3.2.1 Station ALOHA

The climatology of the residual preNO₃ tracer is presented for the upper 250 m at Station ALOHA in Fig. 1a. Residual preNO₃ varies between −1 and 1 µM in the upper ∼200 m, with increasingly positive values below 200 m. The resulting rNPN anomaly (blue colors) is a persistent feature within the γ_n=24.2–25.2 neutral-density layer at a depth of ∼100 to 200 m. Seasonal rPPN anomalies (red colors) are observed immediately above the γ_n=24.2 neutral-density horizon within the euphotic zone, 0 to ∼100 m. Pulses of rPPN anomalies penetrating the bottom of the rNPN anomaly layer are also observed that coincide with shoaling of the γ_n=25.2 neutral-density horizon above 200 m.

The monthly averaged climatology of residual preNO₃ at Station ALOHA for all data in the years 1989–2016 is presented in Fig. 1b. The subsurface rNPN anomaly is observed to grow in magnitude between the months of May/June and October/November. There also exists an increase in the magnitude of the rPPN anomaly within the euphotic zone (0–100 m), concomitant with the rNPN anomaly formation at ∼100–200 m depth in summer.

Volumetric rates of rNPN anomaly formation at Station ALOHA are estimated at 2.8±1.4 and 3.0±1.5 µmol N m⁻³ d⁻¹ for the γ_n=24.2–24.7 layer using r_POM values of 6.9 and 10.6, respectively (Table 1). Slightly lower rates of rNPN anomaly formation of 1.6±0.8 and 2.5±1.4 µmol N m⁻³ d⁻¹ are estimated for the deeper γ_n=24.7–25.2 layer at Station ALOHA (annual regressions are provided in Figs. S1 and S2). Depth- and time-integrated, the estimates of rNPN anomaly formation become 28.3±9.6 (r_POM=6.9) and 17.9±7.4 (r_POM=10.6) mmol N m⁻² yr⁻¹ for the γ_n=24.2–24.7 layer and 18.1±8.8 (r_POM=6.9) and 13.7±7.8 (r_POM=10.6) mmol N m⁻² yr⁻¹ for the deeper γ_n=24.7–25.2 layer. Total rNPN formation rate estimates for Station ALOHA are between 31.6±10.8 and 46.4±13.0 mmol N m⁻² yr⁻¹ (Table 1).

Figure 1(a) Climatology of the residual preNO₃ tracer [µM] in the upper 250 m at Station ALOHA (22.75^∘ N, 158^∘ W) for a value of r_POM=10.6. Black contour lines show neutral density, γ_n=24.2, 24.7, and 25.2. (b) The monthly averaged residual preNO₃ [µM] climatology for all data in the years 1989–2016.

Download

Figure 2(a) Climatology of the residual preNO₃ tracer [µM] in the upper 200 m at the BATS station (31.67^∘ N, 64.17^∘ W) for a value of r_POM=10.6. Black contour lines show neutral density, γ_n=25.8 and 26.3. (b) The monthly averaged residual preNO₃ [µM] climatology for all data in the years 1993–2016.

Download

Table 1Residual negative preNO₃ (rNPN) and residual positive preNO₃ (rPPN) anomaly formation rates at the stations ALOHA and BATS.

Download Print Version | Download XLSX

The volumetric rate of rPPN anomaly formation within the euphotic zone at Station ALOHA is estimated at 3.3±1.1 and 2.4±0.8 µmol N m⁻³ d⁻¹ using r_POM values of 6.9 and 10.6, respectively (Table 1, Fig. S3). These rates are approximately equivalent within error of those estimated for the rNPN formation rate within the γ_n=24.2–24.7 layer immediately below in the water column. Depth- and time-integrated, the estimate of rPPN anomaly formation is 61.2±20.2 (r_POM=6.9) and 43.5±10.5 (r_POM=10.6) mmol N m⁻² yr⁻¹ (Table 1). The euphotic-zone-integrated rPPN anomaly is approximately 33 % higher than the estimated total integrated rNPN anomaly within the combined γ_n=24.2–25.2 layer.

3.2.2 BATS

The climatology of the residual preNO₃ tracer in the upper 200 m at BATS is presented in Fig. 2a. Residual preNO₃ varies between −1 and 1 µM in the upper 200 m, and seasonal subsurface rNPN anomalies at ∼80 to 160 m depth are observed in most but not all years from 1993 to 2016. Similar to Station ALOHA, seasonal pulses of rPPN anomalies are observed immediately above the γ_n=25.8 neutral-density horizon within the euphotic zone, 0 to ∼80 m. Pulses of rPPN anomalies penetrating from below the γ_n=26.3 neutral-density horizon are present in a few years but are much less frequent than observed in the Station ALOHA climatology. rPPN anomalies are observed immediately below the rNPN layer beginning at ∼160 m.

