Article pubs.acs.org/est
Transport Zonation Limits Coupled Nitrification-Denitrification in Permeable Sediments Adam J. Kessler,*,† Ronnie N. Glud,‡,§,∥,⊥ M. Bayani Cardenas,# and Perran L. M. Cook† †
Water Studies Centre, Monash University, Clayton, Victoria 3800, Australia University of Southern Denmark and NordCEE, 5230 Odense M, Denmark § Greenland Climate Research Centre, Greenland Institute of Natural Resources, 3900 Nuuk, Greenland ∥ Arctic Research Centre, Aarhus University, 8000 Aarhus C, Denmark ⊥ Scottish Marine Institute, Scottish Association of Marine Science, Oban, Argyll PA37 1QA, Scotland # Geological Sciences, University of Texas at Austin, Austin, Texas 78712, United States ‡
ABSTRACT: Measurement of biogeochemical processes in permeable sediments (including the hyporheic zone) is difficult because of complex multidimensional advective transport. This is especially the case for nitrogen cycling, which involves several coupled redox-sensitive reactions. To provide detailed insight into the coupling between ammonification, nitrification and denitrification in stationary sand ripples, we combined the diffusion equilibrium thin layer (DET) gel technique with a computational reactive transport biogeochemical model. The former approach provided high-resolution two-dimensional distributions of NO3− and 15N−N2 gas. The measured two-dimensional profiles correlate with computational model simulations, showing a deep pool of N2 gas forming, and being advected to the surface below ripple peaks. Further isotope pairing calculations on these data indicate that coupled nitrification-denitrification is severely limited in permeable sediments because the flow and transport field limits interaction between oxic and anoxic pore water. The approach allowed for new detailed insight into subsurface denitrification zones in complex permeable sediments.
■
INTRODUCTION
In permeable sediments, transport of solutes is dominated by advection − transport by the bulk flow of pore water through the sediment.12 On the continental shelf this advection is mainly caused by pressure gradients over ripples or mounds generated by interaction of flow of the overlying water column with topographic features; the direction of this flow is upward at ripple peaks or mounds, while at ripple troughs downward flow is observed.13 This advection results in a two- or threedimensional distribution of solutes − and biogeochemical processes − throughout the sediment. Recent developments in sensors have been able to capture this complex chemical distribution. Planar optodes have been employed both in laboratory experiments and in the field to show the oxic, wellflushed region below a ripple trough in sandy sediments.7,14 Other works have shown distributions of other redox-sensitive species such as Fe, Mn, NO3−, and NH4+, although even these measurements are usually based on interpolation between a small number of vertical profiles.15 An important finding has been that in rippled or uneven permeable sediment environments, this advective flow is able to break the redox seal, with
1
Despite covering the majority of continental shelves, permeable (sandy) sediments are poorly understood in terms of their biogeochemical characteristics and function. Traditional aquatic biogeochemistry has focused on cohesive (muddy) sediments2 (although we note the exception of hyporheic zone research in streams, which has received extensive attention, for example on subsurface exchange,3 O2 uptake,4 and nutrient retention5). These cohesive sediments are characterized by redox stratification and a relatively well-understood redox cascade, driven by diffusion of solutes through the sediment.6 In these systems, the sediment is typically capped by a layer of oxidized solutes (O2, NO3−, FeIII, etc.), which prevents deep, reduced solutes from reaching the surface. Diffusion is typically too slow for reduced solutes to be transported through this layer before being at least partially oxidized, a phenomenon known as the redox seal.7 This results in strong interaction between the reduced and oxidized species, such as that exhibited in coupled nitrification-denitrification. Cohesive sediments have therefore been extensively modeled, based on these easily reproducible behaviors.8 The main complication of this transport is burrow ventilation by burrowing fauna,9 which can break the redox seal, though this too has been extensively modeled.10,11 © 2013 American Chemical Society
Received: Revised: Accepted: Published: 13404
July 26, 2013 October 8, 2013 November 13, 2013 November 13, 2013 dx.doi.org/10.1021/es403318x | Environ. Sci. Technol. 2013, 47, 13404−13411
Environmental Science & Technology
Article
two- or even three-dimensional activity distribution in permeable sediments.15 For nongaseous analytes, diffusive equilibrium in thin layer (DET) gels have long provided an excellent method for collecting high-resolution pore water profiles of common analytes including NO3−, NH4+, Fe, and other metals.29−31 It has also been employed to measure Fe concentrations in two dimensions by cutting the film into squares.32 A somewhat similar two-dimensional “peeper” device, with discretized wells filled with water, has been applied to assess two-dimensional phosphate distribution in mud33 or phosphate and Fe in an aquifer.34 Diffusive gradient in thin layer (DGT) gels, a similar method to DETs which incorporate a chemical sink to concentrate analytes and improve detection limits, has also been used in two dimensions to measure phosphorus, S2‑, and Fe(II), or all three simultaneously.35 If thin enough, DGTs can be placed over a planar O2 optode to correlate species measured using DGT with O2 concentration.36 Stief et al. (2010)37 recently used diffusive equilibrium in thin layer (DET) gels with 15N isotope labeling to measure the vertical distribution of dissimilatory nitrate reduction to ammonium (DNRA) activity. To our knowledge, there is no example of a two-dimensional DET device in the literature used to measure either nitrogen dynamics or for any gaseous analyte. Here we use a combined two-dimensional DET gel and numerical flow and transport modeling approach to demonstrate the breaking of the redox seal by nitrogen species in permeable sediments and the resultant inhibition of coupled nitrification-denitrification in such systems.
