Cross-basin differences in the nutrient assimilation characteristics of induced phytoplankton blooms in the subtropical Pacific waters

. To better understand the nutrient assimilation characteristics of subtropical phytoplankton, deep water addition 15 incubation experiments were carried out on surface waters collected at seven stations across the subtropical North and South Pacific Ocean. These deep water additions induced phytoplankton blooms with nutrient drawdown at all stations. The drawdown ratios of dissolved inorganic nitrogen (DIN) to phosphate (PO 4 ) varied from 14.1 to 30.7 at the PO 4 -replete stations in the central North Pacific (CNP) and eastern South Pacific (ESP). These ratios were similar to the range represented by the canonical Redfield ratio (16) through to typical particulate N:P ratios in the surface subtropical ocean (28). 20 In contrast, lower DIN:PO 4 drawdown ratios (7.7-13.3) were observed in induced blooms at the PO 4 -depleted stations in the western North Pacific (WNP). The DIN:PO 4 drawdown ratios in the PO 4 -replete ESP were associated with eukaryote-dominated blooms, while those in PO 4 -depleted WNP were associated with eukaryotic and cyanobacterial blooms. The surplus PO 4 assimilation, relative to DIN, by phytoplankton in the WNP was not expected based on their typical cellular N:P ratio, and was likely due to the high PO 4 uptake capability as induced by low PO 4 -adapted phytoplankton. The low and high 25 P* (=PO 4 -DIN/16) regimes geographically corresponded to the low and high DIN:PO 4 drawdown ratios in the WNP and the CNP or ESP, respectively. The basin-wide P* distribution in the oligotrophic Pacific surface waters showed a clear regional trend from low in the WNP (<50 nM) to high in the ESP (>100 nM). These results suggest that the subtropical phytoplankton blooms as observed in our experiments could be an important factor controlling P* as well as the commonly recognized dinitrogen fixation and denitrification characteristics. and Chl a play a common role in light-harvesting. The DVchl a :Tchl a ratios were mostly 0.4-0.5, but the lower ratios were observed at Stations 18 and 21 in the ESP (0.1 and 0.2). Cell densities of surface diatoms were consistently low (3-88 cells L -1 ) at all stations. Author contribution. FH designed the incubation experiments. KF, HS, HO, and PWB designed the sampling schemes across the subtropical North and South Pacific. FH, TK, JK, and EMSW collected nanomolar nutrient data. SY-T, FH, and JK collected DON and DOP data. FH collected phytoplankton data. FH and IT performed the Fe-enrichment incubation experiment. FH wrote the manuscript. All authors reviewed and approved the manuscript.


