We estimated the relative contribution of atmosphere (ic Nitrogen (N) input (wet and dry deposition and N fixation) to the epipelagic food web by measuring N isotopes of different functional groups of epipelagic zooplankton along 23°W (17°N-4°S) and 18°N (20-24°W) in the Eastern Tropical Atlantic. Results were related to water column observations of nutrient distribution and vertical diffusive flux as well as colony abundance of Trichodesmium obtained with an Underwater Vision Profiler (UVP5). The thickness and depth of the nitracline and phosphocline proved to be significant predictors of zooplankton stable N isotope values. Atmospheric N input was highest (61% of total N) in the strongly stratified and oligotrophic region between 3 and 7°N, which featured very high depth-integrated Trichodesmium abundance (up to 9.4×104 colonies m-2), strong thermohaline stratification and low zooplankton δ15N (~2‰). Relative atmospheric N input was lowest south of the equatorial upwelling between 3 and 5°S (27%). Values in the Guinea Dome region and north of Cape Verde ranged between 45 and 50%, respectively. The microstructure-derived estimate of the vertical diffusive N flux in the equatorial region was about one order of magnitude higher than in any other area (approximately 8 mmol m-2 d 1). At the same time, this region received considerable atmospheric N input (35% of total). In general, zooplankton δ15N and Trichodesmium abundance were closely correlated, indicating that N fixation is the major source of atmospheric N input. Although Trichodesmium is not the only N fixing organism, its abundance can be used with high confidence to estimate the relative atmospheric N input in the tropical Atlantic (r2 = 0.95). Estimates of absolute N fixation rates are two- to tenfold higher than incubation-derived rates reported for the same regions. Our approach integrates over large spatial and temporal scales and also quantifies fixed N released as dissolved inorganic and organic N. In a global analysis, it may thus help to close the gap in oceanic N budgets.
Citation: Sandel V, Kiko R, Brandt P, Dengler M, Stemmann L, Vandromme P, et al. (2015) Nitrogen Fuelling of the Pelagic Food Web of the Tropical Atlantic. PLoS ONE 10(6): e0131258. https://doi.org/10.1371/journal.pone.0131258
Editor: Arga Chandrashekar Anil, CSIR- National institute of oceanography, INDIA
Received: December 3, 2014; Accepted: May 31, 2015; Published: June 22, 2015
Copyright: © 2015 Sandel et al. This is an open access article distributed under the terms of the Creative Commons Attribution License, which permits unrestricted use, distribution, and reproduction in any medium, provided the original author and source are credited
Data Availability: The relevant raw data sets (zooplankton isotopic composition, hydrography, nutrient and UVP5 data) are uploaded to PANGAEA: CTD: http://doi.pangaea.de/10.1594/PANGAEA.834588; UVP5: http://doi.pangaea.de/10.1594/PANGAEA.842405; Nutrients: http://doi.pangaea.de/10.1594/PANGAEA.842498; Zooplankton: http://doi.pangaea.de/10.1594/PANGAEA.842500.
Funding: This work is a contribution of the German Research Foundation (DFG) supported project SFB754 (www.sfb754.de) and involved the Federal Ministry of Education and Research (BMBF) joint projects RACE (03F0651B) and SOPRAN (03F0462A, 03F0611A, 03F0662A). LS was supported by the Chair VISION from the National Center for Scientific Research (CNRS) and University Pierre and Marie Curie (UPMC).
Competing interests: The authors have declared that no competing interests exist.
Substantial uncertainties remain in oceanic nitrogen (N) budgets, especially in the tropical North Atlantic, that continue to stimulate critical reevaluation of diazotroph contribution to marine productivity [1–3]. The marine N cycle is closely coupled to the global carbon budget via primary production. The availability of several elements may limit oceanic primary production (N, P, Fe, Si, etc.) with N being typically the most important limiting nutrient on biological timescales and in large areas of the world´s oceans [4, 5]. Therefore, N availability largely determines the oceans capacity to act as a source or sink of atmospheric CO2. Regenerated N or inorganic nutrients that have been recycled in the upper ocean can support a large fraction of total primary production, but new N inputs are required to compensate N losses from surface waters [4, 6]. N losses from surface waters are mainly due to export of particulate matter by sinking and active transport via vertical migration of consumers . Pelagic microbial N loss processes require suboxic to anoxic conditions  and are therefore generally considered of minor importance in the tropical Atlantic, where water column O2 concentrations usually exceed 40 μmol kg-1 .
The principal sources providing bioavailable N in the euphotic zone in the open ocean are vertical diffusive flux due to diapycnal mixing [1, 9], biological N fixation [1, 9, 10, 11] and atmospheric deposition [12, 13]. Especially in warm, stratified, oligotrophic waters, the fixation of atmospheric N by a variety of diazotrophs (such as Trichodesmium spp., diatom-associated cyanobacteria or UCYN-A, [14, 15]) represents a major source of new N for biological production in the mixed layer. In the equatorial Atlantic region, the dominant role of vertical mixing for supplying nutrients to the euphotic zone has long been recognized (e.g. ). Both observations and models confirm elevated chlorophyll and phytoplankton concentrations to be present throughout most of the year (e.g. [17, 18]). Nevertheless, recent findings challenge the general notion that N fixation is low in upwelling regions. Sohm et al. found high rates of N fixation in or near the Benguela Upwelling System , and Subramaniam et al. reported elevated N fixation rates in the equatorial Atlantic during the upwelling period . Studies in the subtropical North Atlantic have demonstrated that depth-integrated N fixation rates by Trichodesmium can exceed the estimated vertical diffusive flux of NO3- locally [1, 9]. Nevertheless, estimates of N fixation and vertical diffusive N flux do not cover the N demand of new production in a study conducted in the subtropical Northeast Atlantic , potentially because vertical diffusive N flux, N fixation or both were underestimated or because wet and dry deposition of N were not taken into consideration when calculating the atmospheric N input. We here use a combination of a biogeochemical tracer quantifying the relative contribution of atmospheric N input and direct measurements of vertical diffusive N flux to provide estimates of the total atmospheric N input to the pelagic food web for the ETNA.
