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 Gulf Stream provides a ‘nutrient stream,’ an advective flux of nutrients carried in sub-surface waters, redistributing nutrients from the tropics to the mid latitudes. There is a dramatic downstream strengthening in the full depth, volume and nitrate transport diagnosed from synoptic measurements along three sections: 32 Sv and 300 kmol s−1 at 27°N, increasing to 66.7 Sv and 940 kmol s−1 at 35.5°N, and 149.5 Sv and 2100 kmol s−1 at 36.5°N; the transport estimates have uncertainties reaching 10% of their values. The transport-weighted nitrate concentration carried by the Gulf Stream generally decreases downstream in light layers (σθ < 26.8), but increases in denser layers between 35.5°N and 36.5°N. The fraction of nutrients from regeneration within the upper thermocline slightly decreases downstream between 35.5°N and 36.5°N. Hence, the downstream variations in the nitrate concentrations carried by the Gulf Stream are probably due to lateral exchange along isopycnals, rather than diapycnal transfer or biological consumption and regeneration. The Gulf Stream also transports dissolved organic nitrogen (DON) northward within lighter upper waters, providing up to 26% of the total nitrogen transport. An accompanying model study reveals coherent nitrate and DON streams, including significant lateral exchange, running from the tropics along the western boundary of the subtropical gyre and following the separated jet along the inter-gyre boundary. The combined nitrate and DON transport along light surfaces (σθ < 26.8) remains within the subtropical gyre, while the larger nitrate transport along denser surfaces passes into the subpolar gyre, sustaining high-latitude productivity.
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 Gulf Stream, evident as a plume of warm water, provides an important poleward heat transport as part of the overturning and gyre circulation of the North Atlantic. What is more surprising is how the same Gulf Stream without a clear signature in surface nutrient concentrations is associated with a poleward transport of nutrients. Pelegri and Csanady  first set out a view of the Gulf Stream providing a ‘nutrient stream’ and a conduit of nutrients to higher latitudes in the North Atlantic. This nutrient transport is achieved by a sub-surface maximum in the nutrient flux, defined by the product of geostrophic velocity and nutrient concentration, centered at a depth of 500 m. The elevated concentration of nutrients within the Gulf Stream primarily originates from outside the subtropical gyre, passing from the tropical Atlantic [Reid, 1994; Palter and Lozier, 2008] and probably ultimately from Sub-Antarctic Mode Water and Antarctic Intermediate Water formed in the Southern Ocean [Sarmiento et al., 2004; Williams et al., 2006].
 Given the canonical picture of the nutrient stream provided by Pelegri and Csanady , we wish to assess how robust this view is, since the original analysis did not include direct velocity measurements and relied on combining nutrient measurements and the geostrophic shear relative to 2000 m. In our study, we compare measurements of the volume and nutrient transport of the Gulf Stream; in the Florida Strait at 27°N, and off Cape Hatteras at 35.5°N and 36.5°N. The downstream variation in nutrient transport is controlled by the variations in the volume transport and the transport-weighted nutrient concentration. In addition, we include the transport of dissolved organic nitrogen (DON); this DON contribution is potentially important, since DON provides the dominant pool of available nitrogen in surface waters of the oligotrophic subtropical gyres, as recently mapped by Torres-Valdés et al. .
 The paper is set out in the following manner: firstly, the details of the measurement programme and volume transports are provided and, secondly, the accompanying transports of inorganic and organic nutrients are estimated and discussed; and finally, an eddy-permitting circulation model is used to illustrate the larger-scale pathways of the nitrate and DON streams, and provide a basin-scale perspective.
2. Volume Transport Estimates
 In order to estimate the downstream changes in volume and property transports in the Gulf Stream, three sections are analyzed (Figure 1): two crossings of the boundary current, at the separation from the coast at 35.5°N and downstream at 36.5°N in May 2005 [McDonagh et al., 2007], as well as a crossing of the Florida Strait at 27°N in May 2001.
2.1. Velocity and Transport off Cape Hatteras
 At the two sections off Cape Hatteras, 35.5°N and 36.5°N, velocity profiles were determined by combining thermal-wind estimates from hydrography with a barotropic component derived from direct velocity estimates from a Lowered-ADCP [McDonagh et al., 2007]. Small uniform velocity and transport offsets are applied, −0.27 cm s−1 and −0.54 Sv across 35.5°N and −0.33 cm s−1 and −0.66 Sv across 36.5°N, to ensure a zero net volume transport across the entire width of the Atlantic basin (see McDonagh et al.  for details of the measurements for the entire section across the basin).
 The surface velocities of the Gulf Stream reach 1.57 m s−1 at 74.13°W, 35.5°N (between stations 20 and 21), where the water column reaches depths of 3000 m (Figure 1a), then weakens slightly downstream reaching 1.47 m s−1 at 70.06°W, 36.5°N (between stations 5 and 6), where the water column reaches depths of 4300 m. The northward shear in the Gulf Stream is associated with a westward shoaling of the main thermocline; σθ surfaces between 26.5 and 27.8 shoal onshore by several hundred meters over the continental slope (Figure 2, contours).
