Phytoplankton size-class contributions to new and regenerated production during the EXPORTS Northeast Pacific Ocean field deployment

Accurate quantification of ocean carbon export (i.e., the transport of carbon from surface to depth in the ocean) is essential to understanding the oceanic carbon storage capacity and its role in mitigating anthropogenic global climate change. Ocean carbon export can be estimated by coupling ground-truthed remote sensing observations with global models to assess current and future global projections. Crucial to these estimates is good knowledge of the processes which influence how much primary production makes its way from sunlit regions of the ocean to depth, which is a function of the sources of nutrients that fuel phytoplankton growth and the characteristics of the plankton community (both autotrophic and heterotrophic components). The NASA EXport Processes in the Ocean from RemoTe Sensing (EXPORTS) program was launched with the goals of advancing remote sensing technologies and interpretations to predict carbon export from the surface ocean (Siegel et al., 2016; Siegel et al., 2021). EXPORTS represents a comprehensive effort to couple carbon export with measurements of primary production, plankton community composition, biogeochemical controls, and heterotrophic processes to develop a holistic view of oceanic dynamics during two field deployments, the first of which was conducted in the Fall of 2018 in the Northeast Pacific Ocean.

Ocean Station Papa (Station P) in the Northeast Pacific Ocean is the terminal site for one of the longest running ocean time series (Line P Program) with measurements recorded since the 1950s (Harrison, 2002). Station P is located at 50°N and 145°W within a high nutrient (nitrate), low chlorophyll (HNLC) region where low annual rates of primary production typically range from 5 to 18 mol C m^–2 yr^–1 (Wong et al., 1995; Boyd and Harrison, 1999). In the late 1980s, the Iron Hypothesis was proposed to explain the lack of production in many HNLC regions including the Northeast Pacific Ocean (Martin and Fitzwater, 1988; Boyd et al., 1996; Boyd et al., 1998). Initial large-scale iron enrichment experiments (IronEX I and II) were conducted in the Equatorial Pacific Ocean, demonstrating the stimulation of large diatom blooms following iron addition to the surface waters (Martin et al., 1994; Coale et al., 1996). Since then, numerous experiments (e.g., SEEDS I and II, SERIES, SOIREE, EisenEx, SOFEX) have verified the extent to which phytoplankton are limited by iron availability and respond to Fe additions in HNLC regions around the globe.

New production represents the proportion of total primary production supported by inputs of new nitrogen into the euphotic zone, which in the Northeast Pacific Ocean is predominantly supplied from deep waters in the form of nitrate (NO₃^–; Whitney and Freeland, 1999). In a steady-state system where phytoplankton biomass is not rapidly changing, the amount of new production within the euphotic zone is thought to be in equilibrium with carbon export. Thus, estimates of carbon export potential may be derived through measurements of new production (Eppley and Peterson, 1979). Yet, the prevalence of utilization and growth on regenerated forms of nitrogen in such forms as ammonium (NH₄⁺) and urea in large regions of the ocean in place of NO₃^– was highlighted by Dugdale and Goering (1967) and then further characterized by Eppley and Peterson (1979) and continues to play a crucial role in global production. The balance between primary production supported by NO₃^– versus NH₄⁺ and other forms of recycled nitrogen (referred to as regenerated production) can be described by the ratio of new production to total production (i.e., new production + regenerated production), where ratios greater than 0.5 indicate a predominance of new production and ratios less than half indicate a predominance of regenerated production. This ratio is commonly referred to as the f-ratio (Eppley and Peterson, 1979). The differentiation between new and regenerated production has led to the study of the mechanisms behind, and influence of, different nitrogen-based metabolic strategies in oceanic regimes and what they mean for carbon export potential.

Iron limitation prevents efficient utilization of macronutrients such as nitrate, phosphate and silicic acid along with dissolved inorganic carbon (DIC) uptake by phytoplankton, particularly in larger-sized phytoplankton such as diatoms, due to the role of iron as a cofactor for enzymes involved in photosynthesis assimilation pathways and nitrogen assimilation pathways. This inefficiency leads to the predominance of smaller-sized phytoplankton that mostly rely on recycled forms of nitrogen. The iron requirement for NH₄⁺ uptake is substantially lower (1.6x; Raven, 1988) than that of NO₃^– uptake, causing a preference for NH₄⁺ utilization by phytoplankton in these Fe-limited regions (Price et al., 1994; Varela and Harrison, 1999). Additionally, high rates of regenerated production have been observed in small-celled phytoplankton with low intracellular chlorophyll contents and larger surface area-to-volume ratios which enhance nutrient acquisition (Greene et al., 1992; Marchetti et al., 2006a).

Previous studies at Station P have observed distinct trends between small (<5 µm) and large-celled (>5 µm) phytoplankton, in terms of both biomass and productivity. The small size class constitutes picophytoplankton (0.2–2 µm) whereas the large size class constitutes microphytoplankton (>20 µm). Nanophytoplankton (2–20 µm) are mostly comprised of small cells (e.g., haptophytes and chlorophytes), although some may be considered large cells depending on the group considered. Under typical, iron-limited conditions, small-celled phytoplankton, specifically cyanobacteria, chlorophytes, and haptophytes, numerically dominate the assemblage (92%), and larger-celled phytoplankton such as diatoms and dinoflagellates are present in low concentrations (8%; Varela and Harrison, 1999; Marchetti et al., 2006c). Productivity data from 1992–1997 showed a consistently low f-ratio (ranging from 0.05 to 0.50) with regenerated production almost always outpacing new production and very minimal seasonal variability (Varela and Harrison, 1999; Peña and Varela, 2007). From these data, pico- and nanophytoplankton are recognized as typically dominating the phytoplankton community at Station P, performing most of the regenerated production. The EXPORTS field campaign provided the unique opportunity to quantify the production rates and f-ratios of small and large-celled phytoplankton independently and within a Lagrangian framework to assess their contributions to total productivity trends and how these contributions may vary over short (i.e., days to a month) time scales.

A primary goal of the EXPORTS Northeast Pacific Ocean field campaign was to collect an expansive series of in situ data to describe biological, chemical, and physical processes during the summer months when net primary production (NPP) is at its annual peak. To assess the processes involved in carbon sequestration via the rates and controls on primary production, we measured size-fractionated rates of ¹³C-DIC isotopic uptake (¹³C-ρDIC) and ¹⁵N-NO₃^– isotopic uptake (¹⁵N-ρNO₃^–), and concentrations of chlorophyll a (chl a), particulate carbon (PC), and particulate nitrogen (PN) as proxies for plankton biomass over an observational period of 24 days. From these measurements, we calculated size-fractionated and whole phytoplankton community f-ratios and carbon and nitrogen assimilation rates (¹³C-VDIC, ¹⁵N-VNO₃^–) in order to assess average values and temporal patterns. Our size-fractionated data allowed us to evaluate the balance between small and large cells and their nitrogen uptake preferences and strategies in order to assess biomass standing stock, temporal productivity trends, and community structure, thus leading to a better understanding of their contributions to carbon export potential. A Lagrangian-like sampling scheme allowed our data to be evaluated as one distinct water mass naturally evolving over the course of the observation period. By removing significant shifts in abiotic factors associated with sampling different water masses, we avoided confounding our observations and enabled them to be evaluated holistically, and ultimately, allowing for the evaluation of our primary productivity measurements in the context of calculating carbon export potential out of the euphotic zone in this region.

