EXPLORING THE ROLE OF HYDROLOGIC RESIDENCE TIME AND CHEMISTRY IN THE PROCESSING OF NITRATE AT THE SEDIMENT - WATER INTERFACE By Tyler Barbee Hampton A THESIS Submitted to Michigan State University in partial fulfillment of the requirements for the degree of Geological Sciences Master of Science 2018 ABSTRACT EXPLORING THE ROLE OF HYDROLOGIC RESIDENCE TIME AND CHEMISTRY IN THE PROCESSING OF NITRATE AT THE SEDIMENT - WATER INTERFACE By Tyler B arbee Hampton The concentrations of inorganic nitrogen, including nitrate (NO 3 - ), are fundamental control s on the trophic state of aquatic ecosystem s . Excess NO 3 - degrades drinking water quality , and therefore there is a need to understand processes that remove inorganic nitrogen . Controls on NO 3 - removal at the sediment - water interface (SWI) of aquatic ecosystems include both biogeochemical and hydrologic conditions, however the relative importance and interact ions of these controls are poorly understood. This thesis explore s these controls on NO 3 - removal using a series of in - situ experiments involving both biogeochemical and hydrologic manipulations of the SWI in both lake and stream settings. Specifically, ma nipulati ve experiments altered dissolved organic carbon (DOC ) and NO 3 - concentrations, as well as physical hydrologic residence times. T he fate of NO 3 - in t hese manipulation experiments was traced by pairing isotopically labeled 15 N - NO 3 - tracer experiments with controlled variable - head infiltrometer rings to isolate the sediment - water system and control the hydrology of the SWI. With these experiments, I was able to isolate biogeochemical versus hydrologic controls on rates of NO 3 - remova l and denitrification rates. I found that increasing NO 3 - and DOC concentrations increase d NO 3 - rem oval and denitrification rates in the SWI , but that increases in physical residence time had a strong er effect on increasing NO 3 - removal and denitrification rates , especially under conditions where DOC and NO 3 - availability were not limiting . iii ACKNOWLEDGEMENTS from my mentors, peers, friends, and family. First and foremost, I want to thank my advisor, Jay Zarnetske, for his guidance throughout this project. I also want to thank my committee members Stephen Hamilton and Nathaniel Ostrom for their feedback on the direction of thi s research and on this thesis. Many thanks also to coauthors on this work, including Martin Briggs and Kamini Singha, whose support I have appreciated throughout my graduate school career. I am very grateful for the help of Farzaneh MahmoodPoor Dehkordy an d Courtney Scruggs who implemented the design of these studies and without whom this work would not have been possible. I owe many thanks to my labmates, including Joseph Lee - Cullin, Sydney Ruhala, Stephen Plont, and Sinchan Roy Chowdhury; who all offered a combination of expertise, institutional knowledge, feedback, field assistance, and emotional support. Furthermore, I would like to offer my appreciation to all my peers in the Department of Earth and Environmental Sciences and across Michigan State Unive rsity for their support throughout my degree and in my growth as a person during this time. In particular, I am grateful to the fostering of an environment that has destigmatized mental health issues and encouraged me to seek help for issues that affect ap proximately half of graduate students. Thanks also to my foster dogs, who gave so much love: Noodle, Indigo, Leia, Jack, Fiona. This research and derived presentations were funded or supported by the National Science Foundation, the Kellogg Biological Stat ion, the Geological Society of America, the Society for Freshwater Science; and the Graduate School, the Council of Graduate Students, and the Department of Earth and Environmental Sciences at Michigan State University. Finally, to my family and friends, w ho have shown so much love and support, and encouraged me to pursue my path at every step: thank you. iv TABLE OF CONTENTS LIST OF TABLES vi LIST OF FIGURES vii KEY TO ABBREVIATIONS xi CHAPTER 1: INTRODUCTION 1 CHAPTER 2: EXPERIMENTAL MODIFICATIONS OF REACTANT AND HYDROLOGIC CONTROLS ON NITROGEN PROCESSING: RESULTS FROM F LOW - THROUGH LAKEBED SWI SEDIMENTS 6 2.1. Introduction 6 2.2. Materials and Methods 10 2.2.1. Site Description 10 2.2.2. Experimental Setup 12 2.2.3. N, C, and Residence Time Manipulations 15 2.2.4. Porewater Sampling and Laboratory Methods 17 2. 2.5. Calculations 19 2.2.6. Sediment Characterization 22 2.3. Results and Discussion 23 2.3.1. Sediment Characterization 23 2.3.2. Hydrologic and Chemical Setting 25 2.3.3. Experimental Outcomes 27 2.3.4. Residence Time Controls N Cycling 35 2.4. Conclusions 38 CHAPTER 3: EXPERIMENTAL MODIFICATIONS OF HYDROLOGIC FLUX AND RESIDENCE TIME REVEAL CONTROLS ON NITROGEN PROCESSING IN THE SEDIMENT - WATER INTERFACE OF A HEADWA TER STREAM 39 3.1. Introduction 39 3.2. Materials and Methods 44 3.2.1. Site Description 44 3.2.2. Experimental Setup 46 3.2.3. Porewater Sampling Methods 49 3.2.4. Laboratory Analyses 51 3.2.5. Calculations of Residence Time and Reaction Rates 51 3.3. Results 52 3.3.1. Hydrologic and Chemical Setting 52 3.3.2. Solute Removal with Depth 55 3.3.3. Scaling by Residence Time 61 3.4. Discussion 65 3.4.1. Biogeochemical Reaction Rates in the SWI Controlled by Residence Time 65 3.4.2. Role of Anoxic Microzones and POC 67 3.4.3. Implications of Dynamic Stream Hydrology for N export 69 v 3.5. Conclusions 72 CHAPTER 4: SYNTHESIS & IMPLICATIONS 74 APPENDICES 78 APPENDIX A : 2016 Snake Pond Sampling Data 79 APPENDIX B : 2017 Sawmill Brook Experimental Data 83 REFERENCES 90 vi LIST OF TABLES Table 1: Details of the Snake Pond experiments. Details are provided on the addition of reactants and changing flux rates. Concentrations of the reactant tanks are reported, as well as the measured pump rate from the tank into the surface water of the injection ring. The addition rate is the concentrat ion in the addition tank multiplied by the pump rate. The hydraulic flux through the ring is also reported for each experiment, either directly measured or interpolated. 17 Table 2: Parameter values for calculation of Bunsen solubility coefficients for N 2 and N 2 O gasses in water. Values from Weiss (1970) and Weiss and Price (1980). 21 Table 3: Details of the Sawmill Brook Experiments. The measured hydraulic flux through the ring is reported for each experiment. 48 Table 4: Tabulated data from Figure 12. Concentrations are as either µmol/L or nmol/L. Experiment is listed, with A for Ambient or the experiment number. Type is the mean with standard deviation of points in parentheses, when more than one measurement was available. 80 Table 5: Tabulated data from Figure 13, as well as data for SO 4 2 - and NO 2 - . Rates are calculated as in Section 2.2.5. For the ambient profile, rates are calculated between 0 and 18 cm depth. For gases (O 2 , N 2 , and N 2 O), rates are between 9.5 and 19.5 cm depths. For NO 3 - and DOC, rates are ca lculated between 9.5 and 19.5 cm for Experiments 1 - 2, and between 0 and 19.5 cm for Experiments 3 - 5. 82 Table 6: Sediment core dat a from Snake Pond. Also including Loss on Ignition results from Figure 11. 82 Table 7: Tabulated data from Figure 20 and Figure 22. Concentrations are as either µmol/L or nmol/L. Experiment is listed, with the flux rate and ring (US or DS). Type is the mean with standard deviation of points in parentheses, when more than one measurement was available. 84 Table 8: Tabulated data from Figure 21 and Figure 23. Rates are calculated as in Section 2.2.5, as µmol/L/h or nmol/L/h, with mean ± standard deviation, when multiple points were available for the calculation. 86 Table 9: Injection and Flush porewater velocities from the Sawmill Brook Experiments. Velocities were determined by the depth (10 or 20 cm) divided by the median arrival time of th e conductivity plume at that depth, as shown in Figure 27. 87 vii LIST OF FIGURES Figure 1: Diagram of the sediment - water interface (SWI). In streams and lakes, groundwater and surface water exchange and interact with landscape fluxes of nitrogen (Minnesota Department of Natural Resources, Division of Ecological Services, 2003). 2 Figure 2: Schematic of the Cape Cod groundwater system. This diagram gives regional hydrologic context (A) to individual groundwater flow - through lakes like S nake Pond, as shown in (B). Groundwater entering the lake is generally poor in labile carbon and high in inorganic N as NO 3 - from anthropogenic groundwater pollution. We sampled the groundwater recharge, or outflow side of the lake, where oxidation of orga nic matter depletes the recharging water of oxygen, depicted by the red box. 9 Figure 3: Map of Massachusetts and Snake Pond. (A) . Cape Cod is a 100 - km - long peninsula that extends into the Atlantic Ocean. (B) Snake Pond, with sampling site shown on the south shore. Map units are in km. Map Projection is UTM, Zone 19T. 11 Figure 4: Site picture from Snake Pond. View is looking northeast, with the injection ring installed in the pond sediments offshore in the center frame. 11 Figure 5: Schematic of injection ring . A 55 - cm diameter plastic drum with open ends is inserted into the lakebed sediments to 22 cm - induced by experimentally elevating the hydraulic head in the injection ring (shown in the schematic as dH). Four steel piezometers (USGS MINIPOINT design) are inserted into the sediments and water is pumped from them at ~2.5 mL/min, so as not to disrupt the hydraulic flow field. In - line from the piezometers are dissolved O 2 and electrical conductivity flow - through sensors. 12 Figure 6: Schematic of Tracer Additions at Snake Pond. Before the experiments, water from the lake was pumped into a 1.89 m 3 holding tank. Using a series of pumps and float switches (FS) to maintain steady water levels, water was pumped into an intermediat e bucket and then into the injection ring. Tracers were added using a peristaltic pump at a rate of ~3 mL/min. 13 Figure 7: Breakt hrough curves for extracting residence time. Specific conductivity was measured at 9.5 and 14.5 cm depth within the injection ring sediments. Panel (A) is for the onset of the experiments, and panel (B) is for the transition from high flux rate to low flux rate, corresponding with a replacement of the injection ring water (Cl - labeled) with fresh lake water, Points along breakthrough curves are for the median time of arrival for the conductivity plume for each depth, in cm. For the injection: 0.53 h at 9.5 cm and 0.79 h at 14.5 cm. For the flush: 0.29 h at 9.5 cm and 0.53 h at 14.5 cm. 14 Figure 8: Conceptual Diagram of Snake Pond Exp erimental Modifications. The three natural controls on SWI N processing are microbial community composition, porewater reactant chemistry, and hydrologic transport. By confirming with a 15 NO 3 - tracer that a microbial community capable of N transformations was present, Experiments 2 - 4 interrogated how changing reactant chemistry changes N removal, and Experiment 5 tested the hypothesis that hydrologic residence time is a key control. 16 viii Figure 9: Particle size distribution for shallow Snake Pond cores. 8 cores were collected on 7/26/2016 and sampled at 2 cm depth. The 10th percentile and 50th percentile particle sizes are shown for each core. 24 Figure 10: Particle size distribution for a deep core from Snake Pond. A core was collected on 7/8/2016 to 11 cm dep th. Depth intervals are shown, with the 10th percentile and 50th percentile particle sizes for each interval. 24 Figure 11: Loss on Ignition results from Snake Pond core. A core collected on 7/9/2016, sampled at 2 cm intervals up to 11 cm depth. (A) % mass lost on ignition, sorted by particle size greater than and less than 500 µm. (B) Percent LOI converted to mass loss per cm 3 . 25 Figure 12: Concentration profiles with depth below the sediment - water interface for Snake Pond. Concentrations (C) are shown at st 15 N/ 15 N max . The five experiments are described in Table 1. Species shown: NO 3 - , O 2 15 N 2 , and N 2 O (symbol legend in figure). Error bars are for the standard deviation of three replicates at each depth, when available; some error bars are within the size of the plotted point. 26 Figure 13: Biogeochemical flux rates in the Snake Po nd experiments. Rates of O 2 , NO 3 , DOC measured as NPOC, N 2 , and N 2 O, across the five experiments (Table 1). For the ambient profile (A), rate is calculated between 0 and 18 cm depth. For gases (O 2 , N 2 , and N 2 O), rates are between 9.5 and 19.5 cm depth. For NO 3 - and DOC, rates are calculated between 9.5 and 19.5 cm for Experiments 1 - 2, and between 0 and 19.5 cm for Experiments 3 - 5. Rates are calculated as in Section 2.2.5. Error bars are based off the standard deviation of the concentrations of 3 samples at each depth. 29 Figure 14: Maps of Massachusetts, the Ipswich River Watershed, and Sawmill Brook. ( A) Study region within Massachusetts, USA with the (B) Ipswich River Watershed, showing the study site at Sawmill Brook and the nearest USGS stream gage (01101500) on the Ipswich River at South Middleton, MA. (C) Topographic map of the Sawmill Brook study reach and site (marked with star). Map units are kilometers. Datum is UTM zone 19T. 43 Figure 15: Detailed Plan - view Site Map of Sawmill Book. Stream lev el and banks were (downstream) (see Figure 5 for ring details). The groundwater (GW) and background (ambient) locations were sampled for groundwater and SWI porewate r samples, respectively. The SWI ambient samples were taken at sediment depths of 2.5, 5, 7.5, 10, 15, 20 cm depths, and the groundwater sample was taken at 60 cm depth. The site labeled as stage (Figure 14) shows the location of our in - stream pressure log ger. Map units are meters. Coordinates are for UTM zone 19T. 44 Figure 16: Site picture from Sawmill Brook. View is looking east , downstream, with ring US in the foreground and ring DS in the background. Additional equipment not described herein, such as the orange wiring seen in the image, are associated with a concurrent geophysical (electrical resistivity imaging) study. 45 Figure 17: Schematic of Tracer Addition at Sawmill Brook. Before the experiments, water from the stream was pumped into two holding ta nks. 15 NO 3 - was added to both tanks, and NaCl ix to only one. Using a series of pumps and float switches (FS) to maintain steady water levels, water was pumped into an intermediate bucket and then into the injection ring. An aerator was in the intermediate bu cket to keep water oxygenated during sometimes long residence times in the bucket at low flux rates. There was no aerator in the holding tank. 48 Figure 18: Conceptual Diagram of Sawmill Brook Experimental Modifications. The three major natural controls on SWI Function on N are microbial community composition, porewater reactant chemistry, and hydrologic transport cond itions. By confirming with a 15 NO 3 - tracer that a microbial community capable of N transformations was present and seeking to isolate the role of reactant chemistry on N removal, Experiments 1 - 4 systematically changed hydrologic flux, with two experiments exploring an oxic or anoxic regime, to observe changes in N removal as a result of changes in hydrologic controls. 49 Figure 19: S tream stage and chemical conditions during sampling at Sawmill Brook. (A) Stage monitored at the site on Sawmill Brook over the study dates. Sampling periods of the four experiments are shown by red bars, with the commencement of each injection as the begi nning of the bar and the end of sampling for that experiment as the end of the bar. Experiment order was 2 m/d, 3 m/d, 0.8 m/d, 1.2 m/d. Total conductivity (TC; not temperature corrected) from the site surface water is also shown, with points showing spot - checks of TC with a handheld probe. (B) NO 3 - concentrations in the stream, shown by connected lines, and concentrations in the tank interquartile range for samples coll ected in the stream, tanks, and rings. Whiskers are to the minimum and maximum. Points are outliers. (C) DOC concentrations as shown in panel C. (D) Oxygen (O 2 ) saturation and concentration are shown for the surface water from the two SWI rings. The boxplo t shows O 2 during the sampling periods (red boxes). (E) Temperature as monitored at the stream stage site and within the two SWI rings. 54 Figure 20: Concentration over depth and residence time from Sawmill Brook experiments. Concentrations of O 2 , NO 3 - , and DOC over flowpath length (A) and over porewater residence time (B). Symbols denote data from Rings US and DS. The color of each line and point denotes the flux rate (see inset legend in the lower left panel), where lighter shade denotes the lower flux rates. Error bars are the standard deviation of concentration when repeated samples were possible. 56 Figure 21: Removal rates for O 2 , NO 3 - , and DOC from the Sawmill Brook Experiments. Symbols denote ring US or DS. Error bars are based on the standard deviation of concentrations used in the rate calculations, where multiple samples are available. More negative values indicate increasing removal rates. Positive values indicate accumulati on along the flowpath. 57 Figure 22: 15 N tracer conditions of N 2 and concentrations of N 2 O from Sawmill Brook Experiments. 15 N iso topic enrichment of N 2 and concentrations of N 2 O over flowpath length (A) and over porewater residence time (B). Symbols denote data from Rings US and DS. The color of each line and point denotes the flux rate, where lighter shade denotes the lower flux ra tes. Error bars are the standard deviation of concentration when repeated samples were possible. 58 Figure 23: Production rates of N 2 and N 2 O from Sawmill Brook Experiments. Symbols denote ring US or DS. Error bars are based on the standard deviation of concentrations used in x the rate calculation, where multiple samples are available. Positive values indicate accumulation along the flowpath. 59 Figure 24: Residence time over flux rate in the Sawmill Brook experiments. Residence time is shown at 20 cm depth, o r the bottom of the SWI flowpath, for each flux rate experiment. 62 Figure 25: Concentrations over Damköhler number for O 2 from Sa wmill Brook Experiments. Concentrations of O 2 , NO 3 - , N 2 15 N 2 over the Damköhler number for oxygen removal ( ), calculated for each ring in each experiment. 63 Figure 26: Removal rates over Removal Velocities for O 2 (left) and NO 3 - (right) from Sawmill Brook Experiments. Removal rates are reported as in Figure 21 and calculated in Section 2.2.5 and Removal velocities are calculated as in Section 3.2.5. Blue lines are linear regression lines. Circles are from ring DS and triangles are from ring US. 64 Figure 27: Conductivity Breakthrough Curves from the Sawmill Brook Experiments. Data are plotted as electrical conductivity over time for the first th ree experiments, for both 10 and 20 cm depth on the injection (A) and flush (B) phases of the high - conductivity injections. Time is normalized to the beginning of injection or flush. Red lines mark the initial concentration and plateau; the orange line mar ks the median concentration between those two, and the green line marks the time of median arrival time, which was used to calculate velocities as in Table 9. 88 xi KEY TO ABBREVIATIONS SWI Sediment - water interface N Nitrogen N R Reactive Nitrogen NO 3 - Nitrate N 2 Di - nitrogen gas N 2 O Nitrous oxide gas NH 3 Ammonia C Carbon DOC Dissolved Organic Carbon POC Particulate Organic Carbon NPOC Non - Purgeable Organic Carbon DO Dissolved Oxygen Cl - Chloride SO 4 2 - Sulfate USGS United States Geological Survey US EPA United States Environmental Protection Agency SIF Stable Isotope Facility, University of California Davis °C Degrees Celsi us PVC Polyvinyl - Chloride HDPE High Density Polyethylene EC Electrical Conductivity VHG Vertical Head Gradient 1 CHAPTER 1: INTRODUCTION Anthropogenic inputs of reactive nitrogen to landscapes have steadily increased since the beginning of the 20 th century following the advent of industrial nitrogen fixation for fertilizer production and the proliferation of nitrogen fixing crops, resulting in an up to 20 - fold increase in fluxes of nitrogen to the ocean (Howarth et al. , 1996) . Nitrate (NO 3 - ) loading in surface waters is now considered one of the top global threats to ecosystems and humanity (Rockström et al. , 2009; Steffen et al. , 2015) . Fortunately , only about 2 5% of anthropogenic reactive nitrogen (N R ) inputs to the continent s is exported by rivers to the ocean s, with the balance either sequestered or removed (Howarth et al. , 1996; Boyer et al. , 2006) . Freshwater ecosystems perform a critical ecosystem service by contributing to this N R retention and removal : the freshwater continuum is estimated to remove ~50% of N R that enters water bodies before export to the oceans (Galloway et al. , 2004) . Lakes and rivers are estimated to contribute a similar proportion of anthropogenic N R removal (Seitzinger et al. , 2006) . The dominant removal pathway of N R in freshwaters is denitrification, the microbially - mediated anaero bic reduction of dissolved inorganic N as NO 3 - to di - nitrogen (N 2 ) and nitrous oxide (N 2 O) gases (Payne, 1973; Tiedje et al. , 1983) . Most denitrifiers are f acultative aerobes and thus are thought to only perform denitrification when oxygen (O 2 ) becomes limiting (Tiedje et al. , 1984; Mosier et al. , 2002) , though studies have shown that this process can occur in predominantly oxic soil environments (Robertson and Kuenen, 1984; Lloyd et al. , 1987; Robertson and Tiedje, 1987; Lloyd, 1993) . The sediment - water interface (SWI), a n important ecotone between surface and groundwater ecosystems (Boulton et al. , 1998; Boano et al. , 2014) ( Figure 1 ), is a hospitable environment for denitrifying microbes. The mixing of these two waters in the SWI and aerobic respiration of DOC results in the depletion of O 2 supplies, with NO 3 - supplied by in situ 2 nitrification or by external surface - water or groundwater inputs. Studies have focused on both fluvial (Marzadri et al. , 2011; Zarnetske et al. , 2011a, 2012; Harvey et al. , 2013) and lacustrine (Whitmire and Hamilton, 2005; Burgin and Hamilton, 2008; Smith et al . , 2015; Stoliker et al. , 2016) settings to examine the role of the SWI in NO 3 - removal and denitrification. Despite research covering a range of ecosystems and impact levels, questions remain regarding the relative impact of external chemical versus ph ysical conditions on SWI biogeochemical function. In other words, the relative importance of reaction and transport controls to NO 3 - removal and denitrification in the SWI remains poorly understood . Figure 1 : Diagram of the sediment - water interface (SWI). In streams and lakes, groundwater and surface water exchange and interact with landscape fluxes of nitrogen (Minnesota Department of Natural Resources, Division of Ecological Services, 2003) . 