 At separation, the Gulf Stream partly overlies the deep western boundary current (Figure 1a) with the bottom flow directed southward, reaching 20 cm s−1 (at station 20). Further north at 36.5°N, the Gulf Stream has separated from the coast and is offshore of the deep western boundary current; the surface and bottom currents are then aligned together, both directed northeastward, with the bottom current typically 20% of the strength of the surface flow.
 The Gulf Stream transport is estimated by integrating the normal velocity with cross-sectional area of the section, based upon the geostrophic velocity (including the barotropic component from the bottom tracking ADCP). In order to focus on the Gulf Stream contribution, we sum over the stations where there is both a northward/northeastward transport above 2000 m and over the full depth, which include stations 20 to 27 along 35.5°N and stations 7 to 34 along 36.5°N (Figure 1a). The Gulf Stream transport over the upper 2000 m reaches 66.7 Sv at the separation point at 35.5°N and then increases downstream reaching 112.1 Sv at 36.5°N.
2.1.1. Comparison of Transport Estimates
 Our transport estimates over the upper 2000 m are broadly consistent with other diagnosed values (Figures 1a and 1b): Richardson and Knauss  report 63 Sv based on three repeat crossings in July 1967; further south, Z. Szuts (personal communication, 2007) reports a transport of 65 Sv based upon an absolute velocity profiler survey in 1992 [Sanford et al., 1996]; and between our sections, Halkin and Rossby  estimate 87.8 ± 17.3 Sv based upon sixteen repeat synoptic velocity sections from Pegasus between 1980 and 1983. Our estimate at 36.5°N is though higher than the nearby estimate of 99 Sv over the upper 2000 m from a single mooring at 68°W from Johns et al. .
 If instead the Gulf Stream transport is defined by the northward moving water over the full water column, then the transport estimates increase to 82.7 Sv at 35.5°N and to 149.5 Sv at 36.5°N (Table 1). These full depth estimates do not include the opposing southward transport in the deep western boundary current, which reaches −33.6 Sv at 37°N and −22.8 Sv at 35.5°N (Figure 1a, black arrows), and the southward transport in the slope waters to the west of station 7 along 36.5°N. Our estimate of the full depth transport is though higher than the full depth estimate of 113 ± 8 Sv using a streamwise co-ordinate and a 26 month average of a single mooring at 68°W by Johns et al.  and is only comparable to the estimate of 147 Sv at 60°W by Hogg .
Table 1. Diagnostics of Volume Transport (Sv), Nitrate Transport (kmol s−1) and Transport-Weighted Nitrate Concentration (μmol kg−1) in Different Density Classes for the Gulf Stream at 27°N, 35.5° and 36.5°Na
Uncertainties in the transports are typically 10%.
σθ < 26.2
26.2 ≤ σθ < 26.5
26.5 ≤ σθ < 26.8
26.8 ≤ σθ < 27.1
27.1 ≤ σθ < 27.3
27.3 ≤ σθ < 27.5
27.5 ≤ σθ
σθ < 27.1
σθ < 27.5
2.1.2. Uncertainty in the Transport Estimates
 Our estimates of the Gulf Stream transport are based upon synoptic measurements and should be viewed as a snapshot, rather than a long-term average. To provide a lower bound estimate of the uncertainty of the Gulf Stream, a volume budget is conduced for a closed box formed west of 69°W by the two crossings of the Gulf Stream in Figure 1a, using the previous repeat of the section in 1981 to extend the northern section to the coast. The volume transports into this box are out of balance by 7.5 Sv. We assume that this mismatch represents a lower bound of the uncertainty in the full depth transport and assign an uncertainty for the full depth Gulf Stream transport of 149.5 Sv at 36.5°N of twice this imbalance of 15 Sv. Hence, we estimate the uncertainty as 10% of the magnitude of the Gulf Stream transport, which is likewise applied to the other transport calculations.
2.2. Velocity and Transport at Florida Strait
 In the Florida Strait at 27°N, the velocity profile is calculated from the hydrography measured at 8 stations (Figure 1c), using the horizontal density gradient via thermal wind and the slope of the sea surface via geostrophy, which leads to a northward transport of 31.5 Sv. In order to match independent cable measurements of the depth-integrated transport of 32.06 Sv for May 2001 [Rousset and Beal, 2010], a small increase to the barotropic transport is made of 0.56 Sv. The northward flow reaches 1.5 m s−1 at the surface and more than 0.5 m s−1 at a depth of 500 m between 79.8°W and 79.3°W (Figure 2a).