The monthly averaged climatology of residual preNO₃ at BATS for all data in the years 1993–2016 is presented in Fig. 2b. The subsurface rNPN anomaly is observed to first appear beginning in April/May and grow in magnitude through the end of the calendar year. The rNPN feature starts shallow at a depth of ∼60 to 100 m, which deepens over the summer months to depths of ∼80 to 160 m by year's end. There also exists a layer of rPPN anomaly at 0 to ∼80 m depth that increases in magnitude, concomitant with the rNPN anomaly formation at ∼80–160 m depth in summer and autumn. Late-winter convective overturn of the upper water column in late January/February at the BATS station (Hansell and Carlson, 2001) erases both the subsurface rNPN and 0–80  m rPPN anomaly.

The volumetric rate of rNPN anomaly formation at the BATS station is estimated at 5.5±2.7 and 3.8±3.1 µmol N m⁻³ d⁻¹ for the γ_n=25.8–26.3 layer using r_POM values of 6.9 and 10.6, respectively (Table 1, Fig. S4). Depth- and time-integrated, the estimate of the rNPN anomaly formation rate at the BATS station is between 46.0±39.3 and 87.1±41.0 mmol N m⁻² yr⁻¹ (Table 1). The volumetric rate of rPPN anomaly formation within the euphotic zone at the BATS station is estimated at 5.8±1.2 and 4.1±0.8 µmol N m⁻³ d⁻¹ using r_POM values of 6.9 and 10.6, respectively (Table 1, Fig. S5). These rates are approximately equivalent within error of those estimated for the rNPN formation rate within the γ_n=25.8–26.3 layer immediately below in the water column. Depth- and time-integrated, the estimate of rPPN anomaly formation is 82.1±13.8 (r_POM=6.9) and 61.8±12.2 (r_POM=10.6) mmol N m⁻² yr⁻¹ (Table 1). The euphotic-zone-integrated rPPN anomaly is approximately balanced (r_POM=6.9) or 33 % higher (r_POM=10.6) than the estimated integrated rNPN anomaly within the γ_n=25.8–26.3 layer.

Figure 3(a) Climatology of the residual prePO₄ tracer [µM] in the upper 250 m at Station ALOHA (22.75^∘ N, 158^∘ W) for a value of r_POM=10.6. Black contour lines show neutral density, γ_n=24.2, 24.7, and 25.2. (b) The monthly averaged residual prePO₄ [µM] climatology for all data in the years 1989–2016.

Download

Figure 4(a) Climatology of the residual prePO₄ tracer [µM] in the upper 200 m at the BATS station (31.67^∘ N, 64.17^∘ W) for a value of r_POM=10.6. Black contour lines show neutral density, γ_n=25.8 and 26.3. (b) The monthly averaged residual prePO₄ [µM] climatology for all data in the years 1993–2016.

Download

3.3.1 Station ALOHA

The climatology of the residual prePO₄ tracer is presented for the upper 250 m at Station ALOHA in Fig. 3a. Residual prePO₄ varies between −0.04 and 0.2 µM with positive values observed throughout the upper 250 m of the time series with exception of a period exhibiting small (∼0 to −0.04 µM) residual negative prePO₄ anomalies at ∼125–175 m in the years 2002–2004. The monthly averaged climatology of residual prePO₄ at Station ALOHA for all data in the years 1989–2016 is presented in Fig. 3b. Positive residual prePO₄ (0–0.2 µM) is observed throughout the upper 250 m within the monthly averaged climatology.

3.3.2 BATS

The climatology of the residual prePO₄ tracer in the upper 200 m at BATS is presented in Fig. 4a. Residual prePO₄ varies between −0.2 and 0.2 µM in the upper 200 m. Seasonal subsurface residual negative prePO₄ anomalies at ∼80 to 160 m depth are observed in all years from 1993 to 2016 with the exception of 2015. Seasonal pulses of residual positive prePO₄ anomalies (∼0–0.1 µM) are observed immediately above the γ_n=25.8 neutral-density horizon within the euphotic zone, 0 to ∼80 m.

The monthly averaged climatology of residual prePO₄ at BATS for all data in the years 1993–2016 is presented in Fig. 4b. The subsurface residual prePO₄ anomaly is present throughout the seasonal cycle although it is observed to grow in magnitude beginning in May continuing through the end of the calendar year into January/February, before springtime vertical overturning circulation resets the residual prePO₄ concentration in the upper 200 m. The residual negative prePO₄ feature starts shallow at a depth of ∼80 m in May/June, which deepens over the summer months to depths of ∼80 to 200 m by year's end. There also exists a layer of residual positive prePO₄ anomaly at 0 to ∼80 m depth that develops during April–November, concomitant with the seasonally increasing residual negative prePO₄ anomaly at ∼80–200 m depth in summer and autumn.

