Notice: Wiley Online Library will be unavailable on Saturday 27th February from 09:00-14:00 GMT / 04:00-09:00 EST / 17:00-22:00 SGT for essential maintenance. Apologies for the inconvenience.
 The spatial extent of N2 fixation in the Atlantic Ocean is examined by determining the isotopic composition of N in suspended particulate organic nitrogen (δ15N PONsusp). The samples were collected from zonal and meridional transects of the Atlantic Ocean during a 3-year period. There is a consistent depleted δ15N PONsusp signal extending over the center of the northern subtropical gyre, which partly coincides with a region where the tracer N* increases westward following the gyre circulation. This nonconservative behavior of N* implies that N2 fixation is responsible for the depleted δ15N PONsusp. A mixing model suggests that N2 fixation over parts of the northern gyre provides up to 74% of the N utilized by phytoplankton. However, since the PONsusp represents only a small fraction of the total N pool, N2 fixation probably only plays a minor role in supplying new N to the euphotic zone in the surface waters of the northern subtropical gyre.
If you can't find a tool you're looking for, please click the link at the top of the page to "Go to old article view". Alternatively, view our Knowledge Base articles for additional help. Your feedback is important to us, so please let us know if you have comments or ideas for improvement.
 The relative abundances of the naturally occurring stable isotopes of N, 15N and 14N (δ15N), have the potential to identify sources of N utilized by phytoplankton and subsequent transformations through the water column (Figure 2b). A unique data set is presented here of the isotopic composition of suspended particulate organic nitrogen (δ15N PONsusp) collected over 3 years (2003 to 2006) in the Atlantic Ocean from four meridional transects (as part of the Atlantic Meridional Transect program [see Robinson et al., 2006]) and two zonal transects across the North Atlantic basin (Figure 1b). This data set provides a large scale context to understand the different nitrogen supply mechanisms to the primary producers in the surface ocean and complements the Bermuda Atlantic Time series Study (BATS) nitrogen isotopic data [Altabet, 1988; Knapp et al., 2005].
 The aims of this study were: (1) to determine the δ15N PONsusp in surface waters (upper 150 m) along transects of the North Atlantic and to identify its large-scale spatial, seasonal and inter-annual variability and (2) to interpret the δ15N PONsusp signals in terms of the sources of N utilized by phytoplankton, particularly focusing on N2 fixation and to identify its relative importance.
2.1. Stable Nitrogen Isotopes as a Tool in Identifying N Pathways
 The study of δ15N in water has revealed how the phototrophic assimilation of nitrogen operates and its dynamics in aquatic ecosystems [Fogel and Cifuentes, 1993]. While the two stable isotopes, 14N and 15N (99.634% and 0.366% by atoms respectively) behave chemically in a similar way to one another [Owens, 1987], during uptake of inorganic N by marine microorganisms, 14N is preferentially consumed relative to the heavier 15N isotope leading to biological fractionation in favor of the lighter isotope in the organisms cell [Montoya and McCarthy, 1995]. However, in environments such as the oligotrophic regions of the Atlantic Ocean where nitrogen is limiting, nitrogen is completely sequestered by the autotroph, there is no fractionation and the isotopic composition of the source-N is reflected within the organism [Altabet and McCarthy, 1985] (Figure 2b).
