Contrasting biogeochemistry of nitrogen in the Atlantic and Pacific oxygen minimum zones

Introduction Conclusions References


Introduction
Nitrogen is a key limiting element for biological productivity and occupies a central role in ocean biogeochemistry (Gruber, 2008).Although most chemical forms of nitrogen in the ocean are bio-available (i.e.fixed nitrogen or "fixed-N") the most abundant form, N 2 is generally not.The sources of fixed-N include river inputs, atmospheric deposition and N 2 fixation (Duce et al., 2008;Gruber, 2008).Sinks of fixed-N, producing N 2 , include microbial denitrification and anammox processes, both requiring very low (i.e.suboxic) [O 2 ] (Devol, 2008).Hence suboxic Oxygen Minimum Zones (OMZs) are the oceanic regions especially associated with denitrification (Cline and Richards, 1972;Codispoti et al., 2001;Ward et al., 2009) and anammox (Lam et al., 2009;Thamdrup et al., 2006;Hamersley et al., 2007;Galan et al., 2009) and play a particularly important role in the global nitrogen cycle as sites of N sinks from the ocean.They are located typically in areas of upwelling with high productivity which exhibit complex cycling of nutrients (Helly and Levin, 2004).
Absence of ammonium in suboxic OMZs that should have accumulated from organic matter breakdown could be indicative of anammox, while the consumption of N 2 O requires the denitrification process (Naqvi et al., 2010).
Although OMZs are found in all major ocean basins, their associated nitrogen cycle processes can vary dramatically due to contrasting minimum oxygen levels.In the Eastern Tropical South Pacific (ETSP), suboxic Published by Copernicus Publications on behalf of the European Geosciences Union.
Cruises carried out during the collaborative research project SFB-754 (www.sfb754.de) of the German Research Foundation and the BMBF supported project SOPRAN (Surface Ocean Processes in the Anthropocene: www.sopran.pangaea.de)provided unique opportunities to sample nitrogen species in these two contrasting regions.Here we present a comparison of N 2 O concentration and stable nitrogen isotope distributions, which were measured in both OMZs to highlight the similarities and differences in nitrogen cycling between the two regions.

Sampling and analytical methods
In the Pacific OMZ, samples were collected onboard the R/V Meteor (M77 Legs 3 and 4) in December 2008 and January 2009 (Fig. 1).In the Atlantic OMZ, samples were collected on the R/V L'Atalante (cruise leg ATA03) during February 2008 from Dakar to Cape Verde Islands and on the R/V Meteor (M80) in December 2009 covering the region south to Cape Verde Islands.At each station, water samples were collected using 12 l Niskin bottles on a CTD rosette system equipped with temperature, pressure, conductivity and oxygen sensors.Nutrients and oxygen were determined onboard according to Grasshoff et al. (1999).Triplicate water samples were taken from the CTD/rosette casts and were analyzed for dissolved N 2 O onboard using a static equilibration method.For details concerning the N 2 O method, see Walter et al. (2006).
Water samples for δ 15 N nitrate and nitrite analysis were collected in 125 ml HDPE bottles and kept frozen prior to analysis.For logistical reasons, samples from the M77 cruise that contained low to negligible levels of nitrite ([NO − 2 ] < 0.1 µmol l −1 ) were acidified and stored at room temperature, whereas samples with significant [NO − 2 ] were kept frozen prior to the δ 15 N-NO − 2 analysis.Aliquots of these samples were treated in the laboratory with sufficient sulfanilic acid to remove [NO − 2 ] prior to δ 15 N-NO − 3 analysis with any remaining sample stored at room temperature.
The isotopic composition of dissolved nitrate (δ 15 N-NO − 3 ) and nitrite (δ 15 N-NO − 2 ) was measured using the Cdreduction/azide method (McIlvin and Altabet, 2005) with addition of NaCl as described by Ryabenko et al. (2009).All the samples from the Atlantic study region and 50 % of the Pacific samples were analyzed at the IFM-GEOMAR in Germany, while c.50 % of the Pacific samples were analyzed at SMAST in the USA, using the same method.The only difference was that at SMAST N 2 O produced by this method was analyzed isotopically, whereas at IFM-GEOMAR N 2 O was converted on-line to N 2 for isotopic analysis.The detection limit at IFM-GEOMAR was 0.2 µmol l −1 of nitrate or nitrite with the precision of the δ 15 N measurements being ±0.2 ‰.The detection limit at SMAST was slightly higher, 0.5 µmol l −1 , with the same precision of 15 N measurements.
The analyses of the Pacific deep water samples (>1500 m) from both labs gave near identical values of ) of 5.69 ± 0.7 ‰ (IFM-GEOMAR; n = 31) and 5.62 ± 0.4 ‰ (SMAST; n = 8) respectively.The Fisher test showed that we can merge the two data sets with a confidence level >95 %.The resulting mean value of 5.64 ± 0.7 ‰ (n = 39) lies between previously published values of 6.5 ‰ Voss (2001) and4.5 ‰ Sigman (1997) for the deep North Pacific Ocean.Deep waters of the Atlantic showed very similar values as those of the Pacific (5.3 ‰ ± 0.5 for > 2000 m, see Supplementary material, Table 1).

