Large freshwater-influx-induced salinity gradient and diagenetic changes in the northern Indian Ocean dominate the stable oxygen isotopic variation in Globigerinoides ruber

The application of stable oxygen isotopic ratio of surface-dwelling planktic foraminifera Globigerinoides ruber (white variety; δ¹⁸O_ruber) to reconstruct past hydrological changes requires a precise understanding of the effect of ambient parameters on δ¹⁸O_ruber. The northern Indian Ocean, with its huge freshwater influx and being a part of the Indo-Pacific Warm Pool, provides a unique setting to understand the effect of both the freshwater-influx-induced salinity and temperature on δ¹⁸O_ruber. Here, we use a total of 400 surface samples (252 from this work and 148 from previous studies), covering the entire salinity end-member region, to assess the effect of freshwater-influx-induced seawater salinity and temperature on δ¹⁸O_ruber in the northern Indian Ocean. The analysed surface δ¹⁸O_ruber mimics the expected δ¹⁸O calcite estimated from the modern seawater parameters (temperature, salinity, and seawater δ¹⁸O) very well. We report a large diagenetic overprinting of δ¹⁸O_ruber in the surface sediments, with an increase of 0.18 ‰ per kilometre increase in water depth. The freshwater-influx-induced salinity exerts the major control on δ¹⁸O_ruber (R²=0.63) in the northern Indian Ocean, with an increase of 0.29 ‰ per unit increase in salinity. The relationship between temperature- and salinity-corrected δ¹⁸O_ruber (δ¹⁸O_ruber−δ¹⁸O_sw) in the northern Indian Ocean [$T=-\mathrm{0.59}\cdot ({\mathit{\delta }}^{\mathrm{18}}{\mathrm{O}}_{\mathrm{ruber}}-{\mathit{\delta }}^{\mathrm{18}}{\mathrm{O}}_{\mathrm{sw}})+\mathrm{26.40}$] is different than reported previously, based on the global compilation of plankton tow δ¹⁸O_ruber data. The revised equations will help create a better palaeoclimatic reconstruction from the northern Indian Ocean by using the stable oxygen isotopic ratio. The entire data set (newly generated and previously published) used in this work is available both as a Supplement to this article and at PANGAEA (https://doi.org/10.1594/PANGAEA.945401; Saraswat et al., 2022).

1 Introduction

The stable oxygen isotopic ratio (δ¹⁸O) of biogenic carbonates is one of the most extensively used marine palaeoclimatic proxies (Mulitza et al., 1997; Lea, 2014; Metcalfe et al., 2019; Saraswat et al., 2019). Even though it was initially suggested that the oxygen isotopic fractionation in biogenic carbonates is largely driven by temperature (Urey, 1947), subsequent work revealed that, besides temperature, the salinity and carbonate ion concentration of ambient seawater also affects the biogenic carbonate δ¹⁸O (Vergnaud-Grazzini, 1976; Spero et al., 1997; Bemis et al., 1998; Spero et al., 1997; Bijma et al., 1999; Mulitza et al., 2003). On longer timescales, the global ice volume contribution to the glacial–interglacial shift in marine biogenic carbonate δ¹⁸O is the largest (∼1.0 ‰–1.2 ‰) at a majority of the locations (Shackleton, 1987, 2000; Lambeck et al., 2014). The ice volume changes have induced well-defined globally similar shifts in biogenic carbonate δ¹⁸O during the last several million years. Therefore, the regional evaporation–precipitation, runoff, and temperature changes are reconstructed from the global ice-volume-corrected biogenic carbonate δ¹⁸O (Wang et al., 1995; Kallel et al., 1997; Schmidt et al., 2004; Saraswat et al., 2012, 2013; Kessarkar et al., 2013).

