The eastern subarctic North Pacific, an area of high nutrients and low chlorophyll, has been studied with respect to the potential for iron to control primary production. The geochemistry of zinc, a critical micronutrient for diatoms, is less well characterized. Total zinc concentrations and zinc speciation were measured in near-surface waters on transects across the subarctic North Pacific and across the Bering Sea. Total dissolved zinc concentrations in the near-surface ranged from 0.10 nmol L−1 to 1.15 nmol L−1 with lowest concentrations in the eastern portions of both the North Pacific and Bering Sea. Dissolved zinc speciation was dominated by complexation to strong organic ligands whose concentration ranged from 1.1 to 3.6 nmol L−1 with conditional stability constants (K′ZnL/Zn′) ranging from 109.3 to 1011.0. The importance of zinc to primary producers was evaluated by comparison to phytoplankton pigment concentrations and by performing a shipboard incubation. Zinc concentrations were positively correlated with two pigments that are characteristic of diatoms. At one station in the North Pacific, the addition of 0.75 nmol L−1 zinc resulted in a doubling of chlorophyll after 4 days.
 Co is preferentially utilized over Zn by some phytoplankton species or vice versa [Sunda and Huntsman, 1995; Saito et al., 2002; Xu et al., 2007; Saito and Goepfert, 2008], which suggests that the ratio of Zn to Co may influence species composition in the ocean [Sunda and Huntsman, 1995]. While this hypothesis has not been explicitly tested, the ability of Zn to influence species composition has been shown. In an incubation in the eastern subarctic N. Pacific, the addition of Zn and Fe together (+Zn/+Fe) resulted in a higher proportion of phytoplankton in the smaller size fraction (0.2–5 μm) compared to the addition of Fe alone [Crawford et al., 2003]. The community in the +Zn/+Fe addition had a higher abundance of small diatoms and small flagellates and less coccolithophores and ciliates than the Fe alone addition [Crawford et al., 2003]. Zn may influence not only which taxa dominate but which species within a taxa are most successful. In a bottle incubation experiment in the sub-Antarctic zone near New Zealand, the addition of Zn caused a community shift from a large colonial pennate diatom to a smaller, less-silicified, solitary pennate diatom species [Leblanc et al., 2005].
 Zn deficiency can also affect the extent of calcification of E. huxleyi, a ubiquitous marine coccolithophore [Schulz et al., 2004]. Under Zn limitation, E. huxleyi cells become more heavily calcified due to a slowing of growth rates with no corollary decrease in calcium carbonate (CaCO3) production rate [Schulz et al., 2004]. This effect, along with the potential of Zn to influence species composition, suggests that Zn may be an important determinant of the organic carbon:CaCO3 rain ratio. At low Zn:Co, coccolithophores and small cyanobacteria would be expected to dominate over diatoms based on their growth rates at low Zn:Co [Sunda and Huntsman, 1995]. This community shift would decrease the organic carbon:CaCO3 due to coccolithophores' CaCO3 shells. In addition, the coccolithophores growing under low Zn would be more highly calcified than those experiencing replete levels of Zn, further decreasing the organic carbon:CaCO3 rain ratio.
 Culture studies with the synthetic ligand ethylenediaminetetraacetic acid (EDTA) have shown that Zn bioavailability is related to the Zn free ion (Zn2+) concentration rather than the total Zn concentration [e.g., Anderson et al., 1978]. Natural organic ligands in the surface ocean strongly bind Zn [Donat and Bruland, 1990; Bruland, 1989; Ellwood and van den Berg, 2000; Ellwood, 2004; Jakuba et al., 2008]. These ligands dominate the speciation of the total Zn pool, with Zn2+ typically accounting for 5% or less of the total dissolved Zn. The inventory of Zn binding ligands can be extremely dynamic with production and removal of ligands occurring on timescales of 1 day [Lohan et al., 2005]. Zn speciation measurements of shipboard incubation experiments suggest that the phytoplankton community can survive at lower Zn2+ than is known to limit some large phytoplankton species in cultures [Ellwood, 2004; Lohan et al., 2005]. However, smaller diatoms such as Thalassiosira oceanica are able to growth at Zn2+ abundances below those that limit larger phytoplankton species [Sunda and Huntsman, 1992, 1995] suggesting that low zinc could cause species composition effects or that diatoms under these conditions can meet their Zn requirement by substituting Co or Cd.
 This study examines total dissolved Zn concentrations and Zn speciation across the subarctic N. Pacific and Bering Sea and the potential influence of Zn on phytoplankton growth, species composition and nutrient utilization. The eastern part of the subarctic N. Pacific, the Gulf of Alaska, is characterized by low levels of chlorophyll and high nutrient concentrations. Fe fertilization experiments have demonstrated the ability of Fe to regulate primary production in this region [e.g., Martin and Fitzwater, 1988; Boyd et al., 2004]. In the western part of the subarctic N. Pacific, there are relatively high rates of atmospheric deposition of Fe [Duce and Tindale, 1991] and a high seasonal export by diatoms [Honjo, 1997], though portions of Western Subarctic Gyre can also experience Fe limitation [Suzuki et al., 2005]. In this area, the east-flowing North Pacific Current originates from the Kuroshio Extension in the central Pacific. Enhanced vertical mixing events associated with variability in the strength of the North Pacific Current and with eddies of the system can bring nutrient-rich waters to the surface [Cummins and Freeland, 2007; Chu and Kuo, 2010] which likely contributes to the observed gradient in Zn concentrations from east to west noted by Fukuda et al. .
