Nitrogen (N2) fixation is an important process that fuels export production in the North Pacific Ocean, as evidenced by seasonally low δ15N of sinking organic nitrogen (N) at the Hawaii Ocean Time series station. However, relatively few direct measurements of N2 fixation exist across the North Pacific. On two cruises there in fall 2002 and summer 2003, the abundance and N2 fixation rate of Trichodesmium spp. and Richelia, as well as bulk water samples, were measured. Trichodesmium spp. were only detected in the area near the Hawaiian Islands, in similar densities on both cruises. Despite similar densities, the areal N2 fixation rate of Trichodesmium spp. in fall 2002 was nearly four times greater than in summer 2003 at stations proximal to the Hawaiian Islands. In the central North Pacific Gyre far from the Hawaiian Islands, where Trichodesmium spp. was not present, whole water N2 fixation rates were relatively high (∼100 μmol N m−2 d−1). Presumably unicellular diazotrophs were responsible for activity there. Our studies show a geographical variation in the dominant diazotroph in the North Pacific Subtropical Gyre in the summer with Trichodesmium being dominant around the Hawaiian Islands, Richelia associated with diatoms to be found in high numbers to the south of the islands while unicellular diazotrophs dominated to the west, away from the islands and evidence from the literature suggests iron may play a role.
If you can't find a tool you're looking for, please click the link at the top of the page to "Go to old article view". Alternatively, view our Knowledge Base articles for additional help. Your feedback is important to us, so please let us know if you have comments or ideas for improvement.
 The export of organic matter from the surface to the deep ocean in oligotrophic waters is dependent on the supply of new nutrients, especially nitrate, to the euphotic zone [Eppley and Peterson, 1979]. Vertical eddy diffusion of nitrate was previously thought to be the most important new nitrogen (N) source [Eppley and Peterson, 1979]. However, recent studies have shown dinitrogen (N2) fixation, the process by which a limited number of organisms can access the large but biologically unavailable pool of N2 gas by reducing it with the nitrogenase enzyme [Postgate, 1998], can be of equal importance as a new nitrogen source in oligotrophic waters of the North Atlantic [Capone et al., 2005] and Karl et al.  found that N2 fixation can fuel up to half the new production at the Hawaii Ocean Time series (HOT) site. Considering that carbon dioxide (CO2) is, on average, brought up in Redfield proportion with nitrate from below the thermocline, N2 fixation, in addition to atmospheric deposition and N introduced by rivers, are the only sources of truly new N that can promote a net sequestration of CO2 from the atmosphere to deeper waters [Eppley and Peterson, 1979]. This effect may be indirect, through the introduction of new N then available to the food web, or in some cases direct, as has been observed for diatom-diazotrophic associations (DDAs) [Subramaniam et al., 2008]. Thus, it is important to quantify this source of N to oligotrophic oceans and understand what factors contribute to differences seen on spatial and temporal scales.
 The North Pacific Subtropical Gyre (NPSG) is a highly oligotrophic, stratified system that has been well studied by the ongoing Hawaii Ocean Times-series program. Decadal scale patterns that have been discerned include the shift in the dominant primary producers from eukaryotes to smaller, photosynthetic prokaryotes [Karl et al., 2001], a decrease in the inventory of soluble reactive phosphorus in the euphotic zone [Karl and Tien, 1997] and a concurrent increase in N2 fixing cyanobacteria [Karl et al., 1997]. Geochemical estimates show that about half of particulate N export at station ALOHA is supported by N coming from N2 fixation [Dore et al., 2002] and recent work by Grabowski et al.  report rates ranging from 20 to 307 μmol m−2d−1.