The δ¹⁸O of surface-dwelling planktic foraminifera Globigerinoides ruber (δ¹⁸O_ruber) is often used to reconstruct past surface seawater conditions (Pearson, 2012; Saraswat et al., 2012, 2013; Mahesh and Banakar, 2014). Therefore, continuous efforts are made to understand the factors affecting δ¹⁸O_ruber (Vergnaud-Grazzini, 1976; Mulitza et al., 1997, 1998, 2003; Waelbroeck et al., 2005; Mohtadi et al., 2011; Horikawa et al., 2015; Hollstein et al., 2017; Sánchez et al., 2022). The depth habitat of G. ruber in the tropical Atlantic Ocean has been inferred from its stable oxygen isotopic ratio (Farmer et al., 2007). The change in the stable oxygen isotopic ratio of planktic foraminifera, including G. ruber, is suggested as a proxy to reconstruct the upper water column stratification in the tropical Atlantic Ocean, based on the good correlation between δ¹⁸O and the ambient seawater characteristics (Steph et al., 2009). A few studies suggested a difference in the δ¹⁸O of the various morphotypes of G. ruber (sensu stricto and sensu lato) and attributed it to their distinct ecology and depth habitat (Löwemark et al., 2005). However, a recent study from the Gulf of Mexico suggested a similar ecology and depth habitat for both the G. ruber morphotypes (Thirumalai et al., 2014). The northern Indian Ocean, being influenced by huge freshwater influx and being part of the Indo-Pacific Warm Pool (De Deckker, 2016), provides a unique setting to understand the effect of large salinity and temperature changes on δ¹⁸O_ruber. Earlier, Duplessy et al. (1981a) measured the δ¹⁸O of the living G. ruber specimens collected from the water column and the dead ones recovered from surface sediments of the northern Indian Ocean. A similar study from the Red Sea and adjoining western Arabian Sea suggested that G. ruber calcifies its test in isotopic equilibrium with the ambient seawater, thus tracking the interannual subtle change in the salinity and temperature (Kroon and Ganssen, 1989; Ganssen and Kroon, 1991).

The temperature influence on δ¹⁸O_ruber is well defined (Shackleton and Vincent, 1978; Multiza et al., 1998, 2003). The effect of freshwater-influx-induced changes in ambient salinity on δ¹⁸O_ruber is, however, debated (Dämmer et al., 2020). With the extensive use of δ¹⁸O_ruber to reconstruct regional evaporation–precipitation changes, especially from the monsoon-dominated tropical oceans, it is imperative to understand the precise influence of ambient salinity on δ¹⁸O_ruber. The ambient seawater pH, carbonate ion concentration (Bijma et al., 1999), and presence/absence of symbionts (Jørgensen et al., 1985) also affect the isotopic composition of G. ruber. However, a limited glacial–interglacial variability in these parameters is masked by the dominance of temperature and freshwater-influx-induced salinity changes in the oxygen isotopic ratio of G. ruber. Additionally, the diagenetic changes, especially dissolution, also substantially alter the original isotopic composition of the foraminifera shells (Berger and Killingley, 1977; Wu and Berger, 1989; Lohmann, 1995; McCorkle et al., 1997; Wycech et al., 2018). The dissolution preferentially removes the lighter-oxygen sections of the shells, thus increasing the whole shell δ¹⁸O_ruber (Berger and Gardner, 1975; Lohmann, 1995; Weinkauf et al., 2020). The studies based on the comparison of ambient parameters with the isotopic composition of living specimens collected in plankton tows may neither address the complete range of the changes in the isotopic signatures during the sinking of the tests from the surface waters post-death nor its subsequent deposition in the sediments at the bottom of the sea. As the fossil shells are the sole basis for finding out the isotopic ratio of the ambient seawater in the past, the effect of diagenetic changes including the dissolution on foraminifer's oxygen isotopic ratio has to be properly evaluated. Here, we assess the influence of freshwater-influx-induced strong salinity gradient, ambient temperature, depth-induced dissolution, and other associated parameters on the stable oxygen isotopic ratio of the surface-dwelling planktic foraminifera G. ruber (white variety) in the surface sediments of the northern Indian Ocean.

2 Ecology of Globigerinoides ruber (white)

Globigerinoides ruber is a spinose planktic foraminifera inhabiting the mixed layer waters, throughout the year, in the tropical–subtropical regions (Guptha et al., 1997; Kemle-von-Mücke and Hemleben, 1999; Stainbank et al., 2019). It is one of the dominant planktic foraminifera in the northern Indian Ocean (Bé and Hutson, 1977; Bhadra and Saraswat, 2021), with its relative abundance being as high as ∼60 % (Fraile et al., 2008). Its test is medium to low trochospiral, and it hosts algal symbionts (Hemleben et al., 1989). Globigerinoides ruber prefers to feed upon phytoplankton (Hemleben et al., 1989) and is dominant in oligotrophic warmer water, with the optimal temperature being 23.5 ^∘C (Fraile et al., 2008). However, it is amongst a few planktic foraminifera species that can tolerate a wide range of salinity (22–49 psu) and temperature (14–31 ^∘C; Hemleben et al., 1989; Guptha et al., 1997). Two varieties of G. ruber, namely the white and pink, are common in the world oceans. However, the pink variety of G. ruber became extinct in the Indian and Pacific oceans at ∼120 kyr during the Marine Isotopic Stage 5e (Thompson et al., 1979).

