Anammox , denitrification and fixed-nitrogen removal in sediments of the Lower St . Lawrence Estuary

Introduction Conclusions References


Introduction
The Laurentian Great Lakes-St.Lawrence drainage basin covers about 1.32 × 10 6 km 2 and is home to approximately 35 million North Americans.The St. Lawrence River-Estuary provides the second largest freshwater discharge (11 900 m 3 s −1 ) to the ocean in North America and is subject to extensive anthropogenic N loading from urban, industrial and agricultural sources (Gilbert et al., 2007).
In estuarine systems, N often limits primary production (Capone et al., 2008), and coastal eutrophication, resulting from nitrogen loading to rivers and estuaries, is a growing global concern (Cloern, 2001;Capone et al., 2008;Breitburg et al., 2009).In a stratified body of water, eutrophication is most often reflected by increased microbial oxygen demand and decreased oxygen availability to both benthic and pelagic organisms (Cloern, 2001;Breitburg et al., 2009).Eutrophication has been implicated in the progressive development of hypoxic bottom waters in the Lower St. Lawrence Estuary (LSLE) over the last century (Thibodeau et al., 2006;Gilbert et al., 2005;Gilbert et al., 2007).
The ability of a system to buffer anthropogenic N loading and resist the ensuing eutrophication rests largely on its capacity to remove fixed forms of N through the production and loss of N 2 gas (Capone et al., 2008).Two biogeochemical reactions, denitrification and anammox (see Fig. 1 for a schematic representation of the sedimentary N-cycle) account for nearly all N 2 production and fixed-N loss from marine and freshwater ecosystems (Canfield et al., 2005;Capone et al., 2008).The microorganisms responsible for these reactions are highly sensitive to oxygen, and marine N 2 production is therefore largely confined to anoxic environments, including coastal sediments and oxygen minimum zones (Capone et al., 2008;Canfield et al., 2005).Bottom waters over much of the LSLE are hypoxic (Gilbert et al., 2005) with O 2 concentrations as low as 50 µmol l −1 , but water column denitrification is only known to occur at O 2 concentrations < 4 µmol l −1 (Codispoti et al., 2001), and the enzymes responsible for complete denitrification exhibit varying degrees of sensitivity to O 2 (Zumft, 1997).Anammox bacteria are believed to be more O 2 tolerant, but they still appear to require O 2 concentrations below 10 µmol l −1 (Kuypers et al., 2005;Jensen et al., 2008).Thus, most fixed-N loss in the LSLE likely occurs in the underlying sediment.