The distinct sources of nitrogen to the pelagic food web have characteristic δ15N signatures. Atmospheric N is defined to have a δ15N value of 0‰ and diazotroph N fixation produces isotopically depleted biomass with δ15N values as low as -1 to -2‰ [10, 11, 21]. Inorganic N compounds in dust have a slightly lower δ15N signal of about -3‰ , whereas deep water nitrate in the Atlantic has a δ15N signature of approximately 4.5‰ . Therefore, atmospheric N input results in a much lower biomass δ15N than biomass fuelled by nutrient rich deep water. Trophic fractionation then results in a relative increase in the heavy isotope during the transfer of N to higher trophic levels . High zooplankton δ15N values ranging approximately between 8 and 12‰ occur in (and close to) upwelling areas, where biological production is principally supported by vertical mixing and advection of nutrient rich subsurface water (e.g. in the California Current system  and in the Eastern Tropical Atlantic ), whereas low zooplankton δ15N values between 1 to 5‰ have been found when Trichodesmium as a conspicuous diazotroph was present in high abundances [10, 11, 21]. The δ15N signature can therefore be used to trace the release of atmospheric derived N into the marine food web. Similarly, the δ13C signal is a tracer of food web structure, but its global distribution (i.e. isoscape) also shows considerable latitudinal variation, with maximum values >-20‰ observed in the tropical oceans [25–27]. Particularly heavy δ13C values around -13‰ were reported in Trichodesmium [28, 29] and hence the upper margin of the δ13C isoscape may be determined by N fixation. This would require exclusive Trichodesmium grazers like Macrosetella gracilis and Miracia efferata to have a fixed δ13C signature and that δ13C and δ15N of these organisms could be used to pinpoint the isotopic baseline of C and N entering the food web via N fixation.
In this study, the differential impact of atmospheric derived versus upwelled inorganic nitrogen to the food web of the tropical Atlantic was assessed. Atmospheric derived N is here defined as the sum of wet and dry deposition via dust and rain, as well as N fixation by diazotrophs. The vertical diffusive N flux was estimated from nutrient profiles, the ocean’s stratification, and concurrently collected microstructure shear data. Stable nitrogen isotopic signatures of zooplankton were used to estimate the relative input of atmospheric N to the surface waters. Estimates of vertical diffusive N flux and relative atmospheric input were then combined to yield absolute estimates of atmospheric N input. Where possible, estimates of wet and dry deposition from literature data were used to also estimate absolute N fixation for a given area. Trichodesmium abundance was determined using an Underwater Vision Profiler 5 to test the hypothesis that Trichodesmium abundance can serve as an indicator of atmospheric N input in the tropical Atlantic and to calculate the potential N fixation rate of Trichodesmium.
Material and Methods
Sampling was conducted along a N-S transect from 15°N to 5°S at 23°W and along an E-W transect from 20 to 27°W at 18°N in the eastern tropical Atlantic (Fig 1) during R/V “Maria S. Merian” cruise MSM 22 (October 24—November 23, 2012). For sampling in the exclusive economic zone of Cape Verde, permission was granted by the Cape Verdean Ministry of Foreign Affairs. Work conducted in international waters did not require a specific permit and did not involve endangered or protected species. Satellite derived rainfall rates (NASA tropical rainfall measuring mission) and back trajectories of air masses for the region and time frame of our observations were downloaded from http://trmm.gsfc.nasa.gov/ and http://ready.arl.noaa.gov/HYSPLIT.php, respectively. Oceanographic observations were conducted using a 24-niskin bottle rosette with a Seabird SBE 11plus CTD equipped with a Dr. Haardt fluorescence probe and an Underwater Vision Profiler 5 (UVP 5, serial number 001). The fluorescence probe was cross-calibrated with regularly conducted chl-a measurements. Chl-a samples were 0.2-mm filtered (25 mm Whatman GF/F), the filters frozen at -80°C for over 5 h, extracted in 90% acetone and measured against a blank in a Turner Trilogy fluorometer calibrated with a chl-a standard dilution series. Trichodesmium distribution and abundance were quantified at 111 stations with the UVP5. This imaging tool allows in situ quantification of particles >60 μm and plankton >500 μm with high vertical resolution [30, 31]. Thumbnails of all objects > 500 μm were extracted using the ZooProcess software . Imaged Trichodesmium were identified by a computer-assisted method  and the identification validated by experts. The observed volume of each image was 0.93 L. On average 11.6 (±3.09) images were recorded per m depth and the mean sampling volume for the upper 200 m of the water column was 2.16 m3.
The inset shows the average satellite rainfall (mm h-1) along 23°W within the longitude range 22°W to 24°W for November 2012. Black lines denote the sampled transects at 23°W and 18°N.
Water column sampling was carried out at 25 stations. Samples for dissolved inorganic macronutrients (NO3- + NO2-, PO43-) were taken at eight depths within the upper 300 m (fixed depths 250, 150, 100, 80, 60, 40, 20, and 10 m), frozen at -80°C and stored at -20°C until later analysis in the home laboratory. Dissolved water column nutrients (NOx, PO4) were measured according to Grasshoff using a Quaatro autoanalyzer . Depth (Z50) and thickness (H) of nitracline and phosphocline was determined following Hauss et al. by fitting sigmoid regressions of NO3- and PO43- concentrations as a function of depth down to 150 m .
Zooplankton was collected with vertical tows of a 200 μm WP2 plankton net from 100 m to the surface and a number of widely distributed and frequently occurring species were chosen to represent four distinct trophic groups and sampled when available [34–36]. They comprised epipelagic copepods and juvenile euphausiids (Table 1). Individuals were identified, rinsed with distilled water, transferred into pre-weighed tin cups (5x9 mm, Hekatech), dried at 50°C for at least 48 hrs, weighed and prepared for elemental analysis of carbon and nitrogen amounts as well as their stable isotope ratios (δ13C and δ15N). See S1 Table for a complete summary of samples. Due to their small size, Macrosetella gracilis were collected on precombusted GF/F filters and packed into a tin capsule. Analysis was conducted as gas chromatography-combustion-isotope ratio mass spectrometry (GC/C/IRMS) at the UC Davis stable isotope facility (California, USA). Stable isotope ratios are reported with reference to a standard and expressed in parts per mil (‰) according to the formula: , where X is the respective element, H gives the heavy isotope mass of that element, and R is the ratio of the heavy to the light isotope .
A multiple linear regression model with zooplankton δ15N depending on depth of nitracline (Z50N), nitracline thickness (HN), and the difference between phosphocline and nitracline thickness (HP-HN) and all significant interactions was used to predict zooplankton δ15N for omnivore and carnivore epipelagic copepods as these groups could be sampled throughout the investigation area. Backward stepwise model simplification was used to identify significant predictors. Since the intercept of the multiple regression represents maximum δ15N under strong upwelling conditions (i.e. when vertical diffusive N flux providing 100% of N available to biological production), it was used as δ15NRef to calculate atmospheric contribution to zooplankton biomass (%NAtm). We applied the simple isotopic mixing model introduced by Montoya et al. : , where δ15NRef is a baseline reference δ15N for zooplankton exclusively consuming NO3-fuelled POM and δ15NAtm is a baseline value for atmospheric inputs (via dust deposition and diazotrophy). δ15NAtm was assumed to be -2‰, reflecting the mean isotopic signature measured for diazotrophs and dust and therefore providing estimates of atmospheric contribution to zooplankton biomass [10, 13]. To explore the relationship between Trichodesmium abundance and the contribution of atmospheric N input to zooplankton biomass, we excluded the station where an anticyclonic mode-water eddy was sampled (18°N 20°W).