3 Northern Indian Ocean

The Indian Ocean, with its northern boundary in the tropics, includes two hydrographically contrasting basins, namely the Arabian Sea and Bay of Bengal (BoB; Fig. 1). The excess of evaporation over precipitation generates a high-salinity water mass that spreads throughout the surface of the northern Arabian Sea, with its core as deep as ∼100 m (Shetye et al., 1994; Prasanna Kumar and Prasad, 1999; Joseph and Freeland, 2005). Other high-salinity water masses from both the Persian Gulf and Red Sea enter the northern Arabian Sea at deeper depths between 200–400 and 500–800 m, respectively (Rochford, 1964). A strong upwelling along the western boundary of the Arabian Sea during the summer monsoon season brings cold, nutrient-rich subsurface waters to the surface (Chatterjee et al., 2019). The weak upwelling during the same season is also reported in the southeastern Arabian Sea (Smitha et al., 2014).

Figure 1 The sea surface temperature (SST; ^∘C; Locarnini et al., 2018) and sea surface salinity (SSS; psu; Zweng et al., 2018) in the northern Indian Ocean during the monsoon (August–September) and non-monsoon (April–May) months. The major rivers draining into the northern Indian Ocean are marked by thin blue lines. The map has been prepared by using Ocean Data View software (Schlitzer, 2018).

The surface water is relatively fresher in the BoB, as the majority of the rivers from the Indian subcontinent drain here, with the total annual continental runoff amounting to 2950 km³ (Sengupta et al., 2006). Additionally, the total annual precipitation over the BoB is 4700 km³, and the evaporation is 3600 km³ (Sengupta et al., 2006). The high-salinity water in the Arabian Sea is transported into the BoB, and the fresher BoB water mixes with the high-salinity water in the Arabian Sea by the seasonally reversing coastal currents (Shankar et al., 2002). The upwelling during summer is restricted to only the northwestern part of the BoB (Shetye et al., 1991). The upwelling combined with the convective mixing during the winter season in the northeastern Arabian Sea (Madhupratap et al., 1996), in addition to eddies in the BoB (Prasanna Kumar et al., 2004; Sarma et al., 2020), result in very high primary productivity in both basins (Qasim, 1977; Prasanna Kumar et al., 2009). The high primary productivity and freshwater capping induce strong stratification and restricted circulation that create oxygen-deficient zones (ODZs) at the intermediate depth in both the Arabian Sea (Rixen et al., 2020; Naqvi, 2021) and BoB (Bristow et al., 2017; Sridevi and Sarma, 2020). The Arabian Sea ODZ, however, is comparatively thicker and more intense, leading to denitrification (Naqvi et al., 2006), which has not yet been reported from the BoB (Bristow et al., 2017).

The equatorial Indian Ocean forms a part of the Indo-Pacific Warm Pool, with the sea surface temperature at >28 ^∘C throughout the year (Vinayachandran and Shetye, 1991; De Deckker, 2016). The marginal regions of the BoB are comparatively warmer due to the freshwater influx from the rivers. The riverine influx shoals the mixed layer and thickens the barrier layer, a buoyant layer separating the thermocline from the pycnocline, in the BoB (Howden and Murtugudde, 2001). The riverine influx flows as a low-salinity tongue all along the eastern margin of India (Chaitanya et al., 2014). The annual average sea surface salinity (SSS) is <34 psu throughout the BoB, increasing from the head bay towards the south. In contrast to that, SSS remains at >35 psu almost throughout the year in the Arabian Sea (Rao and Sivakumar, 2003). The excess of evaporation over precipitation, due to the dry northeasterly winds, leads to the highest salinity in the northern BoB during the winter (Rao and Sivakumar, 2003).

