We report the first direct estimates of N2 fixation rates measured during the spring, 2009 using the 15N2 gas tracer technique in the eastern Arabian Sea, which is well known for significant loss of nitrogen due to intense denitrification. Carbon uptake rates are also concurrently estimated using the 13C tracer technique. The N2 fixation rates vary from ∼0.1 to 34 mmol N m−2d−1 after correcting for the isotopic under-equilibrium with dissolved air in the samples. These higher N2 fixation rates are consistent with higher chlorophyll a and low δ15N of natural particulate organic nitrogen. Our estimates of N2 fixation is a useful step toward reducing the uncertainty in the nitrogen budget.
 The Arabian Sea is one of major open-ocean oxygen-minimum zones (OMZ) in the world oceans, like the Eastern Tropical Pacific [Codispoti, 2007]. The Arabian Sea is biologically one of the most productive oceanic regions, due to the strong seasonal input of nutrients to the surface during the monsoons [e.g., Gandhi et al., 2010a]. Located on the east of the Arabian and southwest of the Thar deserts, it experiences a seasonal reversal of winds twice a year. During the summer monsoon (June–August), intense southwesterly winds cause nutrients to upwell from the deeper western Arabian Sea to the surface, while during the winter monsoon (December–February), cool, dry winds from the northeast induce convective mixing and increase productivity in the northeastern Arabian Sea [Madhupratap et al., 1996; Smith, 2001; Prakash and Ramesh, 2007; Kumar et al., 2010; Singh et al., 2010]. In addition, during spring and autumn inter-monsoons (mainly April–May and September–November), Trichodesmium (Nitrogen fixers – a filamentous cyanobacterium – found throughout the warm oligotrophic oceans) blooms occur in the eastern and central Arabian Sea when the sea is calm and bio-available nitrogen relatively scarce at the surface [Devassy et al., 1978; Capone et al., 1998; Parab et al., 2006]. In addition to higher primary production these blooms contribute to ‘new’ N input by fixing N2 under oligotrophic (i.e., NO3− concentration below the detection limit) conditions [Capone et al., 1997]. In addition to Trichodesmium blooms, the presence of diazotrophic γ-proteobacteria in the Arabian Sea has also been reported [Bird et al., 2005]. Based on nitrogen isotopic data of surface suspended particles Gandhi et al. [2010b] showed that recently fixed nitrogen by Trichodesmium could contribute as high as ∼79% of the nitrogen in surface suspended particles in the northeastern Arabian Sea. Bange et al.  have estimated the N2 fixation rate in the Arabian Sea to be 3.3 Tg N y−1. Patchiness and temporal variability restrict the precise estimates of such rates; thus, more frequent and widespread measurements are needed for a better quantification of the marine nitrogen budget.
 Denitrification and the seasonal occurrence of N2 fixing microorganisms (e.g., Trichodesmium) makes the Arabian Sea one of the most important global basins for testing oceanic nitrogen balance [Naqvi, 1987; Capone et al., 1998]. Higher surface productivity causes oxygen depletion in the deeper ocean and develops the largest open ocean OMZ in the Arabian Sea; this further leads to loss of bio-available nitrogen [Bange et al., 2000]. About 20–40% of global oceanic bio-available nitrogen is lost by this region (only ∼2% of the global ocean area) through denitrification [Bange et al., 2000, Bange et al., 2005]. The global nitrogen loss terms significantly exceed the current estimates of N2 fixation and other gain terms for nitrogen [Codispoti, 2007]. This imbalance indicates either a nonsteady state of the oceanic fixed-N inventory, or an over-estimation of loss processes and/or an under-estimation of nitrogen gain [Codispoti, 2007, and references therein]. However, isotopic signatures in sediments show that the oceanic nitrogen budget is in a steady state [Altabet, 2007]. Therefore, the direct estimation of N2 fixation is more important than believed previously [Codispoti, 2007].