 The Bering Sea is split into a deep western basin whose surface waters are characterized by low summertime primary productivity and a wide continental shelf to the east that is highly productive, particularly at the shelf break with decreasing productivity toward the coast [Springer et al., 1996]. There is evidence that the deep western portion of the Bering Sea is an HNLC limited by Fe [Tyrrell et al., 2005; Aguilar-Islas et al., 2007].
2.1. Sample Collection and Handling
 Samples were collected in the N. Pacific and Bering Sea aboard the R/V Kilo Moana in the summer of 2003 (Figure 1). Water samples were collected using 10 L Teflon-coated Go-Flos (General Oceanics) on a Kevlar hydrowire. Seawater was collected from the Go-Flo bottles by connecting Teflon tubing to both ports and over-pressuring the top port with filtered ultra-high purity nitrogen gas. Seawater passed directly from Teflon tubing through an all plastic sandwich filter rig with acid-cleaned 0.4μm, 142 mm diameter polycarbonate membrane filters and into rigorously acid-washed (including sequential soaks in warm Citranox detergent, J.T. Baker Instra-analyzed HCl, J.T. Baker Instra-analyzed HNO3, and pH 2 Seastar HCl) low density polyethylene (LDPE) or Teflon bottles. Samples for total dissolved Zn analysis (LDPE bottles) were acidified to approximately pH 2 by the addition of 2 mL HCl (Seastar) per liter of seawater. Samples for Zn speciation analysis (Teflon bottles) were stored refrigerated (4°C) until analysis (typically within 10 days). All manipulation of the samples occurred in a laminar flow bench inside clean laboratories.
2.2. Total Dissolved Zn Analysis
 Total dissolved zinc (ZnT) concentrations were measured using isotope dilution and magnesium hydroxide pre-concentration followed by analysis using inductively coupled plasma mass spectrometry (ICP-MS) afterWu and Boyle  and Saito and Schneider . Centrifuge tubes (15 mL polypropylene, Globe Scientific) were cleaned by soaking in 2N HCl (J.T. Baker instra-analyzed) at 60°C for 48 h followed by rinsing 5 times with pH 2 HCl (J.T. Baker instra-analyzed) and once with pH 2 HCl (Seastar). Finally, tubes were filled to a positive meniscus with pH 2 HCl (Seastar) and capped until use. At the time of analysis, tubes were rinsed once with sample and then filled to approximately 13.5 mL (exact volume determined gravimetrically). Samples were then spiked with66Zn (98.9% as 66Zn, Cambridge Isotope Laboratories, Inc.) to an estimated 66Zn:64Zn ratio of 9. This ratio minimizes error magnification [Heumann, 1988]:
The added 66Zn spike was allowed to equilibrate with the samples overnight. The following day, 125 μL of ammonia (Seastar) was added to each tube. After 90 s, the tube was inverted and after an additional 90 s, tubes were centrifuged for 3 min at 3000 × g (3861 rpm) using a swinging bucket centrifuge (Eppendorf 5810R). The majority of the supernatant was carefully decanted. Tubes were then respun for 3 min to firm pellet and the remaining supernatant was shaken out. Pellets were dissolved on the day of ICP-MS analysis using 0.5–1.5 mL of 5% nitric acid (Seastar). ZnT concentrations were calculated as
where f64 is the fraction of the abundance of 64Zn over the abundance of all the Zn isotopes and R is the ratio of 66Zn:64Zn. To measure the procedural blank, 1 mL of low zinc surface seawater was treated the same as samples, and calculations were performed as though it was a 13.5 mL sample (Zn contribution from the 1 mL is considered negligible). The average blank value was 0.12 nmol L−1 with a detection limit of 0.09 nmol L−1 (calculated as three times the standard deviation of the blank). The daily procedural blank value was subtracted from measured sample values.
 The efficiency of the magnesium hydroxide precipitate at scavenging Zn was tested. One mL aliquots of acidified seawater (n = 3) were equilibrated with the radioisotope 65Zn (approximately 0.5 μCi) for 2 h. Thirty μL of ammonia was added to each sample and after 1.5 min, the samples were centrifuged for 3 min. The amount of 65Zn was quantified in the seawater before precipitation, in the precipitate, and in the supernatant using a sodium iodide detector. The percent of 65Zn that was captured, on average, by the magnesium hydroxide pellet was 96% and the fraction remaining in the supernatant was 2%. The accuracy of the method was evaluated by measuring a NASS-5 seawater standard (National Research Council of Canada). The NASS standard has a certified value of 1.56 ± 0.60 nmol L−1 Zn. The value obtained by this method (1.00 ± 0.03 nmol L−1) was within the specified range of the certified value.
 ICP-MS measurements were made using a Thermo Finnigan ELEMENT2 in medium resolution mode, which was sufficient to resolve64Zn from the potential interference peak due to Mg-Ar. Another potential interference at mass 64 is that of64Ni, which is a minor isotope of Ni (natural abundance of less than 1%). This interference was not measured explicitly. In many cases, the contribution of the mass 64 signal due to Ni will not be significant relative to 64Zn, however at the lowest Zn concentrations measured here, the contribution due to Ni may be significant. Total dissolved Ni concentrations were measured in this study using voltammetric methods (see below). The potential influence of the 64Ni on the ZnT concentrations calculated here is discussed in the Results section.