Published by Copernicus
Rates of sediment N 2 production in the LSLE have been: (1) measured directly using the original isotope-pairing technique (IPT) (Wang et al., 2003); (2) estimated from NO − 3 and N 2 fluxes (Thibodeau et al., 2010;Katsev et al., 2007) and water column nitrogen deficits (Thibodeau et al., 2010); and (3) derived from diagenetic modeling (Katsev et al., 2007).Although there is variability in the reported rates, the diverse methods used yield a generally coherent picture of fixed-N removal in the LSLE: relatively high rates of N 2 production in the sediment with in situ benthic nitrification contributing significantly to the NO − 3 supply.The most recent study suggests that fixed-N removal through sedimentary N 2 production is nearly sufficient to balance nitrate inputs from the St. Lawrence River and that little nitrate exits to the Gulf of St. Lawrence (Thibodeau et al., 2010).Despite our relatively comprehensive understanding of the LSLE N-budget, the different fixed-N removal pathways have yet to be determined (Thibodeau et al., 2010) and the importance of anammox is unknown.Accurate partitioning of N-removal pathways is now possible with a recent refinement of the original isotope-pairing technique (IPT) to determine anammox rates in sediments (Risgaard-Petersen et al., 2003;Trimmer and Nicholls, 2009;Trimmer et al., 2006).Our ability to predict productivity, eutrophication, and hypoxia and their relationships in the LSLE depends on our knowledge of the specific biogeochemical processes involved.
The ubiquity of anammox in continental shelf sediments and the deep sea is becoming clear, but the factors regulating its relative importance to total N 2 production remain poorly known (Trimmer and Nicholls, 2009;Thamdrup and Dalsgaard, 2008;Francis et al., 2007).In shelf and deep-sea sediments, the importance of anammox to total N 2 production is positively correlated with water depth (Thamdrup and Dalsgaard, 2002;Trimmer and Nicholls, 2009).This correlation was explained by the progressive decrease with depth in the delivery of reactive organic matter to the sediments (Thamdrup and Dalsgaard, 2002;Dalsgaard et al., 2005); heterotrophic denitrification would be limited by the availability of these organic substrates, and the chemoautotrophic anammox process should be comparably insensitive.Anammox activity is also modulated by temperature (Dalsgaard and Thamdrup, 2002;Rysgaard et al., 2004) and the supply of nitrite either produced in situ via nitrification or diffusing from overlying water (Meyer et al., 2005;Risgaard-Petersen et al., 2005;Trimmer et al., 2005).
Anammox has also been detected in a number of estuaries (Trimmer et al., 2003(Trimmer et al., , 2005;;Meyer et al., 2005;Rich et al., 2008).The most comprehensive study to date (Nicholls and Trimmer, 2009) reports that anammox is important to N 2 production in numerous estuaries of the UK with a maximum contribution of 11 % in the Medway.In the UK estuaries, the contribution of sedimentary anammox to N 2 production is positively correlated with nitrate concentrations in the overlying waters and with sediment organic content (Nicholls and Trimmer, 2009).Given the ubiquity of the anammox reaction in marine sediments and its importance to N 2 production in UK estuaries, it is also likely important to N 2 production in the LSLE.However, most of the historical information on anammox activity is based on slurry incubations, which translate poorly to in situ rates, and the heterogeneity of estuarine ecosystems precludes reliable extrapolation of data from UK estuaries to estuaries in general (Capone et al., 2008).
An alternative pathway for N 2 production, through the direct oxidation of NH + 4 by (hydr)oxides of Fe and Mn in sediment of the LSLE, has also been proposed (Luther et al., 1997;Anschutz et al., 2000).Although thermodynamically favorable (Luther et al., 1997), conclusive evidence for the operation of this pathway in the environment remains elusive.Early tests of this pathway found no evidence for Mn-dependent NH + 4 oxidation in Mn-rich Skagerrak sediments but instead yielded early evidence for anammox in natural environments (Thamdrup and Dalsgaard, 2000).More recently, Fe-dependent NH + 4 oxidation has been reported in wetland soils (Clement et al., 2005;Shrestha et al., 2009) and wastewaters (Park et al., 2009), but the veracity of these reports remains untested and their significance is unknown.Porewater profiles in deep Indian Ocean sediments have recently provided indirect evidence for the oxidation of NH + 4 by sulfate, despite the marginal thermodynamic yield of this reaction (Schrum et al., 2009).The discovery of soluble Mn(III) species in the anoxic waters of the Black Sea and Chesapeake Bay (Trouwborst et al., 2006) and in the porewaters of the LSLE (Madison et al., 2011) raises the possibility that an additional oxidant, with the thermodynamic potential to oxidize NH + 4 to N 2 , NO − 2 or NO − 3 in the absence of O 2 , may play a role in the N-cycle.Overall, the available evidence for alternative pathways of fixed-N conversion to N 2 is inconclusive and warrants further investigation.In this work, we report quantitative rate measurements of anammox and denitrification, we partition the fixed-N removal reactions, and test for alternative pathways to N 2 in sediments from the Lower St. Lawrence Estuary.

Site description
The 300 km long, 50 km wide, and 0.3 km deep Lower St. Lawrence Estuary (LSLE) occupies the landward portion of the Laurentian Trough, a glacial bathymetric feature that extends 1200 km landward from the edge of the continental shelf (Fig. 2).Due to its great depth, the water column in the LSLE is permanently stratified with net seaward flow in the surface layer and net landward flow in the bottom layer (Saucier et al., 2003).Sediments in the channel are composed of fine-grained particulates (pelites) with, on average, 60 % clay, 35 % silt and 5 % sand (Nota and Loring, 1964).The sediments are dark yellowish-brown in the first 1-3 cm below the sediment-water interface, reflecting the presence of detrital and authigenic ferric iron [Fe(III)] and manganic [Mn(IV)] minerals (Loring and Nota, 1968;Lyle, 1983;Konig et al., 1997).Below this oxidized layer, the sediments are dark greenish-grey (Loring and Nota, 1968).

Sampling
All samples were collected during a cruise in the Lower St. Lawrence Estuary (LSLE) on the R/V Coriolis II in July of 2009.Surface and bottom water samples were collected using a 12×12-l Niskin bottle/CTD rosette (SeaBird SBE 911).The core used for the incubations was recovered at Station 23: 48 • 42.032 N, 68 • 39.171 W; 345 m depth.Overlying water O 2 , NO − 3 and soluble reactive phosphate (SRP) concentrations were 63 µmol l −1 , 34 µmol l −1 and 2.7 µmol l −1 , respectively.Sediments were recovered with minimal disturbance using an Ocean Instruments Mark II box corer (20 × 30 × 50 cm).The sediments were sectioned at various intervals in a glove box continuously flushed with N 2 to avoid oxidation artifacts (Edenborn et al., 1986), and porewaters were extracted with a N 2 overpressure using modified "Reeburgh-type" squeezers (Reeburgh, 1967;Mucci et al., 2000).Sediments for intact core incubations were subsampled from a second box core with 6 acrylic tubes (5.2 cm in diameter and 60 cm in length).Fresh bottom water was added to these sub-cores to replace the water lost during box core recovery and subsequent sub-coring.

