Standard hydrographic measurements of temperature, salinity, and chlorophyll (Chl) fluorescence were made using a Seabird SBE 911 conductivity-temperature-depth (CTD) unit combined with a fluorometer mounted to a rosette sampler equipped with 24 10 L Teflon-coated Niskin bottles during a cruise from August 7 to 17, 2016, aboard the R/V Hugh R. Sharp. The cruise was comprised of eight cross-shelf transects (41 stations in total) that spanned a region between 33.5 and 38° N latitude at approximately 0.5° intervals. Stations were occupied in coastal, continental shelf, and oceanic waters in the MAB and SAB (Figures 1A,B). Station depths (Figures 1C,D) ranged from 10–3,450 m with most stations located on the continental shelf (<200 m) and in the Gulf Stream. Stations were also occupied in waters influenced by the shelf-break jet (200–1,000 m) and in the Slope Sea (>1,000 m). At each station, water for the determination of chemical and biological properties was collected at various depths using Niskin bottles (see below).

Nutrient samples at each depth were collected directly from Niskin bottles through a 0.2 μm capsule filter (Pall Supor^®) using a peristaltic pump at pressure ≤5 mm Hg. Filtrate was collected directly into two 50 mL sterile polypropylene centrifuge tubes (Falcon^®) to measure NO₃⁻, nitrite (NO₂⁻), orthophosphate (simplified as PO₄, sum of HPO₄²⁻ and PO₄³⁻), and urea concentrations. Filtrate was also pumped into three 2 mL sterile polypropylene tubes for cyanate determinations, two 15 mL OPA-treated polypropylene tubes (Falcon^®) for NH₄⁺ analyses, and two 40 mL combusted amber glass vials for analysis of total dissolved free primary amine. To minimize contamination from filtration and between samples, filters and tubing were rinsed thoroughly with site water before sample collection. NO₃⁻ + NO₂⁻, NO₂⁻, PO₄, and urea samples (duplicates) were stored at 4°C until analysis within 48 h of their collection using a nutrient autoanalyzer (Astoria-Pacific, Inc., United States) according to the manufacturer’s specifications. The method detection limits for NO₃⁻ + NO₂⁻, NO₂⁻, PO₄, and urea were 0.14 μmol L⁻¹, 0.07 μmol L⁻¹, 0.03 μmol L⁻¹, and 0.08 μmol L⁻¹, respectively. NH₄⁺ samples (duplicates) were kept at 4°C until analysis within 24 h of collection. The concentrations of NH₄⁺ were measured onboard using the OPA-fluorescence method of Holmes et al. (1999) with a spectrofluorometer. The method detection limit was 10.0 nmol L⁻¹. Cyanate samples (triplicates) were stored in liquid nitrogen aboard the ship and at -80°C once samples were returned to the land-based laboratory. Cyanate concentrations were measured by high-performance liquid chromatography (HPLC) using a precolumn fluorescence derivatization method (Widner et al., 2013; Widner and Mulholland, 2017). The method detection limit was 0.4 nmol L⁻¹ (Widner et al., 2013). Concentrations of total dissolved free primary amine were measured using the method of Aminot and Kérouel (2006) to estimate ambient dissolved free amino acids (DFAA) concentrations, because the two measurements are generally in agreement (Kirchman et al., 1989). The method detection limit was 4 nmol L⁻¹.

Whole water samples were collected from three target depths at each station: near surface, a depth above the Chl maximum (with majority of them near the bottom of the mixed layer), and at the depth of the Chl maximum. Water from Niskin bottles was drained into individual 10 L carboys and then transported to the ship-board laboratory, where the water samples in each carboy were mixed and sub-samples (0.2–1.4 L) were collected onto pre-combusted GF-75 filters (Whatman^®, nominal pore size 0.3 μm) in triplicate, for analysis of particulate nitrogen (PN) and carbon (PC) concentrations and the natural abundance of ¹⁵N and ¹³C. The filters were stored in cryovials and immediately frozen and stored in a freezer at -20°C until analysis. Prior to their analysis, filters were dried at 40°C, pelletized in tin capsules, and analyzed on a Europa 20/20 isotope ratio mass spectrometer equipped with an automated N and C analyzer. The average detection limit of the mass spectrometer was 0.0018 and 0.0005 atom% for ¹⁵N and ¹³C, respectively; these values were derived based on three times the standard deviation (3 × SD) of the atom% of 12.5 μg N and 100 μg C standards analyzed with each sample run (40 in total).