 One important diazotroph found in tropical and subtropical waters is Trichodesmium spp. Trichodesmium spp. is a colony forming diazotrophic cyanobacterium that is cosmopolitan in warm, oligotrophic waters [Capone et al., 1997]. Global estimates show that N2 fixation by this genus could account for up to 80 Tg N yr−1, about half of current geochemical estimates of total N2 fixation (ranging from 100 to 200 Tg N yr−1) [Carpenter and Capone, 2008]. Clearly, quantification of N2 fixation by Trichodesmium spp. is important in deriving global and local estimates of N2 fixation. The occurrence of Trichodesmium spp. in the NPSG has been reported in the past [Letelier and Karl, 1996; Mague et al., 1977]. However, it was found predominantly as free trichomes in the NPSG [Letelier and Karl, 1996], rather than as the colonies that are most prevalent in some other locations such as the Sargasso Sea [Carpenter et al., 2004].
 While Trichodesmium spp. is an important contributor to marine N2 fixation, it still cannot account for all geochemically estimated N2 fixation. Another important contributor is Richelia, a symbiont associated with diatoms such as Hemiaulus spp. and Rhizosolenia spp. These diatom diazotroph associations (DDAs) have been commonly noted in the NPSG [Church et al., 2008; Scharek et al., 1999; Venrick, 1974; Villareal and Carpenter, 1989]. N2 fixation by unicellular cyanobacteria in the 3–10 μm phytoplankton size fraction has also been identified in these waters [Zehr et al., 2001] and areal rates of N2 fixation in this fraction can equal or exceed those measured for Trichodesmium spp. [Montoya et al., 2004].
 We undertook two cruises as part of the Marine Nitrogen fixation and Tropospheric Responses to Aeolian Inputs (MANTRA) Biocomplexity project to quantify the dominant diazotroph and N2 fixation rates and study their geographical distribution in the NPSG. We measured Trichodesmium spp. and Richelia abundance, N2 fixation by these organisms as well as bulk N2 fixation measurements. At station lacking larger diazotrophs, these measurements could be attributed to unicellular diazotrophs.
2.1. Sample Collection
 Samples were collected on two cruises in the NPSG. Measurements were carried out in September and October of 2002 aboard the R/V Kilo Moana (MP6) and in July and August 2003 aboard the R/V Roger Revelle (MP9). Samples were largely collected around the Hawaiian Islands, but on leg 1 of MP9, a transect to 175 °E allowed collection of samples and rate measurements across a wide area of the NPSG.
2.2. Trichodesmium spp. and Richelia Counts
Trichodesmium spp. counts were done using the method previously described by Carpenter et al. . Briefly, the entirety of a 10 L Niskin bottle with water collected from depths corresponding to 100%, 55%, 28%, 10% and 1% of surface light (light depths) was drained onto a 47 mm, 10 μm polycarbonate filter placed in a swinex filter holder and attached to the spout of the Niskin bottle. Epifluorescent microscopy was used to enumerate Trichodesmium spp. colonies, free filaments and the number of Richelia heterocysts on the filters, either on board, or preserved in 0.4% paraformaldehyde and frozen until counted in the lab. The number of Trichodesmium spp. trichomes per colony was determined at each station by placing 10 representative colonies into a vial with filtered seawater and shaking to disaggregate the colonies. This sample was then filtered onto a 10 μm Nuclepore filter and the trichomes counted under a microscope. The trichome abundance was then calculated by multiplying the colony abundance by the number of trichomes per colony. Samples for trichome counts were taken all the way to 175°E and back during MP9, but no Trichodesmium spp. was detected by this method west of the station at 161°W. For the purposes of comparisons between the two cruises, we excluded the observations west of 167 °W from our averages (Data Set S1).
2.3. N2 Fixation Measurements
Trichodesmium spp. samples were collected by towing a 1 m, 202 μm mesh net (0.5–1 knot) at a depth of 15–20 m or by towing a small 64 μm mesh net at the surface. Previous work had demonstrated that there are no systematic differences in activity for samples collected by hand in situ and by towing as described [Carpenter et al., 1987]. Colonies were isolated with a plastic transfer loop and placed in a filtered seawater rinse before picking them into incubation bottles. To collect free trichomes, the diluted plankton tow was allowed to briefly sit while the trichomes rose to the surface and clumped together. The aggregation could then be gently removed with a plastic loop or wide bore plastic pipette and placed into filtered seawater to make a slurry, diluted to a specific volume, and used for experiments. Samples were checked under a dissecting microscope for presence of other organisms before use and a subsample saved for trichome density determination.
