INTRODUCTION

Top Introduction Discussion References

Due to increasing nitrogen concentrations in surface waters (6) and eutrophication of coastal waters (14, 26, 51), the quantification of denitrification rates in sediments of lakes, rivers, and estuaries has gained importance. Different methods for measuring denitrification have been developed: the acetylene inhibition technique (41), measurement of nitrate disappearance (2), calculation of nitrate fluxes into the sediment from pore water profiles (20), the ¹⁵N nitrate dilution method (16), direct measurement of N₂ production (40), the nitrogen mass-balance approach (1), and the ¹⁵N isotope pairing technique (IPT) (23). All of these methods have advantages but also potential problems and limitations (38, 39).

The acetylene inhibition technique is a very simple method. It is widely used, and its sensitivity has been improved by N₂ measurements based on N₂/Ar ratios. However, many different studies have documented that the acetylene method systematically underestimates denitrification (20). Several causes may be responsible for artifacts: (i) inhibition of coupled nitrification-denitrification by acetylene (19), (ii) incomplete blockage of N₂O reductase at low nitrate concentrations (3), (iii) incomplete blockage of N₂O reductase when sulfide is present (3), (iv) diffusion of N₂O toward deeper sediment layers and reduction to N₂ (39), and (v) catalytic oxidation of NO into NO₂ (4). The disappearance of nitrate can overestimate denitrification because nitrate may not only be denitrified but also reduced to ammonia (16, 20) or assimilated (36). On the other hand, it may also underestimate denitrification because it does not consider coupled nitrification-denitrification. The calculation of nitrate fluxes into the sediment from pore water profiles has the same problems as described for the nitrate disappearance method. In addition, the resulting fluxes across the sediment-water interface and the diffusive boundary layer often underestimate denitrification, because of insufficient vertical resolution of the profiles and because only diffusive and not turbulent transport is considered when the fluxes are calculated (11, 21). The ¹⁵N nitrate dilution method with subsequent measurement of ¹⁵N nitrate disappearance and ¹⁵N ammonia production quantifies denitrification and nitrification but still neglects coupled nitrification-denitrification and assimilation. Direct measurement of N₂ production has the advantage to include coupled nitrification-denitrification but requires a very sensitive analysis because of the small N₂ production compared to the N₂ background. The mass-balance approach on larger systems such as whole lakes may lead to considerable errors due to a combination of the errors of each term in the mass-balance.

The ¹⁵N IPT has the advantage that denitrification of both NO₃ diffusing from the overlaying water and NO₃ from nitrification within the sediment can be quantified.

The purpose of this study is to briefly describe the principles of the ¹⁵N IPT, to review the different applications of the ¹⁵N IPT in sediments, to point out the advantages and the limitations of the method, and to assess the main research topics, which benefit from different applications of the method. The review covers publications between 1992 and 2000.

	PRINCIPLES OF THE 15N IPT

Denitrification in the sediment can occur at the expense of NO₃ from the water column (D_W) or of NO₃ produced within the sediment by nitrification (D_n). These two pathways can be analyzed by the ¹⁵N IPT, which was developed by Nielsen (23). The method relies on stable isotope tracers. The natural abundance of nitrogen isotopes is 99.64% of ¹⁴N and 0.36% of ¹⁵N.

¹⁵NO₃ tracer is added to the sediment overlying water. This ¹⁵NO₃ mixes with the ¹⁴NO₃ present in the water column and in the upper sediment layer. Denitrification of this nitrate mixture (D^tot) produces N₂ molecules with possible molecular masses of 28, 29, and 30 according to the isotopic signature of the tracer mixture.