the upwelling under ripple peaks providing a direct transport pathway between the deep, reducing sediment, and the overlying water.7 The implications of this flow-dependent redox seal for biogeochemical reactions − especially coupled processes − are not yet fully known. Nitrogen cycling, in particular, has not been well studied in permeable sediments. Bioavailable nitrogen is known to be a key driver of primary production and is effectively removed from aquatic ecosystems as N2 by only two mechanisms: denitrification and anaerobic ammonium oxidation by nitrite (anammox). In most benthic environments, denitrification − the stepwise reduction of NO3− to N2 − is considered to be the most important removal mechanism.16,17 Coupled nitrificationdenitrification (Dn) often accounts for a large percentage of denitrification, involving the obligatory aerobic oxidation of NH4+ to NO3− followed by denitrification in an anoxic part of the sediment.18 We have hypothesized previously that one consequence of breaking the redox seal is that coupled nitrification-denitrification is limited in permeable sediments.19 This conclusion contrasts with other studies, which have suggested that coupled nitrification-denitrification increases with increasing pore water advection;20 however, these conclusions were based on column experiments which have an artificial and unrealistic flow field, which may not replicate environmentally relevant conditions.21 Quantifying biogeochemical processes in permeable sediments is challenging. In cohesive sediments, bulk denitrification rates can be easily measured in chambers using acetylene inhibition methods,22 15N isotope pairing,23 or direct measurements such as N2:Ar ratios.24 These measurements, however, do not properly replicate advective flow in these sediments and are not easily applicable to permeable sediments.21 Nonetheless, in recent years several studies of biogeochemical rates and fluxes over permeable sediments have still employed benthic chambers,25 which may provide results which are not representative of environmental conditions and thus may be of limited relevance.14 Instead, flume tanks should be used, which can reproduce realistic advection and provide better insight. Previously, we used 15N labeling techniques in a flume tank to measure bulk rates and to guide the development of a numerical flow and transport model for denitrification in permeable sediments, and it is from this model that we predict that the redox seal inhibits coupled nitrification-denitrification, but this effect has never been directly measured and quantified.19 One-dimensional profiles of nitrogen species, usually captured with electrodes or vertical dialysis samplers, are standard tools in nitrogen cycle biogeochemistry but do not fully capture the two-dimensional solute distribution in sands. In cohesive sediments, the use of N2O microsensors in combination with the acetylene inhibition method allow vertical profiles of denitrification activity to be obtained; however, this method poorly estimates actual denitrification rates and is best used as a relative tool.26 Profiles of N2 have also been obtained using membrane inlet mass spectrometry (MIMS); however, these involve production of customized inlet probes for a MIMS instrument and have relatively poor sensitivity owing to high background concentrations of N2.27 All of the above methods can be coupled with inverse modeling, such as the “Profile” software package, to infer profiles of biogeochemical activity from concentration profiles. 28 None of these approaches, however, would be able to resolve the complex
■
METHODS 2D Gel Preparation. In order to measure in two dimensions, a support plate was constructed from acrylic. The plate was a 10 cm × 3.5 mm square plate with a 10 × 10 array of 9.5 mm square, 2.5 mm deep wells (Figure 1a). Each well had a 2 mm round hole drilled through the bottom, which was used to remove the gel at the end of the experiment. The size of the wells (∼0.23 mL) was chosen so that the concentration of 15N2 in each well would be within the detection limit of our Isotope Ratio Mass Spectrometer, based on the 15N2 concentrations observed in a previous study.19 While this has effects on both equilibration time and smearing, these were unavoidable due to instrumental detection