Microstructure shear and temperature profiles were collected using a loosely-tethered profiler (MSS 90D-II) manufactured by Sea&Sun Technology . The profiler was equipped with two shear sensors (airfoil), a fast temperature sensor (FP07), an acceleration sensor, tilt sensors and standard CTD sensors. It was adjusted to descent at 0.5–0.6 ms-1. Two to five repeat profiles were collected following a CTD profile at each station from the surface to down to 1000 m depth (S2 Table).
High-frequency shear fluctuations measured by the airfoils were used to estimate the local dissipation rate of turbulent kinetic energy (ε). Wavenumber spectra were calculated from one-second ensembles of shear data (1024 individual measurements). ε was then determined by integrating the shear spectrum using the relationship for isotropic turbulence where μ is the dynamic viscosity of seawater, the vertical shear of horizontal velocity fluctuations, and the shear wavenumber spectrum. The lower wavenumber kmin was set to 3 cpm while the upper cutoff number kmax was varied by iteration between a maximum value of 30 cpm and a minimum value of 14 cpm when dissipation was low . The loss of variance due to incomplete integration was compensated by extrapolating the observed spectrum in the neglected wavenumber band using the theoretical Nasmyth spectrum . Similarly, loss of variance resulting from spatial averaging due to the finite size of the sensor tip was corrected following Macoun & Lueck . Turbulent eddy diffusivities (Kρ) were calculated from ε and the buoyancy frequency (N), as Kρ = ΓεN-2 . Mixing efficiency, Γ, was set to 0.2. Stratification (N2 = g/ρo∙dρ/dz–g2∙c-2, g—earth gravity, dρ/dz–vertical gradient of potential density, ρo—reference potential density, c–speed of sound) was calculated from the CTD data using the adiabatic levelling method . Overall the same procedure as described by Schafstall et al. was used . The upward flux F of NO3- + NO2- across the nitracline (Z50N) due to turbulent mixing was calculated by multiplying the eddy diffusivity with the average vertical nitrate and nitrite concentration gradient between 25 and 75% of the subsurface concentration, FNO3- + NO2- = Kρ d(NO3-+NO2-)/dz.
Absolute atmospheric N input (wet and dry deposition, as well as N fixation) was calculated for each distinct oceanographic area as (diffusive N flux / % diffusive N flux)* % atmospheric N.
Rainfall rates and air mass back trajectories
Satellite derived rainfall rates during the cruise along the 23°W section were high between 3 and 7°N and almost zero within the remaining sampling area (Fig 1). Backtracking of air masses from 1000 m height above the 23°W transect shows that stations south of 7°N were influenced by southeasterly winds stemming from the open Southeast Atlantic. Stations north of 7°N were influenced by northeasterly winds from the Sahara and Sahel region (Fig 1). Back trajectories for 0 and 500 m height were similar (data not shown).
Hydrography and water column biogeochemistry
Stations were grouped according to the following oceanographic areas: 3–5°S—oligotrophic South Atlantic (OSA), 3°S-3°N—equatorial upwelling region influenced by strong diapycnal mixing (EU), 3–7°N—oligotrophic North Atlantic (ONA), 7–15°N—Guinea Dome (GD), along 18°N—north of Cape Verde (NCV; Fig 2). South of approximately 3°S, the water column was highly stratified, but lacked superficial fresher water. Around the equator (3°S-3°N), a comparatively shallow and intense chl-a maximum and elevated vertical shear of horizontal velocity were observed due to the presence of the eastward Equatorial Undercurrent and westward South Equatorial Current (not shown). Between 3°N and 7°N, the water column was highly stratified, featuring a superficial “lens” of very low salinity and a deep chl-a maximum. In the GD region, the pycnocline was considerably shallower than in the southern portion of the transect, and the chl-a maximum was approximately as shallow and intense as in the EU. Along the 18°N transect (Fig 2, right panels), the deep chl-a maximum was generally shallower than at the 23°W transect and shoaling towards the eastern margin (Fig 2H). Within an anticyclonic mode water eddy at 19°40'W identified from shipboard ADCP data (not shown) the chl-a maximum extended to the surface.
Along 23°W, a pronounced Trichodesmium bloom around 5°N extended to a depth of about 80 m with a clear peak around 40 m (Fig 3A). Water column integrated areal abundances of up to 9.4 × 104 colonies m-2 were observed in this area. North of 10°N and in the equatorial region, abundance was lower but Trichodesmium was present in all profiles. South of 2°S only few colonies were observed, with some profiles being entirely void of Trichodesmium. Trichodesmium abundance on the 18°N transect was highly variable and the most conspicuous peak with up to 5.5 x 104 colonies m-2 was found within the anticyclonic mode-water eddy at the easternmost station at 30 to 35 m depth (Fig 3B).
The spatial distribution of macronutrients was closely related to pycnocline depth (Figs 2 and 3). Mean near-surface (10m) concentrations of dissolved inorganic N (combined NO2-/NO3-) often reached the detection limit of 0.004 μmol L-1. North of about 7°N along the 23°W transect, a shallow nitracline with minimum depth of about 20 m was observed. Between 3 and 7°N, the nitracline was generally deeper and characterized by steep vertical gradients. At the equator, the vertical gradient was less sharp compared to off-equatorial locations and nutrient depletion reached deep into the water column. Along the 18°N transect, the nitracline ascended from west to east and its vertical extension decreased concomitantly (Figs 3 and 4). Mean near-surface values of dissolved inorganic phosphate (DIP) were 0.18 ±0.045 μmol L-1 and 0.14 ±0.021 μmol L-1 for the 23°W and 18°N transect, respectively (Figs 3 and 4). Along both transects, nitracline and phosphocline depths were highly correlated (0.96 cor, p<0.0001, Pearson`s test), with the nitracline being below the phosphocline with the exception of two stations.
Stable isotopes in zooplankton
Along the 23°W transect, the zooplankton δ15N values varied widely and ranged from approximately 1.5‰ in Trichodesmium feeders to values exceeding 8‰ in carnivores at some stations. Trichodesmium grazers Macrosetella gracilis and Miracia efferata were only found at stations where Trichodesmium was present. Lowest zooplankton δ15N values were found around 5°N and increased toward north and south, with a peak at the southernmost station sampled (Fig 4C). Along the 18°N transect, δ15N values increased towards the east for all species, except in presumable Trichodesmium feeders with values as low as 2.3(±0.29)‰ throughout the entire transect (Fig 4D). At most stations, δ15N values were highest in carnivores. Lowest values were observed in Trichodesmium feeders (if present). δ15N values of omnivores and euphausiids mostly ranged in between those extremes, but no consistent order among them was evident across stations.