From the production rate of ²⁹N₂ (r₂₉) and ³⁰N₂ (r₃₀) it is possible to calculate denitrification of ¹⁵NO₃ (D₁₅) as follows:

(1)

For the calculation of the denitrification rate D₁₄ of unlabeled ¹⁴NO₃, Nielsen (23) derived the relation:

(2)

The total denitrification rate in the sediment is, of course:

(3)

The denitrification rate of ¹⁵NO₃ (D₁₅) allows us to calculate denitrification of the ¹⁴N/¹⁵N nitrate mixture diffusing from the water column into the sediment (D):

(4)

where  represents the isotopic nitrate enrichment during the incubation. It can normally be expressed as:

(5)

where the brackets indicate concentrations and the subscripts a and b refer to after and before the ¹⁵NO₃ tracer addition, respectively.

Finally, coupled nitrification-denitrification within the sediment (D_n) can be estimated by the difference:

(6)

If we assume that D follows a linear increase with higher tracer concentrations, this rate can now be extrapolated back to tracer-free conditions in order to obtain the natural denitrification rate D_W with nitrate diffusing from the water column (15):

(7)

All of the parameters mentioned here are depicted schematically in Fig. 1. Note that to simplify the figure we neglected that nitrogen is naturally enriched with a small amount of ¹⁵N.

View larger version (143K): [in this window] [in a new window]   FIG. 1.   Schematic representation of the transformation rates during a 15NO3 tracer experiment. D is the total denitrification of nitrate from the water column. Dw is the denitrification of nitrate from the water column without tracer addition. Coupled nitrification-denitrification is given as Dn. Dtot refers to the total denitrification rate during the tracer experiment. The specific denitrification rates of 15N and 14N nitrate are given as D15 and D14, respectively, and the production rates of N2 with masses of 28, 29, and 30 are represented by r28, r29, and r30, respectively.

In the special case wherein there is initially no nitrate in the water column, the added ¹⁵NO₃ is not diluted. Therefore, D corresponds to D₁₅, D_W is zero, and D_n equals D₁₄.

The reader is referred to the methodological studies (15, 23) for more details about the derivation of these equations.

	APPLICATION OF THE 15N IPT TECHNIQUE

The ¹⁵N IPT technique has basically been applied in four different ways: (i) in batch-mode assays (5, 7, 9, 10, 12, 13, 15, 17, 18, 19, 20, 23, 24, 25, 27, 28, 29, 30, 34, 35, 42, 43, 44, 45, 46, 47, 49, 50, 52, 53), (ii) in benthic flux chambers (20, 24), (iii) in enclosures (34), and (iv) in flow through systems (32, 33, 36, 37). A wide range of aquatic systems have been investigated thus far. The experimental conditions, and the resulting denitrification rates are summarized in Table 1.

Batch-mode assays. The batch-mode assay (Fig. 2a) developed by Nielsen (23) has been applied to quantify denitrification in sediments of streams (23, 30), lakes (20, 34, 42, 45, 47, 52), estuaries (5, 7, 9, 13, 17, 25, 27, 28, 29, 35), coastal waters (12, 15, 24, 43, 44, 46, 49, 50, 53), shelf sediments (19), shallow reservoirs (10), and mangrove forests (18). In all studies, undisturbed 3- to 15-cm-long sediment cores (internal diameter, 2.6 to 20.0 cm) overlaid with 5 to 40 cm of water were sampled. In most studies the cores were submersed uncapped into a tank containing 10 to 70 liters of bottom water. In some studies the bottom water was replaced with artificial water (23, 45, 47, 52). The water overlying the sediment was stirred with a magnetic stirrer (mostly 2 to 10 cm above the sediment) coupled to an external rotating magnet. Lohse et al. (19) switched the momentum of these magnets from clockwise to anticlockwise rotation every 5 s to minimize the pressure gradient at the sediment water interface. Jensen et al. (15) circulated the tank water with a pump. The water temperature was held constant at in situ conditions.

