4.1. Nickel Associations With Diatoms
 Diatoms collected from the equatorial Pacific Ocean had significantly higher Ni:P ratios than other co-occurring cell types. Elevated diatom Ni:P ratios were also documented with SXRF analyses in the Southern Ocean, where diatom Ni:P ratios were 4.3- and 9.9-fold higher than in co-occurring autotrophic and heterotrophic flagellates (Figure 2) [Twining et al., 2004]. To our knowledge, these are the first published reports of systematically elevated Ni in diatoms from either laboratory or field studies. Ratios of Ni:P have been reported for suspended marine particulate material (0.21–0.86 mmol/mol [Bruland et al., 1991]), and these bulk ratios fall between the ratios measured in the various cell types with SXRF.
 To investigate the possible drivers of elevated Ni content in diatoms, cell quotas were divided into internal and frustule-associated fractions by examining the response of cells to added Fe and Si in the carboy incubations. The pennate diatoms produced more intracellular biomass (e.g., ribosomes, mitochondria, cytosol) in response to Fe but did not alter silicification. The increases in intracellular biomass are shown by changes in P quotas (Figure 3), as well as by changes in cellular S. Sulfur is primarily associated with cysteine and methionine moieties within proteins or polyamines [Roberts et al., 1955] and can be used as an alternate biomass proxy [Baines et al., 2011]. Although S can also be found as dimethylsulfoniopropionate (DMSP) in diatoms under Fe stress [Sunda et al., 2002], the diatoms in these incubations had been released from Fe limitation, and thus cellular DMSP would be expected to decrease rather than increase. Sulfur quotas showed no change in the control treatment during the incubations but increased 2.4-fold in response to Fe (9.9 ± 0.8 fmol S cell−1 in + Fe compared to 4.1 ± 0.2 fmol S cell−1 in control), similar to P. Therefore the observed increases in Ni cell−1 are taken as broadly representative of the Ni:P stoichiometry of the internal organic matter of the diatoms. Similarly, diatoms deposited more opal into their frustules in response to silicate additions but did not synthesize more internal biomass As with P, S quotas were essentially unchanged in response to Si (4.6 ± 0.3 fmol cell−1 in + Si compared to 4.1 ± 0.2 fmol cell−1 in control) [Baines et al., 2011]. Therefore the observed increases in Ni cell−1 are interpreted as representative of the Ni:Si stoichiometry of the diatom frustule.
 As a check on the consistency of the Ni:P and Ni:Si ratios derived from the incubation experiment, these ratios were used to calculate Ni quotas for diatoms collected from stations across and along the equator in the Pacific Ocean. Ni quotas (mol cell−1) for these cells were estimated by multiplying Ni:Pintby the cell-specific P quota (mol cell−1) and Ni:Sifrustby the cell-specific Si quota (mol cell−1). Calculated Ni quotas were, on average, 85 ± 8% (1 SE) of the measured Ni quotas, and ‘internal’ and ‘frustule’ Ni each accounted for 50 ± 1% of the total cellular Ni. As an additional check on the derived Ni ratios, the Ni:P of the intracellular material added by the diatoms (Ni:Pint = 0.52 ± 0.10 mmol/mol) was 41% of the whole-cell Ni:P ratios of these diatoms at the start of the incubation experiment (1.27 ± 0.35 mmol/mol), similar (within error) to the approximately 50% fraction of Ni estimated to be associated with intracellular biomass of equatorial Pacific diatoms. It thus appears that there are two distinct fractions of Ni in diatoms: Ni associated with intracellular machinery and Ni associated with the frustule.