4 Materials and methodology

The surface sediments were collected all along the path of the seasonal coastal currents in the northern Indian Ocean (Fig. 2; Table S1 in the Supplement). The samples from the Ayeyarwady delta shelf in the northeastern BoB were collected during the India–Myanmar Joint Oceanographic Studies on board the ocean research vessel (R/V) Sagar Kanya (SK175). A total of 110 surface sediment samples were collected from the water depths, ranging from 10 to 1080 m, on the Ayeyarwady delta shelf (Ramaswamy et al., 2008). The multicore samples were also collected at regular intervals in transects running perpendicular to the coast from the western BoB during cruise SK308, on board R/V Sindhu Sadhana (cruise SSD067), and on board R/V Sindhu Sankalp (cruise SSK35). A total of 84 surface samples (including 71 multicore samples and 13 grab samples from sandy sediments) were collected from the inner shelf to outer slope region of the eastern margin of India during cruise SK308 (Suokhrie et al., 2021; Saalim et al., 2022). These samples from the western BoB represent the lowest-salinity region in the northern Indian Ocean (Panchang and Nigam, 2012). The multicore samples collected between 25 and 2980 m in the Gulf of Mannar and the region to the west of it (43 samples on board R/V Sindhu Sadhana during cruise SSD004) represent the zone of cross-basin exchange of seawater between the BoB and the Arabian Sea (Singh et al., 2021). The spade core samples collected from the southeastern Arabian Sea (R/V Sagar Kanya during cruise SK117 and SK237) are located close to the distal end of the low-salinity BoB water intruding into the Arabian Sea. The multicore samples (13 in number) collected during SSD055 cruise, from the northeastern Arabian Sea, represent the warm saline conditions. We also collected spade core surface samples from the Andaman Sea, on board R/V Sindhu Sankalp (cruise SSK98). A total of 252 samples had sufficient G. ruber for an isotopic analysis (Table 1). The new data were augmented with 148 previously published core-top studies (e.g. Sirocko, 1989; Prell and Curry, 1981; Duplessy et al., 1981a; Duplessy, 1982) mainly compiled by Waelbroeck et al. (2005). Therefore, a total of 400 surface sample data points were used for this study.

Figure 2 Location of the core-top samples analysed in this study (black-filled triangle is cruise SSD055, red-filled square is cruise SK117, blue-filled square is cruise SK237, blue-filled inverted triangle is cruise SSD004, blue-filled circle is cruise SSD067, red-filled circle is cruise SK308, black-filled square is cruise SSK035, red-filled triangle is cruise SSK098, and black-filled circle is cruise SK175), and the previously published core-top values (black plus sign) compiled from the northern Indian Ocean. The background contours are temperature (^∘C) (a) and salinity (psu) (b), with the scale on the right. Major rivers draining into the northern Indian Ocean are marked by thin blue lines. The map was prepared by using Ocean Data View software (Schlitzer, 2018).

Table 1 Details of the timing, region, and the number of surface sediment samples collected in each expedition used in this study (SK is Sagar Kanya, SSD is Sindhu Sadhana, and SSK is Sindhu Sankalp).

The surface sediment samples (0–1 cm) were processed following the standard procedure (Suokhrie et al., 2022). The freeze-dried sediments were weighed and wet-sieved by using a 63 µm sieve. Part of the coarse fraction (>63 µm) was dry-sieved by using 250 µm and 355 µm sieves. For δ¹⁸O analysis, 10–15 well-preserved shells of the G. ruber white variety were picked from 250 to 355 µm size range. We picked G. ruber sensu stricto wherever sufficient specimens were available. Unfortunately, several samples yielded a very small carbonate fraction. In such samples, we picked a mixed population of G. ruber to have sufficient specimens for isotopic analysis. The δ¹⁸O_ruber was measured by using Thermo Scientific MAT 253 isotope ratio mass spectrometer coupled with a Kiel IV automated carbonate preparation device. The samples were analysed at the Alfred Wegener Institute for Polar and Marine Research, Bremerhaven (MARUM, University of Bremen, Bremen, Germany) and the Stable Isotope Laboratory (SIL) at Indian Institute of Technology, Roorkee, India. Reference material from the National Bureau of Standards (NBS 18/19) limestone was used as the calibration material, and a secondary in-house standard was run after every five samples to detect and correct the drift. The precision of oxygen isotope measurements was better than 0.08 ‰. The δ¹⁸O_ruber data generated on the newly collected surface sediments was augmented with the published core-top δ¹⁸O measurements in the northern Indian Ocean. A total of 400 surface sediment data points (252 from this work and 148 from the previous studies) were used to understand the factors affecting δ¹⁸O_ruber in the northern Indian Ocean (Table S1 in the Supplement). The annual average sea surface temperature and salinity of the top 30 m of the water column at the respective sample locations was downloaded from the World Ocean Atlas (Boyer et al., 2013). The salinity and temperature at the core location was extrapolated from the nearby grid points using the live-access server at the National Institute of Oceanography, Goa, India.