View larger version (104K): [in this window] [in a new window]   FIG. 2.   Set up of the application of the 15N IPT. The gray area corresponds to the sediment, the dotted area indicates the water without tracer, and the hatched area shows the water with tracer addition. (a) Batch-mode assay. The figure represents a water reservoir (labeled 1) with the incubation container filled with water (labeled 2), with three cylinders containing sediment, water, and a magnetic stirrer closed with rubber stoppers (black). (b) Flux chamber. The figure represents a lander lowered at the sediment surface and attached to a buoy; the flux chamber (labeled 1) is inserted in the sediment with two empty and two filled sampling syringes (labeled 2) and the magnetic stirrer. (c) Water enclosure. The figure represents the cylinder (labeled 1) pushed into the sediment and the plastic bag (labeled 2), which is open toward the sediment and toward the water surface. (d) Flow through system. The figure represents a water reservoir (labeled 1), with the incubation chamber (labeled 2) flowed through by a water flux and containing sediment, water, and a magnetic stirrer.

Benthic flux chambers. Application of the ¹⁵N IPT to benthic chambers (Fig. 2b) is based on the same principles as the batch-mode assay. However, addition of ¹⁵NO₃ and sampling occur in situ at the sediment surface and not in a water-filled container. Nielsen and Glud (24) used this technique to study coastal sediments, and Mengis et al. (20) adapted it to study deep lake sediments. The flux chambers contained one to two chambers, covering a sediment area between 400 and 900 cm² and enclosing 3 to 12 liters of water. ¹⁵NO₃ was added from a syringe to the sediment overlaying water reaching a concentration of 35 and 50 µM (24) and 119 µM (20). During the experiment the water volume was continuously mixed with a magnetic stirrer, keeping the diffusive boundary between 0.4 and 1 mm. Oxygen was measured with sensors, and water samples were taken at regular time intervals. Nielsen and Glud (24) incubated samples for only 3 h, ensuring that oxygen was not depleted by more than 25%. Risgaard-Petersen et al. (33) and Rysgaard et al. (37) showed that the concentrations of the N species did not change linearly with time if the O₂ concentration decreased by more than 20%. Mengis et al. (20) incubated samples for 60 h and, although oxygen was depleted by 75%, NO₃, NH₄⁺, and N₂ still changed linearly with time. After incubation the flux chambers were retrieved on deck. Water samples for determination of ²⁹N₂ and ³⁰N₂ were filled in exetainers (12.4 ml) containing 250 µl of ZnCl₂ (50% [wt/wt]) without entrapping gas bubbles and closed gastight (24). To extract ²⁸N₂, ²⁹N₂, and ³⁰N₂ entrapped in the sediment Nielsen and Glud (24) took sediment cores from the chamber, mixed the sediment and the overlaying water gently, and sampled the mixture as described above. Mengis et al. (20) did not sample the slurry and calculated the denitrification from the increase of labeled dinitrogen in the sampled water volumes. Mengis et al. (20) also included the isotopic composition of NO₃ and NH₄⁺ in their analysis in order to obtain a mass balance of ¹⁵N.

Water enclosures. Similar to the flux chambers, enclosures (Fig. 2c) are setups permitting in situ measurements. However, enclosures integrate over larger spatial scales. Mixing of the water is promoted by wave forces and not by a magnetic stirrer. The enclosure of Risgaard-Petersen et al. (34) consisted of a plastic bag that was open on two sides, fixed to a metal cylinder (inner diameter, 150 cm; height, 120 cm), and pushed into the sediment of a shallow lake.

Flow through systems. In contrast to the applications of the ¹⁵N IPT discussed thus far, flowthrough systems (Fig. 2d) are operated at steady-state conditions (32, 36). Sediment cores were incubated in gastight glass chambers connected to a thermostated continuous flowthrough system. Artificial freshwater or seawater was supplied via the inflow. The flow rate ranged from 17 to 300 ml h¹. Nitrate concentrations of the inflows ranged from 50 to 200 µM, with about 13% to 99% as ¹⁵NO₃. Risgaard-Petersen et al. (32, 33) and Rysgaard et al. (36, 37) adjusted the oxygen and the N₂ concentration in the inflowing water with a gas-mixing system. The water in the chamber was gently mixed by a magnetic stirrer not disturbing the sediment. Preincubation for reaching steady-state conditions lasted between 7 and 14 days, and the incubation itself lasted between 12 and 32 days. During the incubation inlet and effluent were sampled periodically. In the study of Rysgaard et al. (36), the outflowing water passed through a 50-ml flask that was heated to 75°C to strip the gases. Gas samples for analyses of isotopic composition of N₂ were taken with a syringe through a butyl stopper. Oxygen concentrations were measured with a microsensor (36, 37).

