Production of N2 through Anaerobic Ammonium Oxidation Coupled to Nitrate Reduction in Marine Sediments

In the global nitrogen cycle, bacterial denitrification is recognizedas the only quantitatively important process that converts fixednitrogen to atmospheric nitrogen gas, N₂, thereby influencingmany aspects of ecosystem function and global biogeochemistry.However, we have found that a process novel to the marine nitrogencycle, anaerobic oxidation of ammonium coupled to nitrate reduction,contributes substantially to N₂ production in marine sediments.Incubations with ¹⁵N-labeled nitrate or ammonium demonstratedthat during this process, N₂ is formed through one-to-one pairingof nitrogen from nitrate and ammonium, which clearly separatesthe process from denitrification. Nitrite, which accumulatedtransiently, was likely the oxidant for ammonium, and the processis thus similar to the anammox process known from wastewaterbioreactors. Anaerobic ammonium oxidation accounted for 24 and67% of the total N₂ production at two typical continental shelfsites, whereas it was detectable but insignificant relativeto denitrification in a eutrophic coastal bay. However, ratesof anaerobic ammonium oxidation were higher in the coastal sedimentthan at the deepest site and the variability in the relativecontribution to N₂ production between sites was related to largedifferences in rates of denitrification. Thus, the relativeimportance of anaerobic ammonium oxidation and denitrificationin N₂ production appears to be regulated by the availabilityof their reduced substrates. By shunting nitrogen directly fromammonium to N₂, anaerobic ammonium oxidation promotes the removalof fixed nitrogen in the oceans. The process can explain ammoniumdeficiencies in anoxic waters and sediments, and it may contributesignificantly to oceanic nitrogen budgets.

INTRODUCTION

Top

Introduction

The global nitrogen cycle is characterized by the maintenanceof a small pool of fixed, or combined, nitrogen in continuousexchange with the huge reservoir of atmospheric dinitrogen,N₂. Microbial nitrogen fixation and industrial nitrogen fixationare the most important processes that convert N₂ to fixed nitrogen,while bacterial denitrification, i.e., anaerobic reduction ofnitrate to dinitrogen coupled to the oxidation of organic matteror reduced iron or sulfur species (40, 48), is recognized asthe only significant process that regenerates N₂ (18, 33). Itis the balance between these source and sink terms that controlsthe global inventory of fixed nitrogen. Because of the frequentrole of nitrogen as a limiting nutrient for primary production(47), processes that affect the availability of fixed nitrogenare important regulators of ecosystem function and global biogeochemistry(8, 33). For example, denitrification counteracts the effectsof eutrophication in coastal waters (37), and on a global scale,glacial to interglacial variations in denitrification rateshave been suggested to explain changes in nitrogen availabilityin the oceans, thereby forcing global rates of net primary productionand concentrations of atmospheric CO₂ (7, 39).

In current marine nitrogen budgets, bacterial denitrificationis the prevailing sink for nitrogen and most denitrificationtakes place in the seafloor (7, 13, 27). Here, nitrogen fordenitrification is delivered mainly through a combination ofammonification, the degradative release of ammonium from organicmaterial, and nitrification, the oxidation of ammonium via nitriteto nitrate (3, 37). The overall regulation of this ammonification-nitrification-denitrificationpathway is relatively complex due to the involvement of bothaerobic and anaerobic microbial processes (37), and consequently,the relative importance of N₂ as a product depends stronglyon the environmental conditions (27).

There are indications, however, that ammonium may be oxidizedanaerobically in marine sediments with either nitrate or manganeseoxides as the oxidant (1, 2, 12, 17, 25, 36). By shortcuttingthe pathway to N₂, such reactions could support a more efficientconversion of fixed nitrogen than can be achieved through thewell-established sequence of reactions. Ammonium oxidation coupledto manganese reduction has been suggested to occur in continentalshelf sediments either through a direct reaction to N₂ (25)or through oxidation to nitrate followed by denitrification(1, 17). Neither of these processes has been demonstrated directlyin natural sediments, however, and they could not be detectedin anoxic, manganese oxide-rich sediment with a sensitive ¹⁵N-labelingtechnique (44).

