Spatiotemporal variations of nitrogen isotopic records in the Arabian Sea

Available reports of dissolved oxygen, δN of nitrate (δNNO3 ) and δ N of total nitrogen (δNbulk) for trap material and surface/downcore sediments from the Arabian Sea (AS) were synthesized to explore the AS’ past nitrogen dynamics. Based on 25 μmol kg dissolved oxygen isopleth at a depth of 150 m, we classified all reported data into northern and southern groups. By using δNbulk of the sediments, we obtained geographically distinctive bottom-depth effects for the northern and southern AS at different climate stages. After eliminating the bias caused by bottom depth, the modern-day sedimentary δNbulk values largely reflect the δNNO3 supply from the bottom of the euphotic zone. Additionally to the data set, nitrogen and carbon contents vs. their isotopic compositions of a sediment core (SK177/11) collected from the most southeastern part of the AS were measured for comparison. We found a one-step increase in δNbulk starting at the deglaciation with a corresponding decrease in δCTOC similar to reports elsewhere revealing a global coherence. By synthesizing and reanalyzing all reported down core δNbulk, we derived bottom-depth correction factors at different climate stages, respectively, for the northern and southern AS. The diffusive sedimentary δNbulk values in compiled cores became confined after bias correction revealing a more consistent pattern except recent 6 ka. Such high similarity to the global temporal pattern indicates that the nitrogen cycle in the entire AS had responded to open-ocean changes until 6 ka BP. Since 6 ka BP, further enhanced denitrification (i.e., increase in δNbulk) in the northern AS had occurred and was likely driven by monsoon, while, in the southern AS, we observed a synchronous reduction in δNbulk, implying that nitrogen fixation was promoted correspondingly as the intensification of local denitrification at the northern AS basin.


Introduction
Biogeochemical processes of nitrogen in the ocean are intimately related to various elemental cycles (synergistically modulate atmospheric CO 2 and N 2 O concentrations), hence the feedback on the climate being on a millennial time scale (Gruber, 2004;Falkowski and Godfrey, 2008;Altabet et al., 2002).Though oxygen deficient zones (ODZs) occupy only ∼ 4 % of ocean volume, the denitrification process therein contributes remarkably to the losses of nitrate, leaving excess P in the remaining water mass to stimulate N 2 fixation while entering the euphotic zone (Morrison et al., 1998;Deutsch et al., 2007) and thus controlling the budget of bioavailable nitrogen in ocean.Denitrification leaves 15 NO − 3 in residual nitrate (Sigman et al., 2001), whereas N 2 fixation introduces new bioavailable nitrogen with low δ 15 N values (Capone et al., 1997) into ocean for compensation.The Arabian Sea (AS) -one of the three largest ODZs in the world ocean with distinctive monsoon driven upwelling -accounts for at least one-third of the loss of marine fixed nitrogen (Codispoti and Christensen, 1985) playing an important role in the past climate via regulating atmospheric N 2 O concentration (Agnihotri et al., 2006) or nitrogen inventory to modulate CO 2 sequestration through a biological pump (Altabet, 2006).
Sedimentary nitrogen isotope, measured as standard δ notation, with respect to standards of atmospheric nitrogen, is an important tool to study the past marine nitrogen cycle.
Previous measurements of δ 15 N bulk in various cores and surface sediments in the AS showed the following points: (1) near-surface NO − 3 in the AS is completely utilized in an annual cycle, resulting in small isotopic fractionation between δ 15 N of exported sinking particles and δ 15 N of NO − 3 supplied to the euphotic zone (Altabet, 1988;Thunell et al., 2004); (2) monsoon-driven surface productivity and associated oxidant demand were regarded as the main control on water column denitrification in the past (Ganeshram et al., 2000;Ivanochko et al., 2005); (3) sedimentary δ 15 N bulk primarily reflects the relative intensity of water column denitrification in this area (Altabet et al., 1995(Altabet et al., , 1999)); (4) oxygen supply at intermediate depth by the Antarctic intermediate waters (AAIWs) can modulate the denitrification intensity in the northern AS (Schulte et al., 1999;Schmittner et al., 2007;Pichevin et al., 2007).Among previous research, the geographical features in sedimentary δ 15 N bulk between the north and south basins of the AS have not been discussed, particularly on the basis of bottom-depth effect, which might be different during glacial and interglacial periods.
In this study, a sediment core (SK177/11) collected from the slope of the southeastern AS was measured for organic C and N contents and their stable isotopes.We synthesized previous hydrographical and isotopic data, such as dissolved oxygen (DO), N * (N * = NO − 3 16 × PO 3− 4 + 2.9; Gruber and Sarmiento, 2002), and δ 15 N of nitrate, as well as trapped material and surface/downcore sediments, among which surface and downcore sediments may have experienced more intensified diagenetic alteration.Based on the subsurface of a DO concentration of 25 µmol kg −1 isopleth at 150 m, the data sets in the AS were separated into north and south basins by time span (glacial, Holocene and modern) for comparison.We aim to (1) investigate the geographic and glacialinterglacial differences in bottom-depth effect and to (2) retrieve extra information from sedimentary δ 15 N bulk by removing basin/climate stage specific bottom-depth effects, thus better deciphering the environmental history of the Arabian Sea.
blow during winter from a high-pressure cell of the Tibetan Plateau, whereas heating of the Tibetan Plateau in summer (June to September) reverses the pressure gradient leading to warm and moist southwesterly winds and   precipitationmaximum.In the present day, the SW monsoon is much stronger than its northeastern counterpart.The spatial distribution of DO at a depth of 150 m for the AS is shown in Fig. 1a (World Ocean Atlas 2009, http:// www.nodc.noaa.gov/OC5/WOA09/woa09data.html), which shows a clear southwardly increasing pattern with DO having increased from ∼ 5 to > 100 µmol kg −1 , and the lowest DO value appears northeast of the northern basin.As den- itrification, the dominant nitrate removal process generally occurs in the water column, where DO concentration ranges from 0.7 to 20 µmol kg −1 (Paulmier et al., 2009).The intensity of denitrification was reported to descend gradually, corresponding to the DO spatial pattern from the northern to the southern parts of the AS, and did not become obvious at 11 or 12 • N (Naqvi et al., 1982).As indicated by the upper 2000 m N-S transect of DO (Fig. 1c), a southwardly decrease in ODZ thickness can be observed and the contour line of 5 µmol kg −1 extends to around 13 • N. Since the nitrate source is mainly from the bottom of the euphotic zone at around 150 m, we postulate a geographically distinctive sedimentary δ 15 N bulk underneath ODZs.Thus, an isopleth of 25 µmol kg −1 DO at 150 m is applied as a geographic boundary to separate the northern from the southern part of the AS basin.The interface where DO concentration changed from 20 to 30 µmol kg −1 was such a transition zone.On the other hand, the bottom layer of the ODZ moves shallower toward the south, as shown previously by Gouretski and Koltermann (2004).Accordingly, the bottom oxygen content may also be a factor to influence the degree of alteration in sedimentary δ 15 N bulk .
As mentioned in the introduction, nitrate is removed via denitrification in ODZs resulting in excess P to stimulate N 2 fixation.In Figs.1d, e and f, we presented the N-S transect of nitrate and N * (for both the upper 2000 m and 300 m) in January.Even though there is nitrate in the surfacewater (Fig. 1d), as mentioned earlier, near-surface NO − 3 in the AS is completely utilized in an annual cycle (Altabet, 1988;Thunell et al., 2004).Furthermore, negative N * (P excess) throughout the water column represents a nitrate deficit, and the lowest N * value appears at ∼ 300 m at 18-20 • N, where DO is < 1 µmol kg −1 .Meanwhile, a gradual southwardly increase in N * can be observed for upper 100 m (Fig. 1f) and the isopleth of N * of −4 deepens southward with the highest N * (−2) appearing at ∼ 10-12 • N. The volume expansion of high N * water, as well as a simultaneous increase in N * , strongly indicate an addition of bioavailable nitrogen when surface water is traveling southward.