Zooplankton δ13C values ranged from a minimum of -23.0‰ in Trichodesmium feeders to -18.7‰ in omnivore species (Fig 5). Spatial differences in zooplankton δ13C were inversely related to δ15N values, with low values in the equatorial region and in the northern part of the 23°W transect that contrasted a peak of high values at about 5°N. At all stations present, δ13C values were lowest in Trichodesmium grazers. Among feeding types, δ15N and δ13C were negatively correlated (Fig 5). The slope of the corresponding linear regression was -0.38(±0.08) in presumable direct consumers of diazotroph biomass (M. gracilis and M. efferata), which was significantly different (ANCOVA, α = 0.05) from that in all other groups, with values of -1.9(±0.1), -1.7(±0.1) and -1.9(±0.2) for carnivores, omnivores and euphausiids, respectively.
Microstructure shear-derived estimates of diffusive NO3-flux into the mixed layer were highly variable among individual stations, ranging between 50 and 23000 μmol m-2 d-1.This variability results from sporadic occurrence of elevated turbulence in the upper thermocline due to breaking of internal waves or shear instability of the zonal equatorial currents in the case of the EU region (see S1 Fig for eddy diffusivity vertical and latitudinal distribution). Turbulent dissipation rates were elevated by up to 4 orders of magnitude during an active mixing event. Due to elevated turbulence in the upper thermocline of the EU region and despite the weak vertical dissolved inorganic N gradient (Fig 3), the average diffusive flux was with 8200 μmol m-2 d-1 about one order of magnitude higher than in all other considered regions (Table 2). Average eddy diffusivities here were between six and twentyfold higher. In all other areas, eddy diffusivities were close to the background value of 1x10-5 m2s-1 and vertical diffusive N fluxes varied between 400 and 1000 μmol m-2 d-1.
Contribution of atmospheric N input
For both omnivores and carnivores, δ15N was significantly negatively correlated with nitracline depth Z50N, nitracline thickness HN, and with the interaction HN:HP-HN (HP: phosphocline thickness, Table 3). δ15N was significantly positively correlated with HP-HN and with the interaction Z50N:HN. The remaining interactions were not significant in the model. The intercept of multiple linear regression was lower for omnivores (11.08±0.95), where the model explained 57% of the variability found, than for carnivores (12.78±1.76), where the model accounts for 51% of the variability. Overall δ15N of all zooplankton samples was significantly negatively correlated with depth-integrated Trichodesmium abundance (p<0.0001).
Relative contribution of atmospheric N input (wet and dry deposition, as well as N fixation) as calculated with δ15N of zooplankton biomass ranged from 23% in the southern part of the 23°W transect to a maximum of 71% around 5°N (Fig 4E). Atmospheric N input estimates for omnivores and carnivores where always closely related. Relative contribution of atmospheric N input was minimal south of the equator, peaked at about 5°N and slightly declined again further towards the north. Along 18°N, the significance of atmospheric N input decreased from west to east for omnivores (Fig 4F). Atmospheric N input to the marine foodweb was described as a nonlinear function of depth-integrated Trichodesmium abundance (r2 = 0.95, Fig 6).
The aim of this study was to estimate the relative contribution of new N to the pelagic food web by evaluating the vertical diffusive nitrate flux and the atmospheric input (the sum of wet and dry deposition, as well as N fixation). This was achieved by combining observed parameters varying on very different spatial and temporal scales. Measurements of the vertical distribution of macronutrients and the turbulence in the thermocline were combined to calculate associated vertical diffusive nitrate fluxes . The isotopic composition of mesozooplankton body tissue was used to estimate relative atmospheric N [10, 11, 24]. At the smallest spatial and temporal scales of the parameter space, microstructure profiles represent a momentary snapshot of the ocean . At the largest scales, isotopic signatures of different zooplankton ecotypes represent time-integrating in situ tracers of the biological processing of organic compounds in the water column . Hence, the large variations in zooplankton nitrogen isotopes we found along the 23°W and 18°N transects indicate major changes in the source of nutrients. Observations of Trichodesmium abundance support this notion. Combining these methods to quantify the sources of nitrogen to the pelagic food web has the potential to provide a holistic view of global nitrogen budgets [10, 11].
The location of the deep chl-a maximum and the nutriclines mostly coincides with the pycnocline and is located between 50 and 90 m depth. Shallowest locations of these features are within the EU and GD region, albeit with large differences in vertical mixing due to different local physical processes: the diffusive nutrient flux is more than an order of magnitude higher in the equatorial upwelling region compared to the open ocean upwelling region, i.e. the GD. The shallow pycnocline in the GD is related to the positive wind stress curl in this region causing upward velocities due to Ekman divergence and a shoaling of the thermocline. At the equator, dynamic instabilities within the current system (eastward Equatorial Undercurrent and westward South Equatorial current) superimposed on the mean vertical shear flow enhance turbulence generation and thus diapycnal mixing in the upper thermocline (e.g. [44, 45]), resulting in a low vertical nutrient gradient, but a comparatively deep nutricline. Finally, the shoaling of the pycnocline in the mode water eddy is a consequence of the geostrophic adjustment due to the rotational velocities within the eddy. The nitracline was generally deeper and thinner than the phosphocline, suggesting that non-diazotroph primary production in the euphotic zone was first-of-all limited by N rather than P.
It is generally acknowledged that spatial differences in diffusive N flux are largely due to the high variability in eddy diffusivity rather than NO3-gradients . The average Kρ values determined in this study, despite being a single realization in time and space, are consistent with previously reported eddy diffusivities in the respective regions. In the upper thermocline away from the equator, our Kρ estimates are in the range of 0.7-2x10-5 m2s-1. This is in agreement with reported Kρ estimates for ONA and GD region at similar longitudes of 1x10-5 m2s-1 from multi-cruise microstructure observations [46, 47] as well as from a deliberate tracer release experiment . At the equator, previously reported average Kρ estimates vary on seasonal and annual time scales . However, the overall magnitude of our Kρ measurements again compares well to results from previous sampling campaigns: Hummels et al. reported Kρ estimates 2-3x10-4m2s-1 for the equatorial region at 23°W from microstructure data collected during two November cruises , while Fernández-Castro et al. reported average Kρ of 1.5x10-4 m2s-1 from a December cruise at the equator at about 30°W .