2.3. Downstream Transport Changes of the Gulf Stream
 There is marked downstream increase in the transport of the Gulf Stream: over the upper 2000 m, the transport increases from 32.06 Sv at Florida Strait to 66.7 Sv at 35.5°N and 112.1 Sv at 36.5°N. Between Florida Strait and Cape Hatteras, the Gulf Stream transport increases by a factor of 2, and between 35.5°N and 36.5°N by a further factor of 1.7 over a distance of only 350 km. The Gulf Stream growth of 45.4 Sv between 35.5°N and 36.5°N gives a downstream accumulation rate of 13.0 Sv per 100 km. In comparison, Halkin and Rossby  calculated a comparable, though slightly larger, downstream accumulation rate of the Gulf Stream over the upper 2000 m of 15.4 Sv per 100 km.
 This inflation of the Gulf Stream transport is even more marked over the full water column, increasing from the same value at Florida Strait, to 82.7 Sv at the separation point at 35.5°N and then further to 149.5 Sv for the separated current at 36.5°N (Table 1). Between Florida Strait and Cape Hatteras, the full depth Gulf Stream transport increases now by a factor of 2.6, and between 35.5°N and 36.5°N by a further factor of 1.8. This transport variation of the Gulf Stream is due to the presence of two re-circulating gyres flanking the separated jet, including tight subtropical and subpolar recirculations, as described by Hogg . These recirculations are revealed in barotropic flows extending to the seafloor, rather than baroclinic changes in the density structure across the separated jet. Hence, the recirculating gyres lead to the Gulf Stream transport increasing after separation and then eventually decreasing in strength in the basin interior.
 While the transport estimates over the upper 2000 m are more comparable with other published values, we prefer to focus on the estimates of the full depth transport due to the importance of the deep flows in transporting nutrients.
3. Nutrient Transports
 For each of the sections crossing the Gulf Stream, we now assess how the nutrient transport varies downstream from the combination of the velocity and nutrient measurements.
3.1. Inorganic and Organic Nutrient Measurements
 For the 27°N section, dissolved inorganic nitrogen (NO3− + NO2−, here after NO3−) and dissolved inorganic phosphorus (PO43−) were analyzed in the laboratory using a TRAACs autoanalyzer. Samples were stored frozen (−20°C) until analysis could be performed. The detection limit for both nitrate and phosphate was 0.01 mmol m−3. Total dissolved nitrogen (TDN) was determined using a home built, high temperature oxidation system utilizing chemiluminescence detection [Hansell and Carlson, 2001]. Dissolved organic nitrogen was calculated by subtracting the inorganic nitrogen from the total nitrogen, DON = TDN − (NO3− + NO2−). A total of 113 samples were collected from 8 stations along the transect with depths ranging from the surface to near bottom, with the maximum depth sampled of 740 m.
 For the 35.5°N and 36.5°N sections, nitrate and phosphate were analyzed onboard using a Skalar autoanalyzer with limits of detection of 0.15 and 0.03 mmol m−3 for nitrate and phosphate. TDN was determined using a UV method with a detection limit of 0.45 mmol m−3 [Torres-Valdés et al., 2009]. DON was estimated as above and 12 samples of organic nutrients were collected at each station throughout the water column down to the sea floor or depths of 5000 m.
3.2. Nutrient Transfer
 The signature of the Gulf Stream in transferring nutrients is assessed in terms of three related quantities: (i) the nutrient flux from the product of the horizontal velocity, density and nutrient concentration, ρvN, in units of mol s−1 m−2; (ii) the nutrient transport from the cross-sectional area integrated nutrient flux, ∫ ρvNdxdz, in units of kmol s−1 and (iii) the transport-weighted nutrient concentration, ∫ ρvNdxdz/∫ ρvdxdz, in units of μmol kg−1; x is taken to be the distance along the sections, ρ is the density, v is the velocity normal to the sections, and N is a generalized nutrient.
3.2.1. Nitrate Flux
 In all three sections, there is a consistent pattern of nitrate-depleted light waters overlying nitrate-rich denser waters, which broadly follow density surfaces (Figure 2b). The nutrient-depleted waters are less than 10 μmol kg−1 in σθ layers lighter than 26.5, while there is a band of particularly nitrate-rich waters, greater than 20 μmol kg−1, between σθ = 27 and 27.5 along 35.5°N and 36.5°N. The nutrient surfaces broadly follow the density surfaces and shoal westward across each section.
 As highlighted by Pelegri and Csanady , the Gulf Stream has a sub-surface maximum in the nitrate flux, ρvNO3−, reaching 20 mmol s−1 m−2 at a depth range of 300 m to 500 m below the surface (Figure 3a); while the velocity of the boundary current is a maximum at the surface, the surface concentrations of inorganic nutrients are often low or exhausted over the subtropical gyre. The nitrate flux is strongest in the northern section where the Gulf Stream has separated from the coast and strengthened through the presence of the neighboring recirculating gyres.