The Northeast Pacific Ocean EXPORTS field deployment was conducted in August and September 2018, onboard the R/V Roger Revelle and the R/V Sally Ride. Most of the measurements presented here were performed on the R/V Revelle. The cruise consisted of a 24-day Lagrangian study beginning at Ocean Station Papa (50°N, 145°W), combining autonomous platforms and in situ sampling of ocean parameters associated with ecosystem dynamics and carbon export. During the cruise, the observation period began on August 16, 2018, Julian Day (JD) 228, and ended on September 7, 2018, JD 250 (Figure 1). The observational period was preceded and followed by continued observations from autonomous vehicles (gliders, AUVs) and BioArgo floats and an extra day of sampling on board the CCGS John P. Tully as part of the Canadian contribution to EXPORTS and the Line P time series in late September 2018 (JD 264). Primary productivity casts were performed just prior to dawn (04:00–05:00 local time, with one cast at 03:00–04:00, August 20, and one cast at 05:00–06:00, September 5) every other day, beginning on JD 228, using a trace metal clean (TMC) rosette system containing 11 pre-cleaned Go-Flo bottles. Seawater was collected at five depths throughout the euphotic zone corresponding to 65%, 38%, 20%, 10%, and 1% of incident irradiance (I_o). On three sampling days (between JD 238 and 242), sample depths were changed to 40%, 20%, 10%, 5%, and 1% of I_o in order to measure biological rate processes at a particle maximum at approximately 50 m, as observed using a Becton Dickinson Influx Cell Sorter (BD-ICS) flow cytometer measuring discrete seawater samples following protocols of Graff and Behrenfeld (2018). Seawater was dispensed into pre-cleaned bottles within a TMC sampling van where positive pressure was applied to Go-Flo bottles. On JD 248, inclement weather prevented the deployment of the TMC rosette, so seawater was obtained from the ship’s near-surface underway system.

Following the observational period, nitrate (NO₃^–), phosphate (PO₃^4–), and silicate (Si(OH)₄) concentrations were measured at UC Santa Barbara according to protocols outlined in Siegel et al. (2021). Dissolved iron sampling and analysis are described in Siegel et al. (2021) and Mellett and Buck (2020). Ammonium (NH₄⁺) concentrations were measured on ship using the orthophthaldialdehyde (OPA) fluorescence method (Holmes et al., 1999; Taylor et al., 2007). Urea concentrations were analyzed on shore according to a diacetyl monoxime method modified from Mulvena and Savidge (1992). All EXPORTS data are available on the SeaWiFS Bio-optical Archive and Storage System (SeaBASS; https://seabass.gsfc.nasa.gov/).

Triplicate 500-mL HDPE dark bottles were filled with seawater from each depth and stored in the dark at 4ºC until filtration, typically within 2 hr of sample collection. For size-fractionated chl a measurements, 400 mL of seawater were gravity-filtered onto a 5-µm polycarbonate filter and then vacuum-filtered onto a GF/F filter (Whatman 0.7-µm nominal porosity) set up in a filter-series cascade. Total concentrations were calculated as the sum of these two size fractions. Filters were frozen at −20ºC until chl a extractions were performed. For extractions, 6 mL of 90% ethanol (HPLC-grade) were added to each filter in a 20-mL glass scintillation vial and kept at −20ºC in the dark for 24 hr prior to being analyzed on a Turner Designs 10-AU fluorometer following Graff and Rynearson (2011). Triplicate size-fractionated samples allowed for determination of standard deviations and estimation of uncertainty of the samples. All chl a extractions and measurements were performed on the ship and within 1 week of sample collection.

Two sets of triplicate 1-L acid-cleaned polycarbonate bottles were filled with seawater collected from the TMC rosette system. One set of incubation bottles was used for short-term (6-hr) incubations where seawater was collected at the three light depths in the upper water column primarily located within the mixed layer. The second set of incubation bottles was used for longer-term (24-hr) incubations where seawater was collected at all five depths throughout the euphotic zone. Bottles were acid-soaked (10% HCl) for at least 24 hr and rinsed with Milli-Q water and seawater prior to filling. The samples incubated for 6 hr were used to estimate gross primary productivity (GPP), and the samples incubated for 24 hr were used to estimate net primary productivity (NPP; Table 1).

For ¹³C-DIC uptake rate (¹³C-ρDIC) measurements, NaH¹³CO₃ isotope (Cambridge Isotope Laboratories, 99 atom % ¹³C) was added to each bottle to achieve a stable isotope concentration of 180 µmol L^–1 that was anticipated to be approximately 10% of ambient DIC, assuming concentrations between 1800 and 2000 µM. For ¹⁵N-NO₃^– uptake rates (¹⁵N-ρNO₃^–), ambient NO₃^– concentrations were estimated by an in situ ultraviolet spectrophotometer (ISUS) attached to the underway system. The volume of Na¹⁵NO₃^– isotope (Cambridge Isotope Laboratories, 98 atom % ¹⁵N) addition corresponded to approximately 10% of measured ambient NO₃^– concentration at each depth obtained from the prior day’s cast. All stable isotope additions were performed within a metal-free flow hood located within a TMC positive-pressure plastic bubble constructed on the ship. After inoculations with stable isotopes, samples from 65–5% I_o depths were immediately placed in irradiance-controlled (via neutral-density screening) on-deck surface seawater flow-through incubators. Samples from the 1% I_o were placed in irradiance-controlled on-deck incubators cooled to approximately 5ºC using circulating chillers to simulate cooler temperature at this irradiance depth below the mixed layer (Figure S1).

Following incubation, samples were removed from the incubators and placed in dark plastic bags within a walk-in refrigerator room; filtration commenced immediately. Of the triplicate bottles collected at each depth, a single replicate was filtered directly onto a pre-combusted (450ºC for 4 hr) GF/F filter (Whatman 0.7-µm nominal porosity) to represent the whole sample whereas the other two replicates were first filtered onto a 5-µm polycarbonate filter and the filtrate passing through this filter was then filtered onto a pre-combusted GF/F filter. Large (>5 µm) cells captured on the polycarbonate filter were rinsed onto a pre-combusted GF/F filter using 0.2-μm-filtered seawater. Duplicate samples were averaged to represent our two size fractions and the whole sample was kept separately so that each depth had three independent uptake measurements. Comparison between the sum of the two size-fractionated samples and the whole sample showed good agreement, suggesting minimal biomass losses during the size-fractionation process (Figure S2). A time-zero blank filter was collected three times during the cruise and subtracted from the ¹³C-ρDIC calculations. These blanks measured no detectable nitrogen isotope retention, indicating no time-zero ¹⁵N-ρNO₃^– requiring subtraction. All filters were stored in acid-cleaned 30-mm Petri dishes, sealed with vinyl tape, and frozen at −20ºC until further processing. Onshore, filters were dried overnight at 60ºC and pelletized in combusted tinfoil squares. A combusted, unfiltered Whatman GF/F was also used for a filter blank in the uptake rate calculations and treated in the same manner as the seawater samples. Samples were analyzed via mass spectrometry at the UC Davis Stable Isotope Facility. Atom % ¹³C, atom % ¹⁵N, PC and PN concentrations for samples, blanks, and standards were measured, thus enabling the calculation of ¹³C-ρDIC and ¹⁵N-ρNO₃^– as described by Slawyk et al. (1997) and Dauchez et al. (1995). Samples for discrete DIC measurements used for the ¹³C-ρDIC uptake rate calculations were collected aboard the R/V Sally Ride and analyzed at Monterey Bay Aquarium Research Institute according to O’Sullivan and Millero (1998) and Dickson et al. (2007). Samples for dissolved NO₃^– concentrations used for the ¹⁵N-ρNO₃^– calculations were collected aboard the R/V Sally Ride and analyzed at the UC Santa Barbara Marine Science Institute Analytical Lab.

