Nitrogen Fixation by Trichodesmium and unicellular diazotrophs in the northern South China Sea and the Kuroshio in summer

Distribution of diazotrophs and their nitrogen fixation activity were investigated in the northern South China Sea (nSCS) and the Kuroshio from July 16th to September 1st, 2009. N2 fixation activities in whole seawater and <10 μm fraction at the surface were measured by acetylene reduction assay. Higher activities were observed at the East China Sea (ECS) Kuroshio and the nSCS shelf. The nSCS basin showed a low N2 fixation activity. The <10 μm fractions (unicellular diazotrophs) contributed major portion to the whole-water activity in the survey time, indicating that nanoplanktonic cyanobacterias were the major diazotrophs in the survey area. Daily N2 fixation rates of Trichodesmium ranged from 0.11 to 9.83 pmolNtrichome−1 d−1 with an average of 4.03 pmolNtrichome−1 d−1. The Luzon Strait and the ECS Kuroshio had higher N2 fixation rates of Trichodesmium than the nSCS shelf and basin. Calculated activities of Trichodesmium at most stations were moderately low compared with that of the whole-water. The contribution of N2 fixation by the whole-water to primary production ranged from 1.7% to 18.5%. The estimated amount of new nitrogen introduced by Trichodesmium contributed up to 0.14% of the total primary production and 0.41% of the new production in the Luzon Strait.


Results
Hydrographic conditions. In nSCS, sea surface temperature (SST) ranged between 25.3 °C and 30.5 °C with an average of 29.3 ± 0.8 °C. Surface sea salinity ranged from 28.4 to 33.8 with an average of 33.2 ± 0.7. Most regions in nSCS were characterized by high temperature and high salinity (Fig. 1). In East China Sea, regions that heavily influenced by the Kuroshio also had high temperature and high salinity (Fig. 1). In nSCS during survey time, surface NO 3 − + NO 2 − , SRP, and silicate concentrations were lower than 0.15 μmol L −1 , 0.1 μmol L −1 , and 3 μmol L −1 respectively. The nutrient concentrations of most stations at the survey region were undetectable, especially at the Luzon Strait. In the East China Sea Kuroshio region, surface NO 3 − + NO 2 − , SRP, and silicate concentrations were also very low or undetectable.
Horizontal distribution of the N2 fixation activity. N 2 fixation rate of whole-water at the surface ranged from 1.14 to 10.40 nmol N L −1 d −1 with an average of 4.89 nmol N L −1 d −1 . N 2 fixation rate of the <10 μm fraction at the surface ranged from 0.63 to 7.61 nmol N L −1 d −1 with an average of 3.67 nmol N L −1 d −1 . N 2 fixation rate of whole-water changed a lot among different survey stations. High N 2 fixation of whole-water at the surface was observed in the ECS Kuroshio and the nSCS shelf. The nSCS basin showed low N 2 fixation activity. These rates were in the same order of magnitude as those measured under non-bloom conditions in the East China Sea, the southern Yellow Sea and the nSCS coastal upwelling 27,28 . The high N 2 fixation rates in the nSCS shelf were probably attributed to diazotrophic blooms in sampling stations. N 2 fixation activity in the <10 μm fraction exhibited a similar distribution to that of the whole-water in survey region (Fig. 2). Our experiments showed that unicellular diazotrophs contributed significantly to N 2 fixation in both the SCS and the Kuroshio. Vertical distribution of the N2 fixation activity. The whole-water N 2 fixation rate at Sta. SEATS ranged from 0.76 to 8.50 nmolNL −1 d −1 and the average was 4.33 nmolNL −1 d −1 (Fig. 3). The >10 μm fractions N 2 fixation rates decreased with the light intensity except for the 10% light depth, while the <10 μm fractions N 2 fixation rates increased with the light intensity in the middle three light depth. Light intensity is an important factor that determines the distribution of diazotrophs in the water column. Trichodesmium usually prefers the top sunlight of 75 m, but unicellular cyanobacteria tend to appear at deeper layers and down up to ~150 m 29 . The <10 μm fractions N 2 fixation rate of 100% and 1% light depth were undetectable in our study. The rare occurrence of unicellular cyanobacteria at the SEATS station has been verified by molecular study 30 . The growth of unicellular cyanobacteria in this station could be limited by iron resources due to the competition from non-diazotrophs. The maximum whole-water N 2 fixation rate was observed in the 10% light depths in our study, and the result was different from the previous study that the maximum N 2 fixation was in the surface layer and decreasing with depth in the lower euphotic zone 31 . Notably, the main factor resulted in this situation was the N 2 fixation rate by >10 μm fractions in the 10% light depths were larger than other depths extraordinarily. The same phenomenon was also observed in the nSCS 15 and North Pacific 32 . A possible reason generated for this phenomenon probably due to the result of the downward flux of Trichodesmium spp 33 .