Slurry incubations
Sediment slurries were prepared by mixing sediment from the top 2 cm of the box core with an equivalent volume of bottom water that was previously purged with ultra-highpurity He gas to remove O 2 and N 2 .The sediment slurry was subsequently purged with He for an additional 12 h to remove residual N 2 gas and allow NO − 3 present in the bottom water and sediment porewaters to be consumed.Following this 12-h period, the sediment slurry was transferred, with no headspace, into ninety 12-ml gas tight vials (Exetainers, LabCo).Isotopic labels, substrates, and specific inhibitors were added as shown in Table 1.The predicted 15 Nlabeled N 2 products of the individual experiments for a given process are presented in Table 2.The sediment slurries were incubated at 4 • C, close to the in situ bottom water temperature of 4.7 • C, mixed periodically by inversion, and sacrificed over an interval of 36 h.Upon sacrificing, 1 ml of slurry was removed from the Exetainer using a needle and syringe and replaced with He gas and 200 µl of a 37 % formaldehyde solution to stop microbial activity.The withdrawn sediment slurry (1 ml) was filtered directly through a 0.2 µm pore size syringe filter and the filtrate was frozen for later analysis.The formaldehyde-fixed sediment slurry was stored upside down in the Exetainers until isotopic analysis.

Intact core incubations
Our intact core incubations followed the refined IPT protocol described by Trimmer and Nicholls (2009), in which the isotopic composition of NO  1. Slurry incubation conditions (label additions) and labeled N 2 production rates: series A received 10 µl of a 100 mmol l −1 solution of 15 N-NH + 4 ; series B received 10 µl of a 100 mmol l −1 solution of 15 N-NO − 3 ; series C received 10 µl of a 100 mmol l −1 solution of 15 N-NO − 3 and 10 µl of a 100 mmol l −1 solution of 15 N-NH + 4 ; series D received 10 µl of a 100 mmol l −1 solution of 15 N-NH + 4 and 200 µl of a 10 mmol l −1 solution of allylthiourea (ATU), a specific inhibitor of nitrification; series E received 10 µl of a 100 mmol l −1 solution of 15 N-NH + 4 and 200 µl of a freshly prepared 40 mmol l −1 solution of Mn(III)-pyrophosphate; and series F received 10 µl of a 100 mmol l −1 solution of 15 N-NH + 4 , 10 µl of a 100 mmol l −1 solution of 15 N-NO − 3 and 200 µl of a 10 mmol l −1 solution of ATU.
Treatment 15 N-NO − Table 2. Predicted outcomes for slurry incubations (assuming no DNRA).The bold X's mark the observed labeling.The emphasized row highlights the combination of processes (denitrification and anammox) operative in the LSLE sediments (data in Fig. 4).W = Denitrification, X= Anammox, Y = Mn and Fe dependent NH + 4 oxidation to N 2 , Z= Mn and Fe dependent NH + 4 oxidation to NO x .
is determined from the isotopic composition of N 2 O, which is produced as an intermediate during denitrification, but not during anammox.Unlike previous versions of the IPT protocol, which relied either on slurry incubations or a concentration series of intact core incubations to estimate the isotopic composition of NO − 3 in the NO − 3 reduction zone, the refinement permits both denitrification and anammox rates to be calculated using a single set of intact sediment cores without slurries.Following sub-coring and replacement of the overlying water, magnetic stirring devices were inserted into the tubes and suspended 3 cm above the sediment-water interface.Each sub-core was allowed to stand and re-equilibrate at 4 • C for approximately 12 h to near in situ temperatures while the overlying water was stirred.After the equilibration period, 1.5 ml of a 100 mmol l −1 solution of 15 N-NO − 3 was added to the overlying water.Following an additional 6 h, the overlying water was sampled for the determination of N species, and the sub-cores were sealed with no headspace using thick butyl rubber stoppers.The 6 sub-cores were periodically sacrificed over the next 34 h, upon which stirring was halted and the stoppers carefully removed.The overlying water was sampled for NO x and NH + 4 analyses.Subsequently, the top 2 cm of the sediment were gently mixed into the overlying water using a plastic rod.A syringe, fitted with a steel canula, was used to withdraw the slurry from > 10 cm below the water surface to avoid atmospheric N 2 contamination.This slurry was transferred to 12-ml Exetainers, which were flushed with ∼ 36 ml (3 × the Exetainer volume) of slurry.To stop microbial activity, 200 µl of 37 % formaldehyde was added and the Exetainers closed without headspace.Samples were collected using this same technique for membrane inlet mass spectroscopy (MIMS) (Kana et al., 1994).Triplicate samples were collected in 7-ml glass tubes, fixed with 25 µl 0.1 mol l −1 HgCl 2 and sealed with a ground-glass stopper.These tubes were submerged in water and kept cold until analysis.An additional portion of the slurry was transferred to a plastic centrifuge tube and frozen for later analysis of 15 N-NH + 4 .Rate calculations based on these measurements are described in Appendix A.