4.2.1 TEPs

Transparent exopolymer particles are the polysaccharide-rich exudate of phytoplankton that accumulate in the size range <1 to >100 µm in aquatic systems (Mari et al., 2017). TEPs are both sticky and positively buoyant in seawater (Azetsu-Scotte and Passow, 2004), leading to aggregation and flotation towards the surface, with large enrichments of TEPs present in the sea surface microlayer (Wurl et al., 2009). Being comprised of nearly pure saccharide material, TEPs are a carbon-rich and essentially N-deficient pool of non-sinking particulate organic carbon formed within the euphotic zone of marine systems by phytoplankton production (Alldredge et al., 1993). TEP production has been hypothesized to contribute to “carbon overconsumption” in low-nutrient oligotrophic marine ecosystems (Toggweiler, 1993) where organic matter is produced in non-Redfield, C-rich or N-poor proportions. Due to their positive buoyancy (Mari et al., 2017), TEPs have been viewed as a non-contributor to upper-ocean carbon export; however, some portion of the TEP pool is associated with sufficient ballasting particles (e.g., clays, biogenic minerals) to achieve negative buoyancy and may comprise a slowly sinking (a few meters per year) pool of organic carbon exported below the euphotic zone (Mari et al., 2017). TEP production and remineralization stoichiometry has the correct sense – O₂ production without Redfieldian nitrate drawdown within the euphotic zone and O₂ consumption without Redfieldian nitrate accumulation within the mesopelagic layer – to contribute to both the observed rPPN anomalies within the euphotic zone and rNPN anomalies in the upper mesopelagic layer at the stations ALOHA and BATS.

We use field data of TEP concentrations near the stations BATS and ALOHA and a few simplifying assumptions to test for the importance of this process as a contributor to the dual rNPN and rPPN anomalies. Our estimates of the potential for TEP cycling to contribute to the observed euphotic-zone rPPN and subsurface rNPN anomaly formation rates assume (1) that TEPs are pure carbohydrate (CH₂O) without any N content, such that TEP production/remineralization C:O₂ stoichiometry can be assumed 1 : 1 (e.g., ${\mathrm{CO}}_{\mathrm{2}}+{\mathrm{H}}_{\mathrm{2}}\mathrm{O}\iff {\mathrm{CH}}_{\mathrm{2}}\mathrm{O}+{\mathrm{O}}_{\mathrm{2}}$), and (2) that there exists a sufficiently large and dense TEP pool to sink rapidly enough to contribute to annual carbon export and O₂ remineralization budgets. Both assumptions, if incorrect, will overestimate the TEP contribution. In conjunction with TEP cycling, Coomassie-stainable particles are transparent protein-containing particles (Long and Azam, 1996) that can reach similar concentrations within the oligotrophic euphotic zone as TEPs (Cisternas-Novoa et al., 2015). Thus, the assumption that transparent exopolymer particle cycling delivers only C to the subsurface to drive rNPN anomaly formation may not be entirely valid. In addition, Mari et al. (2017) note that elevated TEP concentrations likely enhance material retention in the euphotic zone. Thus, our calculations that follow tend to maximize the contribution of TEPs.

Cisternas-Novoa et al. (2015) measured TEP concentration depth profiles in the Sargasso Sea northeast of Bermuda on five separate occasions from February 2012 to June 2013. The profiles show a ∼10 µg XG eq. L⁻¹ (xanthan gum equivalent) concentration excess between the upper 100 m and the 100–200 m layer (Fig. 15 in Cisternas-Novoa et al., 2015). We take this upper 100 m excess to represent the fraction of the euphotic-zone TEP pool that is exported annually to the upper mesopelagic 100–200 m depth layer by either physical sinking of a fraction of the TEP pool that becomes negatively buoyant or being delivered to the subsurface with wintertime vertical mixing. This assumption is supported by the observation that the euphotic-zone-to-mesopelagic TEP concentration gradient is nearly erased in the winter profile, presumably due to wintertime convective mixing, and reappears following the spring bloom, remaining relatively unchanged throughout the late-spring/summer month profiles. TEPs in xanthan gum equivalents can be converted to µg C L⁻¹ units using a 0.63 conversion factor (Engel, 2004) and again into µM units, yielding a Sargasso Sea upper-ocean concentration excess of ∼0.5 µM TEP-C. Integrating over the surface to 80 m depth layer (the depth where rPPN anomalies switch to rNPN anomalies), we obtain a potential TEP production and export flux of 42 mmol TEP-C m⁻² yr⁻¹ from the euphotic zone into the upper mesopelagic layer. Assuming 1 : 1 C:O₂ TEP production/remineralization stoichiometry, an O₂ production flux in the absence of nitrate drawdown of 42 mmol O₂ m⁻² yr⁻¹ can be ascribed to TEP production within the euphotic zone, and the same flux can be ascribed to O₂ consumption without concomitant nitrate accumulation from TEP remineralization in the mesopelagic layer. We can convert this O₂ flux to a residual preNO₃ equivalent using Eq. (1) and the values in Table 1, yielding estimates of 3.2–4.5 mmol N m⁻² yr⁻¹ residual preNO₃ anomaly formation rate equivalents, depending on the choice of r_POM. Thus, TEP formation within the euphotic zone of the Sargasso Sea has the potential to explain 5.2 %–5.4 % of the estimated rPPN anomaly formation rate and 5.1 %–7.0 % of the estimated rNPN anomaly formation rate within the γ_n=25.8–26.3 layer (Table 2).