Introduction
The surface waters of the subtropical oceans are characterized by strong stratification, low nutrients, and low phytoplankton biomass (Karl, 2002). In this regime, primary production is largely sustained by regenerated production (f-ratio: ~0.1, Dugdale and Goering, 1967;Eppley and Peterson, 1979), and driven by small phytoplankton such as the picocyanobacteria 35 Prochlorococcus and Synechococcus (Waterbury et al., 1979;Chisholm et al., 1988). Despite the persistent oligotrophic regime, phytoplankton blooms with large diatoms and cyanobacteria occur occasionally in the subtropical oceans and have large impacts on new production and export production (Benitez-Nelson et al., 2007;McGillicuddy et al., 2007;Dore et al., 2008;Wilson and Qiu, 2008;Karl et al., 2012;Villareal et al., 2012;Hashihama et al., 2014). The mechanisms that bring about the development of these blooms are not simple, but fundamentally they involve nutrient supply with physical forcing 40 Toyoda and Okamoto, 2017).
The nutrient supply to surface subtropical oceans is important for many aspects of biogeochemical cycling and foodweb dynamics as it drives new production and net community production (Sarmiento and Gruber, 2006;Saito, 2019).
Seasonal variations in dissolved inorganic carbon and dissolved oxygen in the subtropical oceans highlight the net productive systems, which are potentially sustained by intermittent nutrient supply from deep water (Michaels et al., 1994;45 Dore et al., 2003;Johnson et al., 2010). Deep water contains high amounts of nutrients such as nitrate (NO3), phosphate (PO 4 ), and silicic acid (Si(OH) 4 ), and their supply into the surface ocean temporarily alleviates phytoplankton nutrient stress.
Several ship-based experimental studies indicate that deep water additions to subtropical surface waters have induced phytoplankton blooms (Mahaffey et al., 2012;Lampe et al., 2019;Robidart et al., 2019). These studies highlight the shifts in phytoplankton community structure, growth characteristics, and gene expression during the bloom development. However, 50 although nutrient assimilation characteristics are important mechanisms driving the net production, they were not fully described, for example the drawdown ratios (e.g., ∆NO3:∆PO4 and ∆Si(OH) 4 :∆NO 3 ), in these studies.
Phytoplankton N:P stoichiometry is generally based on the canonical ratio of 16 (Redfield, 1958). However, subtropical phytoplankton have higher N:P cellular ratios than Redfield and its mean value for subtropical waters is 28 (Martiny et al., 2013). This higher ratio suggests that subtropical phytoplankton assimilate nutrients with higher N:P ratios than 16. If 55 subtropical phytoplankton assimilate the upwelled deep water nutrients which have nearly Redfield NO 3 :PO 4 ratios (~16, Fanning, 1992), PO 4 -excess waters would remain at the surface. The PO 4 anomaly (P*=PO 4 -NO 3 /16) in the upper 120 m of the water column has indeed positive values throughout the subtropical oceans (Deutsch et al., 2007). As with N* (Gruber and Sarmiento, 1997;Deutsch et al., 2001), P* is recognized to be controlled by dinitrogen (N 2 ) fixation and denitrification (Deutsch et al., 2007), but it may also be influenced by phytoplankton uptake of the upwelled deep water nutrients. 60 Subtropical phytoplankton utilize not only NO 3 and PO 4 but also nitrite (NO 2 ), ammonium (NH 4 ), dissolved organic N (DON), and dissolved organic P (DOP). Amongst them, the concentrations of DON and DOP are one to three orders of magnitude higher than those of dissolved inorganic N (DIN: the sum of NO 3 , NO 2 , and NH 4 ) and PO 4 in subtropical surface waters (Karl and Björkman, 2015;Sipler and Bronk, 2015). The majority of DON and DOP is likely refractory, but the 3 bioavailable forms such as urea, amino acid, and ATP play important roles in sustaining primary production in the inorganic 65 nutrient-depleted subtropical waters (Kanda et al., 1985;Zubkov et al., 2004;Casey et al., 2009;Hill et al., 2011;Shilova et al., 2017;Björkman et al., 2018). Thus, the dynamics of alternative nutrients other than NO 3 and PO 4 should be considered when examining the nutrient assimilation characteristics of subtropical phytoplankton blooms.
Along with N and P assimilation, Si is also assimilated during diatom blooms. Several field studies reported anomalous Si(OH) 4 removal relative to NO 3 and PO 4 at the sites of diatom blooms in the subtropical oceans (Benitez-Nelson et al., 70 2007;Hashihama et al., 2014). These Si(OH) 4 removals were not accompanied by stoichiometrically equivalent N and P removals as with a typical Si:N:P ratio of 16:16:1 (Redfield, 1958;Brzezinski, 1985). Given the linkages between Si and other elemental cycles, it is important to understand Si dynamics in the subtropical oceans. However, the Si dynamics cannot be fully explored from snapshot observations in the field (Hashihama et al., 2014), and to understand them experimental validations are required. 75 In this study, our aim was to reveal N, P, and Si assimilation characteristics of subtropical phytoplankton blooms as induced by deep water additions. The onboard bottle incubation experiments were conducted across the subtropical Pacific Ocean, which has a large geographical variation in surface PO4 from very low (<10 nM) in the western North Pacific (WNP) to high (>100 nM) in the eastern South Pacific (ESP) Martiny et al., 2019;.
Since subtropical phytoplankton respond to nanomolar increases in nutrient concentrations (Garside, 1985;Eppley and 80 Renger, 1988;Eppley et al., 1990), we used sensitive liquid waveguide spectrophotometry for measuring nanomolar NO 3 , NO 2 , NH 4 , PO 4 , and Si(OH) 4 . The nanomolar nutrient data enabled us to calculate accurate stoichiometric ratios for N, P, and Si. Along with the inorganic nutrients, we also examined DON and DOP variations during the incubation experiments.
These deep-water addition experiments successfully induced phytoplankton blooms, while the nutrient drawdown ratios showed geographical patterns concomitant with the surface PO 4 distributions. Here we conclude by discussing the 85 mechanism of the regionally different drawdown ratios and its possible influences on P* distribution in the subtropical Pacific Ocean.

Study areas and water sampling
Observations were conducted at seven stations in the subtropical North and South Pacific Ocean (Table 1 and Fig. 1). Station 90 A in the WNP was occupied in July 2010 during the R/V Tansei Maru KT-10-13 cruise. Stations 2-21 were occupied along the transect from the WNP (2 and 5) to the ESP (15, 18, and 21) through the central North Pacific (CNP, 8) for the period from December 2011 to January 2012 during the R/V Hakuho Maru KH-11-10 cruise. Water sampling was performed using a conductivity-temperature-depth (CTD) system (Sea-Bird Electronics) equipped with HCl-cleaned Teflon-coated Niskin-X bottles (General Oceanics). Water samples for incubation experiments were collected from 10 m depth (hereafter referred to 95 as 'surface') at all stations. At Stations A and 2 in the WNP, deep water from 1500 m depth was also collected.

Deep-water addition incubation experiments
Surface waters collected from 10 m depth were poured into 2.3 L HCl-cleaned polycarbonate bottles and then 50 mL of deep water was immediately added to the triplicate bottles (2.1% v/v deep water addition). Filtration of the deep water was not 105 conducted to avoid NH 4 contamination. Although the deep water contains heterotrophs, their influences on the results of incubation experiments were expected to be at negligible levels because, in addition to the large dilution of deep water (2.1% of the total water), prokaryotic abundance and production are generally much lower in the deep water than in the surface water as exponentially decreasing with depth (Yokokawa et al., 2013). The deep waters collected at Stations A and 2 were added to the surface water samples at Stations A and 2-21, respectively. The deep waters for the experiments at Stations 5-21 110 were kept frozen (-20 °C) until used at each station. Nutrient concentrations in the deep waters at Stations A and 2 were 37.1 and 39.0 μM DIN, 2.1 and 2.9 μM PO4, and 134.5 and 140.5 μM Si(OH) 4 , respectively. For all experiments, triplicate 5 control bottles were prepared. Both the treated and control bottles were incubated for 48-96 h (Table 1) in an on-deck incubator with flowing surface seawater, which was shaded with appropriate sheeting to give 30% of full sunlight so as to mimic the ambient photosynthetically active radiation (PAR) condition at 10 m depth. The ambient PAR on deck was 115 continuously monitored by a LI-COR quantum sensor (LI-190R) with a data logger (LI-1400), and its mean values during the incubation periods including day and night times are presented in Table 1. During the incubation periods (48-96 h), the bottles were subsampled 5-6 times for nanomolar nutrients, 3 times for DON and DOP, and 1 time (at the end point) for phytoplankton. Initial phytoplankton samples at time-zero were collected in single or duplicate directly from the Niskin-X bottles. At the end point, water volumes in the bottles just prior to final sampling were approximately 1.8 L. The DON and 120 DOP samples were not collected at Station A during the KT-10-13 cruise. To assess any significant decrease or increase in the concentrations of nanomolar nutrients, DON, and DOP during the incubation periods, linear regression analyses were performed. In addition, Student t-test and paired t-test were performed to determine significant differences between the measured parameter values. In this paper, the significance is reported where p<0.05. In the calculation of nutrient drawdown ratio, its error was estimated based on the error propagation rule (Miller and Miller, 1993). 125