Analyses
Porewater O 2 was measured with a Unisense PA2000 picoammeter and a Unisense "Clark" type microelectrode fitted with a stainless steel needle tip to prevent breakage.This electrode was calibrated with two points: Lower St. Lawrence Estuary bottom water saturated in O 2 by vigorous stirring in ambient atmosphere, and an anoxic, alkaline ascorbate solution.The detection limit for O 2 was 0.2 µmol l −1 , calculated from the standard deviation of five background measurements taken in an anoxic, alkaline ascorbate solution.Samples for the measurement of N species concentrations were transported on dry ice back to the NordCEE lab in Denmark and stored frozen until analysis.NH + 4 concentrations were measured using a gas-exchange, flow-injection method (Hall and Aller, 1992) with a detection limit of 0.1 µmol l −1 and a reproducibility of 5 % RSD.Combined NO − 3 and NO − 2 concentrations (NO x ) were determined by chemiluminescence (Braman and Hendrix, 1989), (NO x analyzer model 42c, Thermo Environmental Instruments Inc.) with a detection limit of < 10 nmol l −1 and a reproducibility of better than 5 % RSD.The slurry samples designated for isotopic analyses were maintained in their Exetainers at room temperature, and upside down when possible.The isotopic composition of N 2 was determined by injecting 25-50 µl of headspace gas into an in-house built injection system.Following injection, CO 2 was trapped using Ascarite (III), N 2 and N 2 O separated using a Poropak R GC column, and the sample stream passed through a reduction reactor to reduce N 2 O to N 2 and O 2 to H 2 O. H 2 O was trapped on Mg perchlorate, and the sample stream was introduced using a Conflo III to a Thermo Electron DELTA V plus IR-MS operated in continuous-flow mode.N 2 was measured at masses 28, 29, and 30.Similarly, the N isotopic composition of N 2 O was measured by injecting 200-1000 µl of headspace gas, but the reduction reactor was bypassed and isotopic measurements were made on masses 44, 45, and 46.Changes in N 2 concentrations and N 2 / Ar ratios were measured directly using MIMS (Kana et al., 1994).Measurements of 15 N-NH + 4 were conducted by converting NH + 4 to N 2 following oxidation by hypobromite, as described by Rysgaard and Risgaard-Petersen, (1997).In the case of the slurry incubations, NH + 4 was extracted in a 2 mol l −1 KCl solution prior to hypobromite oxidation and isotopic analysis.The reactive Mn and Fe (hydr)oxide content of the sediment used for our slurry incubations (the upper 2 cm of the sediment core) was determined using 1 M hydroxylamine-HCl and citrate-dithionite sequential, selective extractions (Poulton and Canfield, 2005).

Porewater profiles
Porewater profiles of O 2 , NO x and NH + 4 are shown in Fig. 3.After re-establishing thermal equilibrium over several hours open to the ambient atmosphere, O 2 concentrations in the water overlying the sediment-water interface were between 40 and 60 µmol l −1 .These values are similar to those measured in the bottom waters using both the oxygen sensor (Seabird SBE-42) on the CTD and Winkler titration (Grasshoff et al., 1999;Gilbert et al., 2005).Dissolved oxygen concentrations decreased logarithmically and became undetectable (< 0.2 µmol l −1 ) 6 to 9 mm below the sediment-water interface (SWI).These values are consistent with O 2 profiles measured previously (Anschutz et al., 2000;Luther et al., 1998;Katsev et al., 2007).The NO x concentration was 23 µmol l −1 in the bottom waters and decreased from 3.5 µmol l −1 in the 0-0.5 cm sediment sampling interval to 0.8 µmol l −1 in the 0.5-1.0cm depth interval.Traces of dissolved NO x (< 1 µmol l −1 ) were detected throughout the core and a small peak in NO x was observed between 6 and 8 cm below the SWI.Ammonium was undetectable (< 0.5 µmol l −1 ) in the bottom waters, increased below the sediment-water interface, and reached a maximum of 115 µmol l −1 approximately 17 cm below the SWI.The NO x and NH + 4 profiles are consistent with previous measurements (Katsev et al., 2007;Anschutz et al., 2000).