 Though the Arabian Sea is a well known region for N2 fixation, only one study has reported N2 fixation rates in the central Arabian Sea [Capone et al., 1998] and no estimates are available yet from the eastern Arabian Sea where such blooms occur every year during the spring [Devassy et al., 1978; Sarangi et al., 2005; Parab et al., 2006]. In the present study, for the first time we have investigated N2 fixation rates in the eastern Arabian Sea using 15N2 gas (99% enriched in 15N) tracer technique to obtain more precise estimates [e.g., Montoya et al., 1996].
2. Materials and Methods
 The C2H2 reduction (AR) and 15N2 methods are widely used to measure N2 fixation rates. The AR method measures the total nitrogenase activity, while the 15N2 method yields a measure of the net incorporation of N2 in the biomass [Capone et al., 2005]. However, in oligotrophic systems, the latter is more sensitive than the former [Zehr and Montoya, 2007]. Here we have used the 15N2 gas tracer method as the bloom of Trichodesmium in the Arabian Sea flourishes under oligotrophic conditions [e.g., Capone et al., 1998]. Therefore, our estimates based on the 15N2 method might be more precise for the study area. The 13C tracer technique was also concurrently employed to quantify the total productivity.
 Sampling for measurements of N2 fixation and carbon uptake rates was carried out at seven different stations on board ORV Sagar Kanya (Cruise # SK 258) during the spring inter-monsoon (16th April to 1st May 2009, station locations shown in Figure 1). The sampling region was chosen based on previous reports of appearance of Trichodesmium blooms during spring in the region [Devassy et al., 1978; Sarangi et al., 2005; Parab et al., 2006]. Water samples were collected using a CTD rosette fitted with Go-Flo bottles; care was taken to avoid trace metal contamination. For carbon uptake measurements, six sampling depths were chosen to cover the euphotic zone (depth of the 1% surface light intensity), corresponding to 100, 80, 64, 20, 5 and 1% of surface irradiance. Light intensity and chlorophyll were measured using a hyperspectral radiometer (Satlantic Inc.) at each location. Once the samples were collected in pre-cleaned polycarbonate bottles (Nalgene, USA), labeled NaH13CO3 tracer containing 99 atom% 13C was added to the samples. For N2 fixation (using 15N2 gas containing 99 atom% 15N) measurements, water samples were collected at four different depths, corresponding to the surface, 5 m, 10 m and 15 m, as Trichodesmium is known to be restricted to the surface layers due to its positive buoyancy [Capone et al., 1997, 1998; Shiozaki et al., 2010]. Pre-cleaned ∼1.25 L polycarbonate bottles (Nalgene, USA) were used for the deck incubation. All the bottles were filled to overflow before being sealed with a leak-proof septum cap. This was followed by addition of 15N2 gas tracer to individual samples. Into all the bottles, 2 ml of 15N2 gas (Cambridge Isotope Laboratories, Massachusetts, USA) was injected using a chromatographic gas-tight syringe (Hamilton, UK). The technique of Montoya et al.  was followed. For both carbon and N2 fixation measurements, samples were collected in duplicate from each depth. After the addition of tracers, short time (4 h) incubations were performed according to the JGOFS protocol [United Nations Educational and Cultural Organization, 1994]. Tracer-added bottles were covered with well-calibrated neutral density filters to simulate the irradiance at the depths from which the water samples were taken. They were kept in a plastic tubs (1 m × 50 cm × 20 cm) on the deck; seawater from a depth of 6 m was circulated during the incubation between 10:00 and 14:00 Hrs local time. As it is well known that Trichodesmium releases recently fixed nitrogen in the form of ammonium and dissolved organic nitrogen (DON) [Mulholland et al., 2004, 2006] and sometimes even exceeding the net accumulation of N in biomass [Mulholland and Capone, 2001], which might underestimate the N2-fixation rate measured from longer incubation periods. To minimize this, we kept the incubation period short (4 h symmetric to the local noon). After incubation, bottles were transferred to the ship-board laboratory and all samples were filtered sequentially through pre-combusted (4 h at 400°C) 25 mm diameter and 0.7 μm pore size Whatmann GF/F filters, washed with filtered seawater, dried in an oven at 50°C overnight and stored for further mass spectrometric analysis.