Determinations of nanomolar nutrients
Water samples for nanomolar nutrients were collected in 30 mL HCL-cleaned polypropylene tubes and were frozen at -20 °C until analysis. The concentrations of NO3, NO 2 , NH 4 , PO 4 , and Si(OH) 4 were determined using an automated liquid waveguide spectrophotometric system equipped with 50-100 cm liquid waveguide capillary cells (LWCC, World Precision Instruments) Hashihama et al., 2014;Hashihama et al., 2015). The 130 detection limits for NO 3 , NO 2 , NH 4 , PO 4 , and Si(OH) 4 were 3, 2, 6, 3, and 11 nM, respectively. Although we collected triplicate water samples from the incubated bottles, triplicate analytical data were not available for several samples.

Determinations of DON and DOP
Water samples for DON and DOP were collected in HCL-cleaned polypropylene tubes after removing particulate matter by filtering through pre-combusted Whatman GF/F filters. The samples were frozen at -20 °C until analysis. Total dissolved N 135 (TDN) and P (TDP) were quantified by a persulfate oxidation method (Hansen and Koroleff, 1999) with a QuAAtro TN-TP analyser (SEAL Analytical) (Yasui et al. 2016;Yasui-Tamura et al., 2020). Concentration of DON was derived from the difference between TDN and DIN concentrations, and that of DOP was derived from the difference between TDP and PO 4 concentrations. As with the nutrients (2.3), triplicate analytical data on DON and DOP were not obtained for some samples.

Phytoplankton pigment analysis 140
Phytoplankton pigment analysis was performed using high-performance liquid chromatography (HPLC). Water volumes of 440-3000 mL were filtered onto GF/F filters, and the filter samples were immediately frozen in liquid nitrogen and stored in a deep freezer (-80 °C) until analysis. Pigment analysis was conducted using the method of Zapata et al. (2000) with a HPLC 6 system (Hashihama et al., 2008). Six phytoplankton pigments -chlorophyll a (Chl a), divinyl chlorophyll a (DVchl a), 19ˊbutanoyloxyfucoxanthin (But-fuco), fucoxanthin (Fuco), 19ˊ-hexanoyloxyfucoxanthin (Hex-fuco), and zeaxanthin (Zea) -145 were quantified from the peak area calibrated against that of standard pigments (DHI Water and Environment). Total chlorophyll a (Tchl a: the sum of Chl a and DVchl a) was used as an index of total phytoplankton biomass. Total fucoxanthin (Tfuco: the sum of But-fuco, Fuco, and Hex-fuco) was used as a representative marker of eukaryotic phytoplankton. Zea and DVchl a were markers of cyanobacteria and Prochlorococcus, respectively.

Microscopic analysis of phytoplankton 150
Water volumes of 100-1000 mL seawater samples were fixed with neutralized formalin at a final concentration of 1% (v/v).
The fixed samples were concentrated through sedimentation in a land-based laboratory. Diatom species were identified and enumerated under an inverted microscope (Utermöhl, 1958).

P* determination using nanomolar nutrient data
Surface P* at the experimental stations was calculated using the measured nanomolar PO 4 and DIN data through an equation 155 P*=PO 4 -DIN/16. In addition, to reveal basin-wide distribution of surface (≤10 m) P* over the oligotrophic Pacific area (40° N-40° S), we assembled nanomolar (<1000 nM) data sets of PO 4 and NO 3 plus NO 2 (N+N), most of which were previously published by the authors in this study Kitajima et al., 2009;Shiozaki et al., 2009;Sato et al., 2010;Shiozaki et al., 2010;Girault et al., 2013;Sato et al., 2013;Hashihama et al., 2014;Shiozaki et al., 2014;Girault et al., 2015;Sato et al., 2015;Sato et al., 2016;Shiozaki et al., 2016;Shiozaki et al., 2017;Ellwood et al., 160 2018;Horii et al., 2018;Shiozaki et al., 2018;Martiny et al., 2019;Sato and Hashihama, 2019;Yamaguchi et al., 2019;Yamaguchi et al., 2020;Jiang et al. in press). We also included several unpublished data sets collected by F. Hashihama and T. Kodama. These data sets were obtained by using the liquid waveguide spectrophotometry for PO4 and N+N (Woodward, 2002;Hashihama et al., 2009). Since surface NH 4 concentrations were typically low at the sub-nanomolar level and the NH 4 data were relatively limited compared to the PO 4 165 and N+N data, we did not use the NH 4 data to show the basin-wide distribution of P*. Thus, the P* in this case was calculated through an equation P*=PO 4 -N+N/16.