Material and method
A sediment gravity core, SK177/11 (8.2 • N and 76.47 • E), was collected at water depths of 776 m on the continental slope off the southwest coast of India (Kerala) during the 177th cruise of ORV SagarKanya in October 2002.Although the core MD77-191 locates further south in the AS (Bassinot et al., 2012), SK177/11 is, so far, the southernmost core with reference to δ 15 N record.The 3.65 m long core was subsampled at interval of 2 cm for upper 1 m and of 5 cm for the rest (open circles in Fig. 2a).There is a distinct boundary at ∼ 1.7 m, above which the core consists mainly of brownish gray clayey sediments.Neither distinct laminations nor tur- bidities can be observed by visual contact immediately after collection or at the time during sub-sampling (Pandarinath et al., 2007).All sub-samples were freeze-dried and ground into powder in an agate mortar with pestle.Sand was almost absent (< 1 wt %) throughout the core.
The calendar chronology for core SK177/11 was based on seven accelerator mass spectrometry (AMS) radiocarbon ( 14 C) dates of bulk organic matter (Fig. 2a).Calendar years were calculated using calibration CALIB 6.0 with a reservoir age correction of 402 years (Stuiver et al., 1998;Reimer et al., 2009).Details on the 14 C age controlling points were presented in Table 1.Given that the AMS 14 C dates of SK177/11 were obtained on total organic carbon (TOC), we may not be able to avoid the mixture of organics of different ages during transport (Mollenhauer et al., 2005) or interference by preaged organics sourced from land (Kao et al., 2008).However, besides the reservoir age correction, due to higher TOC content (range: 2.2-5.5 %) of sediments and their marinesourced organic carbon, as confirmed by stable C isotope data and C / N ratio, shown in Figs.3b and e, we are confident that our age model is reliable and less likely affected by age heterogeneity.
Bulk sedimentary nitrogen content and δ 15 N analyses were carried out using a Carlo-Erba EA 2100 elemental analyzer connected to a Thermo Finnigan Delta V Advantage isotope ratio mass spectrometer (EA-IRMS).Sediments for TOC analyses were acid-treated with 1N HCl for 16 h, and then centrifuged to remove carbonate.The acid-treated sediments were further dried at 60 • C for TOC content and δ 13 C.The nitrogen isotopic compositions of acidified samples were obtained at the same time for comparison.Carbon and nitrogen isotopic data were presented by standard δ notation with respect to PDB (Pee Dee Belemnite) carbon and atmospheric nitrogen.USGS 40, which has certified δ 13 C of −26.24 and δ 15 N of −4.52 ‰ and acetanilide (Merck) with δ 13 C of −29.76 and δ 15 N of −1.52 ‰ were used as working standards.The reproducibility of carbon and nitrogen isotopic measurements is better than 0.15 ‰ .The precision of nitrogen and carbon content measurements were better than 0.02 and 0.05 %, respectively.Meanwhile, the acidified and non-acidified samples exhibited identical patterns in δ 15 N (not shown) with mean deviation of 0.3 ‰.