Trichodesmium spp. N2 fixation measurements were carried out using the acetylene (C2H2) reduction method described by Capone . Ten colonies in 10 ml GF/F filtered seawater or 10 ml of a free trichome slurry was placed into acid cleaned 14 ml glass serum vials. Twenty μM ethylenediaminetetraacetate (EDTA) was added as it extends the life of the Trichodesmium spp. during incubation, allowing for the collection of more data points, but does not affect the initial acetylene reduction rate [Burns et al., 2006]. Vials were sealed with silicone rubber stoppers and 1 ml of instrument grade C2H2, sparged through deionized water, was added. The production of ethylene (C2H4) in the headspace of the vial was monitored over time courses of 6–10 h by flame ionization detection gas chromatography. Triplicate vials were incubated at 100%, 55%, 28%, 10% and 1% of surface irradiance in on deck incubators with flowing surface seawater to control temperature. The C2H2 reduction rate was derived from the linear regression of incubation time versus C2H4 concentration in the vials. The C2H2 reduction rate was then converted to N2 fixation rate using a conversion factor of 3 C2H4:1 N2 [Capone et al., 2005] then multiplied by 2 (as there are 2 ammonium molecules released from the splitting of N2 gas).
 To calculate areal rates of N2 fixation by Trichodesmium spp., per colony or per trichome rates at each light level were multiplied by colony or trichome abundance at the same light depth and a trapezoidal integration was used to calculate N2 fixation in a square meter of the water column. At some stations, low abundance of colonies in the net tows did not allow measurement of N2 fixation at all light levels. In such cases, N2 fixation was measured for as many light levels as possible (always concentrating on the higher light levels) and a cruise average (C2H2 reduction rate per colony; see Table 1) was used as the rate for the other light levels. Similarly, due to low biomass, we were not able to measure N2 fixation rates of free trichomes at every station. On MP6, the cruise average for each light level for the 9 stations sampled was multiplied by the abundance of free trichomes. However, during MP9 we did not collect enough measurements of N2 fixation by free trichomes to confidently apply this method, and therefore, Trichodesmium spp. N2 fixation rates are based on colonies only. Estimates of free trichome N2 fixation were made by multiplying free trichome densities by the per trichome N2 fixation rate of colonies.
Table 1. Colony and Free Trichome Specific Acetylene Reduction Ratesa
Percent of Surface Irradiance
Average Depth (m)
C2H2 Reduction (nmol col−1 h−1 or pmol trichome−1 hr−1)
Number of Stations (n)
Standard errors are shown in parentheses.
MP6 Free Trichomes
 N2 fixation measurements were also carried out using tracer 15N2 uptake on large water samples [Montoya et al., 2004; Montoya et al., 1996]. Duplicate samples were collected from depths corresponding to 100%, 55%, 28%, 10% and 1% of surface light using a CTD rosette with 10 L Niskin bottles attached. Water from the Niskin bottles was gently drained with a sampling tube into 4.5 L polycarbonate bottles. No prefilter was used. Air bubbles were removed and bottles were sealed with septa lined caps, followed by the addition of 3 ml of 99% 15N2. Incubations were conducted in deck board incubators covered with blue screening to simulate the light depth of collection and flowing surface seawater. The first four light depths, where the large majority of the N2 fixation occurred, fell above the thermocline and the in situ temperature at the sampling depth was within about 1°C of the surface water used to maintain the on deck incubators. Thus the difference between incubation and in situ temperatures were not likely to appreciably affect our calculations of areal rates of N2 fixation.
 After 24 h incubation, the whole content of the bottle was filtered onto precombusted GF/F filters. Filters were dried and pelletized, then the δ15N of the sample was analyzed on a VG IsoPrime mass spectrometer interfaced to a Eurovector elemental analyzer at the USC stable isotope facility. Standards were routinely analyzed during sample runs and included acetanilide for C and N elemental mass and ammonium sulfate and glycine for δ15N. Uptake of 15N2 in the samples was calculated as described by Montoya et al. . Areal rates were calculated using trapezoidal depth integration. Multivariate cross correlation matrices were calculated using JMP software.