Trichodesmium fixation rate. Hourly N 2 fixation rates of Trichodesmium were converted to daily rates with day length, assuming that Trichodesmium fixed N 2 only during the daytime 34 . Daily N 2 fixation rates of  Trichodesmium ranged from 0.11 to 9.83 pmolNtrichome −1 d −1 with an average of 4.03 pmolNtrichome −1 d −1 . The maximum rate of 9.83 pmolNtrichome −1 d −1 was recorded in Sta. E404 in the Luzon Strait for an experiment containing longitudinal colonies (cultivated from 8:00 am to 2:00 pm). The second maximum rate of 9.64 pmolNtrichome −1 d −1 was also recorded in Sta. E404 for an experiment containing radial colonies (cultivated from 8:00 to 10:00 am). In general, the Luzon Strait and ECS Kurosio had higher N 2 fixation rates by Trichodesmium than the nSCS shelf and basin. The variations of N 2 fixation rates by Trichodesmium were consistent with that of the Trichodesmium abundances (Fig. 4). In the ECS Kuroshio, Sta. PN04 showed a low N 2 fixation rate by Trichodesmium because the experiment was conducted at night (11:00 pm to 1:00 am of the next day).
A time-series assay was also conducted at Sta. E404. N 2 fixation rates by Trichodesmium radial colonies were measured at 8:00-10:00, 10:00-14:00, 14:00-16:00 and 16:00-18:00. The maximum N 2 fixation rate occurred in the 8:00 to 10:00 am. The N 2 fixation rate by Trichodesmium decreased sharply from 8:00 am to 6:00 pm, and the rate dropped off rapidly to near zero prior to the onset of the dark period. The N 2 fixation rate during 4:00 to 6:00 pm was almost undetectable (Fig. 5). The results generated from field incubation were consistent with previous study that high N 2 fixation rates were measured for ~6 hours surrounding the middle of the photoperiod and N 2 fixation declined sharply in the latter light period 35 . Trichodesmium standing crop and their contribution to total Chl a. In most instances, the individual free trichomes were the absolute dominant population in global ocean 4 . These free trichomes could form macroscopic colonies or aggregates which known as 'puffs' and 'tuffs' in extreme oligotrophic conditions 36 . In the present study, three living forms of Trichodesmium were observed during the survey time: the free trichome, the radial colony and the longitudinal colony. Likewise, the primary form of Trichodesmium in most stations were free living trichomes rather than colonies. This trait was in better agreement with previous study of Trichodesmium in the SCS and southern ECS 4,37 .
The average number of trichomes per radial colony and longitudinal colony in all stations were 80 and 24, respectively. The average number of cells per trichome was 60. This data was typically higher than previous reports in the SCS that colonial forms of Trichodesmium were usually composed of less than 10 trichomes per colony 37 . Our values were nevertheless lower than other oceans such as tropical North Atlantic Ocean (mean = 98 ± 11, range = 5-153 spp./colony), station ALOHA in subtropical eastern North Pacific Ocean (mean = 182, range = 10-375 spp./colony) and subtropical eastern North Atlantic gyre (mean = 112 ± 47 for puffs and 49 ± 16 for tufts, range = 57-156 for puffs and 31-63 for tufts spp./colony) [38][39][40] . Trichodesmium integrated concentrations ranged from 2.18 to 726.17 × 103 trichomes m −2 and the average was 177.65 × 103 trichomes m −2 . The Luzon Strait and the ECS Kuroshio had higher Trichodesmium abundances than the nSCS shelf and basin. Vertical distribution of Trichodesmium showed that from the surface to the bottom the abundances of Trichodesmium decreased dramatically (Figs 5 and 6).