On JDs 246 and 250, ammonium (NH₄⁺) uptake rate (¹⁵N-ρNH₄⁺) measurements were performed using a single replicate from each of the long-term incubations at each of the five depths throughout the euphotic zone. One-liter polycarbonate bottles were inoculated with ¹⁵NH₄Cl (Cambridge Isotope Laboratories, 98 atom % ¹⁵N) to achieve a final concentration of 0.1 μmol L^–1 and incubated, filtered for size-fractionation and analyzed according to protocols described above for ¹³C-ρDIC and ¹⁵N-ρNO₃^–. Given that stable isotope inoculations of ¹⁵NH₄Cl were greater than 10% of the ambient concentrations (ranging from approximately 30 to >100%), these measured rates are considered potential maximum uptake rates as opposed to absolute in situ uptake rates. ¹⁵N-ρNH₄⁺ was calculated according to Dauchez et al. (1995).

On JD 264, in addition to ¹³C-ρDIC and ¹⁵N-ρNO₃^– incubation experiments, the Line P program conducted ¹⁵N-ρNH₄⁺ and ¹⁵N-ρurea uptake rate experiments. Similar to the EXPORTS cruise, samples were collected from a standard CTD rosette and incubated for 24 hr on deck with surface water flow-through incubators. ¹⁵NaNO₃^– isotope additions were calculated based on previous years ambient NO₃^– concentrations. The limit of detection for measuring ¹⁵N-ρurea is 0.10 µmol L^–1, so ¹⁵N-ρurea was added at a concentration of 0.11 µmol L^–1. Samples were also analyzed at the UC Davis Stable Isotope Facility with ambient NH₄⁺ concentrations calculated onboard using a Turner 10-AU fluorometer and ambient NO₃^– concentrations measured at the Institute of Ocean Sciences following the protocol of Barwell-Clarke and Whitney (1996). Line P ¹³C-ρDIC (Cambridge Isotope Laboratories, 99 atom % ¹³C), ¹⁵N-ρNO₃^–, ¹⁵N-ρNH₄⁺, and urea isotope (Cambridge Isotope Laboratories, 98 atom % ¹⁵N for all nitrogen isotopes) uptake rates were calculated according to Hama et al. (1983) and Dugdale and Wilkerson (1986) for ¹³C-ρDIC and ¹⁵N-ρN, respectively.

New production rates were estimated by converting ¹⁵N-ρNO₃^– rates to C units using the C₁₀₆: N₁₆ ratio (Redfield, 1958). Rates of regenerated production (mmol C m^–2 d^–1) were calculated in two manners, either as ¹⁵N-ρNH₄⁺ or using Equation 1:

If 6.6 * ¹⁵N-ρNO₃^– exceeded ¹³C-ρDIC, regenerated production was assumed to be negligible (i.e., zero). Depth-integrated rates of ¹³C-ρDIC, new production, and regenerated production were calculated through linear extrapolation of the rate of the surface-most (65% I_o or 40% I_o) sample depth to the surface and trapezoidal integration of discrete samples through the water column. Uncertainty for duplicate and triplicate samples was estimated as the average percent range of all of the samples per depth included in the integration and standard deviation, respectively.

The ratio of new production to total primary production (referred to as the f-ratio; Eppley and Peterson, 1979) was calculated in two manners:

Mixed layer f-ratios were estimated from the 6-hr incubations due to the potential uncoupling of DIC and NO₃^– uptake rates over the 24-hr period (Harris and Riley, 1956; Dugdale and Goering, 1967). This uncoupling is illustrated by substantial differences between 24 hr versus 6 hr ¹³C-ρDIC yet only moderate differences between 24 hr versus 6 hr ¹⁵N-ρNO₃^– when both measurements are converted to daily rates (Figure S3). Differences between f-ratios calculated using 24-hr incubation data (NPP) versus f-ratios calculated using 6-hr incubation data (GPP) were noteworthy at 18.8 to 51.0% (Table S2; see discussion below).

To calculate GPP in the mixed layer, rates measured in the 6-hr incubations were extrapolated to daily rates by determining the proportion of 6-hr photosynthetically active radiation (PAR) compared to the 24-hr PAR (Equation 4). Incident PAR was obtained from a HOBO PAR Smart Sensor loggers attached to one of the on-deck incubators. Levels of PAR that would result in GPP saturation via photoinhibition (i.e., PAR values greater than the threshold for photoinhibition, E_k) were predicted from photosynthesis-irradiance curves (J. Fox, personal communication). To account for potential overestimation of GPP in our PAR-based extrapolation method, the maximum 24-hr PAR was constrained to the maximum level of PAR that occurred within our 6-hr incubations (Figure S4).

Numerous assumptions are made when using this approach to calculate rate measurements. Previous studies have questioned whether 24-hr incubations are measuring NPP or a rate somewhere between NPP and net community production (NCP; Marra, 2002; 2009; Barber and Hiking, 2007). Similarly, 6-hr incubations have been shown to measure a rate between GPP and NPP (Marra, 2002; 2009). Therefore, while our definition of the f-ratio varies slightly from previous analyses, we expect that the proposed ratio more accurately maintains the balance between new and regenerated production while removing some of the biases in nitrate uptake due to the decoupling of carbon and nitrogen processes in the dark over a 24-hr period. Additionally, for this analysis, we chose to use the Redfield ratio to convert nitrogen to carbon units instead of the measured PC: PN ratio due to the high amounts of non-algal associated particulate matter at Station P. This conversion via the Redfield ratio rather than our experimental PC: PN ratio may slightly overestimate rates of new production, underestimate regenerated production calculated via Equation 1, overestimate regenerated production calculated as ¹⁵N-ρNH₄⁺, and overestimate the f-ratio. Isotope dilution of the tracer pool could also lead to underestimation of regenerated production in ¹⁵N-ρNH₄⁺ experiments. Ammonium regeneration rates were not measured; however, given that additions of ¹⁵N-NH₄⁺ achieved a labeling of 19.24–71.96 atom% with a mean and standard deviation (SD) of 55.58 ± 15.06 atom% (n = 32), large underestimations in our experiments due to dilution are unlikely.

Chlorophyll-normalized DIC assimilation rates (¹³C-VDIC) and chlorophyll-normalized NO₃^– assimilation rates (¹⁵N-VNO₃^–) were calculated for both size-fractionated and whole samples. Rates were calculated by dividing volumetric measurements of ¹³C-ρDIC and ¹⁵N-ρNO₃^– by the corresponding chl a concentrations. For within the mixed layer and below the mixed layer assimilation rate comparisons with ¹³C-ρDIC or ¹⁵N-ρNO₃^–-based new production, the average assimilation rate for the sample day was used.

Consistent with the EXPORTS goal of conducting a Lagrangian study, 10 of 11 sampling days (between JDs 228 and 246) displayed relatively constant physical water properties suggesting sampling occurred within a single water mass (Figure 1; Seigel et al., 2021). Over the duration of the observational period, mixed-layer temperatures and salinities varied marginally, ranging between 13.5 and 14.6ºC and between 32.24 and 32.34, respectively. Daily integrated PAR ranged from 10 to 40 mol photons m^–2 d^–1 with maximum values occurring on JD 239. Euphotic zone depths (z_eu) were estimated based on the depth at which PAR was 1% of I_o. Throughout the observational period, z_eu varied from 70 to 90 m with an average (± SD) depth of 78 ± 6 m (n = 149; Siegel et al., 2021). Mixed layer depths were calculated as the first depth where potential temperature changed by 0.2ºC relative to temperature at 5 m (McNair et al., n.d.). Mixed layer depths ranged from 25.6 to 37.5 m with an average depth of 32.3 ± 3.7 m (n = 26).