3.1. Trichodesmium spp. Abundance
Trichodesmium spp. was present both as aggregates (colonial form) and as free trichomes on both MP6 and MP9 cruises (Figure 1). Trichodesmium spp. abundances in colony form were very similar and not statistically significantly different (p > 0.55) between the two cruises for the stations sampled (5.6 ± 0.8 trichomes × 106 m−2 during MP6 and 4.8 ± 1.0 trichomes × 106 m−2 during MP9) while free trichomes averaged 1.6 ± 0.3 trichomes × 106 m−2 during MP6, about twice the abundance of free trichomes during MP9. However, the average abundances of free trichomes from the two cruises were not significantly different from each other (p > 0.06). Across all stations, the percentage of total trichomes that were found in colonies averaged 70% on MP6 and 74% on MP9. The data ranged widely around these means, and the median values were 78% and 83% (Data Set S1). The abundance of Trichodesmium spp. increased in our samples toward the Hawaiian Islands on the first leg of MP9, from the station at 167°W toward station ALOHA, just north of Oahu. Trichodesmium spp. was not detected west of 167°W.
Katagnymene (recently renamed Trichodesmium spiralis) occurred in relatively high densities on MP6 with an average of 0.47 ± 0.07 trichomes × 106 m−2, but at much lower levels (0.09 ± 0.02 trichomes × 106 m−2) on MP9. In contrast, Richelia spp. associated with the diatoms Hemiaulus and Rhizosolenia occurred in average densities of 1.3 ± 0.4 × 106 heterocysts m−2 on MP6 while at much higher levels on MP9 (average 5.2 ± 2.1 cells × 106 m−2). Depth integrated abundances were in excess of 107 heterocysts m−2 at 4 stations on this cruise.
3.2. Trichodesmium spp. N2 Fixation
Trichodesmium spp. colonies and free trichomes exhibited high levels of nitrogenase activity, measured as C2H2 reduction, on both cruises (Figure 1 and 2 and Data Set S1). During MP6, N2 fixation by colonies was greater than by free trichomes. However, free trichomes fixed proportionally more N than colonies did when compared to their abundance (free trichomes were 23% of all Trichodesmium spp., but accounted for 31% of total N2 fixation). This was probably due to the fact that the free trichomes on this cruise included a considerable number of Katagnymene with much larger trichomes than those contained in other Trichodesmium spp. colonies (data not shown). During MP9, only N2 fixation by colonies was measured. While abundances of Trichodesmium spp. were nearly the same between MP6 and MP9, depth integrated N2 fixation rates (colony plus free trichome for MP6 and colony only for MP9) were about five times greater, on average, in the fall (MP6) compared to the summer (MP9; 227 ± 26 μmol N m−2 d−1 versus 43 ± 12 μmol N m−2 d−1). If we add in an estimate for free trichome N2 fixation on MP9 (based on per trichome N2 fixation rates of colonies, multiplied by free trichome abundance) areal N2 fixation rates only increase slightly, to 50 μmol N m−2 d−1. The difference in N2 fixation rates between the two cruises can mostly be explained by the difference in colony normalized C2H2 reduction rates. The average per colony rate on MP6 was greater at all light depths than on MP9 (Table 1 and Figure 3). In addition, C2H2 reduction during MP9 was barely measurable at 10% and 1% of surface irradiance. Finally, water column integrated N2 fixation rates by colonies of Trichodesmium spp. increased along a transect from unmeasurable (for lack of colonies) west of 167° W to 47 μmol N m−2 d−1 at station ALOHA during leg one of MP9 (Figure 2).
3.3. Bulk N2 Fixation
 N2 fixation was also measured using the uptake of 15N2 into particulate matter in relatively large volumes (>4L) of water on the two cruises. In general, highest rates were at or near the surface and decreased at lower depths in the water column (Data Set S2). On MP6, two stations showed exceptionally high rates of activity. However, bulk N2 fixation rates were comparable on the two cruises when excluding the two extremely high values.