Hydrographic setting of the two study regions
In the North Equatorial Atlantic region, the eastward flow of the North Equatorial Counter Current (NECC) and North Equatorial Under Current (NEUC) supplies oxygen rich waters to the Atlantic OMZ (Glessmer et al., 2009).The water mass distribution in this Atlantic OMZ study region (Fig. 1b) is affected by the Cape Verde Frontal Zone, which marks the boundary between North and South Atlantic Central Waters (NACW, SACW).This separates well-ventilated waters of the subtropical gyre in the north from less-ventilated waters to the south.Our CTD data (Fig. 2) show water mass properties <500 m intermediate between those of NACW and SACW.Antarctic Intermediate Water (AAIW: 3-5 • C, S = 34.5) is found at ∼1000 m with North Atlantic Deep Water (NADW) filling the depth range between ∼1000 and  4000 m (Fig. 2b).A summary of the water masses properties found in both study regions is presented in Supplementary material, Table 1.
In the South Pacific, the Equatorial Undercurrent (EUC), Southern Subsurface Counter Currents (SCCs), and the Southern Intermediate Counter Currents (SICC) supply 13 • C Equatorial Water (13CW, 25.8 < σ θ < 26.6) to the eastern Pacific OMZ (Stramma et al., 2010).The westward flowing South Equatorial Current (SEC) may recirculate some 13 • water from the OMZ by returning eastward as the South Subsurface Counter Current at 3-5 • S (Schott et al., 2004).The origin of 13CW is remote from the equator (Qu et al., 2009) mostly as Subantarctic Mode Water (SAMW; (Toggweiler et al., 1991)) and transports very oxygen depleted waters to the OMZ, due to its relative old age.The South Pacific Subtropical Underwater (STUW) is a likely O 2 source from the south which is centered on the σ θ = 25.0 isopycnal and is well-ventilated across nearly the full width of the subtropical gyre (O'Connor et al., 2002).The low-salinity water (S < 34.5) found between c. 500 and 1000 m southward of 10 • S is Antarctic Intermediate Water (AAIW).South Pacific Deep Water (1.2-2 • C) is found between c. 1500-3000 m and is underlain by Lower Circumpolar Water (LCPW) (Fiedler and Talley, 2006) (Fig. 2a).
Because of the differences in hydrography and significantly lower oxygen supply, the Pacific OMZ is much  (Karstensen et al., 2008).The resulting very low oxygen concentrations favor metabolic pathways that convert nitrogen from biologically reactive "fixed" forms (for example nitrate, nitrite or ammonium) to N 2 via denitrification and/or anammox.