Macronutrients were expectedly high and relatively invariant over the course of the observational period, consistent with the characterization of the region as HNLC. Euphotic zone NO₃^– concentrations ranged from 8.10 to 17.10 µmol L^–1 with an average (± SD) of 11.12 ± 2.71 µmol L^–1 (n = 62), PO₄^3– concentrations ranged from 0.80 to 1.40 µmol L^–1 with an average of 1.04 ± 0.18 µmol L^–1 (n = 115), Si(OH)₄ concentrations ranged from 13.00 to 22.80 µmol L^–1 with an average of 16.90 ± 1.89 µmol L^–1 (n = 115), and NH₄⁺ concentrations ranged from 0 to 1.08 µmol L^–1 with an average of 0.15 ± 0.12 µmol L^–1 (n = 72; Figure S5; Siegel et al., 2021). Dissolved iron was only calculated in the mixed layer where it ranged from 0.014 to 0.086 nmol L^–1 with an average of 0.043 ± 0.025 nmol L^–1 (K. Buck and S. Burns, personal communication).

Phytoplankton biomass in the mixed layer varied approximately 3-fold throughout the observational period, with chl a concentrations ranging from 0.12 to 0.32 µg L^–1 with an average (± SD) concentration of 0.23 ± 0.06 µg L^–1 (n = 34; Figure 2). The small size fraction accounted for 47–90% of the total chl a concentrations with an average proportion of 68% (Figures 2 and 3B). Mixed layer chl a concentrations displayed slight increases in the concentration of the small size fraction at the beginning of the observational period (0.06 µg L^–1 higher than average) and the large size fraction towards the end of the observational period (0.1 µg L^–1 higher than average) (Figure 3A). Total chl a concentrations throughout the euphotic zone ranged from 0.12 µg L^–1 near the surface on JD 238 to a maximum of 0.35 µg L^–1 at a depth of 68 m on JD 242. A deep chl a maximum was present below the mixed layer over the majority of the observation period with highest concentrations ranging from 0.30 to 0.35 µg L^–1 on JDs 240–244. These concentrations are within the historical range for this time of year and region (Harrison, 2002; and references therein). Depth-integrated total chl a areal amounts throughout the euphotic zone ranged between 13.8 and 21.2 mg m^–2 with an average total amount of 17.6 ± 2.2 mg m^–2 (n = 11; Table 2). Surface satellite chl a concentrations were highest between JDs 244 and 248 (Figure 1). However, because the maximum chl a concentrations during the observation period occurred below the mixed layer, these values are not well-represented with remote sensing observations.

Mixed layer concentrations of particulate carbon and particulate nitrogen displayed similar trends to chlorophyll a, ranging from 3.0 to 7.3 and 0.6 to 1.2 µmol L^–1, respectively, throughout the observation period, with an average (± SD) concentration of 4.9 ± 1.2 and 0.8 ± 0.2 µmol L^–1 (n = 34, each), respectively. Discrete size-fractionated PC within the mixed layer ranged from 2.7 to 5.4 µmol L^–1 in the small size fraction and from 0.2 to 2.3 µmol L^–1 in the large size fraction (Figure 3C and D). Size-fractionated PN ranged from 0.5 to 0.9 µmol L^–1 in the small fraction and from 0.1 to 0.4 µmol L^–1 in the large fraction. Both PC and PN increased in the mixed layer from the beginning of the observation period onward, with maximum concentrations occurring from JDs 240 to 246 (Table 2). Depth-integrated PC through the euphotic zone ranged from 203 to 335 mmol C m^–2 with an average of 273 ± 42.0 mmol C m^–2 (n = 11; Table 2). Depth-integrated PN ranged from 40.0 to 60.1 mmol N m^–2 with an average of 51.2 ± 6.1 mmol N m^–2 (n = 11). The average PC: PN ratio from discrete samples was 5.3 mol C mol N^–1 with slightly higher ratios near the surface (5.4 mol C mol N^–1) compared to depth (5.1 mol C mol N^–1). PC: chl a ratios varied substantially throughout the water column and over the observational period with the majority of ratios greater than 50 (Reynolds, 2006). The average total PC: Chl a ratio was 276 ± 65.4 (n = 56). The average small fraction ratio was substantially (>200) greater than that of the large fraction at 347 ± 87.2 (Figure S6) with overall highest ratios for both size fractions occurring in the top 10–15 m. However, the percentage of phytoplankton-associated PC compared to total PC as estimated using flow cytometry was low at approximately 15% (McNair et al., n.d.).

Data collected on JD 264 at Station P during the Canadian Line P cruise indicated high phytoplankton biomass concentrations coinciding with increased net primary productivity as compared to the EXPORTS observation period. Surface total chl a concentrations were 0.33 µg L^–1, which was 0.10 µg L^–1 higher than the EXPORTS average at the surface (Figure 1). Euphotic zone depth-integrated chl a concentrations were within the range observed during EXPORTS at peak biomass (JDs 240–250), at 20.2 mg m^–2 (Table 2; Figure S7). Size-fractionated chl a indicated a continued predominance of the small fraction accounting for 71% of the total chl a. However, depth-integrated PC and PN were significantly higher (≥ 9 standard deviations and ≥ 7 standard deviations for PC and PN, respectively) than the average EXPORTS concentrations. Depth-integrated PC and PN concentrations were 666 mmol C m^–2 and 93.9 mmol N m^–2, respectively, and the PC: PN ratio on JD 264 was 7.1 mol C mol N^–1 (Table 2; Figure S7).

Net primary productivity (based on ¹³C-ρDIC from 24-hr incubations) was 2.5-fold higher within the mixed layer compared to depths below the mixed layer (Figure 4). The average (± SD) mixed layer NPP was 0.5 ± 0.2 µmol L^–1 d^–1 (n = 34), reaching a maximum rate of 0.8 µmol L^–1 d^–1 on JD 240. Below the mixed layer, the average NPP was 0.2 ± 0.2 µmol L^–1 d^–1 (n = 22) with all NPP values declining appreciably by 70 m (1% of I_o), and depth-integrated NPP declining steadily through the observational period (Figures 4 and S8). Within the mixed layer, large fraction contributions to total NPP increased by 21% over the observational period. Below the mixed layer, the small fraction also consistently dominated NPP, accounting for 97% of the total DIC uptake rate. Total euphotic zone depth-integrated NPP ranged from 17.3 to 30.2 mmol C m^–2 d^–1 over the observational period (Figure 5B). Trends in total NPP were tightly coupled to those in the small fraction, accounting for 91.4% of NPP throughout the euphotic zone (Figures 4 and 6B).

Data from the Canadian Line P cruise on JD 264 collected at the EXPORTS site showed higher euphotic zone depth-integrated NPP than observed during the EXPORTS observation period, measuring 41.5 mmol C m^–2 d^–1 (Table 2; Figure S9a). However, the small fraction contribution appeared comparable at 71%, although on the low end from what was observed during EXPORTS. Higher NPP on JD 264 is consistent with the higher phytoplankton biomass observed in the region following the EXPORTS observation period.

Daily nitrate uptake rates (based on volumetric ¹⁵N-ρNO₃^– from 24-hr incubations) exhibited similar spatiotemporal trends as NPP. NO₃^– uptake rates within the mixed layer ranged from 9.5 to 92.3 nmol L^–1 d^–1 with an average (± SD) rate of 37.7 ± 21.7 nmol L^–1 d^–1 (n = 34). Below the mixed layer, NO₃^– uptake rates decreased substantially with average values 5-fold lower than within the mixed layer. Within and below the mixed layer, the small size fraction contributed approximately 72% of the total NO₃^– uptake.