 The integrated areal average from MP9, 183 μmol N m−2 d−1, exceeded the average from MP6, 143 μmol N m−2 d−1 (Data Set S1). This observation runs counter to the Trichodesmium spp. N2 fixation rates measured with C2H2 reduction. It may suggest that, on the whole, organisms other than Trichodesmium spp. (perhaps unicellular diazotrophs) were responsible for a large part of the N2 fixation during MP9 and that Trichodesmium spp. may have been responsible for much of the total N2 fixation on MP6. Still, Trichodesmium spp. C2H2 reduction rates greatly exceeded bulk 15N2 uptake on MP6. This could arise from a sampling bias against the large and patchy Trichodesmium spp. in the 15N2 uptake assays. Alternatively, recent evidence suggests that 15N2 uptake assays, which introduce the substrate as a bubble (as used here) rather than as a dissolved gas, may underestimate actual rates [Mohr et al., 2010]. At the stations on leg 1 of MP9 where Trichodesmium spp. and Richelia were not detected in our counts, we attribute the N2 fixation measurements to unicellular diazotrophs. At these remote central gyre stations, N2 fixation averaged 137 μmol N m−2 d−1.
Trichodesmium spp. is also known to occur in other areas of the North Pacific, the South China Sea (SCS) and in the Kuroshio [Carpenter, 1983]. Reports of Trichodesmium spp. abundance show an increase in density from 0.44 × 106 trichomes m−2 in the spring up to about 5.3 × 106 trichomes m−2 in the SCS in the summer and fall [Chen et al., 2003], and abundance in the Kuroshio is about ten times greater than in the SCS. These abundances are similar to the abundances reported in our study, suggesting that seasonal N2 fixation rates in the western Pacific Ocean may be as high as near the Hawaiian Islands. In contrast, recent studies in the western part of the NPSG, along 149°E and 155°E have reported very low abundances of Trichodesmium [Kitajima et al., 2009; Shiozaki et al., 2009].
 In some locations around the globe where Trichodesmium spp. has been enumerated, trichomes are often found predominantly in aggregates or colonies [Carpenter et al., 2004, and references therein]. During our two cruises in the North Pacific, free trichomes were more important than reported elsewhere, accounting for 20–30% of the total, suggesting that some factor in this region of the North Pacific either promotes free trichomes or prevents colony formation. A three year study of Trichodesmium spp. abundance at station ALOHA reported that free trichomes were about 80% of the total [Letelier and Karl, 1996] and a recent study in the eastern North Atlantic also noted a general preponderance of free trichomes [González Taboada et al., 2010].
4.2. Comparison to Direct Measurements
 The density of direct measurements of Trichodesmium spp. N2 fixation rates in the North Pacific are relatively low (Table 2), at least when compared to the North Atlantic Ocean. Capone et al.  calculated water column N2 fixation from two studies in the NPSG to be 33 μmol N m−2 d−1 (subtropical [Mague et al., 1977]) and 134 μmol N m−2 d−1 (tropical [Gunderson et al., 1976]). Karl et al.  used biomass abundance and an assumed cell specific N2 fixation rate to estimate N2 fixation from the ALOHA site to be about 140 μmol N m−2 d−1 and from the C2H2 reduction method to be 85 μmol N m−2 d−1. These are the only Trichodesmium spp. specific, water column integrated N2 fixation rates in the NPSG that we are aware of, and they are remarkably similar to each other and the rates presented in this study. Areal N2 fixation by Trichodesmium in the South China Sea was estimated at 126 μmol N m−2 d−1 [Saino, 1977] as calculated by Capone et al. .