 The nitrate stream is also clearly revealed by the nitrate flux plotted versus σθ surfaces, rather than versus depth. There are fluxes greater than 10 mmol s−1 m−2 along σθ surfaces between 26.5 to 27.5 along the path of the Gulf Stream (Figure 3b), but no clear signal of a northward nitrate transport on σθ surfaces lighter than 25.8 or denser than 27.7.
3.2.2. Nitrate Transport
 The nitrate transport, defined by the area-integrated nitrate flux over all layers, increases from 303 kmol s−1 at 27°N to 940 kmol s−1 at 35.5°N and 2059 kmol s−1 at 36.5°N (Table 1), representing downstream increases by factors of 3 and 6.8 relative to the Florida Strait respectively. This downstream increase in nitrate transport is primarily controlled by the inflation in the volume transport, which increases by factors of 2.5 and 4.7 respectively.
 Given that the nitrate transport increases at a slightly greater rate downstream than the volume transport, there are also more subtle changes in the nitrate concentration carried in the Gulf Stream. The transport-weighted nitrate concentration is diluted downstream for all layers between 27°N and 35.5°N, and for layers lighter than σθ = 26.8 between 35.5°N and 36.5°N. Conversely, in denser layers, there is an opposing downstream increase in the transport-weighted nitrate concentration between 35.5°N and 36.5°N. Hence, the greater rate by which the nitrate transport increases downstream, relative to the volume transport, is achieved through the increasing nitrate concentration carried in the denser, deep waters.
3.2.3. Phosphate and Excess Nitrate Transfer
 The phosphate distribution and phosphate fluxes broadly vary in a similar manner to nitrate: there are phosphate-rich deep waters and a sub-surface phosphate flux linked to the boundary current, resembling Figure 3. The phosphate transport again increases downstream from the inflation of the volume flux and the transport-weighted phosphate concentration decreases downstream for all layers (Table 2). This downstream decrease in phosphate concentration differs with the downstream increase in nitrate concentration in σθ layers denser than 26.8 between 35.5°N and 36.5°N.
Table 2. Diagnostics of Phosphate Transport (kmol s−1), Together With Transport-Weighted Phosphate (μmol kg−1) and Excess Nitrate, N* = NO3− − 16PO43− (μmol kg−1), in Different Density Classes for the Gulf Stream at 27°N, 35.5°N and 36.5°N
 A more discerning measure of how the nitrate and phosphate fluxes depart from each other is given by the excess nitrate, N* = [NO3−] − 16[PO43−]. There are generally relatively-enriched nitrate concentrations and positive N* within lighter waters, σθ ≤ 27.1, within the Gulf Stream and offshore in the subtropical gyre. Conversely, there are relatively depleted nitrate concentrations and negative N* onshore of the Gulf Stream, in shelf waters and in the subpolar gyre, as well as in denser waters.
 Following this distribution of N*, the northward flux of N* has a characteristic dipole pattern: a northward flux of positive N* associated with the Gulf Stream for σθ ≤ 27.1 at 27°N and 35.5°N, and for σθ ≤ 27.3 at 36.5°N, together with a northward flux of negative N* in denser waters (Table 2).
3.2.4. DON Transfer
 DON has a generally opposing vertical structure to that of nitrate. There are elevated concentrations of DON in the lighter, near surface waters over the upper 150 m and reduced concentrations in denser, deep waters (Figure 4a). There are also relatively low DON concentrations along σθ = 26.5 to 27 along 35.5°N and along σθ = 27.2 to 27.3 at 36.5°N.
 Given the general surface intensification of DON and the increase in velocity towards the surface, the flux of DON is a maximum at the surface and coincident with the path of the Gulf Stream (Figure 4b). The DON flux reaches 6 to 8 mmol m−2 s−1, and is confined within typically the upper 300 m and σθ surfaces lighter than 26.5 at 35.5°N and 36.5°N.
 The northward DON transport, integrated over all layers, increases downstream from 107 kmol s−1 at 27°N, to 289 kmol s−1 at 35.5°N and 563 kmol s−1 at 36.5°N (Table 3). This downstream increase in DON transport occurs at a slightly greater rate than that of the volume transport, reflecting a downstream increase in transport-weighted DON for most layers (Table 3); for example, within the σθ = 26.2 to 26.5 layer, the transport-weighted DON increases from 2.79 to 3.45 and 4.0 μmol kg−1 at 27°N, 35.5°N and 36.5°N respectively. Hence, relatively DON-rich waters are progressively entrained into the downstream Gulf Stream.