 At each station, 2 L surface seawater was collected for measuring the nitrogen isotopic composition (δ15N) of natural particulate organic nitrogen (PON), before the commencement of 15N tracer measurements. 100 ml of each sample was separately collected for nutrient measurements using a SKALAR autoanalyzer. Analyzer (Flash EA 1112 series, CE instruments, Italy, interfaced with a Finnigan Delta Plus Continuous Flow Mass Spectrometer) via ConFlo III is used to measure particulate organic nitrogen (PON) and carbon (POC) and atom% 15N and atom% 13C in the samples. For the calculation of N2 fixation rate, we use the equation of Montoya et al.  as follows: Volumetric rate of N2 fixation
where AN2 = 15N enrichment of N2 available for fixation, APN0 = 15N enrichment of PON at the start of experiment, APNf = 15N enrichment of PON at the end of experiment, t = time of incubation (4 h), [PN]f = concentration of PON at the end of the experiment. In the above equation all the terms are measured except AN2, which is defined as
Ambient nitrogen (dissolved nitrogen gas) concentrations were calculated using the method of Weiss . When 2 ml 15N2 gas injected to 1.25 L natural sample (ambient concentration 8.16 ml/L, at salinity 35 psu and temperature 30°C [Weiss, 1970]), theoretical value of AN2 is ∼17%. It has been observed recently that the AN2 calculated as above leads to an underestimation of N2 fixation rates [Mohr et al., 2010]. This is caused by the slow equilibration and low solubility of the injected 15N2 gas with the dissolved gas in the water. Therefore, we have corrected our values using the graphs presented by Mohr et al. . As shown by Mohr et al. [2010, Figure 1], after 4 h of incubation only ∼40% of the actual N2 fixation rates were obtained. Thus all the measured N2 fixation rates were multiplied by 2.5. Depth-integrated N2 uptake rates were calculated by trapezoidal integration. Probable errors in the factors involved in the estimation of N2 fixation estimation could be due to (i) Incubation time (3–4%) (ii) Isotopic composition of PON (APN0 and APNf) ± 0.2‰ (iii) AN2 − 3% (iv) PON concentration 4%. Hence, the maximum error in the volumetric rate could be 6% (as explained by Montoya et al. ).
 Carbon uptake rates are estimated likewise following Slawyk et al. . Two ml NaH13CO3 of 0.2 m mol/ml concentration added to 2 L samples. Ambient dissolved inorganic carbon (DIC) concentration (2 mM) for carbon uptake calculations was taken from Sabine et al. . Final theoretical % labeling of 13C is ∼10%. Nitrogenase activity in Trichodesmium exhibits diel variation and remains active only during near dawn to near dusk [Capone et al., 1990], similar to the carbon uptake [Berman-Frank et al., 2001]. Therefore, the daily uptake rates of N2 and carbon uptake are calculated by multiplying hourly values by 12 [Scanlan and Post, 2008]. The maximum differences in the duplicate mass-spectrometric measurements of PON and POC were found to be less than 10%. The coefficients of variation in atom% 15N and atom% 13C measurement were less than 1%. Standards IAEA-NO3 (KNO3, #213) and IAEA-N-2 ((NH4)2SO4, #342) for nitrogen, and Australian National University (ANU) sucrose for carbon, were routinely analyzed to check the accuracy of the isotopic measurements, in addition to internal laboratory isotope standards to check the isotopic reproducibility [see Kumar and Ramesh, 2005; Prakash et al., 2008; Singh, 2011].
3. Results and Discussion
 Sampling was done wherever Trichodesmium bloom was sighted (shown in Figure 1 and stations are named as NF1 to NF8) in the region. At NF3, located more toward the open ocean than the others, no bloom was observed, and therefore N2 fixation measurement was not carried out.
3.1. Environmental Conditions During Spring 2009
 Euphotic depth was less than 40 m at all the stations but NF3. The photosynthetically active radiation (PAR) values at sea surface varied from 826 to 1616 μmol m−2s−1 (Table 1). The highest and the lowest PAR at surface were found at NF4 and NF3, respectively. Overall, average PAR at surface during the study period was 1272 μmol m−2s−1.