Table 2. Estimates of N2 Fixation in the North Pacific by Both Direct and Geochemical Means
 N2 fixation rates have also been measured over a wide area of the North Pacific using 15N2 uptake on large water samples. On a transect from Hawaii to Southern California, N2 fixation by unicellular diazotrophic cyanobacteria in the <10 μm fraction averaged 520 μmol N m−2 d−1, while multiple samplings at station ALOHA ranged from 11 to 103 μmol N m−2 d−1 [Montoya et al., 2004]. Dore et al.  measured N2 fixation on whole water samples at station ALOHA over a one year period and found that it ranged from about 25–125 μmol N m−2 d−1; the highest rate occurred in July and was linked to the >10 μm fraction, which includes Trichodesmium spp. For the stations east of 161°W, our bulk rates of 143 μmol N m−2 d−1 during MP6 and 216 μmol N m−2 d−1 during MP9 fall within the range of these other studies. These studies suggest that Trichodesmium spp. is an important contributor to water column N2 fixation in the area near Hawaii, but that N2 fixation by unicellular diazotrophs can at times far exceed that by Trichodesmium spp. in the region of the Islands. N2 fixation is predominantly associated with the <10 μm fraction in the region west of Hawaii [Montoya et al., 2004]. Bonnet et al.  recently reported on a transect along the equator from 140°W to 145°E. Through most of the transect, rates were low and dominated by the <10μm size fraction. Trichodesmium spp. became important at the extreme western end of the transect, near Papua New Guinea.
 Several recent studies have reported rates of nitrogen fixation on bulk water and in the small size fraction in meridional transects in the western Pacific southeast of Japan. Using a high sensitivity acetylene reduction assay, Kitajima et al.  reported rates of 1 to 9 nmol N L−1d−1 in the summer along 149°E and from 5 to 20 nmol N L−1d−1 during the winter cruise along 155°E. The <10μm fraction accounted for most of the activity.
 In total, these studies suggest that N2 fixation by unicellular diazotrophs dominate the open ocean areas of the North Pacific, while organisms in the large size fraction (Trichodesmium and Richelia) are important in coastal areas and near islands, although from these studies it is impossible to say whether or not there is seasonality to this overall pattern. Across the North Pacific, the N2 fixation rates measured in small diazotrophs are consistently similar to the average reported here (137 μmol N m−2 d−1). Potential factors contributing to the dominance of unicellular diazotrophs in much of the Pacific will be discussed below.
4.3. Comparison to Geochemical Estimates
 A number of geochemically based estimates of N2 fixation have been reported. At station ALOHA, an N:P mass balance yielded a N2 fixation estimate of 93 μmol N m−2 d−1 [Karl et al., 1997]. N2 fixation was also estimated there by measuring the δ15N of sinking particulate N (PN) at 150m and using a 2 end-member mixing model with N2 fixation (−1‰) and nitrate diffusing from below the thermocline (6.5‰) as the two sources of N to exported PN. This method yielded estimates ranging from 100 to 400 μmol N m−2 d−1 over the study period from May 2000 to July 2001 [Dore et al., 2002]. The same technique applied to yearly data at station ALOHA gave estimates from 85 to 230 μmol N m−2 d−1 over the period from 1990 to 2000, with no consistent pattern of increase or decrease over time [Dore et al., 2002]. Deutsch et al.  used a mass balance approach in the Pacific Ocean to create a N budget for the basin. The basin scale N2 fixation rate was estimated from the difference between the source and sink values and calculated to be 59 ± 14 Tg N y−1 for the basin. Averaging this rate over the area from 30°N to 30°S gives an areal rate of 107 μmol N m−2 d−1. A mass balance of total organic P and N in the NPSG gave an estimate of 219 μmol N m−2 d−1 [Abell et al., 2000]. These estimates are all of the same order of magnitude as our instantaneous calculated rates from Trichodesmium spp. However, it is important to note that Trichodesmium spp. is not ubiquitous and subject to seasonal changes, while geochemical estimates are integrated over longer time scales.