Table 3. Diagnostics of DON Transport (kmol s−1) and Transport-Weighted DON Concentration (μmol kg−1), and the Relative Contribution of DON Transport to the Total Nitrogen Transport in Different Density Classes for the Gulf Stream at 27°N, 35.5°N, and 36.5°N
DON Transport ∫ ρvDONdxdz (kmol s−1)
Transport-Weighted DON ∫ ρvDONdxdz/∫ ρvdxdz (μmol kg−1)
DON Transport/Total Nitrogen Transport
σθ < 26.2
26.2 ≤ σθ < 26.5
26.5 ≤ σθ < 26.8
26.8 ≤ σθ < 27.1
27.1 ≤ σθ < 27.3
27.3 ≤ σθ < 27.5
27.5 ≤ σθ
σθ < 27.1
σθ < 27.5
3.2.5. Total Nitrogen Transfer
 The northward transport of total nitrogen is provided from the sum of the nitrate and DON transport; the DON transport accounting for 26%, 23% and 21% of the total nitrogen transport at 27°N 35.5°N and 36.5°N respectively (Table 3).
 Perhaps more importantly, the distribution of the DON and nitrate transport contributions differ. The DON transport contributes more than 30% of the total nitrogen transport within the uppermost waters, occurring within individual layers lighter than σθ = 26.2 at 27°N, lighter than 26.5 at 35.5°N, and lighter than 26.8 at 36.5°N (Table 3).
3.3. How Are Downstream Variations in Nutrient Transport Along the Gulf Stream Controlled?
 The downstream increase in nutrient transport is primarily controlled by the volume flux: when the Gulf Stream strengthens, then the corresponding nutrient fluxes increase. However, there are more subtle variations in the nutrient transport due to changes in the nutrient concentrations carried in the Gulf Stream: a downstream weakening in the transport-weighted nitrate in light layers and a downstream strengthening in dense layers between 35.5°N and 36.5°N.
 There are several processes that might play a role:
 1. Diapycnal mixing is expected to provide an upward flux of nutrients, transferring nutrients from the dense waters to the light waters. However, this signal from diapycnal mixing is the exact opposite of what is seen here along the Gulf Stream, instead a downstream weakening in nutrient concentrations occurs in the light, upper waters. While the strong velocity shears in the Gulf Stream imply enhanced diapycnal mixing, the effect of this process is either masked by other processes or there is a more complicated structure to the diapycnal mixing in this region.
 2. Biological consumption and regeneration might play a role, reducing inorganic nutrient concentrations within the euphotic zone and increasing inorganic nutrient concentrations at depth through the respiration of organic matter. On timescales of several days to a week, nitrate can be effectively consumed by phytoplankton. In comparison, surface waters might take from 3 to 4 days to cover the 350 km between 35.5°N and 36.5°N, assuming surface velocities from typically 1 to 1.5 m s−1, so it is unclear whether there is sufficient time for the biological cycling to play a dominant role.
 3. The downstream variations in transport-weighted nutrient concentration might reflect the effect of lateral exchange along isopycnals: a dilution in the lighter, upper waters through relatively nutrient-poor waters being added and an enrichment in the denser, deeper waters through a supply of nutrient-rich waters.
 To identify whether biological consumption and regeneration or lateral exchange is occurring, we consider the nitrate distributions along 35.5°N and 36.5°N in more detail. Nitrate is separated into preformed and regenerated components [Redfield et al., 1963],
and the fraction of nitrate from regeneration is defined by the relative magnitude of the regenerated and total nitrate concentrations in a water parcel below the euphotic zone, NO3reg/NO3−.
 The regenerated nitrate concentration is estimated from the apparent oxygen utilization, AOU, defined as the difference between the saturated and measured oxygen concentrations, NO3reg = −AOU/R, where the Redfield ratio R is taken as N : O2 = 16 : −170 [Takahashi et al., 1985].
 The regenerated nitrate has high values between the σθ = 27 and 27.5 surfaces along both sections (Figure 5a), while elsewhere there are relatively low values. At 35.5°N, the fraction of nitrate from regeneration is relatively large in the upper waters with a volume-weighted average value of 0.56 between the σθ = 26 and 27.5 surfaces (Figure 5b, bottom). Thus, more than half of the nitrate supplied in the upper waters has been provided through biological regeneration at 35.5°N, although this biological supply is likely to have occurred upstream rather than locally due to the strong flows. However, further north at 36.5°N, the fraction of nitrate from regeneration is slightly reduced with a volume-weighted average value of 0.49 between the σθ = 26 and 27.5 surfaces (Figure 5b, top). Hence, there is a slight downstream weakening in the fraction of nitrate from regeneration in the upper waters along the Gulf Stream.
 Accordingly, the downstream changes in the transport-weighted concentration of nitrate are probably due to the lateral transfer of waters along the Gulf Stream: an influx of nitrate-depleted upper waters and nitrate-enriched deeper waters are required. Whether this transport-dominated view holds is now explored using an isopycnal model study.
4. A Model Study of the Nutrient Stream
 The presence and pathways of the nutrient streams in the Atlantic are now investigated in a process manner using an eddy-permitting coupled physics and nitrogen model integrated for several decades [Roussenov et al., 2006; Torres-Valdés et al., 2009].
