High N2 Fixation in and Near the Gulf Stream Consistent with a Circulation Control on Diazotrophy

The stoichiometry of physical nutrient supply may provide a constraint on the spatial distribution and rate of marine nitrogen (N2) fixation. Yet agreement between the N2 fixation rates inferred from nutrient supply and those directly measured has been lacking. The relative transport of phosphate and nitrate across the Gulf Stream suggests that 3–6 Tg N year−1 must be fixed to maintain steady nutrient stoichiometry in the North Atlantic subtropical gyre. Here we show direct measurements of N2 fixation consistent with these estimates, suggesting elevated N2 fixation in and near the Gulf Stream. At some locations across the Gulf Stream, we measured diazotroph abundances and N2 fixation rates that are 1–3 orders of magnitude greater than previously measured in the central North Atlantic subtropical gyre. In combination, rate measurements and gene abundances suggest that biogeochemical budgets can be a robust predictive tool for N2 fixation hot spots in the global ocean.


Introduction
Nitrogen is an essential element to all living organisms. However, over much of the surface ocean, bioavailable nitrogen is in scarce supply. Some microorganisms, called diazotrophs, have adapted to this scarcity by fixing N 2 gas dissolved in seawater to its bioavailable forms. In so doing, diazotrophs replenish the global ocean reservoir of fixed nitrogen compounds and exert a key control on marine productivity and the ocean's biological carbon pump (Tyrrell, 1999). Despite this important role in the Earth system, the rates and spatial distribution of biological N 2 fixation are highly uncertain (Landolfi et al., 2018). A quantification of N 2 fixation rates in the subtropical North Atlantic, along with questions about the sources of non-nitrogenous nutrients, such as phosphorus (P) and iron (Fe), needed to fuel such fixation, have provoked extensive debate Gruber & Sarmiento, 1997;Mahaffey et al., 2005;Marconi et al., 2017;Palter et al., 2011). Moreover, biogeochemical cycles in the North Atlantic can take on global importance, since this is the basin in which large-scale ocean circulation unites the nutrients subducted in the Southern Ocean with aeolian iron mobilized from the Northern Hemisphere land masses and deposited at the ocean's surface (Moore et al., 2009).
Because N 2 fixation is energetically costly, diazotrophs are thought to have a competitive advantage only where the supply of bioavailable nitrogen compounds (N) is low relative to that of P and Fe with respect to the demands of non-diazotrophs (Falkowski, 1997;Landolfi et al., 2015;Ward et al., 2013). If non-diazotrophs can exhaust P or Fe before N begins to limit their growth, then there may be no niche for N 2 fixing organisms. Given that the P:N ratio is very low in the subsurface waters of the subtropical North Atlantic (Palter et al., 2011), the phosphate that gets mixed vertically into the euphotic zone is accompanied by more than enough nitrate to be completely consumed by non-diazotrophs before any niche is created for N 2 fixing organisms (Ward et al., 2013). These subsurface nutrient ratios imply that N 2 -fixing organisms in the central gyre may depend on dissolved organic phosphorus to support their growth (Landolfi et al., 2015).
However, conditions near the Gulf Stream are distinct from the broader subtropical North Atlantic. North of the Gulf Stream, phosphate concentrations are in excess of the average non-diazotrophic biological nitrate demand, as indicated by positive values of the metric P* = PO 3 − 4 -NO − 3 /16. Assuming that the average community N:P ratio is 16 (Deutsch et al., 2007;Gruber & Sarmiento, 1997), a niche for N 2 fixers may be created in or near this region. Hydrographic sections occupied across the Gulf Stream consistently show positive P* north of the current (Palter et al., 2011). Likewise, recent measurements have also shown that the Slope Sea, between the Gulf Stream and North American shelf, has elevated concentrations of dissolved iron relative to the subtropical gyre, so that the Fe:N ratio may exceed the cellular requirements of non-diazotrophs near the Gulf Stream (Conway et al., 2018). Finally, along its path from the Florida Strait to an offshore position of 60°W, 38°N, the Gulf Stream transports water with temperatures exceeding 20°C all year, and 23°C for all but the coldest months. Therefore, the transport of P-and Fe-rich waters from the Slope Sea across the warm Gulf Stream may alleviate the nutrient limitation of diazotrophy and mitigate any potential temperature effect of the cooler wintertime subtropical water masses.