 Internal Ni comprised only half of the Ni associated with equatorial Pacific diatoms, but Ni:Pintalone was still higher than whole-cell Ni:P ratios observed in most non-diatoms from the equatorial Pacific and Southern Ocean. In fact, Ni:Pintof diatoms was comparable to mean whole-cell Ni:P measured in cyanobacteria in the equatorial Pacific, which was itself higher than in larger flagellated photoautotrophs (Figure 2). Elevated Ni content of cyanobacteria has been reported in freshwater species [Ji and Sherrell, 2008] and may result from an elevated importance of Ni-SOD to cyanobacterial physiology. Ni-SOD is the only SOD isoform present in all sequenced strains ofProchlorococcus, as well as in at least one strain of Synechococcus [Dupont et al., 2008a]. Ni-SOD is a recently evolved enzyme that appears to have been shared between phyologenetic groups via horizontal gene transfer, and the presence of Ni-SOD in cyanobacteria is always associated with the loss of the Fe isoform of SOD [Dupont et al., 2008a, 2008b; Schmidt et al., 2009]. It has thus been suggested that Ni-SOD may have evolved in response to low Fe availability in the modern ocean [Dupont et al., 2008b]. Open-ocean diatom strains contain Ni-SOD as well [Cuvelier et al., 2010], and elevated Ni:Pintmay reflect preferential usage of Ni-SOD by diatoms relative to other eukaryotic taxa. However, Mn-SOD is also present in diatoms [Wolfe-Simon et al., 2005], and diatoms appear to produce more Mn-SOD under Fe stress [Peers and Price, 2004]. Further research is needed to ascertain the importance of Ni-SOD to diatoms.
 The elevated Ni:Pint of diatoms may indicate that diatoms are supporting growth through uptake and utilization of DON via urease. Indeed, studies in the equatorial Pacific have concluded that diatoms in this region utilize organic N as growth substrate [Price et al., 1991, 1994], and work performed on the same cruise demonstrated rapid diatom growth rates and f-ratios of larger cells between 0.27 and 0.41 [Parker et al., 2011; Selph et al., 2011], indicating that the growth of larger cells such as diatoms is substantially supported by recycled N. However, nitrate utilization was stimulated in the same +Fe incubations in which elevated diatom Ni:P was observed [Brzezinski et al., 2011], and urease may be expressed constitutively [Allen et al., 2011] or even up-regulated by diatoms growing on nitrate [Peers et al., 2000]. Thus, the presence of elevated urease, if observed, may be an equivocal indicator of DON utilization.
 Alternatively, Allen et al. demonstrated that diatoms use an ornithine-urea cycle to enhance their response to episodic N availability. They found that urease levels in the pennate diatomPhaeodactylum tricornutum did not change in response to pulsed N inputs but were consistently elevated. Peers et al. also found urease to be constitutively expressed in two centric diatom species. Thus, the apparently unique role played by the ornithine-urea cycle in diatom physiology, as well as elevated constitutive expression of urease in these organisms, may provide an alternate explanation for the relatively high intracellular Ni:P ratios in diatoms. Furthermore,Dupont et al. suggested that Fe and Ni membrane transporters may compete for surface area in larger cells such as diatoms. They proposed that additions of Fe to Fe-limited phytoplankton should result in the accumulation of Ni, as observed here. It is evident that there are several possible explanations for the elevated intracellular Ni:P ratios of diatoms which remain to be tested.
 Maps of Ni in diatoms (Figure 1) [Twining et al., 2003], as well as the incorporation of additional Ni during diatom silicification, confirm that Ni is also associated with the silica frustule. However, the mechanism by which Ni is incorporated into the frustule has yet to be determined. Silica is polymerized and formed into the intricate nano-scaffold of the frustule within the silica deposition vesicle [Hildebrand, 2008; Round et al., 1990], and thus non-specific absorption of ambient dissolved Ni into the frustule during polymeriziation [e.g.,Sheng et al., 2011] seems unlikely. It is more likely that Ni is contained in biochemical moieties associated with the frustule or is involved in the biochemical processes required to precipitate the frustule and becomes inadvertently trapped within the frustule. It is possible that frustule-associated Ni is contained in ureases spanning the silicalemma of the silica deposition vesicle (SDV). A portion of cellular urease is localized to the outer membrane of some bacterial strains [Baik et al., 2004; Mclean et al., 1985; Phadnis et al., 1996], but in other bacterial strains and in plants urease is present primarily in the cytoplasm [Faye et al., 1986; Mobley and Hausinger, 1989; Mobley et al., 1995]. If present, frustule-associated Ni which is actually contained in membrane-bound ureases could be preferentially preserved by the opal test during water column remineralization. For example, it has been shown that organic molecules such as proteins and amino acids are intimately associated with silica in frustules [Abramson et al., 2009].