According to the Trichodesmium integrated abundances and Chl a concentration per trichome measured at the sampling stations, estimated Chl a contributed by Trichodesmium varied from 0.1% to 58.5% of the total Chl a (Fig. 7). The values in the ECS Kuroshio and the Luzon Strait were significantly higher than that in the nSCS. Likewise, obvious spatial difference was also observed in another ocean. For example, literature have reported that Trichodesmium to account for, on average, about 61%, 18% and 5% of total Chl a in the Caribbean Sea, the

Comparison of N 2 fixation rate of Trichodesmium and other diazotrophs and their contributions
to the primary production. Calculated activities of Trichodesmium at most stations were quite low in the whole-water (<0.1 nmolNL −1 d −1 ) except at Sta. E404 where the N 2 fixation rate by Trichodesmium was as high as 57.7 nmolNL −1 d −1 . On the contrary, unicellular diazotrophs were accounted for a great part of the whole-water activity at most stations in the survey time. N 2 fixation was converted to carbon production using the Redfield ratio (C:N = 6.6:1) for comparison with the primary production. N 2 fixation by the whole-water was calculated  to contribute to 1.7% to 18.5% of the primary production (Table 1). According to the Trichodesmium integrated abundances and nitrogen fixation rates per trichome observed at the sampling stations, estimated integrated nitrogen fixation rates from Trichodesmium varied from 0.0031 to 7.0 μmol N m −2 d −1 . Calculated N 2 fixation rate by Trichodesmium showed it could contribute up to 0.14% of the primary production ( Table 2).

Discussion
In the summer, most regions of the nSCS have high SST except some regions which are influenced by the upstream. The stratified water of the nSCS results in the low supplement of the nutrient from deeper water. Originating from the southeast of Taiwan and east of the Luzon Strait, the Kuroshio is characterized by high temperature, high salinity, high transparency and low nutrients concentrations. As a result, most regions of the nSCS and the ECS Kuroshio were warm, stratified and oligotrophic which favored the growth of the diazotrophs. As an important diazotraph in the oligotrophic ocean and the Kuroshio water species indicator, Trichodesmium's distribution and nitrogen fixation rate in the nSCS and neighboring Kuroshio play an important role in the biological production, especially the new production of the SCS and the Kuroshio. Although conditions in both the SCS and the Kuroshio seem favorable for Trichodesmium growth, the Kuroshio had a higher abundance of Trichodesmium and nitrogen fixation rate than the nSCS. The standing crop and nitrogen fixation rate of Trichodesmium in the  SCS basin is very low (Fig. 4 and Table 2). Shiozaki et al. 26 suggested the high abundance of Trichodesmium in the ESC Kuroshio were ascribable to the supply of Trichodesmium spp. and other diazotrophs from the surrounding islands 26 . The Kuroshio is a western boundary current which originating from the southeast of Taiwan and east of the Luzon Strait, and enters the East China Sea (ECS) northeast of Taiwan. During our investigations, the surface abundance of Trichodesmium in Sta.E404 in the Luzon Strait was two orders of magnitude higher than stations in the nSCS and several times higher than stations in the ESC Kuroshio (Fig. 6). These results proved Trichodesmium spp. would likely to be transported to the mainstream of the Kuroshio from the Luzon Strait. Trichodesmium spp. were traditionally perceived to be the most abundant cyanobacteria in the marine, and it has been demonstrated to contribute the vast majority of N 2 fixation rates and primary production to the ocean 3,42 . However, increasing studies have shown that unicellular diazotrophs often appear to be of equal or greater abundance to the other known N 2 fixers due to the revolution use of polymerase chain reaction (PCR) techniques 7,31 . A DIC budget analysis at the SEATS in the SCS basin demonstrated that unicellular diazotrophs contributed enormously to the N 2 fixation, which were thus an indispensable part of the N cycle study in the SCS 43 . Chen et al. 31 observed that unicellular diazotrophs contributed 65% and 50% of the total N 2 fixation in the SCS and the Kuroshio, respectively 31  In the open