2.3. Total Dissolved Ni Analysis
 Total dissolved nickel (Ni) was also measured here as described in Saito et al. . Briefly, a Metrohm 663 hanging mercury drop electrode stand with a Perfluoroalkoxy Teflon (PFA) sample vessel was interfaced with an Eco-chemieμAutolab system and GPES (General Purpose Electrochemical System, Eco Chemie) software. Dimethylglyoxime (DMG) from Aldrich was recrystallized in Milli-Q (Millipore) water in the presence of 10−3ethylenediaminetetraacetic acid (EDTA; Sigma Ultra) to remove impurities, dried, and redissolved in high performance liquid chromatography (HPLC)-grade methanol. We prepared 1.5 mol L−1sodium nitrite (Fluka Puriss) purified by an overnight equilibration with prepared Chelex-100 beads (BioRad). An N-(2-hydroxyethyl)piperazine-N′-(3-propanesulfonic acid) (EPPS) buffer solution (Fisher) was purified by passing through a column with 3 mL of prepared Chelex-100 beads. For total dissolved nickel analyses, 8.50 mL of ultraviolet light irradiated filtered seawater was pipetted into the Teflon sample cup followed by 50μL of 0.5 mol L−1 EPPS, 20 μL 0.1 mol L−1 DMG, and 1.5 mL of 1.5 mol L−1 sodium nitrite. The μAutolab protocol involved a 3-min 99.999% N2gas purge at 120 kPa, a 90-s deposition time at −0.6 V, a 15-s equilibration period, and a high-speed negative scan from −0.6 to −1.4 V at 10 V s−1. A linear sweep waveform was used on drop size three (0.52 mm2) and a stirrer speed of five. Total nickel concentrations were determined with three standard additions of 1 nmol L−1 using a NiCl2solution prepared in pH 2 Milli-Q water.
2.3. Zn Speciation Determinations
 Zn speciation was determined using an anodic stripping voltammetry method [Jakuba et al., 2008] that was adapted from that of Fischer and van den Berg  for the measurement of total lead and cadmium. Measurements were made using a 663 VA stand (Metrohm) consisting of a glassy carbon rotating disc working electrode (2 mm diameter), a glassy carbon rod counter electrode, and a double junction Ag, AgCl, saturated KCl reference electrode. The electrode was interfaced to a PC (IBM Thinkpad) using an IME663 and μAutolab I (Eco Chemie). Before each day of speciation analysis, the glassy carbon rotating disc electrode (RDE) was manually polished with aluminum oxide. Then, a cyclic voltammetry step was performed cycling the voltage between −0.8 V and 0.8 V 50 times. Reagents used included 4-(2-Hydroxyethyl)-1-piperazinepropanesulfonic acid (EPPS), ammonium thiocyanate, and mercuric chloride. The pH of the EPPS buffer was adjusted to 8.1. EPPS and thiocyanate reagents were run through a pre-cleaned chelex column [Price et al., 1989] to remove trace metal contamination. The thiocyanate reagent is added to improve the reproducibility of the mercury film and to ensure full de-plating of the film between samples [Fischer and van den Berg, 1999].
 Zn speciation titrations were set up in 15 mL Teflon vials (Savillex). In order to prevent wall loss, Teflon vials were equilibrated with specific concentrations of Zn before use and each vial was always used for the same concentration of Zn. Before a titration, the vials were rinsed with 12 mL of sample. Then, 12 mL of sample was added into each vial, along with 1.7 mmol L−1 EPPS. For each sample aliquot, the appropriate concentration of Zn was allowed to equilibrate with the sample in the sample vial for 10 min. Then, the sample aliquot was poured into the voltammetric cell and 4.2 mmol L−1 ammonium thiocyanate and 10.4 μmol L−1mercuric chloride were added. Each sample aliquot was purged with ultra-high purity nitrogen gas for 5 min to remove oxygen. A conditioning potential of 0.6 V was held for 60 s. The mercury film and Zn′ were then deposited at a potential of −1.5 V for 180 s. After a 10 s equilibration time, the voltage was ramped in square wave mode from −1.3 V to 0.6 V. A frequency of 50 Hz was used and the step potential and modulation amplitude were 4.95 mV and 49.95 mV, respectively. A peak in current was evident at −1.1 V representing the concentration of inorganic Zn species (Zn′). The total ligand concentration ([LT]) and conditional stability constant (K′cond,Zn′) with respect to Zn′ were calculated by performing a linearization of the titration data [Ružić, 1982; van den Berg, 1982], where [Zn′]/[ZnL] is plotted versus [Zn′]. The equation for the resulting line is
Hence, [LT] can be calculated as and K′cond,Zn′ as . The Zn2+ concentrations in Table 1 were calculated based on the relationships:
[Zn′] was calculated from equation 3, assuming that [L]T − [Zn]T = [L′] and that [ZnL] = [L]T − [L′]. A value of 2.2 was used for αZn [Turner et al., 1981].
Values in parentheses represent the standard deviation of duplicate or triplicate analyses for ZnT. The majority of the SRP values were measured from sub-samples collected from Go-Flo bottles. Where sub-samples from Go-Flo bottles were unavailable, values are reported from samples collected from Niskin bottles and measured on-board (bold values).
Potentially high level of Ni interference, see section 3 for details.
Samples where the SRP values measured from the Go-Flo sub-samples did not match expected oceanographic depth trend. When available these values were replaced by SRP measured on-board from Niskin bottles. Seesection 3 and Figure 4a.
Contamination suspected, see section 3 for details.