 Researchers have identified several groups of proteins involved in frustule formation, including frustulins and silaffins [Kröger et al., 1996, 1999], and long-chain polyamines are also known to be associated with frustule formation [Sumper and Kröger, 2004]. Silaffin-like proteins are intimately associated with silica in the frustule [Scheffel et al., 2011]. The long-chain polyamines are composed of precursors such as putrescine and ornithine that are produced by the cellular urea cycle. However, urease serves to reduce ornithine production by the urea cycle, lowering concentrations of the precursor of the long-chain polyamines involved in silica polymerization. Thus it is not clear why urease production would increase in concert with heightened silicification. Alternately, urease may be used to adjust pH in support of frustule formation. Urease is used by some bacteria to raise the pH of acid environments [Huynh and Grinstein, 2007]. While the SDV is thought to provide an acidic environment to support silica polymerization [Vrieling et al., 1999], diatoms may need to adjust pH upward at some point during silicification, perhaps to maintain silicic acid in solution at high concentrations prior to polymerization or to adjust SDV pH following frustule formation. A mechanistic molecular explanation for the role of Ni in frustule formation remains to be determined, but we postulate that urease plays some role in silica polymerization and could become associated with the completed frustule as a result.
4.2. Comparisons With Water Column Ni Distributions
 Our finding of separate associations of Ni with the internal and frustule fractions of diatoms is consistent with distributions of dissolved P, Si and Ni in the water column. Vertical profiles of these nutrients from the Atlantic, Pacific and Indian Oceans show a correspondence of Ni concentrations with both phosphate and silicic acid [Bruland, 1980; Morley et al., 1993; Saager et al., 1992; Sclater et al., 1976]. Both Sclater et al.  and Bruland used multiple linear regression to estimate ratios of Ni:P and Ni:Si in the water column, and we have extended this analysis to open-ocean stations in the Indian Ocean (Table 2). Coefficients of determination (r2) range from 0.892 to 0.996, demonstrating the close relationship between these dissolved nutrients. Data from the two North Pacific studies show relatively close agreement in both ratios, while the Indian Ocean studies differ in the relative magnitude of Ni:P and Ni:Si. Stations sampled by Saager et al.  have relatively low Ni:P (0.38–0.58 mmol/mol) and relatively high Ni:Si (41–69 μmol/mol), while stations sampled by Morley et al.  have approximately fourfold higher Ni:P (0.48–2.03 mmol/mol) but approximately fourfold lower Ni:Si (6–23 μmol/mol) than observed by Saager et al. .
 This may indicate real differences in the cycling of Ni in these areas (although they are both in tropical waters of the western Indian Ocean), but it could also result from analytical differences between the studies that favor one ratio at the expense of the other. All three nutrients are remineralized in the upper 1,000 m of the water column, and the estimated Ni:Si ratio is therefore constrained largely by the relationship between these elements in the deeper waters (as P is remineralized nearly entirely in the upper 1,000 m). The regression model adjusts the estimated Ni:P ratio to best match the upper water column nutrient data within the constraints of the Ni:Si ratio, and therefore the two ratios are negatively correlated with each other across the entire data set. The regression models show that Si is a significantly better predictor of Ni than is P (based on partial sums of squares). Therefore, the differences between the Indian Ocean studies may represent differences in the deep-water remineralization of Ni and Si, with Ni:P estimates changing in response. Statistical comparisons show that differences between studies are more significant than differences between stations, also suggesting that methodological differences may be at play here. In spite of (or perhaps because of) the variations discussed above, mean (±SD) Ni:P and Ni:Si ratios for North Pacific and Indian Ocean stations are not significantly different from each other.
 Comparisons of water column Ni ratios with those in diatoms and other plankton suggest roles for each in Ni export. Based on the close agreement of ocean Ni:Si ratios with that calculated for the frustule fraction of diatoms in the carboy experiments, diatoms appear to be primarily responsible for Ni export to deep waters. However, the situation in the upper water column is less clear. Average water column Ni:P ratios are similar to those in whole diatoms but are twofold above that estimated for internal diatom biomass (0.52 ± 0.10 mmol/mol), which is the fraction assumed to affect Ni:P ratios in the upper 1,000 m where nearly all P remineralization takes place (Table 3). The offset in water column Ni:P may be explained by Ni uptake and remineralization by other plankton with yet higher Ni:P ratios. Synechococcus have elevated Ni:P compared to autotrophic flagellates (Table 1), and work in the North Atlantic has shown Synechococcus Ni:P to increase by nearly an order of magnitude (up to 3.40 mmol/mol) in response to nutrient delivery in mesoscale eddies [Twining et al., 2010]. Cyanobacteria are far more abundant than diatoms in open-ocean settings and may contribute to organic matter export as well [Amacher et al., 2009; Lomas and Moran, 2011; Richardson and Jackson, 2007]. Heterotrophic bacteria contain Ni-SOD, and some groups have additional Ni proteins involved in methanogenesis [Ragsdale, 2009]. As a result, heterotrophic bacteria are likely to have elevated Ni quotas and may contribute to Ni cycling in the upper water column as a food source for heterotrophic protists and through associations with sinking organic aggregates.