	ANALYSES AND CALCULATIONS

Continuous-flow isotope ratio mass spectrometers are typically used to perform analyses for the IPT. The sample preparation of such a modern setup is much simpler than for conventional dual-inlet mass spectrometers. To analyze the isotopic composition of N₂ in water samples, an He headspace is introduced into the exetainers and, after vigorous shaking for a few minutes, more than 98% of the N₂ is found in the gas phase. The gas phase is then injected into a gas chromatograph in line with a triple-collector mass spectrometer to obtain the isotopic composition of N₂.

The concentrations of ²⁹N₂ and ³⁰N₂ in the water or slurry is calculated in the following way:

(8)

where brackets indicate the concentrations and [²⁹N₂] · [²⁸N₂]¹ is the ratio obtained with mass spectrometry. [²⁸N₂] is calculated with Henry's law:

(9)

where K_H(T) is the temperature-dependent Henry constant and p_N2 is the partial pressure of dinitrogen in the atmosphere (0.78 atm).

Calculation of the production rates of ²⁹N₂ (r₂₉) and ³⁰N₂ (r₃₀) depends on the experimental setup. For the time-series incubations of the batch-mode assay, production rates r₂₉ and r₃₀ are obtained from a relation like the following (here shown for r₂₉):

(10)

where m₂₉ is the slope of the linear regression line of [²⁹N₂] plotted against time, A is the surface of the incubated sediment, V_w is the incubated water volume, V_s is the volume of the sediment (both in liters), and  is the sediment porosity.

For the endpoint incubation in the batch-mode assay, r₂₉ and r₃₀ are calculated as follows (here shown for r₂₉):

(11)

where [²⁹N₂]_f and [²⁹N₂]_i represent the final and initial concentrations of ²⁹N₂ and t is the time interval. Initial concentrations can be determined by analyzing the water phase before capping the cores or by scarifying a reference core at the beginning of the experiment and measuring the concentrations in the water-sediment slurry.

The production rates of labeled dinitrogen in the flux chambers should be determined using the formula of the endpoint experiment (equation 11), where [²⁹N₂]_f is the concentration of ²⁹N₂ in the water-sediment slurry of the core subsampled at the end of the experiment.

For calculating r₂₉ and r₃₀ in the enclosure, equation 10 can be used. In this case, when [²⁹N₂] and [³⁰N₂] are plotted against time to obtain m₂₉ and m₃₀, not only the amounts of ²⁹N₂ and ³⁰N₂ measured in the slurry should be considered but also the amount lost to the atmosphere. The method used to estimate the gas loss is described elsewhere (34).

The production rates r₂₉ and r₃₀ in the flowthrough system are calculated as follows (here shown for r₂₉):

(12)

where [²⁹N₂]_out and [²⁹N₂]_in are the concentrations of ²⁹N₂ and ³⁰N₂ in the outlet and in the inlet, respectively. F is the water flow through the chamber.

The calculated production rates of ²⁹N₂ and ³⁰N₂ (r₂₉ and r₃₀) can then be inserted in equations 1 to 6 to calculate the denitrification of NO₃ from the water column and coupled nitrification-denitrification.

	DISCUSSION

Top Introduction Discussion References

Assumptions of the ¹⁵N IPT. The ¹⁵N IPT is based on four assumptions (23, 35), which are briefly discussed below.