In general, diffusive N fluxes determined from observations strongly vary regionally and seasonally. With the exception of the EU region, our values from the tropical Atlantic range between 500 and 1400 μmol m-2 d-1. These fluxes are comparable to estimates (1040 μmol m-2 d-1) determined from observations at 18°N in the eastern tropical Atlantic during the winter season  and fall in the same range as values estimated by Capone et al. (46 to 736 μmol m-2 d-1), who used fixed Kρ values of 1.1 and 3.7*10-5 m2 s-1 . Similar diffusive N fluxes determined from measurements (35 to 1250 μmol m-2 d-1) were also reported from a N-S transect in the Atlantic conducted in April/May 2008 . In that study, however, N fluxes determined at the equator were very low and in the subtropical Atlantic very high. It should be noted that these values were estimated in the mixed layer and their relevance is unclear. Much lower diffusive N fluxes ranging from 35–85 μmol m-2 d-1 were reported by Painter et al. in the subtropical North Atlantic . On the other hand, strongly elevated diffusive N fluxes have been reported from upwelling and coastal regions. Schafstall et al. observed diffusive N fluxes exceeding 10000 μmol m-2 d-1 in the Mauritanian upwelling , accounting for a substantial part of the primary production in that region. These numbers are comparable to our results of 8200 μmol m-2 d-1 determined from the observations within the EU.
The spatial variability in NO3 flux does not fully explain the spatial variation in Trichodesmium abundance, indicating that not only the lack of NO3 supply defines the ecological niche of Trichodesmium, but other environmental constrains such as sea surface temperature, iron and P availability . Likewise, the spatial variability in zooplankton δ15N of specific trophic levels is not solely linked to NO3 flux, providing evidence that major differences in the N source used for biological production occur, with regions either dominated by upwelled or atmospheric N input.
Natural abundance δ15N tracer techniques were used previously to estimate atmospheric N input into the pelagic food web. δ15N in surface NO3 proved difficult to measure at low concentrations . Filtered seston  or size-fractionated zooplankton [10, 11, 21] represent a mixture of trophic levels and functional groups, hampering the estimation of an NRef value. Our approach to use individual zooplankton species to estimate the atmospheric N input to the marine foodweb at a defined trophic level and, thus, fixed NRef seems more straightforward. Furthermore, previous studies used the highest δ15N measurements of their respective cruises or reported in that area as a benchmark for δ15NRef to estimate relative contribution of upwelled N to total N input into the foodweb [10, 11, 21]. The occurrence of Trichodesmium in the OSA (albeit in low numbers and not in all profiles) indicates that even in this region N fixation occurred. Above described approach was therefore not applicable for us and we defined δ15NRef as the regression intercept of the stepwise multiple linear regression model following the approach of Hauss et al. . This gives a theoretical maximum δ15N with upwelled nitrate as the only N source. While the cited authors [10, 11, 21] reported δ15NRef from 5.6‰ to 7.5‰, our results were perceivably higher with 11.1‰ and 12.8‰ for omnivores and carnivores, respectively. These values also coincide with those observed in upwelling regions [23, 24]. Using the model estimates, we found minimal atmospheric contribution (dry and wet deposition, as well as N fixation) south of the equator, still accounting for approximately 25 to 30% of secondary production. Thus, estimates of atmospheric input would be lower by this amount if the observed δ15N at the southernmost station would represent δ15NRef. Atmospheric contribution was with >60% highest in the ONA region. Resulting absolute atmospheric input rates could be estimated at 0.4 mmol N m-2 d-1 in this region. Diazotroph organisms and atmospheric nutrient sources–in particular by wet deposition —are therefore of particular importance in fuelling the pelagic food web in the oceanic desert of the ONA. Atmospheric contribution of N in the EU region was 30 to 40%, but due to the high diffusive N flux, this translates to 4 mmol N m-2 d-1 atmospheric input. From a biogeochemical point of view, the EU therefore represents a substantial source of atmospheric N in the Atlantic Ocean, probably due to the supply of iron and dissolved organic phosphorous via upwelling that is thought to foster N fixation .
Mirroring the spatial pattern of δ15N, zooplankton δ13C was low in the equatorial upwelling area and high in stratified oligotrophic waters around 5°N. At first glance, it seems therefore to represent a similarly meaningful proxy for N fixation. A global pattern of latitudinal gradients in organic δ13C, with a maximum of approximately -18‰ in the tropics and a minimum close to -30‰ in the southern Ocean, has been observed in particulate organic matter , zooplankton  and cephalopods  and was attributed to algal growth rates and water temperature . The upper end member of this range could very well be defined by N fixation. This would, however, require the identification of a baseline signal. The δ13C signature of Trichodesmium ranges between approximately -15 and -12‰ [28, 29] and is higher than that of any other phytoplankton species described. Our data suggest that this signature is not fixed, because the range in δ13C in exclusive Trichodesmium-grazers is as wide as that of other zooplankton (-23 to -19‰) compared to a very narrow δ15N range. δ13C can therefore not serve as another tracer to estimate N fixation and the reason for the variability of δ13C in the observation region remains enigmatic.
The spatial and temporal association between wet and dry deposition and diazotroph blooms due to iron fertilization complicates the distinction of these two N sources (e.g. ), and the δ15N approach fails to discern between biologically fixed nitrogen and nitrate derived from wet and dry deposition. In the majority of studies, the contribution of atmospheric deposition relative to biological N fixation rates in the ETNA was considered to be small (e.g. [11, 21, 24]). Since direct soluble N input would potentially inhibit diazotroph competitiveness, and reported values of leachable total nitrogen (LTN) flux are comparatively low , the importance of Saharan dust pulses rather seems to lie in fertilization of diazotroph growth by iron and phosphorous supply . Although the diversity of N fixing organisms in the region is high [3, 14, 15], high abundance of Trichodesmium likely indicate favorable conditions for N fixing organisms in general. δ15N in omni- and carnivore zooplankton were strongly negatively correlated with Trichodesmium abundance. Only at the eastern margin of the 18°N transect, high Trichodesmium abundance observed in the core of an anticyclonic mode-water eddy (see e.g.  for a characterization of such eddies) was related to a high zooplankton δ15N. This structure featured a comparatively shallow nutricline, high primary productivity as indicated by satellite ocean color, high particle load and a shallow O2 minimum (data not shown). It thus seemed to represent an aging core of oceanic productivity. Diazotrophs such as Trichodesmium apparently take advantage of residual macro- and micronutrients when non-diazotrophs begin to be nutrient limited in the upper mixed layer. As zooplankton have a lower tissue turnover and generation time than phytoplankton , a recent change in the N source from upwelling to N fixation in this transient feature might be reflected in the high Trichodesmium abundance, but not yet in an altered zooplankton δ15N. The otherwise tight correlation of Trichodesmium abundance with omni- and carnivore zooplankton δ15N results in a significant relationship of Trichodesmium abundance and atmospheric N input and also indicates that N fixation is probably the major atmospheric source of N. However, Trichodesmium abundance can never explain all of the atmospheric contribution (even if used as a proxy for all N fixing organisms), as this would translate into an observation of zero dust input and occurrence of Trichodesmium at the same time. These conditions are not expected to co-occur due to iron limitation, resulting in an asymptotic behavior of the curve describing the Trichodesmium impact on the total atmospheric contribution.