Table 1. Latitude, Longitude, Water Depth, Temperature, Salinity, Photosynthetically Active Radiation (PAR), Chlorophyll a, and Ratios of N/P and N/Si at the Surface at the Different Stationsa
Water Depth (m)
Sea Surface Temperature (°C)
PAR (μmol m−2 s−1)
Chlorophyll a (μgL−1)
Temperature and salinity presented here are measured using a bucket thermometer and an Autosal, respectively; however, the values discussed in the text and figures are collected using a CTD.
 Sea surface temperature (SST; °C) decreased significantly from the south to the north, varied from 29 to 30.6°C with the highest value at NF2 (Table 1). In contrast, surface salinity showed the opposite trend, it increased from the south to the north, and was between 34.7 and 35.8, with the highest value at NF5. This north-south gradient in SST and salinity is a common feature of the Arabian Sea and has also been reported earlier [e.g., Madhupratap et al., 1996; Prasanna Kumar and Narvekar, 2005; Prakash and Ramesh, 2007]. The depth where temperature decreases to 1°C is termed as mixed layer depth (MLD), following Prasanna Kumar and Narvekar . MLD varied between 20 and 35 m at all the stations. During spring, winds are usually weak and the strong surface light intensity promotes a strong stratification, which prevents the supply of nutrients from the deep to the surface.
 Depth profiles of nutrients (NO3, NO2, PO4 and SiO4) are shown in Figure 2. Ratios of nitrate to phosphate (N:P) and nitrate to silicate (N:S) observed at the surface waters are listed in Table 1. N:P varies from 0.8 to 16.5 in the region with the maximum at NF2 (Table 1). Except at NF2, the ratio remains lower (N:P < 8) than the Redfield Ratio (C:N:P:: 106:16:1) at all the locations. Similar values are also found in the sub-surface layers. The ratio is as low as <0.1 between 3 and 6 m depth at NF8. In general, it remains between 0.01 and 8 throughout the euphotic zone at all the locations, except at the surface at NF8 and in sub-surface at NF1. At NF1, N:P is quite high as phosphate remains near the detection limit in the sub-surface layers. Nitrogen to silicon ratio (N:Si) is also lower than the Redfield ratio; it varies from 0.02 to 0.3 in the surface layers. Like the N:P ratio, the sub-surface N:Si values are also lower than the Redfield ratio. N:P < 10 and N:Si < 1, observed here, are indicative of potential nitrogen limitation [Harrison et al., 1977]. Such conditions, along with weaker winds, shallow MLD and a calm sea, favor the occurrence of both Trichodesmium erythraeum and Trichodesmium thibautii.
 Iron and phosphorus control N2 fixation [Capone, 2001]. There is no phosphate limitation since phosphate concentration was mostly more than 3 nM (concentration <3 nM is considered as phosphate-limited) here, unlike in the high iron-input regions [Shiozaki et al., 2010]. This could be because parts of the Arabian Sea are considered to be iron-limited [Naqvi et al., 2010], despite the supply of iron from Arabia [e.g., Krishnamurti et al., 1998]. Our data of phosphate concentration and N:P rules out phosphate-limited conditions (except at NF1).
3.3. Chlorophyll a
 A large variation is seen in the surface chlorophyll a values, from <0.23 to as high as >2.55 μg L−1, with the highest at NF7 (Table 1). The locations NF5–7 witness dense blooms of Trichodesmium, under such conditions, surface chlorophyll a concentration as high as >2000 μg L−1 has been reported during May 2003 [Parab et al., 2006] and more than 100 μg L−1 during the springs of 1975 and 1977 [Devassy et al., 1978]. Euphotic-depth-integrated chlorophyll a also shows a similar pattern, as that of surface chlorophyll a. The maximum euphotic-depth-integrated chlorophyll a is observed at NF7, with the lowest at NF6 (although this was one of the stations with a dense bloom). Overall, it varies between 8.2 and 71.3 mg m−2. This variation is consistent with the intensity of Trichodesmium blooms observed and reported earlier from similar occurrences in the region by Matondkar et al. ; however, concentration in excess of 100 mg m–2 chlorophyll a values has been reported earlier [Desa et al., 2005].