Parallel sets of triplicate whole water (0.5–2 L) at the same three target depths mentioned above from 27 stations (Figure 1) were dispensed from the 10 L carboys into acid-cleaned polyethylene terephthalate glycol incubation bottles (Nalgene^™). Nitrogen uptake incubations were initiated by amending incubation bottles with highly enriched (98–99%) ¹⁵N-labeled substrates (Cambridge Isotope Laboratories, Inc., United States), including ammonium chloride (¹⁵NH₄Cl), potassium nitrate (K¹⁵NO₃), potassium nitrite (K¹⁵NO₂), and ¹⁵N-and ¹³C-labeled potassium cyanate (KO¹³C¹⁵N), urea (¹³CO(¹⁵NH₂)₂), and algal amino acid mixture. The final ¹⁵N enrichment after tracer amendments were mostly 5–50%, but some exceeded 50% so should be considered as potential uptake rates. After the tracer additions, bottles were placed in deck-board incubators with underway surface seawater flowing through to achieve near in-situ temperatures and covered with neutral density screens to reduce incident light to approximately 55, 28, 14% of the ambient light levels approximating those observed at the depth of sample collection. The incubations were generally accomplished during daylight hours but there were 7 stations at which incubations were conducted at nighttime. For nighttime incubations, incubators were covered with an opaque tarp after sunset until just before sunrise to prevent the ship’s deck lights from affecting the incubations. After 2–3 h, incubations were terminated by filtration through combusted GF-75 filters at pressure ≤5 mm Hg. Filters were placed in sterile cryovials and stored at -20°C until their analysis. Prior to analysis, samples were dried at 40°C and then pelletized into tin disks. Final PN concentrations and the corresponding atom% enrichment of the PN pool were measured on a Europa 20/20 isotope ratio mass spectrometer as mentioned above. In this study, heterotrophic contributions to the total measured N uptake were not estimated.

Absolute nitrogen uptake (or “transport”) rates were calculated using a mixing model (Mulholland et al., 2006; Eq. 1).

where ( atom % PN ) initial and ( atom % PN ) final represent the ¹⁵N isotopic composition of the particulate pool at the initial and final time points of the incubation period; atom % N source pool is the ¹⁵N isotopic enrichment of the dissolved nitrogen pool after the tracer addition; [PN] represents the concentration of the particulate nitrogen pool; here, we used the average value of the initial and final PN concentrations. Post-incubation measurements confirmed that substrate consumption over the incubation period while significant, was less than 14% (¹⁵NO₃⁻), 18% (¹⁵NO₂⁻), 27% (¹⁵NH₄⁺), 10% (urea), 4% (cyanate), and 25% (amino acid) of initial addition. If 50% of the added substrates were taken up, which occurred 5 out of 77 observations, the calculated uptake rates were excluded from data analysis. Because the tracer enrichment exceeded the recommended 10%, indicating an over-enrichment of these compounds in natural environmental samples, these measurements more accurately reflect potential rather than in situ uptake rates. NH₄⁺ uptake rates were not corrected for “isotope dilution” (Gilbert et al., 1982; Kanda et al., 1987) and may therefore be underestimated. However, the errors would probably be small because of the short incubation times (Harrison and Harris, 1986).

For each target depth, N uptake rates and standard deviations were calculated from triplicate incubations. The average detection limit was 0.01 ± 0.01 nmol N L⁻¹ h⁻¹, but detection limits were calculated individually for each experiment, because calculated N uptake rates are highly dependent on the incubation time and [PN]. The ¹⁵N isotopic composition of the dissolved N pools were not measured due to the lack of reliable methods for all but NO₃⁻ (see Fawcett et al., 2011). We used δ ¹⁵N of 6‰ when ambient NO₃⁻ concentrations were <0.5 μmol L⁻¹, and 2.5‰ when ambient NO₃⁻ concentrations exceeded 0.5 μmol L⁻¹ (Fawcett et al., 2011). For the rest of the N species, the ¹⁵N isotopic composition of the source pool was estimated as the natural abundance of ¹⁵N in atmospheric N or 0‰ (Lipschultz, 2001).

The specific uptake rate of nitrogen was calculated using Eq. 2, which is independent of [PN], or can alternatively be expressed by dividing [PN] using Eq. 1.

To collect phytoplankton pigment samples, 1–4 L of seawater from each of the same three target depths (mentioned above) were filtered onto 25 mm GF/F filters in duplicate under a low vacuum pressure (<5–7 mm Hg). The filters were stored in cryovials and immediately frozen and stored in liquid nitrogen on board and in freezers at -80°C once ashore. Phytoplankton pigments were extracted with 3 mL of 90% acetone for 2 h in the dark at 2–8°C and measured by HPLC in the Ocean Ecology Laboratory at NASA Goddard Space Flight Center, Maryland (Hooker et al., 2012).