General
Dissolved O 2 varies considerably in its depth distribution between the Atlantic and Pacific study regions as shown in Fig. 3. Figure 4 presents typical water column profiles for oxygen and several nitrogen related properties from both study regions (outside of the upwelling zones).Stations 5 (Atlantic, M80) and 84 (Pacific, M77) were chosen due to synoptic availability of N 2 O, δ 15 N and [NO −  (Chavez and Messie, 2009), and contains water with the highest oxygen concentrations as well as a very sharp oxycline at 80-120 m in which [O 2 ] drops from ∼150 to 20 µmol l −1 .The primary nitrite maximum lies close to the base of the euphotic zone and can be a consequence of two processes (Lomas and Lipschultz, 2006): light-limited, incomplete assimilatory reduction of nitrate by phytoplankton and microbial ammonium oxidation to nitrite (i.e. the first step of nitrification).Near-surface N 2 O is close to saturation and increases within the oxycline from ∼12 to ∼45 nmol l −1 .The observed increase in nitrous oxide within the oxycline can be associated with ammonia oxidation (Codispoti, 2010), which leads to an efflux of N 2 O from the mixed layer to the atmosphere via gas exchange.The concentrations of [NO − 3 ] and [NO − 2 ] above 80 m in layer a at this station were below our detection limit for δ 15 N measurement (0.2 µmol l −1 ).Higher near-surface DIN concentrations were observed at other stations along the 86 • W transect and the corresponding δ 15 N-NO − 3 values were as high as 20 ‰. High surface δ 15 N-NO − 3 is likely the result of incomplete nutrient utilization and fractionation during nitrate assimilation (Granger et al., 2004).
2 is generally much lower than δ 15 N-NO − 3 and the difference between the two increases from layer a to layer b.Relative 15 N depletion in nitrite can be explained by isotopic fractionation during nitrate reduction to nitrite.A smaller difference between δ 15 N-NO − 3 and δ 15 N-NO − 2 observed in the oxycline is likely due to nitrification as the fractionation effect of the process is significantly smaller (∼13 ‰) (Casciotti, 2009;Casciotti and McIlvin, 2007) then the one expected for denitrification (∼25 ‰) (Barford, 1999;Granger, 2006).Thus there is evidence for a clear switch from nitrification to denitrification with depth.
Layer b (120 m to 400 m).O 2 concentrations in this layer drop below 5 µmol l −1 and there is a strong increase in [NO − 2 ] towards a "secondary" maximum at the core of OMZ.N 2 O concentrations drop sharply within the OMZ core to ∼10 nmol l −1 and increase again only towards the lower border of the layer.Denitrification is the only N-removal process which is known to consume N 2 O, hence it is likely that both the increase in [NO − 2 ] and the increase and decrease in [N 2 O] within this layer can be attributed to different stages of canonical denitrification , 2008).The vertical profiles, especially the minimum in N 2 O within the OMZ's core, provide strong evidence that all stages of canonical denitrification influence nitrogen speciation in this layer.The observed increase in δ 15 N-NO − 3 and decrease in δ 15 N-NO − 2 at the base of the layer B are also consistent with denitrification, which leaves [NO − 2 ] depleted in 15 N. Interestingly, the difference between δ 15 N-NO − 3 and δ 15 N-NO − 2 values are higher (∼30 ‰) than fractionation factor calculated for N-loss process within OMZ (∼11.4 ‰, see below) but close to the expected value for pure culture studies (28.6 ‰) (i.e.Barford et al., 1999).The reason for this could be the nitrite oxidation, which has an inverse isotopic fractionation effect, leaving δ 15 N-NO − 2 depleted in 15 N (Casciotti, 2009).Nitrite oxidation can appear in nitrificationdenitrification coupled systems or in anammox as a sidereaction (i.e.Straus 1998, van de Graaf 1996).The deep [NO − 2 ] maximum can support anammox, which has been observed in several previous studies of this region (Galan et al., 2009;Hamersley et al., 2007;Lam et al., 2009).