Slurry incubations
The production of 15 N-labeled N 2 in slurry incubations is shown in Fig. 4. 29 N 2 is produced from the coupling of a single unlabeled 14 N atom with a labeled 15 N atom (Fig. 4a), whereas 30 N 2 is produced from the coupling of two labeled 15 N atoms (Fig. 4b).Using combinations of different labeled N species, it is possible to identify the source of N used to produce N 2 .Volume specific rates were calculated by least squares regressions through the linear periods of N 2 production.These rates are presented in Table 1, and a summary of the expected and observed incorporation of 15 N-labeled N into N 2 is presented in Table 2.In treatments A, D, and E, in which the only labeled nitrogen was in the form of ammonium, there was no production of labeled N 2 .In treatments B, C and F, which all contained labeled nitrate, there was abundant production of labeled N 2 .Thus, in these experiments, the production of isotopically labeled N 2 requires the addition of labeled nitrate.Measurements of dissimilatory nitrate reduction (DNRA) in treatment B yielded volume specific rates of 1.7 ± 0.2 × 10 −6 µmol cm −3 h −1 .Details of each incubation series and their interpretation with respect to sediment N transformations are discussed below.

Extractions
The 1 M hydroxylamine HCl extraction of wet sediments from the upper 2 cm of the core liberated 1.5 ± 0.1 µmol Mn g −1 and 37 ± 3 µmol Fe g −1 , and the citratedithionite extraction liberated 0.3±0.03µmol Mn g −1 of wet sediment and 37 ± 4 µmol Fe g −1 of wet sediment.A porosity of 0.87 (Mucci, unpublished results) and a sediment density of 2.65 g cm −3 (Anschutz et al., 2000) yields the volume specific solid phase Mn and Fe (hydr)oxide concentrations presented in Table 3.

Intact core incubations
In intact core incubations, O 2 and NO x enter the sediment from the overlying water.Some of this O 2 is used to drive benthic nitrification, which in turn generates NO x that fuels both denitrification and anammox.As a result, anammox and denitrification proceed in the intact cores without further addition of NO x , which is rapidly consumed in the closed slurries and therefore must be supplemented.The addition of NO − 3 to the intact core incubations is solely to provide the isotopic tracer.The other advantage of using intact sediment cores over slurries is the retention of the sediment structure, which can play an important role in biogeochemical processes (Nielsen et al., 2010).Results of the intact core incubations (Table 4) provide direct measurements of N 2 production rates and the identity of the responsible pathways.Both denitrification and anammox contribute to N 2 production in the Lower St. Lawrence Estuary, and in situ nitrification accounts for a large fraction of the NO x supplied for both pathways.The calculations used to compute the rates we report in this paper are the same as those used by Trimmer and Nicholls (2009) and are summarized in Appendix A.

Discussion
The modern N-cycle and its evolution through time have been recently reviewed (Canfield et al., 2010).A schematic representation of the sedimentary N-cycle is presented in Fig. 1, which also summarizes the rates measured in this study.Of particular importance to the work presented here are the following processes: nitrification, the aerobic transformation of NH + 4 , via NH 2 OH and NO − 2 , to NO − 3 ; denitrification, the anaerobic transformation of NO − 3 , via NO − 2 , NO, and N 2 O, to N 2 ; anammox, the anaerobic transformation of NH + 4 and NO − 2 , via N 2 H 4 , to N 2 ; and dissimilatory nitrate reduction, the anaerobic reduction of NO − 3 , via NO − 2 , to NH + 4 .

Porewater profiles
With nitrate concentrations in the LSLE bottom water on the order of 23 µmol l −1 , the NO x porewater profiles (Fig. 3) show that the surface sediment is a sink for NO − 3 from the overlying water.Undetectable NH + 4 in the overlying water and a strong sub-surface NH + 4 gradient imply a large upward flux of NH + 4 towards the sediment-water interface.In the classical view of the N-cycle, these profiles would be taken to indicate that nitrification in the oxic sediment layer is the likely sink for ammonium and that denitrification, occurring just below the oxygen penetration depth (8-10 mm below the SWI), provides a sink for NO − 3 .Nitrification and denitrification are often tightly coupled near the oxic-anoxic boundary of the sediment with little loss of fixed nitrogen (NO x and NH + 4 ) to the overlying water (Thamdrup and Dalsgaard, 2008).With the recent discovery of anammox in natural environments, the classical view needs to be amended because anammox may serve as a sink for both NH + 4 and NO − 3 via the reactive intermediate NO −