(i) Added ¹⁵NO₃ does not interfere with denitrification of in situ NO₃. This requirement can be tested by adding different amounts of ¹⁵NO₃ to the sediment overlaying water (23). The rates obtained from such experiments (D_w and D_n, Fig. 1) should remain independent of the concentration of the added tracer. In several studies (15, 19, 24, 29, 35, 45), ¹⁵NO₃ was added from a minimum of 10 µM to a maximum of 400 µM without modifying the in situ denitrification rate. However, this makes it not superfluous to test the interference of the addition of ¹⁵NO₃ before every experiment, because the optimal tracer range depends not only on the nitrate concentration but also on the heterogeneity, stability, and activity of the sediment and the incubation process.

(ii) Denitrification of nitrate from the water column (D_w^tot) should increase linearly with the nitrate concentration. The calculation of D_w (denitrification of in situ ¹⁴NO₃) from D_w^tot (denitrification of ¹⁴NO₃ plus added ¹⁵NO₃) depends on this assumption. This assumption can also be tested by adding different amounts of the ¹⁵N tracer. At elevated nitrate concentrations, denitrification may no longer depend linearly on the concentration of nitrate. At high nitrate concentrations saturation effects may occur due to limited supply of electron donors or Michaelis-Menten-type saturation. In addition, the nitrate penetration depth will increase and cause a nonlinear response of denitrification rates to the concentration in the overlying water.

(iii) Labeling of in situ NO₃ with ¹⁵NO₃ in the water column and in the sediment must be homogeneous. Heterogeneity of the sediment and bioturbation can affect the ratio between nitrification and NO₃ flux at different spots in the sediment. As a result, dinitrogen is produced preferentially as isotope pairs ²⁸N₂ and ³⁰N₂. Higher r₃₀ and r₂₈ rates compared to r₂₉ cause an underestimation of D₁₄ and D₁₅ (see equations 1 and 2). However, Nielsen (23) reports that at higher concentrations of ¹⁵NO₃ more of the denitrified ¹⁴NO₃ will be measured as ²⁹N₂. As a result, the possible miscalculation of ²⁸N₂ production will be of less significance compared to situations with lower concentrations of ¹⁵NO₃. Nielsen suggests that the ¹⁵NO₃ concentrations are high enough when D₁₄ becomes independent of the ¹⁵NO₃ concentrations. However, van Luijn et al. (52) argue that in the case of anaerobic microsites that may exist in the aerobic layer, nitrifying and denitrifying bacteria may be ideally positioned, allowing a tightly coupled nitrification-denitrification. Accordingly, around these microsites the condition of uniform labeling is still not fulfilled, even when the ¹⁵NO₃ concentration increases, so that coupled nitrification-denitrification is still underestimated. Middelburg et al. (22) demonstrated in a modeling study that the condition of homogeneous labeling is fulfilled only when nitrification and denitrification occur in distinct separate zones. Small-scale heterogeneity supporting denitrification within the aerobic layer interfere with this condition and may underestimate the denitrification rate. A final resolution over the controversy of the role of microsites is likely to come from studies combining microsensor or tomographic work to characterize sediment heterogeneity. Nielsen et al. (23) and Welsh et al. (53) also suggested an underestimation of coupled nitrification-denitrification when the sediments are colonized by macrophytes. The reason is that coupled nitrification-denitrification can occur close to the roots deep in the sediment remote from the diffusion zone of the ¹⁵N tracer.

(iv) A stable NO₃ concentration gradient across the sediment water interface must be established shortly after ¹⁵NO₃ addition. This is important, because otherwise ¹⁵NO₃ will not be immediately available for denitrification, leading to an underestimation of the initial denitrification rate. A too-short incubation time results in an increasing production rate of ²⁹N₂ and ³⁰N₂. If ¹⁵NO₃ is initially well mixed in the water column, the time to establish a stable gradient depends on the oxygen penetration depth. It determines the diffusion length of nitrate molecules to reach denitrifying bacteria. The time to reach a steady-state gradient depends on microbial activity: it will be longer in oligotrophic than in eutrophic systems and longer in winter than in summer. According to Nielsen (23), the equilibration time was 8 min at an oxygen penetration depth of 1 mm. Dalsgaard (8) developed a model that predicts the equilibration time as a function of the oxygen penetration depth. Due to their long preincubation times, the flowthrough systems exhibit the most stable nitrate gradients.