Atlantic study region
In the Atlantic study region, the oxygen profile has two minima at ∼70 m and ∼400 m (Fig. 3b).The shallow minimum is strongest between Senegal and the Cape Verde Islands and is probably caused by enhanced subsurface remineralization associated with high biological productivity and a shallow mixed layer (Karstensen et al., 2008).The deeper minimum is more prominent south of Cape Verde and is associated with the water mass boundary between Central Water and AAIW (Stramma et al., 2005).The double oxygen minimum (Fig. 3) is therefore caused by the mixing of two water masses from the North and South (NACW and SACW) of the Atlantic region.The profiles from the Atlantic station (Fig. 4, lower panels) are considerably simpler, with fewer subsurface features.Once again, two layers have been distinguished based on oxygen concentration and its influence on dominant nitrogen cycle processes.
Layer a (0-50 m) includes the surface mixed layer which extends to c. 30 m.This layer includes the steepest part of the oxycline, a strong increase in N 2 O with depth, and a primary nitrite maximum which lies at the base of this layer.The δ 15 N of DIN increases steadily throughout this layer and reaches a maximum at a depth close to the primary nitrite maximum.These features can be attributed to a combination of remineralization of organic matter, nitrite excretion by phytoplankton after nitrate reduction and nitrification.In contrast to the Pacific study region, the surface layer has minimum values of δ 15 N in DIN, with some values being strongly negative (e.g.−5.6 ‰ at 20 m).
Examination of near-surface profiles from the Atlantic (Fig. 5) reveals negative values of δ 15 N in DIN within the surface mixed layer at stations located South of Cape Verde and at the TENATSO station.There is a tendency for the values to be most negative at the shallowest depths (20 m) with extremely low δ 15 N values almost always observed in this depth range.Generally below 20 m, both NO − 3 δ 15 N and concentration increase with depth.We argue that the source of nitrate at the very surface of stations with low δ 15 N NO − 3 is from atmospheric deposition (see below, Sect.4.2.5).Nitrite concentration was below the detection limit of 0.02 µmol l −1 , while [NO − 3 ] concentrations in the region are about 0.1-0.5 µmol l −1 .Thus contamination via nitrite cannot be the reason for low δ 15 N values.In regions more influenced by upwelled waters, the near-surface values were higher in the range +4 -+7 ‰ and more consistent with an isotopic signal from upwelled NO − 3 .Even though these concentration levels lie close to our detection limit for δ 15 N measurements (0.2 µmol l −1 ), all surface water samples were measured 5 times and gave reliable values with >95 % reliability and ±0.3 ‰ standard deviation.Further, laboratory tests with dilutions of δ 15 N standards showed no suggestion of any systematic change of measured δ 15 N values with decreasing [NO − 3 ] concentrations.
Apparent exceptions are found at stations 87 at 25 • W and 67 at 30 • W (marked with white and purple crosses in Fig. 5) where low δ 15 N of −4 and −3 ‰ are observed at 50 and 40 m respectively.Corresponding nitrate concentrations are 0.25 and 0.20 µmol l −1 , respectively, and nitrite concentration is below the detection limit.While these values appear to be too deep to be influenced by atmospheric input, in fact the mixed layer is indeed deeper at these stations: 40-50 m instead of 20 m.Thus, we believe, that the low δ 15 N signal at these stations could also originate from atmospheric deposition.To be conservative, we only considered the upper 20m water column for our calculations of nitrogen fluxes in the Table 1 (see Sect. 4.2.5).
At station 1 (TENATSO, marked with black cross in Fig. 5) between 40-60 m isotope signature lay 0 < δ 15 N<5 with nitrate concentrations increasing up to 6 µmol l −1 and nitrite up to 0.55 µmol l −1 .Elevated nitrate and nitrite concentrations having a δ 15 N signature of only few per mil is here likely due to N-fixation, which was observed in this region during other studies (i.e.Bourbonnais 2009).
Layer b (below 50 m) includes the core of the Atlantic OMZ.In contrast to the Pacific OMZ, the Atlantic profiles had no secondary nitrite maximum, and δ 15 N values and N 2 O concentrations remained relatively constant with depth.The N 2 O profiles show no evidence for consumption as was seen in the Pacific.This is a clear indication for the absence of significant denitrification in this region.A slight increase in N 2 O with depth below 50 m can be explained by nitrification (Walter et al., 2006).

Nitrate to Phosphate
Figure 6 presents the NO − 3 to PO 3− 4 relationship (with dissolved oxygen concentrations as the color code) for the Atlantic and Pacific study regions.According to Redfield stoichiometry, the average ocean ratio of N:P is 16:1.Deviations from this ratio can be an indicator for which nutrient sink/source processes are dominating in the ocean region of interest.Waters in the Pacific study region are highly Ndeficient (N:P<16), with the highest deficits found in oxygen minimum waters (purple coloring, Fig. 6a) and associated with the N-removal processes denitrification and/or anammox (Deutsch et al., 2001).Data from the Atlantic study region show strong positive deviations from the 16:1 Redfield stoichiometry, which can be a result of N 2 -fixation (Hansell et al., 2004;Michaels et al., 1996;Gruber and Sarmiento, 1997) and/or nutrient uptake and/or remineralization with non-Redfield stoichiometry (Monteiro and Follows, 2006).Positive deviations from Redfield stoichiometry can also, potentially, be caused by atmospheric deposition of nitrogen (Duce et al., 2008).Note that our treatment of deviations in [NO − 3 ]:[PO 3−