A cross-Gulf Stream supply of P* and Fe would also help explain satellite-based indications (Westberry & Siegel, 2006) and enigmatic video footage (Davis & McGillicuddy, 2006) of elevated abundances of the marine N 2 fixing organism Trichodesmium along the southern fringe of the Gulf Stream. Here, we present direct observations of N 2 fixation across the Gulf Stream, along with gene copy abundances of Trichodesmium and UCYN-A2, and microscopic observations of Trichodesmium and Richelia. With these data, we lend evidence in support of the hypothesis that the region in and near the Gulf Stream fosters high abundances of N 2 fixing organisms and robust rates of N 2 fixation.

Overview of Sampling and Data Collection
The majority of the data presented here were collected during EN596, a 6-day cruise aboard the R/V Endeavor from 25-30 April 2017, which occupied a high-resolution section across the Gulf Stream ( Figure 1). Additional opportunistic sampling was also conducted on a subsequent leg, EN597, from 1-7 May 2017 between Morehead City, NC, and Fort Lauderdale, Florida. The Gulf Stream was identified from satellite sea surface temperature (SST) data, and the hydrographic section on EN596 was intended to cross the Gulf Stream perpendicular to its local trajectory at 74°W. On this hydrographic section, seawater samples were collected at approximately 24 discrete depth intervals from the surface to 1,000 m, and the distance between stations was approximately 10 km.
Here we give an overview of the sampling and data collection, with all of the methodological details explained in Text S1 in the supporting information. Water samples were collected and analyzed for nutrients, nitrogen isotopes, and chlorophyll a at each of the hydrographic stations. Particulate material for DNA extractions was also collected at each hydrographic station and during transits between stations using an automated custom-made seawater sampler (based upon the design of Holser et al., 2011) connected to the ship's uncontaminated seawater system, drawing seawater from a nominal depth of 5 m. The filtration system was programmed to iteratively redirect water through tubing connected to filter holders in order to sample at a high spatial resolution along the cruise track ( Figure 1).
On Leg 1, at Stations 2, 6, and 12 (Figure 1), the entire volume of a PVC rosette bottle (from a depth of 25 m) was gravity filtered through a 47-mm diameter, 10 μm pore size, black polycarbonate filter with a Sterlitech polyester drain disk as per the protocol of White et al. (2018) in order to enumerate large cell-sized diazotrophic taxa. At these same stations, as well as Station 12 on Leg 1 and Stations 5, 8, and 9 on Leg 2, N 2 and carbon fixation rates were measured via the modified bubble release 15 N 2 -tracer method of Jayakumar et al. (2017) and via traditional 13 C methods of Legendre and Gosselin (1997).

Oceanographic Context
As is evident in Figure 1a, the high-resolution section crossed the Gulf Stream where its ribbon of warm (>25°C), surface waters separates the cold (<19°C) Slope Sea waters to the north from the subtropical spring mixed layer (22°C and 23°C) to the south. The Slope Sea is much richer in chlorophyll than the Gulf Stream or subtropical gyre (Figure 1b). Eddies and filaments of elevated chlorophyll concentrations extend hundreds of kilometers southward and eastward from the Gulf Stream into the more oligotrophic central gyre. These warm, high chlorophyll features show how water is stirred across the Gulf Stream and stretched into coherent structures.
Finite Size Lyapunov Exponents (FSLE), calculated from the altimetric surface geostrophic velocity field and made freely available from the AVISO group (d 'Ovidio et al., 2004), identify regions where the mesoscale velocity field stretches tracer into thin tendrils ( Figure 2). More specifically, the large-magnitude ridges of the backward FSLE ( Figure 2) indicate convergent flow lines (Liu et al., 2018) and partition the flow field into regions that have come from a similar origin. Many studies have shown that these ridges correspond with filaments of chlorophyll and other tracers in the ocean, consistent with their convergent nature (e.g., Hernández-Carrasco et al., 2018; Lehahn et al., 2007). The fact that many of these ridges coincide with  Table S1). (b) The MODIS Aqua 4 km satellite chlorophyll (mg m −3 ; colors), averaged over 1-8 May 2017, which is the 8-day average closest to the cruise dates with reasonably cloud-free view of the cruise region. Cloud covered pixels are white. The crosses are stations where qPCR was performed on bottle samples collected at various depths and reported in Table S2. The numbers next to the crosses are the leg and station numbers (labeled as leg.station). Bottles were collected to a depth of 1,000 m for the analysis of nutrients and natural abundance of δ 15 N of nitrate at 12 stations between stations 1.1 and 1.12. All other symbols are the same as in panel a.