Initial conditions
High temperature (21.76-26.91 °C) and high salinity (34.07-36.50) in the surface waters (10 m) of the seven experimental 170 stations indicated that typical subtropical oceanic waters prevailed in the study regions (Table 2). DIN concentrations at the surface were consistently lower than 50 nM, while PO 4 concentrations varied geographically and were extremely low in the WNP (<10 nM; Stations A, 2, and 5), intermediate in the CNP (53 nM; Station 8), and high in the ESP (>100 nM; Stations 15, 18, and 21). Surface P* at these stations showed a geographical variation similar to PO 4 concentrations, and this trend was due to the consistently low concentrations of DIN found at those stations. Si(OH) 4 concentrations were higher in the 175 WNP and CNP (767-1276 nM; Stations A, 2, 5, and 8) than the ESP (427-541 nM; Stations 15, 18, and 21), and DON and DOP concentrations ranged from 3.47 to 4.45 μM and 0.10 to 0.21 μM, respectively.
Tchl a concentrations at the surface were less than 129 ng L -1 with low values at Stations 18 and 21 in the ESP (18 and 3 ng L -1 , Table 2). Tfuco, Zea, and DVchl a all showed geographical variations similar to Tchl a. Tfuco and Zea play roles in light-harvesting and photoprotection, respectively (Falkowski and Raven, 2007). Based on these roles, Tfuco content in 180 eukaryotes should be higher in the low-PAR regime than in the high-PAR regime, while Zea content in cyanobacteria should be higher in the high-PAR regime than in the low-PAR regime. However, Tfuco:Zea ratios (0.4-2.6) were not consistently higher in the low-PAR WNP and CNP (≤284 μmol photons m -2 s -1 , Stations 2, 5, and 8, Table 1) than in the high-PAR WNP and ESP (≥542μmol photons m -2 s -1 , Stations A, 15, 18, and 21, Table 1), indicating that these ratios represent the biomass proportion of cyanobacteria to eukaryotes rather than the results of photoacclimation and/or photoadaptation. The Tfuco:Zea 185 ratios were lower than 1.0 except for Station 18 in the ESP (2.6), where the biomass proportion of eukaryotes to cyanobacteria was relatively high. DVchl a:Tchl a ratios are indices of the biomass proportion of Prochlorococcus to total phytoplankton because DVchl a and Chl a play a common role in light-harvesting. The DVchl a:Tchl a ratios were mostly 0.4-0.5, but the lower ratios were observed at Stations 18 and 21 in the ESP (0.1 and 0.2). Cell densities of surface diatoms were consistently low (3-88 cells L -1 ) at all stations. 190

Phytoplankton response to deep water additions
Mean Tchl a concentrations were three to ten times higher in the treated bottles than observed for the control bottles in all 195 experiments, although no significant difference was observed at Station A due to a highly variable results in the treated bottles (Fig. 2a). These trends in increasing Tchl a indicate that the deep water additions positively induced phytoplankton blooms. The Tchl a differences between the treated and control bottles were greatest at Station A in the WNP (178 ng L -1 ) and Station 15 in the ESP (306 ng L -1 ), and net growth rates of the blooms (as Tchl a, relative to the control) were higher at these two stations (0.70 and 0.58 d -1 ) than other stations (0.27-0.50 d -1 ). Mean Tfuco concentrations were significantly higher 200 in the treated than in the control bottles except for Station A (Fig. 2b). Mean Zea concentrations were significantly higher in the treated bottles in the WNP (Stations A, 2, and 5) and at two sites of the ESP (Stations 15 and 21) (Fig. 2c). Mean DVchl a concentrations were significantly higher in the treated bottles at two sites in the WNP (Stations 2 and 5) and Station 21 in the ESP (Fig. 2d). The pigment concentrations in the control bottles were similar to those in the initial conditions, indicating that changes in pigment concentrations due to photoacclimation during the incubation periods are small. 205 Mean Tfuco:Zea ratios were higher in the treated than in the control bottles, although no significant differences were observed at Stations A and 18 (Fig. 2e). The higher Tfuco:Zea ratios in the treated bottles imply that biomass increases in eukaryotes were relatively large compared to those of cyanobacteria. The ratios of Tfuco:Zea in the treated bottles were higher in the ESP (Stations 15, 18, and 21) than the WNP and CNP (Stations A, 2, 5, and 8), indicating that the proportions of eukaryotes (cyanobacteria) were higher (lower) in the ESP than the WNP and CNP. Mean DVchl a:Tchl a ratios were 210 significantly lower in the treated bottles than the control, except for Stations 18 and 21 in the ESP, where the ratios in the control were quite low (<0.1) as observed in the initial conditions (Fig. 2f).
Cell densities of diatoms were significantly higher in the treated bottles at two sites of the WNP (Stations A and 2) and Station 18 in the ESP (Fig. 2g). An exceptionally high mean density of diatoms (907 cells L -1 ), mostly consisting of Nitzschia longissima, was observed in the treated bottles at Station 15 in the ESP, although no significant difference between 215 the densities in the control and treated bottles was seen. Significant differences (t-test, p<0.05) between the values in the control and treated bottles are depicted by asterisks. Net growth rates, μ (d -1 ) of the phytoplankton blooms (as Tchl a, relative to the control) are denoted in (a) Tchl a. In (g) diatom cell densities, a scale of y-axis at Station 15 (~1500 cells L -1 ) is different from those at other stations (~300 cells L -1 ).