limits. As with the original DET gel technique used to measure dissolved Fe gradients,31 the basic principle involves equilibration of pore water with the gel, which is immobilized once the gel is removed. Polyacrylamide gel was prepared as described previously37 from 80 mL of acrylamide (15%, wt/vol), 40 mL of N,Nmethylenebisacrylamide (2%, wt/vol), 1.5 mL of dipotassium peroxodisulfate (0.11 M), and 120 μL of tetramethylethylenediamine (all Sigma Aldrich). Gels were cast directly into the support plate. The wells were partially filled with gel matrix using a syringe, and a square cellulose acetate support was placed in each well to provide structural integrity. Each well was then filled, and a sheet of plastic was placed over the top to provide a smooth surface for setting. Once set (∼30 min), the gels were kept under water. Once solidified the gels were stable and did not shrink or swell for several weeks, even when placed in seawater. Flume Experiment. For the flume experiment, the gel plate was fitted into a larger plate (∼50 cm long), which served as a new internal wall for the flume (Figure 1b). As such, the gel 13405
dx.doi.org/10.1021/es403318x | Environ. Sci. Technol. 2013, 47, 13404−13411
Environmental Science & Technology
Article
After this preincubation period, a sample was collected for analysis of background water column NO3− concentration, then 15 N-NO3− (Cambridge Isotope Laboratories) was added to the water column of the flume at ∼50 μM final concentration (in the range previously measured in the North Sea in permeable sediments39), and the flume was allowed to incubate for a further 24 h. At this stage, the gel plate was removed from the flume, and each individual well extracted by pushing through the predrilled hole with a metal rod. These gels were quickly transferred to prefilled gastight vials (with 2% (w/w) ZnCl2 purged with He to reduce the nitrogen background) and sealed. This process was recorded on video camera so that the precise air-exposure time of each gel could be determined. All gels were sealed within 20 min of air exposure. After gel preservation the sediment was resuspended to release porewater N2, and a sample for 15N−N2 gas analysis was collected from the overlying water. Slight staining on the polyethersulfone membrane showed the approximate position of the SWI. Calibration. To account for the diffusive loss of N2 from the gels to the atmosphere, a series of calibration experiments were performed. The principle of the calibration was to expose the gel to known 15N2 concentrations in a filled, airtight container and then to expose those gels to air for varied time periods. 15 N−N2 standard solutions at approximately 0.5, 1, 5, and 20 μmol L−1 were prepared by reduction of labeled 15N-NO2−. To convert NO2− to N2, sulfamic acid (Sigma Aldrich) prepared in 5% HCl (Ajax) was added to the standards to achieve a final concentration of 16.5 mM and a final pH of 1.5−2. Standards were then sealed and shaken overnight. The pH of the standard was neutralized using 2 M NaOH (Ajax) to stop the conversion of NO2− to N2, and smaller versions of the gel plate (4 by 2 wells) were inserted and sealed for ∼24 h, which replicates the incubation time in the flume experiment after 15NO3− addition. The gels were then removed and left exposed to the air. After 1, 2, 5, and 20 min, two gels from each plate were sealed into preprepared gastight vials. A water sample from the original standard solution was saved and provided a t = 0 reference. In all, a minimum of 25% of the original concentration was present after 20 min of air exposure. A multivariate regression was then used to fit the observed concentration in the gels as a function of initial concentration and exposure time. This formula could then be inverted to calculate an actual concentration from observed concentration and exposure time. Sample Analysis. Samples for N2 gas were analyzed on a Sercon 20−22 isotope ratio mass spectrometer (IRMS) coupled to a gas chromatograph. Due to the dilution of 15Nenriched N2 in the gastight vial, the detection limit was approximately 0.4 μM for 29N2 and 0.3 μM for 30N2 in terms of original concentrations. NOx (NOx = NO2− + NO3−) were analyzed spectrophotometrically using a Lachat QuikChem 8000 Flow Injection Analyzer (FIA) with a detection limit of