 Trace metal clean water was collected with an air-driven Teflon pump that pumped water from approximately 15 m depth directly into an acid-washed HDPE carboy, housed in a trace-metal free bubble constructed of a HEPA filter and plastic sheeting. Acid-washed polycarbonate bottles were filled with unfiltered seawater from the carboy. A time zero sample for chlorophylla was also collected from the carboy. Additions were made to duplicate polycarbonate bottles as follows: control (no addition), +Fe (2.5 nmol L−1 FeCl3), +Zn (0.75 nmol L−1 ZnCl2), +Zn/+Fe (0.75 nmol L−1 ZnCl2, 2.5 nmol L−1 FeCl3). Bottles were tightly capped and placed in an on-deck water bath supplied with flowing seawater for temperature control. Sunlight was attenuated with bluegel shading (Roscolux 65: Daylight Blue, Stage Lighting Store) to mimic 15 m irradiances. On days 2 and 4, one bottle was removed from the incubator and sampled for chlorophyll. In order to avoid potential contamination by Zn, which is a highly contamination prone element, bottles were sacrificed upon sampling. Samples for analysis of dissolved nutrients from select bottles were also collected, filtered, and frozen.
2.5. Chlorophyll, Nutrients, and HPLC Pigments
 For chlorophyll a analysis, seawater was passed through a GF/F filter. Filters were extracted in 90% acetone overnight at −20°C. Chlorophyll a (Chla) concentrations were corrected for the interference of phaeophytin by the addition of acid. All samples for nutrient analysis were filtered through the trace metal filtration rig described above into acid-cleaned polypropylene tubes and stored frozen until analysis. Analysis was performed by the Ocean Data Center (ODC) at the Scripps Institution of Oceanography for nitrate, nitrite, ammonium, silicic acid, and phosphate. In addition, phosphate analyses were run on board at select stations using the molybdate blue method [Murphy and Riley, 1962]. In general, analyses from ODC are used here, except on several occasions where frozen nutrient values were not oceanographically consistent due to preservation problems [Maher and Woo, 1998, and references therein] and shipboard analyses were substituted. Besides these few outliers, shipboard and frozen samples compared well. Samples were collected for phytoplankton pigment analysis from Niskin bottles on a CTD rosette frame. Analysis of HPLC pigments was performed after Bidigare  and Bidigare and Trees .
3.1. Total Dissolved Zn
 ZnTwas measured in the near-surface at 5 stations in the N. Pacific and 4 stations in the Bering Sea. ZnTconcentrations in the near-surface ranged from 0.10 to 1.15 nmol L−1 (Table 1). Concentrations increased from east to west in the North Pacific, in agreement with reports of integrated ZnT in the upper 100 m of this region [Fukuda et al., 2000]. The ZnT value at Station 1 (0.10 nmol L−1) is within the range reported for the near-surface in that vicinity (0.06–0.25 nmol L−1) [Martin et al., 1989]. During the SEEDS Fe enrichment experiment, which was performed further west of Station 6 (48.5°N, 165°W), near-surface ZnT concentrations were even higher than those observed at Station 6 (1.18–2.35 nmol L−1) [Kinugasa et al., 2005], which may indicate that the east–west trend of increasing ZnT continues beyond our study region.
 Near-surface ZnT concentrations in the Bering Sea were high (∼1 nmol L−1) in the deep western portion of the basin and low in the eastern waters over the Bering shelf. Previous studies of Zn in the Bering Sea have shown elevated ZnT concentrations at depth (3.5 nmol L−1 at 30 m and 20 nmol L−1 at 3000 m) in the western Bering Sea [Fujishima et al., 2001], but low surface ZnT concentrations (<0.41 nmol L−1) [Fujishima et al., 2001; Leblanc et al., 2005]. At Station 8, the mixed layer was only 15 m, and the SRP concentration at 20 m was 1.56 μmol L−1 compared with only 1.25 μmol L−1 at the surface. Thus, the high ZnTat Station 8 is likely a result of sampling below the mixed layer. The mixed layer at Station 9 is about 18–20 m and the near-surface sample from this station was taken from 20 m. Unfortunately, nutrient data is not available for Station 9 to confirm that the high ZnT value was collected from a depth of active nutrient remineralization. The western Bering Sea can experience high dust inputs from Asian deserts, which may contribute to relatively high Zn concentrations in this area [Boyd et al., 1998; Nishioka et al., 2003]. Lower ZnT was observed in the eastern Bering Sea on the Bering Shelf where Chla concentrations were high, possibly reflecting biological uptake.
 As mentioned in the Methods section, the possible interference of 64Ni contributing to our calculated ZnT concentrations was not measured explicitly. The NiTconcentrations in those near-surface samples measured ranged from 3.3 to 8.2 nmol L−1 (Table 1). While this NiT technique has somewhat variable precision due to Ni's slow kinetic properties and the small mercury drop surface of the Metrohm 663 hanging drop mercury electrode, the surface and deep water values reported here were generally consistent to that reported by Bruland  for the Central North Pacific. These Ni concentrations could cause an overestimation of our ZnT values that was generally less than 10% of ZnT values; however at the lowest ZnTvalues the potential for the Ni interference is more significant. For example, in the near-surface at Station 1, the64Ni concentration was 30 pmol L−1, which could account for approximately 30% of the ZnT measured at this depth. The three samples where the 64Ni could account for over 10% of the ZnT concentration are noted in Table 1. The potential contribution of NiT to the deeper ZnT values was generally very low (average possible contribution of less than 3%).
 Profiles of 8–11 depths were collected at three stations in the N. Pacific (Figure 2). Zinc profiles exhibited nutrient-like behavior, in agreement with previous studies in the region [Bruland et al., 1978; Bruland, 1980; Martin et al., 1989; Bruland et al., 1994; Lohan et al., 2002]. At the majority of stations, the near-surface samples collected were close to the base of the surface mixed layer (e.g., Station 4,Figure 3). In the N. Pacific, the seasonal thermocline extended down to roughly 40–150 m. In the Bering Sea, the thermocline occurred over shallower depths (25–50 m).