Sedimentation rate
The age-depth curve was shown in Fig. 2a, in which age dates were evenly distributed throughout the core, although not at a high resolution.In Mollenhauer et al. (2005), the largest age offset between total organic carbon and cooccurring foraminifera is ∼ 3000 years and mostly < 2000 years.Meanwhile, the offset remains more or less constant throughout past 20 ka, regardless of the deglacial transition.The youngest date in our core is 3180 cal ka BP at 58 cm.We may expect younger age on the surface.Thus, if our TOC samples contain any pre-aged organics, as indicated by Mollenhauer et al. (2005), the offset should not be too large to alter our interpretation for the comparison between glacial and Holocene periods.The linear sedimentation rates derived from seven date intervals range from 6 to 20 cm ka −1 (Fig. 2b), with relatively constant value (∼ 6 cm ka −1 ) prior to Holocene, except for the excursion around the last glacial maximum.The linear sedimentation rates started to increase since Holocene and reached 18-20 cm ka −1 when the sea level reached modern-day level.

Nitrogen and carbon contents and their isotopes
Values of δ 15 N bulk ranged from 4.7 to 7.1 ‰ with significantly lower values during the glacial period (Fig. 3a).The δ 15 N values increased rapidly since ∼ 19 ka BP, with a peak at ∼ 15 ka BP, and then started to decrease gradually toward the modern day, except for the low δ 15 N excursion at ∼ 14 ka BP. Figure 3b shows that values of δ 13 C TOC (−21.5 to −18.5 ‰) were consistent with the δ 13 C of typical marine organic matter end member (−22 to −18 ‰; Meyers, 1997).An abrupt decrease in δ 13 C was observed in concert with the dramatic increase in δ 15 N bulk at the start of deglaciation.
Bulk nitrogen content (TN) had a range of 0.23-0.75% (Fig. 3c), and the TOC content ranged from 2.2 to 5.5 % (Fig. 3d).Both TN and TOC showed similar trends over the last 35 ka BP with relatively constant values prior to Holocene and an afterward elevation until the modern day.The upward increasing TOC and TN patterns since Holocene were consistent with the increasing pattern of the sedimentation rate, suggesting a higher organic burial flux induced by enhanced productivity, which had been reported elsewhere in the AS (Altabet et al., 2002).
As for the TOC / TN ratio, higher values appeared during the deglacial transition and the glacial period (Fig. 3e).The highest value coincides with the δ 13 C TOC drop, implying that there is still some influence from terrestrial organics.However, terrestrial organics contain less nitrogen (C / N of 20; Meyers, 1997), thus the δ 15 N did not drop correspondingly.In the first meter (since ∼ 5 ka), the downward decreasing pattern of TOC and TN can also be attributed to syn-sedimentary degradation; if so, a downward increasing in TOC / TN should be evident.However, TOC / TN values varied in a narrow range not revealing a significant increasing trend.Nevertheless, δ 15 N and δ 13 C did not show concomitant variations with C / N in the first meter or throughout the core.The influence from both organic degradation and changes in terrestrial organic input on isotopic signals is thus limited.
Figure 4 shows the scatter plot of TOC against TN.The slope of the linear regression line for TOC against TN (TOC = (6.67 ± 0.22) × TN + (0.99 ± 0.11), R 2 = 0.94, n = 57, p < 0.0001) is 6.67, again indicating that organic matter is mainly marine-sourced.Although this slope is slightly higher than the Redfield ratio of 5.68 (wt / wt), it is lower than that observed on the East China Sea shelf (7.46; Kao et al., 2003).Meanwhile, the intercept of TN is negative when TOC downs to zero, implying that inorganic nitrogen can be ignored in our core.Obviously, if we force the regression through the origin point, TOC / TN values for samples during the Holocene will have the lower ratios reflecting even less contribution from terrestrial organics.