The relative contributions of taxa to the total Chl a (TChl a, the sum of Chl a and DV-Chl a) were calculated using the CHEMTAX program (Mackey et al., 1996). Thirteen diagnostic pigments (Supplementary Table S1) were used to associate the fractions of the TChl a pool with nine phytoplankton groups: diatoms (Diat), dinoflagellates (Dino), haptophytes (Type 8; Hapt_8), haptophytes (Type 6; Hapt_6), chlorophytes (Chlo), cryptophytes (Cryp), Prochlorococcus (Proc), Synechococcus (Syne), and prasinophytes (Pras). Chl a concentrations of pico-eukaryotes using CHEMTAX are the sum of haptophytes (Type 8), haptophytes (Type 6), chlorophytes, prasinophytes and cryptophytes. The HPLC pigment-based phytoplankton community composition determined in this study is reliable at the class level. Although this study did not combine measurements of phytoplankton taxonomic abundance using quantitative cell imaging, flow cytometry, or microscopy, at the class level, there are strong positive correlations across these different methods, except for dinoflagellates (e.g., Kramer et al., 2024).

Water masses were identified (see section 3.1) based on temperature and salinity characteristics, as well as satellite-measured sea surface temperature (SST) and sea surface height (SSH) made from August 4–11, 2016. Moderate Resolution Imaging Spectroradiometer (MODIS) Aqua level-3 SST data (8-day mean) with a horizontal resolution of 4 km was obtained from https://oceancolor.gsfc.nasa.gov/l3/. The gridded level-4 altimeter SSH data with a horizontal resolution of 0.25° × 0.25° were obtained from the Archiving, Validation and Interpretation of Satellite Oceanographic (AVISO⁺) data portal.¹ The SSH contour of 0.2 m in the study area was used to outline the Gulf Stream edge (Muglia et al., 2022). The SSH data were also superimposed on the MODIS SST and bathymetry images to highlight the Gulf Stream path.

Chl a concentrations were estimated from CTD fluorescence (FL) using a linear regression (R² = 0.94, Eq. 3) between measured fluorescence and total Chl a concentrations measured by HPLC.

The P^∗ parameter was calculated following Eq. 4:

All statistical analyses and data visualization in our study were performed using MATLAB (R2023a, MathWorks). Statistical differences between the total (and specific) N uptake rates measured at the three depths (surface, above the Chl maximum, and at the Chl maximum) were conducted via a non-parametric test (Mann–Whitney U test), due to the non-normal distribution of the data. Statistical differences between the total (and specific) N uptake rates among the different oceanic regimes (e.g., MAB, SAB, Slope Sea, and Gulf Stream) were also conducted via the Mann–Whitney U test. In addition, a redundancy analysis was conducted to link the observed N uptake rates with the phytoplankton community and environmental variables such as temperature, salinity, and concentrations of nutrients, Chl, and PN. Within the redundancy analysis, each phytoplankton species was expressed as the percentage of the total phytoplankton community.

The Gulf Stream, characterized by high temperature and salinity and extremely low nutrient content, is the most prominent feature of the western boundary of the North Atlantic coast (Figures 1A,B). South of Cape Hatteras, the Gulf Stream moves north, along the coast, hugging the coastline as a boundary-trapped current. Near Cape Hatteras, where the shelf-break is close to shore and the slope steepens, it separates from the coast and becomes a free jet moving north/northeast. North of Cape Hatteras, the Slope Sea, a narrow band of ocean between the Gulf Stream and the MAB continental shelf waters, forms a frontal boundary at the shelfbreak with the south-flowing subsurface Cold Pool. This Cold Pool constitutes a cold, well-mixed water mass, capped by warmer waters resulting from summertime heating inshore of the shelfbreak front (Figure 2).

Our study area was at the intersection of the southwest flowing MAB shelf water Cold Pool and shelfbreak jet, and the north flowing SAB shelf water and Gulf Stream, thus featuring very distinctive physical (Seim et al., 2022; Figures 2, 3) and biogeochemical properties (Figures 4–6). The contrasts between the four distinct oceanic regimes were identified through sea surface temperature (SST) and sea surface height (SSH) maps (Figure 1), sections of temperature, salinity, and density anomalies along the 8 cross-shelf transects (Figure 2), temperature (T) − salinity (S) diagrams (Figure 3), and sections of chlorophyll and inorganic nutrient concentrations (Figure 4). North of Cape Hatteras, the MAB shelf and Slope Sea waters were significantly cooler (Figures 2A–D). At some MAB nearshore stations, low-salinity waters (S < 32) were observed, likely due to the introduction of low-salinity estuarine waters from the Chesapeake Bay (Flagg et al., 2002). The MAB shelfbreak frontal system (shelf − slope) had a more pronounced temperature and salinity gradient due to interactions with the Cold Pool waters intruding from the north (Figures 2A–D,I). The MAB shelf and Slope Sea water were distinguished based on temperature and salinity characteristics (Figure 3; Todd, 2020). Based on the observed gradients we operationally grouped stations into MAB shelf water, when salinity was less than 34.5, and Slope Sea water, when salinity was between 34.5 and 36 (Figure 3).