Similar to NPP, volumetric new production (based on ¹⁵N-ρNO₃^– converted to carbon units) decreased from the surface to depth, displaying different patterns within and below the mixed layer as described above (Figure 6; Figure S8). Minimum mixed layer depth-integrated rates of new production were 5.3 mmol C m^–2 d^–1 on JD 234, and maximum rates were 15.2 mmol C m^–2 d^–1 on JD 240, consistent with maximum NPP (Figure 5C). New production rates were higher than regenerated production on JDs 240 and 246 (Figure 5C; Table 2). Below the mixed layer, new production was less variable with values ranging from 0.7 to 6.5 mmol C m^–2 d^–1 and exhibited a declining trend over the duration of the observation period. Similar to NPP, small cells contributed to the majority of depth-integrated new production with an average (± SD) proportion of 73.4 ± 4.8% (n = 56; Figure 5C). However, large cell contribution to new production increased by over 16% from midpoint onwards, reaching a maximum of 37% of the total new production by the end of the observation period, coincident with increases in phytoplankton biomass and nitrate assimilation rates (see below).

On the Canadian Line P cruise on JD 264, depth-integrated new production was measured at 23.8 mmol C m^–2 d^–1, which was higher than that observed during EXPORTS (Table 2; Figure 5C; Figure S9b). Similarly, the percent contribution by the small size fraction was slightly higher at 82%. This rate of new production is comparable to the peak observed on JD 240 of EXPORTS but is concomitant with increases in NPP and regenerated production.

As appears to be a consistent feature at Station P, regenerated production was largely driven by the small fraction which accounted for 95–100% of the euphotic zone depth-integrated rates (Figure 5D). Within the mixed layer, regenerated production ranged from 5.2 to 10.9 mmol C m^–2 d^–1 with an average (± SD) of 8.6 ± 1.7 mmol C m^–2 d^–1 (n = 11), and below the mixed layer, values were 1.7x lower (average 5.2 ± 1.8 mmol C m^–2 d^–1, n = 11). Total regenerated production was less variable than NPP and new production, with values ranging from 9.0 to 16.9 mmol C m^–2 d^–1 with an average rate of 13.3 ± 2.5 mmol C m^–2 d^–1 (n = 11; Table 2).

In the mixed layer, the proportion of new versus regenerated production fluctuated between JD 228 and 238, where there was a majority of regenerated production, to between JD 240 and 248, where there was a majority of new production (Figure S8). Overall, regenerated production remained the dominant form of primary production throughout the euphotic zone over the observation period. The transition to higher rates of NPP within the mixed layer between JDs 240 and 248 is likely a consequence of an increase in new production rather than a decrease in regenerated production (see further discussion below). Consistent with rates of NPP and new production, regenerated production below the mixed layer steadily decreased throughout the observation period (Figure S8).

In calculating rates of regenerated production from the Line P cruise consistent with that of the EXPORTS cruise, in the manner of Equation 1, regenerated production was 17.8 mmol C m^–2 d^–1. However, given that ¹⁵N-ρNH₄⁺ and urea uptake rates were measured on the Line P cruise, regenerated production based on these uptake rates was also calculated. The NH₄⁺ uptake- based regenerated production was higher than the maximum regenerated production observed during the EXPORTS cruise at 24.1 mmol C m^–2 d^–1 (Figure S9c). Additionally, urea uptake was comparatively low but still considerable at 13.9 mmol C m^–2 d^–1 (Figure S9d). Therefore, total regenerated production calculated as the sum of ¹⁵N-ρNH₄⁺ and ¹⁵N-ρurea during the Line P cruise was substantially higher at 38.0 mmol C m^–2 d^–1 compared to production based on Equation 1 and throughout the EXPORTS cruise (Table 2; Figure S9c and d). However, both NH₄⁺ and urea uptake rates represent potential rates as ambient NH₄⁺ and urea concentrations were not measured at sea, leading to isotopic additions >10% of ambient concentrations. Incubation experiments may therefore overestimate these uptake rates and corresponding regenerated production. Percent contributions of the small size fraction to total uptake rates were 81% and 82% for NH₄⁺-based regenerated production and urea-based regenerated production, respectively. Despite the caveats of urea uptake experiments, these rates suggest that urea uptake could be a significant portion of total regenerated production at Station P and should not be discounted (Varela and Harrison, 1999).

The different methods used for calculating f-ratios within the mixed layer both had an average total f-ratio indicating a phytoplankton system mostly performing regenerated production (Table 3). The first method (Equation 2) resulted in total discrete mixed-layer f-ratios ranging from 0.12 to 0.44 with an average (± SD) f-ratio of 0.28 ± 0.2 for the small fraction and from 0.27 to 1.66 with an average of 0.82 ± 0.3 for the large fraction (n = 34; Figure 7). Values of f-ratios >1 are likely the result of methodological caveats (see Figures S3 and S10; Table S2) and should be inferred as an almost complete reliance on new production (i.e., NO₃^_ as the dominant N source). The total cruise average f-ratio, calculated from the first method, was consistent with that of the small fraction f-ratio at 0.28 ± 0.3 (n = 34), indicating net production trends were determined by the small cells and that, on average, regenerated production accounted for 72% of total NPP, while new production accounted for 28% of total NPP. Data collected on the Line P cruise on JD 246 and calculated in this manner yielded an f-ratio of 0.62, which is higher than the f-ratios measured during EXPORTS. This difference indicates that while NPP was higher in late September, the balance between new and regenerated production was fairly consistent with what was observed between JDs 228 and 250.

Given that biological rates from JD 264 were determined from incubations over 24 hr, rather than 6 hr, the f-ratio based on Equation 2 (i.e., 0.62) may be overestimated by approximately 38% due to decoupling of carbon and nitrogen uptake in the dark over 24 hrs. By accounting for this overestimation, the f-ratio calculated from this first method would be 0.38, which is within the expected range of EXPORTS f-ratios.

The second method for calculating f-ratio (Equation 3) supports the size-fractionated trends indicated by the first method: the small cells largely utilized NH₄⁺ through regenerated production while the large cells performed higher rates of new production. However, the values were different. Small size fraction f-ratios ranged from 0.15 to 0.42, with an average (± SD) value of 0.30 ± 0.10, and large size fraction f-ratios ranged from 0.17 to 0.65 with an average value of 0.50 ± 0.17 (n = 6 for both size fractions). The second method of calculating an f-ratio on JD 264 yielded an f-ratio of 0.32, more consistent with previous f-ratios and, arguably, a more realistic value. The second method of calculating f-ratios for the EXPORTS cruise samples was applied for only two sample days coinciding with when ¹⁵N-ρNH₄⁺ measurements were performed.

Based on integrated PAR values over the course of the 6 hr-incubation period relative to daily PAR, a daily GPP estimate was obtained. The average (± SD) depth-integrated mixed layer GPP was estimated at 62.1 ± 18.7 mmol C m^–2 d^–1 (n = 11), 3.8-fold higher than the average NPP (Table 3). The GPP exhibited slightly different patterns than 24-hr ¹³C-ρDIC and ¹⁵N-ρNO₃^– with a maximum rate of 86.7 mmol C m^–2 d^–1 occurring on JD 246, 6 days after the other rate measurements. However, the percent contribution of the small and large fractions remained consistent throughout the observation period at 87% and 13%, respectively.