Downward transfer and transformation of N isotopic signal
As mentioned, the signal of sedimentary δ 15 N may be altered under different burial conditions.Altabet and Francois (1994) reported little diagenetic alteration of the nearsurface δ 15 N in the equatorial Pacific, while there was an apparent +5 ‰ enrichment relative to sinking particles in the Southern Ocean, south of the polar front.In the Sargasso Sea, sedimentary δ 15 N also enriched by 3-6 ‰ relative to sinking particles (Altabet et al., 2002;Gruber and Galloway, 2008).The degree of alteration was attributed to particle sinking rate and organic matter (OM) preservation (Altabet, 1988).Gaye-Haake et al. (2005) also suggested that low sedimentation rates benefit organic matter decomposition, resulting in a positive shift in bulk sedimentary δ 15 N comparing to sinking particles in the South China Sea.Finally, Robinson et al. (2012) concluded that oxygen exposure time at the seafloor is the dominant factor controlling the extent of N isotopic alteration.Thus, it is necessary to follow the track of δ 15 N signal to clarify the occurrence of deviation during transfer.
The reported depth profiles of δ 15 N NO 3 in the AS were shown in Fig. 5, in which δ 15 N NO 3 values of water depth deeper than 1200 m range narrowly around 6-7 ‰, which is slightly higher than the global average of the deep oceans ((4.8 ± 0.2) ‰ for > 2500 m, Sigman et al., 2000; (5.7 ± 0.7) ‰ for > 1500 m, Liu and Kaplan, 1989).Below the euphotic layer, δ 15 N NO 3 increases, rapidly peaking at around 200-400 m.The preferential removal of 14 NO 3 by water column denitrification accounts for these subsurface δ 15 N NO 3 highs (Brandes et al., 1998;Altabet et al., 1999;Naqvi et al., 2006).The subsurface δ 15 N NO 3 maximum ranges from 10 to 18 ‰ for different stations, implying a great spatial heterogeneity in water columns denitrification intensity.It is worth mentioning that higher values, in general, appear in the northeastern AS (15 ∼ 18 ‰) (Fig. 5), highlighting that the focal area of water column denitrifi-  Brandes et al., 1998;Altabet et al., 1999;Naqvi et al., 2006).
cation is prone to the northeastern Arabian Sea (Naqvi et al., 1994;Pichevin et al., 2007), also revealed by the DO spatial distribution (Fig. 1a).Contrary to higher denitrification in the northeastern AS, the export production is always higher in the northwestern AS throughout a year (Rixen et al., 1996).Such decoupling between productivity and denitrification was attributed to the oxygen supply by intermediate water exchange besides primary productivity oxygen demand (Pichevin et al., 2007).Note that the δ 15 N NO 3 values at a water depth of 100-150 m, which correspond to the bottom depth of the euphotic zone (Olson et al., 1993), from different stations fall within a narrow range of 7-9 ‰ despite wide denitrification intensity underneath.The rapid addition of new nitrogen, as mentioned earlier, might account for the relatively uniform δ 15 N NO 3 at the bottom of the euphotic layer.Unfortunately, there are no δ 15 N NO 3 profiles or sediment trap data from the southern basin for comparison.
Interestingly, reported δ 15 N of sinking particles (δ 15 N SP ) collected by five sedimentation traps deployed from 500 m throughout a depth of 3200 m ranged narrowly from 5.1 to 8.5 ‰ (Fig. 6), which is slightly lower but overlaps largely with δ 15 N NO 3 values at 100-150 m.Such similarity in δ 15 N NO 3 at 100-150 m and sinking particles strongly indicated that (1) the NO − 3 source for sinking particles was coming from a depth of around 100-150 m, instead of the ODZs (200-400 m) where the maximum δ 15 N NO 3 value occurred (Schäfer and Ittekkot, 1993;Altabet et al., 1999), and (2) little alteration had occurred in δ 15 N SP during sinking in the water column, as indicated by Altabet (2006).Only these five trap stations with nitrogen isotope information were available in the AS (Gaye-Haaake et al., 2005).The trap locations were in the same area but not as far south compared to the δ 15 N NO 3 stations (insert map in Fig. 6).The slightly lower δ 15 N in sinking particles is attributable to their geographic locations (see below), since incomplete relative utilization of surface nitrate has been documented to have a very limited imprint on the δ 15 N signal in the AS (e.g., Schäfer and Ittekkot, 1993).
The uniformly low values of δ 15 N NO 3 at the bottom of the euphotic zone should be a consequence resulting from various processes in the euphotic zone, such as remineralization, nitrification and N 2 fixation.Nevertheless, the distribution pattern of N * (Figs.1e and f) illustrates that there must be an addition of 14 NO 3 into the system to cancel out the isotopic enrichment caused by denitrification.Note that the positive offset in δ 15 N NO 3 ( δ 15 N NO 3 , 6∼ 12 ‰) in ODZs caused by various degrees of denitrification was narrowed down significantly, while nitrate was transported upward.This implies that a certain degree of addition processes, most likely the N 2 fixation, varied in concert with the intensity of denitrification underneath.Since the upwelling zones distribute at the very north and the west of the AS and the upwelled water travels southward (or outward) on the surface, as shown in Fig. 1e, it is reasonable to see the phenomenon of denitrificationinduced N 2 fixation to compensate the nitrogen deficiency.Consistent to this notion, Deutsch et al. (2007) discovered the spatial coupling between denitrification in eastern tropical Pacific (upstream) and N 2 fixation in western equatorial Pacific (downstream).Such a horizontal nitrogen addition process can also be seen clearly in our background information of N * (Fig. 1f).In fact, fixed N had been proved to account for a significant part of surface nitrate in the modern-day AS, where denitrification is exceptionally intense (Brandes et al., 1998;Capone et al., 1998;Parab et al., 2012).
Compared with reported δ 15 N of surface sediments retrieved from trap locations, a significant positive shift in δ 15 N can be seen at the seafloor (Fig. 6).Such a positive deviation can be seen elsewhere in previous reports (Altabet, 1988;Brummer et al., 2002;Kienast et al., 2005) due to prolonged oxygen exposure after deposition (Robinsson et al., 2012) associated with sedimentation rate (Pichevin et al., 2007).Although Cowie et al. (2009) found an ambiguous relation between contents of sedimentary organic carbon and oxygen in deep water, they also noticed the appearance of maximum organic carbon contents at the lower boundary of ODZs, where oxygen content was relatively higher.Accordingly, they believed that other factors controlling the preservation of organic carbon existed, such as the chemical characteristics of organic matter, the interaction between organic matters and minerals, the enrichment and activity of benthic organism or the physical factor, including the screening and water dynamic effect.