Comparing to literature values of wet and dry deposition, absolute N fixation can be estimated. HYSPLIT back trajectories show that OSA and EU are influenced by air masses from the Southeast, which are nearly aerosol free , while the areas ONA, GD and NCV receive dust-laden air from continental West Africa. TRMM rainfall estimates indicate that ONA receives considerable amounts of wet deposition, whereas nearly no rainfall occurred in the other observation regions. It has recently been suggested that the intertropical convergence zone (ITCZ) forms a “biogeochemical divide” of the subtropical Atlantic, washing atmospheric dust (the major dissolved iron source) into micronutrient-depleted surface waters . This is in line with our observations along the 23°W transect, where the high rainfall rates and a superficial lens of fresher water coincided with a massive Trichodesmium bloom at 5°N and the highest relative contribution of >60% atmospheric N to zooplankton biomass. It is unlikely that terrigenous material from the western part of the basin (Amazon River Plume) contributes quantitatively to the low δ15N values at the ONA region, given that only a small fraction of the plume is bound eastward , that the impact of riverine input is very small compared to precipitation at this longitude (, their Fig 7), and that we directly observed both the Trichodesmium bloom and the highest rainfall at 5°N. This led us to the conclusion that the wet deposition in the ITCZ is the main driver of the isotope pattern. In consequence, the observation regions can be divided into three categories: i) OSA and EU with little to no wet and dry deposition during the observation time frame ii) ONA with strong wet deposition that can not be constrained by literature estimates and iii) GD and NCV with very little to no wet deposition during the observation time frame and an almost constant dry deposition that can be constrained by literature estimates. If we assume dust deposition in the OSA and EU areas to be close to zero, the atmospheric contribution of 27 to 35% in these areas would be entirely due to biological N fixation. Very high N fixation rates of up to 4 mmol m-2 d-1 result for the EU region. While some authors consider the atmospheric contribution, and in particular diazotrophy in the equatorial upwelling to be quite low [10, 11], a similar observation has been made by Mouriño-Carballido et al. . Subramaniam et al. even found that N fixation rates around the equator were 2 to 7 times higher during an upwelling event than during non-upwelling conditions and conclude that upwelled waters rich in phosphate and iron promote diazotrophy . As the ONA mainly receives wet deposition, of which the LTN flux is not known, the absolute N fixation input cannot be estimated for this region. At Cape Verde, little variation in LTN flux over the year was reported . If we assume the reported mean annual value of 32.6 μmol m-2 d-1 LTN for the GD and NCV regions, N fixation would account for 350 and 950 μmol m-2 d-1 (Table 2), respectively. In the GD region at 23°W during the same season, a mean of 194 μmol m-2 d-1 was measured in incubation experiments .
If we assume a colony-specific N fixation rate by Trichodesmium of 4.4 nmol d-1 (which is the mean of the incubation studies summarized in ,their Table 7), mean watercolumn N fixation by Trichodesmium based upon UVP5 colony counts would be 1.8, 6.7, 188.6, 21.7 and 69.1 μmol m-2 d-1 in OSA, EU, ONA, GD and NCV, respectively. Differences to our estimates of N fixation for OSA, EU, GD and NCV are likely due to the fact that Trichodesmium represents only a fraction of the diazotroph community [14,15]. Furthermore, we could only quantify colonies >500 μm in the water column. Smaller colonies and single trichomes, as well as accumulations of Trichodesmium at the surface  could not be quantified. The observed very close relation of atmospheric N input and Trichodesmium abundance indicates that UVP5 estimates of Trichodesmium abundance can serve as a very good indicator for the general existence of a niche for diazotrophs, but these abundance estimates cannot be used to quantify the total N fixation in an area.
In general, our indirect estimates of N fixation rates are two- to tenfold higher than rates measured with incubation methods in the same regions [3, 20, 49]. Part of this large discrepancy may be due to uncertainties in direct N fixation rate measurements, as handling of N fixing organisms during shipboard incubations may disrupt N fixation capacity. On the other hand, recent findings suggest bioavailable N compounds in commercial 15N2 gas used for incubations bias the measurement . Furthermore, current techniques only estimate the amount of N fixed in the particulate matter fraction that is obtained after filtration of the incubation volume. Fixed N that is directly released again as dissolved organic or inorganic N during the incubation time is currently not measured in N fixation incubation experiments, but in the open ocean will ultimately be transferred into the pelagic food web via the bacterial loop and therefore needs to be considered. Our approach is incubation independent, integrates over large spatial and temporal scales and also quantifies fixed N lost via exudation and therefore might provide a more realistic overall estimates of N fixation for the OSA, EU, GD and NCV regions. In a global analysis, it may help to close the gap in oceanic N budgets [2, 3] as it yields N fixation estimates that are about 10-fold higher than incubation techniques in some regions. Nevertheless, a direct comparison of incubation and the here used tracer technique would benefit our understanding of the oceanic N cycle.
S1 Fig. Eddy diffusivity (Kρ) section along 23°W (Panel A) and grouped by region (Panel B; colored crosses denote individual profiles, corresponding colored horizontal lines maximum NOx gradient and black line region mean).
Mixed layer data are omitted from plots.
S1 Table. Species and number of individuals sampled at each station (three replicates each).
We thank the crew of RV “Maria S. Merian” for their support during the cruise, Alice Nauendorf and Jannik Faustmann for help with the validation of UVP5 image analysis, Bente Gardeler for nutrient measurements and two anonymous reviewers for insightful comments that helped to improve the paper.
Conceived and designed the experiments: HH RK VS US. Performed the experiments: HH RK VS PB MD. Analyzed the data: HH RK VS MD LS PV. Contributed reagents/materials/analysis tools: HH LS US RK PV. Wrote the paper: VS HH RK US PV LS PB MD.