4.4. Comparison to Vertical Nitrate Flux
 New nitrogen as nitrate is also delivered to the euphotic zone through vertical eddy diffusive flux from the nitricline. Estimates of eddy diffusive NO3− flux in the North Pacific are highly variable and range from 52 to 1760 μmol N m−2 d−1 (Table 3). Vertical NO3− flux for each of the cruises in this study was calculated from the average NO3− gradient and the range of the diapycnal eddy diffusivity coefficient (Kz) reported by Christian et al. . The estimates of diffusive NO3− flux were very similar between MP6 and MP9, ranging from ∼50 to 400 μmol N m−2 d−1. “Event driven” vertical fluxes of NO3− may also contribute another 240 μmol N m−2 d−1 or more when averaged over the year [Johnson et al., 2010]. While the eddy diffusivity coefficient is not well constrained and the frequency of short-term NO3− injection events may vary from year to year, the N2 fixation rates generated in this study are within the range of each of these new N sources, suggesting that, in the stratified period when it occurs, N2 fixation can be as important a source of new N in the NPSG as the diffusive flux of NO3− or “event driven” vertical fluxes. In the winter months, the mixed layer descends to the top of the nitracline (Hawaii Ocean Time series Data Organization and Graphical site, http://hahana.soest.hawaii.edu/hot/hot-dogs/) and deep convective mixing would be the most important N source.
Table 3. Estimates of Vertical Eddy Diffusivity and NO3 Flux in the NPSGa
NO3− Gradient (mmol m−4)
Kz × 105 (m2 s−1)
N Flux (μmol N m−2 d−1)
DGM, depth gradient model; PND, photosynthetic N demand; na, not applicable.
Comparing export and nutrient profiles, assuming steady state.
Measured microstructure temp and shear with ADCP and freefalling vertical profiler.
 In their transect along 155°E, Shiozaki et al.  reported that nitrogen fixation could account for up to 37% of new production (sum of nitrogen fixation and nitrate assimilation) at a station in the subtropical gyre at 24°N.
4.5. Potential Factors Affecting Diazotroph Abundance and N2 Fixation
 Geographic variability, seasonality, and interannual variability could all have an effect on N2 fixation in the NPSG. The surface chlorophyll values (Figure 4c) derived from monthly averages over an 8° square area north of the islands (24–22°N, 154–158°W) are always higher than the average monthly values south of the islands (20–18°N, 156–160°W). The satellite derived chlorophyll concentrations from the north and south boxes also demonstrate the strong seasonal and interannual variability in this region (Figure 4). The MP6 cruise coincided with the annual maximum in chlorophyll concentrations while the MP9 cruise coincided with the summer minimum and July 2003 appears to have one of the lowest chlorophyll concentrations recorded in that region from 2002 to present. Additionally, the δ15N of sinking particulate N at station ALOHA ranged from 1 to 2.5‰ in 2002 and 2.5–4.5‰ in 2003, suggesting the N2 fixation was a much greater contributor to export in 2002 than 2003 (Hawaii Ocean Time series Data Organization and Graphical site: http://hahana.soest.hawaii.edu/hot/hot-dogs/).
 In order to understand what might drive variability in diazotroph abundance and N2 fixation, biological measurements for the two cruises (all data combined) were compared to environmental factors using multiple regression analysis. Trichodesmium spp. abundance was weakly positively correlated to mixed layer depth and the concentration of dissolved inorganic phosphorus (DIP) while N2 fixation by Trichodesmium spp. was weakly negatively correlated to temperature and DIP (see Figure 5 for all comparisons and values). Bulk N2 fixation was also weakly positively correlated to mixed layer depth. Katagnymene abundance was negatively correlated to temperature, while Richelia abundance was positively correlated to temperature, the only biological factors to be significantly correlated to environmental factors (p < 0.5). These results show that it is difficult to determine the factors controlling N2 fixation in the ocean in this manner, and this type of analysis likely suffers from two issues: (1) diazotroph abundance and N2 fixation should be influenced by the conditions preceding the time of measurement, in addition to the conditions at the time of measurement, and (2) this data set may be too small on its own to tease out this information, given the scale of variability commonly encountered in the natural environment.
 Previous work on the factors controlling N2 fixation in the NPSG suggest that PO43− is not a strong limiting factor [Sohm et al., 2008; Zehr et al., 2007]. However, bioassay experiments by Grabowski et al.  with samples from Station Aloha yielded variable results with additions of PO43− and iron (Fe). Similarly variable results were obtained by J. A. Sohm and D. G. Capone (unpublished data) for N2 fixation by Trichodesmium colonies amended with additions of PO4 or Fe in this region. Church et al.  have shown that mesoscale physical variability, in the form of sea surface height anomaly, can have a large impact on N2 fixation rates during the lower nutrient summer months.