Geographically distinctive bottom-depth effects in the modern day
As classified by oxygen content of 25 µmol kg −1 at 150 m, the documented surface sedimentary δ 15 N bulk (Gaye-Haake et al., 2005) was separated into northern and southern groups to examine the geographic difference in bottom-depth effect.Both groups exhibit positive linear relationships between δ 15 N bulk and bottom depth (deeper than 200 m) (Fig. 7a).The regression equations were shown in Table 2. Interestingly, the regressions generally differ statistically from each other in terms of slope and intercept.The slope represents the degree of positive shift of sedimentary δ 15 N due to bottom-depth effect.For the southern AS, the slope is (0.76 (±0.14) × 10 −3 km −1 ), which is close to the correction factor (0.75 × 10 −3 km −1 ) for the world ocean, proposed by Robinson et al. (2012) and further applied by Galbraith et al. (2012).By contrast, the slope for the northern AS is significantly lower (0.55 (±0.08) × 10 −3 km −1 ), implying that the depth-associated alteration in the northern AS is smaller.
The correction factor for bottom-depth effect was predicted to vary in different regions such as that in the South China Sea (Gaye et al., 2009).Since the magnitude of oxygen exposure is the primary control of depth effect (Gaye-Haake et al., 2005;Mobius et al., 2011;Robinson et al., 2012), we attributed this lower slope in the northern AS to relatively higher sedimentation rates (not shown) and lower www.biogeosciences.net/12/1/2015/Biogeosciences, 12, 1-14, 2015  oxygen contents, as indicated by previous research (Olson et al., 1993;Morrison et al., 1999;Brummer et al., 2002).
On the other hand, the intercept for the northern AS regression (8.1 ± 0.2) is significantly higher than that for the southern AS (6.0 ± 0.3).As mentioned above, δ 15 N values of sinking particle resembled the δ 15 N of nitrate sourced from a depth of 100-150 m.According to the depth-dependent correction factor, we may convert sedimentary δ 15 N bulk values at various water depths into their initial condition when the digenetic alteration is minimal to represent the δ 15 N of source nitrate.Higher intercept suggests that a stronger denitrification had occurred in northern AS surface sediments.The 2.1 ‰ lower intercept in the southern AS likely reflects the addition of N 2 fixation in the upper water column while it travels southward.The progressive increase of N * toward the southern AS supports our speculation, although no δ 15 N NO 3 profiles had been published in the southern basin.Future works about δ 15 N NO 3 and δ 15 N SP in the southern AS are needed.
In Fig. 7b, we presented corrected δ 15 N bulk values along with bottom depth for the northern and southern AS surface sediments for comparison.After removing site-specific bias caused by bottom-depth effect, the values and distribution ranges of δ 15 N bulk for both the northern and southern AS became smaller and narrower.For the northern AS, the distribution pattern skewed negatively, giving a standard deviation of 0.88 ‰, falling exactly in the range of 7-9 ‰ for δ 15 N NO 3 (7-9 ‰) at the bottom of the euphotic zone.As a result, the corrected nitrogen isotopic signals in sediments more truthfully represent the δ 15 N NO 3 value at the bottom depth of the euphotic zone.Meanwhile, the statistically significant difference in δ 15 N bulk distribution between the northern and southern AS further confirms the feasibility of our classification by using DO isopleth of 25 µmol kg −1 at 150 m.