3 Many NO 3 - removal pathways in the SWI , includi ng denitrification, are controlled by multiple concurrent factors, including the availability and lability of electron donors such as DOC (Baker et al. , 1999; Zarnetske et al. , 2011b) ; NO 3 - concentrations (Mulholland et al. , 2008) ; SWI flowpath length (Quick et al. , 2016) ; microbial community composition and abundance (Storey et al. , 1999; Farrell et al. , 2013; Stoliker et al. , 2016) ; and physical residence times that in turn are a product of stream - bed morphology and composition and hydrologic conditions (Cardenas, 2008) . Though not often discussed in SWI studies, these controls are interrelated, since hydrodynamics result in a distribu tion of flowpath lengths and residence times (Briggs et al. , 2014b; Marzadri et al. , 2014) governing the transport of reactants to resident microbes and of reaction products downstream or down - gradient. Some p revious studies have suggest ed as much interrelatedness, and hypothesize that hydrologic conditions would be among the most dominant controls on SWI function (Ocampo et al. , 2006; Gu et al. , 2007; Stoliker et al. , 2016) . However, relatively few studies have performed controlled manipulati ve experiments exploring SWI function in a field setting to assess the relative influence of biogeochemical versus hydr ological controls. This thesis explore s the magnitude of major controls on SWI NO 3 - cycling by systematically regulating both hydrologic and biogeochemical conditions in real SWIs. The experiments are centered around a series of controlled variable - head hy drologic manipulation experiments, examining the removal of NO 3 - from infiltrating surface - water along an isolated SWI flowpath. The first set of experiments ( described in Chapter 2 ) were conducted in Snake Pond, MA ( Figure 3 ) during the summer of 2016. This site was chosen to provide a hydrologically stable environment to test these new methods, while also leveraging an environment of interest to othe r N R studies in the region (Barbaro et al. , 2013; Smith et al. , 4 2015) . This study a lso directly addressed hypotheses proposed by Stoliker et al. (2016) on the role of hydrologic variation in controlling N export from lakes. The Snake Pond study sought to address the following main research questions : 1) how do changing concentrations of DOC and NO 3 - influence NO 3 - removal in the SWI; and 2) how do es hydrologic variability ( as invoked by manipulating pressure head) change residence times along a SWI flowpath and the removal of O 2 , DOC, and NO 3 - ? The second set of experiments described in Chapter 3 was conducted in Sawmill Brook, a tributary of the Ipswich River, MA ( Figure 14 ). Following the Snake Pond experiments, the study design focu sed on examining the scaling nature of SWI processing of NO 3 - with vary ing hydrologic residence times, reexamining question 2 described above. While the transition to a fluvial environment posed additional logistical challenges, this study had high relevan ce to studies comparing N R removal and denitrification at the same site to streams across the conterminous United States (Wollheim et al. , 2005; Mulholland et al. , 2008; Beaulieu et al. , 2011) . In these two studies, I find that NO 3 - concentration and labile DOC abundance both stimulate increased NO 3 - removal and denitrification (Mulholland et al. , 2008; Zarnetske et al. , 2011b) , but that hydrologic residence time primarily controls the NO 3 - removal rate (Stoliker et al. , 2016) more than the abundance of NO 3 - or labile DOC. The experiments also specifically examine how these hypothesized controls influence the abundance of two respective end products of denitrification N 2 O and N 2 gases. The nitrogen gas end - products of denitrification are of particular intere st because N 2 O is a potent greenhouse gas (Forster et al. , 2007) and also a strong contributor to recent depletion of stratospheric ozone (Ravishankara e t al. , 2009) , while N 2 is relatively inert. Increased emissions of N 2 O from freshwater environments as a result of anthropogenic NO 3 - loading contribute significantly to global anthropogenic emissions 5 (Galloway et al. , 2004; Beaulieu et al. , 2011) . In addition to evaluating controls on NO 3 - removal, this research examines how the proposed controls affect the conversion of NO 3 - to N 2 O versus N 2 , ultimately affecting the relative amounts of these gases being release d from the SWI. These findings will inform future conceptual - and process - based modeling efforts to study NO 3 - cycling in streams and lakes. Additionally, while not directly part of the scope of the proposed experiments, this research also addresses the likely prominence of less - mobile porosity (e.g., diffusion dominated mass - transfer) in the oxygenated zone of the study SWI sediments (Briggs et al. , 2015). This less - mobile porosity m ay be an important location for SWI denitrification (Briggs et al. , 2015) . F inally, data from these experiments will be beneficial to future numerical flow and transport models, by enhancing understanding of NO 3 - removal in the SWI and generating parameterization and validation data sets. 6 CHAPTER 2: EXPERIMENTAL MODIFICATIONS OF REACTANT AN D HYDROLOGIC CONTROLS ON NITROGEN PROCESSING : RESULTS FROM FLOW - THROUGH LAKEBED SWI SEDIMENTS 2.1. Introduction Excess reactive nitrogen ( N R ) in surface waters is considered one of the top global threats to aquatic ecosystems (Vitousek et al. , 1997; Steffen et al. , 2015) , and freshwater ecosystems perform a critical ecosystem service by removing about 25 % of anthro pogenic N before it is transported to the oceans (Howarth et al. , 1996; Boyer et al. , 2006; Seitzinger et al. , 2006) . Many processes governing N concentrations in freshwater systems take place within sediment - water interfaces (SWI s) (Boulton et al. , 1998; Boano et al. , 2014) . Though the role of SWIs in N processing are often studied in the context of fluvial systems (Zarnetske et al. , 2011a, 2012; Harvey et al. , 2013) , surface - groundwater exchanges in lakes also create the potential for N processing (Chen et al. , 1972; Cherkauer et al. , 1992; Rysgaard et al. , 1993; van Luijn et al. , 1996; Kidmose et al. , 2015; Lewandowski et al. , 2015; Smith et al. , 2015) , particularly in groundwater flow through lakes with strong advective exchange (Rosenb erry et al. , 2015) . SWIs are disproportionately important to N cycling in freshwater systems due to long exposure timescales, mixing of organic and inorganic solutes, and high microbially active sediment surface area (McClain et al. , 2003; Zarnetske et al. , 2012; Abbott et al. , 2016) . The main process the microbially mediated anaerobic reduction of dissolved inorganic N as nitrate (NO 3 - ) to di - nitrogen (N 2 ) and nitrous oxide (N 2 O) gases (Tiedje et al. , 1983) . While N 2 is inert, N 2 O a product of incomplete denitrification is a potent greenhouse gas (Beaulieu et al. , 2011) that has also been implicated in the depletion of stratospheric ozone (Forster et al. , 2007; Rav ishankara et al. , 2009) . B oth biogeochemical and 7 physical conditions may dictate how much and what form of N is ultimately exported from freshwater systems (Zarnetske et al. , 2012) . The b iogeochemical functioning of the SWI with respect to N is dictated by a hierarchy of conditions: 1) transport or exchange of surface waters across the SWI, 2) sufficient N and associated electron acceptors in the SWI waters, and 3) the presence of a microb ial community within the SWI capable of removing N R . N - modifying communities are consistently shown to be ubiquitous in SWI sediments (Sobczak et al. , 1998; Findlay and Sinsabaugh, 2003; Stoliker et al. , 2016) , suggesting that their presence is not a limiting factor and that the resident community will readily process N in the order of the most energetically favorable reactions (Storey et al. , 1999; Burgin and Hamilton, 2007; Burgin et al. , 2011) . Biogeochemical controls on N processin g are important, as NO 3 - concentrations influence both NO 3 - removal and N 2 and N 2 O production (Mulholland et al. , 2008; Beaulieu et al. , 2011; Quick et al. , 2016) . Both the quantity and quality (labi lity) of d issolved organic carbon (DOC) are potential limiting reactant s for denitrification (Sobczak et al. , 2003; Zarnetske et al. , 2011b) . Key to the understanding of overall SWI function in N processing is that parcels of water entering the SWI experience a distribution of residence times and flowpaths (Marzadri et al. , 2014; Briggs et al . , 2015) , and long er residence times result in long er contact times between reactants and microbial communities (Findlay, 1995; Zarnetske et al. , 2012) . Longer SW I residence times and higher oxygen (O 2 ) removal rates enhance rates of NO 3 - removal (Thomas et al. , 2001; Zarnetske et al. , 2012) . Physics ultimately regulates the delivery of solutes to SWI microbial communities as well as solute residence/exposure timescales. This physical transport limitation would occur when shor t residence times constrain the exposure of solutes to microbes even if there is abundant NO 3 - and labile DOC in the water. 8 The interaction between lakes and groundwater affect water chemistry and N cycling in the water column . This is important to water m anagers trying to alleviate excess N pollution. Many studies have focused on groundwater flow - through lakes (Born et al. , 1974, 1979; Anderson and Munter, 1981; Winter et al. , 1998; Winter, 199 9) ; a specific classification of lake common in glacio - fluvial terrains where the lake intersects an aquifer with discharge and recharge zones located along the up - gradient and down - gradient sides of the lake shore, respectively ( Figure 2 ). These lakes are significant in the regional groundwater budget in highly populated and economically valuable regions such as Cape Cod, MA, USA, where approximately 25% of the total groundwater flux passes through lakes (Walter and Whealan, 2004; Walter and Masterson, 2011) , and the U . S . E nvironmental Protection Agency estimates billions of $USD will be spent in the coming years to mitigate N R pollution in lakes, rivers, and coastal bays (US EPA, 2016) . For these groundwater flow - th rough lakes in particular, a previous study showed that the potential for N transformations at the SWI was not limited by microbial community or functional group presence all groups were found to be ubiquitous and instead other environmental variables such as availability of DOC and water residence time may control N removal at the SWI (Stoliker et al. , 2016) . For example, it is expected that changes in lake stage and the adjacent groundwater table create a dynamic hydraulic gradient across lakebed sediments, resulting in variable flowpath orientation and porewater velocities and residence times (Winter, 1999) , which in turn affect the N - processing function of the SWI . 9 Figure 2 : Schematic of the Cape Cod g roundwater s ystem . This diagram gives regional hydrologic context (A) to individual groundwater flow - through lakes like Snake Pond, as shown in (B). Groundwater entering the lake is generally poor in labile carbon and high in inorganic N as NO 3 - from anthropogenic groundwater pollution. We sampled the groundwater recharge, or outflow side of the lak e, where oxidation of organic matter depletes the recharging water of oxygen, depicted by the red box. The objective of this research was to characterize how changing reactant and hydrologic conditions concomitantly influence the fate of NO 3 - passing through the SWI of flow - through lakebed sediments. Using a novel field method of induced vertical recharge through lakebed sediments, we specifically explored how NO 3 - and labile DOC availability influenced microbial respiration rates (e.g. oxygen removal rates) and how changing the system residence time by reducing the hydraulic gradient would affect these rates. We hypothesized that labile carbon supply would limit denitrification, and that reducing the hydraulic gradient would increase removal o f N and C. We also directly addressed previous hypotheses about how seasonal 10 changes in N and C availability and hydraulic gradient through sediments affect biogeochemistry and N pollution (Stoliker et al. , 2016) . 2.2. Materials and Methods 2.2.1. Site Description Snake Pond is a 33 - hectare kettle lake on the Cape Cod Peninsula in Sandwich, Massachusetts, USA ( Figure 3 ). This peninsula consists of several intersecting glacial moraines with outwash pla in deposits extending to the south (Mather et al. , 1942) . The aquifer is composed of permeable sands and gravels (Masterson et al. , 1996) , and like many of the lakes on Cape Cod, Snake Pond is a groundwater flow - through lake with no surface - water inlets or outlets (Winter et al. , 1998) . The lake is adjacent to the Joint Base Cape Cod, which has contaminated regional groundwater with N R (LeBlanc et al. , 1991; Smith et al. , 1991; Repert et al. , 2006; Barbaro et al. , 2013) . Snake Pon d is attractive for studies involving solute manipulation: being near the regional groundwater high point (Massachusetts Division of Fisheries and Wildlife, 1993) , natural total dissolved solutes are relatively low and stable (Ahrens and Siver, 2000) . Therefore, in addition to residence time, potential limitations on net N removal imposed by dissolved N and C availability could readily be tested. The study was conducted in July 2016, with the study site located at the southern, naturally recharging, side of the lake ( Figure 4 ) in a gravel - cobble substrate ( Figure 9 ), about 3 m from the shoreline in shallow wat er. 11 Figure 3 : Map of Massachusetts and Snake Pond. (A). Cape Cod is a 100 - km - long peninsula that extends into the Atlantic Ocean. (B) Snake Pond, with sampling site shown on the south shore. Map units are in km. Map Projection is UTM, Zone 19T. Figure 4 : Site picture from Snake Pond. View is looking northeast, with the injection ring installed in the pond sediments offshore in the center frame. 12 Figure 5 : Schematic of injection ring . A 55 - cm diameter plastic drum with open ends is inserted into the lakebed sediments to 2 2 cm - experimentally elevating the hydraulic head in the injection ring (shown in the schematic as dH). Four steel piezometers (USGS MINIPOINT design) are inserted into the sediments and water is pumped from them at ~2.5 mL/min, so as not to disrupt the hydraulic flow field. In - line from the piezometers are dissolved O 2 and electrical conductivity flow - through sensors. 2.2.2. Experimental Setup The studied lake sediments were isolated from the surrounding environment using a 55 - cm - diameter polyvinyl - chloride (PVC) barrel, which was installed in the lakebed and driven to a depth of 2 2 cm, also serving to ensure vertical flow ( Figure 5 ). Water from the lake was pumped into a 1.89 m 3 holding tank located on the shore, where it was mixed with sodium chloride (NaCl) salt to bring the total conductivity of the water from ~ 60 to 550 µS/cm. 13 Figure 6 : Schematic of Tracer Additions at Snake Pond . Before the experiments, water from the lake was pumped into a 1.89 m 3 holding tank. Using a series of pumps and float switches (FS) to maintain steady water levels, water was pumped into an intermediate bucket and then into the injection ring. Tracers were added using a peristaltic pump at a rate of ~3 mL/min. Using an intermediate bucket and a series of float switches ( Figure 6 ) to maintain a constant water level within the PVC injection ring, water was pumped from the holding tank to the intermediate bucket and into the ring to enhance and control the natural rec harge rate ( Table 1 ) by precisely manipulating the vertical hydraulic gradient (Scruggs et al. , 2016) . The flux rates were chosen so that these experiments could be directly compared to previous SWI - related studies in Ashumet Po nd (Bussey and Walter, 1996; Walter and LeBlanc, 1997; McCobb et al. , 2003; Rosenberry et al. , 2013; Santelli et al. , 2014; Smith et al. , 2015; Stoliker et al. , 2016) , which is about 4.8 km to the south and has similar hydrologic and geologic characteristics, and where downwelling seepage rates have been reported to be as high as 1.7 m/d (Harvey et al. , 2015) . A combination of meas ured water flux rates and specific conductivity (SpC) breakthrough curves ( Figure 7 ) was used to determine vertical flowpath residence times in the var ious 14 experiments. A second series of breakthrough curves was obtained as the high conductivity water was pushed out by new fresh lake water added at the beginning of Experiment 4 (see next section). This was done to match the injection breakthrough curve a t approximately the same flux, before flux was reduced by lowering the hydraulic head in the ring, which was performed for Experiment 5. Ambient porewater samples were collected about 5 m away from the injection rings at a similar distance from the shore. Also adjacent to the injection rings, iButton thermal data loggers (model DS1922L, Maxim Integrated, CA, USA) were installed at a depth spacing of 0.03 m up to 0.11 m. Ambient vertical downwelling flux was calculated using a diurnal signal amplitude attenu ation - based model run by VFLUX2, as in Briggs et al. (McCallum et al. , 2012; Briggs et al. , 2014a; Irvine et al. , 2015) . A ) B ) Figure 7 : Breakthrough curves for extracting residence time. Specific conductivity wa s measured at 9.5 and 14.5 cm depth within the injection ring sediments . P anel (A) is for the onset of the experiments, and panel (B) is for the transition from high flux rate to low flux rate, corresponding with a replacement of the injection ring water (Cl - labeled) with fresh lake water, Points along breakthrough curves are for the median time of arrival for the conductivity plume for each depth, in cm. For the injection: 0.53 h at 9.5 cm and 0.79 h at 14.5 cm. For the flush: 0.29 h at 9.5 cm and 0.53 h at 14.5 cm. 15 2.2.3. N, C, and Residence Time Manipulations Five different experiments were conducted, with sequential reactant additions followed by an increase in residence time. Sampling of porewater was conducted roughly 24 hours after the beginning of each experi ment to allow solute concentration s to reach steady - state . Conservative tracer and dissolved oxygen (O 2 ) profiles at depth were examined to ensure steady - state flow conditions and that dissolved O 2 concentrations had stabilized at the time of sampling. Rea ctants and tracers were drawn from 20 L tanks at about 3 mL/min using a peristaltic pump and added to the recharging lake water within the injection ring ( Figure 6 ) . Each reactant addition involved adding the new reactant or tracer to the injection ring following previous additions . A summary of modifications and expected results are as follows (see also Figure 8 and Table 1 ) : Ambient Profile: The goal was to assess N, O 2 , and C processing in the native lakebed sediments under the background downwelling rates (~0.12 m/d). Experiment 1: The goal was to assess N p rocesses and biogeochemical conditions under increased downwelling rates (1.2 m/d) relative to the Ambient Profile , introduc ing a 15 NO 3 - tracer to track 15 N denitrification products. Experiment 2: The goal was to assess available NO 3 - limitation on N proc essing. We added a NO 3 - amendment, where the original addition of 15 N in the first modification was calculated to achieve about 5 atomic percent (atom %) 15 NO 3 - in this experiment. Experiment 3: The goal was to assess available labile DOC limitation on N processing when NO 3 - is abundant. We added labile DOC (as acetate), at the same time as the 15 N and NO 3 - amendments. 16 Experiment 4: The goal was to assess N processing when NO 3 - and DOC are abundant, under anoxic conditions. This also concluded the salt injection (see previous section). T he injection ring water was replaced with new lake water and the NaCl addition ceased . New lake water was pumped into the intermediate bucket an d injection ring. To achieve the same concentrations in the recharging water as the previous experiment, a slug of solutes was mixed with the fresh lake water, and all three amendments ( 15 N, NO 3 - , C) were pumped into the injection ring throughout the exper iment just as in Experiment 3. The acetate addition rate was increased to eliminate DOC limitation and stimulate O 2 depletion at depth. Experiment 5: The goal was to assess the effect of increased residence time when NO 3 - and DOC are abundant. The hydraul ic head was reduced to just above average lake level, to achieve a downward flux of approximately 0.92 m/d. Figure 8 : Conceptual Diagram of Snake Pond Experimental Modifications. The three natural controls on SWI N processing are microbial community composition, porewater reactant chemistry, and hydrologic transport. By confirming with a 15 NO 3 - tracer that a microbial community capable of N transformations wa s present, Experiments 2 - 4 interrogate d how 17 changing reactant chemist ry changes N removal, and Experiment 5 test ed the hypothesis that hydrologic residence time is a key control. Table 1 : Details of the Snake Pond experiments. Details are provided on the addition of reactants and changing flux rates. Concentrations of the reactant tanks are reported, as well as the measured pump rate from the tank into the surface water of the injection ring. The addition rate is the concentrat ion in the addition tank multiplied by the pump rate. The hydraulic flux through the ring is also reported for each experiment , either directly measured or interpolated. Experiment Abbrev iation Details Pump Rate (mL/min) Addition Rate (µmol/h) Hydraulic Fl ux Rate (m/d) Ambient Conditions Amb Outside injection ring NA NA 0.12 Experiment 1: 15 N Addition 15N 51. 3 mg/L K 15 NO 3 (99% purity) 2.90 87 15 NO 3 - 1.2 Experiment 2: NO 3 - Addition NO 3 1,47 4 mg/L KNO 3 2.90 2.95 87 15 NO 3 - 2600 NO 3 - 1.267 (interpolated) Experiment 3: Acetate Addition N+C 615 mg/L NaAcO 2.90 2.95 3.08 87 15 NO 3 - 2600 NO 3 2800 C 1.356 (interpolated) Experiment 4: 2 nd Acetate Addition N++C 1714 mg/L NaAcO 2.90 2.95 3.08 87 15 NO 3 - 2600 NO 3 - 7800 C 1.45 Experiment 5: Increased Residence Time N++C, R T All added 2.90 2.95 3.08 87 15 NO 3 - 2600 NO 3 7800 C 0.92 2.2.4. Porewater Sampling and Laboratory Methods Sampling at depth was achieved using four stainless steel MINIPOINT samplers, similar to the USGS MINIPOINT system (Harvey et al. , 2013) , installed in the lakebed sediments within the ring. These had an outer diameter (OD) of 3.2 mm, a screened interval 10 mm long, with 3 individual slits ~0.5 mm in width. Samplers were driven to depths of 9.5, 14.5, 19.5, and 24.5 cm below the lakebed. 3.2 - mm OD tubing was attached to the end of each MINIPOINT and the two shallowest lines were fed through two electrical conductivity (EC) micro flow - through cells - cell c ontrol units using sensor - specific calibrations performed at the beginning of the experiment. Flow - through cells equipped with fiber - optic O 2 microsensors attached to a FireStingO2 Optical 18 Oxygen Meter (Pyro Science, Germany) were in - line with the tubing f rom the MINIPOINTs. Ambient porewater data were also collected outside the injection ring approximately 5 m away at the same distance from shore. For these ambient data, MINIPOINT samplers were driven to depths of 1.5, 7, 12, and 18 cm below the lakebed. D uring Experiments 1 - 2, concentrations of the reactants were not measured in the downwelling surface water, however, mass balance calculations based on the reactant addition rates and downwelling water flux indicates that changes between surface and 9.5 cm depth for all solutes in these two experiments were negligible . From the onset of each experimental modification and reactant addition, about 24 hours passed before sampling took place. Triplicate w ater sampl es were obtained in a closed system using perist altic pumps (Cole - Parmer, IL, USA) and syringes, followed by immediate filt ration through a 0.7 - µm glass - fiber filter and 0.2 - µm cellulose - acetate filter into acid - washed amber high - density polyethylene ( HDPE ) bottles (Nalgene, NY, USA) . Water samples were chilled on site and frozen the evening following collection. For dissolved gas samples, 1. 6 mm OD tubing directly from the pump was placed into the bottom of a 12 - mL glass Exetainer (Labco, United Kingdom) and filled for two full volumes. Samples were pre served with 120 µL of 50% w/v zinc chloride solution. Preserving a convex meniscus, the tubing was removed, and the cap was screwed on to prevent any air bubbles in the sample. Gas samples were stored at room temperature in the dark and later shipped to th e Stable Isotope Facility (SIF) at the University of California, Davis, for isotope ( 15 N) analysis of dissolved gases (N 2 and N 2 O). Water samples were later separated into groups for 15 N analysis of NO 3 - , which were shipped frozen to the SIF , and for anion , carbon, and nutrient analysis, which were kept chilled at 4ºC during shipment and prior to analysis at Michigan State University. 19 At the SIF, the sealed 12 - mL glass Exetainers had 4 mL of sample water replaced with a helium headspace, which was then all owed to equilibrate with the remaining 8 mL of sample. S table isotope ratios of nitrogen ( 15 N) in N 2 and N 2 O from the equilibrated headspace gas were measured using a ThermoScientific GasBench + Precon gas concentration system interfaced to a ThermoScientific Delta V Plus isotope - ratio mass spectrometer (Bremen, Germany). Nitrate in water samples was converted to N 2 O by the bacteria l denitrificati on assay and 15 N ratios were measured as stated above for N 2 . A t Michigan State University , a nions were measured with a Dionex ICS - 2100 Ion Chromatography System (ThermoScientific, MA, USA), including chloride (Cl - ), nitrite (NO 2 - ), nitrate (NO 3 - ), and sul fate (SO 4 2 - ). Non - purgeable Organic Carbon (NPOC) and Total Dissolved Nitrogen were measured using a TOC - L total organic carbon analyzer (Shimadzu, Japan) using catalytic oxidation at 720ºC followed by gas chromatographic measurement of CO 2 and chemilumine scence measurement of N O . Samples were also shipped to the USGS in Reston VA for analysis of NH 3 using a Seal AQ2 Discrete Analyzer (Seal Analytical, WI, USA) using method EPA - 103 - A Rev 10. 