Ammonium oxidation coupled to nitrate reduction has been indicatedby deficiencies of ammonium relative to other products of organicmatter mineralization in anoxic, nitrate-containing pore watersof deep-sea sediments, and was hypothesized to have the followingstoichiometry (2, 12, 36):

(1)
A possiblemechanism explaining these observations has been provided bythe discovery in a wastewater bioreactor of bacterially catalyzedammonium oxidation with nitrite in the anammox process (28,46):

(2)
Concomitant organotrophic nitratereduction is the source of the nitrite, and N₂ forms throughthe pairing of one nitrogen atom from ammonium with one fromnitrite (46). The anammox reaction supports autotrophic growthin so far uncultured bacteria from the order Planctomycetales(34, 41; reviewed in references 20, 21, and 24). While thisprocess and organisms performing it have been described froma second wastewater treatment facility (34), there are, to ourknowledge, no reports of their occurrence or activity in naturalenvironments.

Here we provide the first quantification of anaerobic ammoniumoxidation in natural environments. We used nitrogen-15-labeledcompounds to investigate the role of nitrate in ammonium oxidationin sediments from the Baltic-North Sea transition. These siteswere chosen because their biogeochemistry has been describedin detail and because they are representative of a wide rangeof continental shelf sediments.

MATERIALS AND METHODS

Top

Materials and Methods

Sediment cores were collected from three sites with minimaldisturbance of the surface by using a multiple corer (Skagerrak)or by divers (Aarhus Bay) and kept at bottom water temperature(Table 1). Sediment biogeochemistry and microbial processeshave been studied in detail at all three sites (6, 22, 23, 43-45).

View this table:
[in this window]
[in a new window]
TABLE 1. Characteristics of freshly collected sediments

Incubations.
The depth of oxygen penetration was determined with oxygen microsensors(30), and the nitrate distribution in the pore water was analyzedafter centrifugation of sediment sectioned into 0.3- to 0.5-cmdepth intervals. Based on these determinations, sediment tobe used in incubations was subsampled from the zone of nitrateconsumption, i.e., from around the oxic-anoxic interface tothe depth of nitrate depletion (Table 1), with all handlingtaking place in an anoxic glove bag in a room held at bottomwater temperature. Incubations, pore water extraction, and analysesfollowed the procedures described in reference 44. In brief,sediment was incubated unamended or after addition of combinationsof Na¹⁵NO₃, ¹⁵NH₄Cl (both >99.5% ¹⁵N), and their unlabeledanalogues at 50 to 200 nmol/cm³ from $~$ 100 mM stock solutions.The sediment was incubated with no headspace in portions of,typically, 200 cm³ in gas-tight plastic bags stored in an N₂atmosphere (16) and sampled in the anoxic glove bag. For porewater extraction, sediment from the bags was loaded into 15-mlpolypropylene centrifuge tubes that were filled completely andcentrifuged. After centrifugation, 2 ml of the supernatant forN₂ analysis was transferred through a butyl rubber septum toa 6.6-ml glass vial (Exetainer; Labco, High Wycombe, UnitedKingdom) that contained 50 µl of 50% ZnCl₂ as an inhibitorof microbial activity and had previously been flushed with Heor Ar. After N₂ analysis, these samples were also used for determinationof the isotopic composition of NO₃^-. The remaining supernatantwas filtered through a 0.45-µm-pore-size cellulose acetatefilter and stored frozen for analysis of NH₄⁺, NO₂^-, NO₃^-, andthe isotopic composition of NO₂^- plus NO₃^-. Additional incubations(Skagerrak S9 September 2000 and Aarhus Bay) were made by distributionof 9-ml portions of sediment into 12.6-ml Exetainers in whichthe headspace was flushed with He. These vials were sampledafter vigorous shaking followed by centrifugation. For N₂ analysis,a 2-ml sample was withdrawn from the headspace and transferredto a 6.6-ml Exetainer filled with He-degassed water while excesswater was drained off through a hypodermic needle. The porewater supernatant was filtered for other analyses as describedabove. The two incubation procedures produced similar results.

