Water samples were collected using Niskin bottles (bottles were closed by a messenger) from the four different depths (0, 5, 10, and 20 m) within the euphotic zone at three locations (NF-a, NF-b and NF-c that have station depths 42, 37, and 37 m, respectively) over the eastern Arabian Sea, during ORV Sagar Manjusha cruise during 10–14 May 2010 (Figure 1). Duplicate samples were taken from each depth in 1.225 L polycarbonate Nalgene bottles. All the bottles filled without headspace followed by the addition of ¹⁵N₂ gas (bubble method) with the chromatographic gas tight syringe (Montoya et al., 1996). Two milliliters of ¹⁵N₂ gas (99% ¹⁵N enriched gas from Cambridge Isotope Laboratories, Inc. USA) and 1 ml of 0.2 mmol ml⁻¹ NaH¹³CO₃ (99% ¹³C enriched) tracers were injected to each bottle (final enrichment of 16.6% for ¹⁵N and 8.5% for ¹³C) before the start of the incubations, which were performed during 10:00–14:00 h, i.e., symmetric to local noon. Tracer added bottles were covered with the calibrated neutral density filters to simulate the irradiance at the depths from which the samples were taken. After the incubations, the samples were filtered sequentially through pre-combusted (4 h at 400°C) Whatmann GF/F filters (25 mm diameter and 0.7 μm pore size), washed with filtered sea water, 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 of natural particulate organic nitrogen (PON) and carbon (POC).

Elemental analyzer interface with continuous flow mass spectrometer at the Physical Research Laboratory, Ahmedabad was used to measure the PON, POC, atom % ¹⁵N and atom % ¹³C in the samples. Volumetric rate of N₂ fixation were calculated following (Montoya et al., 1996).

Where, A_PN0 = ¹⁵N atom% in PON at the start of experiment, A_PNf = ¹⁵N atom% in PON at the end of experiment, t = time of incubation (4 hrs), [PON]_f = concentration of PON at the end of the experiment and A_Nenrich = ¹⁵N enrichment in the dissolved form after tracer addition at the start of the incubation, which is estimated as:

Surface water samples (at 0–1 m depth) were also incubated in complete dark condition and these estimates are attributed to the presence of heterotrophic diazotrophs and their contribution to N₂ fixation. It has been discovered that in the “bubble method,” only 40% enrichment can be achieved in 4 h incubations, which further results in 40% underestimation in the rates (Mohr et al., 2010). Hence, we multiplied N₂ fixation rates by a constant factor of 2.5 to avoid possible underestimation in the bubble gas technique. Later, it was discovered that the underestimation of rates is community dependent-there is less underestimation for Trichodesmium compared to that for other diazotrophs. Our sampling area witnesses Trichodesmium so underestimation in bubble method was less (Großkopf et al., 2012; White, 2012; Klawonn et al., 2015). We still multiplied our rates by 2.5 because we compared our results with Gandhi et al. (2011), who multiplied by the same factor. Carbon uptake rate is calculated by substituting N and ¹⁵N by C and ¹³C, respectively, in Equations (1) and (2) (Slawyk et al., 1977). Areal rates were calculated from the volumetric rates using the trapezoidal rule of integration, i.e., mean of volumetric rates were multiplied by the corresponding depth interval and then summed for all the depth intervals. Hundred milliliters of each sample was separately collected for nutrient measurements using a SKALAR auto analyzer at the offshore laboratory.

Spatial distribution of sea surface temperature (SST) was plotted using the Gridded High Resolution Sea Surface Temperature: Operational sea surface temperature and sea ice analysis (GHRSST: OSTIA) satellite data (Donlon et al., 2012). The mixed layer depth (MLD) was estimated based on the temperature criterion (0.2°C difference from the SST; de Boyer Montégut et al., 2004). Salinity values were obtained from the CTD data, values ranged from 35.23 to 35.56 at the sampling locations (Figure 2).