2.2.5. Calculations After inserting the injection ring into the lakebed, the recharge flux was increased from the ambient downwelling flux of 0.12 m/d to 1.2 m/d. Flux rate increased to 1.5 m/d over the course of the experiments ( Table 1 ), due to dropping lake stage and changes in the hydraulic gradient between the elevated head in the ring and the lake stage. Median arrival times of the Cl - labeled lake water at 9.5 and 14.5 cm - depths were estimated from SpC breakthrough curves and median porewater velocities were calculated by subtracting the median time of arrival at 9.5 cm from the time at 14.5 cm and dividing by the known separation distance of 5 cm. Velocities for Experim ents 2 and 3 were interpolated, assuming a linear increase in velocity and flux over time. 20 Porosity was calculated by dividing the flux by the porewater velocity, resulting in an effective porosity of 30%. Porewater velocities for Experiment 5, with increa sed residence times, were calculated by assuming the same porosity and dividing the flux of 0.92 m/d by the porosity. Residence times at each depth for each experiment were then calculated by dividing the depth by the calculated porewater velocity for that experiment. Removal rates were calculated as the linear regression of concentration over time as in Lansdown et al. (2015) : ( 1 ) is the removal rate in µmol/L/h, is concentration in µmol/L and is the residence time in hours (h) at a given depth. Concentrations were retrieved directly from the analytical instruments described in Section 0 . For calculations of denitrification rates (N 2 and N 2 O production), rates were based on a linear isotopic mixing model (Ostrom et al. , 2002; Harvey et al. , 2013) . The SIF provided data for concentrations and 15 N enrichment 15 N relative to air) of N 2 and N 2 O gases from the equilibrated helium headspace of the 12 - mL Exetainers. T he concentration of gas in the original liquid sample was calculated as the total mass in the system divided by the liquid volume : ( 2 ) Where is the original dissolved gas concentration in the liquid sample , is the vessel liquid volume of 8 mL, is the reported mass of N 2 or N 2 O in the final equilibrated vessel headspace , and is the mass remaining in the liquid, which can be calculated based on the headspace mass: ( 3 ) 21 Whe re is the equilibrated vessel liquid concentration, is the atmospheric pressure with units of atmospheres , is the vessel headspace volume of 4 mL, and is the Bunsen solubility coefficient for headspace equilibration within the vessel (units of atm - 1 ) . The Bunsen solubility coefficient for N 2 and N 2 O are calculated as a function of equilibrium temperature ( ) in units of Kelvin (Weiss, 1970; Weiss and Pric e, 1980 ; Table 2 ) : ( 4 ) Table 2 : Parameter values for calculation of Bunsen solubility coefficients for N 2 and N 2 O gasses in water. Values from Weiss (1970) and Weiss and Price (1980) . Parameter Value in Value in Isotopic enrichment of N 2 and N 2 15 N relative to air. To conduct a 15 N mass balance, the isotopic mole fraction ( ) was calculated to determine the proportion of reported N 2 and N 2 O mass that originated from the added 15 NO 3 - tracer. First 15 N was converted to the ratio ( ) of 15 N/ 14 N , by standardizing against the natural abundance ratio of 15 N in the environment ( ) (Ostrom et al. , 2016) , then the ratio was transformed into a fraction: ( 5 ) ( 6 ) In the isotopic mixing model , the N composition of porewater sampled at depth i is a mixture of the porewater advected from depth i - 1 and the mass and composition of products of denitrification ( ) between these depths: ( 7 ) 22 where is concentration and is the isotopic mole fraction of 15 N . The mass of the product can be solved for by assuming that the isotopic enrichment of the denitrification product ( ) is equivalent to the enrichment of the 15 N labeled NO 3 - source at depth i - 1 : . ( 8 ) The denitrification rate can be calculated as in the previous secti on by dividing the mass of the product by the difference in residence times between the two depths i - 1 and i : . ( 9 ) For the ambient profile, rates are calculated between 0 and 18 cm depth. For gases ( O 2 , N 2 , and N 2 O), rates are between 9.5 and 19.5 cm depths. For NO 3 - and DOC, rates are calculated between 9.5 and 19.5 cm for Experiments 1 - 2, and between 0 and 19.5 cm for Experiments 3 - 5. 2.2.6. Sediment Characterization An 11 - cm deep core was collected proximal to but outside of the injection ring during the sampling (7/9/2016), and later 8 shallow (~3 cm deep) cores were collected from the surface sediments surrounding the site (on 7/26/2016). Methods of coring and analysis fo llowed Harvey et al. (2013) with only a few exceptions. C ores were collected by pushing a clear polycarbonate cylinder (nominally 4.8 cm internal diameter and 1.6 mm wall) that had been sharpened at one end into the lakebed. Cores were capped with butyl ru bber stoppers and removed from the lakebed. After removal the cores were immediately extruded, sectioned into 1 or 1.5 - cm increments, bagged, placed on ice, and returned to the laboratory. Cores were wet sieved to remove fines from sand and gravel and drie d at 60 degrees C to constant weight. Porosity was determined using dry weight and bulk volume of each core increment assuming a grain density of 2.65 g/cm 3 . The grain size distribution was determined by dry sieving samples through 17000, 4000, 1000, 500, 250, 125, and 63 m diameter sieves on a Gilson Model SS - 3 shaker and 23 weighing each size fraction. The secondary axis of pebbles larger than 17000 m were measured individually. Characteristic grain sizes of gravel, sand, and fines were determined with ref erence to a 4000 m maximum as indicators of the median grain size (D 50 ) and the diameter of the tenth percentile weight fraction (D 10 ). These grain size metrics characterize the finer sediment that fills in between the pebbles, increasing granular surface area and decreasing the hydraulic conductivity of the bulk streambed sediment. The 11 - cm deep core was analyzed for particulate organic carbon (POC) in sediments less than 4000 m by combusting samples at 550 degrees C in a muffle furnace for 24 hours to determine weight fraction after loss on ignition (LOI). The core was divided into 1 - 2 cm intervals and sieved to particle sizes less than and greater than 500 Mass loss was calculated by multiplying the % LOI by the sediment density to obtain mass l oss in grams of organic matter per cm 3 . 2.3. Results and Discussion 2.3.1. Sediment Characterization Sediments at the sampling site were medium and coarse sands with small contributions of fines below 0.5 mm in diameter ( Figure 9 ). Gravel and pebbles also made up a small portion of the sediments. Sediments had a polydisperse nature, with moderate to poor grading. There was generally high variability in sediment char acteristics observed within the 8 shallow cores collected in July. Slight coarsening was observed with depth within the cores ( Figure 10 ). POC was on average 0.43 weight % of the total sediment, with a larger proportion (0.58 wt. %) as LOI Figure 11 sediment but showed a larger percent decrease from 0 to 11 cm ( - 34%) than total POC ( - 15%). 24 Figure 9 : Particle size distribution for shallow Snake Pond cores . 8 cores were collected on 7/26/2016 and sampled at 2 cm depth. The 10th percentile and 50th percentile particle size s are shown for each core. Figure 10 : Particle size distribution for a deep core from Snake Pond. A core was collected on 7/8/2016 to 11 cm depth . Depth intervals are shown, with the 10th percentile and 50th percentile particle size s for each interval. 25 A) B) Figure 11 : Loss on Ignition results from Snake Pond core. A core collected on 7/9/2016, sampled at 2 cm intervals up to 11 cm depth. (A) % mass lost on ignition, sorted by particle size greater than and less than 500 µ m. (B) Percent LOI converted to mass loss per cm 3 . 2.3.2. Hydrologic and Chemical Setting Ambient local downwelling rates were 0.12 m/d adjacent to the experimental ring, as determined by temperature modeling. This modeling used an in - situ measurement of thermal 26 diffusivity of 0.13 m 2 /d, which was derived from changes in paired diurnal signal amplitude and phase with depth (Luce et al. , 2013; Briggs et al. , 2014a; Irvine et al. , 2015) . Ambient dissolved O 2 data showed anoxic or virtually anoxic ( <63 µmol O 2 /L ; Rosamond et al. , 2012) conditions at shallow depths beneath the sediment surface , reaching a concentration of 11 µmol O 2 /L at 7 cm below the se diment surface ( Figure 12 ). Ambient NO 3 - concentrations were low (<1.3 µmol NO 3 /L), but patterns with depth suggested a small zone of nitrification as well as a zone of net NO 3 - removal below the oxic - anoxic transition. Figure 12 : Concentration profiles with depth below the sediment - water interface for Snake Pond . C oncentration s (C) are shown at steady state (~24 h) and expressed as C/Cmax 15 N/ 15 N max . The five experiments are described in Table 1 . Species shown: NO 3 - , O 2 , DO C, 15 N 2 , and N 2 O (symbol legend in figure) . Error bars are for the standard deviation of three replicates at each depth, when available; some error bars are within the size of the plotted point. 27 Concentrations of NH 3 were below detection (<1.43 µmol NH 3 /L) for almost all samples, and thus nitrification was unlikely to contribute significantly to N cycling unless it was closely coupled with denitrification . Removal rates of DOC (between 0 - 18 cm depths ) were low compared to the later experiments, at 10.6 µmol DOC/L /h, with c oncentrations reducing to a bout half from 0 to 18 cm depth ( Figure 12 ). These ambient chemical profiles demonstrate the presence of microbial populations performing aerobic respiration, as rates of O 2 and DOC removal were similar, but 8 times greater on average than the 1:1 O 2 :C molar ratio predicted by the expected stoichiometric relationship (Findlay and Sobczak, 1996) . This could be explained if respiration was utilizi ng particulate organic carbon (POC) in the sediments (Sawyer, 2015; Quick et al. , 20 16) , which is consistent with the decreased POC content with depth in our sediment cores ( Figure 11 ). 2.3.3. Experimental Outcomes Downwelling conditions were maintained within the injection ring for the duration of the experiments. Analysis of data from the 24.5 cm depth MINIPOINT piezometer indicated that it was to o close to the bottom of the 1 - dimensional flow field generated by the experimentally raised hydraulic head, and that our conservative and reactive tracers were being diluted by ambient groundwater. For this reason, data from the 24.5 cm depth are not discussed. The onset of our experiments corresponded with an increase in downwelling flux from an estimated 0.12 m/d to 1.2 m/d and with the addition of 15 NO 3 - to serve as a tracer for denitrification. During Experiment 1, the increased downwelling rate caused the shallow anoxic ( <63 µmol O 2 /L) zone that was present in the ambient sediments to move deeper ( Figure 12 ) . The sampling depth at 9.5 cm remained consistently oxic through Experiments 1 - 5, with an average O 2 concentration of 216 µmol O 2 /L. Despite the reduced efficacy of O 2 removal in 28 Experiment 1, dropping from 85% removal at 18 cm under ambient conditions to 27% removal at 19.5 cm, O 2 removal rates increased by 620% relative to ambient conditions ( Figure 13 ), suggesting that ambient conditions had been transport - limited in terms of O 2 supply. Removal of DOC also decreased from 52% efficacy to 6.4%, despite a 180% increase in DOC removal rate. During Experiment 1, production of N 2 O was also observed, presumably from the reduction of the added 15 NO 3 - tracer. Concentrations of N 2 O at 19.5 cm depth were 9.18 n moles N 2 O - N /L. The production of N 2 O in Experiment 1 indicates that N 2 O escaped during the sequential reduction of NO 3 - to N 2 in denitrification (Firestone and Davidson, 1989; Baulch et al. , 2011) . Sediments were observed to be bulk - oxic (i.e., dissolved O 2 was detected in bulk samples) , so the source of this denitrification byproduct can be attributed local anoxic microzones embedded in the sediment matrix (Triska et al. , 1993; Harvey et al. , 2013; Briggs et al. , 2015; Sawyer, 2015) . 29 Figure 13 : Biogeochemical flux rates in the Snake Pond experiments. Rates of O 2 , NO 3 , DOC measured as NPOC, N 2 , and N 2 O, across the five experiments ( Table 1 ) . For the ambient profile (A), rate is calculated between 0 and 18 cm depth. For gases (O 2 , N 2 , and N 2 O), rates are between 9.5 and 19.5 cm depth. For NO 3 - and DOC, rates are calculated betwe en 9.5 and 19.5 cm for Experiments 1 - 2, and between 0 and 19.5 cm for Experiments 3 - 5. Rates are calculated as in Section 2.2.5 . Error bars are bas ed off the standard deviation of the concentrations of 3 samples at each depth. In Experiment 2, downwelling lake water was amended with NO 3 - , bringing NO 3 - from a background concentration of 1.04 µmol/L to approximately 142 µmol/L. This NO 3 - addition had little effect on O 2 removal or denitrification ( Figure 12 ), suggesting organic carbon, and more likely the availability of labile DOC, wa s a more important limitation on denitrification . Increased NO 3 - concentrations corresponded with an increase in the peak N 2 O concentration (at 14.5 cm) in the sediments by 110%, in agreement with previous studies showing correlations between dissolved NO 3 - and N 2 O concentrations (Firestone and Davidson, 1989; Beaulieu et al. , 2011; Quick et al. , 2016) . In Experiment 2, DOC removal between the 9.5 - 19.5 cm depths also 30 increased by 140% ( Figure 13 ), with the greatest removal rates of DOC for both Experiments 1 and 2 taking place between the depths of 14.5 and 19.5 cm. Still, there was a slight mismatch betwe en the depth intervals of maximum O 2 , NO 3 - , and DOC removal ( Figure 12 ). Consequently, in the subsequent experiments ( Experiments 3 - 4) we tested if denitrification in our interrogated sediments was limited by DOC. In Experiment 3, downwelling lake water was amen ded with labile DOC in the form of acetate (Baker et al. , 1999; Za rnetske et al. , 2011b; Kurz et al. , 2017) , bringing DOC from a background concentration of approximately 250 µmol /L to approximately 370 µmol /L (+47%). Following this labile DOC addition in Experiment 3, rates of O 2 removal (9.5 - 19.5 cm) increased by 77%, and NO 3 - removal (0 - 19.5 cm) increased to 6.36 µmol/ L/h , confirming DOC limitation o f respiration and NO 3 - removal. Even with the addition of both NO 3 - and acetate, conditions remained oxic throughout the sediments in Experiment 3. Under these and subsequent experimental conditions, the majority of NO 3 - and DOC removal took place along flowpaths in the first 9.5 cm beneath the sediment surface , whereas t he highest O 2 removal occurred between 9.5 and 14.5 cm. The removal rate of DO C (0 - 19.5 cm) in Experiment 3 decreased by 93%, but this was largely caused by an depth intervals, between 14.5 and 19.5 cm ( Figure 12 ). Net DOC production has previously been observed in alluvial aquifers, but at much longer residence times than those in our st udy (Helton et al. , 2015) . In Experiment 3 the sediments had transitioned away from having any nitrification signal, conditions remained oxic, and N 2 production was too low to be detectable by our methods. N 2 O production rates (9.5 - 19.5 cm) were not observed to be different from Experiment 2. T he increase in NO 3 - removal was not accompanied by an increase in the proportion of denitrification accounted for by N 2 O , which was only 0.04%. 31 The acetate addition in Experiment 4 produced an 82% increase in DOC concentrations (to 680 µmol C - DOC/L ) in the downwelling lake water. The goal of inducing more anoxia was achieved for the first time in Experiment 4, where the anoxic zone shifted upwards toward the SWI to 14 .5 cm, driven by a 74% increase in the O 2 removal rate (0 - 19.5 cm) . In Experiment 4, the sampling depth s straddled the bulk oxic - anoxic transition between 9.5 and 14.5 cm - depth. During Experiment 4 NO 3 - concentrations increased in the injectate from 160 to 250 µmol/L (+54%) as a function of the shift in injection rate due to a constant experimental water level within the ring and naturally changing lake stage. Thus, Experiment 4 was not solely a DOC mani pulation. The 86% increase in labile DOC supply in Experiment 4 yielded a 1400% increase in NO 3 - removal (0 - 19.5 cm) and 2700% increase DO C removal rate (0 - 19.5 cm). Concentrations of DOC decreased by 20% from surface water to 19.5 cm depth. ( Figure 13 ). At 14.5 cm depth, the percent removal of DOC from the surface water concentration increased from 32% in Experiment 3 to 59% in Experiment 4. Concentrat ions of DOC continued to exhibit apparent productionat depth, suggesting that the true DOC removal rate was higher especially up to 14.5 cm depth. Removal of O 2 also continued to outpace removal of DOC, with >60% of O 2 removal unaccounted for in the remova l of DOC. This suggests that over the duration of our experiments local POC continued to be an important electron donor in aerobic respiration. This is supported by the presence of POC in our sediment core s ( Figure 11 ), and by observations of strong retention of DOC in shallow sediments of Ashumet Pond (Harvey et al. , 2015) . While this labile DOC addition demonstrated that an increased supply of acetate promotes NO 3 - removal, it did not have a strong effect on denit rification rates. 32 In Experiment 4, the N 2 production rate (9.5 - 19.5 cm) increased to 1.8 µmol N 2 - N/ L/h , but only represented 1.9% of the observed NO 3 - removal. In contrast, the N 2 O production rate (9.5 - 19.5 cm) increased by almost 4400%, representing 5.7% of total denitrification (N 2 + N 2 O). Most clearly shown in Experiment 4 was a spatial offset between NO 3 - removal and denitrification ( Figure 12 ), wit h most of the 15 N gas accumulating further along the downwelling flowpath. Concentrations of NO 3 - were low where the highest denitrification rate was observed. In Experiment 5, where residence times were increased by the reduced flux, O 2 concentrations on ly decreased another 5% from surface water conditions when compared to Experiment 4 because O 2 was already close to being entirely depleted at depth ( Figure 12 ). Like the case in Experiment 4, and due specifically to the manipulated flux rate and enhanced concentration of the flux from the reactant drip tanks, measured NO 3 - and DOC concentrations increased in the lake water to 800 µmol NO 3 - /L (+220%) and to 1600 µmol DOC/L (+135%). Under these increased residence time conditions in Experiment 5, NO 3 - removal (0 - 19.5 cm) increased by another 420% to 500 µmol/ L/h . The removal rate for DOC (0 - 19.5 cm) also increased by 370% to 680 µmol /L /h (77% removal at 19.5 cm depth) ( Figure 13 ), with a maximum rate of DOC removal occurring along the first 9.5 cm of flowpath. At 14.5 cm depth, the percent removal of DOC from the surface water concentration increased from 59% in Experiment 4 to 85% in Experiment 5 ( Figure 12 ). Once longer residence time conditions we re introduced ( Experiment 5), denitrification also increased markedly by 2500% to a rate (9.5 - 19.5 cm) of 47 µmol N 2 - N/ L/h, and N 2 O production (9.5 - 19.5 cm) increased by 4100% to a net rate of 4.66 µmol N 2 O - N/ L/h . Therefore, under reactant - replete conditions ( Experiments 1 - 4), measured NO 3 - removal rates were strongly transport - limited. This scenario can be translated to the downwelling sediments of Ashumet 33 Pond, where this transport limitation is likely a significant control on N export. In creased denitrification rates are a result of the sediment becoming anoxic once NO 3 - and DOC additions occurred ( Experiment 4). Specifically, the O 2 removal rates increased with subsequent acetate additions, suggesting increased aerobic processing of DOC t hat depleted the O 2 (Hedin et al. , 1998) . So in the presence of conditions favorable to anoxia, the net NO 3 - removal becomes a function of residence (or exposure) time in the SWI , as sugges ted by theory and reviews of previous studies (Zarnetske et al. , 2012; Abbott et al. , 2016) . In Experiment 5, N 2 production increased to represent 9.4% of the observed NO 3 - removal, and N 2 O production increased to represent 9.1% of total NO 3 - removal . N 2 O production accounted for 0.93% of NO 3 - removal, in agreement with previous observations in aquatic sediments (Beaulieu et al. , 2011) . Taken together, the result s of these experiments demonstrate that while NO 3 - and DOC concentrations are important limiting factors for denitrification and specifically N 2 O production (Bernhardt and Likens, 2002; Mulholland et al. , 2008; Zarnetske et al. , 2011b) , residence time is the most important control on N 2 O production. Despite higher percent increases in reaction rates for NO 3 - and DOC between Experiments 3 and 4, this experimental increase in residence times during Experiment 5 showed the largest magnitude of increases in reaction rates ( Figure 13 ), as well as for denitrification from ambient conditions. Quick et al. (2016) showed in their column experiments that N 2 O accumulation peaks at an intermediate residence time such that oxygen is depleted, and sediments are bulk - anoxic, but where N 2 O does not become the most energetically favorable electron donor to then produce N 2 . O nly 9.5% of NO 3 - removal could be accounted for in production of N 2 and N 2 O in this study. In other studies in lakes, this proportion can vary widely from 63 to 100% (Chen et al. , 1972; Rysgaard et al. , 1993) ; and in one study of streams, up to 87% of 15 NO 3 - added in 34 sediment incubations could be accounted for by biological assimilation (Lansdown et al. , 2012) . Sulfur (S) oxidation - driven reduction of N could also be ev idenced by the observed increasing SO 4 2 - concentrations with depth ( Table 4 ). Burgin and Hamilton (2008) (Burgin and Hamilton, 2008) observed, in a review of studies, that S - driven NO 3 - reduction to N 2 acco unted for on average 25% of NO 3 - removal in streams, and 45% in lakes. The SO 4 2 - concentrations in our SWI may indicate NO 3 - being reduced to N 2 while sulfide is oxidized to SO 4 2 - , because SO 4 2 - concentrations increased with depth in all our experiments. S till, based on the observed SO 4 2 - increases, the S - driven NO 3 - reduction pathway could only account for 33%, 3.7%, and 3.5% of NO 3 - removal in the Experiments 3, 4, and 5, respectively. Further, the stoichiometry of this S - reaction during Experiment 4, bas ed on observed SO 4 2 - production, could predict almost 97% of N 2 production. Predicted N 2 production by this reaction under increased residence times ( Experiment 5) could account for 19% of the total observed N 2 production. Consequently, the 15 N - NO 3 - tracer would have still produced 15 N 2 by this S - based reaction, and it may not be possible to differentiate this S - driven pathway of N 2 production from dissimilatory N reduction via denitrification. In addition, there was a net increase of NO 2 - to 25 µmo l/L at 19.5 - cm depth, while in Experiment 4 NO 2 - concentrations only increased to 13 µmol/L ( Table 4 ). These concentration increases of NO 2 - correspon ded to nitrate reduction rates of 12 and 3.7 µmol/L/h occurring between the sediment surface and 19.5 cm depth for Experiments 4 and 5, respectively. For Experiments 4 and 5, taking into account recovery of N end - products as N 2 , N 2 O, and NO 2 - , 86 and 89% o f NO 3 - removal must be accounted for by some other pathway, such as biological assimilation. 35 The initial ambient conditions and the results of Experiments 1 - 4 suggest that low background concentrations of NO 3 - and DOC created reactant limitations for deni trification . However, it is not until the transport timescales are manipulated that it becomes apparent that this SWI system is also limited by rates of hydrological transport. Importantly, the largest increase i n overall biogeochemical function of these sediments with respect to NO 3 - and DO C removal and denitrification was observed with the experimentally increased residence times. 