Analysis.
Concentrations of ²⁹N₂ and ³⁰N₂ were determined by isotope ratiomass spectrometry (Robo-Prep-G+ in line with TracerMass; EuropaScientific, Crewe, United Kingdom) and calculated as excessabove their natural abundances (44). Concentrations were correctedfor an analytical recovery of 50 to 80%. The N isotope compositionof NH₄⁺ was analyzed after conversion to N₂ with hypobromite(31), and the isotopic composition of NO₃^- plus NO₂^- was determinedafter conversion to NO with acidic V³⁺ (5), followed by reductionto N₂ on hot Cu in line with the mass spectrometer. Nitriteconcentrations were determined spectrophotometrically (14),and NO₃^- plus NO₂^- was determined through chemiluminescenceafter reduction to NO (5). Ammonium was determined by flow injectionanalysis with conductivity detection (15) or spectrophotometricallyby the salicylate-hypochlorite method (4). To determine changesin the total inventory of NH₄⁺ in the sediment, soluble NH₄⁺concentrations were multiplied by 1 + K_NH₄, where K_NH₄ is theadsorption coefficient for NH₄⁺, determined from the initialrecovery of added ¹⁵NH₄⁺ in the pore water (44).

Calculations.
Two approaches were used to quantify the pathways of N₂ production.In incubations with added ¹⁵NO₃^- and unlabeled NH₄⁺, calculationswere based on the accumulation of ²⁹N₂ and ³⁰N₂. Consistentwith the results of the ¹⁵NH₄⁺ experiments (see below), we assumedthat N₂ produced through anaerobic ammonium oxidation consistedof one nitrogen atom from NO₃^- and one from NH₄⁺. Thus, in incubationswith ¹⁵NO₃^-, the process produced ²⁸N₂ and ²⁹N₂ only, and thisat a ratio corresponding to the labeling of the NO₃^- pool, asfollows:

(3)
where A_m represents theproduction of N₂ of mass m through anaerobic ammonium oxidation,A_total = A₂₈ + A₂₉, and F_N is the fraction of ¹⁵N in NO₃^- (typically, $~$ 0.95). Denitrification was assumed to produce ²⁸N₂, ²⁹N₂, and³⁰N₂ through random isotope pairing (29) as follows:

(4)
where D_m denotes production of N₂ through denitrification.With P_m representing the total, measured production of N₂ ofmass m, it follows from equations 3 and 4 and D₃₀=P₃₀ that

(5)

(6)

(7)

(8)
and, by addition of equations 7 and 8,

(9)
In incubations with ¹⁵NH₄⁺, anaerobic ammoniumoxidation was quantified as follows from the consumption of¹⁵NH₄⁺: A_total = - ${Delta}$ [¹⁵NH₄⁺]_corr x F_A^-1, where ${Delta}$ [¹⁵NH₄⁺]_corr isthe change in the ¹⁵NH₄⁺ total sediment concentration correctedfor the consumption due to assimilation as determined in separateincubations and F_A is the fraction of ¹⁵N in NH₄⁺. Nitrate consumptioncoupled to ammonium oxidation was assumed to be equivalent toA_total, and N₂ production from denitrification was calculatedas follows from the excess consumption of NO₃^- plus NO₂^-: D_total= 0.5 x (- ${Delta}$ [NO_x^-] - A_total), where ${Delta}$ [NO_x^-] is the change in theconcentration of NO₃^- plus NO₂^-. In Aarhus Bay, ammonium oxidationwas too small for accurate quantification by this approach inincubations with ¹⁵NH₄⁺. Instead, the process was quantifiedfrom ¹⁵N-N₂ production and related to denitrification as determinedin parallel incubations with ¹⁵NO₃^-.

The relative contribution of anaerobic ammonium oxidation toN₂ production in each experiment was calculated based on threeor more samplings during the incubation. At Skagerrak S9, ratesof the process were determined from linear accumulations overseveral samplings while rates for Skagerrak S6 and Aarhus Bayare minimum estimates because all of the nitrate and nitritewas typically consumed at the second sampling in time courseexperiments, allowing only a two-point determination.