3.4. N2 Fixation and Carbon Uptake Rates
Figure 3 presents the depth profiles of N2 fixation and carbon uptake rates. A large variation in surface N2 fixation rate is observed; it varies from 0.1 to 1125 nM N h−1, with the highest rate at NF6; much higher range than reported by Carpenter and Capone  (0 to 5.4 nM N h−1). Except at NF6, the surface rates are less than 45 nM N h−1. The arithmetic mean is ∼16 nM N h−1, excluding NF6; with the inclusion of NF6, the mean increases to ∼175 nM N h−1. As the variation is of more than an order of magnitude, the geometric mean is a better statistic; this is 13 nM N h−1. Surface values observed at three stations, NF5, NF6 and NF8, are significantly higher than those reported by Capone et al.  and Church et al.  in the central Arabian Sea and the North Pacific Ocean, respectively. Capone et al.  used C2H2 reduction technique and estimated ∼10 nM N h−1 in the top 0.5 m, similar to most of values presented here, but significantly less than those at NF5, NF6 and NF8. In the North Pacific Ocean, surface rates ranged from 0.5 to 11 nM N d−1 [Church et al., 2009], less than the values reported here, if our values are converted to daily rates by multiplying by 12. At all the locations, surface values are much higher than the sub-surface values except at NF2, where NO3 was present at the surface. N2 fixation contributes ∼50–90% to the total column N2 fixation in the top 5 m; this is expected as positive buoyancy keeps Trichodesmium at the surface. Capone et al.  also found about threefold higher N2 fixation in the top 0.5 m than that occurring between 0.5 to 40 m. N2 fixation rate increases from the surface to 5 m depth and then decreases up to 15 m at NF2; a higher rate is observed at this station than at NF1, though these stations are nearby (Figure 1).
 Depth profiles of carbon uptake rates are also shown in Figure 3. Primary productivity (carbon uptake rate) is not measured at NF2. It varies between 91 and 4594 nM C d−1 (with an average of 834 nM C d−1) at the surface. NF6 shows the highest surface productivity, an order of magnitude higher than at the other stations. The lowest surface productivity is seen at NF3 and NF7. Euphotic-depth-integrated carbon uptake rates are presented in Figure 4. Column productivity values range from 19 to as high as 100 mmol C m−2d−1, with the highest at NF6. The average productivity is 46 mmol C m−2d−1. Ryther et al.  have reported a mean productivity of 186 mmol C m−2d−1 (reported value is 2.23 g C m−2 d−1 and converted to mmol C m−2d−1) for the Oman coastal upwelling zone (north of 18°N) compared to 257 mmol C m−2d−1 (reported value, 3.08 gC m−2d−1) reported for the same region by Savidge and Gilpin . Owens et al.  have reported primary productivity of ∼42 mmol C m−2d−1 (reported value, 0.5 gC m−2d−1) at the equator to 25 mmol C m−2d−1 (reported value, 0.3 gC m−2d−1) in the oligotrophic gyre in the central Arabian Sea, and >208 mmol C m−2d−1 (reported value, 2.5 gC m−2d−1) in the upwelling region off the coast of Oman during September–October, 1986. Qasim  reported a very wide range of primary productivity for the northern Arabian Sea (north of 15°N) varying from 0.8 to 501 mmol C m−2d−1 (reported values 0.01 to 6.01 gC m−2d−1) with an average value of 70 mmol C m−2d−1 (reported value, 0.84 gC m−2d−1). He also found the coastal region to be more productive (average 1.33 gC m−2d−1) than the offshore region (0.63 gC m−2d−1). The values reported here are much higher than that observed earlier during spring in the open ocean region of the Arabian Sea; however, they do confirm the mean value reported under similar bloom conditions earlier in the region by Matondkar et al. , i.e., column productivity from 8 to as 1417 mmol C m−2d−1 (reported values 0.1 to 17 gC m−2d−1) with an average of ∼225 mmol C m−2d−1 (reported value 2.7 gC m−2d−1).