The potential contribution of TEP cycling to rPPN and rNPN anomaly formation near the BATS station can be also be estimated using published sinking rates for negatively buoyant TEPs. Again, using the TEP profiles from the Sargasso Sea from Cisternas-Novoa et al. (2015), we estimate a concentration difference of ∼5 µg XG eq. L⁻¹ between 100 and 200 m depth (the approximate depths exhibiting rNPN anomalies). Converting to molar carbon concentration yields a quantity of ∼0.26 µM TEP-C that is apparently remineralized within the rNPN anomaly layer. We apply the TEP sinking rate of 0.04 d⁻¹ from Hamanaka et al. (2002) to estimate the rate of delivery of exported TEPs from the euphotic zone into the upper mesopelagic layer, yielding a TEP supply and remineralization flux of 10 µmol C m⁻³ d⁻¹. Depth- and time-integrated to 15 m and 1 year, i.e., the depth slowly sinking TEPs will sink after 1 year, this flux becomes 56.9 mmol C m⁻² yr⁻¹ or O₂ units assuming 1 : 1 C:O₂ stoichiometry for TEPs. Conversion of this O₂ flux into residual preNO₃ equivalents yields an estimated 4.3–6.0 mmol N m⁻² yr⁻¹ of the observed rPPN and rNPN anomaly formation rate attributable to TEP cycling. Thus, computed using the sinking rate, TEP formation within the euphotic zone has the potential to explain 7.0 %–7.3 % of the estimated rPPN anomaly formation rate and 6.9 %–9.3 % of the estimated rNPN anomaly formation rate within the γ_n=25.8–26.3 layer.

Seasonal depth profiles of TEP concentration are unavailable for the subtropical North Pacific near Station ALOHA. Wurl et al. (2011) measured TEP concentration profiles south of the island of Hawaii during August/September 2009. We use the upper-ocean concentration excess present in these profiles with the TEP sinking rate of 0.04 d⁻¹ to estimate the TEP cycling contribution to rPPN and rNPN anomaly formation rates near Station ALOHA. The 100–200 m concentration excess in TEP-C is ∼1 µM, yielding an export and supply flux of 40 µmol C m⁻³ d⁻¹. With depth and time integration for 15 m and 1 year, this flux becomes 220 mmol C m⁻² yr⁻¹ (mmol O₂ m⁻² yr⁻¹). Conversion of this O₂ flux to residual preNO₃ equivalents using Eq. (1) and values in Table 1 yields an estimated 16.2–21.8 mmol N m⁻² yr⁻¹ of the observed rPPN and rNPN anomaly formation rate potentially attributable to TEP cycling at Station ALOHA. Thus, TEP formation within the euphotic zone has the potential to explain 35.6 %–37.2 % of the estimated rPPN anomaly formation rate and 47.0 %–51.3 % of the total estimated rNPN anomaly formation rate within the combined γ_n=24.2–25.2 layer (Table 2).

4.2.2 Bacterial nitrate uptake

Multiple studies have shown the uptake of nitrate and/or phosphate by heterotrophic bacteria during the remineralization of organic matter or that inorganic nutrients can limit bacterial OM consumption (e.g., Zweifel et al., 1993; Kirchman, 1994; Cotner et al., 1997; Rivkin and Anderson, 1997; Caron et al., 2000; Letscher et al., 2015). Bacterial nitrate uptake to remineralize OM has the effect of creating an NPN anomaly (O₂ consumption without concomitant nitrate accumulation) and thus can only contribute to the observed subsurface rNPN anomaly formation. It cannot contribute to either the formation of the euphotic-zone rPPN anomaly or surface mixed-layer drawdown of DIC. If TEP cycling is an important contributor to oligotrophic ocean C export and shallow subsurface O₂ consumption, heterotrophic bacterial nitrate and/or phosphate uptake is a likely complementary biological process, since the nutrient-deficient content of TEPs could require exogenous uptake of N and P from the seawater media for bacterial production when growing on TEP organic matter. This would result in an imbalance between rPPN and rNPN, in contrast to the results presented earlier.