N 2 O vs. AOU
Property-property plots of N 2 O to apparent oxygen utilization (AOU) (with [O 2 ] as the color code) are presented in Fig. 7, where AOU is the difference between the measured dissolved oxygen concentration and its equilibrium concentration in water with the same physical and chemical properties.
N 2 O is the excess nitrous oxide and is defined as the difference between the measured N 2 O and the equilibrium N 2 O concentration at the time when a water parcel had its last contact with the atmosphere.Because the atmospheric N 2 O mixing ratios have been increasing since 1800, the calculation of excess N 2 O has to take into account the age of the water parcel at the time of the measurement.Freing et al. (2009) showed that the difference in the slopes of N 2 O vs. AOU associated with different ways of calculating excess N 2 O can be as much as 17 %.The methods used include the transit time distribution (TTD) approach, where CFC-12 and SF 6 data are used to calculate a mean-age of a water parcel.Alternatively, a "layer" method (Walter et al., 2004) uses different equilibrium N 2 O concentrations for mixed layer and deep waters.For the sake of simplicity we used here the "contemporary" approach, where the [N 2 O] eq in the upper 500 m is calculated based on the contemporary atmospheric dry mole fraction of N 2 O (N 2 O of 322 × 10 −9 for Pacific data and 323 × 10 −9 for Atlantic data, http://agage.eas.gatech.edu/)(Nevison et al., 2003;Yoshinari, 1976).Although the contemporary method may lead to underestimations of N 2 O vs. AOU slopes of up to 17 % (Freing et al., 2009), it does not affect the qualitative comparison of the Atlantic and Pacific study regions given below.An overall linear relationship of N 2 O to AOU (Fig. 7) was observed previously in both regions (Oudot et al., 1990;Elkins et al., 1978;Nevison et al., 2003).
However, the Pacific relationship has two different slopes for oxygen concentrations below and above 50 µmol l −1 (which corresponds here to an AOU of c. 208 µmol l −1 ).For 5 < [O 2 ] < 50 µmol l −1 the slope of the N 2 O to AOU relation is 0.30 ± 0.050 which is significantly higher than that for [O 2 ] > 50 µmol l −1 (0.10 ± 0.009).This is suggestive of a higher yield of N 2 O per mole NO − 3 produced by nitrification at low oxygen levels (Goreau et al., 1980;Stein and Yung, 2003).In Pacific waters with [O 2 ] < 5 µmol l −1 (AOU of c. 248 µmol l −1 ), N 2 O concentrations decrease again to near-zero values, indicative of the consumption of N 2 O at very low oxygen levels mentioned above.Corresponding changes in slope are not visible in the Atlantic data, likely because there are so few data with [O 2 ] < 50 µmol l −1 .The slopes of the N 2 O vs. AOU relationships for [O 2 ] > 50 µmol l −1 are remarkably similar in both regions: 0.10 ± 0.009 and 0.11 ± 0.003 in the Pacific and Atlantic, respectively.These values lie close to the value of 0.107 reported for the tropical Atlantic by Walter et al. (2006) but lower than the value of 0.211 reported by Oudot (2002).The values from the Oudot (2002) paper, however, should be taken with particular care as the mean atmospheric mixing ratio of N 2 O presented in their paper (316 × 10 −9 ) seems to be unrealistically high in comparison to the mean atmospheric background dry mole fraction of N 2 O at the time of their measurements (308 × 10 −9 , http://agage.eas.gatech.edu/).This higher yield of N 2 O under reduced concentrations of oxygen was observed earlier (Goreau et al., 1980) and was attributed to increasing N 2 O yield when ammonia oxidizing microbes become O 2 stressed.This view was challenged by Frame and Casciotti (2010), who showed that ammonia-oxidizing bacteria do not have increased N 2 O yield under low O 2 conditions under environmentally relevant culture conditions.The most recent findings from both the Atlantic and Pacific oceans indicate however that archaeal ammonia-oxidizers (AOA) rather than bacteria may be key organisms for the production of oceanic nitrous oxide and can exhibit higher production rates under low oxygen conditions (Löscher et al., 2011).
Regarding the Pacific observations at very low O 2 , N 2 O removal provides strong evidence for the occurrence of denitrification given its specificity for this process (Bange et al., 2005).However, the amount of N 2 O removed (c.50 nmol l −1 ) is an order of magnitude lower than the observed amount of NO − 3 removal.Hence it gives no indication of the quantitative significance of this process for overall fixed nitrogen removal (e.g. compared to anammox).
Additional insight into N-loss processes is gained here from nitrogen isotope (δ 15 N-NO − 3 ) and fixed nitrogen deficit (N') data.In order to differentiate between these two processes the correlation between δ 15 N-NO − 3 and the N-deficit was calculated (Fig. 8c).In the core of the OMZ, the δ 15 N of DIN is inversely correlated with N' (mainly negative values) and hence with N-removal, whereas in high-oxygen, near-surface waters, δ 15 N-NO − 3 increases independent of N', reflecting fractionation during NO − 3 assimilation by phytoplankton in the euphotic zone (Granger et al., 2004).