Bottom-depth effect during different climate stages
In order to better decipher the history of δ 15 N NO 3 in the bottom the euphotic zone of the water column, we synthesized almost all available δ 15 N bulk of sediment cores reported for the AS (see Figs. 1a and 1b for locations).Similar to modern surface sediments, northern and southern groups were  35 ka (Fig. 8a).Unfortunately, data points were less in 0-6 ka and there were only three sediment cores in the southern AS: SK177/11 in this study, and NIOP 905 and SO42-74KL in previous studies.
As shown in Fig. 8a, the original δ 15 N bulk from the northern (gray dots) and southern AS (green, blue and red curves) scatter in a wide range from 4.5 to 10.5 ‰ over the entire 35 ka.The pink dots are for the data from core MD-04-2876, which is peculiar since the relatively low δ 15 N bulk values deviated from all other reports in the northern AS.Pichevin et al. (2007) excluded the influences from incomplete nitrate utilization and terrestrial input, thus we still include this core in our statistical analyses.As for the southern cores, the temporal variations of δ 15 N bulk in core SK177/11 and NIOP 905 (red and blue) had a very similar trend distributing at the lower bound of the whole data set.The mean δ 15 N bulk values for SK177/11 and NIOP 905 during the glacial period were almost identical, and the deviation in the Holocene was as small as 0.7 ‰.By contrast, the temporal pattern for δ 15 N bulk of core SO42-74KL (green) resembles that of NIOP 905, yet with an enrichment in 15 N by ∼ 2 ‰ for the entire period.The core SO42-74KL is retrieved from a depth of 3212 m, which is the deepest of the three cores in the southern AS; the positive offset is apparently caused by the bottom-depth effect.Thus, inference should be made with caution when compare sediment cores from different depths.
Below we consider two time spans -0 ∼ 11 ka (Holocene) and 19∼ 35 ka (glacial) -to examine the bottom-depth effect at different climate stages.We ignore the transgression period, which is shorter with more variable in δ 15 N bulk , to avoid bias caused by dating uncertainties in different studies.Also, we will discuss the peculiar patters for 0-6 ka later.The mean and standard deviation of reported δ 15 N bulk values for the specific time span were plotted against the corresponding depth of the core.Accordingly, we obtained the correction factors for glacial and early Holocene, respectively, for the northern and southern AS (Fig. 8b and c).Since only 35 ka was applied in this practice, the long-term alteration (Rewww.biogeosciences.net/12/1/2015/Biogeosciences, 12, 1-14, 2015 ichart et al., 1998;Altabet et al., 1999) is ignored.The regression curves for the modern day (dashed lines) were plotted for comparison.
The difference among regressions of three climate stages in the northern AS (Table 2) is not significant (0.55 × 10 −3 km −1 to 0.70 × 10 −3 km −1 ).However, the regression slopes for the northern AS are significantly lower compared with those obtained from the southern AS for all climate states.This might indicate that the oxygen content in the northern AS is always lower, resulting in a lower degree of alteration of δ 15 N bulk .On the other hand, we may not exclude the effect by sedimentation rate changes over these two stages, which also affect the oxygen exposure time; unfortunately, insufficient sedimentation rate data in the northern AS in previous reports prevent us from implementing further analysis.
As for the southern AS, correction factors are always higher than those in the northern AS.The overall spatialtemporal patterns are consistent with the oxygen distribution in the Arabian Sea (Olson et al., 1993;Morrison et al., 1999;Pichevin et al., 2007) and agree with the view that DO concentration was the dominant factor for organic matter preservation (Aller, 2001;Zonneveld et al., 2010).Meanwhile, the regression slopes remained high from 0.76 × 10 −3 to 1.01 × 10 −3 km −1 over different climate stages in the southern AS, suggesting that environmental situations, and thus those correction factor, change less relatively to that in the northern AS.For SK177/11, the sedimentation rate in Holocene is two-fold higher compared to that in the glacial period.However, the influence caused by the sedimentation rate changes is likely not significant enough to alter the regression slopes for the southern AS, based on the small changes in the slope (0.93 × 10 −3 and 1.01 × 10 −3 km −1 ).