. It has been proposed that NH +
4 can be anaerobically oxidized to N 2 , NO − 2 or NO − 3 by Mn (hydr)oxides or organic complexes of Fe(III) and Mn (Mn(III/IV), which are ubiquitous and abundant in many soils and sediments (Luther et al., 1997;Hulth et al., 1999;Madison et al., 2011).These reactions are thermodynamically favorable under a variety of environmental conditions and could be globally important contributors to N cycling (Luther et al., 1997).The Mn-and Fe-dependent reactions are conceptually consistent with observed N species distributions in sediments of the Lower St. Lawrence Estuary (Luther et al., 1997;Anschutz et al., 2000), and diagenetic models incorporating these reactions accurately reproduce N-species profiles (Katsev et al., 2007).

Slurry incubations and N 2 production pathways
Our slurry incubations constrain the N-transformation pathways operating in St. Lawrence Estuary sediments.In treatment A, which received an addition of 15 N-labeled NH + 4 , there was no production of 29 N 2 or 30 N 2 , and none of the added NH + 4 was converted to N 2 (Fig. 4).This demonstrates the absence of direct oxidation of NH + 4 to N 2 by the Mn(III, IV) or Fe(III) species present in these sediments.In conjunction with the results of treatment B, it also demonstrates that NH + 4 is not oxidized to NO − 3 or NO − 2 because, if it were, 15 N originating from 15 N-NH + 4 would register in the N 2 pool following denitrification of the newly produced 15 N-NO x .Treatment B, which received 15 N-NO − 3 , produced ample 30 N 2 (Fig. 4), confirming active denitrification (discussed further below).It could be argued that the reactive Mn and Fe pools in the sediment were rapidly consumed during the equilibration period prior to our incubations and thus were not available for the oxidation of NH + 4 .We can constrain this argument by considering the size of the reactive Mn(IV) and Fe(III) pools, the potential rates of Mn and Fe reduction from organic matter oxidation, and the duration of the experiments.Assuming that most of the labile organic matter is mineralized within the upper 2 cm of the sediment, the volume specific demand for oxidants can be estimated from the published oxygen uptake rates.Taking the value of 0.43 µmol cm −2 d −1 for the O 2 uptake (Katsev et al., 2007) and normalizing for the stoichiometry of oxic respiration, we estimate maximum volume specific C mineralization rates of 0.22 µmol C cm −3 d −1 .These are maximum rates because of the assumption that anaerobic C mineralization, as would occur in our slurries, would be as rapid as aerobic C mineralization, although other studies indicate that C mineralization rates during Fe(III) and Mn(IV) reduction are slower than during oxic respiration (Magen et al., 2011).Considering the stoichiometry of Mn(IV) and Fe(III) respiration and the reactive Fe and Mn (hydr)oxide concentrations operationally defined by the 1 M hydroxylamine-HCl extractions, we estimate that the reactive Mn(IV) and Fe(III) pools would be exhausted in closed anoxic incubations after 2 and 25 days, respectively.Stated differently, less than 25 % of the total a Calculated with Fick's first law using a temperature and salinity corrected diffusion coefficient, and taking into account tortuosity (Boudreau, 1996).