To assess the potential contribution of bacterial nitrate uptake to our observed rNPN anomaly formation rates at the stations ALOHA and BATS, we examined the literature for estimates of bacterial C production rates and bacterial biomass C:N ratios for the Sargasso Sea and subtropical North Pacific. Carlson et al. (1996) measured bacterial production rates at the BATS station using the 3H-thymidine and 3H-leucine uptake methods with average euphotic-zone rates of 23 pmol C L⁻¹ d⁻¹ by 3H-thymidine and 0.41 nmol C L⁻¹ d⁻¹ by 3H-leucine. Using a bacterial biomass C:N of 5:1 for this region (Gunderson et al. 2002) and integrating for 1 year and over the ∼80–160 m thick layer exhibiting rNPN anomaly yields an estimate of 0.17–3.0 mmol N m⁻² yr⁻¹ bacterial N demand. If we assume that bacteria satisfy all of their N demand via uptake of seawater nitrate in this shallow subsurface layer, bacterial nitrate uptake can explain at most 0.5 %–3.5 % of the estimated rNPN anomaly formation rate at the BATS station in the γ_n=25.8–26.3 layer. This proposed mechanism still requires a process that removes bacterial N from the γ_n=25.8–26.3 layer such that a rNPN anomaly is observed due to NO₃ uptake in the absence of stoichiometric O₂ accumulation. Diel vertically migrating grazers are a candidate mechanism whereby grazing on bacterial biomass and its eventual remineralization to NO₃ is spatially separated from respiration and O₂ consumption.

We can make a similar calculation for Station ALOHA using published 3H-leucine bacterial production rate measurements (Church et al., 2004). The mean bacterial C production rate at 100 m is ∼0.46 nmol C L⁻¹ d⁻¹. Using the same bacterial biomass C : N of 5 : 1 and integrating for 1 year over the ∼100–200 m thick layer exhibiting rNPN anomaly yields an estimated 3.4 mmol N m⁻² yr⁻¹ bacterial N demand. If we again assume that bacteria satisfy all of their N demand from nitrate uptake, this process can explain at most 0 %–7 % of the estimated rNPN anomaly formation rate at Station ALOHA in the combined γ_n=24.2–25.2 layer.

4.2.3 Vertically migrating phytoplankton

From our analyses of the available published data, remineralization of N-poor DOM, TEP cycling, or heterotrophic bacterial nitrate uptake cannot quantitatively explain both the seasonal rPPN anomaly formation within the euphotic zone and subsurface rNPN anomaly formation observed at the stations ALOHA and BATS, even with the generous assumptions made. We now examine the vertical migration of phytoplankton down to the nutricline as a potential biological mechanism to explain the dual rPPN and rNPN anomalies.

In the absence of other mechanisms, the potential contribution of vertically migrating phytoplankton to the rNPN and rPPN anomaly features at the two stations can be determined by subtracting the contributions of TEP cycling and bacterial nitrate uptake to rNPN and/or rPPN formation from the total observed rNPN and rPPN anomaly formation rates (presented in Table 2). By contrast, the phytoplankton vertical migration category must provide 16–21 mmol N m⁻² yr⁻¹ (or 45 %–50 %) of rNPN anomaly formation and 27–36 mmol N m⁻² yr⁻¹ (or 59 %–62 %) of rPPN anomaly formation at Station ALOHA (Table 2). At the BATS station, phytoplankton vertical migration must provide 43–78 mmol N m⁻² yr⁻¹ (or 90 %–93 %) of rNPN anomaly formation and 59–76 mmol N m⁻² yr⁻¹ (or 87 %–95 %) of rPPN anomaly formation. These are likely underestimates because the bacterial nitrate uptake and TEP contribution are maximum estimates.

As a rare, giant component of the flora, net collections, in situ observations by divers, and towed optical systems constitute the bulk of observations; hence vertical migrators are somewhat out of the mainstream of current phytoplankton observations. The unique attributes and evidence for vertical migration are briefly discussed here for background, as well as documentation of their presence in both oceans at or near the time series stations. Nitrate transport rates are then summarized for comparison to the required contribution from the residual preNO₃ anomaly flux budgets.

Flagellate motility and cyanobacteria buoyancy control has been recognized for decades as a strategy to exploit spatially disjunct light and nutrient fields (Cullen, 1985; Eppley et al., 1968; Ganf and Oliver, 1982; Hasle, 1950; Steemann Nielsen, 1939) and was suggested for non-flagellated marine species of the genus Pyrocystis (Sukhanova and Rudyakov, 1973; Ballek and Swift, 1986; Rivkin et al., 1984). Rhizosolenia mats (macroscopic, multi-species (up to seven) associations of Rhizosolenia spp.; Villareal and Carpenter, 1989) have been the model for understanding this process (Villareal et al., 2014), with migration inferred for the diatoms Ethmodiscus spp. and large Rhizosolenia spp., as well as the phycomate prasinophyte Halosphaera spp. in addition to Pyrocystis. There is consistent evidence of high-δ¹⁵N composition of Rhizosolenia biomass similar to sub-nitracline N (Villareal et al., 1993, 2014), internal millimolar nitrate pools that can only be acquired by direct uptake at micromolar concentrations (Villareal and Lipschultz, 1995; Villareal et al., 1996; Woods and Villareal, 2008), buoyancy reversals linked to nutrient status (Richardson et al., 1996), nitrate reductase activity (Joseph et al., 1997) induced only when nitrate is the primary N source, ascent/descent rates of multiple meters per hour (Villareal et al., 2014; Moore and Villareal 1996), observation of Rhizosolenia mats from the surface to 305 m (Pilskaln et al., 2005), and compositional difference in floating and sinking mats that mirror physiological changes associated with nutrient depletion and carbohydrate ballasting (Villareal et al., 1996). A key attribute in all giant phytoplankton (10⁷⁺ µm³) migrators is multiple-meters-per-hour ascent/descent rates that permit 50–100 m vertical excursions (Moore and Villareal, 1986). This large vacuole of these giant phytoplankton allows the necessary storage (millimolar concentrations in vacuoles that are 90+ % of total cell volume) for a multiple-day migration and division cycle. Unlike the more widely known zooplankton vertical migration, open-ocean phytoplankton migration is a multi-day cycle (Richardson et al., 1998; Villareal et al., 1996). It is asynchronous and not cued to diel rhythms as in coastal dinoflagellates or vertically migrating zooplankton.