Average chlorophyll-normalized carbon assimilation rates (¹³C-VDIC) displayed a notable difference in magnitude between the small and large cells. The small size fraction exhibited ≥4-fold higher assimilation rates on average compared to the large size fraction. ¹³C-VDIC of the small size fraction throughout the euphotic zone ranged from 1.1 to 2.9 µmol C µg chl a^–1 d^–1 with an observational period average (± SD) rate of 1.8 ± 0.5 µmol C µg chl a^–1 d^–1 (n = 56; Figure 8A). The maximum ¹³C-VDIC was observed on JD 238, approximately 2 days before maximum ¹³C-ρDIC (Figure 8B). ¹³C-VDIC of the large size fraction ranged from 0.3 to 0.6 µmol C µg chl a^–1 d^–1 with an average of 0.4 ± 0.1 µmol C µg chl a^–1 d^–1 (n = 56), achieving a maximum rate on JD 238. Discrete samples within the mixed layer showed overall higher and more variable values ranging by as much as 1.8 µmol C µg chl a^–1 d^–1 for the small size fraction and 0.5 µmol C µg chl a^–1 d^–1 for the large size fraction. Overall, the temporal trends exhibited for ¹³C-VDIC in small cells explain a substantial portion of the ¹³C-ρDIC (r = 0.65; r² = 0.42; p < 0.001) whereas the trends exhibited by the ¹³C-VDIC in the large cells did not explain a substantial portion of the ¹³C-ρDIC variability (r = 0.36; r² = 0.13; p < 0.001).

Average total ¹³C-VDIC from the surface to 1% I_o ranged from 1.02 to 2.80 µmol C µg chl a^–1 d^–1 with an average (± SD) of 1.74 ± 0.51 µmol C µg chl a^–1 d^–1 (n = 56). Total ¹³C-VDIC was tightly coupled with the rates in small cells (r = 0.99; r²= 0.98) throughout the observational period, exhibiting a peak consistent with that of ¹³C-VDIC in small and large cells and with total ¹³C-ρDIC on JD 238. ¹³C-VDIC of the large fraction was loosely coupled with total ¹³C-VDIC over the observation period (r = 0.35; r² = 0.12). Total ¹³C-VDIC correlated positively with total ¹³C-ρDIC (r = 0.51; r² = 0.27).

For chlorophyll-normalized nitrate assimilation rates (¹⁵N-VNO₃^–), small and large cells displayed similar trends through time (r² = 0.78), with minimum values on JD 234 and maximum values on JD 240 for both size fractions. ¹⁵N-VNO₃^– in the small fraction ranged from 71.7 to 207.5 nmol N µg chl a^–1 d^–1, with an average (± SD) value of 117.5 ± 43.1 nmol N µg chl a^–1 d^–1 (n = 56; Figure 8B). ¹⁵N-VNO₃^– in the large size fraction remained consistently lower than the small size fraction with rates ranging from 35.8 to 117.6 nmol N µg chl a^–1 d^–1 and an average value of 75.4 ± 25.8 nmol N µg chl a^–1 d^–1 (n = 56). Within the mixed layer only, discrete measurements of ¹⁵N-VNO₃^– exhibited a similar pattern to ¹³C-VDIC, showing overall higher and more variable values ranging by as much as 174.9 nmol N µg chl a^–1 d^–1 and 129.9 nmol N µg chl a^–1 d^–1 for small and large cells, respectively. Unlike ¹³C-VDIC, ¹⁵N-VNO₃^– appeared substantially related to ¹⁵N-ρNO₃^– in both small (r = 0.81; r² = 0.66; p < 0.001) and large cells (r = 0.58; r² = 0.33; p < 0.001) (Figures 5B and 8C).

Total ¹⁵N-VNO₃^– ranged from 68.9 to 214.9 nmol N µg chl a^–1 d^–1, with an average (± SD) value of 116.3 ± 42.4 nmol N µg chl a^–1 d^–1 (n = 56). Total ¹⁵N-VNO₃^– exhibited similar temporal trends to ¹⁵N-VNO₃^– in both small and large cells. However, the peak total ¹⁵N-VNO₃^– occurred 2 days before the size-fractionated measurements, consistent with peak total ¹³C-VDIC on JD 238, and total ¹⁵N-VNO₃^– appeared substantially more coupled with small (r = 0.99; r² = 0.98) and large cells (r = 0.88; r² = 0.78) compared to ¹³C-VDIC. Total ¹⁵N-VNO₃^– correlated strongly with new production (r = 0.72; r² = 0.52).

Total ¹³C-VDIC and ¹⁵N-VNO₃^– from JD 264 were consistent with those observed over JDs 228–250 at 2.25 µmol C µg chl a^–1 d^–1 and 207.1 nmol N µg chl a^–1 d^–1, respectively. ¹³C-VDIC decreased dramatically through the water column from 3.26 µmol C µg chl a^–1 d^–1 within the mixed layer to 0.23 µmol C µg chl a^–1 d^–1 below the mixed layer. ¹⁵N-VNO₃^– decreased with depth even more appreciably from 305.9 nmol N µg chl a^–1 d^–1 within the mixed layer to 9.5 nmol N µg chl a^–1 d^–1 below the mixed layer.

From our size-fractionated measurements of chl a, ¹³C-ρDIC and ¹⁵N-ρNO₃^–, the percent standard deviation of the mean (PSDM) can be calculated to obtain an estimate of sample variability and data uncertainty. With the exception of ¹³C-ρDIC in the large fraction, the average PSDMs were <14%. The small size fraction chl a PSDM estimates ranged from 3.7 to 13.6% with an average of 7.2% (n = 3 per depth; Table S2). The large size fraction chl a PSDMs were higher, ranging from 6.6 to 25.9%, with an average of 13.4% (n = 3 per depth; Table S2). Size-fractionated uptake rate variability was generally higher. Average PSDMs for the small size fraction ¹³C-ρDIC was 7.1% with 82% of the daily average values falling between 4.7 and 7.1% (n = 2 per depth). PSDM for the large size fraction ¹³C-ρDIC was higher at 31.3% and showed more daily variability with values ranging from 8.0 to 62.6% (n = 2 per depth; Table S2). Average PSDM for the small size fraction ¹⁵N-ρNO₃^– were similar to those of ¹³C-ρDIC at 6.9% with values ranging between 5.7 and 14.0% (n = 2 per depth). Large size fraction PSDM for ¹⁵N-ρNO₃^– were less than those of ¹³C-ρDIC at 8.3% with values ranging from 5.1 to 22.1% (n = 2 per depth; Table S2). Higher variability within the large size fraction uptake rate measurements was anticipated as a result of methodological procedures where large cells needed to be removed from the polycarbonate filter and transferred onto a pre-combusted GF/F filter for onshore analysis, possibly resulting in some loss of cells.

While PSDM estimates for total ¹³C-ρDIC and ¹⁵N-ρNO₃^– could not be estimated due to collection of only one replicate per sampling depth, the general agreement between total values and summed size-fractionated values was high at r² = 0.96 and r² = 0.98 for ¹³C-ρDIC and ¹⁵N-ρNO₃^–, respectively (Figure S2). At higher uptake rates, total values were typically higher than the summed size-fractionated values due to an artifact of some large cell loss during the transferring process between filters as described previously.