RESULTS AND DISCUSSION

Top

Results and Discussion

Detection and stoichiometry.
Nitrate-dependent ammonium oxidation was observed at all threesites. Thus, when ¹⁵N-ammonium and unlabeled nitrate were addedto the sediment, ¹⁵N-ammonium and nitrate were consumed and¹⁵N-labeled N₂ was produced without delay, while in controlincubations without added nitrate there was no production of¹⁵N-labeled N₂ after the rapid initial depletion of the nativenitrate pool (Fig. 1 and 2). Nitrate, although present in concentrationsof >10 µM in freshly collected sediment (Table 3),was consumed during the initial handling of the sediment forthe control incubations and was depleted before completion ofthe first sampling (Fig. 1 and data not shown). In these controls,small amounts of ¹⁵N-labeled N₂ that had already accumulatedat the first sampling after the addition of ¹⁵N-ammonium indicatedthat ammonium oxidation was also coupled with the consumptionof the small native nitrate pool (Fig. 1 and 2). A slight consumptionof ¹⁵N-ammonium in the absence of nitrate could be attributedto bacterial nitrogen assimilation (3). Nitrite accumulatedtransiently during nitrate consumption, and production of ¹⁵N-labeledN₂ and enhanced ¹⁵N-ammonium consumption continued until nitrateand nitrite were depleted (Fig. 1).

View larger version (18K):
[in this window]
[in a new window]
FIG. 1. Anaerobic ammonium oxidation coupled to nitrate reduction in anoxic marine sediment from Skagerrak station S9 amended with ¹⁵N-ammonium and unlabeled nitrate (filled symbols) compared to a control with ¹⁵N-ammonium only (open symbols). Panels: A, time course of concentrations of nitrate (circles) and nitrite (triangles); B, time course of concentrations of ¹⁵N-ammonium (circles) and ¹⁵N in N₂ (triangles). Examples are from May 2000, 1- to 3-cm depth, and are representative of three experiments.

View larger version (22K):
[in this window]
[in a new window]
FIG. 2. Anaerobic ammonium oxidation coupled to nitrate reduction in anoxic marine sediments amended with ¹⁵N-ammonium and unlabeled nitrate (filled symbols), compared to controls with ¹⁵N-ammonium only (open symbols), from Skagerrak station S6 (A) and Aarhus Bay (B). Time courses of the combined concentrations of nitrate and nitrite (squares, right axis), ¹⁵N-ammonium (circles, left axis, Skagerrak S6 only), and ¹⁵N in N₂ (triangles, left axis). Nitrate and nitrite concentrations are only shown for nitrate-amended incubations. In incubations without nitrate addition, endogenous nitrate and nitrite were depleted at the first sampling time. Examples are from 1- to 4-cm and 0.8- to 1.3-cm depths at Skagerrak S6 and Aarhus Bay, respectively. Results in panel A are representative of three experiments, and the error bars in panel B represent the standard deviations of three parallel incubations.

Incubations with ¹⁵N labeling of ammonium, nitrate, or bothrevealed that during nitrate-dependent ammonium oxidation, nitrogenoriginating from ammonium was paired with nitrogen originatingfrom nitrate. Thus, in incubations with ¹⁵N-ammonium and unlabelednitrate, more than 97% of the ¹⁵N transformed to N₂ was pairedwith ¹⁴N to form N₂ with a molecular weight of 29 (²⁹N₂) (Fig.3A). The ¹⁵N percentage in ammonium decreased during the experimentsdue to dilution with ¹⁴N from ammonification (3, 44), but ammoniumwas labeled with more than 50% ¹⁵N during most of the reaction,so with random pairing of ammonium-N into N₂, less than 50%of the transformed ¹⁵N would have been combined with ¹⁴N toform ²⁹N₂ (29, 46). In combination with the nitrate dependenceof ammonium oxidation, this indicated that nitrate was the sourceof ¹⁴N in ²⁹N₂.