10.1029/2020GL089103 elevated chlorophyll and SST structures on the subtropical side of the Gulf Stream ( Figure 1) suggests that these features have been, at least in part, passively advected by the surface circulation. An illustration of such a feature is the tongue of high chlorophyll extending southeastward from the Gulf Stream at 74°W and about 35°N to 33°N, along a clear ridge of the backward FSLE ( Figure 2), a feature that is also visible as a tongue of warm SST (Figure 1a). Figure 2 shows the distribution of N 2 fixation rates and Trichodesmium nifH gene copy abundances across the Gulf Stream. This section is shown versus latitude in Figure 3, along with the nifH gene copy abundances of the unicellular cyanobacterium UCYN-A2, above the Gulf Stream velocity as a function of depth and latitude. The highest Trichodesmium and UCYN-A2 nifH gene abundances are found very near the swiftest velocities of the Gulf Stream, while the highest N 2 fixation rate was measured at a station more than 10 km south of the zero velocity isotach and, therefore, outside of the Gulf Stream. The satellite chlorophyll concentration and FSLE ridges ( Figure 2) indicate that the high Trichodesmium and UCYN-A2 abundances coincide with regions of lateral convergence, suggesting that the circulation field may passively accumulate the cells.

Diazotroph Abundance and N 2 Fixation Rates
Trichodesmium is known be positively buoyant and may be prone to accumulate while it floats in convergent surface circulation (Detoni et al., 2016;Walsby, 1975). Indeed, diazotroph gene abundance was much higher at 5 m than any other depth (Table S3), consistent with the hypothesis of passive accumulation of the cells, but also typical of the vertical distribution of cyanobacterial diazotrophs (decreasing with depth). Cell counts determined via epifluorescence microscopy at 25 m were much lower (≤14 Trichodesmium cells L −1 and ≤437 Richelia heterocysts L −1 ) but did independently confirm the presence of these large cell-sized diazotrophs. We note that due to known polyploidy and limitations in qPCR assays we would not expect 1:1 relationship with gene copies and cell counts (White et al., 2018).
While vertical profiles of gene copies could suggest physical accumulation of at least Trichodesmium, we cannot rule out the possibility that conditions within the Gulf Stream fostered active Trichodesmium growth at locations where its abundance was high, as there were no incubations or microscopic observations precisely colocated with the highest gene abundances. From the underway data at 5 m, Trichodesmium showed as the most consistently dominating diazotroph in the area. Its abundances were measured at 10 4 nifH gene copies L −1 at all the stations both within the Gulf Stream and south of its swiftest central velocity (Figures 2 and 3). UCYN-A2 abundances were up to 5,000 gene copies L −1 in the Gulf Stream, with the notable exception of the station closest to the center of the current measuring 60,000 gene copies L −1 . UCYN-A2 has often been considered a coastal phylotype, but the understanding of its global distributions is evolving (Thompson et al., 2014). The g24774A11 diazotroph phylotype was found consistently at 'detected but not quantifiable' (DNQ) levels along the high resolution sampling transect, with the exception of the northernmost stations beyond the Gulf Stream north wall, where it was not detected in the samples. The g24774A11 distributions thus support the idea that Gulf Stream waters are enhancing diazotroph fitness.
One hypothesis to explain the disconnect between the N 2 fixation rates and gene abundances is that in the areas where highest rates were measured, the diazotrophic cells experienced conditions more favorable for fixation, resulting in greater N 2 fixation activities per cell. Conversely, in areas with reduced N 2 fixation rates, the diazotroph cells may have been senescent. Alternatively, a relatively fast local reduction in cell-specific N 2 fixation rates could have been induced by a local injection of dissolved inorganic nitrogen (DIN) to the system. Trichodesmium can switch to nitrate uptake when it becomes available, while reducing its N 2 fixation rates (e.g., Holl & Montoya, 2005). In addition, it is possible that such disconnect between rates and gene abundances may arise by changing diazotroph community composition across the transect. In this study we screened for the presence of the known key oceanographic diazotroph groups; yet it is possible that some of the active diazotrophs were missed. Amplicon sequencing data (not shown) targeting the nifH gene containing communities was processed from the stations and showed presence of many non-cyanobacterial (presumably heterotrophic) diazotrophs in addition to the ones quantified with qPCR. These groups could have potentially contributed to rates, although evidence is scarce in terms of their actual contributions to N 2 fixation activity in the oceanic surface waters (e.g., Moisander et al., 2017).