ocean, many factors can influence the activity and diversity of diazotrophs. Water temperature is a major factor controlling the distribution of diazotrophs 44 . Trichodesmium is distributed where temperatures are between 20 °C and 30 °C and thrives when temperature is 25 °C or warmer 45,46 . By comparison, unicellular diazotrophs often present in deeper and colder waters than Trichodesmium 29 . In the present study, however, the correlation between size-fractioned N 2 fixation and temperature was not significant in both the nSCS and the Kuroshio. This might be attributed to the unusually strong wind and highly turbulent surface during summer cruise 31 . Diazotrophs, especially Trichodesmium, usually thrives in relatively stable environment. Previous works had demonstrated the importance of wind as a driving force for promoting the growth of Trichodesmium and diatoms 47,48 . Generally it has been found that highest concentrations of Trichodesmium occur during prolonged calm wind (<10 knots ≈ 5.14 m/s) conditions whereas the highest diatom concentrations occur during windy (>15 knots ≈ 7.72 m/s) conditions. Calm waters with low turbulence benefit bundle formation and tend to enhance N 2 fixation 4 . During the survey time, influenced by the typhoon "Swan" and "Morakot", a much lower abundance of Trichodesmium was detected at Sta. SEATS (1.08 × 10 3 trichomes m −2 ) compared with the previous study 4 . Besides, diazotrophs growth rates are presumably limited by the availability of non-N nutrients, iron and phosphate are key factors limiting diazotrophs N 2 fixation and growth rates 49 . Iron is a important cofactor of nitrogenase, and it plays an crucial role in the synthesis and expression of nitrogenase in diazotroiphs 50 . According to literatures, dissolved Fe (dFe) in surface waters ranged from 0.2-0.3 nmol L −1 in the SCS basin stations and 1.8-43.2 nmol L −1 in the northwest SCS coastal stations 28,37 . It was suggested that Fe availability may limit the rates of N 2 fixation in the SCS basin, while it was not happened in the nSCS coastal stations. As for the phosphate, it is a major component of cells which could also limit the growth of diazotroiphs 29,51 . In the present study, the phosphate (P) concentration in most of the areas was very low or undetectable. However, both Trichodesmium and unicellular diazotrophs had form a number of mechanisms to overcome P limitation. A key characteristic of Trichodesmium is the presence of gas vesicles, which could provide buoyancy and help them migrate vertically in the water column to optimize light intensity or to obtain P and other nutrients from deep waters 3 . Unlike with Trichodesmium, unicellular diazotrophs could not migrate vertically in the water column, but they had been proved to own a robust capacity for scavenging phosphorus in oligotrophic systems due to the presence of high-affinity phosphate scavenging systems 52 .
In addition, nitrogenase is instantaneously and irreversibly inactivated by oxygen, N 2 fixation is therefore a strictly anaerobic process 53 . However, many studies have demonstrated that both natural populations and laboratory cultures of Trichodesmium could fix N 2 exclusively in the daylight hours when photosynthetic activity takes place simultaneously 54,55 . Our result in Sta. PN04 and Sta. E404 also proved that Trichodesmium presented to be a daytime N 2 fixer. There are several hypotheses which aims to explain this phenomenon, but it is still an enigma how Trichodesmium can evolve O 2 while simultaneously fixing N 2 . The most popular hypothesis is that Trichodesmium can use a combination of spatial and temporal separation strategy of N 2 fixation and photosynthesis within the photoperiod 35,42,55 . Staal et al. 55 pointed that light stimulation of nitrogenase activity was most occurred at low O 2 concentration, while light stimulation became gradually less important at high O 2 concentration 55 . This result suggested the reduction of N 2 fixation in the latter stages of the light period was primarily attributed to the high net production of O 2 in the latter stages. Furthermore, nitrogenase activity can be recovered after a dark incubation when the cells were subsequently incubated in the light. Similarly, low O 2 concentration is in favor of the recovery of nitrogenase activity 55 .