Insights from temporal changes in geographic
δ 15 N bulk distribution Based on the earlier comparison among δ 15 N NO 3 , sinking particles and surface sediments, we recognized that the regression intercept is representative of the nitrogen isotope of nitrate source at a depth of 100 m.Therefore, the regressionderived intercepts given in Table 2 can be used to infer the δ 15 N NO 3 source at different climate stages, while the slopes can be used as correction factors to eliminate the positive shift in δ 15 N bulk caused by bottom depth; by doing this, we can get the original signal of δ 15 N bulk prior to alteration.We applied the correction factor to be equal to (bottom depth −100 m) × slope, ignoring the sea level changes during the different climate stages.Noticeably, the regression intercepts for both the northern and southern AS are higher in the Holocene compared to those in the glacial period, indicating the intensified isotopic enrichment in δ 15 N NO 3 in the entire AS in Holocene.Such increment is almost the same to be ∼ 1.5 ‰, which is similar to the increase in the eastern tropical North Pacific, but slightly smaller than that in the eastern tropical South Pacific (Galbraith et al., 2012).The 120 m sea level increase, which may induce only a 0.1 ‰ offset, cannot be the reason for such a significant increase of average δ 15 N bulk during the Holocene.Moreover, deviations between the northern and southern AS at the respective climate stage are almost identical (1.0 ‰ for Holocene and 0.9 ‰ for glacial), indicating a synchronous shift in the relative intensity of denitrification and N 2 fixation over the basin to keep such a constant latitudinal gradient of subsurface δ 15 N NO 3 .
The intermediate water formation near the polar region controls the oxygen supply to the intermediate water and thus the extent of denitrification on global scale and the stoichiometry of nutrient source to the euphotic zone (Galbraith et al., 2004).Lower glacial-stage sea surface temperature may increase oxygen solubility, while stronger winds in high-latitude regions enhance the rate of thermocline ventilation.The resultant colder and rapidly flushed thermocline thus lessened the spatial extent of denitrification and, consequently, N fixation (Galbraith et al., 2004).Therefore, such a basin of wide synchronous increase in δ 15 N bulk is likely a global control.The lower intercepts in glacial time (4.3 ‰ for the south and 5.3 ‰ for the north), which are similar to the global mean δ 15 N NO 3 (4.5-5‰, Sigman et al., 1997), illustrate a better ventilation of intermediate water during glacial time in the Arabian Sea (Pichevin et al., 2007).In fact, the AAIWs penetrate further northward over 5 • N in the present day and even during the late Holocene (You, 1998;Pichevin et al., 2007).Since the δ 13 C of autochthonous particulate organic carbon is negatively correlated to [CO 2 (aq)] in the euphotic zone (Rau et al., 1991), the sharp decrease of δ 13 C TOC in SK177/11 at the start of deglaciation (Fig. 3b) may infer the timing of a rapid accumulation of dissolved inorganic carbon driven by the shrinking of oxygenated intermediate water (Pichevin et al., 2007) or enhanced monsoon-driven upwelling (Ganeshram et al., 2000); both facilitate the promotion of denitrification.Nevertheless, the mirror image between δ 15 N and δ 13 C TOC pro?les revealed their intimate relation, of which the variability was attributable to the change of physical processes.
The intercepts of the northern AS increase continuously from 5.2 to 8.1 from glacial through to modern day, indicating the strengthened intensity of denitrification relative to nitrogen fixation in the northern AS (Altabet, 2007).When we take a close look at the temporal pattern of corrected δ 15 N bulk for long cores (Fig. 9), we can see an amplified deviation since 6 ka, during which δ 15 N bulk increases continuously in the northern AS, whereas it decreases in the southern AS.Note that the northern most core, MD-04-2876, also followed the increasing trend in recent 6 ka even though its δ 15 N bulk values deviated from all other cores.Such opposite trends indicate that the controlling factors on the nitrogen cycle in the northern AS were different from that in the southern AS, which means that localized enhancement in specific process had occurred.Besides the oxygen supply to the intermediate water, the intensity of water column denitrification varies with primary productivity (Altabet, 2006;Naqvi et al., 2006).Strong summer monsoon and winter monsoon drive upwelling or convective mixing enhances primary productivity, which, in turn, intensifies denitrification (Altabet et al., 2002;Ganeshram et al., 2002).However, it was also reported that primary productivity did not correlate well with water column denitrification underneath during the Holocene in some parts of the northern AS (Banakar et al., 2005, and references therein).Regardless of the declining summer monsoon strength since 5500 ka (Hong et al., 2003), primary productivity in the northern AS seemed to have increased.Similar to the patterns observed for TOC and TN in this study, productivity indicators (TOC and Ba / Al ratios), reported by Rao et al. (2010) in the core SK148/4 located near our SK177/11, also increased gradually since the Holocene.Incomplete nitrate consumption can hardly explain the decreasing pattern for all three cores in the southern AS, where upwelling intensity is much less relative to that in the north.Moreover, lower TOC / TN ratios observed in Holocene in SK177/11, as mentioned earlier, rule out the influence of terrestrial organic input.Therefore, a spatial coupling of denitrificationdependent N 2 fixation is the more plausible cause of the decreasing δ 15 N bulk pattern (Deutsch et al., 2007).
We suggested that the intensified supply of excess phosphorous (phosphorus in stoichiometric excess of fixed nitrogen) toward the southern AS to stimulate N 2 fixation, subsequently responsible for the decreasing δ 15 N bulk pattern in the southern basin.The intensification in excess phosphorous supply can be driven by enhanced upwelling or intensified subsurface water column denitrification or both.According to the increasing pattern in δ 15 N bulk and primary productivity in the northern AS, synergetic processes are suggested.The upwelled water in the northern AS basin brings up low N / P water to the surface for non-diazotrophs to uptake.If we assume complete consumption, the remaining excess phosphorous after complete consumption will be transported toward the south by clockwise surface circulation and advection.Therefore, N 2 fixation in the southern AS acts as feedback to balance denitrification changes in the northern AS.This phenomenon is similar to the illustration for the spatial coupling of nitrogen inputs and losses in the Pacific Ocean, proposed by Deutsch et al. (2007).The question as to why such forcing to expand the N-S deviation had not occurred before 6 ka warrants more studies.