Nutrient drawdown
Following the phytoplankton blooms, DIN and PO 4 concentrations in the treated bottles at all stations showed significant 225 linear decreases as a function of the incubation times (r 2 >0.82, Figs. 3a and 3b). In contrast, the concentrations in the control bottles at all stations showed no significant trends. The DIN decreases were largely ascribed to NO 3 decreases (Fig. A1a), and interestingly, NO 2 concentrations in the treated bottles at all stations significantly increased with time (Fig. A1b). There were also significant linear increases in the control bottles for NO 3 at Stations 8 and 15, and NO 2 at Station 18 but these changes were quite small (<13 nM). NH 4 concentrations in the treated and control bottles showed no significant trends with 230 time, but occasionally high standard deviations (>100 nM) were observed (Fig. A1c).
Drawdown rates (slopes of linear regression lines in Figs. 3a and 3b) of DIN and PO 4 in the treated bottles varied from 1.38 to 7.37 nmol N L -1 h -1 and from 0.13 to 0.71 nmol P L -1 h -1 , respectively. These rates were not significantly different from the rates derived from the first 36-48 h incubation data (paired t-test, p>0.05). The drawdown rates were relatively low (≤1.49 nmol N L -1 h -1 and ≤0.15 nmol P L -1 h -1 , respectively) in the WNP during winter (December, Stations 2 and 5) where 235 relatively low mean PAR was observed during the incubation periods (≤284 μmol photons m -2 s -1 , Table 1). Differences between the control-corrected mean concentrations of DIN and PO 4 at the start and end points of the incubation (∆DIN and ∆PO 4 ) varied from 123 to 749 nM and from 9 to 53 nM, respectively (Table 3). The values of ∆DIN and ∆PO 4 normalized by the incubation times (h) were almost identical to the drawdown rates of DIN and PO 4 in the treated bottles (r 2 =0.99 and r 2 =0.91 in linear regressions, respectively). 240 Unlike the DIN and PO 4 concentrations, Si(OH) 4 concentrations in the treated bottles at all stations did not show any significant linear decreases with the occasionally high standard deviations (>500 nM) (Fig. 3c). The insignificant trends in Si(OH) 4 concentrations were also observed in the control bottles at all stations. Difference between the control-corrected mean Si(OH) 4 concentrations at the start and end points (∆Si(OH) 4 ) showed net drawdown values ranging from 7 to 465 nM, although mean Si(OH) 4 concentrations in the treated bottles at the start and end points were not significantly different except 245 at Station A (Table 3).

Nutrient drawdown ratio
In all experiments, the enriched nutrients in the treated bottles were not fully taken up by phytoplankton during the incubation periods (Fig. 3). Therefore, we assessed the nutrient drawdown ratios using ∆DIN, ∆PO 4 , and ∆Si(OH) 4 (Table 3). 260 ∆DIN:∆PO 4 ratios varied from 7.7 to 30.7. Although 95% confidence intervals of these ratios were large (12.6-82.8), the relatively lower ratios (≤13.3) were convergent in the WNP (Stations A, 2, and 5). ∆Si(OH) 4 :∆DIN ratios varied from 0.03 to 2.96 with most stations less than 1 except for Stations 5 and 8. However, these ratios, except at Station A (0.36), involved uncertainties due to no significant decreases in Si(OH) 4 concentrations in the treated bottles during the incubations.
We also evaluated the drawdown characteristics using the control-corrected mean concentrations of DIN and PO 4 at 265 sampling points during the incubation periods. A plot of PO 4 against DIN showed strong negative linear relationships (r 2 >0.75), except for Station 21 (r 2 =0.23) (Fig. 4). Here, the drawdown ratio of DIN to PO 4 was expressed as 1/slope of the linear regression, and ranged from 10.2 to 31.7 with 95% confidence intervals between 1.8 and 41.6. These drawdown ratios were almost identical to the ∆DIN:∆PO 4 ratios in Table 3   seven stations. Linear regression lines with their parameters (1/slope, PO 4 -intercept, and r 2 ) at seven stations were denoted by the different colours. Errors of 1/slope and PO 4 -intercept represent 95% confidence intervals.

DON and DOP
DON and DOP concentrations in both the treated and control bottles did not show any significant increase or decrease as a function of the incubation time, except for DOP in the treated bottles at Station 8 in the CNP (Figs. A2a and A2b). At this 280 station, the DOP concentrations in the treated bottles significantly increased with time, although the change over 48 h (0.03 μM) was smaller than that from 0 to 24 h (-0.07 μM) in the control bottles. Overall, the DON and DOP concentrations in the treated bottles were similar to those in the control bottles, indicating that the deep water did not contain significant amounts of DON and DOP and its additions did not alter DON and DOP regimes during the incubation periods.

Basin-wide P* distribution in the oligotrophic Pacific 285
The assembled surface N+N and PO 4 data (Figs. A3a and A3b) revealed a detailed surface P* distribution over the oligotrophic Pacific Ocean (Fig. 5). The distributional pattern of P* was similar to that of PO 4 (Fig. A3b), mainly due to a broad area with low N+N (<100 nM) (Fig. A3a). The P* showed a clear west-east gradient from <50 nM in the western basin to ~500 nM in the eastern basin. In the western basin, the extremely low P* (<10 nM) was found in the WNP. Stations A, 2, and 5 were located within this low P* area. Station 8 was in the intermediate P* area (50-100 nM) in the CNP, while 290 Stations 15, 18, and 21 were within the high P* area (>100 nM) in the ESP.  (Table 3).