There is a long history of observations at Station P with a record of net primary productivity estimates dating back to the 1960s. Over this period, several notable changes have occurred in NPP, with an initial trend towards enhanced net primary productivity including two appreciable increases between the 1960–1970s and the 1990s (of 9 mol C m^–2 yr^–1; Table 4; Welschmeyer et al., 1993; Wong et al., 1995) and between the mid-1980s/early 1990s and 1997 (5 mol C m^–2 yr^–1; Table 4; Welschmeyer et al., 1993; Wong et al., 1995; Boyd and Harrison, 1999). However, during the SERIES iron enrichment experiment in 2002, NPP was substantially lower (by 14 mol C m^–2 yr^–1; Table 4; Harrison, 2002; Marchetti et al., 2006b). When extrapolating our average daily rate of NPP to an estimated annual rate (assuming our summer rates are 2x as high as winter rates) a value of 6.3 mol C m^–2 yr^–1 was obtained, which is a reduction of 11.7 mol C m^–2 yr^–1 from peak rates measured in 1997 (Table 4). This type of simple extrapolation smooths subseasonal and spatial variability but suggests that our annual NPP estimates are most in line with those values measured in the 1960–1970s and the 1980s being approximately 1 mol C m^–2 yr^–1 and 3 mol C m^–2 yr^–1 higher, respectively. The Line P cruise occupation of the EXPORTS site, sampled 2 weeks after the EXPORTS cruise, detected rates 10 mmol m^–2 d^–1 higher than our highest rates, highlighting the spatial variability of productivity and phytoplankton biomass when evaluated outside of the Lagrangian framework.

Historical rates of gross primary production at Station P are far less common than those of NPP. However, our average (± SD) yearly GPP of 17.0 ± 5.1 mol C m^–2 yr^–1 (n = 11) and GPP: NPP ratio of 3.8 is higher than the GPP of 11.7 mol C m^–2 yr^–1 and the GPP: NPP ratio of 2.3 reported by Wong et al. (1995) averaged over the period of 1960–1976 (Table 4). Our GPP: NPP ratio is higher than the canonical 2.7 (Halsey and Jones, 2015; and references therein), suggesting inefficient assimilation of organic carbon into phytoplankton biomass and, potentially, the development of specialized physiological traits to deal with iron limitation at Station P (Halsey and Jones, 2015). Additionally, PC: chl a ratios that are up to 10 times larger than the canonical 50:1 ratio, while partially inflated due to high concentrations of detritus and other non-algal PC sources (Reynolds, 2006), are suggestive of physiological stress and/or iron limitation (Sunda and Huntsman, 1997; Figure S6).

Station P has been characterized historically as a region dominated by regenerated production with typical f-ratios between 0.25 and 0.32 (Varela and Harrison, 1999; Peña and Varela, 2007). Our observation period average f-ratio of 0.28 is consistent with a predominance of regenerated production and low carbon export efficiency (13 ± 5%, n = 12; Buesseler et al., 2020a). However, our high-resolution sampling and size-fractionated measurements revealed that while the overall f-ratio was low, f-ratios varied through the water column spatially and temporally, and ratios were differentiated by size class with large cells having a much higher average f-ratio, indicating an almost complete dependency on nitrate-supported new production. The contributions of ammonium, urea and nitrate uptake measured from the Line P cruise were 39%, 23%, and 38%, respectively. These percent contributions to total primary productivity were similar to utilization rates and relative preference indices previously measured at Station P, at 50%, 23%, and 27%, respectively (Varela and Harrison, 1999) and suggest that despite higher rates of NPP observed during the Line P cruise, the overall balance of new versus regenerated nitrogen use at Station P remained fairly consistent across space and time. When extrapolating our average daily rates to annual rates, accounting for seasonality, rates of new production from 1992 to 1994 (2.8 mol C m^–2 yr^–1) were comparable to our rates of 2.7 mol C m^–2 yr^–1 (Table 4; Varela and Harrison, 1999). Emerson (2014) reported a similar average net community production from a variety of different methods including steady-state new production estimates of 2.3 mol C m^–2 yr^–1 (Table 4). The consistency between f-ratios and rates of new production across studies indicates that the balance between new production and regenerated production on an annual basis has remained fairly consistent over time. Additionally, numerous measurements of NCP derived from the EXPORTS cruise are comparable to our rates of new production (Niebergall et al., n.d.), verifying a predominantly steady-state system and minimal amount of nitrification within the mixed layer during our observation period (Dugdale and Goering, 1967; Eppley and Peterson, 1979; A Santoro, unpublished data). Measurements of NCP from the Line P cruise were comparable to measurements of new production at 13.5 mmol C m^–2 d^–1 and 11.4 mmol C m^–2 d^–1, respectively, thus further supporting a mostly steady-state system (R Hamme, personal communication, 2021).

While our data clearly indicate a system dominated by small-celled phytoplankton fueled by regenerated production, the increase in net primary production, new production, and large-cell contributions leading up to JD 240 is best explained by a physiological response to some ecosystem forcing, likely a change in iron supply or light availability. The consistency of increases and peaks of net primary production and new production with high nitrate assimilation rates from JD 234 to 240 suggest that more new production in both the small and large cells drove this positive deviation from baseline rates of net primary production with both small-celled and large-celled nitrate assimilation rates increasing during this time. Given the high concentrations of ambient NO₃^–, NH₄⁺, phosphate, and silicate throughout the cruise (Siegel et al., 2021) and the classification of Station P as an HNLC region, the observed physiological response likely resulted from an ephemeral relax in degree of iron limitation consistent with the increase in production observed from JD 234 to 240.

Recent studies have shown that picoeukaryotic phytoplankton can exhibit similar responses (increased growth and assimilation rates) to temporary alleviation of iron limitation as do larger-sized phytoplankton such as diatoms (Botebol et al., 2017). The high rates of nitrate assimilation among large cells during the peak production period is particularly noteworthy because this increased rate led to the highest percent contribution to net primary productivity in large cells for the entire observation period, with percent contribution increasing from 28.3% to 37.0% from JD 240 to 250. Nitrate uptake rates and f-ratios exhibited differential trends between the size fractions with significant differences for the small size fraction (p < 0.001, r = 0.69; r² = 0.50) and large fraction (p < 0.001, r = 0.50; r² = 0.22; Figure 9). Small cells performed the majority of nitrate uptake, yet this uptake only accounted for a small proportion of their total production, as indicated by their low f-ratios. Despite f-ratios on average being high (>0.5) for the large cells, the low percent contribution of large cells to total nitrate uptake prevented deviation from the dominant form of production of the total community, i.e. the regenerated production performed primarily by small cells. Small-celled and large-celled nitrate assimilation rates had an average difference in magnitude of 0.33 µmol N µg chl a^–1 d^–1 (only 0.01 µmol N µg chl a^–1 d^–1 more than one standard deviation) but exhibited similar temporal trends, suggesting that changes in environmental conditions were driving shifts in new production.

Iron limitation exerts significant control on the phytoplankton assemblage in the Northeast subarctic Pacific Ocean, particularly in larger cells (Martin and Fitzwater, 1988; Boyd and Harrison, 1999; Marchetti et al., 2006a; Marchetti et al., 2006b). While small cells engage in regenerated production, in part due to the lower iron requirement of NH₄⁺ uptake (Price et al., 1994; Boyd and Harrison, 1999), diatoms exhibit reduction strategies for iron requirement (such as utilizing the iron-free proteins flavodoxin or plastocyanin), but overall, large-sized phytoplankton cell densities remain low (La Roche et al., 1993; McKay et al., 1997; Peers and Price, 2006). When iron limitation is alleviated, the phytoplankton assemblage transitions from a community predominantly of cyanobacteria and nanoflagellates to one dominated by diatoms, which have been observed to upregulate nitrogen assimilation genes and subsequently grow and assimilate nitrate faster than other phytoplankton groups (Marchetti et al., 2012; Lampe et al., 2021). The observed increase in percent contribution of new production by large cells during EXPORTS may be representative of the ecological response to natural fluctuations in iron bioavailability as observed in other HNLC regions.