 The key sources of N are now briefly discussed in terms of their isotopic signal. Atmospheric N2 made available via nitrogen fixation gives rise to an isotopically depleted signal of −2‰ to 0‰ for the δ15N PONsusp, since no fractionation occurs during uptake [Minagawa and Wada, 1986]. NO3− found in rain has an isotopic signal of −2.1‰ to −5.9‰ [Hastings et al., 2003] which could also result in depleted δ15N PONsusp. In contrast, deep nitrate is isotopically enriched with values in the Atlantic Ocean ranging between 4.5‰ and 6‰ [Knapp et al., 2005; Liu and Kaplan, 1989; Sigman et al., 1997]. The enriched signal of nitrate reflects the remineralization of PON, principally from the suspended particulate organic nitrogen (PONsusp) rather than the sinking pool (PONsink). The more abundant PONsusp pool becomes isotopically enriched with depth through progressive decomposition and removal of dissolved N species depleted in 15N [Altabet, 1988]. The δ15N PONsink is also initially enriched with depth [Altabet, 1996], although δ15N PONsink then becomes isotopically depleted deeper in the ocean interior, either due to a loss of an enriched N fraction or by sorption of 14N rich compounds [Altabet et al., 1991; Nakatsuka et al., 1997; Voss et al., 1996]. The isotopic signature of dissolved organic nitrogen (DON) is also important, since aside from molecular N2, it is the largest available nitrogen pool in the surface waters of the oligotrophic gyres (Figure 2a) [Berman and Bronk, 2003; Jackson and Williams, 1985; Mahaffey et al., 2004]. At BATS, δ15N values for DON were consistently found to be in the range of 3.5‰ to 4.5‰ in the upper 100 m over 12 months. However, the stability in both concentrations and the N isotopic composition of DON suggests that this DON pool is mainly recalcitrant and not assimilated by phytoplankton [Knapp et al., 2005].
2.2.1. Sampling Protocol
 Suspended particulate organic nitrogen samples on AMT 12, 14, 16, 17 and 36°N cruises were collected using Stand Alone Pumps (SAPs; Challenger Oceanic; Figure 1b and Table 1). Two GF/F filters (293 mm diameter; Whatman; pre-combusted at 400°C, >4 hours) were placed on the filter bed with the top filter being used for the analysis. The SAPs were deployed to depths of 50, 100 and 150 m and pumped ∼1000 L seawater in 90 min. The filters were wrapped in precombusted (400°C > 4 hours) aluminum foil and frozen (−20°C) until further analysis in the laboratory. Prior to analysis, filters were lyophilized (−60°C; 10−2 Torr, 24 hours).
Table 1. Cruise Details and Methods of PONsusp Collectiona
 PON samples collected on the 24°N cruise (Figure 1b and Table 1) were from Niskin bottles fired in surface waters (10 to 25 m). Seawater (8 L) was filtered through GF/F filters (47 mm diameter; Whatman; precombusted at 550°C, 4 hours) using an acid (HCl) washed glass filtration unit (1 L). Filters were then folded and placed into precombusted aluminum foil (550°C; 4 hours) and then frozen (−20°C) until further analysis in the laboratory. Prior to analysis, filters were dried (60°C; 24 hours) and stored in a dessicator [Landolfi, 2005].
2.2.2. Determination of PON, POC, and Chlorophyll a
 Concentrations of particulate organic carbon (POC) and particulate nitrogen (PN) were determined from freeze-dried SAPS filters [Kiriakoulakis et al., 2004]. Analyses of the filters were carried out in duplicate (analytical error <10%; CEInstruments NC 2500 CHN analyzer), on aliquots of known area (∼130 mm2), using the acid HCl vapor method of Yamamuro and Kayanne . Chlorophyll a concentrations were determined from Niskin water bottle samples and the underway non-toxic water supply using a fluorometric assay of the acetone extract of particulate material collected on a GF/F filter [Welschmeyer, 1994]. Chlorophyll a concentrations were also obtained from an in situ fluorometer fitted on the CTD rosette sampler frame, data being calibrated against the field samples.
2.2.3. Determination of δ15N PONsusp
 Isotopic analyses of the PONsusp were carried out directly on filter aliquots (314 mm2) without decarbonation. The AMT12, 14, 16, 17 and 36°N samples were analyzed using a Thermo Finnigan Delta-Plus Advantage isotope ratio mass spectrometer (IRMS), coupled to a Costech Instruments ECS 4010 elemental combustion system (“EA”) via a Thermo Finnigan Conflo III open-split interface. The EA was equipped with a Costech Zero Blank autosampler, utilizing a 32-position, large-capacity sample wheel. For 24°N samples a Eurovector 3028-Ht elemental analyzer connected to a GV Isoprime mass spectrometer was employed. Sample δ15N ratios were measured relative to atmospheric nitrogen, where
with the samples calibrated against a variety of primary standards, and normalized for weight-to-δ bias and calibration offset using a secondary isotope ratio standard (acetanilide).The samples for the AMT and 36°N cruises were analyzed in duplicate and the precision is based upon taking the variation of the mean of the two measurements made and was found to be ±0.5‰. The accuracy of the analysis was determined as ±0.4‰ by combining errors associated with measuring both the primary standards and the acetanilide standard.