2.3.4. Residence Time Controls N Cycling Our results agree with previous findings that de nitrification is limited by labile DOC supply (Baker et al. , 2000b; Zarnetske et al. , 2011b; Quick et al. , 2016) . In this study we also address the role that residence time p lays in the biogeochemical function of a system receiving water with the same initial concentrations and ratios of DOC and NO 3 - flowing through a fixed SWI volume at different rates . The likelihood that a SWI flowpath will transition from net nitrification to net denitrification increases after the residence time increases to the point where dissolved O 2 is depleted and anoxia can develop, often represented in a Damköhler number framework for O 2 (Zarnetske et al. , 2012; Briggs et al. , 2014b; Marzadri et al. , 2014) . The Damköhler framework acknowledges that at longer residence times denitrification becomes more likely , but it is incomplete in capturing other limitations, such as reactant limitations, on denitrification. A complicat ing factor is that at longer residence times, the DOC source is also more likely to be exhausted, especially the labile forms as they are removed preferentially, concentrating the more recalcitrant DOC compounds along longer flowpaths (Zarnetske et al. , 2011b; Lansdown et al. , 2015; Quick et al. , 2016) . This increases the likelihood of DOC limitation of the se cond half of the denitrification reaction and would increase the likelihood of 36 more N 2 O reduction relative to N 2 production because additional carbon electron donors are needed to get from N 2 O to N 2 (Hedin et al. , 1998; Quick et al. , 2016) . These dynamics between N 2 O versus N 2 production are indicated in our second acetate addition ( Experiment 4) and increased residence time ( Experiment 5) experiments. Here we observed increased proportion s of N 2 O production representing total denitrification and NO 3 - removal relative to the N 2 , corresponding with an over 4000% increase in total N 2 O production ( Figure 13 ). Similar to the results of Experiment 5, Lansdown et al. (2015) found that residence time was a principal control on denitrification and NO 3 - in deep stream sediments, where their deeper sed iments accounted for 81% of observed subsurface NO 3 - removal in their study. D enitrification products containing tracer 15 N were observed in the SWI even at depths whe re bulk oxic conditions were present, ( Experiments 1 - 3). These observations imply the pr esence of anoxic microzones embedded in bulk oxic pore waters that facilitate denitrification , a process long proposed to occur in unsaturated soils (Reddy and Patrick, 1975; Sexstone et al. , 1985; Kravchenko et al. , 2017) and in stream sediments (Triska et al. , 1993; Zarnetske et al. , 2011a; Harvey et al. , 2013; Lansdown et al. , 2014, 2015) . The heterogeneity in sediment porosity characteristics com mon in SWI environments can result in a broad distribution of residence times along the flowpaths, with smaller throated pores having longer residence times and creating pore volumes more likely to become anoxic (Briggs et al. , 2015) . Given the nature of the sediments observed in our SWI ( Figure 12 , Table 4 ), there are certainly a distribution of more - and less - connected pore volumes. Briggs et al . (2015) modeled anoxic microzones across a range of changing hydrologic flow rates and O 2 removal rates and showed that small portion of porosity (3 - 5%) was consis tently anoxic, with slower flow rates and shorter threshold time to anoxia , resulting in the highest proportion of microzones. Along with physical sediment grain 37 heterogeneity, buried POC has also been shown to result in localized anoxic zones (Kr avchenko et al. , 2017) , and enhanced microbial activity (Sobczak et al. , 1998) . Recognition that denitrification rates and residence times can be highly variable across small spatial sca les (Harvey et al. , 2013; Lansdown et al. , 2015) further emphasizes the potential importa nce of microzone contribution to total flowpath denitrification, especially N 2 O production, which was observed in our study. This study is one of the first field demonstrations of the dynamic biogeochemical functioning of groundwater flow through lake SWI sediments (as represented by N and C removal and denitrification rates). It also clearly shows that the functioning of these SWI s can be dramatically changed by altering hydraulic gradients and thus residence times (Stoliker et al. , 2016) . Consequently, any environmental factor that changes local or regional hydraulic flux (e.g., seasonal - or management - induced variable lake stage or regional water table) will change the biogeochemical function of the SWI and impact the abundance of NO 3 - an d N 2 O mass moving through and out of these important inland waters. This dependence has been observed in comparable studies done in rivers, where it has been demonstrated that changing river stage changes the fate of NO 3 - in the SWI of a river (Gu et al. , 2008) . 38 2.4. Conclusions Pa st studies in groundwater flow - through lakes have shown that there is not a fundamental microbial limitation on the transformation of N R in these coupled surface and groundwater systems because denitrifiers are ubiquitously distributed and facultatively ae robic (Stoliker et al. , 2016) . Consequent ly, the controls on the fate of N R in SWIs have been hypothesized to be primarily via limitations on the supply and reactant exposure timescales. Here we directly tested this hypothesis in the SWI of a lake and show that while labile carbon limitations are important, the overall net effect of physical transport timescales is a more dominant control on the fate of N R . Future studies can explore the optimal condition of multiple reactants and multiple residence times by conducting more field - and lab - based re sidence time manipulation experiments . Overall, we established that the transport limitation interacts with the reaction limitation, including increasing the anoxic domain and volume where denitrification can occur. Consequently, the fate of N R in these co upled lake and groundwater systems will vary primarily with hydrological processes that regulate hydraulic gradients driving surface - groundwater exchanges, and thus the transport timescales of reactants through lake SWIs. To investigate the scaling nature of transport timescales and SWI N processing, a follow - up study was conducted and is described in Chapter 3. 39 CHAPTER 3: EXPERIMENTAL MODIFICATIONS OF HYDROLOGIC FLUX AND RESIDENCE TIME REVEAL CONTROLS ON NITROGEN PROCESSING IN THE SEDIMENT - WATER INTERFACE OF A HEADWATER STREAM 3.1. Introduction Human activity has dramatically altered the global nitrogen (N) budget , impacting nearly all aquatic ecosystems on the planet (Vitousek et al. , 1997) . This N manipulation is caused p rincipally by the conversion of atmospheric N 2 to N - based fertilizers through the Haber - Bosch process, but also by altering atmospheric NO x concentrations and thus N deposition through the burning of fossil fuels (G alloway et al. , 2004) . There is large uncertainty around the global fluxes of anthropogenic reactive N (N R ) from landscapes to the oceans, but freshwater ecosystems are highlighted as both important transporters and sinks of N (Schlesinger et al. , 2006) . The proportion of ant hropogenic N R inputs to landscapes that is ultimately removed by freshwater ecosystems before reaching the oceans has been estimated to range from 8 - 50% (Howarth et al. , 1996; Galloway et al. , 2004; Boyer et al. , 2006; S eitzinger et al. , 2006) . Despite this uncertainty, most of this transformation likely occurs in headwater streams (Peterson et al. , 2001; Thomas et al. , 2001; Bernhardt and Likens, 2002; Seitzinger et al. , 2002) . Headwater streams make up the majority of river network length (Downing et al. , 2012) , and have the highest proportion of sediment contact area to surface flow area (Anderson et al. , 2005; Gardner and Doyle, 2018) . Consequently , a key locus of N transformations in smaller rivers and streams is the sediment water interface (SWI): often called the hyporheic zone, which is the zone of exchange between stream water and groundwater (Boulton et al. , 1998; Boano et al. , 2014) . The SWI is a dynamic ecotone that provides many ecosystem services, including its role in denitrification, or the microbially - mediated reduction of oxidized forms of N , most abundantly present as nitrate ( NO 3 - ) , to N 2 gas (Duff and Triska, 1990; Triska et al. , 1993) . While the 40 denitrification process is inhibited in the presence of dissolved oxygen, SWIs with significant stream water exchange and oxygenation have nonetheless been shown to beco me anoxic and create significant sinks of N (Findlay, 1995; Harvey et al. , 2013) . A byproduct of the denitrification reaction is nitrous oxide (N 2 O), which is an interm ediary product of the reduction of NO 3 - , and which can then be further reduced to N 2 . The fraction of denitrified N R that is released as N 2 O in sediments has been reported to be ~ 1% (Mulholland et al. , 2008; Beaulieu et al. , 2011) , however streams and rivers have been shown to account for 10 - 20% of the recent increase in atmospheric N 2 O concentrations due to anthropogenic activity (Seitzinger and Kroeze, 1998; Kroeze et al. , 1999; Beaulieu et al. , 2011) . This is of concern because N 2 O is a strong greenhouse gas, with 300 times the warming potential as CO 2 (Forster et al. , 2007) , and in addition it is a significant contributor to atmospheric ozone depletion (Ravishankara et al. , 2009; Syakila and Kroeze, 2011) . It has also been shown that N 2 O emission rates are higher in headwater streams relative to large rivers (Seitzinger and Kroeze, 1998; Marzadri et al. , 2017) , and though the causes for this are yet to be revealed, recent modeling suggests it can be attributed to SWI process ing of landscape N R (Marzadri et al. , 2017) . A key chal lenge in upscaling understanding of SWI function to entire fluvial networks, specific ally the contribution of SWI to N processing rates and NO 3 - , N 2 , and N 2 O export from headwaters , is the unique and dynamic interplay of reaction chemistry and hydrodynamics in the SWI (Triska et al. , 1993; Zarnetske et al. , 2012; Lansdown et al. , 2015; Liu et al. , 2017) . Adding to the complexity of e fforts to charac terize large - scale N processing in stream SWIs is that sediment conditions and stream flows can be extremely heterogeneous in space and time (Marzadri et al. , 2014) . Consequently, studies attempting to characterize SWI function s such as N process ing face difficulty in characterizing individual controls in natural settings b ecause the 41 mixing of reactants (e.g., nutrients, oxidants) between surface and groundwaters is fundamentally dependent on the direction and magnitude of exchange flows (Triska et al. , 1990, 1993; Brunke and Gonser, 1997; Zimmer and Lautz, 2014; Danczak et al. , 2016) , For example the physical exchange flow between streams and the SWI have been documented to vary up to 5 orders o f magnitude, and the fraction of total stream discharge passing through the SWI can be very high relative to surface flow depending on substrate and sediment depths (Boulton et al. , 1998; Anderson et al. , 2005; Tonina et al. , 2016) . This flow variability can be further divided among variability of flowpath length and residence times during both steady - state and variable flow conditions (Wörman et al. , 2002; Anderson et al. , 2005; Kaufman et al. , 2017) . The large range in f luxes and residence times in the SWI can be contrasted with documented changes in nutrient and reactant chemistry rates and timescales across catchments, which typically only vary 1 - 2 orders of magnitude (McGuire et al. , 2014; Abbott et al. , 2018; Ruhala et al. , 2018) . These large ranges in controls on biogeochemical reaction rates in the SWI are rar ely reconciled in field observations or experiments . There have been significant efforts by multiple disciplines to explore and model the function of the SWI across a range of spatiotemporal scales , but significant questions remain from this large body of research as to whether reactive versus hydrologic (transport) controls are most important to the fate of NO 3 - in streams (Sophocleous, 2002; Cardenas, 2015) . Consequently, there is a need for novel field investigation techniques that can estimate the relative importance of these cont rols to the fate of N O 3 - . Based on the documented range of variability in physical transport and reaction rate controls across stream SWIs that have been studied (Zarnetske et al. , 2012) , we hypothesize that physical parameters (hydrologic exchange timesca les ) are the master control on SWI biogeochemical function in N processing. To test this 42 hypothesis, we used a novel field method of controlled vertical exchange of known surface water chemistry through the SWI of a headwater stream that carries anthropogenically increased NO 3 - concentrations . Using 15 N - NO 3 - as a tracer , we monitored the transformation of stream water NO 3 - as it passed through SWI flowpaths . We hypothesized that under stable biogeo chemical inputs, changing the SWI rec harge flux rate , and thus porewater residence times , would result in substantial changes in aerobic respiration ( as indicated by oxygen removal ) and NO 3 - removal through the SWI . This study attempts to bridge results from controlled lab experiments on stream sediments (Quick et al. , 2016; Liu et al. , 2017) with more natural, but less well constrained, in - situ studies of NO 3 - processing (Zarnetske et al. , 2011a; Lansdown et al. , 2015) . Specifically, this study was conducted in a natural headwater stream SWI setting , but experimentally constrained NO 3 - tracer and flow conditions made it more feasible to assess how syst ematic ally chang ing hydrologic residence time controls biogeochemical functioning of the SWI using NO 3 - as a reactive solute . 43 Figure 14 : Map s of Massachusetts, the Ipswich River Watershed, and Sawmill Brook . ( A) S tudy region within Massachusetts, USA with the ( B) Ipswich River Watershed, showing the study s ite at Sawmill Brook and the nearest USGS stream g age ( 01101500 ) on the Ipswich River at South Middleton, MA. ( C) T opographic map of the Sawmill Brook study reach and site (marked with star) . Map units are kilo meters. Datum is UTM zone 19T. 44 3.2. Materials and Methods 3.2.1. Site Description The study reach and SWI experimental site are located in Sawmill Brook , which is a first order tributary to the Ipswich Rive r, in Burlington, Massachusetts ( Figure 14 ). The Ipswich River drains a watershed of 404 km 2 composed of mixed forest and urban land uses, and is unde rlain by Pleistocene glacial deposits (Carlozzi et al. , 1975; Briggs et al. , 2010) . This watershed has been previously investigated for N R removal in streams because there is significant N R contamination from the surrounding heavily urbanized headwaters (Williams et al. , 2004) . The site at Sawmill Brook ( Figure 15 ) drains a 4.1 km 2 watershed with 72% urban/residential land use, and 25% impervious surface cover (Wollheim et al. , 2005) . Figure 15 : Detailed Plan - view S ite Map of Sawmill Book. S tream level and banks were surveyed . Ring and (downstream) (see Figure 5 for ring details ). The groundwater ( GW ) and background ( ambient ) locations were sampled for groundwater and SWI porewater samples, respectively. The SWI ambient samples were taken at sediment depths of 2.5, 5, 7.5, 10, 15, 20 cm depths, and the groundwater sample was taken at 60 cm depth . The site labeled as stage ( Figure 14 ) shows the loca tion of our in - stream pressure logger . Map units are meters . Coordinates are for UTM zone 19T. 45 Figure 16 : Site picture from Sawmill Brook. View is looking east, downstream, with ring US in the foreground and ring DS in the background. Additional equipment not described herein, such as the orange wiring seen in the image , are associated with a concurrent geophysical (electrical resistivity imaging) study. This location was selected in part because it allowed us to conduct our SWI manipulation experiments in a stream that is relevant to previous extensive stream N studies, including the Lotic Intersite Nitrogen eXperiments II (LINX II), which were conducted there between 2003 and 2005 (Mulholland et al. , 2008, 2009; Hall Jr et al. , 2009) . The stream channel was highly incised into native glacial deposits ( Figure 16 ). Two dominant benthic sedi ment types were present at the site: 1) eroded sand and clay from the native surficial till, with high amounts of organic debris, and 2) deposits of road sand carried from road crossings upstream (see Figure 16 ). Due to the high percentage of impervious surface in the catchment, the stream is very flashy. A large precipitation event of approximately 75 mm on the evening of July 12 caused a fast response in the stream of approximately 0.46 m in stage ( Figure 19 ) , but the stage returned to close to base level within 6 h . 46 3.2.2. Experime ntal Setup The methods for this study are consistent with those presented in Chapter 2 and are only slightly modified for the stream SWI setting of this study. Hence, they are briefly reviewed here with highlights of the key differences between Chapter 2 a pproach and this study . Our study took place from July 10 18, 2017. The studied stream sediments were isolated from the surrounding environment using two open ended 55 - cm - diameter polyvinyl - chloride (PVC) barrels, which were installed in the stream bed a nd driven to a depth of 20 cm ; ensur ing vertical 1 - directional flow ( Figure 5 ). Individual sites were located by verifying downwelling or neutral flow from the stream to the groundwater, using a large 55 - cm - diameter steel seepage meter (Rosenberry and LaBaugh, 2008) . They were also selected to target the two benthic sediment types , with the upstream injection ring (Ring US) placed in the organ ic - rich sands , and the downstream injection ring (Ring DS) placed in the road sand deposits . The two rings were 2.1 meters apart ( Figure 15 ) . Immediate ly before the experiments were conducted, stream w ater was pumped into 1.9 - m 3 and 1.1 - m 3 holding tanks located on the stream bank ( Figure 17 ) . In both tank s , stream water was mixed with K 15 NO 3 (99% purity) to create the NO 3 - injectate enrichment of 5 atom - percent 15 N, assuming a background concentration of 96 µmol NO 3 - /L ( W. Wollheim, personal communication). In only one tank , stream water was also mixe d with sodium chloride (NaCl) to bring the total conductivity of the water from a background of approximately 1000 µS/cm to approximately 2000 µS/cm. This NaCl concentration was required to accurately observe breakthrough curves of the higher conductivity water moving through the sediments and thereby characterize hydrologic flow conditions during the experiments . Using two intermediate bucket s and a series of float switches , we were able to maintain a constant water level and hydrologic flux rate within ea ch PVC injection ring ( Figure 5 , Figure 17 ) . The water was pumped from each 47 holding tank to the intermediate bucket an d into the ring to enhance and control the rate of downwelling of surface water through the SWI within the ring. In this way, we were able to systematically create a range of stable hydrologic flux rates and resulting hydrologic residence times under which to conduct tracer experiments to test our hypotheses ( Figure 18 ) . An aerator was placed in the intermediate buckets to keep water oxygenated during w hat were sometimes long residence times in the bucket at low flux rates. There was no aerator in the tank. I nitially , the h ydraulic conditions in the ring were sustained with a hydraulic flux set to 3 m/d , which was repeatedly confirmed by monitoring the injection rate through the intermediate bucket ( Figure 17 ) . Flux rates were modified over four experiments ( Table 3 ) , from 2 m/d ( Experiment 1, July 11) to 3 m/d ( Experiment 2, July 1 4 ) , then 0.8 m/d ( Experiment 3, July 1 5 ) , and finally to 1.2 m/d ( Experiment 4, July 1 6 ) , which represents four realistic and systematic changes in SWI residence times, respectively. The storm event on July 12 briefly delayed the experiments, and stream conditions were allowed to stabilize before experiments resumed. During each experiment, h igh conductivity water from the NaCl - labelled tank was injected, until porewater conductivity up to 20 cm depth within the injection ring stabilized, and then the water in the ring was immediately evacuated and quickly filled with lower conductivity water, which then continued to flow into the ring at the same flux rate, allowing for assessment of two conductivity breakthrough curves at each flux rate. A combination of measured injection ring flux rates and specific conductivity (SpC) breakthrough curves wa s used to determine vertical flowpath residence times throughout the experiments, with the porewater velocity for each experiment was derived from the average of the injection and flux curves (detailed below in Section 3.2.5 ). 48 Table 3 : Details of the Sawmill Brook Experiments. The measured hydraulic flux through the ring is reported for each experiment. Experiment Flux Rate (m/d) Injection time and Sampling time (Y - M - D H:M) Experiment Duration (h) Experiment 1 2.0 2017 - 07 - 11 19:35 2017 - 07 - 12 13:10 17.6 Experiment 2 3.0 2017 - 07 - 14 13:15 2017 - 07 - 15 08:40 19.4 Experiment 3 0.8 2017 - 07 - 15 09:45 2017 - 07 - 16 13:10 27.4 Experiment 4 1.2 2017 - 07 - 16 21:30 2017 - 07 - 17 09:45 12.3 Figure 17 : Schematic of Tracer Addition at Sawmill Brook . Before the experiments, water from the stream was pumped into two holding tank s . 15 NO 3 - was added to both tank s , and NaCl to only one. Using a series of pumps and float switches (FS) to maintain steady water levels, water was pumped into an intermediate bucket and then into the injection ring. An aerator was in the intermediate bucket to keep water oxygenated durin g sometimes long residence times in the bucket at low flux rates . There was no aerator in the holding tank. 49 Figure 18 : Conceptual Diagram of Sawmill Brook Experimental Modifications. The three major natural controls on SWI Function on N are microbial community composition, porewater reactant chemistry, and hydrologic transport conditions. By confirming with a 15 NO 3 - tracer that a microbial community capable of N transformations wa s present and seeking to isolate the role of reactant chemistry on N removal, Experiments 1 - 4 systematically changed hydrologic flux, with two experiments exploring an oxic or anoxic regime , to observe changes in N removal as a result of changes in hydrologic controls. Site stream flow conditions were monitored with a Levelogger pressure - temperature datalogger (Solinst, ON, Canada) . The datalogger was suspended in the stream water column at the location indicated in Figure 15 . This logger provided high - resolution information on stage, but also stream - water total conductivity (TC) and temperature. Measurements of TC and temperature were also spot - checked and validated using a OrionStar handheld probe (Thermo Scientific, MA, USA). 3.2.3. Porewater Sampling Methods Measurement of dissolved oxygen (O 2 ) concentration and temperature within the ring water was accomplished by two MiniDOT loggers (Precision Measurement Engineering, CA , USA) suspended beneath the surface. Porewater sampling methods were identical to those presented in Section 0 . Briefly reviewed here, porewater s ampling at depth was achieved using four stainless steel minipoint samplers, similar to the USGS MINIPOINT system utilized by 50 Harvey et al. (2013) . These had an outer diameter (OD) of 3.2 mm, a screened interval 10 mm long, with 3 individual slits 0.4 - 0.7 mm in w idth. The samplers were driven to depths of 5, 10, 15, and 20 cm below the stream sediments within each ring. 3.2 - mm OD tubing was attached to the end of each minipoint, and the 10 and 20 cm sample depths were fed through two electrical conductivity (EC) m icro flow - automatically by the micro flow - cell control units using sensor - specific calibrations performed at the beginning of the experiment. Flow - through cells equipped with fiber - optic oxygen microsensors attached to a FireStingO2 Optical Oxygen Meter (Pyro Science, Germany) were in - line with the tubing from the 10 and 20 cm MINIPOINTs . For each experiment, porewater sampling took place once the lower conductivity flush was at steady - state (i .e., reaching plateau curve) . Water sampling was accomplished in a closed system by peristaltic pumps (Cole - Parmer, IL, USA), and water was collected in syringes, followed by immediate filtering through a 0 .7 µm glass fiber filter and 0.2 µm cellulose acetate filter into acid washed amber HDPE bottle s (Nalgene, NY, USA). Water samples were chilled at 4ºC on site and then frozen within 8 h of collection . For dissolved gas samples, 1.6 mm OD tubing directly fr om the pump was placed into the bottom of a 12 mL glass E xetainer (Labco, United Kingdom) and over - filled for two full volumes before collecting the sample volume . These gas s amples were preserved with 120 µL of 50% w/v z inc c hloride solution. Preserving a convex meniscus, the tubing was removed, and the cap was screwed on to prevent any air bubbles in the sample. Gas samples were stored at room temperature in the dark, and later shipped to the Stable Isotope Facility (SIF) at the University of California, Davis, for isotope ( 15 N) analysis of dissolved gases (N 2 and N 2 O). Water samples were later separated into groups for 15 N analysis of NO 3 - , which were shipped frozen to the SIF, 51 and for nutrient, anion, and carbon analysis, which were never frozen but kept chilled at 4ºC during shipment and prior to analysis at Michigan State University. 