Phytoplankton blooms following deep water additions
Our study confirms that deep water additions induced phytoplankton blooms in various regions of the subtropical Pacific 300 Ocean. Such induced blooms have also been reported at Station ALOHA in the subtropical North Pacific Ocean (Mahaffey et al., 2012). The ALOHA experiments revealed that Tchl a-based growth rates following deep water additions were higher in boreal summer than boreal winter. A similar seasonal trend was observed in the North Pacific stations reported here; net growth rates were higher at the westernmost site in July (Station A, 0.70 d -1 ) than other sites in December (Stations 2, 5, and 8, 0.27-0.50 d -1 ) (Fig. 2a). The low growth rates in winter could be explained by low assimilation rates of nutrients (Fig. 3). 305 These results suggest that the magnitude of a subtropical phytoplankton bloom is regulated by additional seasonal factors such as PAR (Table 1).
The net phytoplankton growth rates were relatively low at Stations 18 and 21 in the ESP (0.43 and 0.40 d -1 , respectively, Fig. 2a), even for austral summer conditions with high PAR (Table 1). Since the ESP is known as a low dust deposition area, growth by the resident phytoplankton is considered to be limited by iron (Fe) (Jickells et al., 2005;Blain et al., 2008;310 Wagener et al., 2008;Moore et al., 2013). However, a Fe-enrichment incubation experiment at Station 18 demonstrated that there was no significant difference between the Fe-enriched and control bottles for phytoplankton, nutrients, DON, and DOP during a 96 h incubation, and also these parameter values were little changed from their initial values as the initial values lay within standard deviations of the mean values in the Fe-enriched and control bottles (Appendix methods, Table A1). Similar experimental results were reported by Bonnet et al. (2008) and they suggested that phytoplankton growth in the ESP is 315 limited by N rather than Fe. Furthermore, the deep waters used in this study were from a depth (1500 m) within North Pacific intermediate water that typically contains high dissolved Fe (>0.6 nM, i.e., not iron-limiting) (Nishioka et al., 2013;Nishioka et al., 2020). These lines of evidence imply that surface phytoplankton in the ESP and their bloom formation were not primarily limited by Fe. Although grazing by zooplankton is a possible factor controlling phytoplankton net growth, it was suggested to be not strong in the case of the subtropical phytoplankton blooms following deep water additions 320 (Mahaffey et al., 2012). There is further research required to understand the factors controlling the bloom development in the ESP.
The induced phytoplankton blooms in this study were accompanied by changes in community structure. Several nutrient enrichment experiments conducted in the subtropical oceans have demonstrated significant increases in eukaryotic phytoplankton following nutrient enrichments, particularly of N Moore et al., 2008;Mahaffey et al., 325 2012;Shilova et al., 2017;Rii et al., 2018;Lampe et al., 2019;Robidart et al., 2019). Similar blooms dominated by eukaryotic phytoplankton were also observed at most stations in this study as evidenced by the Tfuco increases (Fig. 2b). In addition, significant increases in cyanobacteria (Zea and DVchl a) following deep water additions were also observed particularly in the WNP and CNP (Figs. 2c and d). The relative proportions of cyanobacteria to eukaryotes in the treated bottles were lower in the ESP than the WNP and CNP (Fig. 2e). At Stations 18 and 21 in the ESP, the low proportions of 330 Prochlorococcus at time zero (Table 2) might influence the low proportions of cyanobacteria in the induced phytoplankton blooms. On the other hand, the relative increases of cyanobacterial abundances in the North Pacific experiments are likely driven by seasonal phytoplankton response to nutrient enrichment. Mahaffey et al. (2012) reported that there were less increases in eukaryote abundances in winter than in summer at Station ALOHA following deep water additions. Because of this, the relative increases in cyanobacteria in the winter-time North Pacific (Stations 2, 5, and 8) might be significant. 335 However, although Station A in the WNP was occupied in summer, we did not observe any significant eukaryotic bloom following deep water addition (Fig. 2e). This opposing trend at Station A could be due to regional differences of seasonal phytoplankton responses to nutrient enrichments, as summer eukaryotic blooms frequently occur in the eastern basin compared to the western basin in the North Pacific (Wilson, 2011;Villareal et al., 2012;Hashihama et al., 2014;Jiang et al. in press). 340