Conclusions
The available data showed that values of δ 15 N NO 3 at the bottom of the euphotic zone (∼ 150 m) were similar to δ 15 N SP , implying that the source of nutrients for sinking particulate organic matter was largely derived from a depth of around 150 m.Values of sedimentary δ 15 N bulk were obviously higher than δ 15 N SP in surrounding areas, suggesting that such a shift of sedimentary δ 15 N bulk occurred after deposition.It is necessary to remove site-specific bias of δ 15 N bulk values caused by bottom depth to retrieve the original signal before alteration.As a result, the corrected nitrogen isotopic signal in sediments could be representative of the value of δ 15 N NO 3 at the bottom depth of the euphotic zone.The bottom-depth effects in the northern AS vary during different climate stages, but the variation is always lower than such effects in the southern AS in general.The modern surface δ 15 N bulk values can be separated statistically into northern and southern AS groups, reflecting a special coupling of denitrification to the north, and N 2 fixation to the south.This phenomenon is supported by the reported modern-day N * distribution.As for historical records, the offset in δ 15 N bulk between the southern and northern AS remained relatively constant (1.0 ‰ for early Holocene and 0.9 ‰ for glacial) prior to 6 ka, indicating a synchronous shift in the relative intensity of denitrification and N 2 fixation over the basin to keep such a constant latitudinal gradient of subsurface δ 15 N NO 3 .However, this offset expanded gradually since 6 ka, likely due to more localized intensifications in denitrification and N 2 fixation had occurred, respectively, in the northern and southern Arabian Seas.The spatial coupling of nitrogen inputs and losses in the Arabian Sea was proposed, yet the www.biogeosciences.net/12/1/2015/Biogeosciences, 12, 1-14, 2015 question as to why the driving force did not expand the N-S deviation before 6 ka warrants more studies.

Figure 2 .
Figure 2. (a) Plot of calendar age against depth; (b) Linear sedimentation rate (indicates the 14 C age controlling points).

Figure 3 .
Figure 3. Temporal variations of (a) stable isotopic compositions of bulk nitrogen (δ 15 N); (b) stable isotopic compositions of total organic carbon (TOC) (δ 13 C); (c) contents of total nitrogen; (d) total organic carbon; (e) TOC / TN ratio.Horizontal dashed lines are references for low value periods.

Figure 4 .
Figure 4. Scatter plot of the total organic carbon content against total nitrogen.Redfield field ratio of 5.68 is shown in line.Bold dashed line stands for regression.Red, purple, green and blue dots represent the late Holocene, early Holocene, deglacial and glacial periods, respectively.

Figure 5 .
Figure 5. Depth profiles of nitrogen isotope of nitrate (δ 15 N NO 3 ) in water columns; data not marked are all from August; the location of Station January 1995 overlaps with Station SS3201 (data digitized fromBrandes et al., 1998;Altabet et al., 1999;Naqvi et al., 2006).

Figure 6 .
Figure 6.Vertical profiles for nitrogen isotope of nitrate (crosses in inserted map), sinking particles (inverse triangles in map) and trap-corresponding surface sediments.Data for sediment traps and surface sediments are from Gaye-Haake et al. (2005).Depth profile of δ 15 N NO 3 follows that in Fig. 5.

Figure 7 .
Figure 7. (a) Non-corrected δ 15 N values of modern surface sediments against corresponding bottom depth in the northern and southern Arabian Sea (see text for N-S boundary).Regression lines were shown in dashed and solid lines, respectively, for the northern and southern AS.(b) Corrected surface sedimentary δ 15 N values against water depth.

Figure 8 .
Figure 8.(a) Temporal variations of non-corrected δ 15 N bulk values of all reported cores in the AS.Data shown in curves are for cores in the southern Arabian Sea (red for SK177/11, blue for NIOP 905 and green for SO42-74KL), dots in gray are for the northern part and pink dots are for core MD-04-2876.Mean values of δ 15 N for fixed periods against corresponding water depths for cores in the (b) northern and (c) southern Arabian Sea.Pink and indigo blue are for the Holocene and glacial periods, respectively.Error bars represent the standard deviation for mean δ 15 N bulk .The dashed regression lines for modern surface sediments are shown for reference.

Figure 9 .
Figure 9. Temporal variations of corrected δ 15 N bulk values of all reported cores in the AS.Gray and black dots are for the northern and southern AS, respectively.Pink dots are specifically for core MD-04-2876.The deglacial period is in shadow because non proper equations for bottom-depth effect correction.The upper panel is the blow-up for the Holocene period.The intensified deviation trends since 6 ka were marked by bold dashed lines.

Table 2 .
Linear equations of bottom-depth effect during different climate stages.