2.2.4. Determination of Inorganic Nitrate and Phosphate
 Inorganic nitrate and phosphate concentrations in seawater samples collected on AMT12, 14, 16 and 17 were analyzed using a Technicon segmented flow colorimetric autoanalyzer [Woodward and Rees, 2001]. Analytical precision was ±2–4% with reproducibility errors in the same range. For 24°N and 36°N cruises, inorganic nitrate and phosphate were measured using a Skalar SanPlus continuous flow autoanalyzer [Sanders and Jickells, 2000].
2.3. Validity of Data: Trophic Tests
 PONsusp is derived from phytoplankton, but may also be modified by organisms from higher trophic levels by grazing and the microbial loop. When N is assimilated by zooplankton, there is an enrichment of ∼3‰ [Minagawa and Wada, 1986], which is significant when using isotopic data to infer N sources to the phytoplankton. In order to determine if higher trophic particulate material has affected the measured δ15N in the present study, the relationships between POC and PON:δ15N PONsusp with chlorophyll a were examined. Organisms of higher trophic levels feeding on suspended particles significantly alter both POC and PONsusp, leading to enhanced values of POC/Chlorophyll a and PON/Chlorophyll a [Waser et al., 2000]. If the PONsusp material measured is principally derived from heterotrophs rather than autotrophs, a positive relationship would be expected between these ratios and δ15N PONsusp. At a confidence level of 95% (P < 0.05; T-test), no statistically significant positive or negative relationship was found in the individual or collective cruise data sets, which implies that the measured δ15N PONsusp is representative of autotrophic PON (Figure 3).
3. Nitrate and Isotopic N Distributions
3.1. Nitricline Structure of the Atlantic Ocean
 The AMT cruise transects pass through distinct biogeochemical regimes of the Atlantic Ocean as revealed by the variations in the nitrate concentrations of the upper 300 m of the water column (Figure 4). The nitricline varies meridionally, reflecting the background gyre structure and pattern of convection (Figures 4a–4d). The nitricline is deepest over the central parts of the subtropical gyres where there is downwelling and subduction (30°N and 25°S), and is shallowest in the tropics where upwelling occurs (10°S to 15°N). The nitricline also shallows poleward from the subtropical gyres, typically outcropping at 40°S in the south Atlantic and north of 45°N in the northern subpolar gyre.
 There is a general westward deepening of nutrient surfaces below the 1 μM isopleths within the northern subtropical gyre, reflecting the deepening of the thermocline through Ekman pumping (Figures 5a and 5b). The increasing depth of the nitricline from the eastern to western basin boundary is more pronounced along the 24°N than the 36°N transect. In addition, at the eastern boundary of the 24°N transect (Figure 5a) there are higher concentrations of nitrate, which decrease considerably westward into the centre of the oligotrophic gyre. A similar pattern is apparent along the 36°N transect (Figure 5b), although concentrations of nitrate are lower at the eastern edge of the gyre and are elevated in the western boundary current.
3.2. Spatial Variability of the δ15N PONsusp in the Atlantic Ocean
 The δ15N PONsusp has been measured in the Atlantic Ocean between 45°S and 50°N (AMT12, 14, 16 and 17) and along 24°N and 36°N. In order to consider the general variations in the δ15N PONsusp, the data have been averaged separately above and below the nitricline (previously defined at 1 μM).
3.2.1. Spatial Variability Above the Nitricline
 For the meridional sections (Figure 6a), there is a consistent depleted signal (−1 to 0‰) in the northern basin between 7°N and 32°N for both the Northern Hemisphere spring (AMT12, 14 and 16) and autumn (AMT17). In the southern gyre, enriched values of 5‰ are found at 40°S and generally become depleted to reach a minimum of ∼2‰ at 12°S. A small depleted region (∼0‰) is found at 32°S. These data represent the Southern Hemisphere autumn (AMT12 and 14), whereas the more enriched values at a similar location derive from the Southern Hemisphere spring. Therefore, while δ15N PONsusp in the northern gyre is consistently depleted, the southern gyre appears to show some seasonality. The δ15N PONsusp is enriched at the equator (4‰) and along the northern flanks of the northern gyre (4–6‰).