4.1. Model Formulation and Closures
 The model study employs an isopycnic model (MICOM 2.7) [Bleck and Smith, 1990]: there are 15 isopycnal layers in the vertical plus a surface mixed layer with variable density and a horizontal resolution of 0.23° on a Mercator grid. The model domain extends from 35°S to 65°N and from 98.5°W to 19°E with the topography taken from ETOP05 [General Bathymetric Chart of the Oceans, 2003] averaged within the model grid. At the northern and southern boundaries, sponge layers are incorporated below the mixed layer. Outside the mixed layer and these sponge layers, the only diapycnal mixing is achieved via a mixing coefficient, κ = 10−7 m2 s−2/N, varying with buoyancy frequency N, which typically gives 4 × 10−5 m2 s−1 for N ∼ 2.5 × 10−3 s−1 within the main thermocline. In addition, the isopycnic model employs isopycnic mixing of tracers and thickness diffusion (for details, see Torres-Valdés et al. ).
 The dynamical model is initialized from Levitus climatology [National Oceanographic Data Center, 1998] and integrated for 60 years forced by NCEP monthly-mean winds and surface fluxes in a repeating climatological mean year mode. Rivers are added as freshwater fluxes at corresponding coastal grid points. From year 60, the dynamical model is coupled with the nitrate/phosphate model with a set-up similar to Roussenov et al.  and Williams et al. , and integrated for another 40 years.
 The physical model is coupled to a simplified nitrogen model including nitrate and dissolved organic nitrogen (DON) pools, consisting of semi-labile and refractory components, and particulate organic matter (PON). The organic matter in the model is only formed over the euphotic zone, which has a constant thickness of 100 m. The PON is assumed to fallout and be remineralized in the interior with a vertical remineralization scale of 200 m [Jenkins, 1998]. The dissolved inorganic, labile and refractory organic nitrogen are transported by the circulation with life times in the euphotic zone of typically 6 months for the semi-labile component and 6–12 years for the refractory component, depending on the amount of solar radiation.
 The rate of consumption of the inorganic nutrients follows Michaelis-Menten kinetics with a dependence on sunlight and the inorganic nutrient concentrations [Roussenov et al., 2006]. The biological consumption of inorganic nutrients is assumed to form a combination of PON and DON, in a 50:50 ratio for nutrient-rich waters (NO3− > 5 μmol l−1) and 35:65 ratio in nutrient-poor waters. The formation of DON is further split into semi-labile and refractory components.
 The initial nitrate is taken from climatology [Conkright et al., 1994], while the semi-labile DON is initialized to be 0 and the refractory DON initialized as 1 μmol l−1. Along the southern closed boundary, nutrients are continuously relaxed to the initial condition within the buffer zones, and there are no atmospheric or riverine inputs of nutrients, nor any loss of nutrients on the seafloor through burial.
 The resulting model distributions of inorganic and organic nutrients, export production and the effects of organic nutrients in sustaining export are reported by Torres-Valdés et al. . Our focus here is how the isopycnal fluxes of nitrate and DON vary within the vicinity of the western boundary and extend over the rest of the basin.
4.2. Modeled Nutrient Stream Pathways
 The network of western boundary currents and separated jets within the model lead to nutrient streams extending over the domain, evident as regions of enhanced advective fluxes of nitrate and DON.
4.2.1. Close to the Western Boundary
 Along the western boundary at 35.5°N, the model has northward velocities reaching 1.5 m s−1 at the surface and weakening to 0.1 m s−1 or less at depths of 500 m or more (Figure 6a). This range of velocities occur along σθ surfaces lighter than σθ = 27.0.
 The modeled northward nitrate flux reaches 10 mmol s−1 m−2 along σθ surfaces between 26.8 and 27.3, which corresponds to a depth range between 300 m to 1000 m (Figure 6b). In contrast, the northward DON flux is a maximum at the surface, reaching 8 mmol s−1 m−2, and is carried along σθ surfaces generally lighter than σθ = 26.8 and depths less than 500 m.
 This model structure for the nutrient stream is broadly similar to the observed signals at 35.5°N (Figures 2 and 3), there are comparable surface velocities and similar distribution to the nutrient fluxes, although the model deeper velocities are weaker than seen in the data and lead to weaker, advective nitrate fluxes at depth.
4.2.2. Downstream Variation in Transport-Weighted Nitrate Concentration
 The nitrate stream is now examined along the western boundary, separated into the nitrate flux carried in light layers, σθ < 26.8, and denser layers covering the core of the nitrate stream, 26.8 < σθ < 27.5. In the data, there is an opposing downstream variation in the transport-weighted nitrate within these two density ranges, (Table 1): a downstream dilution in light layers and an opposing increase in concentration in dense layers between 35.5°N and 36.5°N. Now consider the corresponding downstream variations in the modeled nitrate stream.