View larger version (16K):
[in this window]
[in a new window]
FIG. 3. Accumulation of ¹⁵N in ²⁹N₂ (open circles) and ³⁰N₂ (filled circles) during parallel anoxic incubations of marine sediment amended with ¹⁵N-ammonium and unlabeled nitrate (same experiment as Fig. 1) (A), ¹⁵N-nitrate (B), and ¹⁵N-ammonium and ¹⁵N-nitrate (C). Percentages of labeling with ¹⁵N-ammonium and ¹⁵N-nitrate were 30 to 80% and 90%, respectively. During all experiments, unlabeled ammonium was generated through ammonification. Examples are from May 2000, Skagerrak S9, 1- to 3-cm depth, and are representative of two or more replicates.

Similarly, in incubations with ¹⁵N labeling of nitrate by morethan 90% and no labeling of ammonium, up to 70% of the ¹⁵N-nitrateconverted to N₂ was found in ²⁹N₂ (Fig. 3B). This distributiondeviated strongly from that expected during conventional denitrification,which under these conditions would produce less than 10% ²⁹N₂through random isotope pairing (equation 4; reference 29), andthe distribution indicated that most of the ¹⁴N in ²⁹N₂ originatedfrom ammonium generated through ammonification. A lesser productionof ³⁰N₂ in the ¹⁵N-nitrate experiments revealed that denitrificationproceeded simultaneously with the anaerobic ammonium oxidation.Further confirmation of the pairing of nitrogen from nitrateand ammonium was obtained through ¹⁵N labeling of both nitrateand ammonium (Fig. 3C), which resulted in the highest yieldof ³⁰N₂. In this case, the lesser production of ²⁹N₂ resultedmainly from the reaction of endogenous ¹⁴N-ammonium with ¹⁵N-nitrate.

The transformations of nitrate and ammonium to N₂ are accompaniedby oxidation state changes of -5 and +3, respectively (equation1). Thus, the observed one-to-one combination of nitrogen fromthese two sources in N₂ requires more reducing power than isdelivered by ammonium and is not in agreement with the stoichiometrypreviously proposed for anaerobic ammonium oxidation in marinesediments (equation 1). Instead, the observed patterns of isotopepairing were similar to those reported for the anammox process(equation 2), including the slight production of ³⁰N₂ from ¹⁵N-ammonium(Fig. 3A) (46). Transient nitrite accumulation was observedin the incubations at both Skagerrak sites (Fig. 1), and itis likely that nitrite was the oxidant for ammonium while thereduction of nitrate to nitrite was coupled to the oxidationof organic matter in the sediments, in analogy to the transformationsoccurring in the wastewater reactor that supports anammox (41,46). We are aware of no other processes that may occur in marinesediments and can explain our results well (see also reference46). At Aarhus Bay, all nitrate and nitrite was consumed atthe first sampling only 20 min after nitrate had been added(Fig. 2) and nitrite production could therefore not be demonstrated.Hence, we cannot exclude the possibility that anaerobic ammoniumoxidation at this site was limited by low nitrite concentrations.

Further investigations are needed to determine whether organismsrelated to the known anaerobic ammonium oxidizers carry outanaerobic ammonium oxidation in the marine environment. Criticalconditions for the anammox process in the wastewater reactorinclude a temperature of at least 20°C (42) and a cell densityof the active bacterium of 10¹⁰ to 10¹¹/cm³ (41). In the sedimentsexplored here, anaerobic ammonium oxidation proceeded at 6 to7°C and total bacterial numbers were about 10⁹/cm³, whichis typical of fine-grained marine sediments (32, 35). Thesedifferences suggest that different organisms are active in thetwo systems. Bacteria related to the anammox planctomycetes,but of unknown physiology, have been detected in natural environments,including marine sediments (24, 34), and are potential candidatesfor anaerobic ammonium oxidation in the seafloor.

Quantitative significance.
Rates of anaerobic ammonium oxidation and denitrification werecalculated either based on the different patterns of isotopepairing observed during the two processes or from the measuredconsumption of ¹⁵N-ammonium (see Materials and Methods). Therelative importance of the two processes showed no dependenceon the concentration of either nitrate or nitrite as these changedduring the incubations (Fig. 4). Hence, we expect that the elevatednitrate and nitrite concentrations during our experiments didnot enhance the importance of ammonium oxidation with nitrate.Likewise, similar contributions of the two processes were observedin experiments in which ammonium was added and those in whichammonium was only available from ammonification in the sediment(Fig. 5).