Giant phytoplankton are found throughout the warmer oceans of the world and are specifically noted at both Station ALOHA and the BATS station (Figs. S11–S13). The required characteristics of buoyancy reversals, high-internal-nitrate pools, and rapid ascent/descent have been found in multiple taxa from the Atlantic and Pacific oceans (Villareal et al., 2014). The taxa Halosphaera (Villareal and Lipschultz, 1995), Ethmodiscus (Villareal and Lipschultz, 1995), Rhizosolenia (Villareal and Lipschultz, 1995), and Pyrocystis (Swift et al., 1973; Rivkin et al., 1984; Villareal and Lipschultz, 19965) occur in the Sargasso Sea and all exhibit positive buoyancy and have millimolar internal nitrate pools. In the Sargasso Sea, the non-motile dinoflagellate Pyrocystis has been the most extensively studied by Elijah J. Swift's lab in the 1980s. A “once in a generation” migration cycle (Sukhanova and Rudyakov, 1973) was proposed for Pyrocystis based on differences in vegetative and reproductive cell maxima (∼60 and 120 m, respectively). Vegetative stages were positively buoyant (up to 0.2 m h⁻¹; Kahn and Swift, 1978), while reproductive stages were negatively buoyant at up to 18 m h⁻¹ in the Mona Passage, southwest N. Atlantic (Swift et al., 1976). In the Sargasso Sea, carbon budgets (Rivkin et al., 1984) and buoyancy changes upon nutrient depletion and addition (Ballek et al., 1986) all supported a vertical migration life cycle. In single-cell analysis from field material collected off Bermuda, Pyrocystis internal nitrate concentrations ranged from 0 to 8 mM, with floating cells containing significantly higher concentrations (p=0.05) than sinking cells (Villareal and Lipschultz, 1995). Such elevated internal nitrate can only be acquired at µM nitrate concentrations found in the nutricline. Abundance maxima in the upper ∼200 m of the Sargasso Sea/southwest N. Atlantic range from 100 to 200 cells m⁻³, with typical values reported of 18–75 cells m⁻³ (Rivkin et al., 1984), 10–50 cells m⁻³ (Ballek and Swift, 1986) and 10–30 cells m⁻³ (Swift et al., 1981). While the other migrating taxa are found in the Sargasso Sea near Bermuda, there are no data on abundance in the N. Atlantic other than for Ethmodiscus. Ethmodiscus was found at every station sampled on three cruises in the southwest N. Atlantic (n=18; ∼17–25^∘ N, 66–77^∘ W), with abundance ranging from 0.03 to 4.7 cells m⁻³. Free-living chains of Rhizosolenia were routinely isolated by hand collection with scuba at Bermuda (Moore and Villareal, 1996a, b; Richardson et al., 1996) and were widely noted by Carpenter et al. (1977) in the Caribbean. Cleve (1900) reported R. castracanei (a migrating giant diatom found both free-living and in Rhizosolenia mats) to occur in the Sargasso Sea as well as the tropical Atlantic, Caribbean Sea, Florida current to the Grand Banks of Newfoundland, and the Azores.

In the Pacific, the range of Ethmodiscus, Pyrocystis, and Rhizosolenia mats includes the study site at Station ALOHA (Figs. S11–S13). Unlike the Atlantic, where mats are generally very rare (∼0.001 mats m⁻³; Carpenter et al., 1977), Rhizosolenia mats dominate in the Pacific. Abundance ranges up to several hundred mats per cubic meter, depending on the mat size and enumeration methods (Villareal et al., 1996, 1999, 2014). Quantitative net surveys of Ethmodiscus north of Hawaii yielded abundance ranges of <0.1 to >2 cells m⁻³ (Villareal et al., 2007). Both Halosphaera and Pyrocystis were present but not quantified. Further to the west, Soviet-era literature documents Pyrocystis at 10–50 cells m⁻³ in the upper 200 m (Sukhanova and Rudyakov, 1973; Sukahova, 1973). Free-living giant Rhizosolenia were observed by divers at all stations sampled in the N. Pacific (Tracy A. Villareal, unpublished observation) and overlapped 100 % with Rhizosolenia mats (Fig. 2 in Villareal et al., 2014); R. castracanei was reported across the Pacific Ocean (Fig. 8 in Semina and Levashova, 1977). It should be noted that net collections disrupt Rhizosolenia mats (Alldredge and Silver, 1982), and the large-diameter species such as R. castracanei in the mats will appear as free chains and cells in the literature.