Biogeosciences
b Calculated as the difference between denitrification driven by a diffusive NO − 3 flux from the water column and the total diffusive NO − 3 flux.c Calculated as the sum of sediment nitrification and 0.5 of the anammox rate.N 2 production determined using MIMS: 10.19 ± 1.31 µmol m −2 h −2 .
reactive Mn(IV) and < 2 % of reactive Fe(III) would have been consumed during our 12-h incubation period at these C mineralization rates.It is unlikely therefore that the supply of reactive Mn(IV) and Fe(III) would have limited Mn(IV) and Fe(III) dependent NH + 4 oxidation during our experiments.We also tested the hypothesis that organically-complexed Mn(III) species, which have recently been discovered in the Black Sea and Chesapeake Bay (Trouwborst et al., 2006) and quantitatively measured in the LSLE sediment porewaters (Madison et al., 2011), could serve as oxidants for NH + 4 .Treatment E received both 15 N-labeled NH + 4 and Mn(III)-pyrophosphate at a concentration of 635 µmol l −1 .As with treatment A, neither 29 N 2 nor 30 N 2 were generated during the incubation (Fig. 4), demonstrating that Mn(III)pyrophosphate is not an effective oxidant of NH + 4 in the LSLE sediments, even at relatively high Mn(III) concentrations.It could be argued that Mn(III)-pyrophosphate is a strong complex which may not be kinetically reactive or (bio)available for NH + 4 oxidation.Even though the Mn(III)pyrophosphate complex stability constant is poorly constrained (Klewicki and Morgan, 1998), information on its reactivity can be gleaned from published experimental data.For example, the complex reacts readily with Fe(II) and HS − and can be used as an electron acceptor in the respiration of simple organic acids by Shewanella putrefaciens MR1 (Kostka et al., 1995).Thus, the available experimental evidence attests that both the kinetic reactivity and the bioavailability of Mn(III)-pyrophosphate make it an appropriate analogue of natural Mn(III) complexes.Our slurry experiments thus provide no evidence for the coupling of Mn(III/IV) or Fe(III) reduction with the oxidation of NH + 4 to N 2 , NO − 2 or NO − 3 .We conclude that these reactions are unlikely to take place in the LSLE sediments.
In contrast, the slurry incubations reveal that anammox occurs at high rates in LSLE sediments.In the absence of 14 N-NO 3 , which was completely consumed during the 12-h pre-equilibration period, denitrification cannot produce 29 N 2 in treatment B, which only received 15 N-labeled NO − 3 .In other words, there was no 14 N-NO − 3 available during denitrification to pair with the 15 N-NO − 3 and form 29 N 2 .Thus, we attribute the observed 29 N 2 formation (Table 1) to the anammox reaction, which in our experiment produced 29 N 2 at rates of 6.6±0.7×10−4 µmol cm −3 h −1 , coupling 15 N-NO − 2 , produced from added 15 N-NO − 3 , with naturally-occurring 14 N-NH + 4 .As the sediment was diluted 1 : 1 with seawater, we can scale these rates up by a factor of two to estimate in situ, volume specific, anammox rates of 1.32 ± 0.14 × 10 −3 µmol cm −3 h −1 .
Dissimilatory NO − 3 reduction to NH + 4 could in principle produce 15 N-NH + 4 from the 15 N-NO − 3 added, which would translate to 30 N 2 production via anammox and a corresponding underestimation of total anammox rates by only considering the 29 N 2 pool.Similarly, denitrification based on 30 N 2 production would be overestimated.However, our measured rates of dissimilatory NO − 3 reduction to NH + 4 are two orders of magnitude lower than the denitrification and anammox rates.Thus, dissimilatory NO − 3 reduction to NH + 4 has an insignificant effect on our estimates of anammox and denitrification rates.From the data presented in Fig. 4b, we can estimate maximum N 2 production rates from denitrification of 3.3±0.6×10−4 µmol cm −3 h −1 (Table 1), and in situ rates of 6.6±1.2×10−4 µmol cm −3 h −1 .In our slurry incubations, anammox would therefore account for ≥ 67 % of the total N 2 production.
We also tested for NH + 4 limitation of anammox and the possibility that trace leakage of oxygen into the Exetainers might cause nitrification.To test for NH + 4 limitation of anammox, 15 N-NH + 4 and 15 N-NO − 3 were added in treatment C. Both 29 N 2 and 30 N 2 production rates were statistically equivalent to those in treatment B ( 15 N-NO − 3 only, Fig. 4).This demonstrates that NH + 4 was not limiting for anammox and confirms that little 15 N-NH + 4 is incorporated into the 30 N 2 pool during anammox.This validates our measurements of dissimilatory nitrate reduction rates, which are very low, and confirms that 30 N 2 production is exclusively due to denitrification.Lack of 29 N 2 and 30 N 2 production in treatments A, D, and E demonstrates that nitrification rates in our slurry experiments are insignificant.Otherwise, any nitrification would be recorded in the 29 N 2 and 30 N 2 pools due to subsequent denitrification.The addition to treatment F of allylthiourea (ATU), a specific inhibitor of nitrification that blocks the oxidation of NH + 4 to NO − 2 (Hall, 1984), resulted in a weak, but statistically significant stimulation of both anammox and denitrification (Table 1).The reasons for this stimulation are unclear, but one possible explanation could be that ATU is used as an electron donor or carbon substrate by denitrifying or anammox bacteria.
The results of our slurry incubations demonstrate that Mn(III/IV)-and Fe(III)-dependent ammonium oxidation are not a significant component of the sedimentary N-cycle at Station 23 in the LSLE.Anammox, on the other hand, is an important pathway for N 2 production.The estimated volume specific denitrification and anammox rates are 0.66 and 1.3×10 −3 µmol cm −3 h −1 , respectively.Assuming that these rates are representative of the upper 2 cm of sediment and that all N 2 production occurs within this interval, these rates translate to area-specific rates of 13 and 26 µmol m −2 h −1 for N 2 production through denitrification and anammox, respectively (see Appendix for calculation details).