In contrast, SAB shelf waters south of Cape Hatteras in August were difficult to distinguish from Gulf Stream waters based on temperature and salinity alone (Figure 3; Seim et al., 2022), due to the strong influences from the adjacent Gulf Stream, limited riverine discharge, and seasonal warming. Therefore, there wasn’t a clearly defined frontal zone where SAB shelf waters and the Gulf Stream water interacted (Figures 1B, 2E–H,M–P,U–X). Thus, the SAB shelf water was considered to be waters extending from the coastline to the 200 m isobath. Notably, upwelling occurred at the SAB shelfbreak, including Stations 23, 28, and 35, that was accompanied by lower subsurface water temperatures (Figures 2E,G,H), isopycnal uplifting at σ = 24.5 kg m⁻³ (Figures 2U,W,X), and higher nutrient concentrations (Figures 4M,O,P,U,W,X).

At inner shelf stations where water depths were <30 m, Chl maxima were observed near the bottom, while at outer shelf and offshore oceanic stations, subsurface Chl maxima were prominent features (Figures 4A–H). The depth of the subsurface Chl maximum deepened from shelf to offshore, it was 21 ± 7 m in the outer shelf of the MAB, 33 ± 4 m in the Slope Sea, and 102 ± 11 m in the Gulf Stream (Figures 4A–H; also see Figure 3 in Selden et al., 2021). At shelfbreak stations south of Cape Hatteras where upwelling occurred, e.g., Stations 23, 28, and 35, subsurface Chl maxima were deeper, appearing at 52, 82, and 54 m, respectively (Figures 4E,G,H; Supplementary Figures S2A,G,H). In addition, the Chl concentrations within subsurface Chl maxima were highly variable among different water masses—higher in both the MAB and SAB shelf water and extremely low within the Gulf Stream (Figures 3A, 4A–H).

Distributions of NO₃⁻ and PO₄ (sum of HPO₄²⁻ and PO₄³⁻) concentrations were described in Selden et al. (2021). Briefly, NO₃⁻ was mostly depleted throughout both the MAB and SAB during the period of the survey, however, high PO₄ concentrations, relative to NO₃⁻ and the Redfield ratio, were observed in shelf waters in the MAB, unlike the SAB where both NO₃⁻ and PO₄ concentrations were depleted (Figures 4I–L,Q–T, 5). At Chl hotspots (>2 μg Chl L⁻¹) in the MAB, e.g., at Stations 2 and 8 (Figures 4A,B), the high Chl concentrations appeared to be located at depths at the top of the nitra- and phospho-clines (Figures 4I,J; Supplementary Figures S1B,C). At Stations 23, 28, and 35, higher subsurface NO₃⁻ and PO₄ concentrations were also observed (Figures 4M,O,P,U,W,X; Supplementary Figures S2A,G,H), as a result of upwelling of deeper nutrient-rich water. In the Gulf Stream, both NO₃⁻ and PO₄ concentrations were depleted in the upper 150–200 m. Supplementary Figures S1, S2 show detailed depth profiles of Chl and nutrient concentrations at representative stations.

In general, NO₂⁻ concentrations in the study area ranged from below the detection limit (0.07 μmol L⁻¹) to 0.92 μmol L⁻¹ (Figures 6A−H). For most stations, NO₂⁻ concentrations were not detectable throughout the water column, and subsurface NO₂⁻ maxima were very ephemeral (Figures 6A–H; Supplementary Figures S1, S2). As for NO₂⁻, NH₄⁺ concentrations ranged from below the detection limit (10 nmol L⁻¹) to 960 nmol L⁻¹ (Figures 6I–P). We observed that NH₄⁺ concentrations were often highest near the bottom (up to 0.96 μmol N L⁻¹) at shallow inner-shelf stations. At offshore stations where the water depths were greater, two types of vertical NH₄⁺ distributions were found, those with and without the presence of NH₄⁺ maxima (e.g., Supplementary Figures S1, S2). NH₄⁺ concentrations on the SAB shelf were much lower compared to the MAB shelf (Figures 6M–P), and NH₄⁺ maxima were absent at some stations in the SAB. At stations in the Gulf Stream, NH₄⁺ concentrations were mostly below the detection limit, and both the primary NO₂⁻ and NH₄⁺ maxima were absent or barely detectable (Figure 6). Cyanate concentrations varied from below the detection limit (0.4 nmol L⁻¹) to 25 nmol L⁻¹ (Figures 6Q–X). Similar to NH₄⁺, higher cyanate concentrations were observed in the bottom waters of the MAB shelf (e.g., Stations 6 and 13, and Supplementary Figures S1A,G) suggestive of a sedimentary source. Like NH₄⁺, cyanate maxima were ephemeral, occurring in the subsurface waters at some stations (e.g., Stations 1, 2, 19, 20, and 34) but absent at others (e.g., Stations 12, 25, 35, and 41). Supplementary Figures S1, S2 show typical profiles of cyanate. Urea concentrations were below the detection limit (0.08 μmol L⁻¹) for all samples. Total dissolved primary amines concentrations ranged 0.05–0.4 μmol N L⁻¹, with the mean concentration of 0.17 ± 0.08 μmol N L⁻¹ at depths above and at the depth of Chl maximum (data not shown).