 The zonal section along 24°N also shows a depleted signal of −1‰ from 56°W to 45°W (Figure 6c). There are enriched values (6‰) on the western boundary of the gyre at 80°W. However, on the eastern side of the gyre at 30°W, δ15N PONsusp is depleted, with values of 0‰ to 1‰. Along 36°N, δ15N PONsusp values are more enriched (Figure 6c) than along 24°N and δ15N PONsusp increases eastward from 2.2‰ at 60°W to a maximum of 6.8‰ at 28°W.
3.2.2. Spatial Variability Below the Nitricline
 Meridional sections of the δ15N PONsusp reveal a more enriched signal, as expected, below the nitricline (Figure 6b) compared with above the nitricline. The southern gyre (3.7–5.5‰) is slightly more enriched than the northern gyre (3–4‰), with even greater enrichment (6–8‰) being observed at the equator, at the southern flanks of the southern gyre and at the northern flanks of the northern gyre. A zonal section of the δ15N PONsusp again reveals samples being slightly more enriched below the nitricline (Figure 6d) (only shown for 36°N data) compared with above the nitricline, and there is the same general pattern of enrichment from west to east.
4. How Are the δ15N PONsusp Signals Controlled?
4.1. Enriched Signals
 At the equator, the enriched δ15N PONsusp signal coincides with the nitricline (>1 μM) shoaling to depths as shallow as 40 m and is attributed to the wind-induced upwelling of nitrate rich waters. Deep Atlantic nitrate has a δ15N range of 4.5‰ to 6‰ at depths greater than 500 m [Knapp et al., 2005; Liu and Kaplan, 1989]. On the northern flank of the northern gyre, isotopic enrichment might result from deep NO3− supplied by convection at the end of the winter. On the southern flank of the southern gyre, the enriched δ15N PONsusp implies a source of deep NO3−, probably supplied as part of the northern transport of intermediate waters from the Southern Ocean.
 The enrichment in isotopic values with depth in the oligotrophic gyres shows a strong relationship with the nitricline. Phytoplankton lying at or below the nitricline may utilize deep nitrate leading to enriched δ15N PONsusp and this may explain the isotopic gradient in the photic zone. Alternatively, the enrichment of δ15N PONsusp with depth may simply reflect its greater degree of degradation [Altabet, 1988].
 The enrichment in δ15N PONsusp from the Southern Hemisphere autumn to the Southern Hemisphere spring in the south Atlantic can be attributed to deepening of the mixed layer. During the winter months the mixed layer thickens and entrainment of deeper waters leads to a supply of nutrients to surface phytoplankton [Polovina et al., 1995]. The more enriched δ15N PONsusp during the Southern Hemisphere spring (AMT17) in the southern gyre might be the result of uptake of the deep enriched nitrate that was entrained into the euphotic zone during the previous winter.
4.2. Depleted Signals
 The low values of δ15N PONsusp within the northern and southern oligotrophic gyres of the Atlantic Ocean might reflect three distinct N sources.
 1. One possible source is uptake of remineralized compounds excreted by zooplankton, principally NH4+ [Biggs, 1977; Corner and Davies, 1971], which are depleted in 15N. In oligotrophic regions, where phytoplankton take up and assimilate all nitrogen excreted by zooplankton, there is little or no overall isotopic fractionation [Checkley and Miller, 1989] and thus, δ15N PONsusp is depleted.
 3. Uptake of isotopically light N supplied to the surface ocean by atmospheric deposition could give rise to depleted δ15N PONsusp [Hastings et al., 2003].