Isotope fractionation and N-loss in the Pacific OMZ
The reduction of nitrate to nitrite is the first step of the denitrification process and is also an essential source of NO − 2 for fuelling anammox (Lam et al., 2009).We will next examine the isotope fractionation signal associated with this reduction step.The kinetic isotope fractionation factor can be represented as either α r = 15 R/ 14 R or ε r = (1-α) × 1000, where 15 R and 14 R are the rates of denitrification for 15 NO − 3 and 14 NO − 3 , respectively.An effective or "apparent" value for the fractionation factor for nitrate reduction (ε r ) can be calculated through application of the Rayleigh model to the field data or "diffusive" model of Brandes (1998), where diffusive transport is included.Model of Brandes (1998) require knowledge or estimations of denitrification rates and coefficient of eddy diffusivity for fractionation factor calculation.For the Peruvian OMZ neither of those two parameters are known however, thus we calculation ε * with the Rayleigh model.This model assumes removal from a closed pool of nitrate with constant isotopic fractionation.Hence: where f is the fraction of consumed NO − 3 , f = [NO − 3 ]/(16×[PO 3− 4 ]), the ε * is an "apparent" fractionation factor, in this case for a nitrogen removal process.Least squares fitting of all data from the Pacific OMZ (i.e.[O 2 ] < 50 µmol l −1 ) is shown in Fig. 10a, with the "apparent" isotopic enrichment factor (ε * ) estimated to be +11.4‰ (standard error of the fit is 0.7, Fig. 10).The data are scattered between relationships defined by ε * = 5 and 25 ‰ (assuming a common initial value for δ 15 N initial of 5.2 ‰).This value of ε * of +11.5 ‰ is significantly lower than values estimated from data from the Eastern Tropical North Pacific (22.5-30 ‰) and Arabian Sea (22-25 ‰) (Brandes et al., 1998;Sigman et al., 2003;Voss et al., 2001) and from denitrifier cultures (28.6 ‰) (Barford et al., 1999).However the value lies close to a values determined 30 years ago for 2 stations off southern Peru using much less sensitive analytical techniques (13.8 ‰) (Liu, 1979).
Separating data for shelf and offshore stations (Fig. 9b) results in fits with significantly different values of ε * of 7.6 ‰ and 16.0 ‰, respectively.Similar observations of low ε d have been made in Santa Barbara Basin as compared to the open ETNP (Sigman et al., 2003).This was attributed to a larger contribution from sedimentary denitrification input www.biogeosciences.net/9/203/2012/Biogeosciences, 9, 203-215, 2012 into the water column in the Basin, which has a significantly smaller fractionation effect of 0-5 ‰ due to control of overall NO − 3 removal rate by transport through the sediments (Brandes and Devol, 2002, Lehmann et al., 2007).