The analysed δ¹⁸O_ruber data were compared with the expected δ¹⁸O calcite to ascertain whether the G. ruber properly represents the ambient conditions. For the expected δ¹⁸O calcite, the δ¹⁸O_sw was calculated from the ambient salinity by using a compilation of the regional seawater salinity and its stable oxygen isotopic (δ¹⁸O_sw) ratio data for the entire northern Indian Ocean (5^∘ S to 30^∘ N). The seawater salinity and corresponding δ¹⁸O_sw data were downloaded from Schmidt et al. (1999; version 1.22) and augmented with other regional data sets (Delaygue et al., 2001; Singh et al., 2010; Achyuthan et al., 2013).

5 Results

The oxygen isotopic ratio of G. ruber varied from a minimum of −3.82 ‰ to the maximum of −1.09 ‰ in the surface sediments of the northern Indian Ocean (Fig. 3). The most depleted δ¹⁸O_ruber was in the eastern BoB, and the most enriched values were in the western Arabian Sea.

Figure 3 The variation in Globigerinoides ruber δ¹⁸O (‰) in the surface sediments of the northern Indian Ocean. The stations are marked by a black-filled circle. The lowest δ¹⁸O_ruber is in the riverine-influx-influenced northern Bay of Bengal, and the highest is in the evaporation-dominated central and western Arabian Sea. The major rivers are marked with thin blue lines. The map was prepared by using Ocean Data View software (Schlitzer, 2018).

The east–west gradient in δ¹⁸O_ruber was also evident in its significant correlation (R²=0.5, n=400) with the longitude (Fig. 4a). However, δ¹⁸O_ruber did not have any systematic latitudinal variation (Fig. 4b).

Figure 4 The variation in Globigerinoides ruber δ¹⁸O (‰), with the corresponding longitude (a) and latitude (b), in the surface sediments of the northern Indian Ocean. The correlation between δ¹⁸O_ruber and the latitude of the sample is insignificant.

A significant correlation (R²=0.14; n=400) was observed between the water depth and δ¹⁸O_ruber (Fig. 5). δ¹⁸O_ruber increased with increasing depth. The increase was gradual, without any abrupt change.

Figure 5 The relationship between water depth and the oxygen isotopic ratio of the mixed-layer-dwelling Globigerinoides ruber in the surface sediments of the northern Indian Ocean. The trend line signifies the relative enrichment of δ¹⁸O_ruber shells in surface sediments, with increasing water depth. The empirical relationship between the water depth and δ¹⁸O_ruber is represented with an equation. The δ¹⁸O_ruber is significantly correlated (R²=0.14; n=400) with the water depth.

The uncorrected δ¹⁸O_ruber was significantly correlated (R²=0.63; n=400) with the ambient salinity (Fig. 6a). However, the relationship between uncorrected δ¹⁸O_ruber and ambient temperature was not as robust (R²=0.18; n=400; Fig. 6b). A large scatter ($\sim -\mathrm{3.8}$ ‰ to −1.4 ‰) was observed in the δ¹⁸O_ruber of the samples collected from a narrow range of ambient temperature (28–29 ^∘C; Fig. 6b).

Figure 6 The relationship between the stable oxygen isotopic ratio of the mixed-layer-dwelling Globigerinoides ruber and annual average mixed layer salinity (a) and temperature (b) in the northern Indian Ocean.

As the northern Indian Ocean includes two contrasting basins, the δ¹⁸O_ruber–salinity relationship was explored for both the Arabian Sea and the BoB. A significant δ¹⁸O_ruber–salinity relationship was observed for both the Arabian Sea (R²=0.28; n=205) and BoB (R²=0.14; n=195; Fig. 7). We report a different δ¹⁸O_ruber–salinity relationship in these two basins. δ¹⁸O_ruber increased with increasing salinity in both the BoB and the Arabian Sea.