 A plot of ZnT versus soluble reactive phosphorus (SRP), shows good agreement between this study and Martin et al.  from the N. Pacific in the vicinity of Station 1 (Figure 4a), with a few exceptions. At high SRP and Zn, four data points fall above the trend (Station 4, 500 m; Station 4, 3000 m; Station 6, 1000 m; Station 6, 3000 m). They occur at depths in the profile (1000–3000 m) where SRP concentrations are expected to remain relatively constant with depth, yet the SRP in these samples is significantly lower than the two depths surrounding it or the depth above it in the case of the 3000 m points. SRP was measured on board the ship and on frozen samples for this depth. The shipboard value of 1.7 μmol L−1 SRP is significantly higher than the frozen sample of 1.2 μmol L−1 and is a value that matches the trend. The anomalously low SRP values suggest that in some cases SRP may have been lost during storage before analysis by ODC, a potential problem of sample storage [Maher and Woo, 1998, and references therein]. If the SRP values for these four data points are replaced with the average value for the two surrounding depths or the value above it in the case of 3000 m, the two shallower samples (Station 4, 500 m; Station 6, 1000 m) fall back in line with the expected trend (Figure 4b). The final two data points (Station 4, 3000 m; Station 6, 3000 m) fall closer to the line but still seem to have higher than expected ZnTconcentrations. The 3000 m samples were collected with Go-Flo bottles attached to the ship's CTD rosette frame, so they may have been contaminated. A final point that seems to fall somewhat off the trend is that at 1.2μmol L−1 SRP and 2.7 nmol L−1 ZnT (Station 6, 80 m). Once again, when the ODC SRP value is replaced with the shipboard SRP value, the data point conforms with the trend.
 A tight correlation was observed between silicic acid (Si) and ZnT concentrations (Figure 5, R2 = 0.93). The highest Zn values again fall above the trendline–the likely contamination of these samples has already been noted. The Zn/Si relationship is generally linear, but data from both this study and Martin et al. trend slightly above the linear relationship in the mid-values.
 Zinc concentrations reached maximum concentrations in deep waters of 10–10.5 nmol L−1 (excluding the presumably contaminated 3000 m data) in agreement with Martin et al. who reported similar concentrations at 1500 m for VERTEX stations T-5 thru T-8 andFujishima et al.  who found deep water ZnT concentrations across the subarctic N. Pacific to be 10–11 nmol L−1. In the central North Pacific, between 20–30°N, ZnT concentrations reached maximums of only about 9.5 nmol L−1 [Bruland et al., 1978; Bruland, 1980; Bruland et al., 1994]. The higher deep Zn values in the subarctic N. Pacific relative to the central N. Pacific are consistent with the increasing time for remineralized Zn to accumulate in deep waters as they move northward. At Station 4, ZnTconcentrations from the same profile were also measured using isotope dilution, magnesium hydroxide pre-concentration coupled with an anion exchange resin, and the results from that method showed excellent agreement with this study [John, 2007].
3.2. Zn Speciation
 Zn speciation titrations were performed on near-surface samples for stations 1–8. The concentration of Zn binding ligands (LT) ranged from 1.1 to 3.6 nmol L−1in the near-surface with an average concentration of 2.3 nmol L−1 (Table 1). The concentration of LT followed a general trend of increasing concentrations from east to west in the N. Pacific, similar to ZnT concentrations (Figure 6). The highest concentration of LT was observed at Station 8 in the western Bering Sea. The K′cond,Zn′ of the natural organic ligands ranged from 109.3 to 1011.0 with an average value of 1010.2 (Table 1). The resultant Zn2+ concentrations ranged from 2–71 pmol L−1, which represented 3% on average of the total Zn concentration. Where multiple depths of a profile were analyzed for speciation, the Zn binding ligands had similar concentrations and conditional stability constants at each depth (Table 1). Free Zn2+concentrations in the near-surface were very low at the eastern North Pacific stations (2–7 pmol L−1), similar to previous studies in this area [Bruland, 1989; Donat and Bruland, 1990; Lohan et al., 2005]. The Zn2+ concentrations at the westernmost station in the North Pacific and in the Bering Sea were much higher at 33 and 71 pmol L−1, respectively.
 The lowest Zn2+ concentrations were observed in the eastern portion of the subarctic N. Pacific where the majority of previous studies have focused. Ligand concentrations of 1.0–2.2 nmol L−1 that had K′cond,Zn′'s of 1010.6–1011.2 were observed in the central and northeast North Pacific by these workers. However, the ligand concentrations were higher and the K′cond,Zn′'s lower here than those reported by Bruland . The differences cancel out when determining free Zn2+, hence the agreement. K′cond,Zn and [LT] observed in this study are more similar to those in subantarctic waters near New Zealand where the [LT] ranged from 1.3–2.5 nmol L−1 and K′cond,Zn′ was between 109.7 and 1010.4 [Ellwood, 2004].
 Culture studies have shown that E. huxleyi can exude Zn binding ligands [Vasconcelos et al., 2002]. In this study, though the concentrations of LT followed a trend similar to Chla, there was not a significant relationship between Chla and [LT] (R2 = 0.51, p > 0.1).