Temperature profiles from a portable CTD at sampling locations NF-a, NF-b, and NF-c showed that MLD varied between 17 and 20 m (Figure 2). Surface temperature at NF-b was ~1.4°C less than that at the other two stations. Temperature measurements obtained from CTD at surface level were reproduced by the satellite remote sensing data images (Figure 3). Salinity values obtained from CTD varied from 35.23 to 35.56 at the surface with maximum at NF-b and minimum at NF-c. NF-b is possibly influenced by Western India Coastal Current (WICC), which brings convecting mixing driven colder water from the north to the south during pre-monsoon (Schott and McCreary, 2001) as evidenced in the temperature profile (Figure 1).

Dissolved oxygen from the same CTD casts showed a dip of 2 mg L⁻¹ at 3 m depth at sampling location NF-c, whereas there was no vertical gradient in oxygen in the upper 20 m at locations NF-a and NF-b (Figure 2). Oxygen showed a sudden decline at all the stations below 20 m depth. Lower temperature at NF-b was associated with the detectable SiO₄ values, whereas the dip in the dissolved oxygen was associated with the maximum NO_x (NO2− + NO3−), phosphate (PO43−) values and higher N:P ratios at NF- c (Table 1).

Nutrient concentrations were higher at the surface and decreased with depth, with maximum surface value at NF-c (Figure 4). NO3− varied from 0.37 to 2.42 μM while PO43− ranged from 0 to 0.71 μM (Table 1). At sampling station NF-b, SiO₄ values were readily detectable with maximum at surface (1.69 μM) and decreased with depth (Table 1). We calculated P^* that indicates PO43− concentration deviations from the Redfield ratio (N:P = 16):

Positive P^* indicates PO43− concentrations in excess compared to NO3− (Deutsch et al., 2007). Surface water P^*-values were mostly in excess (except at 20 m at NF-c) and varied from −0.06 to 0.56 μM (Table 1). P^*-values at surface showed increasing trend from NF-a to NF-c, and decreased with depth at all the three locations. PO43− and NO3− maxima at NF-c could be due to the upwelling of subsurface water. Low P^* at NF-a could also be either due to the consumption of phosphate by diazotrophy or due to the upwelling, which starts during May (Gupta et al., 2016) as evidenced in the temperature profile at NF-a (Figure 2). Upwelling would create low oxygen just below MLD (Gupta et al., 2016; Sudheesh et al., 2016), as observed at NF-a (Figure 2). However, high P^*-values would have been expected in low oxygen water but deeper water, as evidenced from the P^* profiles (Table 1), has low P^*—possibly because of the remineralization of diazotrophy dominated organic matter (high N:P ratio) in the deeper depths. PO43− was below the detection limit at 20 m depth at NF-c yet there was N₂ fixation. This could again be attributed to the PO43− consumption by diazotrophs.

N:P ratio was less than the Redfield ratio (positive P^*, Table 1) at all stations with maximum at NF- c, which could have resulted in nitrogen limitation and further would have facilitated of N₂ fixation. N₂ fixation is controlled by iron and PO43− availability (Wu et al., 2000; Capone, 2001; Sañudo-Wilhelmy et al., 2001; Mulholland, 2007). At these sampling locations, N₂ fixation may be regulated by iron input as PO43− is not the limiting nutrient in the northern Indian Ocean as suggested by P^*-values (Shiozaki et al., 2014). Iron limitation is likely, as some part of the western Arabian Sea has been highlighted as a high nutrient low chlorophyll (HNLC) region (Naqvi et al., 2010; Moffett et al., 2015). Changes in the response of external forcing (e.g., seasonal monsoon and upwelling events) and inputs (e.g., aeolian and river influx) have specific control on N₂ fixation over the basin (Mulholland and Capone, 2009). POC and PON varied between 15 and 73 μM and 3–51 μM, respectively (Table 1), but did not show any correlation with C and N₂ fixation rates. However, the highest values of POC, PON, C, and N₂ fixation rates were observed at the surface at NF-a.