Figure 7 The relationship between the stable oxygen isotopic ratio of the mixed-layer-dwelling Globigerinoides ruber and the annual average mixed layer salinity in the Arabian Sea and Bay of Bengal.

The data set to derive the regional salinity–δ¹⁸O_sw relationship comprised of a total of 750 stations, with salinity varying from 20.92 to 40.91 psu (Schmidt et al., 1999). The data set also covered a large range of δ¹⁸O_sw, varying from a minimum of −2.45 ‰ to the maximum of 2.02 ‰ (Figs. 8 and 9). The measured δ¹⁸O_ruber was strongly correlated (R²=0.56; n=400) with the expected δ¹⁸O_calcite, as estimated by using the salinity–δ¹⁸O_sw relationship and the ambient temperature. However, the relationship between seawater temperature and δ¹⁸O_ruber–δ¹⁸O_sw was not very robust. It should, however, be noted here that the stratigraphic information is not provided for most of the core tops. The core-top sediments can represent older time slices when the sedimentation rates are low or when older sediments are exposed due to erosional processes. This does not matter so much if the Holocene is present and stable. However, in the Indian Ocean, large Holocene δ¹⁸O variations are expected due to variations in monsoon precipitation. Therefore, the uncertain age of the core tops can affect the results stated above.

Figure 8 The surface seawater oxygen isotopic ratio (‰) in the northern Indian Ocean. The black-filled circles are the seawater sample locations compiled from previous studies. The thin blue lines are the major rivers draining in the northern Indian Ocean. The map was prepared by using Ocean Data View software (Schlitzer, 2018).

Figure 9 The relationship between the surface water oxygen isotopic ratio and salinity in the northern Indian Ocean (5^∘ S–30^∘ N), Arabian Sea, and the Bay of Bengal. The data points are from Schmidt et al. (1999), Delaygue et al. (2001), Singh et al. (2010), and Achyuthan et al. (2013).

6 Discussion

6.1 Expected versus analysed δ¹⁸O

The seawater δ¹⁸O data are required to estimate the expected δ¹⁸O carbonate. The seawater δ¹⁸O, however, was not measured. Therefore, the salinity–δ¹⁸O_sw relationship established from the previous regional seawater isotope and salinity measurements was used (Fig. 8). The salinity–δ¹⁸O_sw relationship varies seasonally and from region to region (Singh et al., 2010; Achyuthan et al., 2013; Tiwari et al., 2013). Therefore, it was difficult to choose the appropriate salinity–δ¹⁸O_sw relationship. Initially, all the data points were clubbed to establish the salinity–δ¹⁸O_sw relationship. By comparing the measured δ¹⁸O_sw with the ambient salinity, we established the following relationship for the entire northern Indian Ocean (north of 5^∘ S latitude; R²=0.57; n=750; Fig. 9):

$$\begin{array}{l}{\mathit{\delta }}^{\mathrm{18}}{\mathrm{O}}_{\mathrm{sw}}=\mathrm{0.16}\cdot \text{salinity}-\mathrm{4.94}\\ \text{Northern Indian Ocean }(R^{\mathrm{2}}=\mathrm{0.57}).\end{array}$$

Previously, a large difference in the slope of salinity–δ¹⁸O_sw equation has been reported from the Arabian Sea and the BoB (Delaygue et al., 2001; Singh et al., 2010; Achyuthan et al., 2013). Therefore, we also plotted the salinity–δ¹⁸O_sw separately for the Arabian Sea and BoB (Fig. 9). The salinity–δ¹⁸O_sw relationship for these two basins was represented by the following equations:

$$\begin{array}{l}{\mathit{\delta }}^{\mathrm{18}}{\mathrm{O}}_{\mathrm{sw}}=\mathrm{0.10}\cdot \text{salinity}-\mathrm{2.84}\\ \text{Arabian Sea }(R^{\mathrm{2}}=\mathrm{0.31},n=\mathrm{375})\\ {\mathit{\delta }}^{\mathrm{18}}{\mathrm{O}}_{\mathrm{sw}}=\mathrm{0.13}\cdot \text{salinity}-\mathrm{4.23}\\ \text{Bay of Bengal }(R^{\mathrm{2}}=\mathrm{0.31},n=\mathrm{375}).\end{array}$$