Temporary increases in bioavailable growth-limiting nutrient concentrations are often not detected from in situ measurements due to the rapid rates at which they are taken up and assimilated by phytoplankton (Goldman et al., 1979). Increases in bioavailable iron are correlated with increases in PAR which can affect the photolability of iron speciation (Anderson and Morel, 1982). However, a correlation between carbon and nitrate assimilation rates in both size fractions with integrated daily PAR during the observation period was not apparent, as would be expected if photolability were a significant driver of changes in iron speciation (Figure 10; Table S3). Previous studies have indicated that small increases (≥0.1 nM) in bioavailable iron concentrations can result in a physiological response (i.e., increases in C and N assimilation rates) from phytoplankton that may not be large enough to result in substantial shifts in the phytoplankton assemblage (Hutchins and Boyd, 2016).

Grazing rates and shifts in zooplankton assemblages indicate that top-down control plays an important role in maintaining the small-celled phytoplankton standing stock in the NE Pacific Ocean (Landry et al., 1993; McNair et al., n.d.). The temporal patterns in nitrate assimilation for both small and large-sized phytoplankton suggest that the variability in net primary production trends over the course of our observation period were driven by physiological changes due to increased iron bioavailability as well as a possible alleviation of top-down control via grazing pressure. The stronger positive correlation between new production and nitrate assimilation of large cells and weaker correlation between NPP and carbon assimilation in this size fraction may indicate that these phytoplankton performing new production are not being grazed as rapidly as those performing regenerated production. The concept of small phytoplankton groups predominantly performing regenerated production being controlled largely by top-down forcing and large phytoplankton groups predominantly performing new production being controlled largely by bottom-up forcing (specifically Fe limitation) is consistent with previous findings and is referred to as the Ecumenical Iron Hypothesis (Morel et al., 1991; Landry et al., 1993; Muggli et al., 1996). Our observed patterns of new production during EXPORTS are consistent with the hypothesis of generally low rates of new production in HNLC regions being supported by recycled iron (Rafter et al., 2017) and highlight the importance of ammonium utilization even at high ambient nitrate concentrations (Wan et al., 2018).

Despite our low average NPP compared to global estimates (Henson et al., 2011; Li and Cassar, 2018), we measured appreciable fluctuations in primary production over the observational period. The sequential increase in carbon and nitrate assimilation rates, new production and NPP from JD 234 to 240 indicates a period of environmental change resulting in a physiological response by phytoplankton culminating in an increase in phytoplankton biomass. The initiation of the phytoplankton response coincided with a deepening of the mixed layer (by 8.4 m since the beginning of the observational period; Figure 5A). The concomitant increase of these parameters and the maintenance of a low f-ratio suggest that deviations from the ecological baseline NPP trends (identified here as JDs 232–240) are correlated to increased rates of new production, a relationship that we found to be significant (r = 0.9; r² = 0.80; p < 0.001). This finding suggests that the most likely driver for this increase in new production is a change in bioavailable iron preceding the beginning of this period of increased productivity, which coincided with the deepening of the mixed layer. The Canadian Line P data measured on JD 264 support the classification of Station P as a largely small cell dominated, recycled nutrient system despite temporally variable rates of NPP. The overall consistency of our NPP, new production, and f-ratios with historical data suggests that the variability in production that we observed over the course of EXPORTS is likely part of a natural cycle of biological variability in this low iron region.

Consistent with previous studies (Boyd and Harrison, 1999; Harrison, 2002; Marchetti et al., 2006a; Marchetti et al., 2006b; Marchetti et al., 2006c), our results indicate that general bottom-up controls (i.e., phytoplankton growth regulated by light, temperature, nutrient availability, etc.) on primary productivity trends are determined by the metabolic activities of the small cells which engage primarily in regenerated production. The low f-ratios and dependence on recycled forms of nitrogen in this region have significant implications for the food web structure and biological carbon export potential, which are regulated primarily by efficient grazing (Price et al., 1994; McNair and Menden-Deuer, 2020), higher rates of bacterial respiration (Hutchins et al., 1993; Guidi et al., 2016; C Carlson and B Stephens, personal communication, 2021), and low-to-modest transfer efficiency of carbon export to depth (Mouw et al., 2016; Buesseler et al., 2020b; McNair et al., n.d.). As observed in our study, the evidence of the significant role that small-sized phytoplankton could play in new production and carbon export, particularly in open ocean regions, is growing (e.g., Juranek et al., 2020). As global climate change continues to warm the oceans, a transition toward more ecosystems characterized by small-celled phytoplankton is projected (Rousseaux and Gregg, 2015). The proportions of primary production performed by the different phytoplankton taxa we observed during EXPORTS may become more common place over time, leading to an expansion of low efficiency carbon export regions dominated by small phytoplankton that are punctuated by episodes of enhanced new production performed by both small and large phytoplankton.

All EXPORTS data presented here are available at the SeaWiFS Bio-optical Archive and Storage System (SEABASS; seabass.gsfc.nasa.gov; doi: 10.5067/SeaBASS/EXPORTS/DATA001). All Line P data presented here are available through the University of Victoria Scholars Portal Dataverse (https://dataverse.scholarsportal.info/; doi.org/10.5683/SP3/NSZMKI).

The supplemental files for this article can be found as follows:

Figures S1–S11. Tables S1–S3. Docx

We would like to thank the EXPORTS project leaders, D. Seigel and I. Cetinic, R/V Roger Revelle chief scientists D. Steinberg and J. Graff, the captain and crew of the R/V Revelle, and other members of the EXPORTS science team. In particular, we thank J. Fox for estimates of the daily light saturation parameter, A. Fassbender for DIC concentrations, the EXPORTS Hydro team for nutrient data, K. Buck and S. Burns for dissolved iron data, and R. Hamme for estimates of NCP from the Line P cruise.

EXPORTS work was supported by NASA Grant 80NSSC17K0552 to AM, SG, and NC and NASA Grant 80NSSC18K1431 to AS. Line P work was supported by a Discovery Grant Individual from the Natural Sciences and Engineering Research Council (NSERC) of Canada to DV and an NSERC Undergraduate Student Research Award (USRA) to SK.

The authors declare no competing interests.

Contributed to concept and design: AM, SG, NC.

Contributed to acquisition of data: AM, SG, NC, WG, SMK, OT, DEV, AES, AN, GS.

Contributed to analysis and interpretation of data: MGM, AM, WG, SMK, DEV, AN.

Drafted and/or revised the article: MGM, AM, WG, SMK, OT, DEV, AES, NC, SG, AN, GS.

Approved the submitted version for publication: MGM, WG, SMK, OT, DEV, AES, NC, SG, AN, GS, AM.

How to cite this article: Meyer, MG, Gong, W, Kafrissen, SM, Torano, O, Varela, DE, Santoro, AE, Cassar, N, Gifford, S, Niebergall, AK, Sharpe, G, Marchetti, A. 2022. Phytoplankton size-class contributions to new and regenerated production during the EXPORTS Northeast Pacific Ocean field deployment. Elementa: Science of the Anthropocene 10(1). DOI: https://doi.org/10.1525/elementa.2021.00068

Domain Editor-in-Chief: Jody W. Deming, University of Washington, Seattle, WA, USA

Associate Editor: Jean-Éric Tremblay, Department of Biology, Université Laval, Québec, Canada

Knowledge Domain: Ocean Science

Part of an Elementa Special Feature: Accomplishments from the EXport Processes in the Ocean from RemoTe Sensing (EXPORTS) Field Campaign to the Northeast Pacific Ocean