 The [LT] at Station 8 in the Bering Sea (3.6 nmol L−1) is the highest reported value of Zn binding ligands in the open ocean measured by voltammetric methods to the authors' knowledge. Ligand concentrations as high as 6 nmol L−1 have been reported in bottle incubations after the addition of Zn [Lohan et al., 2005], and much higher concentrations (>20 nmol L−1) have been observed in coastal waters and phytoplankton cultures [van den Berg, 1985; Vasconcelos et al., 2002]. Station 8 had the highest ZnT concentrations of any station where Zn speciation measurements were performed (0.89 nmol L−1). The pool of Zn binding ligands is dynamic and can change on short timescales in response to the addition of Zn [Lohan et al., 2005]. The high LT concentrations may reflect a phytoplankton community that is well adapted to synthesizing additional ligands in response to atmospheric dust, which can be deposited to the Bering Sea from Asian deserts [Duce and Tindale, 1991].
3.3. Shipboard Incubations
 Results from the shipboard incubation that began at Station 5 are presented (Figure 7). At time zero of the incubation, total Chla concentrations were 0.99 μg L−1. The dominant phytoplankton pigments at Station 5 were hexanoyloxyfucoxanthin, fucoxanthin, chlorophyll c, and chlorophyll b. This pigment signature is consistent with an initial phytoplankton community dominated mainly by diatoms, prymnesiophytes, and green algae [Mackey et al., 1998]. The near-surface ZnT concentration at Station 5 was 0.65 nmol L−1. Chlorophyll a values remained constant in the control treatment over 4 days (Figure 7). In the +Zn treatment, Chla concentrations increased over the control by 25% on day 2 and reached twice the value in the control on day 4. An even more pronounced effect was seen in the treatments where Fe was added (+Fe, +Zn/+Fe). In these treatments, Chla increased by between 60 and 80% over the control on day 2 and reached over 3 times the Chla values observed in the control on day 4. At the end of the experiment, nutrients were most depleted in the treatments where Fe was added. Nutrient concentrations in the +Zn treatment were intermediate between those of the control and the Fe additions. The drawdown of nutrients in the incubation bottles matched the Chla trend (Table 2). Incubation experiments that included Zn additions were performed at 7 other stations. No increase in response to Zn addition was observed at any other station (Table 3). Previous shipboard incubations in the N. Pacific have shown little effect on total Chla concentration due to the addition of Zn [Coale, 1991; Crawford et al., 2003] or no effect [Lohan et al., 2005; Leblanc et al., 2005]. There is no obvious explanation for the anomalous result at Station 5. Near-surface ZnT concentrations at Station 5 were relatively high, and the concentration of Zn2+ was not particularly low at 7 pmol L−1. The pigment distribution at Station 5 was very similar to that at Station 6, where Zn addition had no effect. Fe contamination is a possible explanation for the positive result to Zn addition at this site. However, no growth was observed in the control bottles or in a treatment of 500 pmol L−1 added Co (M. Saito et al., unpublished data, 2003). Thus it seems unlikely that contamination would have randomly occurred in the two +Zn bottles that were sacrificed at successive time points and not in any of the control or +Co bottles.
Table 2. Time Final (Day 4) Nutrient Concentrations and Nutrient Drawdown Ratios From the Shipboard Incubation Performed at Station 5 in the North Pacific
NO3 (μmol L−1)
SRP (μmol L−1)
Si (μmol L−1)
Table 3. Summary Table of Incubation Experiments Performed in the North Pacific and Bering Sea During the Summer of 2003a
Stations 1–6 are the subarctic North Pacific; stations 8–10 are the Bering Sea. An + indicates that a treatment resulted in an increase in chlorophyll aabove a no-addition control, whereas a – indicates no increase observed. Where spaces are empty, the treatment was not performed at that station. +Fe additions were performed by Saito et al. (unpublished data, 2003) at stations 1, 3 and 4. A chlorophylla increase due to Fe addition was observed at stations 3 and 4 but not at station 1.
3.4. Chlorophyll, Nutrients, and HPLC Pigments
 In the N. Pacific, near-surface SRP was lowest in the east and increased significantly between Station 1 and Station 3 and then remained relatively constant (Table 1). In the Bering Sea, SRP was highest in the west and decreased significantly on the Bering shelf. There were twelve samples where SRP was measured both by the Popp group at sea and from frozen samples by the ODC. When the SRP values from both analysts are plotted against one another, the vast majority of the samples fall within 10% of the 1:1 line (data not shown). Two exceptions are the samples from Station 4, 60 m and Station 6, 80 m. In both cases, the values measured by the ODC (1.20 μmol L−1, 1.23 μmol L−1) were significantly lower than those measured at sea (1.63 μmol L−1, 1.74 μmol L−1). The ODC SRP values for these two samples were also lower than would be expected based on the SRP concentrations from the shallower depths of the same profile. For both these samples, the values measured at sea are used in Table 1 and in Figures 4 and 9.
 Nitrate plus nitrite values in the near-surface were also very low at Station 1 (<0.1μmol L−1) and much higher and more consistent over the western half of the N. Pacific transect (∼15 μmol L−1). Silicic acid followed the same trend (Station 1: 0.5 μmol L−1; Sts. 4–6: 14 μmol L−1).