The continuous flux of G. ruber throughout the year (Guptha et al., 1997), and the accumulation of shells in the sediments over a large interval, implies that the salinity–δ¹⁸O_sw relationship based on data representing all seasons will provide a better estimate of the average δ¹⁸O_ruber, as recovered from the sediments (Vergnaud-Grazzini, 1976). The expected δ¹⁸O_sw was calculated by using these equations and the annual average mixed layer salinity at the stations for which δ¹⁸O_ruber data were available. The mixed layer was defined as the top 25 m of the water column, following Narvekar and Prasanna Kumar (2014). Although the mixed layer depth varies regionally and during different seasons, the average mixed layer depth was used to compare the calcification conditions. A correction factor of 0.27 ‰ was applied to convert δ¹⁸O_sw from Standard Mean Oceanic Water (SMOW) scale to the Pee Dee Belemnite (PDB) scale (Hut, 1987). The expected δ¹⁸O calcite was then estimated from the calculated δ¹⁸O_sw and the annual average mixed layer temperature by using the equation proposed by Mulitza et al. (2003). We also estimated the expected δ¹⁸O calcite by using the high light equation of Bemis et al. (1998), as G. ruber δ¹⁸O is better described with the high light equation (Thunell et al., 1999). The choice of equation used to estimate the expected δ¹⁸O calcite did not make any difference, other than a small offset. The difference between expected δ¹⁸O calcite estimated by using palaeotemperature equation of Mulitza et al. (2003) and the high light equation of Bemis et al. (1998) varied from −0.33 ‰ to −0.41 ‰. The expected δ¹⁸O calcite estimated by using the Mulitza et al. (2003) palaeotemperature equation provided values close to the measured G. ruber δ¹⁸O. From the scatterplot (Fig. 10), it was clear that the analysed δ¹⁸O_ruber was significantly correlated (R²=0.56; n=400) with the expected δ¹⁸O calcite, suggesting that G. ruber closely represents the ambient conditions in the entire northern Indian Ocean. The expected δ¹⁸O calcite estimated by using the separate Arabian Sea and BoB salinity–δ¹⁸O_sw equations was also similarly correlated with the analysed δ¹⁸O_ruber.

Figure 10 The scatterplot of expected δ¹⁸O calcite (as estimated from the ambient salinity–temperature) and the analysed δ¹⁸O_ruber. The two are significantly correlated (R²=0.56), suggesting that Globigerinoides ruber correctly represents the ambient conditions. The blue diamonds are the samples collected from the Bay of Bengal, and the black diamonds represent the samples collected from the Arabian Sea. The dotted line represents the 1:1 relationship between the measured and expected δ¹⁸O.

The deviation of the expected δ¹⁸O_calcite from the observed δ¹⁸O_ruber (δ¹⁸O_residual) can be because of several factors, including the difference in the ambient conditions at the time of the secretion of the primary calcite during the lifetime and the digenetic changes post-death and during the burial. The observed δ¹⁸O_ruber was close to the expected δ¹⁸O_calcite in the shallower waters, especially the BoB, Andaman Sea, and northeastern Arabian Sea (Fig. 11). The difference was large in the deeper Arabian Sea and the equatorial Indian Ocean. Globigerinoides ruber is suggested to inhabit chlorophyll maximum for the easy availability of food (Fairbanks and Wiebe, 1980). In such a scenario, δ¹⁸O_ruber is expected to be higher due to lower temperatures and lower light levels at relatively deeper depths (Spero et al., 1997). The depth of chlorophyll maximum is shallower in the marginal marine waters of both the BoB and the Arabian Sea (Sarma and Aswanikumar, 1991; Madhu et al., 2006). If G. ruber thrived at chlorophyll maximum depths, then the δ¹⁸O_ruber should be enriched in heavier isotope and thus δ¹⁸O_residual should be negative. The positive δ¹⁸O_residual in the shallower regions, however, suggests that G. ruber thrives in the warmer upper parts of the mixed layer. Alternatively, the large influence of the heavier-oxygen-isotope-depleted freshwater δ¹⁸O dominates the chlorophyll maximum influence on the observed δ¹⁸O_ruber in the shallower regions of the northern Indian Ocean. The concentration of positive δ¹⁸O_residual values close to the riverine influx regions confirms the strong influence of the heavier-oxygen-isotope-depleted freshwater δ¹⁸O in modulating δ¹⁸O_ruber in the northern Indian Ocean. The negative δ¹⁸O_residual at deeper stations is attributed to a combination of factors including deeper chlorophyll maximum depth habitat of G. ruber, the reduced influence of freshwater, a lower sedimentation rate resulting in the mixing of older and younger fauna, and post-depositional digenetic changes.