N and P drawdown characteristics
Our incubation experiments have revealed consistent linear decreases in DIN and PO4 concentrations, at nanomolar levels, along with the development of phytoplankton blooms (Figs. 3a and 3b). Additionally, accurate measurements of nanomolar inorganic N species detected the consistent increases in NO 2 concentrations in the treated bottles in which there were large decreases in NO 3 concentrations. This trend was particularly prominent at Station A (Figs. A1a and A1b). These NO 2 345 increases could be due to in vitro nitrification of NH 4 and/or incomplete assimilation of NO 3 by phytoplankton (Lomas and Lipschultz, 2006). While the factors controlling regional or seasonal differences in these NO 2 increases remains unknown, we demonstrate that sensitive measurements for multiple nutrients enable us to detect trace, but important, biogeochemical dynamics in response to the addition of nutrient-rich deep water.
Unlike the DIN and PO 4 , DON and DOP concentrations did not show any consistent changes over time despite the 350 occurrence of the phytoplankton blooms (Fig. A2). Although natural phytoplankton in the nutrient-depleted oligotrophic oceans show high affinity to DON and DOP (Karl and Björkman, 2015;Sipler and Bronk, 2015), the DON and DOP concentrations here did not indicate net drawdown in either the treated or control bottles. For the treated bottles, phytoplankton N and P demands were largely met by the enriched inorganic nutrients -which were not exhausted during the incubation periods (Figs. 3a and 3b). The DON and DOP dynamics during the development of phytoplankton blooms appear 355 to be in an equilibrium between uptake and release, as seen under ambient conditions like those in the control bottles (Fig.   A2).
Although the induced phytoplankton blooms in this study were solely dependent on DIN and PO4, regionally unique ∆DIN:∆PO 4 ratios were unveiled (Table 3). The ∆DIN:∆PO 4 ratios showed a geographical trend with relatively lower ratios in the PO 4 -depleted WNP (7.7-13.3) than in the other PO 4 -replete regions (14.1-30.7) although there were large variation in 360 95% confidence intervals (12.6-82.8). While the ratios in the PO 4 -replete regions were similar to the range (16-28) from the Redfield ratio to the subtropical particulate N:P ratios (Redfield, 1958;Martiny et al., 2013), those in the WNP were convergently lower than 16. Cellular N:P ratios in the subtropical phytoplankton are higher in cyanobacteria (25-35) than for eukaryotes (16) (Martiny et al., 2013). Because the phytoplankton blooms in the WNP were composed of communities with a relatively high proportion of cyanobacteria (Figs. 2e and 2f), the lower ∆DIN:∆PO 4 ratios (7.7-13.3) could not be explained 365 by phytoplankton cellular N:P ratios. Since macro-scale (>2000 km) exhaustion of PO 4 in the WNP is coupled with N 2 fixation Martiny et al., 2019), N 2 fixation by diazotrophic cyanobacteria might potentially meet some of phytoplankton N demand. However, assuming a relatively high N 2 fixation rate in the WNP (5 nmol N L -1 d -1 , , the contributions of N 2 fixation to ∆(DIN+N 2 ) were small (4-16%), leading to still lower ∆(DIN+N 2 ):∆PO 4 ratios (8.9-13.8) than 16. 370 Other explanations for the regionally unique ∆DIN:∆PO 4 ratios we observed (Table 3) are based on resident communities being 'primed' for rapid P acquisition. Lomas et al. (2014) reported that phytoplankton in the western subtropical North Atlantic have active PO 4 transporters which are rapidly induced under severely PO 4 -depleted condition. In addition, several studies have reported that microbial genes for high-affinity PO 4 transporter (pstSCAB) are enriched in the PO 4 -depleted regions in the western North Atlantic and WNP compared to the PO 4 -replete regions in the CNP and ESP 375 (Coleman and Chisholm, 2010;. These studies indicated that the phytoplankton in the PO 4 -depleted regions possess a high PO 4 uptake capability. Since N perturbation to phytoplankton little alters their cellular N quota (Moreno and Martiny, 2018), a high cellular P accumulation by the high uptake capability could induce the surplus drawdowns of PO 4 relative to DIN in the incubation experiments of the PO 4 -depleted WNP. Since DOP uptake and release in the treated bottles in the WNP were balanced (Fig. A2), the assimilated PO 4 would be sustained in the cellular P 380 components such as polyphosphate, that typically accumulates in particulate P in the PO 4 -depleted regions such as the western North Atlantic and WNP (Martin et al., 2014;.

Si and N drawdown characteristics
Previous field studies have reported that natural diatom blooms in the subtropical North Pacific were accompanied by high Si(OH) 4 :DIN drawdown ratios (>1) (Benitez-Nelson et al., 2007;Hashihama et al., 2014). However, in our incubation 385 experiments, most of the ∆Si(OH) 4 :∆DIN ratios were less than 1 and involved no significant values of ∆Si(OH) 4 (Table 3).
Furthermore, there was no significant ∆Si(OH) 4 even with a large increase (806 cells L -1 relative to the control) in diatom stocks at Station 15 (Table 3 and Figs. 2g and 3c). This increased diatom density is comparable to that in the natural bloom in the WNP reported by Hashihama et al. (2014). This mismatch between increased stocks and little change in ∆Si(OH) 4 implies that ∆Si(OH) 4 :∆DIN ratio is not so high in an early stage of diatom blooms. Si(OH) 4 uptake and release would be 390 balanced in the early bloom phase. In addition, low ∆Si(OH) 4 :∆DIN ratio in the early bloom phase was also observed in a mesoscale Fe enrichment experiment in the northeastern part of the subarctic Pacific, suggesting no Fe limitation of diatoms in the early bloom phase (Boyd et al., 2005). Probably, in a Fe-or DIN-depleted late stage of the bloom, selective Si(OH) 4 removal by diatoms (>1 of ∆Si(OH) 4 :∆DIN) occurs through putative biogeochemical processes such as selective Si export (Si pump), anomalous Si uptake associated diatom physiology, and/or Si uptake supported by N 2 fixation (Dugdale and 395 19 Wilkerson, 1998;Takeda, 1998;Boyd et al., 2005;Benitez-Nelson et al., 2007;Brzezinski et al., 2011;Krause et al., 2013;Hashihama et al., 2014).