 Total Chlaconcentrations in the near-surface were lowest in the eastern N. Pacific (Table 1). Chlorophyll a increased moving west from Station 1 to Station 5. Moderate Chla values were observed for stations 6 through 10 in the western N. Pacific and Bering Sea, and the highest Chlaconcentration encountered was at Station 11 on the Bering Shelf. The most abundant phytoplankton marker pigments (i.e., non-Chla) were hexanoyloxyfucoxanthin, fucoxanthin, chlorophyll c, and chlorophyll b. The pigment signatures agree with previous examinations of phytoplankton pigments in the North Pacific and are consistent with phytoplankton communities containing prymnesiophytes, diatoms, pelagophytes, and green algae. In the northeast subarctic Pacific at Ocean Station Papa (50°N, 145°W), the phytoplankton community is typically dominated by small prymnesiophytes (such as E. huxleyi), pelagophytes, small diatoms, and cyanobacteria, with larger diatoms and other large phytoplankton contributing 10–30% of the total Chla [e.g., Boyd and Harrison, 1999; Thibault et al., 1999]. The waters of the northwest subarctic Pacific often host phytoplankton communities with abundant prasinophytes and a higher relative abundance of diatoms than in the eastern subarctic North Pacific [e.g., Suzuki et al., 2002; Obayashi et al., 2001]. Hexanoyloxyfucoxanthin was the most abundant marker pigment at Sts. 1, 2, 3, and 10, while fucoxanthin was dominant at Sts. 4, 5, 6, 8, and 11. This pattern meshes with the above studies and indicates that at the eastern North Pacific stations the community was likely dominated by prymnesiophytes, while at the western North Pacific stations, diatoms were more likely to dominate. The higher concentrations of fucoxanthin at the western North Pacific stations is consistent with the proliferation of diatoms in waters containing relatively high levels of macro- and micronutrients.
 Phytoplankton pigment concentrations were compared to Zn concentrations from those samples in the upper thermocline where Zn speciation data were available. This criterion excludes data from St 6, 40 m and St 4, 100 m, where significant mixing with deep waters is evidenced by the more than doubling of the total Zn concentrations relative to mixed layer values. There was not a significant relationship (p > 0.05) between ZnT concentration and total Chla (Figure 8a). However, for two major pigments found in diatoms, fucoxanthin (p < 0.01) and chlorophyll c (p < 0.05), positive relationships were observed with ZnT (Figures 8b and 8c), which may indicate the relative importance of Zn to this fraction of the phytoplankton community. Diatoms have high Zn quotas relative to other taxa [Sunda and Huntsman, 1995]. The relationship observed here is consistent with Zn being an essential micronutrient for diatoms.
 Significant correlations (p < 0.01) were also observed between both SRP and silicic acid and fucoxanthin (Figure 9) but nitrate was not as strongly correlated. When pigment concentrations were compared to the free Zn2+ concentration rather than total Zn concentration, no significant correlations were observed (Figure 10). Instead, aside from two high points, most of the data were below 10 pmol L−1, regardless of pigment concentration, suggesting that at very low concentrations, free Zn2+ is relatively invariant with biology within the capabilities of this methodology.
 Two minor pigments, alloxanthin and chlorophyllide a, also exhibited positive correlations with ZnT (data not shown). Alloxanthin is a marker for cryptomonads and was found in very low quantities (average relative contribution to total Chla ∼1%). Chlorophyllide a is a degradation product of Chla and was observed in somewhat higher concentrations (average relative contribution to Chla < 9%). Degradation of Chla to chlorophyllide a can occur during copepod grazing and during the filtration of some diatoms [Klein and Sournia, 1987, and references therein]. These relationships are based on limited data, and interpreting pigment data can be difficult because organisms contain multiple pigments and the pigment composition of field organisms does not accurately match those of cultured organisms.
 A prominent feature of the data presented here is the strong gradient in zinc concentration and zinc speciation in surface waters on the east–west transect. Increasing zinc concentrations toward the west were matched by increasing ligand concentrations, leading to relatively little variability in free Zn2+ in the mixed layer until the westernmost station (Station 6). ZnT at Station 6 is higher in surface waters, and free Zn2+ increases to values seen only in the subsurface further east. Together, these results show an emerging picture of a large region in the central and eastern N Pacific where mixed layer free Zn2+ is less than 10 pmol L−1. This region is bounded by waters underlying the mixed layer, where ZnT and free Zn2+are much higher, and by nutrient-rich surface waters in the west. Increasing Zn levels toward the west are associated with an increase in biomass and diatoms, as indicated by Chla and fucoxanthin data. These increasing zinc abundances may be one of the factors, with the increased macronutrients, that contributes to this trend toward higher primary productivity in the west. Establishing a causal relationship between Zn and biological activity is premature, since Zn additions to bottle incubations resulted in a positive response at only one station.
 Free Zn2+ is presumed to be the biologically available form [Sunda and Huntsman, 1992, 1995] so the observation that total dissolved Zn has a stronger correlation with pigment parameters is surprising. However, for acquisition of some micronutrients like Fe and Co, some phytoplankton appear to be capable of accessing the organically complexed forms [Maldonado and Price, 1999; Saito et al., 2002], hence free ion concentrations may be less relevant than total concentrations for some species because of these high affinity uptake systems that render organically complexed forms biologically available. Correlations here also reflect the fact that ZnT can be measured much more precisely than the scarcer free Zn2+, making it difficult to look for correlations between the latter and other parameters.
 In spite of these limitations, our free Zn2+ data enable us to compare field data with culture studies, which are typically carried out using metal ion buffers like EDTA, where effects (e.g., specific growth rates, cellular Zn quota) are reported as a function of free Zn2+. Comparison of our free Zn2+ data with phytoplankton culture studies suggest that Zn is a limiting micronutrient for some phytoplankton, in particular larger diatoms [Sunda and Huntsman, 1995], and is low enough to induce biochemical responses such as substitution of Co for Zn [Yee and Morel, 1996; Xu et al., 2007; Saito et al., 2008].