3.2.4. Laboratory Analyses Stable isotope ratios of nitrogen ( 15 N) in gas were measured using a ThermoScientific GasBench + Precon gas concentration system interfaced to a Thermo Scientific Delta V Plus isotope - ratio mass spectrometer (Bremen, Germany). Nitrate in water samples was converted to N 2 O by the bacteria denitrification assay and 15 N ratios were measured as stated above. Anions were measured by a Dionex ICS - 2100 Ion Chrom atography System (ThermoScientific , MA, USA ), producing concentrations for chloride (Cl - ), nitrite (NO 2 - ), nitrate (NO 3 - ), and sulfate (SO 4 2 - ). Non - purgeable Organic Carbon (NPOC), and Total Nitrogen (TN), were measured using a TOC - L total organic carbon a nalyzer (Shimadzu , Japan ) catalytic oxidation at 720ºC followed by gas chromatographic measurement of CO 2 and chemiluminescence measurement of N O . 3.2.5. Calculations of Residence Time and Reaction Rates The SpC was measured at 10 and 20 - cm depth within each of t he two injection rings. For each of the four experiments, flux was varied, and the injection of higher conductivity water resulted in one breakthrough curve , and the following flush of lower conductivity water resulted in a second ( see Figure ) . Median porewater velocities (see Table 9 ) for each injection/flush were calculated by subtracting the two median times of arrival and dividing by the separation distance. The average of the injection/flush velocities for both 10 and 20 cm sampling depth for each ring was used as the velocity to calculate residence time at 5, 10, 15, and 20 - cm depth s for each experiment. Solute removal rates were calculated as the linear regression of change of concentration over median residence tim e, as in Section 2.2.5 . Rates, unless otherwise specified, are calculated 52 between surface water (0 cm) and the 20 cm depth. Calculations of d enitrif ication rates are outlined in Section 2.2.5 . For each ring and experiment, residence time with depth was also transformed using the O 2 removal rates to calculate Damköhler numbers, consistent with the method of Zarnetske et al . (2012) , which help illustrate if the SWI potential for bulk oxic or anoxic ( <63 µmol O 2 /L) conditions is limited by transport or reaction timescales. For each ring and ex periment, the removal velocity of O 2 ( ) was calculated as the slope of the linear regression of the natural log of O 2 concentration over time. The removal velocity is equivalent to the 1 st order decay constant . F or each depth, Damköhler values ( ) were calculated by multiplying the residence time ( ) at each depth by the O 2 removal velocity: ( 10 ) was calculated by the same method, using a linear regression of the log of NO 3 - concentration over time to calculate the NO 3 - removal velocity. 3.3. Results 3.3.1. Hydrologic and Chemical Setting Stream water chemistry remained relatively steady across the experimental periods ( Figure 18 ) . Concentrations of NO 3 - and DOC were on average 67 ± 8 µmol NO 3 /L and 490 ± 80 µmol DOC/L ( Figure 19 ) during the experimental periods . Background stream - water conductivity was approximately 1000 µS/cm . S tream Cl - concentrations were strongly correlated with stream water TC (R 2 >0.99) , suggesting that variation in TC was largely due to the variable influence of road salt application as NaCl in the watershed s (Kaushal et al. , 2018) . Concentrations of Cl - in stream water were on average 8000 µmol/L, as opposed to groundwater concentrations of 1400 µmol/L, providing a good contrast between the two waters e xchanging through the SWI at the site. Groundwater contained an average of 21 ± 2 µmol NO 3 /L and 355 ± 53 44 µmol DOC/L. The storm event on July 12 resulted in a large increase of stream stage , corresponding with dilution of both stream - water TC and NO 3 - , and an increase in DOC , and concentrations slowly returned to the pre - event conditions, while stage recovered much faster ( Figure 19 ). Removal of stream water to fill the tanks before mixing injectate for the remaining experiments was delayed until stream stage and chemistry returned to previous conditions . This is reflect ed in the concentrations of the tank water being highly stable, relative to the stream water ( boxplots: Figure 19 ). Concentrations of the tank water th at was injected into the rings were on average 77 ± 5 µmol NO 3 - /L and 450 ± 70 µmol DOC/L, suggesting that after mixing with our tracers, the injectates were representative of background chemistry measured in the stream during the experimental periods ( Figure 19 ). Tank concentrations of water drawn from the stream did not change considerably before and after the storm. Shifts in the concentrations of NO 3 - were minimal between the holding tanks and surface water in the injection rings. Nitrate c oncentrations within the surface water of the rings were on average 76 ± 4 µmol/L. An increase in DOC concentrations was observed from the tanks to the ring surfac e water, with an average ring concentration of 520 ± 100 µmol DOC/L. The higher carbon concentrations could have been due to leaching of carbon from leaves and woody debris captured within the rings at the sediment - water interface, which was more prevalent in Ring US. Concentrations of dissolved O 2 in the injection ring water were on average 220 ± 40 µmol/L ( Figure 19 ), and stayed above 200 µmol/L durin g the sampling periods (red bars: Figure 19 ) except for in ring US during Experiment 4 (flux 1.2 m/d), when the concentration dropped to 140 µmol/L. Th is was due to a temporary overnight failure of the aerator system within the intermediate bucket that kept the injection water oxygenated. 54 Exp 1 Exp 2, 3, 4 A) B ) C ) D) E) Figure 19 : S tream stage and chemical conditions during sampling at Sawmill Brook. (A) Stage monitored at the site on Sawmill Brook over the study dates. Sampling periods of the four experiments are shown by red bars, with the commencement of each injection as the beginning of the bar and the end of sampling for that experiment as the end of the bar . Experiment order was 2 m/d, 3 m/d, 0.8 m/d, 1.2 m/d. Total conductivity (TC ; not temperature corrected ) from the 55 site surface water is also sh own , with points showing spot - checks of TC with a handheld probe. ( B ) NO 3 - concentrations in the stream, shown b y connected lines, and concentrations in the tank Boxplot s on the right show the median and interq uartile range for samples collected in the stream, tanks, and rings. Whiskers are to the minimum and maximum. Points are outliers. ( C ) DOC concentrations as shown in panel C. (D) Oxygen (O 2 ) saturation and concentration are shown for the surface water from the two SWI rings. The boxplot shows O 2 during the sampling periods (red boxes). (E) Temperature as monitored at the stream stage site and within the two SWI rings. 3.3.2. Solute Removal with Depth During each of the four hydrologic flux rate experiments , Ring s US and DS showed similar behavior . The O 2 concentrations at all depth s during Experiments 1 - 2 ( higher fluxes of 3 and 2 m/d) remained greater than 50 µmol O 2 /L ( Figure 20 ). During Experiments 3 - 4 ( lower fluxes of 1.2 and 0.8 m/d) porewaters were anoxic ( <63 µmol O 2 /L) at 5 cm and deeper, except for in ring DS at 0.8 m/d flux, where the 5 cm concentration was 2 8 µmol O 2 / L but was 0.0 µmol O 2 /L at 10 cm and deeper. Removal of NO 3 - showed similar trends to O 2 , with an average percent removal of 33% of the injection concentration for Experiments 1 - 2 , and 79% for Experiments 3 - 4 . Concentrations of DO C did not sugges t strong removal , and concentrations increased by 34% on average from the injection concentration. Removal of DOC was only seen for the 3 m/d flux rate (Experiment 2) , with 19% removal on average for the two rings . When accounting for changing residence times with hydrologic flux, the linear removal rates of O 2 between 0 and 20 cm depths showed similar behavior between Experiments 1 and 2 and between Experiments 3 and 4 but increased with increasing residence times. Depth - integrated r emoval rates of O 2 de creased from 112 during Experiments 1 - 2 to 70.6 µmol O 2 /L/h during Experiments 3 - 4 ( Figure 21 ) , but O 2 depletion occurred at very shallow depths in Exp eriments 3 - 4, and O 2 removal rates between 0 and 5 cm were 78% higher in Experiments 3 - 4 . There were slight differences between rings US and DS, with 38 % higher O 2 removal rates on average for ring DS than US for each flux rate. Removal rates of NO 3 - were 7.7% higher for 56 Experiments 3 - 4 than Experiments 1 - 2 ( Figure 21 ). In contrast to the trend between the two rings for O 2 removal, ring US had NO 3 - remov al rates that were 47 % higher than those in ring DS for each flux experiment, and the difference was most pronounced during Experiments 1 - 2 (i.e., shorter residence times), with 76 % higher values than in ring DS on average, than for Experiments 3 - 4 (i.e., longer residence times), at only 29 % higher than in ring DS ( Figure 21 ). A) B) Figure 20 : Concentration over depth and residence time from Sawmill Brook experiments. Concentrations of O 2 , NO 3 - , and DOC over flowpath length (A) and over porewater residence time (B). Symbols denote data from R ings US and DS . T he color of each line and point denotes the flux rate (see inset legend in the lower left panel ), where l ighter shade denotes the lower flux rates. Error bars are the standard deviation of concentration whe n repeated samples were possible . 57 Figure 21 : Removal rates for O 2 , NO 3 - , and DOC from the Sawmill Brook Experiments . Symbols denote ring US or DS. Error bars are based on the standard deviation of concentrations used in the rate calculation s , where multiple samples are available. More negative values indicate increasing removal rates. Positive values indicate accumulation along the flowpath. The NO 3 - tracing showed e vidence for denitrification during each hydrologic flux rate experiment . The N 2 concentrations were on average 450 ± 60 µmol N 2 /L, which would be predicted by water i n equilibrium with the atmosphere at a pressure of 1 atm and a water temperature of 30°C (Weiss, 1970) . Temperature measurements within the injection ring water , however , suggested that average temperature was 19.4 ± 2.4°C during da ytime hours ( Figure 19 ). Temperatures during evening hours were warmer on average , at 20.4 ± 2.9°C. There was no notable difference in temperature bet ween the two rings (average difference 0.18 ± 0.45°C). Water temperatures in the tanks were not monitored but are expected to have varied on a diel cycle, resulting in the potential for slight degassing , especially of O 2 and N 2 , from tanks. H owever , O 2 concentrations measured continuously in the injection rings indicated that the aerators in the intermediate bucket s and mixing upon entering the injection rings kept water stable and sufficiently oxygenated ( 229 ± 31 µmol O 2 /L) . Along each SWI flowpath, t he N 2 concentrations increased on average 2.2%, however , in one instance, the concentration was observed to decrease by 11% in ring DS during Experiment 1 (2 m/d flux) , and was observed to increase by 14% in ring US during Experiment 2 ( 3 m/d flux; Figure 22 ). 58 A) B) Figure 22 : 15 N tracer conditions of N 2 and concentrations of N 2 O from Sawmill Brook Experiments. 15 N isotopic enrichment of N 2 and concentrations of N 2 O over flowpath length (A) and over porewater residence time (B). Symbols denote data from Rings US and DS. The color of each line and point denotes the flux rate, where lighter shade denotes the lower fl ux rates. Error bars are the standard deviation of concentration when repeated samples were possible. 59 Figure 23 : Production rates of N 2 and N 2 O from Sawmill Brook Experiments . Symbols denote ring US or DS. Error bars are based on the standard deviation of concentrations used in the rate calculation, where multiple samples are available. Positive values indicate accumulation along the flowpath. Despite relatively small changes in total dissolved N 2 mass, enrichment of 15 N 2 was observed along each flowpath, especially for Experiments 3 - 4 (long residence times), with enrichment increasing to 0.494 atom % on average; higher than for Experiments 1 - 2 with an average of 0.374 atom %, which is closer to natural abundance of 15 N. In ring US for Experiments 1 - 2, 15 N 2 was present at 20 cm depth with an abundance of 0.408 and 0.421 atom %, respectively ( Figure 22 ). The same decrease in total N 2 mass along the flowpath for ring DS during Experiment 1 was reflected during the same experiment in a 6.8% decrease in 15 N 2 mass. During Experiment 2 for both rings and Experiment 1 only for ring US, 15 N 2 mass increased by 14% on average from 0 to 20 cm. The increase was larger for Experiments 3 - 4 for both rings, by 60% on average. These increases in 15 N 2 mass are reflected in denitrification rates, which were on average 1. 3 µmol N 2 - N/L/h for Experiments 1 - 2, and 4.3 µmol N 2 - N/L/h for Experimen ts 3 - 4 ( Figure 23 ). The N 2 O concentrations increase d with depth for all experiments ( Figure 22 ) . C oncentrations of N 2 O closely mirrored 15 N - N 2 O enrichment patterns, which generally increased 60 from 1 to 7.5 atom % for most experiments, with the exception of ring DS during Experiment 2 ( 3 m/d flux ) enrichment only rose to about 5 atom %, and during Experiment 4 ( 1.2 m/d flux ) where enrichment decreased with depth from over 7 to below 1 atom % ( Figure 22 ). Concentrations of N 2 O were higher in ring US th an ring DS for each experiment by 2.4 - fold on average, and peak concentratio ns for each experiment were 4.9 - fold higher on average in ring US. Concentrations of N 2 O increased with depth for all experimen ts, between 0 and 20 cm, by 7 - fold on average. T his increase with depth was greater for ring US, at a 7 . 9 - fold average increase, than ring DS at a 6.1 - fold average increase. Despite this, for Experiment 4 ( 1.2 m/d flux ) in both ring s , tota l N 2 O concentrations decreased from 0 to 20 cm, however 15 N - N 2 O enrichment suggested that for ring US in Experiment 4 there was still net N 2 O production via reduction of the injected 15 NO 3 - tracer. Production of N 2 O via denitrification was highest for ring US in Experiment 1 at a rate of 0. 27 µmol N 2 O - N/L/h ( Figure 23 ). For Experiments 2 - 4 for all rings and Experiment 1 only in ring DS , the N 2 O production rate was on average 0.0 39 ± 0. 054 µmol N 2 O - N/L/h. Rates were higher in ring US than in ring DS , at 2.3 - and 9.5 - fold higher for Experiments 1 and 2 ( 2 m/d and 3 m/d fluxes) , respectively. The N 2 O production rates for Experiments 3 - 4 were not substantially different between the two rings ( 5.8 % difference on average). N 2 O production rates were much lower during Experiments 1 - 2 than during 3 - 4 , at 0. 01 ± 0. 06 and 0. 13 ± 0. 11 µmol N 2 O - N/L/h, respectively. Although N 2 O production rates were small, there were distinct patterns with depth for N 2 O. For Experiments 1 - 2 , the highest N 2 O concentrations with depth f or each experiment were at 20 cm. The 15 N - N 2 O enrichment peaked at 5 cm - depth for Experiment 4 (1.2 m/d flux) , and at either 10 or 15 cm - depth for Experiment 3 (0.8 m/d flux) . For specific 5 cm depth intervals within each of the injection rings, the maximu m N 2 O production rates were not substantially different between the averages of Experiments 1 - 2 61 and 3 - 4 (1.3% different). For Experiments 1 - 2 , 5 cm interval production rates were highest in the deeper portion of the sediments, while the highest rates were in the shallowest depth interval for Experiments 3 - 4 . The N 2 O removal rates showed larger differences between Experiments 1 - 2 and 3 - 4 , with the largest 5 c m - interval removal rates 2 7 - fold higher on average for the Experiments 3 - 4 than for the Experiments 1 - 2 . Production of N 2 accounted for 14% of NO 3 - removal on average for Experiments 1 - 4, and production of N 2 O accounted for 0.4% of NO 3 - removal on average. For production of N 2 O, this percentage was higher in Experiments 1 - 2 (0.7%) than in Experiments 3 - 4 (0.09%), whereas for production of N 2 the reverse was true. During Experiments 1 - 2 production of N 2 accounted for 5% of NO 3 - removal, but this increased to 22% on average for Experiments 3 - 4. The proportion of NO 3 - removal accounted for by total denitrification (N 2 + N 2 O) increased from 6% to 22% on average from Experiments 1 - 2 to Experiments 3 - 4. The e stimated proportion of NO 3 - removal unaccounted for by denitrification then ranged from 75% (Experiment 3 ring DS) to 94% (Experiment 1 ring US). 3.3.3. Scaling by Residence Time Residence times with depth for each experiment were calculated by assuming a uniform porewater velocity through the sediments for each experiment in each ring , which is possible given the constrained flow field within the ring . Th at velocity was calculated from the average of the injection and flush breakthrough curves of high conductivity water at both 10 and 20 cm depth in each ring. The total residence time of each ring, as the residence time from 0 to 20 cm depth, increased wit h decreasing flux rate. For ring US, total residence time increased approximately linearly with decreasing flux rate, while for ring DS total residence time increased exponentially with decreasing flux rate. For ring US, total residence times increased fro m 1.3 h 62 at 3 m/d flux to 3.2 h at 0.8 m/d flux, while for ring DS total residence times increased from 1.0 h to 4.4 h ( Figure 24 ) . A 3.7 - fold decrease in flux resulted in a 2.5 - fold increase in residence times in ring US, but in a 4.4 - fold increase in ring DS. Assuming seepage flux is directly proportional to porewater velocity by a factor of the porosity, this exponential increase in residence times would be expected for a uniform porosity: as flux approaches zero , residence times will approach infinity. Figure 24 : Residence time over flux rate in the Sawmill Brook experiments. Residence time is shown at 20 cm depth, or the bottom of the SWI flowpath, for each flux rate experiment. In Experiments 1 - 2 , with porewater residence times up to 2 h, O 2 concentrations remained in the oxic range (>16 mol/L) , whereas in Experiments 3 - 4 , anoxia was achieved at residence times as early as 0.7 h ( Figure 20 ) . O 2 did not show a threshold res ponse where after the same specific residence time, concentrations were fully anoxic (as in Briggs et al ., 2015). For NO 3 - , removal was governed by different removal velocities (see Section 3.2.5 ) and increased with increasing porewater residence times. Concentrations of 15 N - N 2 increased with residence time. Concentrations of N 2 O and 15 N - N 2 O were higher at intermediate residence times, generally around 1 to 3 h , but for some experiments concentrations peaked earlier and decreased along the 63 flowpath, and , for ring DS during Experiment 3 , concentration peaked at 3.3 h and then decreased only by half at 4.4 h . Figure 25 : Concentrations over Damköhler number for O 2 from Sawmill Brook Experiments . Concentrations of O 2 , NO 3 - , N 2 O 15 N 2 over the Damköhler number for oxygen removal ( ) , calculated for each ring in each experiment. The Damköhler number ( ) was used as a nondimensional approach to scale residence times by the O 2 removal velocity , using Equation 10 . The approach more clearly illustrates the threshold patterns in N processing rates . Average O 2 removal velocity was 3.5 - fold higher for Experiments 3 - 4 than for Experiments 1 - 2 . This res ulted in compression of the O 2 concentrations for Experiments 1 - 2 in early space ( Figure 25 ), with values up to 1.5, while Experiments 3 - 4 extended into values from 7.5 to 11.5. The O 2 concentrations , when scaled by , show a characteristic removal pattern , meaning that anoxia occurs after a 64 specific threshold. =1 means O 2 supply and demand timescales are in unit y and larger values indicate conditions conducive to anoxia (Zarnetske et al. , 2012) . A noxia occurred at a threshold value =1.9 , beyond which 15 N 2 enrichment increased the most, along with decreases in NO 3 - concentrations. Removal velocities for NO 3 - were also calculated ( Figure 26 ) . Although removal velocities represented by a first - order reaction k value are generally thought to be more represen tative of biogeochemical processes in the SWI (Hedin et al. , 1998; Zarnetske et al. , 2012) , we found that removal rates for NO 3 - as calculated by Lansdown et al. (2015) were appropriate representations of SWI function . The relationship between O 2 removal rates and removal velocities was less clear, hence examining the data in space . Figure 26 : Removal rates over Removal V elocities for O 2 ( left) and NO 3 - ( right ) from Sawmill Brook Experiments . Removal rates are reported as in Figure 21 and calculated in Section 2.2.5 and Removal velocities are calculated as in Section 3.2.5 . Blue lines are linear regression lines. Circles are from ring DS and triangles are from ring US. Resulting values ranged from 0.16 to 0.59 for the Experiments 1 - 2 , and from 1.03 to 1.65 for Experiments 3 - 4 . The N 2 O concentrations did not show a relationship to or , as for each experiment the peak N 2 O concentration occurred at an intermediate r anging from 0.7 to 8.5. The average peak N 2 O concentration in both rings for Experiment 1 was 65 6% higher than the average peak concentration for Experiment 3, and the outflow was 3.4 - fold higher in Experiment 1 versus Experiment 3 . Thus , under t h e condition s of Experiment 1 , with shorter residence times, there was a daily export of 948 µmol N 2 O - N/d per ring , which is 8 .5 - fold greater than the export for Experiment 3 , which had the longest residence times. 