N₂ fixation rates varied from 4 to 238 nM N h⁻¹, while carbon uptake rate ranged between 16 and 1628 nM C h⁻¹ (Figure 5). Our N₂ fixation rates were higher than those synthesized by Capone et al. (2008) in the world oceans (0–5.4 nM N h⁻¹). The highest values for both carbon uptake and N₂ fixation rate for light incubation fixation (Figure 5) were observed at NF-a lying south of the other two sampling locations (Figure 1). N₂ fixation rates were higher in dark incubations compared to the light incubations at the surface level, except for NF-a (44.93, 63.32, and 164.7 nM N h⁻¹ at NF-a, NF-b, and NF-c, respectively, Figure 6). Whereas carbon uptake rates in the light incubations were an order of magnitude higher than in the dark incubations (Figure 6). Higher N₂ fixation rates for the dark incubations compared to light might be attributable to the presence of heterotrophic species at the surface. Heterotrophs contributed up to 52% to the total N₂ fixation (estimated from light and dark incubations, assuming light incubations correspond to phototrophic and dark to heterotrophic). Heterotrophic contribution to N₂ fixation could be even higher in the deeper waters due to favorable conditions for them. These higher values associated with dark incubation suggest that heterotrophic N₂ fixers might play an important role in fixing N₂ over the eastern Arabian Sea. In the surface waters of the Arabian Sea, the DNA and RNA recovered during the southwest monsoon periods were also categorized as those of heterotrophic bacteria (Jayakumar et al., 2012; Bird and Wyman, 2013). Therefore, our results suggest that an active N₂ fixation by heterotrophic bacteria could occur in the surface water of the eastern Arabian Sea (Shiozaki et al., 2014).

Euphotic-depth integrated autotrophic N₂ fixation and carbon uptake rates were highest at NF-a, where N₂ fixation rate was 8.4 ± 2.8 mmol N m⁻² d⁻¹ (standard deviation of duplicate samples) and carbon uptake rate was 74.7 ± 17.7 mmol C m⁻² d⁻¹. Our areal rates (8 mmol N m⁻² d⁻¹) were the highest ever observed in anywhere in world oceans (Table 2) and comparable to those measured by Gandhi et al. (2011) in the similar region-suggesting that the high rates during the spring 2009 were not episodic rather it could be a regular phenomenon during the spring in the Arabian Sea. High rates in the eastern Arabian Sea are probably due to the availability of both the essential nutrients for diazotrophs, i.e., iron and PO43−. Iron is available because this region unlike the western Arabian Sea is close to the Thar desert, whereas the western Arabian Sea is recognized to be an HNLC region (Naqvi et al., 2010). PO43− is available because upwelled water brings PO43− rich (compared to NO3−) water as NO3− is lost in denitrification. So the Arabian Sea is unique basin for sustaining diazotrophy and the higher rates of N₂ fixation. On the other hand, the other major oceans, like the Atlantic does not have enough phosphate (Wu et al., 2000), while the Pacific does not have enough iron (Behrenfeld and Kolber, 1999).

Gandhi et al. (2011) showed an overall increasing trend in the N₂ fixation rate from south to north, whereas in this present study, it was in the opposite direction. N₂ fixation supplies a large portion of the new N that supports marine productivity (Capone, 1997; Karl et al., 2002; LaRoche and Breitbarth, 2005; Mahaffey et al., 2005). Estimated contribution of new nitrogen to N₂ fixation in the present study is about 92% of new nitrogen, which is the same as that reported by the previous study (Gandhi et al., 2011).

Most observations around the world oceans are indirectly derived from geochemical estimates that are based on sub-surface nutrient distribution (Table 2). We have directly measured N₂ fixation using ¹⁵N₂ tracer, which is still the best available method despite its inherent problems of incomplete dissolution of the bubble (Großkopf et al., 2012). Our findings of high N₂ fixation in this region have large implications in understanding marine C, N, and cycles. If we extrapolate these high rates over the Arabian Sea, then we might find the missing nitrogen inputs (Codispoti, 2007). But we require to conduct more N₂ fixation experiments using the dissolution method in this region for extrapolation. In addition to seasonal upwelling, N₂ fixation might play a role in making the Arabian Sea perennially productive (Singh and Ramesh, 2015). Higher N₂ fixation and its degradation in the subsurface waters may also be useful in understating the phosphorous cycle-low P^* at the deeper waters and higher P^* at the surface-contrary to what is expected from the denitrified waters.

The authors declare that the research was conducted in the absence of any commercial or financial relationships that could be construed as a potential conflict of interest.