- 1. Capone DG, Burns JA, Montoya JP, Subramaniam A, Mahaffey AC, Gunderson T, et al. (2005) Nitrogen fixation by Trichodesmium spp.: An important source of new nitrogen to the tropical and subtropical North Atlantic Ocean. Global Biogeochem Cycles 19,
- 2. Codispoti LA (2007) An oceanic fixed nitrogen sink exceeding 400 Tg N a-1 vs the concept of homeostasis in the fixed-nitrogen inventory. Biogeosciences 4: 233–253
- 3. Großkopf T, Mohr W, Baustian T, Schunck H, Gill D, Kuypers MMM, et al. (2012) Doubling of marine dinitrogen fixation rates based on direct measurements. Nature 488: 361–364, pmid:22878720
- 4. Gruber N, Sarmiento JL (1997) Global patterns of marine nitrogen fixation and denitrification. Global Biogeochem Cycles 11: 235–266.
- 5. Agusti S, Duarte CM, Vaqué D, Hein M, Gasol JM, Vidal M (2001) Food-web structure and elemental (C, N and P) fluxes in the eastern tropical North Atlantic. Deep-Sea Res II 48(10): 2295–2321.
- 6. Fernández A, Grana R, Mouriño-Carballido B, Bode A, Varela M, Domínguez-Yanes JF, et al. (2012) Community N2 fixation and Trichodesmium spp. abundance along longitudinal gradients in the eastern subtropical North Atlantic. ICES J Mar Sci 70: 223–231,
- 7. Francis CA, Beman JM, Kuypers MMM (2007) New processes and players in the nitrogen cycle: the microbial ecology of anaerobic and archaeal ammonia oxidation. ISME J. 1: 19–27, pmid:18043610
- 8. Brandt P, Banyte D, Dengler M, Didwischus S-H, Fischer T, Greatbatch RJ, et al. (2015) On the role of circulation and mixing in the ventilation of oxygen minimum zones with a focus on the eastern tropical North Atlantic. Biogeosciences 12: 489–512.
- 9. Painter SC, Patey MD, Forryan A, Torres‐Valdes S (2013) Evaluating the balance between vertical diffusive nitrate supply and nitrogen fixation with reference to nitrate uptake in the eastern subtropical North Atlantic Ocean. J Geophys Res: Oceans 118: 5732–5749.
- 10. Montoya JP, Edward JC, Douglas GC (2002) Nitrogen fixation and nitrogen isotope abundances in zooplankton of the oligotrophic North Atlantic. Limnol Oceanogr 47: 1617–1628
- 11. Landrum JP, Altabet MA, Montoya JP (2011) Basin-scale distributions of stable nitrogen isotopes in the subtropical North Atlantic Ocean: Contribution of diazotroph nitrogen to particulate organic matter and mesozooplankton. Deep-Sea Res I 58: 615–625
- 12. Spokes LJ, Yeatman SG, Cornell SE, Jickells TD (2000) Nitrogen deposition to the eastern Atlantic Ocean. The importance of south‐easterly flow. Tellus B 52: 37–49.
- 13. Baker AR, Weston K, Kelly SD, Voss M, Streu P, Cape JN (2007) Dry and wet deposition of nutrients from the tropical Atlantic atmosphere: Links to primary productivity and nitrogen fixation. Deep-Sea Res I 54: 1704–1720.
- 14. Langlois RJ, LaRoche J, Raab PA (2005) Diazotrophic diversity and distribution in the tropical and subtropical Atlantic Ocean. Appl Environ Microb 71: 7910–7919. pmid:16332767
- 15. Foster RA, Subramaniam A, Zehr JP (2009) Distribution and activity of diazotrophs in the Eastern Equatorial Atlantic. Env Microb 11: 741–750.
- 16. Voituriez B, Herbland A (1979) The use of the salinity maximum of the Equatorial Undercurrent for estimating nutrient enrichment and primary production in the Gulf of Guinea. Deep-Sea Res I 26: 77–83.
- 17. Christian JR, Murtugudde R (2003) Tropical Atlantic variability in a coupled physical–biogeochemical ocean model. Deep-Sea Res II 50: 2947–2969.
- 18. Wang X, Murtugudde R, Hackert E, Marañón E (2013) Phytoplankton carbon and chlorophyll distributions in the equatorial Pacific and Atlantic: a basin-scale comparative study. J Mar Sys 109: 138–148.
- 19. Sohm JA, Hilton JA, Noble AE, Zehr JP, Saito MA, Webb EA (2011) Nitrogen fixation in the South Atlantic Gyre and the Benguela upwelling system. Geophys Res Lett 38(16).
- 20. Subramaniam A, Mahaffey C, Johns W, Mahowald N (2013) Equatorial upwelling enhances nitrogen fixation in the Atlantic Ocean. Geophys Res Lett 40:1766–1771,
- 21. Mompeán C, Bode A, Benítez-Barrios VM, Domínguez-Yanes JF, Escánez J, Fraile-Nuez E (2013) Spatial patterns of plankton biomass and stable isotopes reflect the influence of the nitrogen-fixer Trichodesmium along the subtropical North Atlantic. J Plankton Res 35: 513–525
- 22. Fry B (2006): Stable isotope ecology. Springer. 308 pp.
- 23. Rau G, Ohman MD, Pierrot-Bult A (2003) Linking nitrogen dynamics to climate variability off central California: a 51 year record based on 15N/14N in CalCOFI zooplankton. Deep-Sea Res II 50: 2431–2447.
- 24. Hauss H, Franz JMS, Hansen T, Struck U, Sommer U (2013) Relative inputs of upwelled and atmospheric nitrogen to the eastern tropical North Atlantic food web: Spatial distribution of δ15N in mesozooplankton and relation to dissolved nutrient dynamics. Deep-Sea Res I 75: 135–145,
- 25. Rau G, Sweeney R, Kaplan I (1982) Plankton 13C/12C ratio changes with latitude: differences between northern and southern oceans. Deep-Sea Res I 29: 1035–1039.
- 26. Takai N, Onaka S, Ikeda Y, Yatsu A, Kidokoro H, Sakamoto W (2000) Geographical variations in carbon and nitrogen stable isotope ratios in squid. J Mar Biol Assoc UK 80: 675–684
- 27. McMahon KW, Hamady LL, Thorrold SR (2013) A review of ecogeochemistry approaches to estimating movements of marine animals. Limnol Oceanogr 58:697–714,
- 28. Carpenter EJ, Harvey HR, Fry B, Capone DG (1997) Biogeochemical tracers of the marine cyanobacterium Trichodesmium. Deep-Sea Res I 44: 27–38.
- 29. Holl CM, Villareal TA, Payne CD, Clayton TD, Hart C, Montoya JP (2007) Trichodesmium in the western Gulf of Mexico: 15N-fixation and natural abundance stable isotope evidence. Limnol Oceanogr 52: 2249–2259.
- 30. Picheral MG, Stemmann L, Karl DM, Iddaoud G, Gorsky G (2010) The Underwater Vision Profiler 5: An advanced instrument for high spatial resolution studies of particle size spectra and zooplankton. Limnol Oceanogr Methods 8: 462–473.