3.4. Discussion 3.4.1. Biogeochemical Reaction Rates in the SWI C ontrolled by Residence Time Our study shows that the biogeochemical function of the stream SWI, especially for NO 3 - removal , is controlled by residence time when reactant concentrations in input water are stable . D ecreasing SWI hydrologic fluxes by 73% res ulted in longer residence times , which increased removal of O 2 and NO 3 - and increased rates of production of denitrification end - products ( Figure 21 , Figure 23 ). This same increase in NO 3 - removal and denitrification rates with increasing residence times has been previously observed in laboratory per fusion column studies (Gu e t al. , 2007; Bourke et al. , 2014; Liu et al. , 2017) , where controlled manipulations of hydrologic flux were paired with stable simulated river water influx chemistry . In our study, th e removal rates of NO 3 - increased with increasing residence times as predicted by previous experimental and modeling studies (Hedin et al. , 1998; Marzadri et al. , 2011; Zarnetske et al. , 2012; Quick et al. , 2016) . Unlike these previous studies that were unable to fully decouple biogeochemical (Hedin et al. , 1998; Zarnetske et al. , 2012) and hydrologic variability or conduct their work in field settings (Marzadri et al. , 2011; Quick et al. , 2016; Liu et al. , 2017) , we were able to clearly confirm and quantify the impact of variable residence time on the fate of O 2 and NO 3 - in a stream SWI. Overall, concentrations of NO 3 - decreased consistently once entering the SWI , showing no threshold response related to co - occurring oxic conditions. Removal of NO 3 - was greatest in 66 shallow sediment intervals ( Figure 20 ) , supporting previous findings in other SWI sediments that indicated that the very upper most sediment interval is the most reactive for NO 3 - (Inwood et al. , 2007; Harvey et al. , 2015) . Concentrations of O 2 and NO 3 - both showed dec lines with depth that were characteristic of a 1 st order removal reaction , but the removal velocity was not constant across our experiments ( Figure 20 , Figure 26 ). Nitrate removal was consistently enhanced by increased residence times. Removal rates of NO 3 - , production rates of N 2 , and the proportion of removed NO 3 - explained by production of denitrification end - products (N 2 and N 2 O) also increased when the S WI sediments were fully anoxic below 5 cm , which occurred during Experiments 3 - 4 ( fluxes of 0.8 and 1.2 m/d). The proportion of NO 3 - removal accounted for by N 2 + N 2 O production increased with residence times, from 6% to 22% on average from Experiments 1 - 2 to Experiments 3 - 4 . Still, up to 94% of the NO 3 - removal was unaccounted for by denitrification end - products across the experiments , suggesting alternative pathways for removal of NO 3 - in the SWI. Biologic al assimilation may account for a large pro portion of observed NO 3 - removal during our experiments. Our findings that most of NO 3 - removal was unaccounted for by denitrif ication ha s also been observed in most other studies in streams. For example, Lansdown et al . (2012) found that up to 87% of 15 NO 3 - removal observed during sediment incubations could be accounted for by biological assimilation. Mulholland et al . (2008) also found that , across 72 streams, the total stream reach denitrification account ed for less than 16 % of total NO 3 - removal at over half of their sites. Even c ontrolled mesocosm studies , where there is a large degree of control and precision for the N budgets, have found that denitrification could not account for 4 0 - 7 0% of NO 3 - removal (Stelzer et al. , 2015) . I t is possible that once NO 3 - was assimilated into biomass, that N could be re - mineralized and ultimately denitrified (Seitzinger et 67 al. , 2002; Hall Jr et al. , 2009) . In addition, certain sulfur bacteria are known to be able to take up and store NO 3 - for later use in dissimilatory metabolic transformations (Burgin and Hamilton 2007). Unfortunately, measuring all these different N poo ls was outside of the scope of this work because this study was focused more on the complete N r removal pathway of denitrification. Lastly, Wollheim et al . (2005) also show that at Sawmill Brook study site that , despite high N R loading, 65 - 85% of N R is retained in the catchment . Our study suggests, that the SWI of Sawmill Brook is effective at removing NO 3 - , but that only a small fraction of the removal is likely due to immediate and direct denitrification in the top 20 cm of the SWI. 3.4.2. Role of Anoxic Microzones and POC Interesting ly , w e documented large fluxes of both denitrified N 2 and N 2 O from our studied sediments where porewater O 2 concentrations were bulk - oxic ( >50 µmol O 2 /L ) , adding to the growing list of studies documenting this phenomenon of an ae r obic microbial metabolism occurring in bulk oxic conditions , as spe cifically related to leaf particulates in soils (Kravchenko et al. , 2017, 2018) , and further implicating anoxic microzones as important denitrification sites in stream sediments (Triska et al. , 1993; Zarnetske et al. , 2011a; Harvey et al. , 2013; Briggs et al. , 2015; Kravchenko et al. , 2017) . We also confirmed results from studies suggesting that N 2 O emissions would peak at intermediate residence times in sediments (Firestone and Tiedje, 1979; Firestone et al. , 1980; Quick et al. , 2016; Liu et al. , 2017) . Not only did N 2 O concentrations peak at intermediate residence times in each depth profile ( Figure 22 ), but N 2 O production rates were also highest at intermediate system residence times, as in the 2 m/d flux rate experiments ( Figure 23 ). During Experiment 1 ( 2 m/d flux rate ) , N 2 O production represent ed the largest percentage of NO 3 - removal of any of the experiments, at 1.2%. In contrast , N 2 production was the highest proportion of NO 3 - re moval during Experiments 3 - 4 , at 22%. Despite higher NO 3 - removal rates 68 during Experiments 3 - 4 , N 2 O export was 7.7 - fold higher from the rings during Experiments 1 - 2 . N 2 O export was also 10 - fold higher on average from ring US than from ring DS. This corresponded with 53 % higher NO 3 - removal rates on average in ring US for each flux rate, and with observations of woody debris and POC in the stream sediments beneath ring US. It is unlikely that the POC in ring US contributed to much more DOC production, or contributed to increases in O 2 removal (Stelzer, 2015) , b ecause DOC concentrations showed accumulation on average along the flowpath, but no difference in a ccumulation between the rings . Buried POC in stream sediments have been shown to be important to enhancing microbial activity even when not contributing to higher DOC concentrations (Sobczak e t al. , 1998) . Despite the POC in ring US, ring DS was observed to have O 2 removal rates 2 9 % higher on average across the experimental flux rates. Instead, POC with in the sediment matrix likely reduced the SWI porosity and thereby promoted the presence of reactive anoxic microzones (Sexstone et al. , 1985; Briggs et al. , 2015; Kaufman e t al. , 2017; Kravchenko et al. , 2017) , thus indirectly contributing to the higher NO 3 - removal and N 2 O emissions. During Experiment 1, the POC in the sediments of ring US may have leached the most labile forms of carbon first (Stelzer et al. , 2014) because this was the experiment with the largest hydrologic flux , however , from Experiment 1 to 2 O 2 removal rates increased by 73 % for the t wo rings, suggesting O 2 removal was not limited by labile DOC supply via leaching of POC. N 2 O production is known to be promoted by more recalcitrant DOC sources, where the reduction of N 2 O to N 2 is limited by DOC supply, leading to N 2 O export (Burford and Bremner, 1975; Quick et al. , 2016) . Th is further suggests that the high production rates of N 2 O during Experiment 1 were not due to DOC sources and were likely instead due to increased abundance of anoxic microsites. Based on the SWI pore - network model of Briggs et al. (2015), an increase in f lux 69 rate (as in Experiment 2 , increase from 2 to 3 m/d) will cause a decrease in the proportion of reactive less mobile porosity, specifically in the proportion of flow - dependent less mobile porosity (i.e., flux rate tends to directly correlate with the ab undance of microzones) . With fewer anoxic microsites, production of N 2 O would decrease , as was observed in Experiment 2. With the onset of complete anoxia at depth in Experiments 3 - 4, the further reduction of N 2 O to N 2 was not limited by reaction site abundance, and would be controlled at the flowpath scale by electron donor abundance (Betlach and Tiedje, 1981; Quick et al. , 2016) . Production of N 2 was highest at the longest residence times ( Figure 23 ) , suggesting more complete denitrification (i.e., higher N 2 /N 2 O ratios) can be achieved with additional exposure time (Marzadri et al. , 2017) . Although we did not monitor porewater deeper than 20 cm along each flowpath, due to limitation of the experimental apparatus, these shallow vertical flowpaths produced by the rings provide an analogue for hyporheic flowpaths of comparable length or residence time. Contrary to the assumptions of many models incorporating hydrodynamics and biogeochemical processes (e.g., Marzadri et al. , 2014) , denitrification wa s not inhibited in bulk - oxic sediments, and N 2 O production can actually be highest from flowpaths of intermediate length (Quick et al. , 2016) that may be mostly bulk - oxic ( Figure 23 ). This is a key finding that highlights N processing conditions that are poorly characterized or not even accounted for in current stream N and SWI studies and models. 3.4.3. Implications of Dynamic Stream Hydrology for N export Variable flow events , such as those produced by storm induced runoff , have been shown to result in increased chemical mixing of stream SWIs (Zimmer and Lautz, 2014) , however the net effect on SWI exchang e flow is less clear, with some studies suggesting that exchange flow can either increase or decrease at higher stream discharges depending upon a large range of 70 hydrogeomorphic conditions in the stream sediment and reach (Wondzell and Swanson, 1996; Wondzell, 2011; Boano et al. , 2014) . Still o ther studies have documented no relationship between hyporheic exchange rates and discharge (Ward et al. , 2012, 2016) , and many studies have found large spatial variability in SWI flux around specific reach features (Lautz and Fanelli, 2008; Briggs et al. , 2013; Smidt et al. , 2015) . As we saw from the hydrologic data from our experiments, there were different relationships between biogeochemical flux es and residence times within our two SWI rings ( Figure 24 ), and sometimes large disparities of O 2 , NO 3 - , and denitrification reaction rates at the same hydrologic flow - through rates ( Figure 21 , Figure 23 , Figure 26 ). This heterogeneity has been the source of inquiry for several empirical m odels (Marzadri et al. , 2014; Tonina et al. , 2016) suggest ing that heterogeneity of streambed and reach - scale morphology unto itself has a large influence on hydrologic residence time distribu tions as well as NO 3 - removal and N 2 O production. Recent modeling incorporating process - based particle - tracking simulations by Li et al. (2017) starts to address this role of different scale s of heterogeneity by showing that spatial heterogeneity of SWI flow s and exchange s with surface water can enhance both NO 3 - removal and denitrification. While it may be true that net SWI flow is relatively small compared to overall stream flow out of watersheds in all but the smallest and steepest stream reaches (Wondzell, 2011) , our results show that even if a small fraction of SWI flowpaths achieve s the appropriate balance of reaction and transport timescales , rates of NO 3 - removal and production of the potent greenhouse gas, N 2 O , will be high . Th e existence of many SWI flow path with NO 3 - removal and N 2 O production along stream reaches may explain the high levels of dissolved N 2 O that have been previous observed at the reach scale (Mulholland et al. , 2008; Beaulieu et al. , 2011; Wollheim et al. , 2014) . This conclusion has previously been suggested by Quick et al. (2016), appropriately 71 r eferencing a where the balance of the controlling timescales lead s to flowpaths that are net exporters of N 2 O, whereas flowpaths that are too short are generally not anoxic enough and thus not conducive to denitrification, and flowpaths that are too long become net consumers of N 2 O. SWI exchange generally produces a power - law distribution of flowpath residence times (Haggerty et al. , 2002; Cardenas, 2008; Sawyer and Cardenas, 2009) , with a strong bias towards flowpaths of short residence times, but also a large tail of longer residence time flowpaths. These more abundant , shorter SWI flowpaths ar e generally dominated by oxic conditions (Baker et al. , 2000a; Arnon et al. , 2007) , which has been assumed to lead to lower emissi ons of N 2 O (Marzadri et al. , 2011; Quick et al. , 2016) , but here we show that the highest N 2 O emissions we re observed in Experiment 1 in ring US under oxic conditions ( >100 µmol O 2 /L) , equaling an export of 948 µmol N 2 O - N/d. Thus, the range of porewat er residence time s and SWI conditions where net N 2 O production is possible may be much wider than previously estimated. The role of residence time heterogeneity in increased N 2 O production under oxic conditions would vary depending on streambed morphology , which we do not address with our small - scale studies (Boano et al. , 2014; Tonina et al. , 2016) . Sawmill Brook is a stream with heavily urbanized headwaters, and it has been recognized that these urbanized catchments have higher stream NO 3 - concentrations and N 2 O emissions than catchments draining native or unmanaged successional vegetation (Mulholland et al. , 2008; Beaulieu et al. , 2011) . Still, the role of urbanization in altering the physical flow regime in the stream and streambed has not yet been detangled from proposed influences of elevated N R inputs. While we do not know the Nr effects of land use in Sawmill Brook, we do know that our stud y targeted streambed morphologies that were a direct result of road sand application in the catchment (Paul and Meyer, 2001; Finkenbine J. K. et al. , 2007) , and it is not known how 72 denitrification in nati ve sediments compares to altered , increasingly sandy, bed composition . Research on a comparable environment , where a once clay - bedded stream ended up with a new sandy bed originating from road sand application, found that benthic algal growth was inhibited by constant bed movement, and thus respiration was higher within the sediment than the benthic zone or water column (At kinson et al. , 2008) . Our study and th at of Atkinson et al. (2008) highlight a potentially exciting research gap of how highly reactive SWI conditions are altered by human - introduced sand to stream channels . 3.5. Conclusions The role of the SWI in NO 3 - removal has been emphasized by other studies as being dependent on the interplay between reactant supply (e.g., buried POC) and transport timescales (Ocampo et al. , 2006; Gu et al. , 2007; Zarnetske et al. , 2012) . Similarly, r iparian zones have also been emphasized as an important stream - sediment interface in facilitating NO 3 - removal (Hedin et al. , 1998; Vidon and Hill, 2004) . This study found that under varying hydrologic conditions O 2 and NO 3 - removal were enhanced at lower hydrologic flow - through rates and thus longer hydrologic residence times. The shift towards longer residence times also resulted in increased N 2 production via denitrification (by 3.3 - fold) and an increase in the proportion of NO 3 - removal accounted for by denitrification end - products (by 4 .5 - fold). N 2 O production peaked at intermediate porewater residence times, and longer residence time flowpaths exhibited further reduction of N 2 O to N 2 , such that N 2 O export also peaked at intermediate flux rates. N 2 O export was 12 - fold greater from two h igh flux rate experiments (3 and 2 m/d) than from two low flux rate experiments (1.2 and 0.8 m/d). N 2 O export was also 2.5 - fold greater for all experiments from the experimental ring underlain by sediments with POC and woody debris, as opposed to well sort ed sand s . These experiments also reveal measurable denitrification and N 2 O production 73 in bulk - oxic sediments (as in Experiments 1 - 2 ), emphasizing the importance of less - mobile porosity and anoxic microzones in heterogeneous stream sediments. Models relying on bulk properties and threshold - type behavior of O 2 inhibition of denitrification should consider the role of anoxic microzones in contributing to increased N 2 O emissions from streams. Since these experiments maintained constant inflow chemistry, we are able to definitively show that hydrologic residence time is an important control on SWI biogeochemical reaction rates with respect to N transformations, an important function of the SWI that affects NO 3 - transport to downstream groundwaters and surface waters . 74 CHAPTER 4: SYNTHESIS & IMPLICATIONS This thesis focused on the role of the sediment - water interface (SWI) in processing of nitrogen (N) across freshwater landscapes. The central research questions addressed in Chapters 2 and 3 were: 1) how do changing conc entrations of dissolved organic carbon (DOC) and nitrate (NO 3 - ) influence NO 3 - removal in the SWI; and 2) how do hydrologic conditions ( downwelling rates ) change residence times along a SWI flowpath and affect rates of removal of oxygen (O 2 ), DOC, and NO 3 - ? To address these questions, two field studies were conducted during the summers of 2016 and 2017. Chapter 2 outlines experiments conducted in Snake Pond, Massachusetts, where I manipulated both reactant chemistry and hydrologic residence times entering v ertical flowpaths through the lakebed SWI sediments within open - bottom mesocosms . I found that when NO 3 - and DOC abundance were varied in the SWI sediments, changes in hydrologic residence time actually had the largest effect on SWI N processing ( NO 3 - removal and denitrification) , not the abundance of NO 3 - and DOC . Longer residence times resulted in increased O 2 , DOC, and NO 3 - removal , as well as denitrification and nitrous oxide (N 2 O) production. While the Snake Pond experiments described in Chapter 2 explored the relationships between NO 3 - and DOC concentrations and NO 3 - removal rates in a lentic sediment , subsequent experiments conducted in Sawmill Brook, MA, and described in Chapter 3, solely explored how hydrologic flow - through affected biogeochemical function , and specifically NO 3 - removal, in a lotic setting . The results of the Sawmill Brook experiments show that NO 3 - removal does clearly increase with hydrologic residence times, with longer residence times promoting increased NO 3 - removal . While O 2 removal increased with residence time , NO 3 - removal and denitrification occurred frequently in oxic conditions before O 2 removal was complete along flowpaths. This implicates anoxic microzones providing embedded anaerobic processing in ot herwise bulk oxic 75 SWI porewater . Further more , while NO 3 - removal rates were highest at long residence times, shorter residence times where bulk porewater was oxic resulted in the largest production and export of N 2 O from the SWI sediments at Sawmill Brook . Results from the Snake Pond Experiments (Chapter 2) also highlighted the role of particulate organic carbon (POC) in facilitating O 2 and NO 3 - removal, including via denitrification (Stelzer et al. , 2015) . In other studies POC has been highlighted as supplying an additional DOC source, based on budgets constructed for O 2 and NO 3 - removal and molar ratios of those reactions (Findlay and Sobczak, 1996) . These experiments corroborated findings that POC abundance decreased with depth (Inwood et al. , 2007; Harvey et al. , 2015) , and t his had the effect of highly reactive shallow sediments, with less modification with depth, especially under anoxic conditions. The Sawmill Brook Experiments (Chapter 3) targeted two sediment types (one with and one without abundant POC) with two identical experimental rings and the same hydrologic flow - through rates, and found different relationships between flow - through and residence time in each ring , as well as much higher NO 3 - removal and N 2 O production in sediments with the most POC. Thus, these exper iments indicate that POC can play a role as a source of DOC for reactions, but also a modifier of hydrologic flux rates , which can either promote or inhibit other controls on N processing in the SWI. By exploring the change in NO 3 - removal and denitrificat ion rates as a function of reactant (NO 3 - and DOC) concentrations in the Snake Pond experiments , and by solely manipulating flow - through rates in the Sawmill Brook experiments, the relative importance of biogeochemical and hydrologic factors wa s directly interrogated. Overall, I found that modifications of hydrologic flux in these two distinct SWIs had the largest effect O 2 and NO 3 - removal, denitrification, and N 2 O production rates more than increasing the concentrations of 76 NO 3 - and labile DO C . The results presented in Chapters 2 and 3 show that relatively small reductions in hydrologic flux (40 - 70%) resulted in large increases in rates of processes such as denitrification (7 00 - 40 00% ) . With DOC and NO 3 - abundance limitations removed, hydrologic residence time is the dominant control on SWI biogeochemical function s that are driven by solutes in water moving through the SWI . The results of this thesis raise important points to be addressed in future SWI s tudies. M any similar research questions to those addressed here concerning the role of chemical versus hydrologic controls have been addressed through controlled column studies (Gu et al. , 2007; Bourke et al. , 2014; Quick et al. , 2016; Liu et al. , 2017) and mesocosm experiments (Sobczak et al. , 2003; Stelzer et al. , 2014, 2015; Kurz et al. , 2017) . The methods used in this thesis present some advances over these previous experiments, in that I have independent ly manipulat ed both chemical and hydrologic conditions in SWI sediments in situ . While my experiments have highlighted that cont rolled changes in both chemical inputs and hydrologic conditions resulted in changes in SWI function, other research questions have yet to be addressed. Specifically: 1) How does the naturally - occurring gradient of DOC qualit y ( lability ) affect NO 3 - removal and N 2 O production? Decreased DOC lability was hypothesized to result in increased N 2 O production by incomplete denitrification (Quick et al. , 2016) , but this was not substantiated by qualitative measurements of DOC lability, such as spect rophotometric analysis (McKnight et al. , 2001; Cory et al. , 2011) . 2) How do seasonal changes in DOC quality, paired with hydrologic variability in stage or flow, result in changing SWI function (Stoliker et al. , 2016) ? While my experiments addressed the role of hydrologic residence time in NO 3 - processing in the absence of DOC abundance limitations, determining how these two controls interact on a seasonal 77 basis requires long er - term sampling. It is also not known if microbial community composition in a natural setting responds to these controls over these time scales (Storey et al. , 1999; Li et al. , 2017; Kim et al. , 2018) . 3) What is the scaling nature of the importance of less - mobile porosity in contributing to the abundance of anoxic microsites in SWI sediments under varying flow regimes (Briggs et al. , 2015) ? These experiments identify that our studied SWI sediments were large sources of N 2 O under fully oxic conditions, implicating anoxic microzone s as important for N 2 O production . 4) What was the importance of POC in contributing to N 2 O production? POC can play a role as a DOC source to fuel aerobic and anaerobic respiration. However, it may also be possible that it impedes convective flow and there by facilities diffusion - dominated transport and the development of anoxic microzones (Sawyer, 2015) . There was also a large disparity in the hydrologic flow - through rate s achieved within the rings with the same hydraulic head conditions, which implicates that the large fraction of POC in the sandy sediment of one ring played a role in regulating overall f low conditions through the SWI. 5) H ow does human alteration of the shallow SWI in urban environments, specifically through increased erosion and contribution of non - native sandy sediment such as from road sand application (Atkinson et al. , 2008) , alter SWI biogeochemical function? Further, can we d isen tangle the role of the physical alteration of SWI sediments and resid ence time distributions from anthropogenic N loading in urban catchments (Wollheim et al. , 2005; Mulholland et al. , 2008) , and which control is important for N 2 O emissions from streams (Beaulieu et al. , 2011) ? 