Our results were comparable with global ocean, and higher than other study in the same area which might be caused by the discrepancy between the two methods (Table 3). On the basis of a C: N = 6.6, N 2 fixation by the whole-water was calculated to contribute to 1.7% to 18.5% of the primary production in surface waters, suggesting that N 2 fixation plays a very important role in supporting surface phytoplankton N demand in some stations. Such values were exceeded the previous studies in the nSCS, and were comparable with N input by biological N 2 fixation in other oceans 27,28,56 . The variation of N 2 fixation rate by Trichodesmium was consistent with that of the standing crop of Trichodesmium. According to the result of Chen et al. 4 new production from Trichodesmium N 2 fixation was higher in the Kuroshio than the nSCS in summer, we also found a higher N 2 fixation rate by  4 , Trichodesmium could contributed up to 0.4% of the new primary production at the Luzon Strait and <0.1% of the new primary production at the nSCS basin 4 . Compared with the previous studies, we got lower N 2 fixation rate and abundance of Trichodesmium (Table 4).

Conclusion
In conclusion, the abundance of Trichodesmium were higher in the Luzon Strait and ECS Kuroshio than the nSCS shelf and basin, and so does the N 2 fixation by Trichodesmium. It was suggested that the high abundance of Trichodesmium in the ECS Kuroshio were ascribable to the supply of Trichodesmium spp. and other diazotrophs from the surrounding islands. Nevertheless, it is still poorly understood and additional research is needed to resolve this question. According to the size-fractioned N 2 fixation rates, unicellular diazotrophs also contributed significantly to N 2 fixation in both the SCS and the Kuroshio. Further research should focus on the nitrogen and carbon biogeochemical cycles of unicellular diazotrophs in the SCS and the Kuroshio. Trichodesmium presented to be a daytime N 2 fixer with the maximum N 2 fixation rate in the 8:00 to 10:00 am and the rate decreased sharply as time goes by. Free trichomes were the absolute dominant population in the study area, and the radial colony and the longitudinal colony were also observed during the survey time which could be caused by the shortage of nutrient or trace element. The average number of trichomes per colony in present study was lower than values reported in other oceans, and the contribution of Trichodesmium to total Chla presented spatial difference which might be caused by the variation of their dominant phytoplankton. N 2 fixation by the whole-water in some stations contributed a lot to primary production in surface waters, suggesting that N 2 fixation plays a very important role in supporting surface phytoplankton N demand in these areas.

Materials and Methods
Station locations and sampling. The study area was located between 15°N to 35°N and 105°E to 130°E (Fig. 8). A total of 13 sampling stations were studied during the summer of 2009. Surface water was collected using an HCl-rinsed bucket at all the stations. Vertical seawater was collected by using 30 L Go-Flo bottles attached to a rosette multi-sampler, on which the conductivity, temperature, and depth (CTD) probes were installed (Seabird SBE 9/11). Temperature and salinity were profiled vertically by the CTD profiler.
Nutrients and Chl a measurements. Water samples for determination of nutrients were collected into 100 ml HCl-rinsed bottles and stored in 4 °C until analysis. Soluble reactive phosphate (SRP) concentrations were measured using standard molybdenum blue procedure 57 or a home-made ship-board C18 enrichment-flow injection analysis system 58,59 . Nitrate and nitrite concentrations (NO 3 − + NO 2 − ) were measured by reducing NO 3 − to NO 2 − with a Cd column and then determining NO 2 − using the standard pink azo dye method and a flow injection analyzer 60 .
Chl a samples were collected in 1 L bottles and then vacuum filtered (<10 mm Hg) through a 25 mm GF/F filter. Filters were kept in 10 mL vials and pigments were extracted with 90% acetone for 24 h inside a freezer at 4 °C. Chlorophyll concentrations of the extraction were then determined based on the fluorescence technique using a Trilogy (CHL NA, Model # 046) fluorometer on board.
Trichodesmium enumeration and its contribution to Chl a. Water samples for determining the vertical distribution of Trichodesmium were collected from 25 m, 50 m, 75 m, 100 m and 150 m depths in each sampling station. Specimens for enumeration of Trichodesmium were prepared on board by filtering a 1-10 L water samples onto a 20 μm polycarbonate membrane filter (25 mm in diameter). Concentrations of the samples were fixed with 1% neutralized formaldehyde and preserved in the dark until analysis. Trichomes and colonies on the entire filter were counted under an inverted microscope. In order to determine the contribution of Trichodesmium to the total Chl a, 30-50 Trichodesmium colonies were picked up and washed in the filtered sea water. Then the chlorophyll concentrations of the colonies were measured using fluorescence method as described above.