 The depth profiles of carbon uptake rate at NF1 and NF3 are similar. Despite the significant carbon uptake rate, NF1 shows no significant change in N2 fixation with depth and the rates are small (<0.6 nM N h−1). This indicates the possible dominance of non-diazotrophic plankton here, which contributes to carbon uptake but not to N2 fixation. At NF1, nitrate is quite high except at the surface, while phosphate is near the detection limit. Silicate is also significantly higher here. All these conditions favor the non-diazotrophic plankton, which results in lower a N2 fixation rate with moderate carbon uptake rates. Variation in the N2 fixation and carbon uptake rates with depth are similar at NF4. Stations NF5, 6 and 8 showed nitrogen limiting conditions throughout the euphotic zone and witnessed dense blooms of Trichodesmium; however, NF7 (located near NF6) exhibited less N2 fixation than NF5, 6 and 8. This could be due to the relatively higher N:P value at NF7.
Figure 4 presents the column-integrated N2 fixation rates, primary productivity and chlorophyll a estimates. Column-integrated N2 fixation rates vary from ∼0.1 to 34 mmol N m−2d−1 (with an arithmetic average of 5.5 mmol N m–2d–1). The arithmetic average value decreases to ∼0.7 mmol N m−2d−1, if we exclude NF6, the densest bloom station having the highest N2 fixation rate. The geometric mean value for all stations is 0.9 mmol N m−2d−1. Although chlorophyll a is higher at NF7, N2 fixation here is lower than at NF6, probably due to the patchiness of diazotroph abundance. However, NF5 and 8 are near each other (∼20 km) and show similar N2 fixation rates. In the central Arabian Sea, the average column N2 fixation was ∼0.17 mmol N m−2d−1 [Capone et al., 1998].
 It is worth comparing the present N2 fixation rates with those reported for other oceanic region. Average areal rates of N2 fixation in the North Atlantic Ocean, covering different seasons and areas, ranged from about 0.06 to 0.90 mmol N m−2d−1 [Capone et al., 2005]. Chen et al.  reported seasonal variations in N2 fixation rates in the upstream of Kuroshio and South China Sea. They observed N2 fixation rates of 0.002 to 0.168 mmol N m−2d−1 and 0.001 to 0.013 mmol N m−2d−1 during different seasons in the Kuroshio and South China Sea, respectively. In both the basins, rates were found to be the highest during summer. N2 fixation rates have been reported below the detection limit to 0.09 mmol N m−2d−1 along the 155 °E meridian from the equator to 44°N in the western North Pacific Ocean during spring 2007 [Shiozaki et al., 2010]. Church et al.  found depth-integrated N2 fixation rates between 0.02 and 0.31 mmol N m–2d–1 over a three year period in the North Pacific Ocean. N2 fixation observed by us is higher than those observed elsewhere in the world oceans, except at the North Atlantic where rates are comparable. This shows that the Arabian basin fixes a significant amount of atmospheric N2 and hence provides new nitrogen to the ocean. The Arabian Sea is known for higher denitrification (nitrogen loss process) and the present results provide the first direct evidence of higher N2 fixation (nitrogen gain process) in this basin. Therefore, more measurements of direct estimation of N2 fixation are needed to further constrain the nitrogen gain in the basin better.