 Modeled atmospheric fluxes of N to the surface North Atlantic Ocean are only in the range of 0.003–0.007 mol N m−2 a−1 [Dentener et al., 2006; Galloway et al., 2004; Prospero et al., 1996]. In situ aerosol measurements along the AMT transect showed a dry N deposition flux of ∼0.01 mol N m−2 a−1 [Baker et al., 2003] in the same region where there is the depleted signal of δ15N PONsusp in the northern subtropical gyre. These estimates of atmospheric deposition appear to be relatively unimportant in sustaining export production, since they are much smaller than the export flux of 0.42 and 0.56 mol N m−2 a−1 determined for the western subtropical North Atlantic [Jenkins and Goldman, 1985].
 Hence N2 fixation or the uptake of regenerated N are the most likely processes responsible for the depleted δ15N PONsusp. To distinguish between the two processes, the relative size of the NO3− and PO43− pools can be considered. Redfield stoichiometry [Redfield, 1958] is perturbed by N2 fixation in favor of NO3− relative to PO43−. A quasiconservative tracer, N*, which utilizes a linear combination of NO3− and PO43− (2) has been used to define the distribution of N2 fixation and denitrification at the basin scale [Gruber and Sarmiento, 1997],
Gruber and Sarmiento  suggest that high values of N* are due to the remineralization of N-rich organic matter from diazotrophic organisms (see section 4.3.2 for further analysis). Meridional sections of the upper 300 m of the Atlantic over 3 years reveal that high N* values consistently lie below the photic zone in the northern gyre (Figure 7). These signals are in accord with the climatological analysis of Gruber and Sarmiento .
 The localized region of depleted δ15N PONsusp at 32°S found in the austral autumn might be due to a localized patch of N2 fixation or through uptake of remineralized N, since N* values lie between 1 and −1 over the entire southern basin (Figure 7). However, the data set for the South Atlantic is rather limited and it is difficult to provide a reliable explanation for the depleted δ15N PONsusp signal there.
4.3. Synthesis of the Atlantic Ocean
4.3.1. Spatial Extent of the Depleted Isotopic N Signal
 A synthesis of δ15N PONsusp data for samples collected above the nitricline has been made by collating data from the present study (excluding AMT17 which sailed in the Northern Hemisphere autumn) and from AMT10 surveyed in the Northern Hemisphere spring 2000 [Mahaffey et al., 2004, 2003] (Figure 8a). In the northern gyre between 25°W and 55°W, there is an extensive region of depleted δ15N PONsusp, while elevated values occur in the more temperate waters of the Northeast Atlantic and off the west coast of Africa. In the South Atlantic, there are low values of δ15N PONsusp in the center of the subtropical gyre, with more enriched values off the east coast of South America and at the equator.
 The robustness of these signals is assessed by calculating the standard deviation (σ) and the standard error (σ/√n) based upon the number of independent data points (n) (Figures 8c and 8d). Within the northern gyre, the zone of depleted δ15N reveals little variability and low standard errors are implied from the independent data points. Thus, the extensive region of depleted δ15N PONsusp appears to be a robust signal. However, on the eastern side of the north Atlantic basin, there is a large standard deviation and standard error, reflecting how conditions are more variable there with upwelling off the west coast of Africa and convection in the north east Atlantic. The variability of the signals in the South Atlantic is small where there is data, but the data coverage is too limited to infer any robust signals.
4.3.2. Spatial Extent of Elevated N*
 The depleted δ15N PONsusp signal in the northern subtropical gyre coincides with N* having a local maximum between 10°N and 30°N in the centre of the gyre (Figures 9a and 9b). N2 fixation is only reliably implied from N* when this tracer increases following the flow [Deutsch et al., 2001]; that is, N2 fixation leads to enhanced N:P ratios in the underlying thermocline following a moving water mass. Thus elevated N* over the North Atlantic only provides a crude bulk measure of N2 fixation occurring in the basin, instead, it is the gradient in N* created by horizontal transport which is a more accurate indicator of the source region. Comparing our N* signals (Figure 9b) and the latest drifter data (Figure 9c) [Lumpkin and Pazos, 2006] suggests that the local maximum in N* occurs where there is westward flow on the southern flank of the subtropical gyre. Thus, in our view, the most likely region of N2 fixation in the North Atlantic is between 30°W–50°W and 15°N–30°N, where there is both depleted δ15N PONsusp and a westward increase in N* following the gyre circulation.