Primary production. Primary production (PP) was conducted according to Parsons et al. 61 by 14 C incubation method 61 . Seawater samples were collected from surface (Table 1) or five depths of the euphotic layer (Table 2), then prescreened through 200 μm mesh and filled into 250 mL acid-cleaned polycarbonate carboy (Nalgene; USA). Each sample was inoculated with 10 μCi NaH 14 CO 3 before incubation, and carboys which used for incubate subsurface samples were covered with the neutral density filter to simulate the in situ irradiance. Carboys were incubated in an on-deck incubator cooled by running water pumped up from a 5 m depth. After 4 h incubation, seawater samples were filtered through 25 mm diameter GF/F filters, and then the filters were placed in the aluminum foil bags and stored in the −20 °C freezer until analysis. In the laboratory, filters were fumed with HCl to remove residual inorganic carbon before they were moved into the scintillation vials, next the 10 mL scintillation cocktail (Ultima Gold; PerkinElmer; USA) were added. The activity of radioactive was counted using a Tri-Carb 2800TR liquid scintillation counter (Perkin-Elmer; USA).
Size-fractioned N 2 fixation rates measurements. N 2 fixation activity was measured by acetylene reduction assay 3 . Sampling stations were grouped into four sectors according to their bottom depths and locations (Fig. 8): the nSCS shelf (Sta. S209, E503 and D104), the nSCS basin (Sta. E607 and SEATS), the ECS Kuroshio (Sta. DH54 and PN04) and the Luzon Strait (Sta. E404 and E406). All experimental incubations were run in triplicate. 500 ml of surface seawater samples were introduced into a 600 mL HCl-rinsed polycarbonate bottles. After sealing with a butyl rubber stopper, 20 mL of acetylene was injected by replacing the same volume of headspace. Seawater samples that passed through a 10 μm mesh were also prepared for incubation of the surface samples in the same manner as whole seawater samples. Samples were incubated for 24 h in an on-deck incubator cooled by running water pumped up from a 5 m depth. At Sta. SEATS, subsurface samples were also obtained from 100%, 50%, 10%, 3% and 1% light depths relative to that of the surface to determine N 2 fixation at subsurface layers. The light intensity of the subsurface samples was adjusted to the in situ density using neutral density filters. N 2 fixation by Trichodesmium was also measured by acetylene reduction assay. A total of 6 sampling stations were occupied to determine the N 2 fixation rate of Trichodesmium (Fig. 8). Two sites were located in the ECS Kuroshio (Sta. PN04 and DH12), one site in the nSCS shelf (Sta. S209), one site in the Luzon Strait (Sta. E404) and two sites in the nSCS basin (Sta. SEATS and S504). Trichodesmium colonies were collected by a 76 μm mesh phytoplankton net towing gently in the surface water. Trichodesmium colonies were manually picked immediately by using a plastic bacteriological transfer loop and placed into a glass container with GF/F filtered seawater. Approximately 30-50 colonies were transferred to a 25 ml glass vial with 15 ml filtered seawater. The vials were sealed with butyl rubber stoppers. After injection of 2 ml acetylene, vials containing Trichodesmium colonies were incubated for 4-6 h in the daytime. To evaluate N 2 fixation of Trichodesmium, their activity at each station was calculated from their numerical abundance and their N 2 fixation activity per trichome.
At the end of the incubation, the gas phase of the vials was sampled with a 100 μl gastight syringe (Agilent) and manually injected into a flame ionization gas chromatograph (GC6890N, Agilent). The temperatures of injector, detector and oven were set at 250 °C, 200 °C and 60 °C respectively. The carrier gas was helium at the highest purity available at a flow rate of 9 mlmin −1 . The supply of H 2 and air for the FID were 30 mlmin −1 and 300 mlmin −1 , respectively. The column was 20m-long wide-bore fused silica (0.53 mm inner diameter) packed with Porapak U (Agilent). The ethylene production during the incubation was calculated as Capone 62 . The produced ethylene was converted to fixed nitrogen with a molar ratio of 4:1. After the assay, the incubated samples were fixed with 1% neutralized formaldehyde for the later enumeration of Trichodesmium.