Comparison of different applications and methods. Lohse et al. (19) and Svensson (45) compared the acetylene inhibition technique with the ¹⁵N IPT in the batch-mode assay. The ¹⁵N IPT yielded denitrification rates two times (19) and three to eight times (45) higher than the acetylene inhibition technique because acetylene inhibits also nitrification-denitrification, which was the main denitrification mechanism. Similar observations were reported by Seitzinger et al. (39). Van Luijn et al. (52) observed that denitrification rates measured with the ¹⁵N IPT were also smaller than those estimated with the N₂ flux method. In contrast, Risgaard-Petersen et al. (32) observed good agreement of the two methods in the same flowthrough system. These authors concluded that the bad agreement of the two methods described by van Luijn et al. (52) was due to the longer preincubation time necessary for the N₂ flux method, which caused an accumulation of nitrate and ammonia, leading to an overestimation of denitrification. However, results from the N₂ flux experiment of van Luijn et al. (52) agreed well with the nitrogen mass-balance of the studied lake, while the results obtained from the ¹⁵N IPT in the batch-mode assay resulted in lower denitrification rates. These authors proposed an underestimation by the IPT due to the presence of coupled nitrification-denitrification in microsites within the aerobic layer. In contrast, Nielsen et al. (25) found a good agreement between the nitrogen mass-balance of an estuary and results from the ¹⁵N IPT in the batch-mode assay. Similar results were reported in two deep lakes (20). In the study of Steingruber (42), denitrification rates from the batch-mode assay were higher than the results from a mass-balance calculation of a shallow lake. It was proposed that the sampled bottom water for the incubation experiment may not have been representative for the whole lake (differences in nitrate concentrations). Comparison of the ¹⁵N IPT in the batch-mode assay and in an enclosure experiment in a shallow lake yielded denitrification rates 6 to 26 times higher in the enclosure (34). In the same lake the annual average denitrification estimated from whole lake mass-balance calculations was about three times higher than the rates estimated with the ¹⁵N IPT in the batch-mode assay. These authors concluded that the batch-mode assay underestimated in situ denitrification rates in shallow lakes due to lower turbulent mixing and thus transport of nitrate into the sediment. Mengis et al. (20) and Nielsen and Glud (24) compared the application of the ¹⁵N IPT to the use of a benthic chamber with the batch-mode assay. The two methods yielded similar results which also corresponded well to the results from the mass-balance calculations.

Advantages and limitations of different applications of the IPT. Three major limitations of the IPT deserve further attention independent of the mode of application. (i) So far no technique provides satisfactory simulation of the turbulent mixing conditions in shallow lakes. In the future, ¹⁵NO₃ tracer could be added to whole ponds in order to incubate the tracer under the most realistic conditions. (ii) It is still not clear whether an underestimation of denitrification in the case of coupled nitrification-denitrification in microniches of oxic zones can be excluded, when D₁₄ is proved to be independent of the amount of added tracer. More research in order to clarify this issue is needed. (iii) The anaerobic oxidation of ammonium into dinitrogen with nitrate serving as an electron acceptor (anammox reaction), which has been recently reported to occur in sea sediments (T. Dalsgaard, Abstr. ASLO Aquat. Sci. Meet. 2001), can also limit the application of the IPT. Ogilvie et al. (27) suggested that in the presence of the anammox reaction, coupled nitrification-denitrification may be overestimated, while the total denitrification result would be correct. Further research is needed regarding this issue.