grateful for technical assistance from Ashley Winker and Alyssa McCumber. This research was supported by a grant from the University of Wisconsin Water Resources Institute through the Wisconsin Groundwater Coordinating Council and from a grant from the Faculty Development Program at the University of Wisconsin-Oshkosh. Comments from two anonymous reviewers improved the manuscript.