View larger version (14K):
[in this window]
[in a new window]
FIG. 4. Relative contribution of anaerobic ammonium oxidation to total N₂ production (anaerobic ammonium oxidation plus denitrification) as a function of the concentration of nitrate (filled symbols) or nitrite (open symbols) during two parallel experiments from Skagerrak S9 with the initial addition of ¹⁵N-nitrate to $~$ 50 nmol cm^-3 (circles) and $~$ 100 nmol cm^-3 (triangles). The different nitrate concentrations correspond to the consumption of the initial amendment, while nitrite concentrations reflect accumulation followed by depletion (Fig. 1). The contribution of anaerobic ammonium oxidation was calculated for each period between two consecutive samplings and is plotted as a function of the average concentration of nitrate or nitrite during the same period. Error bars indicate the concentration change between the consecutive samplings.

View larger version (23K):
[in this window]
[in a new window]
FIG. 5. Summary of experiments from three locations. Columns and upper axis: relative contribution of anaerobic ammonium oxidation to N₂ production calculated from isotope pairing (black) or ¹⁵NH₄⁺ consumption (gray). Circles and lower axis: rate of N₂ production through coupled ammonium oxidation-nitrate reduction. Conditions for each experiment summarized next to each column include the sediment depth interval (centimeters), the combination of nitrate and ammonium added (n and N, nitrate; a and A, ammonium; uppercase, ¹⁵N labeled; lowercase, unlabeled), and the time of sampling at Skagerrak S9. Sediment from Skagerrak S6 and Aarhus Bay was sampled in autumn 2000.

Anaerobic ammonium oxidation was an important process in theoffshore sediments, accounting for, on average, 24% of the totalN₂ production at Skagerrak S6 and 67% of the total N₂ productionat Skagerrak S9 (Fig. 5). In the coastal bay, the contributionwas only 2% and denitrification completely dominated N₂ production.The results were reproducible at each site, including threevisits to Skagerrak S9 over a 1.5-year interval. The strongdecrease in the relative significance of anaerobic ammoniumoxidation from Skagerrak S9 over S6 to Aarhus Bay occurred inspite of a concurrent increase in the rate of the process (Fig.5); i.e., the decrease resulted from large differences in thevolume-specific rate of denitrification. The variation in denitrificationrates can be explained by differences in the availability ofsubstrates for the denitrifying bacteria. Thus, rates of organiccarbon mineralization in the surface sediment, reflecting organicsubstrate availability, vary similarly among the three sites(6, 23, 43, 44). The variation in the demand for electron acceptorsbetween the sediments was also reflected in the oxygen penetrationdepth (Table 1). In addition to organic matter, ferrous ironand, possibly, iron sulfides may serve as substrates for denitrification(40). Consistent with this, our denitrification rates variedin parallel with concentrations of ferrous iron and solid reducedsulfur species in the surface sediment of the three sites (6,44). Since the reduced inorganic substrates are themselves productsof organic carbon oxidation (6), the extent to which they accumulatein the sediment will depend on the intensity of this process.Thus, although the relative significance of the potential substratesfor denitrification is not known, the reactivity of organicmatter can be used as a general key to substrate availability.

At Skagerrak S9, denitrifiers may have been further limitedby competition for substrates from manganese-reducing bacteria,the activity of which is stimulated by a high manganese oxidecontent of the sediment (6, 44). Ammonium, however, is not asubstrate for manganese reduction at this site (44). Thus, anaerobicammonium oxidation may be particularly favored over denitrificationas a sink for nitrate in manganese oxide-rich sediments. However,high concentrations of manganese oxides are not required forsubstantial contributions from the process, as demonstratedat S6, where manganese oxide concentrations are low (6).

The source of ammonium in the sediments is the degradation oforganic matter (3). At all three sites, the average rates ofanaerobic ammonium oxidation (S9, 30 µM day^-1; S6, 99µM day^-1; Aarhus Bay, 83 µM day^-1; Fig. 1) werehigher than the rates of ammonium accumulation in the pore watermeasured in anoxic, nitrate-free incubations of sediment fromthe same depth intervals as studied here: 5 to 21 µM day^-1at Skagerrak S9 (44), 23 to 57 µM day^-1 at Skagerrak S6(6), and 15 to 32 µM day^-1 at Aarhus Bay (B. Thamdrup,unpublished results). Thus, when nitrate or nitrite is available,anaerobic ammonium oxidation has the potential to consume alarge fraction of the ammonium that is produced in the sediments.