 Separate from the overall patterns of diazotrophy, the apparent shift in N2 fixation by unicellular diazotrophs in the central gyre, to an increase in the contribution by Trichodesmium spp. near the Hawaiian Islands on leg 1 of MP9 is an intriguing pattern suggesting the presence of an important growth factor for Trichodesmium spp. near the islands. Although dissolved Fe (dFe) concentrations were not measured on MP9, dFe was measured in a wide area of the North Pacific on the 2002 Intergovernmental Oceanographic Commission cruise, including a large part of the NPSG. Surface water dFe was about 0.1 nM along a transect at 24°N from roughly 170°E to 165°W, then increased to ∼0.6 nM in the area north of the Hawaiian Island chain [Brown et al., 2005], a remarkably similar pattern to the presence of Trichodesmium spp. While the connection between Trichodesmium spp. and dFe is merely correlative, it may explain why Trichodesmium spp. was only found near the Hawaiian Islands. Despite the lack of Trichodesmium spp. in the central gyre, N2 fixation was still found at appreciable rates there and is thus attributable to unicellular diazotrophs. We suggest, then, that Fe inputs may control the distribution and importance of Trichodesmium spp. and unicellular diazotrophs in the NPSG.
 Other factors could also be important in defining the distribution of diazotrophs. Temperature, salinity and mixed layer depth were not significantly different on leg 1 of MP9 at stations with Trichodesmium spp. compared to those without, while DIP was significantly lower at stations with Trichodesmium spp. present (p = 0.013). This seems counterintuitive, however, rather than being a cause of this difference, this may be an affect of the presence of Trichodesmium spp., as blooms can lead to the drawdown of DIP in the Pacific [Hashihama et al., 2009].
 If the modes of transfer of newly fixed N into the food chain or export into the deep ocean are different between Trichodesmium spp. and unicellular diazotrophs, this distribution could have important biogeochemical consequences. The primary mode of N transfer from Trichodesmium spp. is thought to be through the release of dissolved forms [Capone et al., 1994; Glibert and Banahan, 1988; Mulholland et al., 2004], and Trichodesmium spp. has yet to be identified in sediment traps and has few grazers [O'Neil et al., 1996]. Presently, no information is published on the ability of unicellular diazotrophs to release dissolved N, or the ability of zooplankton to consume them.
 N2 fixation in the subtropical North Pacific Ocean is carried out by Trichodesmium spp., Richelia spp., and unicellular diazotrophs at rates comparable, but somewhat lower, those seen in the North Atlantic. Proximate to the Hawaiian Islands, Trichodesmium spp. was abundant and accounted for much of the N2 fixation, however, over a large spatial area in the NPSG, Trichodesmium spp. and Richelia spp. was not present, or found at very low abundances, during our studies and N2 fixation was thus carried out largely by unicellular diazotrophs, at a relatively high rate of ∼100 μmol N m−2 d−1. We hypothesize that dFe is a controlling factor in the distribution of these different types of diazotrophs. Our studies show that the results reported by various investigators from work around the HOT site can be extended over a much larger geographical region in the summer. We found large diazotrophs such as Trichodesmium and Richelia to be the dominant diazotrophs in the vicinity of the Hawaiian Islands while unicellular diazotrophs seem to dominate to the regions west of the islands.
 We thank the captains, crew, and technical staff of R/V Kilo Moana (MP6) and Roger Revelle (MP9). Our field operations were greatly facilitated by David Karl and the HOT team at University of Hawaii. We also thank Lia Protopopadakis and Michael Neumann for technical support and Claire Mahaffey for assistance with some of the isotope work. This research was funded by grants OCE99-81545 and OCE99-81371 from the Biological Oceanography Program under the BioComplexity in the Environment Program. A.S. and D.G.C. were supported by the NASA Ocean Biology and Biogeochemistry Program. This is LDEO contribution 7423.