In the global diazotroph abundance databases of Luo et al. (2012) and Tang and Cassar (2019), there are no samples of nifH gene abundance measurements directly in the Gulf Stream. In the central subtropical gyre (i.e., near Bermuda, about 32°N), Trichodesmium and UCYN-A abundances are generally below 100 gene copies L −1 and often not detectable at any depth (c.f. Tang & Cassar, 2019, their Figures S1 and S2). Thus, the peak measured gene copies here are 1-3 orders of magnitude above those typically measured in the central North Atlantic subtropical gyre, though they are several orders of magnitude smaller than frequently measured in the tropical Atlantic.
The N 2 fixation rates reported here are also between a factor of two and ten higher than nearby open ocean measurements: with rates between 6 and 18 nmol L −1 day −1 during this April cruise (Figure 1, Table S1), compared to rates consistently below 4 nmol L −1 day −1 measured in the open ocean subtropical North Atlantic gyre in August . These rates are consistent with those newly reported for the northwestern gyre (Tang et al., 2020).

Comparing Measured Rates to Inferences from Biogeochemical Budgets
The strong gradients across the Gulf Stream are a reflection of its role as a sharp dynamical and biogeochemical divide, with high nutrient layers outcropping at the sea surface to the north of the current and near-zero concentrations to the south ( Figure 4). As has been observed on six previous cross-Gulf Stream nutrient sections (Marshall et al., 2009;Palter et al., 2011), P* is enriched on the northern side of the current (Figure 4b).
Additionally, a Geotraces cruise (GA03) recently revealed that the Slope Sea is enriched in iron relative to the Sargasso Sea (Conway et al., 2018). Therefore, the environment near the Gulf Stream has many of the nutrient elements required for diazotrophs to thrive. Moreover, down-front winds and a rich spectrum of mesoscale eddies associated with the Gulf Stream drive a down-gradient flux of P* and iron into the subtropical gyre (Conway et al., 2018;Letscher et al., 2016;Palter et al., 2011;Roussenov et al., 2006;Williams & Follows, 2003;Williams et al., 2011;Yamamoto et al., 2018).
To maintain steady state nutrient ratios in the Sargasso Sea, any net P* supply must be consumed (Deutsch et al., 2007). The most likely P* sink is diazotrophic N 2 fixation, which consumes water column phosphate without drawing down nitrate. The steady state budget equation for P* can be rearranged to solve for N 2 fixation, J fix (N), integrated over the annual maximum mixed layer of depth, H, as in (Palter et al., 2011): Here, the P* supply terms include (1) advection across a P* gradient by the horizontal velocity field u and (2) the lateral mixing of the P * gradient with horizontal mixing coefficient A h , both integrated above the depth H. The third term is the vertical advection of P* across the base of the layer with vertical velocity, w, and the fourth term is the turbulent mixing across a gradient in P* at the base of the layer with vertical mixing coefficient A v . The supply of fixed nitrogen by atmospheric deposition (the fifth term on the right hand side) hull-mounted ADCP. In Figure 4, the location of the highest depth-averaged velocity (36.1°N) is taken as the center of the Gulf Stream, and the horizontal distance axis is transformed to be perpendicular to the current direction at this location. reduces the amount of N 2 fixation required to balance the physical supply of P*. The factor λ represents the number of moles of N that must be fixed to balance each mole of P* supplied to the region (Deutsch et al., 2007). For oligotrophic systems like the Sargasso Sea with low fractions of organic material exported from the mixed layer, λ predominantly reflects the N:P ratio for non-N 2 fixing organisms, and is often elevated above 16 (Mills & Arrigo, 2010;Moreno & Martiny, 2018;Weber & Deutsch, 2010). Thus, a choice of λ = 16 should yield a conservative estimate of the N 2 fixation required to balance the estimated P* supply. An upper bound of the N:P ratio for the subtropical North Atlantic has been estimated at 25 (Galbraith & Martiny, 2015). If we were to use this higher estimate for λ, all of the N 2 fixation rates from the P* budget calculations would be revised upwards by a factor of approximately 1.5.
Using the average P* gradients from six cross-Gulf Stream nutrient sections, similar to those shown in Figure 4b, together with satellite winds and mixing coefficients taken from the literature, Palter et al. (2011) estimated the physical supply of P* by transport across the Gulf Stream, as in Equation 1. A scale analysis showed that atmospheric nitrogen deposition and vertical mixing and advection were small P* sink terms, and were essentially negligible in the budget. Thus, the budget simplified to a balance between the P* lateral transport convergence (Term 1) and the biological N 2 fixation that consumes it (J fix (N)). This budget calculation suggested the cross-Gulf Stream mixing and advection of P* could support subtropical gyre N 2 fixation rates of 4.7 ± 1.6 Tg N year −1 . The new hydrographic and nutrient data collected during EN596 are consistent with these estimates, because the cross-stream P* gradients are very close to the averages reported in Palter et al. (2011). A comparison of the P* transport during EN596 to the earlier work is included in Text S2.