Although the vertical distribution of N uptake rates appeared to be increasing with depth at certain stations and for some N compounds (Figures 7, 8A–C), total absolute N uptake rates in surface waters, above the Chl maximum, and at the Chl maximum were not significantly different (p > 0.05) from each other. Nighttime uptake rates were comparable to those daytime uptake rates observed at the neighboring stations within the same sampling region (Figures 8A–C). Among the six N species examined, urea uptake rates were the highest, accounting for ~27% of the total measured N uptake rates. NO₃⁻ and NH₄⁺ uptake rates were comparable across the study region, and together made up ~40% of the total measured N uptake rates (Figures 7, 8D–F). Amino acids and NO₂⁻ uptake rates were similar to each other, together representing ~30% of the total measured N uptake (Figures 7, 8D–F). Cyanate-N uptake rates were very low, accounting for just 1.2% of the total N uptake on average (Figures 7, 8D–F).

There were remarkable spatial differences in the magnitude of absolute N uptake rates across the different regimes sampled (Figures 8, 9). Total absolute N uptake rates ranged from 0.13–0.69 μmol N L⁻¹ h⁻¹ in the MAB (Figures 8A–C), with combined uptake rates of urea, amino acids, and NO₃⁻ representing ~70% of the total measured N uptake, and uptake of cyanate accounting for only ~0.4% of the total measured N uptake rates (Figures 8D–F). In the Slope Sea, total measured N uptake rates (mean = 0.23 μmol N L⁻¹ h⁻¹, n = 8) were comparable to those observed in the MAB, ranging from 0.11–0.42 μmol N L⁻¹ h⁻¹ (Figures 8A–C). Like the MAB, about 73% of the total observed N uptake in the Slope Sea was from urea, amino acids, and NO₃⁻ (Figures 8D–F). In the SAB, total N uptake rates were lower, ranging from 0.03 to 0.33 μmol N L⁻¹ h⁻¹ (Figures 8A–C). Together, NO₃⁻, urea, and NH₄⁺ uptake comprised ~73% of the total N uptake there (Figures 8D–F). Total measured N uptake rates were significantly greater (p < 0.05, Mann–Whitney U test) in the MAB region (mean = 0.34 μmol N L⁻¹ h⁻¹, n = 15) than the SAB region (mean = 0.13 μmol N L⁻¹ h⁻¹, n = 17). In the Gulf Stream, total measured N uptake rates were the lowest, ranging from 0.004–0.18 μmol N L⁻¹ h⁻¹ (mean = 0.062 μmol N L⁻¹ h⁻¹, n = 32; Figures 8A–C, 9). Cyanate uptake rates within the Gulf Stream were greater than those observed in other regions and represented up to ~11% of the total measured N uptake rates at some stations (Figures 8D–F).

Specific N uptake rates (h⁻¹) that are independent from PN showed a similar trend as the absolute N uptake rates (μmol N L⁻¹ h⁻¹) (Figure 10; Supplementary Figures S3A–C). The average specific N uptake rates were 0.031, 0.026, 0.020, 0.017, 0.016, and 0.0005 h⁻¹ for NO₃⁻, urea, NO₂⁻, NH₄⁺, and amino acids, and cyanate, respectively. Similar to the absolute N uptake rates, the specific uptake rates of urea, NO₃⁻, and NH₄⁺ were higher than those for NO₂⁻, amino acids, and cyanate (Supplementary Figures S3D–F). In vertical profiles, total specific uptake rates did not exhibit significant differences across the three depths examined. Spatially, total specific uptake rates were lowest in the SAB and Gulf Stream (Figure 10; Supplementary Figure S4) and significantly higher in the MAB (p < 0.05, Mann–Whitney U test).

Overall, diatoms, picocyanobacteria (Prochlorococcus and Synechococcus), and haptophytes (Type 8) were the dominant groups in the study area based on CHEMTAX analysis of pigment data (Figure 11). Prasinophytes were also relatively prevalent in the surface mixed layer and at the depth of the chlorophyll maximum in the MAB, Slope Sea, and SAB (Figure 11). The phytoplankton community also displayed distinctive biogeographical patterns with diatoms dominating the phytoplankton community composition in the MAB and Slope Sea, transitioning to picocyanobacterial dominance in the Gulf Stream. Phytoplankton community composition in the SAB appeared to be vertically stratified with picocyanobacteria dominant in surface waters and diatoms becoming increasingly dominant with depth (Figure 11). The Gulf Stream community was characterized by high concentrations of picocyanobacteria in both surface and subsurface waters, however, haptophytes (Type 8) and Prochlorococcus co-dominated at the depth of the chlorophyll maximum. These three groups contributed 93% of the total Chl a in Gulf Stream.