Dust deposition signal in the surface waters south of Cape Verde
In marked contrast to the Pacific study region, most δ 15 N-NO − 3 values from the Atlantic (Fig. 8b) stay close to the ocean average value of 5.2 ‰ (Supplementary material, Table 1 (Sigman et al., 2009)).In part this can be explained by the absence of significant fixed-N removal in this region (N' values remain positive, data not shown (Gruber and Sarmiento, 1997)).Notable also was the complete absence of any trend towards higher values associated with partial nitrate utilization in fully-oxygenated, near-surface waters on the M80 samples (stations south to Cape Verde).Significant increases of δ 15 N (up to 12 ‰) in surface waters were only observed at shallow stations very close to the African coast (data from L'Atalante cruise in 2008, not shown) that are likely associated with partial phytoplankton uptake of upwelled NO − 3 (Altabet, 2001;Altabet and Francois, 1994).Decreasing values of δ 15 N of DIN towards the surface have been reported previously for Monterey Bay (Wankel et al., 2007), and for near-surface samples collected close to the Azores Front (30-35 • N) (Bourbonnais et al., 2009) and at Bermuda (Knapp et al., 2010).The lowest values published from this general region (Bourbonnais et al., 2009) were ∼3.5 ‰ at a depth of 100 m.Our data indicate very similar values at this depth.The relatively low values of 3.5 ‰ were attributed by Bourbonnais et al. (2009) to the effects of nitrogen fixation, which can result in remineralised DIN with typical values of −1 ‰ (−2 ‰ to +2 ‰) (Carpenter et al., 1997;Montoya et al., 2002).The strongly negative δ 15 N values measured in surface waters south of Cape Verde (e.g. down to −5.5 ‰, Fig. 5) have not been observed before in oceanic surface waters and cannot be explained by ammonification and nitrification of organic nitrogen produced by nitrogen fixers.On the other hand, very low values of δ 15 N (∼ −7 ‰) of aerosol nitrate have been measured in samples of atmospheric dust from this region (Baker et al., 2007;Morin et al., 2009).Similarly low, negative values have been measured in samples of atmospheric dust originating in the Sahara that were collected from the eastern Mediterranean (Wankel et al., 2010).Recent work (Knapp et al., 2010) shows that the wet deposition flux of fixed-N at Bermuda can be comparable to estimates of biological N 2 fixation rates in surface waters.The δ 15 N-NO − 3 in wet deposition at Bermuda was significantly lower (−4.5 ‰) then δ 15 N added by oceanic N 2 fixation (−2 to 0 ‰) (Hastings et al., 2003;Knapp et al., 2010).For our study region, dry deposition of dust from the Sahara is likely to dominate the N-flux (Duarte et al., 2006).
The N-flux due to diapycnal mixing of NO − 3 from below in the eastern Atlantic (28 • 30 N, 23 • W) has been estimated to be 140 µmol/m 2 /day and was statistically indistinguishable from the integrated rate of nitrate assimilation (Lewis et al., 1986).Later studies in the oligotrophic north Atlantic and at Cape Verde region come to values of about 7 mg N m −2 d −1 (or ∼500 µmol/m 2 /day) (Klein and Siedler, 1995) and at region close to Mauretania to almost double value of about 1037 µmol/m 2 /day, calculated from the cruises data during high upwelling season (Schafstall et al., 2010).According to Baker et al. (2007)

Fig. 1 .
Fig. 1.Oxygen distribution at 200 m (Schlitzer, R., Ocean Data View, World Ocean Atlas 2005, http://odv.awi.de/en/data/ocean/world ocean atlas 2005/) with CTD station locations in the Pacific (a) and Atlantic (b) study regions referred to in the text.White circles indicate station 84 (a) and station 5 (b).

Fig. 2 .
Fig. 2. T-S diagrams with O 2 color coded for the Pacific (a) and the Atlantic (b) study regions from CTD data collected during the M77, M80 and L'Atalante cruises.