 The nitrate stream follows the current filaments making up the western boundary current (Figure 7): at 25°N, there are two northward-flowing currents, converging together at 29°N, passing along the western boundary through 35.5°N and 36.5°N. The modeled nitrate fluxes have a broadly similar pattern along the western boundary in both density layers with higher values in the more nitrate-rich, denser layers (Figures 7a and 7b); there is though a greater westward influx of nutrients into the western boundary from the gyre interior between 25°N and 27°N in the denser layers.
 Along the light layer in Figure 7a, the transport-weighted nitrate carried by the modeled Gulf Stream is 1.49 μmol kg−1 at 27°N, made up of two current branches carrying 1.09 and 1.83 μmol kg−1 to the west and east of 78.5°W respectively. This transport-weighted nitrate then becomes diluted to 1.324 and 1.318 μmol kg−1 at 35.5°N and 36.5°N respectively. The high value of the transport-weighted nitrate at 27°N is due to a combined westward influx from the gyre interior, supplying 2.1 μmol kg−1, and northward transfer from the tropics, supplying 1.01 and 1.74 μmol kg−1 in the western and eastern current branches. The downstream decrease in transport-weighted nitrate from 27°N to 35.5°N is due to an eastward flux of nutrients from the western boundary into the gyre interior (carrying a transport-weighted nitrate of 2.6 μmol kg−1).
 Along the dense layers in Figure 7b, the transport-weighted nitrate is 23.94 μmol kg−1 at 27°N, made up of the two current branches carrying 21.9 and 24.4 μmol kg−1 to the west and east of 78.5°W respectively, which is very similar to the value of 24.0 μmol kg−1 at 35.5°N. The transport-weighted nitrate concentration then increases to 24.2 μmol kg−1 at 36.5°N, which is due to a westward influx of nitrate-rich waters of 24.8 μmol kg−1 from the subtropical gyre into the boundary current.
 The biological cycling of nutrients can potentially also contribute to the downstream changes in transport-weighted nitrate concentration. In the western region between 27°N and 35.5°N, the gain in nitrate over the euphotic zone is negative, −3.48 kmol s−1, due to biological consumption over the euphotic zone. Over this same region, there is a gain of nitrate due to regeneration of nitrate from PON and DON of 3.40 kmol s−1 for σθ < 26.8 and a deeper regeneration from PON of 0.08 kmol s−1. Further to the north in the western region between 35.5°N and 36.5°N, the gain in nitrate over the euphotic zone is −0.5 kmol s−1 and the regeneration of nitrate from PON and DON is 0.41 kmol s−1 for σθ < 26.8 and a deeper regeneration of nitrate from PON is 0.03 kmol s−1. Hence, the biological transfer of nutrients are much weaker than the lateral flux of nutrient and their horizontal convergences, which can reach several tens of kmol s−1. For example, there is a westward flux of 17.8 kmol s−1 in the light layers and 68.8 kmol s−1 in the dense layers passing from the gyre interior into the western boundary region along 72°W between 35.5°N and 37°N.
 In summary, the modeled transport-weighted nitrate varies downstream in a similar sense to that seen in the data (Table 1), although the magnitude of the modeled changes are much less striking: a slight downstream dilution in the upper layers and strengthening in the deeper layers. There are much weaker contributions from the local biological cycling.
4.2.3. Nutrient Pathways Over the Basin
 The larger-scale pathways of the nutrient stream over the basin are now considered. The nitrate stream is defined in terms of the magnitude of the nitrate transport, integrated either over the cross-sectional area bound by the light layers, σθ < 26.8, or from dense layers, 26.8 < σθ < 27.5, over each 0.23° grid element. Over the lighter layers, there are elevated nitrate transports directed along isopycnals either along coherent zonal bands in the tropics or coincident with the western boundary current, reaching magnitudes of typically 10 kmol s−1 (Figure 8a). For this light layer, the nitrate stream is confined to the subtropical gyre, although part of this nitrate transport directed along the layer passes across the end of winter outcrop into denser waters within the mixed layer (Figure 8a, grey shading).
 Over the denser layer, the nitrate stream is revealed as a continuous region of enhanced nitrate transport directed along isopycnals, reaching 60 kmol s−1 for each grid element, running along the western boundary of the Atlantic, separating along the northern flank of the northern subtropical gyre and circuiting the northern subpolar gyre (Figure 8b). There are also coherent bands of elevated nitrate transport connecting the western boundary and the western side of the subtropics and tropics. The nitrate stream is particularly tightly compressed along the western boundary from 20°S to 15°N, follows several branches between 15°N and 27°N, and then again follows the western boundary until the current separates from the coast just south of 40°N. After separation, the nitrate stream extends into the interior of the basin, circuiting the eastern and northern sides of the subpolar gyre, before returning southward in the westward boundary of the subpolar gyre.
 The volume transport integrated across the basin is directed northward within these light and dense layers, reaching 10 Sv in the light layers and 6 Sv in the dense layers, and represents part of the upper cell of the overturning (Figure 8, right, blue line). The accompanying nitrate transport integrated across the basin is likewise directed northward and reaches 100 and 200 kmol s−1 within the light and dense layers respectively (Figure 8, right, black line).