The vertical profiles of phytoplankton biomass differed between water masses (Figure 11). Diatoms and prasinophytes showed clear subsurface maxima layers in all regions except for the Gulf Stream. For the Gulf Stream, haptophytes (Type 8), and Prochlorococcus showed clear subsurface maxima. Synechococcus was more prevalent in surface waters in all regions and its relative dominance decreased with depth. Other groups [haptophytes (Type 6), chlorophytes, and cryptophytes] also showed vertical variations, but their concentrations were relatively low throughout the water column and in all regions.

Chronic or transiently low P availability can constrain biological productivity in aquatic systems (Trommer et al., 2013; Karl, 2014; Duhamel et al., 2021; Hashihama et al., 2021). In addition, a growing body of literature demonstrates the pervasive occurrence of N − Fe or N − P co-limitation of phytoplankton growth across diverse ocean regimes (Graziano et al., 1996; Saito et al., 2014; Browning and Moore, 2023). Nutrient amendment experiments in the North Atlantic subtropical gyre (low-nutrient, low-chlorophyll) confirmed that combining P with N induces larger increases in chlorophyll and primary productivity than N additions alone, emphasizing the role of dissolved P as a secondary limiting factor on phytoplankton growth (Graziano et al., 1996; Moore et al., 2008; Sedwick et al., 2018). Although DOP might serve as alternative P source when PO₄ is depleted, its concentrations generally decrease with increasing PO₄ stress, as indicated by more negative P* values, yielding the lowest DOP level in regions like North Atlantic (as seen in Liang et al., 2022), thus, the SAB and Gulf Stream provinces might experience both inorganic and organic P stress (Mather et al., 2008).

The Cold Pool contains a reservoir of nutrients that can sustain phytoplankton productivity on the MAB shelf during the spring and summer months (Flagg et al., 1994; Hales et al., 2009). Despite significant differences in temperature and salinity between the MAB and SAB, this study found excess PO₄ in the MAB waters (Figures 4, 5). The positive P* values across the northern MAB suggest an excess PO₄ relative to N and the Redfield Ratio, a trend contrasting with the mostly near-zero or negative P* values observed in the SAB (Figure 5C). This was likely due to: (1) the MAB being influenced by the Cold Pool waters intruding from the sub-Arctic whose source waters have already undertaken denitrification, resulting in an excess PO₄ relative to NO₃⁻ (Harrison and Li, 2007; Fennel, 2010; Sherwood et al., 2021), and (2) the SAB shelf region has more direct interactions with the oligotrophic Gulf Stream and has limited riverine inputs (Andres et al., 2023). Specific N uptake rates were significantly higher in the MAB than the SAB (p < 0.05), showing a positive correlation with ambient P concentrations (r = 0.31, p < 0.05, Figure 12). This result suggests a N-P synergy, such that increased P availability enhances phytoplankton’s ability to assimilate and utilize available N, as has been observed in the northern Gulf of Mexico (Turner and Rabalais, 2013); whereas when P is limited, organisms may not be able to utilize available nitrogen as efficiently, leading to lower apparent nitrogen uptake rates.

Under various growth conditions, phytoplankton employ optimal allocation strategies, allowing them to allocate N and P to specific cellular functions necessary for survival (and growth) and thereby offering them some plasticity in their N:P ratios. For example, phytoplankton N:P ratios can be higher when growing under competitive equilibrium conditions with resource limitations (e.g., N, P, and light) that requires allocating resources to acquisition assembly. Conversely, phytoplankton N:P ratios can be lower during exponential growth when cell metabolism assembly takes priority (Klausmeier et al., 2004; Armin et al., 2023). Observations suggest a nuanced pattern where fast-growing species, like diatoms, have relatively stable but P-rich elemental compositions, while slower-growing species, such as cyanobacteria, exhibit more flexible elemental stoichiometry (Hillebrand et al., 2013). In addition, at low nutrient concentrations, diffusive nutrient transport was found to increase linearly with phosphorous concentrations (Armin et al., 2023). In the context of this study, diatoms dominated both the MAB and SAB regions, hence the demand for P at their fast-growing phase might have been high. Consequently, the synergistic response, higher N uptake rates (and perhaps growth rates) in the MAB where there was excess P, may have been promoted to sustain the increased allocation to biosynthetic and photosynthetic molecules. In contrast, severe P stress in the SAB at the time of the cruise may have limited N uptake, because the deficiency of P can reduce the cell quota of this element and also might impair the uptake and decrease cell quota of N (Persson et al., 2010; Hillebrand et al., 2013; Giménez-Grau et al., 2020).

In the oligotrophic Gulf Stream region, the residing phytoplankton community has been subjected to prolonged N and P starvation, and it exhibited the lowest community-level N uptake. Picocyanobacteria are thought to have higher binding affinity for nutrients overall. Although, Prochlorococcus, the dominant species in this region, are able to regulate high-and low-affinity intracellular P transport systems at different P availabilities (Martiny et al., 2006; Lin et al., 2016), under the severe nutrient starvation in the Gulf Stream where both N and P (and possibly other nutrients) were likely below critical threshold values, it was not surprising to observe the lowest N uptake rates as their physiology may have already attuned to the nutrient-poor pre-condition.