Fig. 3 .
Fig. 3. Oxygen distribution in the Pacific (a) and the Atlantic (b) study regions, as measured on the cruises M77, M80 and L'Atalante.The red lines show the water-column profiles for M77 station 84 in the Pacific and M80 station 5 in the Atlantic

Fig. 4 .
Fig. 4. Typical water column profiles from the SE Pacific OMZ (st.84, 81 • W/14 • S) from the M77 cruise (upper panels) and the NE Atlantic OMZ (st. 5, 20.5 • W/12.3 • N) from the M80 cruise (lower panels).Black lines in the Pacific indicate δ 15 N-NO − 3 and δ 15 N-NO − 2 , and the red line indicates δ 15 N-DIN (calculated as the concentration weighed average of the NO − 3 and NO −

Fig. 6 .
Fig. 6. [NO 3 ]:[PO 4 ] relationships in the Pacific (a) and the Atlantic (b) study regions.The data are color-coded by oxygen concentration.Note that the average [NO 3 ]:[PO 4 ] relationship in the Pacific was calculated for [O 2 ]>50 µmol l −1 .

Figure 8
Figure 8 shows N 2 O vs. δ 15 N-NO − 3 (with color coding indicating oxygen concentration) in the two study regions, which helps to reveal processes responsible for the production or consumption of nitrous oxide.In the Atlantic, the profiles and property-property plots show no evidence of N 2 O consumption and the nitrogen isotope values stay close to the oceanic average of 5 ‰, which is also consistent with a lack of denitrification.As discussed above, the dominant process affecting N 2 O in the Atlantic study region is production due to nitrification.For Pacific oxygenated waters ([O 2 ] > 5 µmol l −1 ) the N 2 O vs. δ 15 N-NO − 3 relationship is similar to that found in the Atlantic.The reason for some very low δ 15 N-NO − 3 values in Atlantic surface water is discussed below.A trend towards high δ 15 N-NO −3 values in the Pacific study region (Fig.8a) can be associated with denitrification at lower O 2 concentrations ([O 2 ] < 5 µmol l −1 , purple coloring) or with nitrate assimilation in near surface waters ([O 2 ] > 200 µmol l −1 , red coloring).These two processes cannot be distinguished in figure8aas the N 2 O is close to zero both for waters with [O 2 ] < 5 µmol l −1 (due to denitrification), and for waters with [O 2 ] > 200 µmol l −1 (due to N 2 O equilibration with the atmosphere).In order to differentiate between these two processes the correlation between δ 15 N-NO − 3 and the N-deficit was calculated (Fig.8c).In the core of the OMZ, the δ 15 N of DIN is inversely correlated with N' (mainly negative values) and hence with N-removal, whereas in high-oxygen, near-surface waters, δ 15 N-NO − 3 increases independent of N', reflecting fractionation during NO − 3 assimilation by phytoplankton in the euphotic zone(Granger et al., 2004).

Fig. 8 .
Fig. 8. δ 15 N-NO − 3 vs.N 2 O in the Pacific (a) and in the Atlantic (b) study areas and (c) the δ 15 N vs. N' distribution in the Pacific.The data are color-coded by oxygen concentration.The nitrogen deficit in the figure 8c was calculated as N' = [NO − 3 ] + [NO − 2 ] -16 × [PO 3− 4 ]).δ 15 N vs. N' data reveal two clear trends in the Pacific study region.

Fig. 9 .Fig. 10 .
Fig. 9. (a) Application of Rayleigh model to assess fractionation in the Pacific OMZ for all waters with [O 2 ] < 50 µmol l −1 .Dashed lines indicate relationships calculated for ε d = 5 and 25 ‰.The average calculated or "apparent" fractionation factor for the entire region is 11.4 ‰.(b) Apparent fractionation factors calculated separately for shelf (stations shallower then 200m) and offshore (stations deeper then 200 m) stations.The shelf stations show a lower apparent fractionation factor of 7.6 ‰, while the value for off-shore stations is 16.0 ‰.
the dry deposition N flux of soluble aerosol at 20 • W in the Atlantic ocean is 80-120 µmol/m 2 /day, while wet deposition is 50-70 µmol/m 2 /day.Duarte (2006), for example, estimated a dry deposition N flux of 280 ± 70 µmol/m 2 /day in tropical Atlantic region, which is significant in comparison to the diapycnal flux.This deposition flux is sufficient to supply the observed DIN inventory of the top 20 m (0.2 µmol l −1 ) within two weeks.The most negative δ 15 N values in surface water were observed at stations south of the Cape Verde Islands, which