 This region of potential N2 fixation is subject to enhanced dust inputs from the Saharan Desert, which provide the limiting nutrients (P and Fe) for marine diazotrophs [Mills et al., 2004] and creates an ideal niche for N2 fixation to occur. In support of this view, Mahaffey et al.  reported a natural fertilization event in the Northeastern Atlantic with coincident signals of depleted δ15N PONsusp, phytopigments indicative of cyanobacteria and prochlorophytes, and elevated N:P ratios in the thermocline.
4.3.3. N2 Fixation: Local or Far-Field Signatures?
 The region of depleted δ15N PONsusp appears to be more confined than the N* signal, which extends over much of the North Atlantic. This mismatch poses the question of whether these independent signals reflect local or far field control? It has been previously suggested that the N* signal could be advected around the northern basin with the gyre circulation [Hansell et al., 2004]. While at the surface, NO3− and PO43− have very short residence times due to their rapid uptake by photoautotrophs, below the euphotic zone these nutrients are not utilized rapidly. Therefore, the NO3− and PO43− can be advected within the thermocline without much modification and, thus, distribute the N* signal around the northern gyre. In turn, the isotopic signal might partially reflect the effect of advection. However, the lifetime of PONsusp is only 30 to 60 days in the euphotic zone [Charette et al., 1999] suggesting that PONsusp might be advected 150 to 300 km, assuming a typical surface velocity of 5 cm s−1 in the gyre interior. Therefore the different extent of the N* and depleted N isotopic signals reflect their different lifetimes.
5. Significance of N2 Fixation Within the North Atlantic
5.1. Role of N2 Fixation in Providing New N to PONsusp
 There has been some controversy surrounding the importance of N2 fixation as a source of new N to the oligotrophic regions of the North Atlantic. At BATS in the Sargasso Sea, the percentage contribution of new N from N2 fixation [Knapp et al., 2005] determined both from direct in situ rate measurements and geochemical estimates varies from 2% [Orcutt et al., 2001] to 8% [Gruber and Sarmiento, 1997]. A simple two-end-member mixing model (3) is applied to the present North Atlantic data set in order to estimate the percentage contribution of new N supplied via N2 fixation through the PONsusp pool (Table 2).
Table 2. Estimates of the Percentage Contribution of New N That N2 Fixation Provides to the PONsusp Pool and Subsequently to the Total N Pool Over Separate Regions of the Northern Subtropical Gyre
Northern Subtropical Gyre
Western Side (65°W–80°W, 20°N–40°N)
Center (30°W–65°W, 7°N–32°N)
Eastern Side (5°W–30°W, 7°N–40°N)
Standard error associated with the δ15N PONsusp, based upon (σ/√n), where σ is the standard deviation and n is the number of independent data points.
 The domain is separated into a region on the western side of the subtropical gyre (65°W–80°W, 20°N–40°N), the center of the oligotrophic gyre (30°W–65°W, 7°N–32°N) and the eastern side of the gyre (5°W–32°W, 7°N–40°N). Within these regions, the average δ15N PONsusp was identified, and the percentage contribution of new N that N2 fixation contributes to the PONsusp pool was then estimated using three boundary δ15N values for deep nitrate (4‰, 4.5‰ and 5‰) and a value of −1‰ for N2 fixation,
 Given the three choices for the boundary values of deep NO3−, the percentage contribution of N from N2 fixation to the PONsusp pool within the central part of the subtropical gyre is estimated to range between 72% (±5%) and 76% (±4%) (Table 2), with the errors representing the standard error of the δ15PONsusp signals. On the western side of the gyre, the contribution of N2 fixation was found to be lower in the range of 27% (±15%) to 40% (±12%) and lower again on the eastern side of the basin from 9% (±7%) to 24% (±6%). Hence, N2 fixation is clearly important over the central part of the northern subtropical gyre in providing new N to the PONsusp pool, but less important on the western and eastern sides of the subtropical gyre. This difference suggests that other processes become important on the western and eastern sides of the gyre: On the western side, eddy transfer of nitrate close to the western boundary current is likely to be important [Williams and Follows, 2003] and, on the eastern side, the lateral transfer of nitrate and DON from coastal upwelling [Roussenov et al., 2006].