78 APPENDICES 79 APPENDIX A 2016 Snake Pond Sampling Data 80 Table 4 : Tabulated data from Figure 12 . C oncentrations are as either µmol/L or nmol/L. Experiment is listed, with A for Ambient or the experiment number. Type is the mean with standard deviation of points in parentheses , when more than one measurement was available. Exp Depth (cm) Time (h) N 2 µM % N 2 O nM % % amb 0 0 414 (0.933) 0.366 (5.26e - 05) 8.97 (1.35) 0.429 (0.00161) 0.379 (0.00856) amb 1.5 0.9 422 (32.2) 0.366 (1.16e - 05) 8.55 (0.706) 0.42 (0.00731) 0.37 (0.000248) amb 7 4.2 417 (0) 0.366 (0) 7.35 (0) 0.408 (0) 0.367 (0) amb 12 7.2 400 (0) 0.366 (0) 2.01 (0) 0.45 (0) 0.37 (0) amb 18 10.8 404 (0) 0.366 (0) 1.64 (0) 0.497 (0) 0.386 (0) 1 0 0 NA NA NA NA NA 1 9.5 0.52 402 (22.7) 0.366 (9.66e - 05) 6.23 (1.51) 0.536 (0.0225) 5.5 (0.905) 1 14.5 0.793 400 (13.5) 0.366 (6.4e - 05) 7.06 (0.694) 0.432 (0.0409) 0.978 (0.182) 1 19.5 1.07 391 (8.86) 0.366 (3.29e - 05) 9.18 (0.921) 1.49 (0.182) 5.11 (1.97) 2 0 0 NA NA NA NA NA 2 9.5 0.506 431 (39.1) 0.367 (1.87e - 05) 10.1 (0.999) 1.09 (0.152) 4.07 (0.054) 2 14.5 0.773 429 (5.91) 0.367 (5.62e - 05) 14.7 (1.69) 1.94 (0.272) 4.08 (0.0403) 2 19.5 1.04 411 (4.94) 0.367 (1.07e - 05) 10.5 (0.651) 1.52 (0.0954) 4.11 (0.0488) 3 0 0 386 (0) 0.367 (0) 0 (0) 2.47 (0) 3.92 (0) 3 9.5 0.49 474 (8.18) 0.367 (6.65e - 05) 4.82 (4.06) 1.18 (0.286) 3.84 (0.00257) 3 14.5 0.748 459 (30.4) 0.368 (4.55e - 05) 7.73 (3.4) 1.38 (0.0241) 3.84 (0.00788) 3 19.5 1.01 427 (16) 0.367 (5.32e - 05) 9.88 (0.584) 1.14 (0.0823) 3.84 (0.000759) 4 0 0 0 (0) 0 (0) 0 (0) 0 (0) 3.92 (0) 4 9.5 0.474 396 (4.02) 0.367 (0.00012) 10.1 (1.6) 1.23 (0.259) 3.85 (0.00358) 4 14.5 0.724 416 (25.5) 0.37 (0.000993) 36.3 (31.8) 2.62 (0.763) 3.85 (0.00246) 4 19.5 0.973 404 (6.2) 0.367 (0.000165) 66.5 (27.1) 3.35 (0.193) 3.85 (0.0044) 5 0 0 0 (0) 0 (0) 0 (0) 0 (0) 3.93 (0) 5 9.5 0.747 474 (22.4) 0.532 (0.0596) 12.6 (15.9) 2.42 (0.662) 2.28 (1.32) 81 Table 4 ( cont d ) 5 14.5 1.14 431 (21.1) 0.631 (0.105) 38.5 (53.9) 2.42 (1.27) 1.31 (1.22) 5 19.5 1.53 433 (24.9) 0.778 (0.0435) 2140 (291) 3.92 (0.0015) 3.86 (0.0444) Exp Depth (cm) Time (h) NO 3 - µM O 2 µM C µM SO 4 2 - µM NO 2 - µM amb 0 0 1.04 (0.0673) 231 (0) 222 (45.1) 54.1 (0.393) 0 (0) amb 1.5 0.9 1.2 (0.0824) 230 (0.442) 166 (4.92) 54 (0.994) 0 (0) amb 7 4.2 1.31 (0) 10.9 (0) 123 (0) 53.5 (0) 0 (0) amb 12 7.2 1.05 (0) 14.4 (0) 116 (0) 53 (0) 0 (0) amb 18 10.8 1.05 (0) 35.6 (0) 107 (0) 56.2 (0) 0 (0) 1 0 0 NA NA NA NA NA 1 9.5 0.52 3.65 (3.55) 225 (1.7) 253 (7.92) 54.5 (0.484) 0 (0) 1 14.5 0.793 4.36 (3.35) 170 (1.38) 238 (9.87) 54.5 (0.449) 0 (0) 1 19.5 1.07 3.97 (0.947) 154 (3.76) 237 (16) 54.9 (0.306) 0 (0) 2 0 0 NA NA NA NA NA 2 9.5 0.506 180 (4.04) 225 (1.84) 226 (6.25) 53.5 (0.888) 0 (0) 2 14.5 0.773 176 (2.98) 148 (3.68) 219 (7.45) 53.6 (0.19) 0 (0) 2 19.5 1.04 183 (4.72) 159 (7.83) 188 (5.16) 54.1 (0.143) 0 (0) 3 0 0 164 (0) 0 (0) 368 (0) 52.7 (0) 0 (0) 3 9.5 0.49 154 (3.42) 236 (5.45) 253 (27) 52.5 (1.35) 0 (0) 3 14.5 0.748 150 (4.06) 108 (2.64) 250 (38.1) 52.4 (0.797) 0 (0) 3 19.5 1.01 158 (2.8) 123 (0.895) 362 (33.6) 54 (0.313) 0 (0) 4 0 0 252 (0) 0 (0) 684 (0) 50.8 (0) 1.47 (0) 4 9.5 0.474 177 (1.93) 195 (4.26) 310 (4.4) 46.6 (11.3) 2.42 (0.215) 4 14.5 0.724 164 (5.91) 8.15 (3.66) 279 (58.1) 52.3 (0.871) 4.29 (1.86) 4 19.5 0.973 159 (3.43) 4.35 (0.281) 544 (65.3) 52.9 (1.21) 12.7 (3.91) 5 0 0 798 (0) 0 (0) 1610 (0) 45.7 (0) 19.3 (0) 82 Table 4 ( cont d ) 5 9.5 0.747 27.9 (23.7) 197 (10) 326 (9.63) 50.2 (0.929) 0 (0) 5 14.5 1.14 27 (23) 4.32 (0.0177) 302 (0.307) 50.6 (0.445) 0 (0) 5 19.5 1.53 30 (22.4) 4.5 (0.00766) 576 (53.8) 62.7 (1.58) 25 (5.45) Table 5 : Tabulated data from Figure 13 , as well as data for SO 4 2 - and NO 2 - . Rates are calculated as in Section 2.2.5 . For the ambient profile, rates are calculated between 0 and 18 cm depth. For gases (O 2 , N 2 , and N 2 O), rates are between 9.5 and 19.5 cm depths. For NO 3 - and DOC, rates are calculated between 9.5 and 19.5 cm for Experiments 1 - 2, and between 0 and 19.5 cm for Experiments 3 - 5. Standard deviation is in parentheses. N 2 - N ( µ M/h) N 2 O - N ( n M/h) NO 3 - ( µ M /h) O 2 ( µ M/h) DOC ( µ M/h) SO 4 2 - ( µ M/h) NO 2 - ( µ M/h) Ambient Conditions - 0.889 ( 0.020 ) - 0.741 ( 0.018 ) 0.000719 ( 0.0673 ) - 18.1 ( 0.0 ) - 10.6 ( 45.1 ) 0.193 ( 0.393 ) 0 (0) Exp 1: 15 N - 1.36 ( 0.09 ) 3.46 ( 0.03 ) 0.596 ( 3.67 ) - 130 ( 4 ) - 29.8 ( 17.8 ) 0.849 ( 0.572 ) 0 ( 0 ) Exp 2: NO 3 - 3.18 ( 0.15 ) 2.24 ( 0.03 ) 5.95 ( 6.21 ) - 123 ( 8 ) - 72.0 ( 8.1 ) 1.22 ( 0.90 ) 0 ( 0 ) Exp 3: N+C - 8.59 ( 0.07 ) 2.44 ( 0.06 ) - 6.36 ( 2.8 ) - 219 ( 6 ) - 5.2 ( 33.6 ) 1.3 ( 0.3 ) 0 ( 0 ) Exp 4: N++C 1.83 ( 0.03 ) 111 ( 1 ) - 96.1 ( 3.4 ) - 381 ( 4 ) - 144 ( 65 ) 2.21 ( 1.21 ) 11.5 ( 3.9 ) Exp 5: Incr. R T 46.8 ( 0.5 ) 4660 ( 10 ) - 501 ( 22 ) - 245 ( 10 ) - 675 ( 54 ) 11.1 ( 1.6 ) 3.71 ( 5.45 ) Table 6 : Sediment core data from Snake Pond. Also including Loss on Ignition results from Figure 11 . Depth Interval (cm) Dry Sediment Mass (g) Porosity (%) Loss on Ignition (%) Sediment POC (g loss/cm3) 0 - 1 39.0 26.7 0.477 0.0126 1 - 2 36.1 26.7 0.428 0.0113 2 - 3 34.2 28.7 0.448 0.0119 3 - 5 69.7 27.4 0.408 0.0108 5 - 7 69.7 35.2 0.384 0.0102 7 - 9 65.0 32.2 0.428 0.0113 9 - 11 67.2 29.9 0.404 0.0107 83 APPENDIX B 2017 Sawmill Brook Experimental Data 84 Table 7 : Tabulated data from Figure 20 and Figure 22 . Concentrations are a s either µmol/L or nmol/L. Experiment is listed, with the flux rate and ring (US or DS). Type is the mean with standard deviation of points in parentheses, when more than one measurement was available. Exp Depth (cm) Time (h) N 2 µM % N 2 O nM % 3 m/d US 0 0 384 (44) 0.367 (3.25e - 05) 19.7 (0.557) 1.38 (0.012) 3 m/d US 5 0.33 421 (34.8) 0.367 (8.68e - 06) 17.3 (1.08) 1.63 (0.0413) 3 m/d US 10 0.66 414 (65.6) 0.371 (0.000209) 39.4 (2.18) 4.91 (0.0214) 3 m/d US 15 0.991 408 (13.8) 0.37 (9.09e - 05) 33.5 (0.268) 4.15 (0.0217) 3 m/d US 20 1.32 440 (56.8) 0.408 (0.00383) 155 (7.52) 7.08 (0.0115) 3 m/d DS 0 0 398 (47.5) 0.368 (0.00237) 23.4 (8.08) 1.98 (1.57) 3 m/d DS 5 0.249 423 (32) 0.367 (1.86e - 05) 12.8 (6.88) 1.51 (0.0108) 3 m/d DS 10 0.499 426 (60.2) 0.367 (3.31e - 05) 17.2 (0.409) 1.53 (0.027) 3 m/d DS 15 0.748 445 (60.2) 0.368 (2.41e - 05) 27.8 (0.268) 3.65 (0.0252) 3 m/d DS 20 0.998 430 (48.7) 0.37 (9.11e - 05) 31.5 (2.18) 4.53 (0.0923) 2 m/d US 0 0 452 (46.4) 0.367 (1.94e - 05) 23.2 (2.46) 0.621 (0.0357) 2 m/d US 5 0.458 423 (10.9) 0.367 (2.48e - 05) 21.5 (0.568) 0.884 (0.0197) 2 m/d US 10 0.917 354 (20) 0.371 (5.72e - 05) 59.4 (0) 5.09 (0.0269) 2 m/d US 15 1.38 342 (25.5) 0.382 (0.00115) 220 (28.2) 6.95 (0.15) 2 m/d US 20 1.83 424 (62.8) 0.421 (0.00215) 499 (41.8) 7.49 (0.0136) 2 m/d DS 0 0 439 (81.9) 0.367 (3.08e - 05) 20.1 (0.757) 0.624 (0.19) 2 m/d DS 5 0.318 468 (0.91) 0.367 (5.22e - 05) 21.3 (0.568) 1.27 (0.0849) 2 m/d DS 10 0.637 364 (46.4) 0.367 (3.64e - 05) 18.7 (0.379) 0.658 (0.00103) 2 m/d DS 15 0.955 369 (30.9) 0.369 (0.000109) 47.2 (1.33) 4.57 (0.0672) 2 m/d DS 20 1.27 390 (20) 0.385 (0.000398) 159 (9.28) 6.97 (0.0117) 1.2 m/d US 0 0 557 (0) 0.368 (0) 18.7 (0) 1.85 (0) 1.2 m/d US 5 0.708 581 (0) 0.41 (0) 319 (0) 8.61 (0) 1.2 m/d US 10 1.42 489 (0) 0.59 (0) 160 (0) 8.56 (0) 1.2 m/d US 15 2.12 470 (0) 0.643 (0) 137 (0) 8.64 (0) 1.2 m/d US 20 2.83 555 (0) 0.59 (0) 8.83 (0) 6.88 (0) 1.2 m/d DS 0 0 483 (0) 0.395 (0) 71.7 (0) 6.88 (0) 1.2 m/d DS 5 0.628 524 (0) 0.513 (0) 211 (0) 6.55 (0) 1.2 m/d DS 10 1.26 614 (0) 0.644 (0) 33.7 (0) 6.92 (0) 1.2 m/d DS 15 1.88 505 (0) 0.643 (0) 0.803 (0) 2.34 (0) 1.2 m/d DS 20 2.51 467 (0) 0.633 (0) 0.268 (0) 0.657 (0) 0.8 m/d US 0 0 495 (50.5) 0.367 (6.24e - 05) 21 (0.674) 0.919 (0.00892) 0.8 m/d US 5 0.8 563 (32.4) 0.431 (0.0982) 400 (208) 7.4 (0.117) 0.8 m/d US 10 1.6 525 (96.9) 0.486 (0.0941) 467 (445) 7.5 (0.0467) 85 Table 7 ( cont d ) 0.8 m/d US 15 2.4 479 (80.3) 0.583 (0.00197) 142 (7.35) 7.61 (0.00828) 0.8 m/d US 20 3.2 516 (86.5) 0.543 (0.00867) 32.5 (6.47) 7.28 (0.113) 0.8 m/d DS 0 0 451 (3.64) 0.367 (0.000252) 17.3 (3.45) 0.916 (0.197) 0.8 m/d DS 5 1.1 455 (29.6) 0.368 (0.000121) 17.9 (8.95) 3.59 (0.368) 0.8 m/d DS 10 2.19 486 (68.4) 0.37 (0.00104) 170 (86.6) 6.59 (0.552) 0.8 m/d DS 15 3.29 509 (76.6) 0.388 (0.0104) 533 (189) 7.18 (0.0766) 0.8 m/d DS 20 4.38 502 (75.8) 0.555 (0.0105) 259 (18.4) 7.23 (0.021) Exp Depth (cm) Time (h) NO 3 - µM % O 2 µM C µM 3 m/d US 0 0 73.4 (0.161) 7.69 (0.453) 228 (0) 571 (27.2) 3 m/d US 5 0.33 67.8 (8.21) 7.64 (0.172) 151 (0) 591 (40) 3 m/d US 10 0.66 51.7 (1.39) 7.61 (0.097) 84.1 (0) 595 (29.8) 3 m/d US 15 0.991 57.4 (1.01) 7.69 (0.0659) 69.5 (0) 533 (42.4) 3 m/d US 20 1.32 41 (0.703) 7.6 (0.156) 55 (0) 549 (31.2) 3 m/d DS 0 0 71.4 (2.78) 7.92 (0.0581) 219 (0) 737 (302) 3 m/d DS 5 0.249 67.2 (0.518) 7.38 (0.0292) 173 (0) 591 (44.6) 3 m/d DS 10 0.499 63.4 (0.838) 7.37 (0.0309) 127 (0) 602 (40.8) 3 m/d DS 15 0.748 62.4 (0.246) 7.36 (0.0195) 95.8 (0) 549 (19) 3 m/d DS 20 0.998 56.1 (2.13) 6.93 (0.752) 64.7 (0) 492 (38.3) 2 m/d US 0 0 78.5 (5.25) 7.45 (0.784) 219 (0) 522 (142) 2 m/d US 5 0.458 71.9 (3.88) 7.77 (0.0386) 180 (0) 659 (38.3) 2 m/d US 10 0.917 58.8 (10.4) 7.62 (0.147) 141 (0) 636 (11.8) 2 m/d US 15 1.38 63.7 (0.456) 7.78 (0.0347) 125 (0) 669 (41.2) 2 m/d US 20 1.83 39.7 (0.456) 7.69 (0.0259) 109 (0) 743 (130) 2 m/d DS 0 0 80.4 (6.96) 7.47 (0.827) 227 (0) 548 (128) 2 m/d DS 5 0.318 76.3 (2.74) 7.5 (0.0266) 192 (0) 644 (83.6) 2 m/d DS 10 0.637 73.1 (2.05) 7.49 (0.0495) 166 (0) 762 (126) 2 m/d DS 15 0.955 75.5 (2.74) 7.44 (0.0403) 130 (0) 617 (85.4) 2 m/d DS 20 1.27 66.3 (0.456) 7.42 (0.0015) 93.8 (0) 765 (83) 1.2 m/d US 0 0 80 (0) 8.89 (0) 142 (0) 489 (0) 1.2 m/d US 5 0.708 24.7 (0) 8.89 (0) 0.01 (0) 890 (0) 1.2 m/d US 10 1.42 16.9 (0) 8.89 (0) 0.01 (0) 941 (0) 1.2 m/d US 15 2.12 19.7 (0) 8.89 (0) 0.01 (0) 662 (0) 1.2 m/d US 20 2.83 12.1 (0) 8.89 (0) 0.01 (0) 664 (0) 86 Table 7 ( cont d ) 1.2 m/d DS 0 0 71.3 (0) 8.46 (0) 229 (0) 466 (0) 1.2 m/d DS 5 0.628 15 (0) 8.46 (0) 0.01 (0) 1030 (0) 1.2 m/d DS 10 1.26 15.8 (0) 8.46 (0) 0.01 (0) 1070 (0) 1.2 m/d DS 15 1.88 16.9 (0) 8.46 (0) 0.01 (0) 1090 (0) 1.2 m/d DS 20 2.51 18.5 (0) 8.46 (0) 0.01 (0) 739 (0) 0.8 m/d US 0 0 76.3 (0.652) 7.86 (0.156) 253 (0) 427 (15.1) 0.8 m/d US 5 0.8 37 (8.81) 7.48 (0.135) 11.3 (0) 789 (36.1) 0.8 m/d US 10 1.6 11.9 (7.92) 6.81 (0.23) 0.01 (0) 863 (80.5) 0.8 m/d US 15 2.4 16.8 (0.945) 6.61 (0.504) 0.01 (0) 612 (37.4) 0.8 m/d US 20 3.2 14.6 (0.566) 1.4 (0.482) 0.01 (0) 805 (17.4) 0.8 m/d DS 0 0 76.2 (1.99) 7.88 (0.118) 273 (0) 427 (41.2) 0.8 m/d DS 5 1.1 54.8 (3.81) 7.18 (0.167) 28.8 (0) 816 (50.8) 0.8 m/d DS 10 2.19 37.7 (5.5) 7.35 (0.0379) 0.01 (0) 845 (45.4) 0.8 m/d DS 15 3.29 16.7 (11.1) 7.25 (0.0211) 0.01 (0) 962 (80.4) 0.8 m/d DS 20 4.38 17.5 (1.3) 5.91 (1.13) 0.01 (0) 624 (56.2) Table 8 : Tabulated data from Figure 21 and Figure 23 . Rates are calculated as in Section 2.2.5 , as µ mol/L/h or nmol/L/h, with standard deviation in parentheses , when multiple points were available for the calculation. Experiment N 2 - N ( µ M /h) N 2 O - N (nM /h) NO 3 - ( µ M /h) O 2 ( µ M /h) DOC ( µ M /h) 3 m/d US 3.77 (0.27) 105 (1) - 24.5 (0.7) - 131 (0) - 16.6 (41.4) 3 m/d DS 1.59 (0.25) 11.1 (0.6) - 15.3 (3.5) - 154 (0) - 245 (305) 2 m/d US 0.929 (0.307) 272 (3) - 21.2 (5.3) - 60 (0) 120 (193) 2 m/d DS - 1.16 (0.31) 115 (1) - 11.1 (7) - 105 (0) 170 (153) 1.2 m/d US 4.86 (0) 1.03 (0) - 24 (0) - 50 (0) 62.1 (0) 1.2 m/d DS 4.96 (0) - 23.2 (0) - 21 (0) - 91 (0) 109 (0) 0.8 m/d US 3.92 (0.51) 8.66 (0.51) - 19.3 (0.9) - 79 (0) 118 (23) 0.8 m/d DS 3.28 (0.38) 53.9 (1.4) - 13.4 (2.4) - 62.3 (0) 44.9 (69.7) 87 Table 9 : Injection and Flush porewater velocities from the Sawmill Brook Experiments. Velocities were d etermined by the depth (10 or 20 cm) divided by the median arrival time of the conductivity plume at that depth, as shown in Figure . Experiment, Ring, Depth Injection Velocity (m/d) Flush Velocity (m/d) Exp1 DS - 10cm 3.60 4.23 Exp1 DS - 20cm 3.30 3.95 Exp1 US - 10cm 2.38 3.21 Exp1 US - 20cm 1.74 3.14 Exp2 DS - 10cm 4.72 5.56 Exp2 DS - 20cm 4.24 4.72 Exp2 US - 10cm 3.50 3.01 Exp2 US - 20cm 3.51 4.51 Exp3 DS - 10cm 1.08 NA Exp3 DS - 20cm 1.11 NA Exp3 US - 10cm 1.51 NA Exp3 US - 20cm 1.49 NA 88 A) B) Figure 27 : Conductivity Breakthrough Curves from the Sawmill Brook Experiments. Data are plotted as electrical conductivity over time for the first three experiments, for both 10 and 20 cm depth on the injection (A) and flush (B) phases of the high - conductivity inje ctions. Time is normalized to the beginning of injection or flush. Red lines mark the initial concentration and plateau; the orange line marks the median concentration between those two, and the green line marks the time of median arrival time, which was u sed to calculate velocities as in Table 9 . 89 Figure 27 A) B) 90 REFERENCES 91 REFERENCES Abbott BW, Baranov V, Mendoza - Lera C, Nikolakopoulou M, Harjung A, Kolbe T, Balasubramanian MN, Vaessen TN, Ciocca F, Campeau A, et al. 2016. Using multi - tracer inference to move beyond single - catchment ecohydrology. Earth - Science Reviews 160 : 19 42 DOI: 10.1016/j.earscirev.2016.06.014 Abbott BW, Gruau G, Zarnetske JP, Moatar F, Barbe L, Thomas Z, Fovet O, Kolbe T, Gu S, Pierson - Wickmann A - C, et al. 2018. Unexpected spatial stability of water ch emistry in headwater stream networks. Ecology Letters 21 (2): 296 308 DOI: 10.1016/j.earscirev.2016.06.014 Ahrens TD, Siver PA. 2000. Trophic Conditions and Water Chemistry of Lakes on Cape Cod, Massachusetts, USA. Lake and Reservoir Management 16 (4): 268 280 DOI: 10.1080/07438140009354235 Anderson JK, W ondzell SM, Gooseff MN, Haggerty R. 2005. Patterns in stream longitudinal profiles and implications for hyporheic exchange flow at the H.J. Andrews Experimental Forest, Oregon, USA. Hydrological Processes 19 (15): 2931 2949 DOI: 10.1002/hyp.5791 Anderson MP, Munter JA. 1981. Seasonal reversals of groundwater flow around lakes and the relevance to stagnation points and lake budgets. Water Resources Research 17 (4): 1139 1150 DOI: 10.1029/WR017i004p01139 Arnon S, Gray KA, Packman AI. 2007. Biophysicochemical process coupling controls nitrogen use by benthic biofilms. Limnology and Oceanography 52 (4): 1665 1671 DOI: 10.4319/lo.2007.52.4.1665 Atkinson BL, Grace MR, Hart BT, Vanderkruk KEN. 2008. Sediment instability affects the rate and location of primary production and respiration in a sand - bed stream. Journal of the North American Benthological Society 27 (3): 581 592 DOI: 10.1899/07 - 143.1 Baker MA, Dahm CN, Valett HM. 1999. Acetate retention and metabolism in the hyporheic zone of a mountain stream. Limnology and Oceanography 44 (6): 1530 1539 DOI: 10.4319/lo.1999.44.6.1530 Baker MA, Dahm CN, Valett HM. 2000a. Anoxia, anaerobic metabolism biogeochemistry of the stream water - ground water interface. In Streams and Ground Waters , Jones JA, , Mullholland PJ (eds).Academic Press: San Diego, California, USA; 259 283. Baker MA, Valett HM, Dahm CN. 2000b. Organic carbon supply and metabolism in a shallow groundwater ecosystem. Ecology 81 (11): 3133 3148 DOI: 10.1890/0012 - 9658(2000)081[3133:OCSAMI]2.0.CO;2 92 Barbaro JR, Walter DA, LeBlanc DR. 2013. Transport of nitrogen in a treated - wastewater plume to coastal discharge areas, Ashumet Valley, Cape Cod, Massachusetts. U.S. Geological Survey Scientific Investigations Report 2013 - 5061. US Geological Survey, Resto n, VA. Available at: http:// pubs.er.usgs.gov/publication/sir20135061 [Accessed 17 March 2017] Baulch HM, Schiff SL, Maranger R, Dillon PJ. 2011. Nitrogen enrichment and the emission of nitrous oxide from streams. Global Biogeochemical Cycles 25 (4) DOI: 10.1029/2011GB004047 Beaulieu JJ, Tank JL, Hamilton SK, Wollheim WM, Hall Jr RO, Mulholland PJ, Peterson BJ, Ashkenas LR, Cooper LW, Dahm CN, et al. 2011. Nitrous oxide emission from denitrification in stream and river networks. Proceedings of the National Academy of Sciences 108 (1): 214 219 DOI: 10.1073/pnas.1011464108 Bernhardt ES, Likens GE. 2002. Dissolved organic carbon enrichment alters nitrogen dynamics in a forest stream. Ecology 83 (6): 1689 1700 DOI: 10.1890/0012 - 9658(2002)083[1689:DOCEAN]2.0.CO;2 Betlach MR, Tiedje JM. 1981. Kinetic Explanation for Accumulation of Nitrite, Nitric Oxide, and Nitrous Oxide During Bacterial Denitrific ation. Applied and Environmental Microbiology 42 (6): 1074 1084 Boano F, Harvey JW, Marion A, Packman AI, Revelli R, Ridolfi L, Wörman A. 2014. Hyporheic flow and transport processes: Mechanisms, models, and biogeochemical implications. Reviews of Geophysi cs 52 (4): 603 679 DOI: 10.1002/2012RG000417 Born SM, Smith SA, Stephenson DA. 1974. The hydrogeological regime of glacialterrain lakes, with management and planning applications: Madison. University of W isconsin Extension Born SM, Smith SA, Stephenson DA. 1979. Hydrogeology of glacial - terrain lakes, with management and planning applications. Journal of Hydrology 43 (1 4): 7 43 DOI: 10.1016/0022 - 1 694(79)90163 - X Boulton AJ, Findlay S, Marmonier P, Stanley EH, Valett HM. 1998. The functional significance of the hyporheic zone in streams and rivers. Annual Review of Ecology and Systematics : 59 81 DOI: 10.1146/annurev.ecolsys.29.1.59 Bourke MF, Kessler AJ, Cook PLM. 2014. Influence of buried Ulva lactuca on denitrification in permeable sediments. Marine Ecology Progress Series 498 : 85 94 DOI: 10.3354/meps10611 Boyer EW, Howarth RW, Galloway JN, Dentener FJ, Green PA, Vörösmarty CJ. 2006. Riverine nitrogen export from the continents to the coasts. Global Biogeochemical Cycles 20 (1) DOI: 10.102 9/2005GB002537 Briggs MA, Day - Lewis FD, Zarnetske JP, Harvey JW. 2015. A physical explanation for the development of redox microzones in hyporheic flow. Geophysical Research Letters 42 (11): 4402 4410 DOI: 10.1002/2015GL064200 93 Briggs MA, Gooseff MN, Peterson BJ, Morkeski K, Wollheim WM, Hopkinson CS. 2010. Surface and hyporheic transient storage dynamics throughout a coastal stream network. Water Resources Research 46 (6) DOI: 10.1029/2009WR008222 Briggs MA, Lautz LK, Buckley SF, Lane JW. 2014a. Practical limitations on the use of diurnal temperature signals to quantify groundwater upwelling. Journal of hydrology 519 : 1739 1751 DOI: 10.1016/j.jhydrol.2014.09.030 Briggs MA, Lautz LK, Hare DK. 2014b. Residence time control on hot moments of net nitrate production and uptake in the hyporheic zone. Hydrological Processes 28 (11): 3741 3751 DOI: 10.1002/hyp.9921 Briggs MA, Lautz LK, Hare DK, González - Pinzón R. 2013. Relating hyporheic fluxes, residence times, and redox - sensitive biogeochemical processes upstream of beaver dams. Freshwa ter Science 32 (2): 622 641 DOI: 10.1899/12 - 110.1 Brunke M, Gonser T. 1997. The ecological significance of exchange processes between rivers and groundwater. Freshwater Biology 37 (1): 1 33 DOI: 10.1046/j.1365 - 2427.1997.00143.x Burford JR, Bremner JM. 1975. Relationships between the denitrification capacities of soils and total, water - soluble and readily decomposable soil organic matter. Soil Biol ogy and Biochemistry 7 (6): 389 394 DOI: 10.1016/0038 - 0717(75)90055 - 3 Burgin AJ, Hamilton SK. 2007. Have we overemphasized the role of denitrification in aquatic ecosystems? A review of nitrate re moval pathways. Frontiers in Ecology and the Environment 5 (2): 89 96 DOI: 10.1890/1540 - 9295(2007)5[89:HWOTRO]2.0.CO;2 Burgin AJ, Hamilton SK. 2008. NO3 driven SO42 - production in freshwater ecosystems: implications for N and S cycling. Ecosystems 11 (6): 908 922 DOI: 10.1007/s10021 - 008 - 9169 - 5 Burgin AJ, Yang WH, Hamilton SK, Silver WL. 2011. Beyond carbon and nitrogen: how the microbial energy economy couples elemental cycles in diverse ecosystems. Frontiers in Ecology and the Environment 9 (1): 44 52 DOI: 10.1890/090227 Bussey KW, Walter DA. 1996. Spatial and temporal distribut ion of specific conductance, boron, and phosphorus in a sewage - contaminated aquifer near Ashumet Pond, Cape Cod, Massachusetts. US Geological Survey; Branch of Information Services [distributor] . Available at: http:// pubs.er.usgs.gov/publication/ofr96472 Cardenas MB. 2008. Surface water - groundwater interface geomorphology leads to scaling of residence times. Geophysical Research Letters 35 (8) DOI: 10.1029/2008GL033753 Cardenas MB. 2015. Hyporheic zone hydrologic science: A historical account of its emergence and a prospectus. Water Resources Research 51 (5): 3601 3616 DOI: 10.1002/2015WR017028 94 Carlozzi CA, King K, Newbold WF. 1975. Ecosystems and resources of the Massachusetts Coast . Massachusetts Coastal Zone Management Program, Boston. Available at: http://masslib - dspace.lon gsight.com/handle/2452/69134 Chen RL, Keeney DR, Graetz DA, Holding AJ. 1972. Denitrification and Nitrate Reduction in Wisconsin Lake Sediments. Journal of Environmental Quality 1 (2): 158 162 DOI: 10.2134/jeq1972.00472425000100020011x Cherkauer DS, McKereghan PF, Schalch LH. 1992. Delivery of Chloride and Nitrate by Ground Water to the Great Lakes: Study for the Door Peninsula, Wisconsin. Groundwater (30): 10.1111/j.1745 - 6584.1992.tb01571.x Cory RM, Boyer EW, McKnight DM. 2011. Spectral methods to advance understanding of dissolved organic carbon dynamics in forested catchments. Forest Hydrology and Biogeochemis try : 117 135 DOI: 10.1007/978 - 94 - 007 - 1363 - 5_6 Danczak RE, Sawyer AH, Williams KH, Stegen JC, Hobson C, Wilkins MJ. 2016. Seasonal hyporheic dynamics control coupled microbiology and geochemistry in Colorado River sediments. Journal of Geophysical Research - Biogeosciences 121 (12): 2976 2987 DOI: 10.1002/2016JG003527 Downing JA, Cole JJ, Duarte CA, Middelburg JJ, Melack JM, Prairie YT, Kortelainen P, Striegl RG, McDowell WH, Tranvik LJ. 2012. Global abundance and size distribution of streams and rivers. Inland waters 2 (4): 229 236 DOI: 10.5268/IW - 2.4.502 Duff JH, Triska FJ. 1990. Denitrifications in sediments from the hyporheic zone adjacent to a small forested stream. Canadian Journal of Fisheries and A quatic Sciences 47 (6): 1140 1147 DOI: 10.1139/f90 - 133 Farrell TB, Quick AM, Reeder WJ, Tonina D, Benner SG, Feris KP. 2013. Carbon availability and the distribution of denitrifying organisms influence N2O pro duction in the hyporheic zone. In AGU Fall Meeting Abstracts L05. Available at: http://adsabs.harvard.edu/abs/2013AGUFM.H31L..05F [Accessed 9 November 2016] Findlay S. 1995. Importance of sur face - subsurface exchange in stream ecosystems: The hyporheic zone. Limnology and Oceanography 40 (1): 159 164 DOI: 10.4319/lo.1995.40.1.0159 Findlay S, Sinsabaugh RL. 2003. Response of hyporheic biof ilm metabolism and community structure to nitrogen amendments. Aquatic microbial ecology 33 (2): 127 136 DOI: 10.3354/ame033127 Findlay S, Sobczak WV. 1996. Variability in removal of dissolved organic carbon in hyporheic sediments. Journal of the North American Benthological Society 15 (1): 35 41 DOI: 10.2307/1467431 95 Finkenbine J. K., Atwater J. W., Mavinic D. S. 2007. Stream health after urbanization. JAWRA Journal of the American Water Resources Association 36 (5): 1149 1160 DOI: 10.1111/j.1752 - 1688.2000.tb05717.x Fires tone MK, Davidson EA. 1989. Microbiological basis of NO and N2O production and consumption in soil. Exchange of trace gases between terrestrial ecosystems and the atmosphere 47 : 7 21 Firestone MK, Tiedje JM. 1979. Temporal Change in Nitrous Oxide and Dinit rogen from Denitrification Following Onset of Anaerobiosis. Applied and Environmental Microbiology 38 (4): 673 679 Firestone MK, Firestone RB, Tiedje JM. 1980. Nitrous oxide from soil denitrification: factors controlling its biological production. Science 208 (4445): 749 751 DOI: 10.1126/science.208.4445.749 Forster P, Ramaswamy V, Artaxo P, Berntsen T, Betts R, Fahey DW, Haywood J, Lean J, Lowe DC, Myhre G, et al. 2007. Changes in atmospheric cons tituents and in radiative forcing. In Climate Change 2007. The Physical Science Basis Cambridge University Press: New York. Galloway JN, Dentener FJ, Capone DG, Boyer EW, Howarth RW, Seitzinger SP, Asner GP, Cleveland CC, Green PA, Holland EA, et al. 2004. Nitrogen cycles: past, present, and future. Biogeochemistry 70 (2): 153 226 DOI: 10.1007/s10533 - 004 - 0370 - 0 Gardner JR, Doyle MW. 2018. Sediment Water Surface Area Along Rivers: Water Column Versus Be nthic. Ecosystems DOI: 10.1007/s10021 - 018 - 0236 - 2 Gu C, Hornberger GM, Herman JS, Mills AL. 2008. Effect of freshets on the flux of groundwater nitrate through streambed sediments. Water resources res earch 44 (5) DOI: 10.1029/2007WR006488 Gu C, Hornberger GM, Mills AL, Herman JS, Flewelling SA. 2007. Nitrate reduction in streambed sediments: Effects of flow and biogeochemical kinetics. Water Resources Research 43 (12) DOI: 10.1029/2007WR006027 Haggerty R, Wondzell SM, Johnson MA. 2002. Power - law residence time distribution in the hyporheic zone of a 2nd - order mountain stream. Geophysical Research Lett ers 29 (13) DOI: http://dx.doi.org/10.1029/2002GL014743 HM, Poole GC, Peterson BJ, et al. 2009. Nitrate removal in stream ecosystems measured by 15N addition experiments: Total uptake. Limnology and Oceanography 54 (3): 653 665 DOI: 10.4319/lo.2009.54.3.0653 Harvey JW, Böhlke JK, Voytek MA, Scott D, Tobias CR. 2013. Hyporheic zone denitrification: Controls on effective reaction depth and contribution to whole - stream mas s balance. Water Resources Research 49 (10): 6298 6316 DOI: 10.1002/wrcr.20492 96 Harvey RW, Metge DW, LeBlanc DR, Underwood J, Aiken GR, Butler K, McCobb TD, Jasperse J. 2015. Importance of the Colmation Laye r in the Transport and Removal of Cyanobacteria, Viruses, and Dissolved Organic Carbon during Natural Lake - Bank Filtration. Journal of Environmental Quality 44 (5): 1413 1423 DOI: 10.2134/jeq2015.03.01 51 Hedin LO, von Fischer JC, Ostrom NE, Kennedy BP, Brown MG, Robertson GP. 1998. Thermodynamic constraints on nitrogen transformations and other biogeochemical processes at soil stream interfaces. Ecology 79 (2): 684 703 DOI: 10.1890/0012 - 9658(1998)079[0684:TCONAO]2.0.CO;2 Helton AM, Wright MS, Bernhardt ES, Poole GC, Cory RM, Stanford JA. 2015. Dissolved organic carbon lability increases with water residence time