The N 2 fixation rates measured in the incubations (Figure 1, Table S1) agree with those that would be inferred from the P* budget, if the P* convergence were evenly distributed over an area 3 × 10 12 m 2 (i.e., the northwestern gyre with a zonal extent of approximately 30°longitude and meridional extent of 10°latitude). We do not know the precise area over which the lateral P* transport converges, since our observations permit the quantification of only the cross-stream transport, which moves P* from the north side of the Gulf Stream toward its swiftly flowing center. The Gulf Stream then advects the P* further downstream, and much of it likely recirculates over a broad region in the northwest gyre. Models (Letscher et al., 2016) and observation-based estimates of the Ekman convergence (Palter et al., 2011) suggest these lateral transports may elevate productivity and/or N 2 fixation in a large area of the northwestern gyre. If so, the per-area fixation rate inferred from the biogeochemical budget would be approximately 0.1 mol m 2 year −1 , or an average volumetric rate of 10 nmol L −1 day −1 spread over the top 25 m of the water column, in line with the measured N 2 fixation rates in the incubations. If robust diazotrophy were to extend throughout a deeper layer or was heightened at the ocean's surface relative to that measured at 25 m, it would imply the P* budget could be balanced by a higher areal rate of N 2 fixation over a smaller region. In addition, it is important to recognize that the P* budget assumes a steady state over an annual cycle, whereas our incubations measure daily rates during a single week in April at just a few stations in a turbulent ocean. We would expect that such fixation would be patchy in space and time, perhaps tracking the mesoscale features that deliver P*-and Fe-rich water to the subtropics. Higher-resolution data would be needed to test whether rates we measured represent an average over this larger region.
The δ 15 N of nitrate is also frequently used to investigate the integrated rate and spatial distribution of N 2 fixation (e.g., Knapp et al., 2008) and is defined as The bottom panel shows two isopycnals, labeled with their potential density anomaly (potential density referenced to the sea surface in kg m −3 , minus 1,000). As in Thomas and Joyce (2010) Newly fixed N has a δ 15 N that is similar to that of atmospheric N 2 , approximately −2‰ to 0‰ (e.g., Carpenter et al., 1997). In contrast, nitrate δ 15 N in the North Atlantic north of ∼30°N and within/beneath the thermocline is typically above 4‰, as documented in Marconi et al. (2017) and in Figure 4c. Therefore, the light isotopic values (2‰ and 3‰) measured on the southern fringe of the Gulf Stream are suggestive of N 2 fixation, though it is unclear whether the isotopically light nitrate is added locally or is advected with the Gulf Stream from the tropics. Preliminary evidence that local N 2 fixation within the subtropical gyre is at least partly responsible for the isotopic signature of these water masses is provided by the fact they have potential densities between 1,026 and 1,027 kg m −3 (Figure 4c). Water mass analysis suggests that these density classes in the Gulf Stream are comprised almost entirely of recirculated subtropical water, with little contribution from the throughput of tropical or Southern Hemisphere sourced water in the Atlantic Meridional Overturning Circulation (Palter & Lozier, 2008;Schmitz & McCartney, 1993;Schmitz & Richardson, 1991;. The total subtropical N 2 fixation inferred from the P* budget and verified with the incubation measurements agrees with a recent analysis by Marconi et al. (2017), who estimate the N 2 fixation in the Atlantic from the divergence of the nitrate δ 15 N transport between two latitudes. Their method takes advantage of the requirement for isotope mass balance, and the assumption that the divergence in δ 15 N meridional transport is principally driven by the addition of isotopically light nitrate due to biological N 2 fixation. The difference between the estimated meridional transport of δ 15 N across 24°N and 48°N suggests that 3-5 Tg N year −1 must be fixed between these latitudes, similar to the total N 2 fixation potentially supported by the cross-Gulf Stream supply of P* (Palter et al., 2011). If these fixation rates were distributed evenly over the entire Atlantic between 24°N and 48°N and above 25 m, it would imply volumetric fixation rate of only 1 nmol L −1 day −1 , about an order of magnitude lower than the incubation results reported herein. Thus, this comparison suggests the likelihood of a region of elevated N 2 fixation in the northwestern subtropical gyre under the influence of the Gulf Stream and its recirculations.