Sunda and Huntsman [1992, 1995] demonstrated a robust relationship between free Zn2+ and Zn:C in phytoplankton cultures, building on earlier studies [Sunda and Guillard, 1976; Anderson et al., 1978]. Using a Redfieldian-type analysis, they estimated Zn:P ratios in phytoplankton biomass through analysis of the dissolved Zn:P data in the upper 1000 m of the North Pacific reported inMartin et al. . While Zn shows the strongest correlation with Si over the entire water column, the strong correlation with P in the upper water column indicates a close coupling of Zn and nutrient remineralization. Therefore, Sunda and Huntsman  argue that these dissolved ratios presumably reflect the Zn:P of remineralized phytoplankton biomass, and using their data for Zn:C in culture, they estimated the corresponding free Zn2+ concentration required to produce a sufficiently large Zn:P ratio. Their calculations suggest that the Zn:P ratio from the nutricline could only arise from diatoms growing with a free zinc abundance of ∼1000 pmol L−1. The estimated “nutricline value” for free Zn2+ is an order of magnitude higher than the highest value reported here and in Lohan et al. . The only source of nutrient-rich, high Zn water to lead to such conditions is substantially deeper than 100 m, yet the mixed layer does not penetrate that depth in this region, even during the winter [Anderson et al., 1969]. Therefore, the Zn must come from elsewhere. One possibility is that episodic deep mixing events, perhaps within boundary regions, lead to conditions with elevated Zn. A major shortcoming of all Zn studies, including ours, is that they do not examine the dynamics of Zn injection and drawdown during bloom periods. For instance, it is possible that deep mixing events preceded the SEEDS experiment in the western North Pacific, when surface Zn was extremely high [Kinugasa et al., 2005].
 The Zn:P ratio calculated from data between 50 to 150 m is much smaller, and yields a smaller cell quota and free Zn2+ abundance of 0.8 to 2 pmol L−1. The free Zn2+ abundances estimated from the shallow dissolved Zn:P ratios are quite similar to that measured in this study in the low picomolar range (Table 1), suggesting that much less bioavailable Zn is in the photic zone and which also is within the range of what would affect phytoplankton species composition [Sunda and Huntsman, 1995].
 In summary, data from a variety of sources, including culture work and measurements of remineralized Zn:P inferred from dissolved Zn and phosphate abundances, provide a mechanism for Zn cycling that is consistent with the free ion model. A principal, if accidental role of organic ligands is to act as a natural Zn buffer. Complexation decreases the rate of “luxury” uptake or non-biological scavenging under conditions of high particle scavenging, leaving a residual concentration available for organisms with high affinity transport systems. From Sunda and Huntsman's work, this would appear to include small diatoms, but not larger ones. Zn limitation will probably induce community shifts to species or biochemistries with lower Zn requirements before total Zn is completely removed. Thus organic complexation provides a useful natural service to organisms with a small, but nonzero Zn requirement.
 The inter-replacement of Co and Zn as micronutrients has been shown in phytoplankton cultures [Saito et al., 2008, and references therein]. Therefore, Co distributions should also be considered in evaluating the biological implications of our data, particularly for station 5, the only location where we had evidence for Zn limitation. A strong depletion in near-surface Co concentrations was observed at Station 5, with some of the lowest Co values observed on this transect (∼30 pmol L−1; M. Saito, manuscript in preparation, 2012). The low concentration of Co is consistent with the biochemical substitution of Co for Zn under low Zn conditions and with the observation that Co draw down in the North Pacific typically occurs when Zn concentrations are depleted relative to SRP concentrations [Sunda and Huntsman, 1995].
 Much of the region we surveyed exhibits Fe limitation to varying degrees [Martin and Fitzwater, 1988; Kinugasa et al., 2005]. The relationship between Fe and Zn limitation has been studied previously in cultures and deck-board incubations. In cultures of the diatomT. weissflogii, Fe and Zn deficiency resulted in cells with higher Si contents, and Zn-stressed cells had a significantly higher Si:NO3 than Zn replete diatoms [De La Rocha et al., 2000]. Franck et al. performed a series of incubation experiments in Fe-limited upwelling regions and observed a decrease in the Si:NO3 utilization ratio when Fe or Fe and Zn together were added, due to a greater stimulation of NO3 uptake versus Si uptake. In contrast to the Fe addition results, the Si:NO3 utilization ratio increased when Zn was added as a result of either a decline in NO3 uptake (Costa Rica and Point Conception, CA stations) or enhanced Si uptake (Big Sur, CA) [Franck et al., 2003]. In this study, a similar decrease in the Si:NO3 utilization ratio was observed in all the metal treatments relative to the control (Table 2), which is consistent with previous Fe addition experiments in the subarctic North Pacific [Takeda, 1998]. The mechanism for the alteration of the nutrient ratios was not explored; however, our results support previous observations that both Zn and Fe limitation may influence a phytoplankton cell's ability to compete for Si [De La Rocha et al., 2000; Franck et al., 2003].
 The authors would like to thank the scientists and crew of the R/V Kilo Moana, particularly Chief Scientist Brian Popp. Brian Popp and Adriana Eskinasy provided the shipboard phosphate data, and Bob Bidigare's group provided the HPLC pigment data. Thanks to Seth John for partnering in the Zn radiotracer experiment and help with sampling. Thanks to Ed Boyle for helpful discussions on the total Zn analysis. Thanks to Gary Fones and Chris Dupont for assistance with sampling and to Dave Schneider and Lary Ball of the WHOI Plasma Mass Spectrometry Facility for assistance with ICP-MS measurements. This research was supported by NSF grant OCE-0136835 and by an EPA STAR Fellowship.