Biogeochemical considerations.
In terms of the rates of benthic respiration and the associatedredox zonation, sites S6 and S9 together delineate a range ofconditions representative of wide areas of the seafloor on thecontinental shelves and upper slopes (e.g., see references 26and 43). Based on a good correlation between rates of benthicrespiration and N₂ production in the North Atlantic (38), wealso expect representative N₂ fluxes from these sediments. Usingthe availability of reactive organic matter as a key to therelative importance of denitrification and coupled ammoniumoxidation-nitrate reduction, as discussed above, we consequentlyexpect the new process to contribute substantially to N₂ productionin such areas. The process may indeed be even more widespread,since it can explain the ammonium deficiencies observed in theanoxic, nitrate-containing zone of hemipelagic and pelagic sediments(2, 12, 36), and it is likely to contribute to similar deficienciesin oxygen-deficient waters such as those of the Black Sea (9,10).

The anaerobic oxidation of ammonium to N₂, as described here,adds a shunt to the conventional sequence of nitrogen transformationsin sediments, which facilitates the removal of fixed nitrogenas N₂. In current models of benthic nitrogen transformations,the ammonification-nitrification-denitrification pathway fromorganic nitrogen to N₂ involves the production of both ammoniumand nitrate, either of which may escape the sediment (3, 27).With coupled ammonium oxidation-nitrate reduction in function,ammonium may be consumed in the anoxic zone of the sediment,thus reducing its chance of escape. Furthermore, the processproduces twice as much N₂ as does denitrification per moleculeof nitrate or nitrite consumed, which augments N₂ productionin sediments in which nitrification rates limit the supply ofnitrate and nitrite. Thus, the process may help explain thevery efficient conversion of ammonium to N₂, which is typicallyinferred for shelf sediments, and which has hitherto been attributedsolely to tight coupling of nitrification and denitrification(11, 19, 37).

Denitrification in shelf sediments constitutes one of the largestsinks for fixed nitrogen in current marine budgets (7, 27).Our results suggest that anaerobic ammonium oxidation couldbe of comparable importance. Budgets of fixed nitrogen in theoceans suggest a current imbalance due to an excess of denitrificationover source terms (8). Because of the role of nitrogen as alimiting nutrient, the disparity has potential implicationsfor primary production and thereby for oceanic CO₂ uptake. Ourfinding of an additional sink for fixed nitrogen appears tofurther exacerbate this imbalance. Global denitrification ratesare, however, very loosely constrained (7, 8, 13, 27). Hence,in future revisions of global marine N₂ production, coupledammonium oxidation-nitrate reduction should be included specificallyand the regulatory characteristics of the process should beconsidered in studies of the dynamics of the nitrogen cycle.

ACKNOWLEDGMENTS

We thank the masters, crews, and scientific parties of the cruisesonboard R/V Arne Tiselius, R/V Gunnar Thorson, and R/V Danafor assistance during sampling. We thank Egon B. Frandsen, AnnaHaxen, Tanja Quottrup, Lilian Salling, and Peter Søholtfor technical assistance; Søren Rysgaard and Lars PeterNielsen for fruitful discussions; and Don Canfield and two anonymousreviewers for comments on the manuscript.

Cruises were sponsored by the EU Large Scale Facility at KristinebergMarine Research Station and by the Danish National ResearchCouncil. Bo Thamdrup was supported by the Danish National ResearchFoundation through the Danish Center for Earth System Science.

FOOTNOTES

* Corresponding author. Mailing address: Danish Center for Earth System Science, Institute of Biology, University of Southern Denmark, Campusvej 55, DK-5230 Odense M, Denmark. Phone: 45 6550 2477. Fax: 45 6593 0457. E-mail: bot{at}biology.sdu.dk.

REFERENCES

Top