For rNPN and rPPN development, the unique characteristics of vertical migration are that cells acquire and store nitrate at depth, transport it into the euphotic zone, and then reduce it internally to biomass concomitant with oxygenic photosynthesis. Nitrate transport calculations by vertical migration have a number of assumptions and caveats, including variability in abundance, particularly for Halosphaera spp. (Villareal et al., 2014). Inclusion of small-sized Rhizosolenia mats missed by divers (Villareal et al., 1999) in estimates yielded rates of 6–444 µmol N m⁻² d⁻¹ (mean $=\mathrm{171}\pm$ SD 187 µmol N m⁻² d⁻¹) at four stations north of Station ALOHA. Assuming a 180-day production period, this converts to 1.1–80 mmol N m⁻² y⁻¹ (mean $=\mathrm{30}\pm$ SD 34 mmol N m⁻² yr⁻¹). Using a generalized gyre abundance (Villareal et al., 2014), four other taxa yielded transport rates of 3.2 (Ethmodiscus spp.), 9.2 (free living Rhizosolenia spp.), 17.0 (Pyrocystis spp.) and 33.2 (Halosphaera spp.) µmol N m⁻² d⁻¹ for a total of 62.5 µmol N m⁻² d⁻¹ (11.2 mmol N m⁻² yr⁻¹). These taxa exhibit basin-specific abundance patterns (i.e., Rhizosolenia mats are rare in the N. Atlantic – Carpenter et al., 1977; Halosphaera viridis has an extensive subsurface presence in the deep Mediterranean in winter – Wiebe et al., 1974), but the generalized abundance and distribution data are consistent with required rNPN and rPPN rates at both the BATS station (43–78 and 59–76 µmol N m⁻² d⁻¹, respectively) and Station ALOHA (16–21 and 27–36 µmol N m⁻² d⁻¹, respectively). Rhizosolenia mats alone may be sufficient to support rNPN and rPPN rates at Station ALOHA. There are several implications of this. Lower abundance or turnover times of this flora will still have substantial effects on rNPN and rPPN. Patchiness in abundance (Villareal, 2007; Villareal et al., 2014) or concentration of buoyant particles by mesoscale fronts (Guidi, 2012) could lead to significant local perturbations on rNPN and rPPN.

Fraga (2001) developed an independent assessment of the impact of vertical migration on the NO tracer (functionally related to our residual preNO₃ tracer) based on first principles of photosynthetic production and biosynthesis of biomass. For ecosystems containing both coastal dinoflagellates and oceanic diatoms, a subsurface minimum in NO and a related tracer, NCO (NO corrected for the stoichiometry of carbohydrate synthesis), formed. For migrating Rhizosolenia, the NO and NCO minimum was at ∼150 m with negative anomalies from ∼80 to ∼200 m (Fig. 6 in Fraga, 2001). The Fraga (2001) formulation also included an explicit term for release of soluble carbohydrate (functionally equivalent to TEPs). Migrating phytoplankton can clearly provide a mechanistic linkage via upward nutrient transport to the observed NO anomalies.

4.3.1 DIC

Vertically migrating phytoplankton can also help explain the observed summertime DIC drawdown (Gruber et al., 1998; Keeling et al., 2004) in the absence of measurable nitrate from the mixed layer at both the stations ALOHA and BATS. The concurrently operating migration cycles of different individuals would continually bring intracellular nitrate-rich migrators into the surface mixed layer, where their oxygenic photosynthesis would draw down DIC and release dissolved O₂, contributing to rPPN anomaly formation within the mixed layer. High-intracellular-nitrate gradients within migrators (millimoles to nanomoles across cell membranes) can also lead to excretion of dissolved nitrate into the water column (Singler and Villareal, 2005). Thus, nitrate leakage from migrators could help explain the observation of a presumed subsurface nitrate source supporting pico-eukaryotic phytoplankton in the mixed layer, even during the stratified summer months at the BATS station (Fawcett et al., 2011). However, nitrogen derived from migrator-mediated nitrate excretion is required to sink out of the mixed layer in some form in order to contribute to summertime net DIC drawdown at these two sites. Regardless of how nitrate is acquired and internally reduced (by migrators or by release and uptake by non-migrating phytoplankton), it will contribute to the observed rPPN.

We can compare our estimated rates of rPPN formation within the euphotic zone with the summertime mixed-layer DIC drawdown at each time series site using assumptions on the appropriate ${\mathrm{O}}_{\mathrm{2}}:\mathrm{C}:\mathrm{N}$ stoichiometry of production. Although seasonal rPPN formation exhibits the largest increase in preNO₃ concentrations at ∼50–100 m depth, seasonal accumulation of residual positive preNO₃ is observed within the mixed layer (∼0–30 m) at both sites (Figs. 1, 2).