Intact core incubations and in situ rates
Although slurry incubations can constrain potential rates and the qualitative importance of different reaction pathways, many biogeochemical reactions are stimulated in such slurries.Incubations with intact sediment cores provide more realistic estimates of in situ rates.The recent development of a method to measure both denitrification and anammox in intact sediment cores allows us to partition N 2 production between these reactions and provides a robust estimate of their in situ rates (Trimmer and Nicholls, 2009).Results of our intact sediment core incubations are consistent with those of our slurry experiments to the effect that both anammox and denitrification are important components of the sedimentary N-cycle in the LSLE (Tables 1 and 4).Nevertheless, the anammox reaction accounts for only 33 % of total N 2 production in intact cores compared to ≥ 67 % in the slurries.Although both denitrification and anammox were stimulated in the slurry incubations relative to the intact core incubations, anammox was stimulated to a much larger extent.In the absence of nitrification, it is likely that the first step in denitrification, the conversion of NO − 3 to NO − 2 , provided 15 NO − 2 to fuel anammox.Given that the initial step of denitrification is energetically the most favorable (Zumft, 1997), complete denitrification may be inhibited under the electron-acceptorlimiting conditions of the slurry incubations, thus augment-ing the relative importance of anammox in the slurry relative to the whole core incubations.Our measurements of N 2 production rates in intact cores are in good agreement with previous measurements and model predictions (Table 4) (Katsev et al., 2007;Thibodeau et al., 2010), though the measurements made by MIMS are slightly, but not irreconcilably, lower (see Table 4 footnote).Despite the agreement between our measured rates and those modeled by Katsev et al. (2007), the modeled rates are based on a different set of biogeochemical reactions than those we observe.The model includes Fe(III) and Mn(IV) dependent NH + 4 oxidation and neglects anammox.These differences would not affect the ability of the model to reproduce current rates because it was calibrated using existing measurements.Model-based predictions of future changes would, however, be unreliable if the active sedimentary processes respond differently than model reactions to environmental changes.Similarly, previous reaction rate estimates were based on different techniques with different assumptions (Wang et al., 2003;Katsev et al., 2007).For example, rates based on NO − 3 fluxes across the sediment-water interface are blind to tightly-coupled, in situ, sedimentary nitrification and denitrification and cannot distinguish between the different possible sinks for NO − 3 (Thibodeau et al., 2010).As anammox may contribute as much as a third of the total N 2 production in sediments of the Lower St. Lawrence Estuary, it should be considered in any predictions about the future of the N-cycle in the LSLE.
In UK estuaries, the importance of anammox to N 2 production correlates positively with the concentration of NO − 3 in the overlying water and with sediment organic carbon content, but apparently not with the reactivity of the latter (Nicholls and Trimmer, 2009).The much greater percentage of N 2 production attributed to anammox in the Lower St. Lawrence Estuary cannot be ascribed to differences in organic carbon content as the sedimentary organic carbon content at our study site varies between 1.2 and 1.7 wt.%, similar to that at Medway (2.0 wt.%), which had the highest percent anammox of all the UK estuaries surveyed.Furthermore, the reactivity of organic carbon in the sediments of the LSLE, as characterized by the pseudo-first order oxic respiration reaction rate (k = 1.8 yr −1 ; Katsev et al., 2007), is broadly comparable to that of the UK estuaries (0.6 yr −1 ; Nicholls and Trimmer, 2009).Whereas NO − 3 concentrations in water overlying the Medway sediments (7-790 µmol l −1 ) are much higher than those in the LSLE (∼ 25 µmol l −1 ), the relationship between NO − 3 concentrations and the importance of anammox to N 2 production does not appear to apply to cross-system comparisons over large geographical distances.
By comparing the isotopic composition of the nitrate in the overlying waters and that of the N 2 produced through denitrification and anammox, we can discriminate between N 2 generated from NO − 3 and NO − 2 diffusing from the overlying water and NO x produced in the sediment via nitrification www.biogeosciences.net/9/4309/2012/Biogeosciences, 9, 4309-4321, 2012

Fig. 1 .
Fig. 1.Schematic illustration of the sedimentary N-cycle in the Lower St. Lawrence Estuary (LSLE).Numbers in parentheses are the rates in µmol m −2 h −1 .

Table 3 .
Extractable solid phase Fe and Mn in upper 2 cm of sediment.