Theoretical studies, culture experiments, and in situ data suggest diverse nutrient uptake strategies in phytoplankton populations and communities (Litchman et al., 2007; Edwards et al., 2012; Lomas et al., 2014). The size of phytoplankton cells influences physiological rates (e.g., nutrient uptake and growth rates), biotic interactions (e.g., grazing) and behaviors (e.g., sinking speed) in the fluid environment (Barton et al., 2013b; Edwards et al., 2015). Larger cells, such as diatoms, tend to have greater maximum uptake rates on a per-cell basis (Sarthou et al., 2005; Litchman et al., 2007; Edwards et al., 2011; Lomas et al., 2014) and larger nutrient storage capacity (Grover, 1991; Stolte and Riegman, 1995; Grover, 2011; Stief et al., 2022). This allows them to quickly take up nutrient pulses (Zimmerman et al., 1987) and/or store nutrients after brief periods of high supply (Bode et al., 1997; Stief et al., 2022), prolonging saturated growth beyond the nutrient pulse (Edwards et al., 2013). This explains why large sized phytoplankton (e.g., diatoms) are favored/selected under high or pulsed nitrate supplies occurring over a relatively long (~2–30 day) periods, e.g., during upwelling events (Litchman et al., 2009). Recent studies on diatom gene expression patterns suggest diatom can rapidly assimilate newly acquired nitrate and reduce carbon requirement to support high growth rates (Lampe et al., 2019; Inomura et al., 2023), highlighting their physiological adaptation to nutrient replete and intermittently replete environments. Consistent with these findings, diatoms were the dominant group in phytoplankton assemblages at both shelf regions, especially at the depths of Chl maximum, coinciding with the top of the nitracline where nutrient diffusivity or diapycnal transport usually occurs (e.g., Hales et al., 2009).

Under nutrient-limited conditions, particularly in stratified systems, phytoplankton communities rely on reduced/regenerated forms of nitrogen for growth (Raimbault et al., 1999; Fawcett et al., 2011). Gene expression patterns under nitrogen starvation reveal a dependence on regenerated nitrogen, with NH₄⁺ and urea transporters highly expressed (Lampe et al., 2019). Smaller phytoplankton cells exhibit higher scaled nutrient binding affinities than larger cells, allowing them to maintain positive growth rates at lower nutrient concentrations (see Figure 1 in Edwards et al., 2012). In subtropical gyres with consistent stratification and weak nutrient resupply to surface waters, smaller phytoplankton cells can draw down ambient nutrient concentrations to critically low levels where larger cells cannot survive (Barton et al., 2013b). Congruent with these observations and our hypothesis, Prochlorococcus was found to be the dominant species in Gulf Stream surface waters. Thriving in these “ocean deserts,” Prochlorococcus demonstrated the capability to deploy multiple biochemical strategies simultaneously to use urea (Moore et al., 2002; Saito et al., 2014), amino acids (e.g., Zubkov et al., 2003), and cyanate (e.g., Kamennaya et al., 2008). A recent model study integrating genomic and molecular datasets detailed the cellular-scale optimization of metabolism and physiology in Prochlorococcus (Casey et al., 2022).

Synechococcus was ubiquitous across our study region, particularly dominant at the shallower euphotic depths (Figures 11D,E). Synechococcus is capable of mobilizing ammonium, nitrite, nitrate, urea, cyanate, and amino acids to support their growth (Lindell et al., 1998; Moore et al., 2002; Kamennaya and Post, 2011; Muñoz-Marín et al., 2020; Sato et al., 2022). In contrast to Prochlorococcus who lacks genes for nitrate uptake and reduction (Moore et al., 2002), the ability to assimilate both oxidized and reduced N may reflect their higher cellular N requirements, especially given the large, N-rich light-harvesting protein complexes (phycobilisomes) that must be maintained (Scanlan, 2003). This could explain the predominance of Synechococcus shallower in the water column (Figures 11D,E) as the low light levels at the base of the euphotic zone may not yield sufficient energy to reduce oxidized N forms (Van Oostende et al., 2017) and transport N and P into outer membrane (Kamennaya et al., 2020). It also implies a trade-off or trait that helps to explain their broader geographical distribution in waters receiving higher irradiance, regardless of nutrient conditions (Partensky et al., 1999).