- 31. Stemmann L, Boss E (2012) Plankton and particle size and packaging: From determining optical properties to driving the biological pump. In: Carlson , C.A.G.S.J. (Ed.), Ann Rev Mar Sci 4: 263–290. pmid:22457976
- 32. Gorsky G, Ohman MD, Picheral M, Gasparini S, Stemmann L, Romagnan JB, et al. (2010) Digital zooplankton image analysis using the ZooScan integrated system. J Plankton Res 32(3): 285–303.
- 33. Grasshoff K, Kremling K, Ehrhardt M (1999) Methods of seawater analysis. 3rd ed. Wiley. 160pp.
- 34. Arashkevich YG (1969). The food and feeding of copepods in the northwestern Pacific. Oceanology 9(61): 5–709.
- 35. Eberl R, Carpenter EJ (2007) Association of the copepod Macrosetella gracilis with the cyanobacterium Trichodesmium spp. in the North Pacific Gyre. Mar Ecol Prog Ser 333: 205–212.
- 36. Berkes F (1975) Some aspects of feeding mechanisms of euphausiid crustaceans. Crustaceana 29(3): 266–270.
- 37. Prandke H, Stips A (1998) Test measurements with an operational microstructure-turbulence profiler: Detection limit of dissipation rates. Aquat Science 60: 191–209.
- 38. Schafstall J, Dengler M, Brandt P, Bange H (2010) Tidal induced mixing and diapycnal nutrient fluxes in the Mauritanian upwelling region. J Geophys Res 115: C10 pmid:20463844
- 39. Wolk F, Yamazaki H, Seuront L, Lueck RG (2002) A new free-fall profiler for measuring biophysical microstructure. J Atmos Ocean Tech 19 (5): 780–793.
- 40. Macoun P, Lueck R (2004) Modeling the Spatial Response of the Airfoil Shear Probe Using Different Sized Probes. J Atmos Ocean Tech 21 (2): 284–297.
- 41. Osborn NS (1980) Determination of the local rate of diffusion from dissipation measurements, J Phys Oceanogr 10: 83–89.
- 42. Fofonoff NP. (1985) Physical properties of seawater: a new salinity scale and equation of state for seawater. J Geophys Res:Oceans 90: 3332–3342
- 43. Moum JN, Rippeth TP (2009) Do observations adequately resolve the natural variability of oceanic turbulence? J Mar Sys 77(4): 409–417.
- 44. Hummels R, Dengler M, Bourlès B (2013) Seasonal and regional variability of upper ocean diapycnal heat flux in the Atlantic cold tongue. Progr Oceanogr 111: 52–74.
- 45. Hummels R, Dengler M, Brandt P, Schlundt M (2014) Diapycnal heat flux and mixed layer heat budget within the Atlantic Cold Tongue. Climate Dynamics 43: 3179–3199.
- 46. Fischer T, Banyte D, Brandt P, Dengler M, Krahmann G, Tanhua T, et al. (2013) Diapycnal oxygen supply to the tropical North Atlantic oxygen minimum zone. Biogeosciences 10: 5079–5093.
- 47. Fernández-Castro B, Mouriño-Carballido B, Benítez-Barrios V, Chouciño P, Fraile-Nuez E, Graña R, et al. (2014) Microstructure turbulence and diffusivity parameterization in the tropical and subtropical Atlantic, Pacific and Indian Oceans during the Malaspina 2010 expedition. Deep-Sea Res I 94: 15–30.
- 48. Banyte D, Tanhua T, Visbeck M, Wallace DW, Karstensen J, Krahman G, et al. (2012) Diapycnal diffusivity at the upper boundary of the tropical North Atlantic oxygen minimum zone. J Geophys Res: Oceans (1978–2012) 117.
- 49. Mouriño-Carballido B, Graña R, Fernández A, Bode A, Varela M, Domínguez J, et al. (2011) Importance of N2 fixation vs. nitrate diffusion along a latitudinal transect in the Atlantic Ocean. Limnol Oceanogr 56: 999–1007
- 50. Ryabenko E, Kock A, Bange HW, Altabet MA, Wallace DW (2012) Contrasting biogeochemistry of nitrogen in the Atlantic and Pacific Oxygen Minimum Zones. Biogeosciences 9: 203–215.
- 51. Schlosser C, Klar JK, Wake BD, Snow JT, Honey DJ, Woodward EMS, et al. (2014) Seasonal ITCZ migration dynamically controls the location of the (sub) tropical Atlantic biogeochemical divide. PNAS 111: 1438–1442, pmid:24367112
- 52. Laws EA, Popp BN, Bidigare RR, Kennicutt MC, Macko SA (1995) Dependence of phytoplankton carbon isotopic composition on growth rate and [CO2aq]: Theoretical considerations and experimental results. Geochim Cosmochim Ac 59: 1131–1138.
- 53. Mingkwan P (2011) The influence of atmospheric organic carbon and organic nitrogen on biogeochemistry of the (sub-) tropical North Atlantic Ocean. PhD Thesis, Univ of Southampton. Available: http://eprints.soton.ac.uk/338872/. Accessed 25 November 2014.
- 54. Karstensen J, Fiedler B, Schütte F, Brandt P, Körtzinger A, Fischer G, et al. (2015) Open ocean dead-zone in the tropical North Atlantic Ocean. Biogeosciences 12: 2579–2605.
- 55. Jennings S, Warr KJ (2003) Environmental correlates of large-scale spatial variation in the δ15N of marine animals. Mar Biol 142: 1131–1140,
- 56. Coles VJ, Brooks MT, Hopkins J, Stukel MR, Yager PL, Hood RR (2013) The pathways and properties of the Amazon River Plume in the tropical North Atlantic Ocean. J Geophys Res: Oceans 118: 6894–6913.
- 57. Mulholland MR, Bernhardt PW, Heil CA, Bronk DA, O'Neil JM (2006) Nitrogen fixation and release of fixed nitrogen by Trichodesmium spp. in the Gulf of Mexico. Limnol Oceanogr 51: 1762–1776.
- 58. Villareal T, Carpenter E (2003) Buoyancy regulation and the potential for vertical migration in the oceanic cyanobacterium Trichodesmium. Microb Ecol 45: 1–10. pmid:12481233
- 59. Dabundo R, Lehmann MF, Treibergs L, Tobias CR, Altabet MA, Moisander PH, et al. (2014) The Contamination of Commercial 15N2 Gas Stocks with 15N–Labeled Nitrate and Ammonium and Consequences for Nitrogen Fixation Measurements. PloS one 9: e110335. pmid:25329300