5.2. Role of N2 Fixation in Providing New N to the Total N Pool
 The isotopic calculation is now extended to estimate the percentage contribution of new N that N2 fixation introduces via PONsusp to the total N pool. The total N pool is defined here by DON, NO3− and PONsusp; while PONsink is important in the export of N out of the euphotic zone, PONsink makes only a minor contribution to the total N pool [Wakeham and Lee, 1989]. Within the centre of the northern gyre, the PONsusp pool represents only ∼6.8% of the total N (Figure 1a and Table 2). Thus the average contribution of new N supplied from N2 fixation via PONsusp to the total N pool is 5.0% (±0.3%) again with the errors representing the standard error of the δ15PONsusp signals for the region (Table 2, assuming 4.5‰ for the deep nitrate). At the western and eastern sides of the gyre the contribution of new N supplied from N2 fixation via the PONsusp are even smaller and only typically reach 0.6% (±0.2%) and 0.8% (±0.3), respectively.
 The contributions of NO3− and DON now need to be considered in the total N pool. DON dominates the total N pool in surface waters accounting for 93–96% across the northern basin, while surface NO3− represents only 0.4–1.9% of the total N pool. While no measurements of δ15N NO3− or δ15N DON were made in the present study, data are available from BATS with typical values of 3.5‰ for surface NO3− and 4‰ for DON (Table 3) [Knapp et al., 2005]. Using these data, applying the two-end-member mixing model (3) to the western side of the northern subtropical gyre suggests that the new N supplied to the total N pool from N2 fixation via DON is 8.2% (±8%) and via NO3− is 0.3% (±0.1%) with the error range now representing the standard deviation of the estimates obtained from the choice of three boundary δ15N values for deep NO3−. Thus our estimate of the total supply of new N from N2 fixation reaches 9.2% (±4.5%) over the western side of the northern subtropical gyre. If the same calculations were applied to the central and eastern regions of the North Atlantic including the DON and NO3− pools there (Table 2), the total supply of new N from N2 fixation would be 13.7% and 9.2%, respectively. However, since it is unlikely that δ15N DON or δ15N NO3− are the same across the entire northern basin, these estimates are speculative. Further work is necessary to determine the stable N isotopic composition of components of the N pool in order to understand the different sources of N utilized by phytoplankton.
Table 3. Two-End-Member Mixing Model (4) Applied to Surface NO3− and DON to Gain an Estimate for the Percentage Contribution of New N That N2 Fixation Makes Via Surface PONsusp, NO3−, and DON Over the Western Side of the Northern Subtropical Gyre
Typical concentrations (averaged above 1 μM nitricline) for region. Range of boundary values used for deep NO3− while δ15DON is 4‰, and δ15N surface NO3− is 3.5‰ (average at 100 m at BATS) [Knapp et al., 2005].
Percent contribution from N2fixation through DON to total N pool
8.2% ± 8.0
Total % contribution from N2fixation to total N pool
5.3. Wider Context
 The isotopic mass balance method employed in this study reveals that nitrogen fixation varies over the northern subtropical gyre, accounting for 72% to 76% of the PONsusp in the central region (30°W–65°W, 7°N–32°N) and 27% to 40% in the western region of the gyre (65°W–80°W, 20°N–40°N). In comparison, using the same method, Montoya et al.  estimates the contribution of nitration fixation to PONsusp of 30–40% over the tropics (25°W–60°W, 0°–15°N) and the western side of the gyre (42°W–80°W, 7°N–28°N); likewise, Capone et al.  estimates a similar contribution of 36% over the western side of the gyre (42°W–80°W, 7°N–28°N). These estimates of Montoya et al.  and Capone et al.  are consistent with our estimates over our western region, but are lower than our estimates over the central region of the subtropical gyre. Montoya et al.  and Capone et al.  used the same two-end-member mixing model (3), but with a more depleted δ15N value for N2 fixation of −2‰, rather than our boundary value of −1‰. Our estimates for nitrogen fixation for the central region only decrease to 60–65% when this lower boundary value of −2‰ is instead used. Thus the contrasting estimates of nitrogen fixation over the central part of the subtropical gyre reflect differences in the isotopic data for PONsusp.