 The average ratio of POC:PON of natural samples was near the Redfield ratio, but with a large variation i.e., 6.5 ± 5.0. This ratio was higher (9.4 ± 5.7) at non-bloom and less-dense bloom stations (NF1–4) stations than at the dense bloom stations i.e., 3.6 ± 1.0. In contrast, carbon to N2 consumption ratio from the incubation experiments was 263 (ranges from 4 to 1391), much higher than the Redfield Ratio. Similar to natural samples, consumption ratio was higher at NF1 and 4 (438) than at NF5–8 (147). Mulholland et al.  have found the ratio between 6.1 and 42.7 in the Gulf of Mexico. Orcutt et al.  reported the ratio as high as 198 for the North Atlantic. The large variations have been attributed to the release of NH4 during incubation [Mulholland and Capone, 2001] as NH4 and DON may become elevated as a result of recent N input by N2 fixation [Devassy et al., 1978; Karl et al., 1992]. Particularly, in the tropical oligotrophic oceans, availability of combined N, NH4 and DON is low and rapidly recycled nitrogen may support much of the nitrogen demand to phytoplankton [Eppley and Peterson, 1979; Bronk et al., 1994]. As incubation was limited to 4 h in the present study, the effect of release of NH4 and DON over the ratio might be small. The other possible reason could be that we might have measured higher photosynthesis and a lower N2-fixation phase of the Trichodesmium bloom as these activities are known to show opposite temporal trends [Berman-Frank et al., 2001]. In this scenario, the estimates of N2 fixation presented here would be on the lower side. The possible presence of microbes that are incapable of fixing N2 could also be the reason for the observed higher consumption ratio. However, it is difficult to pin point the exact mechanism based on the available data set.
3.5. Nitrogen Isotopic Composition (δ15N) of Natural Surface Samples
 Nitrogen isotopic composition (δ15N) is usually reported relative to atmospheric nitrogen
This is a proxy for the existence of N2 fixing microorganisms. Bacteria fix dissolved oceanic nitrogen with little isotopic fractionation and therefore δ15N of a natural sample likely reflects that of its nitrogen source [Altabet and McCarthy, 1985; Kumar et al., 2004, 2005]. Since the reference is air, the expected δ15N of nitrogen fixing bacteria is likely to be close to 0 ‰. δ15N values of surface PON at different stations are shown in the top of Figure 4. They vary between 1 and 6‰ with a minimum value corresponding to the sites with maximum N2 fixation rates i.e., at sampling station at NF6. This confirms that the source of new nitrogen here is dissolved air.
 The δ15N of organic matter is a mixture of various nitrogen sources. Capone et al.  used a simple mass balance equation to estimate the contribution of recently fixed nitrogen to the surface pool of suspended particles. They assumed it to be a two component mixing and took the nitrogen isotopic composition of Trichodesmium as δ15N = 0‰ and that of the nitrate below mixed layer as δ15N = 10‰. Using the same approach i.e., isotopic composition of PON − δPON = δTfT + δNfN (with fT + fN = 1), where δT (0‰) and δN (10‰) are isotopic composition of Trichodesmium and dissolved nitrate, respectively; and fT and fN are the fractions of Trichodesmium and nitrate, respectively; our isotopic data on PON of surface natural samples suggest that recently fixed nitrogen contributes 40 to 90% (with an average of 60%) of the nitrogen to the surface PON. If so, the fraction of primary productivity due to Trichodesmium (depth profiles up to 15 m) can be estimated; we find that the carbon to N2 consumption ratio is accordingly revised downward to an average of 150 (varies from 1 to 707).
3.6. Variability in the N2 Fixation Rates
 A synthesis of the previous and recent estimates of nitrogen fluxes in the Arabian Sea is presented in Table 2. Gandhi et al. [2010b] reported N2 fixation 0.002 to 0.54 mmol N m−2 d−1 associated with Trichodesmium estimated from the abundance of Trichodesmium and specific N2 fixation rates of 1.5 pmol N trichome−1 h−1 for spring 2006. Using these data with taking the total surface area covered by blooms (more than 20% of the area of the Arabian Sea [Capone et al., 1998]) and their persistence (more than 30% of the time [Westberry and Siegel, 2006]) with the estimated N2 fixation rates can be used to extrapolate the nitrogen input by Trichodesmium blooms for the whole region. The estimated annual nitrogen gain for the region by such blooms is ∼0.9 Tg N y−1 in good agreement with that reported by Capone et al.  (∼1 Tg N y−1). However, in the present study, fixation rate at NF6 is ∼50 times higher than at other stations (Figure 4). For calculating the average rate we take the arithmetic mean of data except from NF6, and then take the geometric mean of this value and the data from NF6; we estimate that 5 mmol N m−2 d−1 is likely fixed in the Arabian Sea during a bloom. After extrapolating the results into the area of bloom occurrence and its duration we have estimated that 15.4 Tg y−1 (arithmetic mean of all results is 16.9 Tg N y−1, which is the same within the error; however, geometric mean of all the data suggests 2.6 Tg N y−1) nitrogen is fixed in the Arabian Sea, which is ∼11% of global nitrogen fixation in 2% of the global ocean area. In addition, dry and wet deposition aerosols supply 1.2 ± 0.2 Tg N y−1 in the Arabian Sea and much lesser amount through rivers i.e., 0.1 ± 0.02 Tg N y−1 [Singh and Ramesh, 2011; A. Singh et al., An assessment of the contribution of atmospheric deposition to new productivity in the northern Indian Ocean, submitted to Journal of Geophysical Research, 2011]. A simple calculation suggests that ∼92% of ‘new’ nitrogen is gained through N2 fixation.