Table 3. Comparison of Remineralization of Ni From Sinking Phytoplankton in the Pacific Ocean and Indian Ocean as Measured in the Studies Presented in Table 2 or Calculated From Remineralized Si and P in Each Study Using Ni:Sifrust and Ni:Pintfor Frustule- and Cytosol-Associated Ni Measured in This Studya
|North Pacific Ocean|
|Bruland Sta. H-77||985||3.3||126||7.2||1.7||3.5||5.2||72%||67%|
|Sclater et al.  Sta. 204||1,039||3.2||128||6.4||1.6||3.5||5.2||81%||68%|
|Sclater et al.  Sta. 343||791||2.9||86||5.2||1.5||2.4||3.9||75%||61%|
|Saager et al.  Sta. 6||1,200||2.0||74||5.7||1.0||2.0||3.1||54%||67%|
|Saager et al.  Sta. 7||1,600||2.4||124||6.7||1.3||3.4||4.7||69%||73%|
|Morley et al.  Sta. 3||1,000||2.1||66||4.2||1.1||1.8||2.9||69%||63%|
|Morley et al.  Sta. 7||965||2.7||79||4.2||1.4||2.2||3.6||85%||61%|
 As an alternate approach to examine the role of diatoms in Ni export, we attempted to reproduce the Ni remineralization patterns measured in the North Pacific and Indian Ocean using our estimates of Ni:Pint and Ni:Sifrust (Table 3) Concentrations of remineralized phosphate and silicic acid were calculated for the upper water column (above the phosphate maximum) and the lower water column (between the phosphate maximum and silicic acid maximum). These were multiplied by the Ni:Pint and Ni:Sifrust ratios derived from the carboy experiment, and calculated remineralized Ni was compared to measured levels for these depth regions. Calculated Ni remineralized in the upper water column was consistently ca. 30% below measured Ni remineralization values. This suggests that Ni:Pint may be underestimated by the carboy experiment results (matching the finding of the Ni quota reconstructions discussed previously), and/or that other groups with elevated Ni:P also make significant contributions to Ni flux in the upper water column, as discussed above. In the lower water column, comparisons between calculated and measured values were more variable. In the North Pacific calculated deep water values were about 25% below measured values, but in the Indian measured Ni remineralization ranged from 31% to 400% of measured values. These large discrepancies could be an artifact of relatively small changes in dissolved Ni in the deeper waters (making the calculations more vulnerable to analytical error). They could also be driven by water mass differences, as deep waters are transported laterally via thermohaline circulation and have only partial biogeochemical connection with the overlying waters. It is also interesting to note that although diatoms clearly export Ni to deep waters, frustules are an important source of Ni in the upper 1,000 m as well. Comparison of Ni remineralization from diatom cytosol to Ni remineralization from frustules shows that frustules consistently account for about 2/3 of upper water column remineralization. This follows from the observation that typically 50–75% of all Si remineralization occurs in the upper water column.
 The measurements of Ni, P and Si in diatoms and other plankton from the equatorial Pacific presented here point to an important role for diatoms as mediators of sinking Ni flux in the ocean. This role results both from elevated internal Ni levels in diatoms relative to other eukaryotes, as well as from a unique association between Ni and the opal frustule of diatoms. Diatom Ni flux may be augmented by the sinking of other high-Ni groups such as cyanobacteria and heterotrophic bacteria. While this work clearly implicates diatoms in the ocean Ni cycle, we do not yet have a molecular understanding for the causes of the elevated Ni content of diatoms. High internal Ni:P in diatoms may result from enhanced usage of urease and Ni-SOD, like cyanobacteria, or a Ni protein yet to be identified. The association of Ni with the diatom frustule remains unexplained, as well. Application of genomic and proteomic tools to address these questions should further advance our understanding of the roles of individual plankton groups in ocean trace metal cycles.