However, we caution that the apparent agreement between the North Atlantic N 2 fixation rate estimates of Marconi et al. (2017) and those estimated by Palter et al. (2011) may be coincidental. Because δ 15 N was never measured on the zonal sections used in the divergence estimates, Marconi et al. (2017) used δ 15 N values from cross-over points between the zonal sections and a meridional section in the eastern Atlantic where δ 15 N data were available. This practice neglects the likelihood that the Gulf Stream, which is the conduit of almost all the northward volume transport at these latitudes (Rayner et al., 2011), could have a different nitrate δ 15 N concentration than in the eastern basin, even on the same isopycnals. Indeed, the average Gulf Stream δ 15 N we measured at approximately 36°N/74°W, weighted by velocity and nitrate concentration in each grid cell, is 4.2‰, which is lighter than the purported 4.9‰ of the northward transport at both 24°and 48°from the cross-over point method of Marconi et al. (2017). The inference of N 2 fixation from the δ 15 N transport divergence is extraordinarily sensitive to small differences in the δ 15 N of the northward and southward transport. Therefore, it is crucial that δ 15 N of nitrate be measured at the same time and location as the velocity field to accurately deduce the net meridional transport of the isotope and the implied N 2 fixation rate.

Conclusions
Advection and mixing across the Gulf Stream supply P* and iron to the northwest subtropical gyre (Conway et al., 2018;Palter et al., 2011). A quantification of the P* supply led to the hypothesis that there is likely elevated N 2 fixation in and near the Gulf Stream (Palter et al., 2011). Here we report the results from a cruise designed to test this hypothesis by measuring N 2 fixation and the abundance of diazotrophs in this region. The measurements made on this cruise confirmed the presence of diazotrophs and reveal rates of N 2 fixation in and near the Gulf Stream 1-3 orders of magnitude greater than those previously measured in the central and southern subtropical gyre. These N 2 fixation rates are on the order of 10 nmol L −1 day −1 , consistent with estimates of the fixation needed to balance P* supply, if that supply were consumed within a broad region in the northwestern subtropical gyre. The high resolution measurements also show sharp gradients in diazotrophic abundance that would be missed from more typical station spacing. The patchiness in Trichodesmium abundance possibly indicated that the cells accumulated along convergent structures in the surface circulation, though more sampling would be needed to confirm this speculation.
The sources and sinks of fixed nitrogen in the ocean are thought to be tightly coupled, thereby stabilizing the ocean nitrogen reservoir (Deutsch et al., 2007). One mechanism behind this hypothesized homeostasis is that denitrifying bacteria create a niche for diazotrophs by decreasing nitrogeneous nutrients to concentrations that limit non-diazotrophic competitors. However, the scarcity of iron can decouple denitrification and N 2 fixation , contributing to the mismatch between N 2 fixation inferred from a model-based P* budget and that measured in incubations and estimated from nitrogen isotopes in the tropical Pacific (Knapp et al., 2016). The separation between P* supply and consumption can have important implications for the time scale over which a perturbation to the ocean nitrogen budget might persist. For instance, Moore et al. (2009) proposed that nitrogen fixed in the North Atlantic, where there is relatively abundant iron, is added to the North Atlantic Deep Water; this fixed nitrogen helps balance nitrogen lost in the denitrifying, sub-oxic zones of the Indo-Pacific. In this paradigm, a perturbation to either the removal or inputs of nitrogen could persist over centuries to a millennium, the average time scale over which a parcel of water makes a circuit through the global-scale circulation pathways connecting these distant regions.
These considerations reveal the importance of understanding the rate and spatial distribution of Atlantic N 2 fixation and how phosphate and iron are supplied to support it. Circulation features, like the Gulf Stream, that transport water masses from distant locales into new biogeographical regions, play an important role in these nutrient cycles. Expanding our observations to quantify the mean N 2 fixation rates in and near the Gulf Stream, their spatial and temporal variability, the diazotrophic communities responsible, and the phosphate and iron sources that sustain these communities will ultimately help uncover the stability or variability of the oceans fixed nitrogen reservoir.

Data Availability Statement
Sea surface temperature and MODIS chlorophyll data were downloaded from the NASA JPL site: https:// podaac.jpl.nasa.gov. Biological and hydrographic data are archived in the Biological and Chemical Oceanography Data Management Office (BCO-DMO), http://lod.bco-dmo.org/id/dataset/774288.