Table 2. Developments in the N-Cycle of the Arabian Sea: Results From Earlier, Recent and the Present Study (Nitrogen Fluxes in Tg N y−1)
This is an indirect estimate suggested by Lam et al. , that nitrogen loss through anammox can be equal to that through denitrification.
 The Arabian Sea loses 60 Tg N y−1 through denitrification [Bange et al., 2005]. In addition, it has been suggested that biological production of N2 in the Arabian Sea may exceed estimates based on canonical stoichiometries for denitrification [Devol et al., 2006]. If the suggestion of Lam et al.  is accounted for, the same amount of nitrogen could be additionally lost through anaerobic ammonium oxidation (anammox [Kuypers et al., 2003]).
 The mystery about the missing nitrogen is yet unsolved and it appears that there may be significant nitrogen fixation over the Arabian Sea not limited to spring and fall seasons alone. We believe that Trichodesmium may not be the only species which fixes N2, as the presence of another diazotroph, i.e., γ-Proteobacteria, is also reported in the Arabian Sea [Bird et al., 2005] even in the presence of bio-available nitrogen. In addition, Montoya et al.  reported substantial N2 fixation by smaller microbes. Heterotrophic diazotrophs may also contribute significantly to N2 fixation in and outside the euphotic zone [Mulholland and Capone, 2009, and references therein]. To address these issues, more measurements are required during different months, possibly including other diazotrophs with advanced techniques. Use of nano-scale secondary-ion mass spectrometry (nanoSIMS) in future experiments to describe the role of an individual cell in a group of carbon and N2 fixing microorganisms appears to hold promise [Musat et al., 2008; Ploug et al., 2010]. Estimating N2 fixation in single cell would be capable of identifying other organisms which assimilates the released 15NH4 and 15DON by N2 fixers [Mulholland and Capone, 2001] during longer incubations.
 We report first results from the direct measurements of N2 fixation in the Arabian Sea, useful to constrain the global nitrogen budget. N2 fixation by Trichodesmium occurs mainly in the upper 10 m of ocean surface while carbon uptake takes place throughout the euphotic zone. Our results suggest that significant dinitrogen is fixed (∼11% of the global N2 fixation) during a Trichodesmium bloom in the Arabian Sea. The present estimate shows that the previously estimated N2 fixation rates could be considerably underestimated, however, the fixed nitrogen (15.4 Tg N y−1) is still far less than the estimated nitrogen loss (∼120 Tg N y−1) through denitrification, and assuming an equal rate of loss of nitrogen through anammox. N2 fixation is the most important process among all the nitrogen gain processes and we estimate that the Arabian Sea gains ∼92% of its ‘new’ nitrogen through this process. Lower values of δ15N of PON associated with higher fixation rates confirm the presence of N2 fixing bacteria. Higher N2 fixation rates are consistent with higher carbon uptake rates. Higher consumption ratios suggest the possible under-estimation of N2-fixation or the presence of additional microbes that are incapable of fixing N2.
 We thank all the participants and crew members of ORV Sagar Kanya (SK 258) for their assistance onboard. We also thank the Indian Space Research Organization Geosphere-Biosphere program for funding. Critical comments from two anonymous reviewers helped improve the manuscript. This is INCOIS contribution 89.