in the all uvial aquifer of a river floodplain ecosystem. Journal of Geophysical Research: Biogeosciences 120 (4): 693 706 DOI: 10.1002/2014JG002832 Howarth RW, Billen G, Swaney D, Townsend A, Jaworski N, Lajtha K, Downing JA, Elmgren R, Caraco N, Jordan T, et al. 1996. Regional nitrogen budgets and riverine N & P fluxes for the drainages to the North Atlantic Ocean: Natural and human influences. Biogeochemistry 35 : 75 139 DOI: 10.1007/978 - 94 - 009 - 1776 - 7_3 Inwood SE, Tank JL, Bernot MJ. 2007. Factors Controlling Sediment Denitrification in Midwestern Streams of Varying Land Use. Microbial Ecology 53 (2): 247 258 DOI: 10.1007/s00248 - 006 - 9104 - 2 Irvine DJ, Lautz LK, Briggs MA, Gordon RP, McKenzie JM. 2015. Experimental evaluation of the applicability of phase, amplitude, and combined methods to determine water flux and thermal diffusivity from temperature tim e series using VFLUX 2. Journal of Hydrology 531 : 728 737 DOI: 10.1016/j.jhydrol.2015.10.054 Kaufman MH, Cardenas MB, Buttles J, Kessler AJ, Cook PLM. 2017. Hyporheic hot moments: Dissolved oxyge n dynamics in the hyporheic zone in response to surface flow perturbations. Water Resources Research 53 (8): 6642 6662 DOI: 10.1002/2016WR020296 Kaushal SS, Likens GE, Pace ML, Utz RM, Haq S, Gorman J, Gr ese M. 2018. Freshwater salinization syndrome on a continental scale. Proceedings of the National Academy of Sciences : 201711234 DOI: 10.1073/pnas.1711234115 Kidmose J, Engesgaard P, Ommen DAO, Nilsson B, Flindt MR, Andersen FØ. 2015. The Role of Groundwater for Lake - Water Quality and Quantification of N Seepage. Groundwater 53 (5): 709 721 DOI: 10.1111/gwat.122 81 Kim H, Kaown D, Mayer B, Lee J - Y, Lee K - K. 2018. Combining pyrosequencing and isotopic approaches to assess denitrification in a hyporheic zone. Science of The Total Environment 631 632 : 755 764 DOI: 10.1016/j.scitotenv.2018.03.073 97 Kravchenko AN, Fry JE, Guber AK. 2018. Water absorption capacity of soil - incorporated plant leaves can affect N 2 O emissions and soil inorganic N concentrations. Soil Biology and Biochemistry 121 : 113 119 DOI: 10.1016/j.soilbio.2018.03.013 Kravchenko AN, Toosi ER, Guber AK, Ostrom NE, Yu J, Azeem K, Rivers ML, Robertson GP. 2017. Hotspots of soil N 2 O emission enhanced through water absorption by plant residue. Nature Geoscience 10 (7): 496 DOI: 10.1038/ngeo2963 Kroeze C, Mosier A, Bouwman L. 1999. Closing the global N2O budget: a retrospective analysis 1500 1994. Global Biogeochemical Cycles 13 (1): 1 8 DO I: 10.1029/1998GB900020 Kurz MJ, Drummond JD, Martí E, Zarnetske JP, Lee - Cullin J, Klaar MJ, Folegot S, Keller T, Ward AS, Fleckenstein JH. 2017. Impacts of water level on metabolism and transient storage in vegetated lowland rivers: Insights from a mesoco sm study. Journal of Geophysical Research: Biogeosciences 122 (3): 628 644 DOI: 10.1002/2016JG003695 Lansdown K, Heppell CM, Dossena M, Ullah S, Heathwaite AL, Binley A, Zhang H, Trimmer M. 2014. Fine - sca le in situ measurement of riverbed nitrate production and consumption in an armored permeable riverbed. Environmental Science & Technology 48 (8): 4425 4434 DOI: 10.1021/es4056005 Lansdown K, Heppell CM, Tri mmer M, Binley A, Heathwaite AL, Byrne P, Zhang H. 2015. The interplay between transport and reaction rates as controls on nitrate attenuation in permeable, streambed sediments. Journal of Geophysical Research - Biogeosciences 120 (6): 1093 1109 DOI: 10.1002/2014JG002874 Lansdown K, Trimmer M, Heppell CM, Sgouridis F, Ullah S, Heathwaite AL, Binley A, Zhang H. 2012. Characterization of the key pathways of dissimilatory nitrate reduction and their response to complex organic substrates in hyporheic sediments. Limnology and Oceanography 57 (2): 387 400 DOI: 10.4319/lo.2012.57.2.0387 Lautz LK, Fanelli RM. 2008. Seasonal biogeochemical hotspots in the strea mbed around restoration structures. Biogeochemistry 91 (1): 85 104 DOI: 10.1007/s10533 - 008 - 9235 - 2 LeBlanc DR, Garabedian SP, Hess KM, Gelhar LW, Quadri RD, Stollenwerk KG, Wood WW. 1991. Large - scale natural gradient tracer test in sand and gravel, Cape Cod, Massachusetts: 1. Experimental design and observed tracer movement. Water Resources Research 27 (5): 895 910 DOI: 10.1029/91WR00241 Lewandowski J, Meinikmann K, Nützmann G, Rosenberry DO. 2015. Groundwater the disregarded component in lake water and nutrient budgets. Part 2: effects of groundwater on n utrients. Hydrological Processes 29 (13): 2922 2955 DOI: 10.1002/hyp.10384 Li S, Peng C, Wang C, Zheng J, Hu Y, Li D. 2017. Microbial Succession and Nitrogen Cycling in Cultured Biofilms as Affected by the I norganic Nitrogen Availability. Microbial Ecology 73 (1): 1 15 DOI: 10.1007/s00248 - 016 - 0827 - 4 98 Liu Y, Liu C, Nelson WC, Shi L, Xu F, Liu Y, Yan A, Zhong L, Thompson C, Fredrickson JK, et al. 2017. Eff ect of Water Chemistry and Hydrodynamics on Nitrogen Transformation Activity and Microbial Community Functional Potential in Hyporheic Zone Sediment Columns. Environmental Science & Technology 51 (9): 4877 4886 DOI: 10.1021/acs.est.6b05018 Lloyd D. 1993. Aerobic denitrification in soils and sediments: From fallacies to factx. Trends in Ecology & Evolution 8 (10): 352 356 DOI: 10.1016/0169 - 5347( 93)90218 - E Lloyd D, Boddy L, Davies KJP. 1987. Persistence of bacterial denitrification capacity under aerobic conditions: The rule rather than the exception. FEMS Microbiology Ecology 3 (3): 185 190 DOI: 10.1111/j.1574 - 6968.1987.tb02354.x Luce CH, Tonina D, Gariglio F, Applebee R. 2013. Solutions for the diurnally forced advection - diffusion equation to estimate bulk fluid velocity and diffusivity in streambeds from temperature time series. W ater Resources Research 49 (1): 488 506 DOI: 10.1029/2012WR012380 van Luijn F, Boers PCM, Lijklema L. 1996. Comparison of denitrification rates in lake sediments obtained by the N2 flux method, the 15N is otope pairing technique and the mass balance approach. Water Research 30 (4): 893 900 DOI: 10.1016/0043 - 1354(95)00250 - 2 Marzadri A, Dee MM, Tonina D, Bellin A, Tank JL. 2017. Role of surface and s ubsurface processes in scaling N2O emissions along riverine networks. Proceedings of the National Academy of Sciences : 201617454 DOI: 10.1073/pnas.1617454114 Marzadri A, Tonina D, Bellin A. 2011. A sem ianalytical three - dimensional process - based model for hyporheic nitrogen dynamics in gravel bed rivers. Water Resources Research 47 (11) DOI: 10.1029/2011WR010583 Marzadri A, Tonina D, Bellin A, Tank JL. 2014. A hydrologic model demonstrates nitrous oxide emissions depend on streambed morphology. Geophysical Research Letters 41 (15): 5484 5491 DOI: 10.1002/2014GL06 0732 Massachusetts Division of Fisheries and Wildlife. 1993. Pond maps of Massachusetts: Cape Cod. publication no. 17455 - 60 - 1M - 12/93. Masterson JP, Walter DA, Savoie, J. 1996. Use of particle tracking to improve numerical model calibration and to analyze g round - water flow and contaminant migration, Massachusetts Military Reservation, Cape Cod, Massachusetts. U.S. Geological Survey Water - Supply Paper 2482. Available at: http:// pubs.er.usgs.gov/publ ication/ofr96214 Mather KF, Goldthwait RP, Thiesmeyer LR. 1942. Pleistocene geology of western Cape Cod, Massachusetts. Geological Society of America Bulletin 53 (8): 1127 1174 DOI: 10.1130/GSAB - 53 - 1127 99 M cCallum AM, Andersen MS, Rau GC, Acworth RI. 2012. A 1 - D analytical method for estimating surface water groundwater interactions and effective thermal diffusivity using temperature time series. Water Resources Research 48 (11): W11532 DOI: 10.1029/2012WR012007 McClain ME, Boyer EW, Dent CL, Gergel SE, Grimm NB, Groffman PM, Hart SC, Harvey JW, Johnston CA, Mayorga E, et al. 2003. Biogeochemical hot spots and hot moments at the interface of terrestrial and aquatic ecosystems. Ecosystems 6 (4): 301 312 DOI: 10.1007/s10021 - 003 - 0161 - 9 McCobb TD, LeBlanc DR, Walter DA, Hess KM, Kent DB, Smith RL. 2003. Phosphorus in a ground - water contaminant plume dischar ging to Ashumet Pond, Cape Cod, Massachusetts, 1999 Available at: http://pubs.usgs.gov/wri/wri024306/pdfs/wrir024306.pdf [Accessed 2 April 2017] McGuire KJ, Torgersen CE, Likens GE, Bus o DC, Lowe WH, Bailey SW. 2014. Network analysis reveals multiscale controls on streamwater chemistry. Proceedings of the National Academy of Sciences 111 (19): 7030 7035 DOI: 10.1073/pnas.1404820111 M cKnight DM, Boyer EW, Westerhoff PK, Doran PT, Kulbe T, Andersen DT. 2001. Spectrofluorometric characterization of dissolved organic matter for indication of precursor organic material and aromaticity. Limnology and Oceanography 46 (1): 38 48 DOI: 10.4319/lo.2001.46.1.0038 Minnesota Department of Natural Resources, Division of Ecological Services D. 2003. Healthy Rivers: A Water Course - How Rivers Run - Connections. Minnesota Department of Natural R esources Available at: http:// files.dnr.state.mn.us/assistance/backyard/healthyrivers/course/200/203_130.htm [Accessed 15 January 2017] Mosier AR, Doran J W, Freney JR. 2002. Managing soil denitrification. Journal of soil and water conservation 57 (6): 505 512 Mulholland PJ, Hall Jr RO, Sobota DJ, Dodds WK, Findlay SE, Grimm NB, Hamilton SK, measured by 15N addition experiments: denitrification. Limnology and Oceanography 54 (3): 666 680 DOI: 10.4319/lo.2009.54.3.0666 Mulholland PJ, Helton AM, Poole GC, Hall Jr RO, Hamilton SK, Peterson BJ, Tank JL, Ashkenas LR, Cooper LW, Dahm CN, et al. 2008. Stream denitrification ac ross biomes and its response to anthropogenic nitrate loading. Nature 452 (7184): 202 205 DOI: 10.1038/nature06686 Ocampo CJ, Oldham CE, Sivapalan M. 2006. Nitrate attenuation in agricultural catchments: S hifting balances between transport and reaction. Water Resources Research 42 (1) DOI: 10.1029/2004WR003773 100 Ostrom NE, Gandhi H, Trubl G, Murray AE. 2016. Chemodenitrification in the cryoecosystem of Lake Vida, Victoria Valley, Antarctica. Geobiology 14 (6): 575 587 DOI: 10.1111/gbi.12190 Ostrom NE, Hedin LO, Von Fischer JC, Robertson GP. 2002. Nitrogen transformations and NO3 - removal at a soil stream interf ace: a stable isotope approach. Ecological Applications 12 (4): 1027 1043 DOI: 10.1890/1051 - 0761(2002)012[1027:NTANRA]2.0.CO;2 Paul MJ, Meyer JL. 2001. Streams in the Urban Landscape. Annual Review of Ecology and Systematics 32 (1): 333 365 DOI: 10.1146/annurev.ecolsys.32.081501.114040 Payne WJ. 1973. Reduction of nitrogenous oxides by microorganisms. Bac teriological Reviews 37 (4): 409 452 Peterson BJ, Wollheim WM, Mulholland PJ, Webster JR, Meyer JL, Tank JL, Martí E, Bowden WB, Valett HM, Hershey AE, et al. 2001. Control of nitrogen export from watersheds by headwater streams. Science 292 (5514): 86 90 DOI: 10.1126/science.1056874 Quick AM, Reeder WJ, Farrell TB, Tonina D, Feris KP, Benner SG. 2016. Controls on nitrous oxide emissions from the hyporheic zones of streams. Environmental Science & Techn ology 50 (21): 11491 11500 DOI: 10.1021/acs.est.6b02680 Ravishankara AR, Daniel JS, Portmann RW. 2009. Nitrous oxide (N2O): the dominant ozone - depleting substance emitted in the 21st century. Science 3 26 (5949): 123 125 DOI: 10.1126/science.1176985 Reddy KR, Patrick WH. 1975. Effect of alternate aerobic and anaerobic conditions on redox potential, organic matter decomposition and nitrogen loss in a flooded soil. Soil Biology and Biochemistry 7 (2): 87 94 DOI: 10.1016/0038 - 0717(75)90004 - 8 Repert DA, Barber LB, Hess KM, Keefe SH, Kent DB, LeBlanc DR, Smith RL. 2006. Long - term natural attenuation of carbon and nitrogen within a groundwater plume after removal of the treated wastewater source. Environmental science & technology 40 (4): 1154 11 62 DOI: 10.1021/es051442j Robertson GP, Tiedje JM. 1987. Nitrous oxide sources in aerobic soils: Nitrification, denitrification and other biological processes. Soil Biology and Biochemistry 19 (2): 187 193 D OI: 10.1016/0038 - 0717(87)90080 - 0 Robertson LA, Kuenen JG. 1984. Aerobic denitrification: a controversy revived. Archives of Microbiology 139 (4): 351 354 DOI: 10.1007/BF00408378 Rockström J, Steffen W, Noone K, Persson Å, Chapin FSI, Lambin E, Lenton T, Scheffer M, Folke C, Schellnhuber HJ, et al. 2009. Planetary Boundaries: Exploring the Safe Operating Space for Humanity. Ecology and Society 14 (2) DOI: 10.5751/ES - 03180 - 140232 101 Rosamond MS, Thuss SJ, Schiff SL. 2012. Dependence of riverine nitrous oxide emissions on dissolved oxygen levels. Nature geoscience 5 (10): 715 DOI: 10.1038/ngeo1556 Rosenberry DO, LaBaugh JW. 2008. Field techniques for estimating water fluxes between surface water and ground water. Techniques and Methods 4 - D2. Geological Survey (US). Available at: http://pubs.usgs.gov/tm/04d02/ [Accessed 16 June 2015] Rosenberry DO, Lewandowski J, Meinikmann K, Nützmann G. 2015. Groundwater the disregarded component in lake water and nutrient budgets. Part 1: effects of groundwater on hydrology. Hydrological Processes 29 (13): 2895 2921 DOI: 10.1002/hyp.10403 Rosenberry DO, Sheibley RW, Cox SE, Simonds FW, Naftz DL. 2013. Temporal variability of exchange between groundwater and surface water based o n high - frequency direct measurements of seepage at the sediment - water interface. Water Resources Research 49 (5): 2975 2986 DOI: 10.1002/wrcr.20198 Ruhala SS, Zarnetske JP, Long DT, Lee - Cullin JA, Plont S, Wiewiora ER. 2018. Exploring dissolved organic carbon cycling at the stream groundwater interface across a third - order, lowland stream network. Biogeochemistry 137 (1 2): 105 126 DOI: 10.1007/s10533 - 017 - 0404 - z Rysgaard S, Risgaard - Petersen N, Nielsen LP, Revsbech NP. 1993. Nitrification and Denitrification in Lake and Estuarine Sediments Measured by the 15N Dilution Technique and Isotope Pairing. Applied and Environmental Microbiology 59 (7): 2093 209 8 Santelli CM, Chaput DL, Hansel CM. 2014. Microbial communities promoting Mn (II) oxidation in Ashumet Pond, a historically polluted freshwater pond undergoing remediation. Geomicrobiology Journal 31 (7): 605 616 DOI: 10.1080/01490451.2013.875605 Sawyer AH. 2015. Enhanced removal of groundwater - borne nitrate in heterogeneous aquatic sediments. Geophysical Research Letters 42 (2): DOI: 10.1002/2014GL062234 Sawyer AH, Cardenas MB. 2009. Hyporheic flow and residence time distributions in heterogeneous cross - bedded sediment. Water Resources Research 45 (8) DOI: 10.1 029/2008WR007632 Schlesinger WH, Reckhow KH, Bernhardt ES. 2006. Global change: The nitrogen cycle and rivers. Water Resources Research 42 (3) DOI: 10.1029/2005WR004300 Scruggs CR, Mitzman R, MahmoodPoor Dehkordyt F, Briggs MA, Day - Lewis FD, Lane JW. - domain porosity apparatus to study water exchange with less - mobile Meeting, Graduate Virtual Poster Showcas e, San Francisco, California, 2016. 102 Seitzinger S, Harrison JA, Böhlke JK, Bouwman AF, Lowrance R, Peterson B, Tobias C, Drecht GV. 2006. Denitrification across landscapes and waterscapes: a synthesis. Ecological Applications 16 (6): 2064 2090 DOI: 10.1890/1051 - 0761(2006)016%5B2064:DALAWA%5D2.0.CO;2 Seitzinger SP, Kroeze C. 1998. Global distribution of nitrous oxide production and N inputs in freshwater and coastal marine eco systems. Global biogeochemical cycles 12 (1): 93 113 DOI: 10.1029/97GB03657 Seitzinger SP, Styles RV, Boyer EW, Alexander RB, Billen G, Howarth RW, Mayer B, Van Breemen N. 2002. Nitrogen retention in rivers: model development and application to watersheds in the northeastern USA. Biogeochemistry 57/58 : 199 237 DOI: 10.1007/978 - 94 - 017 - 3405 - 9_6 Sexstone AJ, Revsbech NP, Parkin TB, Tiedje JM. 1985. Direct Measurement of Oxygen Profiles and Denitrification Rates in Soil Aggregates 1. Soil science society of America journal 49 ( 3): 645 651 DOI: 10.2136/sssaj1985.03615995004900030024x Smidt SJ, Cullin JA, Ward AS, Robinson J, Zimmer MA, Lautz LK, Endreny TA. 2015. A Comparison of Hyporheic Transport at a Cross - Vane Structure and Natural Riffle. Groundwater 53 (6): 859 871 DOI: 10.1111/gwat.12288 Smith RL, Bohlke JK, Song B, Tobias CR. 2015. Role of anaerobic ammonium oxidation (anammox) in nitrogen removal from a freshwater aquifer. Environmental Science & Technology 49 (20): 12169 12177 DOI: 10.1021/acs.est.5b02488 Smith RL, Howes BL, Duff JH. 1991. Denitrification in nitrate - contaminated groundwater: occurrence in steep vertical geochemical gradients. Geochimica et Cosmochimica Acta 55 (7): 1815 1825 DOI: 1 0.1016/0016 - 7037(91)90026 - 2 Sobczak WV, Findlay S, Dye S. 2003. Relationships between DOC bioavailability and nitrate removal in an upland stream: An experimental approach. Biogeochemistry 62 (3): 309 327 DOI: 10.1023/A:1021192631423 Sobczak WV, Hedin LO, Klug MJ. 1998. Relationships between bacterial productivity and organic carbon at a soil stream interface. Hydrobiologia 386 (1 3): 45 53 Sophocleous M. 2002. Interactions between groundwater and surfa ce water: the state of the science. Hydrogeology journal 10 (1): 52 67 DOI: 10.1007/s10040 - 001 - 0170 - 8 Steffen W, Richardson K, Rockström J, Cornell SE, Fetzer I, Bennett EM, Biggs R, Carpenter SR, Vr ies W de, Wit CA de, et al. 2015. Planetary boundaries: Guiding human development on a changing planet. Science 347 (6223): 1259855 DOI: 10.1126/science.1259855 Stelzer RS. 2015. Yearlong Impact of Bur ied Organic Carbon on Nitrate Retention in Stream Sediments. Journal of Environmental Quality 44 (6): 1711 1719 DOI: 10.2134/jeq2015.02.0073 103 Stelzer RS, Scott JT, Bartsch LA. 2015. Buried particulate o rganic carbon stimulates denitrification and nitrate retention in stream sediments at the groundwater surface water interface. Freshwater Science 34 (1): 161 171 DOI: 10.1086/678249 Stelzer RS, Scott JT, Bartsc h LA, Parr TB. 2014. Particulate organic matter quality influences nitrate retention and denitrification in stream sediments: evidence from a carbon burial experiment. Biogeochemistry 119 (1 3): 387 402 DOI: 10.1007/s10533 - 014 - 9975 - 0 Stoliker DL, Repert DA, Smith RL, Song B, LeBlanc DR, McCobb TD, Conaway CH, Hyun SP, Koh D - C, Moon HS, et al. 2016. Hydrologic Controls on Nitrogen Cycling Processes and Functional Gene Abundance in Sediments of a Ground water Flow - Through Lake. Environmental Science & Technology 50 (7): 3649 3657 DOI: 10.1021/acs.est.5b06155 Storey RG, Fulthorpe RR, Williams DD. 1999. Perspectives and predictions on the microbial ecol ogy of the hyporheic zone. Freshwater Biology 41 (1): 119 130 DOI: 10.1046/j.1365 - 2427.1999.00377.x Syakila A, Kroeze C. 2011. The global nitrous oxide budget revisited. Greenhouse Gas Measure ment and Management 1 (1): 17 26 DOI: 10.3763/ghgmm.2010.0007 Thomas SA, Valett HM, Mulholland PJ, Fellows CS, Webster JR, Dahm CN, Peterson CG. 2001. Nitrogen Retention in Headwater Streams: The Influence of Groundwater - Surface Water Exchange. The Scientific World Journal DOI: 10.1100/tsw.2001.272 Tiedje JM, Sexstone AJ, Myrold DD, Robinson JA. 1983. Denitrification: ecological niches, competition and survival. Antonie van Leeuwenhoek 48 (6): 569 583 DOI: 10.1007/BF0 0399542 Tiedje JM, Sexstone AJ, Parkin TB, Revsbech NP. 1984. Anaerobic processes in soil. Plant and Soil 76 (1 3): 197 212 DOI: 10.1007/BF02205580 Tonina D, de Barros FPJ, Marzadri A, Bellin A. 2016. Does streambed heterogeneity matter for hyporheic residence time distribution in sand - bedded streams? Advances in Water Resources 96 : 120 126 DOI: 10.1016/j.advwatres.2016.07.009 Triska FJ, Duff JH, Avanzino RJ. 1990. Influence of exchange flow between the channel and hyporheic zone on nitrate production in a small mountain stream. Canadian Journal of Fisheries and Aquatic Sciences 47 (11): 2099 2111 DOI: 10.1139/f90 - 235 Triska FJ, Duff JH, Avanzino RJ. 1993. The role of water exchange between a stream channel and its hyporheic zone in nitrogen cycling at the terrestrial aquatic interface. Hydrobiologia 251 (1 3): 167 184 DOI: 10.1007/BF00007177 US EPA O. 2016. EPA Approves Massachusetts Plan to Protect Cape Cod Waters. US EPA Available at: http:// www.epa .gov/newsreleases/epa - approves - massachusetts - plan - protect - cape - cod - waters [Accessed 13 June 2017] 104 Vidon PG, Hill AR. 2004. Landscape controls on the hydrology of stream riparian zones. Journal of Hydrology 292 (1): 210 228 DOI: 10.1016/j.jhydrol.2004.01.005 Vitousek PM, Aber JD, Howarth RW, Likens GE, Matson PA, Schindler DW, Schlesinger WH, Tilman DG. 1997. Human Alteration of the Global Nitrogen Cycle: Sources and Consequences. Ecological Applic ations 7 (3): 737 750 DOI: 10.1890/1051 - 0761(1997)007%5B0737:HAOTGN%5D2.0.CO;2 Walter DA, LeBlanc DR. 1997. Geochemical and hydrologic considerations in remediating phosphorus - contaminated ground water in a sewage plume near Ashumet Pond, Cape Cod, Massachusetts. US Geological Survey; Branch of Information Services [distributor] . Av ailable at: http:// pubs.er.usgs.gov/publication/ofr97202 Walter DA, Masterson JP. 2011. Estimated Hydrologic Budgets of Kettle - Hole Ponds in Coastal Aquifers of Southeastern Massachusetts. U.S. G eological Survey Scientific Investigations Report 2011 5137. Available at: http:// pubs.usgs.gov/sir/2011/5137/ Walter DA, Whealan AT. 2004. Simulated Water Sources and Effects of Pumping on Surface and Gr ound Water, Sagamore and Monomoy Flow Lenses, Cape Cod, Massachusetts. U.S. Geological Survey Scientific Investigations Report 2004 - 5181. Available at: http:// pubs.er.usgs.gov/publication/sir2 0045181 Ward AS, Fitzgerald M, Gooseff MN, Voltz TJ, Binley AM, Singha K. 2012. Hydrologic and geomorphic controls on hyporheic exchange during base flow recession in a headwater mountain stream. Water Resources Research 48 (4) DOI: 10.1029/2011WR011461 Ward AS, Schmadel NM, Wondzell SM, Harman C, Gooseff MN, Singha K. 2016. Hydrogeomorphic controls on hyporheic and riparian transport in two headwater mountain streams during base flow recession. Water Reso urces Research DOI: 10.1002/2015WR018225 Weiss RF. 1970. The solubility of nitrogen, oxygen and argon in water and seawater. Deep Sea Research and Oceanographic Abstracts 17 (4): 721 735 DOI: 10.1016/0011 - 7471(70)90037 - 9 Weiss RF, Price BA. 1980. Nitrous oxide solubility in water and seawater. Marine chemistry 8 (4): 347 359 DOI: 10.1016/0304 - 4203(80)90024 - 9 Whitmire SL, Hamilton SK. 2005. Rapid removal of nitrate and sulfate in freshwater wetland sediments. Journal of Environmental Quality 34 (6): 2062 2071 DOI: 10.2134/jeq2004 .0483 Williams M, Hopkinson C, Rastetter E, Vallino J. 2004. N budgets and aquatic uptake in the Ipswich River basin, northeastern Massachusetts. Water Resources Research 40 (11) DOI: 10.1029/2004WR003172 Winter TC. 1999. Relation of streams, lakes, and wetlands to groundwater flow systems. Hydrogeology Journal 7 (1): 28 45 DOI: 10.1007/s100400050178 105 Winter TC, Harvey JW, Franke OL, Alley WM. 1998. Groun d water and surface water: a single resource . DIANE Publishing Inc. Available at: http:// pubs.usgs.gov/circ/circ1139/ [Accessed 25 March 2018] Wollheim WM, Harms TK, Peterson BJ, Morkeski K, Hopkinson CS, Stewart RJ, Gooseff MN, Briggs MA. 2014. Nitrate uptake dynamics of surface transient storage in stream channels and fluvial wetlands. Biogeochemistry 120 (1 3): 239 257 DOI: 10.1007/s10533 - 014 - 9993 - y Wollheim WM, Pellerin BA, Vörösmarty CJ, Hopkinson CS. 2005. N Retention in Urbanizing Headwater Catchments. Ecosystems 8 (8): 871 884 DOI: 10.1007/s10021 - 005 - 0178 - 3 Wondzell SM. 2011. The role of the hyporheic zone across stream networks. Hydrological Processes 25 (22): 3525 3532 DOI: 10.1002/hyp.8119 Wondzell SM, Swanson FJ. 1996. Seasonal and Storm Dynamics of the Hyporheic Zone of a 4th - Order Mountain Stream. I: Hydrologic Processes. Journal of the North American Benthological Society 15 (1): 3 19 DOI: 10.2307/1467429 Wörman A, Packman AI, Jo hyporheic zones on longitudinal transport of solutes in streams and rivers. Water Resources Research 38 (1) DOI: 10.1029/2001WR000769 Zarnets ke JP, Haggerty R, Wondzell SM, Baker MA. 2011a. Dynamics of nitrate production and removal as a function of residence time in the hyporheic zone. Journal of Geophysical Research: Biogeosciences 116 (1): 72 DOI: 10.1029/2010JG001356 Zarnetske JP, Haggerty R, Wondzell SM, Baker MA. 2011b. Labile dissolved organic carbon supply limits hyporheic denitrification. Journal of Geophysical Research: Biogeosciences 116 (March 2010): G04036 DOI: 10.1029/2011JG001730 Zarnetske JP, Haggerty R, Wondzell SM, Bokil VA, González - Pinzón R. 2012. Coupled transport and reaction kinetics control the nitrate source - sink function of hyporheic zones. Water Resources Research 48 (11): W11508 DOI: 10.1029/2012WR013291 Zimmer MA, Lautz LK. 2014. Temporal and spatial response of hyporheic zone geochemistry to a storm event. Hydrological Processes 28 (4): 2324 2337 DOI: 10.1002/hyp.9778