Possible influences of the subtropical phytoplankton blooms on P* distribution
The present study is the first to reveal basin-wide distributions of surface P* in the oligotrophic Pacific Ocean using nanomolar N+N and PO 4 data (Fig. 5). The distributional pattern of surface P* coincided with that obtained in the upper 120 400 m using micromolar NO 3 and PO 4 data (Deutsch et al., 2007), indicating that a relatively homogenous P* condition prevails throughout the water column from the surface to 120 m depth. Based on the conventional concept, the upper layer P* is likely controlled by N 2 fixation and denitrification (Deutsch et al., 2007). However, by comparing surface P* distribution with experimentally determined ∆DIN:∆PO 4 ratios across the subtropical Pacific, we see insights into additional controls on P*. The low (≤13.3) and high (≥14.1) ∆DIN:∆PO 4 ratios geographically corresponded to the low and high P* in the WNP 405 (<50 nM, Stations A, 2, and 5) and the CNP and ESP (>50 nM, Stations 8, 15, 18, and 21), respectively (Fig. 5).
A comparison of the PO 4 -intercepts that were determined from the deep-water addition experiments with surface P* (from ambient concentrations) was conducted (Fig. 6). This cross-comparison was valuable since both represent metrics for excess PO 4 that remains after DIN exhaustion. The PO 4 -intercepts and P* are hereafter referred to as 'bloom P*' and 'ambient P*', respectively. The bloom and ambient P* showed a similar geographical trend both being relatively low in the 410 PO 4 -depleted WNP (Stations A, 2, and 5) when compared with the PO 4 -replete regions (Stations 8,15,18,and 21). The one to two orders of magnitude higher ambient P* than bloom P* in the PO 4 -replete regions suggests that, rather than phytoplankton uptake, denitrification (and also anammox) has a more pronounced influence on setting excess PO 4 in those regions. This trend may be particularly important in the ESP which is the vicinity of an oxygen minimum zone with active denitrification and anammox conditions (Paulmier and Ruiz-Pino, 2009). Alternately, N 2 fixation may exert an influence on 415 ambient P* in the PO 4 -depleted WNP. However, several studies reported that directly measured N 2 fixation rates are not consistently high in the WNP compared to other subtropical Pacific regions (Shiozaki et al., 2009;Shiozaki et al., 2010;Bonnet et al., 2017;. Given that natural phytoplankton blooms in the subtropical oceans have a large impact on nutrient dynamics through new production (Benitez-Nelson et al., 2007;McGillicuddy et al., 2007;Dore et al., 2008), the surplus PO 4 removal by phytoplankton bloom as observed in our experiments might play a significant role in 420 maintaining low ambient P* in the WNP. Furthermore, we found the unique result that bloom P* showed negative values in the PO 4 -depleted WNP (Stations A, 2, and 5), while the ambient P* did not exhibit negative values (Fig. 6). The difference between the bloom P* and ambient P* largely depends on the different N:P consumption ratios of ≤13.3 and 16, respectively. If the low N:P consumption ratios (≤13.3) are consistently dominant in the PO 4 -depleted WNP, alternative P sources other than PO 4 would be required to fully 425 exhaust DIN. Since lower DOP concentrations and higher alkaline phosphatase activity were observed in the WNP, compared to other subtropical Pacific regions , active DOP utilization in the WNP likely contributes to the DIN exhaustion. These perspectives suggest that, in the studies on subtropical nutrient 20 biogeochemistry using N:P stoichiometry, the bioavailable fraction of DOP could be an important factor as well as DIN and PO 4 . 430 Figure 6. Comparison of ambient P* (blue) and bloom P* (red) at seven stations. The bloom P* is identical to the experimentally determined PO 4 -intercept in Fig. 4. Error bars denote 95% confidence intervals. The ambient P* at Stations 8 and 21 have no error bar due to no triplicate data set. 435

Conclusions
By applying highly sensitive analytical methodology, we have revealed nutrient drawdowns and their ratios during the developments of phytoplankton blooms as induced by deep water additions to the surface water of the oligotrophic Pacific Ocean. The ∆DIN:∆PO 4 ratios showed a geographical variation from low in the PO 4 -depleted WNP to high in the PO 4replete ESP. While the ∆DIN:∆PO 4 ratios in the PO 4 -replete regions were similar to the range from the Redfield ratio to 440 typical subtropical particulate N:P ratio (16-28), those in the PO 4 -depleted regions (7.7-13.3) could not be expected from the conventional phytoplankton N:P ratios. The lower ∆DIN:∆PO 4 ratios were likely due to the high PO 4 uptake capability of low PO 4 -adapted subtropical phytoplankton. The regional trend in ∆DIN:∆PO 4 ratios was aligned with that of ambient P* in the oligotrophic Pacific. Although it remains necessary to examine nutrient assimilation characteristics in natural phytoplankton blooms, the regional variation in ∆DIN:∆PO 4 ratios as observed in our experiments appears to at least control 445 21 basin-scale ambient P* distribution in addition to conventional N 2 fixation and denitrification (also anammox). We have also demonstrated that accurate measurements of nanomolar nutrients are powerful tools in investigating trace nutrient dynamics.
Further application of these tools to the field and experimental studies would be beneficial for understanding of nutrient biogeochemistry in the oligotrophic ocean.

Appendix methods
The Fe-enrichment incubation experiment was conducted using the surface water (10 m depth) collected at Station 18 in the ESP (Table 1 and Fig. 1). Water sampling was performed using HCl-cleaned Teflon-coated Niskin-X bottles (General Oceanics) on a CTD system (Sea-Bird Electronics) attached to a titanium-armored cable. This sampling procedure succeeded in avoiding Fe contamination as reported previously (Shiozaki et al., 2018). The surface water was poured into 1.19 L 455 polycarbonate bottles and then Fe was enriched as iron chloride (FeCl3, Iron Standard Solution Fe 1000, Wako) at the final concentration of 1.8 nM. The triplicate bottles for either Fe-enrichment or control were prepared. These bottles were precleaned sequentially with neutral detergent, 1 M HCl, and 0.3 M hot HCl (for Analysis of Poisonous Metals, Wako), and filled with pure water for a day (Takeda and Obata, 1995). Both the Fe-enriched and control bottles were incubated for 96 h in the on-deck incubator as described in 2.2. After 96 h, the incubated bottles were sampled for nanomolar nutrients, DON, 460 DOP, and Tchl a. Initial samples for nanomolar nutrients, DON, DOP, and Tchl a were collected in duplicate directly from the Niskin-X bottles. The samples for nanomolar nutrients, DON, and DOP were processed and analysed as described in 2.3 and 2.4. For Tchl a here, a water volume of 100 mL was filtered onto GF/F filters, and the filter samples extracted with N,Ndimethylformamide (DMF, Wako) were analysed using a Turner Design fluorometer (Suzuki and Ishimaru, 1990). Student ttest was performed to determine significant differences (p<0.05) between the measured parameter values in the Fe-enriched 465 and control bottles.   Data availability. The data are available upon request to the corresponding author (Fuminori Hashihama).