4.2.4. DON Stream
 The DON stream is concentrated along light layers, σθ < 26.5, much lighter than the core of the nitrate stream within σθ layers between 26.8 and 27.3. There is a coherent DON transport running along the western boundary from 20°S to the separation of the Gulf Stream, just before 40°N, and then extends eastward following the separated jet along the inter-gyre boundary (Figure 9). The DON transport reaches a magnitude of 35 kmol s−1 for each grid element along the core of the boundary current, more than three times larger than the nitrate transport along the same layer, but typically half as strong as the nitrate transport in the denser waters (Figures 8a and 8b). The bulk of the DON stream remains within the subtropical gyre, rather than extend into the subpolar gyre. Again, there are zonal bands of elevated DON transport on the western side of the tropics and subtropics, suggesting two way exchange of DON between the basin interior and western boundary current.
 The DON transport integrated across the basin within the light σθ layers is directed northward: increasing from 20 kmol s−1 at 20°S to nearly 50 kmol s−1 in the tropics, decreasing to less than 10 kmol s−1 north of 50°N (Figure 9, right). The DON transport is again mainly confined within the tropics and subtropical gyres. The DON transport is typically half as strong as the corresponding nitrate transport (Figures 8, right, and 9, right), the nitrate transport being more important in the deeper, thicker layers.
 In summary, the model reveals coherent streams of nitrate and DON, being transferred by the western boundary currents and separated jets over the basin. The nitrate and DON streams differ in that the nitrate stream is concentrated in denser layers, so the core of the nitrate stream preferentially passes into the subpolar gyre, while the DON stream is concentrated in lighter layers and preferentially remains within the subtropical gyre.
 As first highlighted by Pelegri and Csanady , the Gulf Stream leads to an accompanying ‘nutrient stream’, a region of a high nutrient flux, ρvN, directed along isopycnals, where v is the velocity along the boundary current, N is the nutrient concentration and ρ is density. Our observational study reveals the nutrient transport of the Gulf Stream strengthening downstream from Florida Strait to off Cape Hatteras. This downstream variation is primarily achieved by a downstream inflation in the volume transport together with a slight dilution in the nutrients carried in the upper waters and an increase in nutrient concentration in the deeper waters. The fraction of nitrate from regeneration, NO3reg/NO3−, weakens downstream from 0.56 along 35.5°N to 0.49 along 36.5°N for the upper thermocline waters (σθ < 27.5). Hence, the downstream variation in nitrate concentrations carried by the current is probably due to lateral exchange along density surfaces, possibly together with a smaller opposing contribution from diapycnal transfer, rather than reflecting the effects of biological cycling.
 The core of the nutrient stream lies between σθ = 26.5 and 27.3 surfaces, so that the bulk of the nutrient stream is directed along isopycnals outcropping in the subpolar gyre [Pelegri et al., 1996]. Nutrients are advected along isopycnals and transferred from the thermocline into the downstream winter mixed layer through the reverse of the subduction process (Figure 10) [Williams et al., 2006]. The nutrient stream then acts to maintain nitrate concentrations over the downstream mixed layer and sustain biological productivity over the subpolar gyre. There is a much smaller nitrate transfer along σθ surfaces remaining within the subtropical gyre [Jenkins and Doney, 2003], which can help sustain the relatively low nutrient concentrations in the subtropical thermocline [Palter et al., 2005].
 The Gulf Stream also leads to a northward transport of DON providing 21% to 26% of the total nitrogen transport. The DON stream is carried in lighter layers remaining within the tropics and subtropics. This advective transfer leads to DON becoming the dominant pool of available surface nitrogen in surface waters over much of the subtropical gyre, as recently mapped over the Atlantic by Torres-Valdés et al. . The transfer of semi-labile DON, together with any accompanying dissolved organic phosphorus, helps sustain biological productivity over the subtropical gyre, particularly on its eastern side close to productive upwelling zones [Roussenov et al., 2006; Torres-Valdés et al., 2009].
 In summary, nutrient streams are evident along the western boundary of the North Atlantic subtropical gyre. They provide advective pathways of inorganic and dissolved organic nutrients, primarily transferring inorganic nutrients to the subpolar gyre and DON to the oligotrophic waters of the subtropical gyre. This transfer ultimately maintains nutrient concentrations within the downstream winter mixed layer and helps sustain biological productivity at higher latitudes.
 The work was supported by UK NERC grants, NER/O/S/2003/00625 and NE/D011108/1, and US NSF OCE 0752972. We thank the scientists, crew and technical staff who collected the data used here. We are grateful to M. Follows for assistance in the early stages of this work and K. Lancaster for drafting Figure 10, as well as two anonymous reviewers for constructive feedback that strengthened the manuscript.