 On previous AMT cruises (1–8), the nitrogen-fixing cyanobacteria Trichodesmium spp. was found to be most abundant to the north of the equator (0°–15°N). Tyrrell et al.  estimated the nitrogen fixation provided >20% of total new N input in these tropical waters. This estimate is greater than our estimates of nitrogen fixation providing ∼9% of the total N pool in the western North Atlantic and the speculative 14% in the central part of the subtropical gyre, but these difference might easily be accounted for by the different regions, timing and uncertainties in the studies.
 In our view, N2 fixation leads to the extensive depleted δ15N PONsusp over the North Atlantic subtropical gyre, as well as the westward increase in N* following the gyre circulation along 20°N. This view is in accord with independent studies suggesting that N2 fixation occurs in the Atlantic Ocean and provides a significant source of N phytoplankton to the subtropical Northern gyre [Capone et al., 2005; Gruber and Sarmiento, 1997; Michaels et al., 1996]. However, the importance of N2 fixation in providing an area-integrated flux is uncertain, since estimates range from 0.03–0.3 mol N m−2 a−1 (see Capone et al.  for full details). This uncertainty reflects the difficulty of providing area-averaged estimates for a process associated with blooms, which are spatially heterogeneous and often short-lived.
 A unique data set of stable isotopic data for δ15N PONsusp is presented for the Atlantic Ocean. Over the central part of the subtropical northern gyre, there was an extensive signal of depleted δ15N PONsusp. This signal was persistent and robust, being observed over 3 years both in Northern Hemisphere spring and autumn. The PONsusp signal was primarily autotrophic rather than heterotrophic, based upon interpretation of the isotopic signals, particulate organic carbon, nitrogen and chlorophyll a levels.
 The extensive depleted isotopic signal might be formed either by N2 fixation or recycling. N2 fixation appears to be the more plausible explanation, since the depleted signals occur in a region where there is a westward increase in the quasi-conservative tracer N* following the gyre circulation along 20°N.
 Estimates of the percentage contribution of new N that N2 fixation introduces via the PONsusp pool to the total N pool have been calculated using a two-end-member mixing model. This estimate typically reaches ∼74% in the centre of the oligotrophic gyre. Since the phytoplankton are included in the PONsusp pool, this underlines the importance of N2 fixation as a source of new N to the primary producers. However, PONsusp accounts for only ∼6.8% of total N in the center of the gyre, so the contribution to the total N pool of new N from N2 fixation is only 5%. If this estimate is repeated at the western side of the gyre including isotopic values for NO3− and DON [Knapp et al., 2005], then ∼9% of new N can be accounted for by N2 fixation. This estimate is likely to be only slightly higher over the central part of the subtropical gyre, but further isotopic values are needed for DON and surface NO3− to provide a reliable estimate. It seems, therefore, that N2 fixation can be an important source of N to the phytoplankton, but is unlikely to account for the mismatch in the supply of N and the observed export production for the subtropical gyre. There is clearly a need for further work in obtaining measurements of δ15N NO3− and DON in the North Atlantic in order to identify the relative importance of the different pathways of N through the marine system.
 This study was supported by the UK Natural Environment Research Council through the Atlantic Meridional Transect consortium (NER/O/S/2001/00680) and the 36°N consortium (NER/O/S/2003/00625). We would like to thank Katie Chamberlain, Sinhue Torres, and Tim Lesworth for their assistance in the analysis of nutrients, Alex Poulton for the analysis of chlorophyll. a, the officers, crew and UKORS staff of RRS James Clark Ross, RRS Discovery, and RRS Charles Darwin. Thanks go also to Robert Petty at the Marine Science Institute Analytical Lab, University of California, for isotopic analysis. This is contribution 150 of the AMT Programme.