Haptophytes (Type 8) were the dominant group of pico-eukaryotes (Figure 11) and comprised nearly 50% of the total Chl a at the depth of Chl maximum in the Gulf Stream (Figure 11F). This pattern is consistent with observations from oligotrophic areas where haptophytes (Type 8) occupied niches characterized by elevated mean nitrate and phosphate concentrations but low mean temperature and irradiance (Xiao et al., 2018). Studies also noted that diatoms and haptophytes competitively interact in natural environments (Alexander et al., 2015b). Ecologically, diatoms are generally r-strategists, characterized by rapid growth rates under suitable conditions. In contrast, haptophytes are likely K-strategists, characterized by slow growth rates (Endo et al., 2018). This distinction is further supported by a metatranscriptomic study in an oligotrophic ocean (Alexander et al., 2015b), where diatoms increased growth-related transcriptional activity under nutrient-rich conditions and haptophytes decreased. Our findings corroborate this pattern, with low diatom but high haptophyte abundances (Type 8) at the depth of the Chl maximum in the Gulf Stream (Figure 11F).

The influence of light, temperature, nutrient availability, and their interactions on primary producers are critical factors in shaping phytoplankton communities and associated biogeochemical processes (MacIsaac and Dugdale, 1972; Edwards et al., 2015, 2016; Maguer et al., 2015). In general, the North Atlantic region experiences N limitation (Graziano et al., 1996). Several studies have reported uptake rates for various nitrogen compounds, including NO₃⁻ (Elskens et al., 1997; Rees et al., 2006; Painter et al., 2008b), urea (e.g., Painter et al., 2008a), NH₄⁺ (Rees et al., 1999; Donald et al., 2001; Rees et al., 2006), and NO₂⁻/amino acids (e.g., Fuhrman, 1987; Filippino et al., 2011), across a wide space of Atlantic Ocean. In addition, through WOCE and JGOFS expeditions, Harrison et al. (1996) examined N uptake kinetics across the northern Atlantic and found the uptake of NO₃⁻ and NH₄⁺ by phytoplankton were both concentration-and temperature-dependent. Reay et al. (1999) found that phytoplankton’s affinity for NO₃⁻ was indeed strongly dependent on temperature and consistently decreased at temperatures below their optimum temperature, however, the affinity for NH₄⁺ showed no clear temperature dependence. Subsequently, Smith et al. (2009) suggested that nutrient uptake patterns of phytoplankton are better explained as a trade-off between uptake capacity and affinity, and that phytoplankton cells acclimate to varying nutrient concentrations, modifying their apparent half-saturation constants for nutrient uptake. It is now understood that the uptake rates of nutrients is a manifestation of phytoplankton physiological state and demonstrate a degree of plasticity (Kwon et al., 2022).

Using both pair-wise correlations (Heatmap) and multivariate statistical analysis (redundancy analysis, RDA), we found that both temperature and salinity exhibited negative correlations with total N uptake and uptake rates of individual N compounds, whereas PN and Chl concentrations were positively correlated with total N uptake and uptake rates of individual N compounds (Figure 12). There was no significant relationship between NO₃⁻ concentrations and total N uptake or between NO₃⁻ concentrations and the uptake rates of other N compounds in the study area. PO₄ concentrations showed positive correlations with the total N uptake and the uptake of individual N compounds, supporting the earlier suggestion that elevated P concentrations contribute to higher N uptake rates through synergistic effect or N-P co-limitation. In addition, distinct patterns were observed between phytoplankton species compositions and environmental factors. For example, there were negative correlations between temperature and salinity and the abundances of diatoms and pico-eukaryotes, but positive correlations between temperature and salinity and Prochlorococcus and Synechococcus fractions, corroborating what we know about these groups’ physiological preferences. Synechococcus dominance also showed (weak) positive correlations with cyanate uptake rates, highlighting this organisms capability to use cyanate as a nitrogen source, a pattern also observed in the North Pacific subtropical gyre (Sato et al., 2022).

In the RDA plot (Figure 13), the first RDA axis (RDA 1) captured nearly 95% of the variability related to the explanatory variables. The coefficient of RDA 1 for total N uptake was 0.93. Temperature, salinity, and Prochlorococcus exhibited the highest negative correlations (r < −0.6) with total N uptake, and this was particularly pronounced in observations made in the Gulf Stream. As discussed in section 4.2, elevated temperatures can lead to strong stratification, limiting nutrient resupply and availability. Prochlorococcus is well adapted to nutrient-poor environments, exhibits inherently low nitrogen uptake rates.

PN, phosphate, dinoflagellates, prasinophytes, and cryptophytes demonstrated the highest positive correlations (r > 0.5) with total N uptake. It remains unclear why dinoflagellates and cryptophytes exhibited stronger correlation with N uptake than diatoms, despite not being the dominant species. This analysis reinforces the notion that phosphorus serves as the secondary or co-limiting factor, particularly in the SAB and MAB. For individual N uptake rates, ammonium and cyanate uptake rates showed negligible correlations with any measured variables (although absolute cyanate uptake rates were ubiquity low). Correlations with the remaining individual N uptake rates were obvious but insignificant, further indicating that uptake processes are complex and influenced by phytoplankton physiological status, nutrient pre-conditioning, and the relative rates of nutrient resupply.