At Station ALOHA, net community production during the 6-month summer-to-fall period represents ∼40 % of the 2.3 mol C m⁻² yr⁻¹ ANCP (Keeling et al., 2004), yielding a summertime mixed-layer DIC drawdown of ∼920 mmol C m⁻². Our estimated rate of euphotic-zone rPPN formation during this same seasonal period is 33–81 mmol N m⁻² (Table 1). The summertime DIC drawdown can be converted to residual preNO₃ nitrogen equivalents using the literature range in O₂:C stoichiometries of production of 1.0–1.6 (Paulmier et al., 2009) and the DOM- and POM-weighted O₂:N stoichiometries of production (using the values in Table 1), yielding values of 7.8–14.3 for the C:N of summertime new production at Station ALOHA. Summertime DIC drawdown converted to preNO₃ equivalents is 64–118 mmol N m⁻². Thus, our estimated rPPN formation within the euphotic zone can potentially explain 28 % to >100 % of the mixed-layer DIC drawdown at Station ALOHA.

The estimated rate of euphotic-zone rPPN formation during summer to fall at the BATS station is 50–96 mmol N m⁻² (Table 1). Net community production during this seasonal period is ∼45 % of the 2.3 mol C m⁻² yr⁻¹ ANCP (Gruber et al., 1998), yielding a summertime mixed-layer DIC drawdown of ∼1035 mmol C m⁻². Conversion of the summertime DIC drawdown to preNO₃ nitrogen equivalents following a similar approach to our calculation at Station ALOHA uses a DOM- and POM-weighted C:N of 7.9–14.8 for summertime new production at the BATS station, yielding 70–131 mmol N m⁻². Thus, our estimated rPPN formation within the euphotic zone can potentially explain 38 % to >100 % of the mixed-layer DIC drawdown at the BATS station.

4.3.2 prePO₄

We also investigated for the presence of subsurface residual negative prePO₄ anomalies at the stations ALOHA and BATS. Station ALOHA does not exhibit a subsurface residual negative prePO₄ anomaly (Fig. 3), with the exception of a small anomaly present in the years 2002–2004. There do exist a subsurface residual negative prePO₄ anomaly and euphotic-zone residual positive prePO₄ anomaly at the BATS station (Fig. 4), present throughout the time series and monthly averaged climatology, which are coincident in both seasonal timing and depth/density layer location as the rNPN and rPPN anomalies (Fig. 2). Carbon-to-phosphorus ratios exhibit their global maxima in the subtropical North Atlantic for phytoplankton biomass (Martiny et al., 2013), dissolved organic matter (Letscher and Moore, 2015), and exported organic matter (Teng et al., 2014). The hypothesized phosphorus limitation of this basin (Ammerman et al., 2003; Moore et al., 2004) may explain the contrasting observations of the presence of a residual negative prePO₄ anomaly at the BATS station in the North Atlantic but not at Station ALOHA in the North Pacific. Because of the many observations of elevated C:P across organic matter pools in the North Atlantic, we tested for the theoretical value of R${}_{-{\mathrm{O}}_{\mathrm{2}}:\mathrm{P}}$, the stoichiometric ratio of O₂ consumed to phosphate released during the remineralization of organic matter, needed to eliminate the subsurface residual negative prePO₄ anomaly at the BATS station. This theoretical value of R${}_{-{\mathrm{O}}_{\mathrm{2}}:\mathrm{P}}$ was found to be $\sim \mathrm{1000}:\mathrm{1}$ (Fig. S10), which is $\sim \mathrm{4}\times$ higher than the inferred R${}_{-{\mathrm{O}}_{\mathrm{2}}:\mathrm{P}}$ for the North Atlantic from an inversion of subsurface O₂ and phosphate data (DeVries and Deutsch, 2014).

Figure 5Seasonal climatology of the residual preNO₃ tracer [µM] at 150 m depth from the World Ocean Atlas (2013) 1^∘ climatologies of O₂ anomaly and nitrate. Residual preNO₃ is calculated using values of f_DOM=0.5, r_DOM=20.0, and r_POM=8.75.

Download

We hypothesize that P-limited or P-stressed vertically migrating phytoplankton also take up phosphate at the nutricline in combination with nitrate to contribute to both the observed rNPN and rPPN and the subsurface residual negative prePO₄ anomaly present at the BATS station. Limited sampling in the waters between Hawaii and California indicated N:P ratios were not significantly different between sinking and ascending Rhizosolenia mats ($\mathrm{N}:\mathrm{P}\sim \mathrm{26}$–30) while C:P ratios were significantly different (p=0.05; C:P sinkers $=\mathrm{388}\pm \mathrm{66}$; C:P ascending $=\mathrm{221}\pm \mathrm{43}$). This is consistent with simultaneous uptake of N and P at depth but carbon consumption at depth relative to the surface. Further data on phosphate composition of vertically migrating phytoplankton are needed to confirm our hypothesis.