Figure 11 The difference in the expected δ¹⁸O_calcite and observed δ¹⁸O_ruber (δ¹⁸O_residual) in the surface sediments of the northern Indian Ocean. The grey-filled squares are the sample locations. The thin blue lines are the major rivers draining in the northern Indian Ocean. The thin black lines mark the contours at 0.25 ‰ intervals. The map was prepared by using Ocean Data View software (Schlitzer, 2018).

6.2 Latitudinal and longitudinal variation in δ¹⁸O_ruber

We report a strong (R²=0.50) longitudinal influence on δ¹⁸O_ruber. A similar relationship with the latitudes is missing. The strong longitudinal signature in δ¹⁸O_ruber is attributed to the large salinity gradient. The huge freshwater influx in the BoB reduces the SSS in the eastern Indian Ocean. The lack of major rivers in the western Arabian Sea results in strong low- to high-salinity gradient from east to west. Although the equatorial and nearby regions are a part of the Indo-Pacific Warm Pool, the limited temperature variability is evident in the insignificant latitudinal influence on δ¹⁸O_ruber.

6.3 Diagenetic alteration

We found a strong diagenetic overprinting of δ¹⁸O_ruber in the northern Indian Ocean (Fig. 5). The enrichment of δ¹⁸O_ruber with increasing water depth suggests either dissolution, leading to the preferential removal of chambers with a higher fraction of the lighter oxygen isotope (Wycech et al., 2018), or secondary calcification under comparatively colder water (Lohmann, 1995; Schrag et al., 1995). The increase in planktic foraminifera δ¹⁸O with increasing depth is a common diagenetic alteration throughout the world oceans (Bonneau et al., 1980). Interestingly, the extent of the increase in δ¹⁸O_ruber with depth in the northern Indian Ocean is much smaller (0.18 ‰ per 1000 m) than that reported for the same species from the Pacific Ocean (0.4 ‰ per 1000 m; Bonneau et al., 1980). However, the increase in δ¹⁸O_ruber with depth in the northern Indian Ocean is continuous, unlike the abrupt shift in δ¹⁸O (0.3 ‰–0.4 ‰ between the depths above and below the lysocline) of another surface-dwelling planktic species, namely Trilobatus sacculifer, as observed in the western equatorial Pacific (Wu and Berger, 1989). The smaller increase in δ¹⁸O_ruber with depth is attributed to the shallower habitat of G. ruber as compared to T. sacculifer. The chamber formation at different water depths implies increased heterogeneity in the T. sacculifer shells, with those formed at warmer surface temperature being more susceptible to dissolution as compared to those formed at deeper depths during the gametogenesis phase (Duplessy et al., 1981b; Wycech et al., 2018). The chambers in G. ruber are formed at a similar depth, and therefore, the increase in δ¹⁸O_ruber is continuous, while those of T. sacculifer are precipitated at different depths, and therefore, the shift in δ¹⁸O occurs after a particular depth. The increase in δ¹⁸O_ruber with depth is mainly due to the partial dissolution of the more porous and thinner parts of the shells secreted at warmer temperatures, as such parts are comparatively more susceptible to dissolution (Berger, 1971). The increase in δ¹⁸O_ruber with depth is similar in both the Arabian Sea and BoB.

Additionally, the gradual decrease in the sedimentation rate with increasing depth and distance from the continental margins can also induce a depth-related trend in δ¹⁸O_ruber. The bioturbation disturbs the top few centimetres of the sediments (Gerino et al., 1998), mixing older shells with comparatively younger shells (Löwemark and Grootes, 2004). In high sedimentation rate regions of the shelf and slope, the mixing is restricted to the shells deposited in a shorter, climatologically stable interval. However, in the deeper regions, it is likely that the shells deposited during the colder glacial interval or deglaciation with relatively higher δ¹⁸O_ruber are mixed with the younger shells, as these are available close to the surface due to the low sedimentation rate (Broecker, 1986; Anderson, 2001). The mixing of shells with a relatively higher δ¹⁸O with the modern shells having lighter δ¹⁸O can also result in the depth-related increasing trend of δ¹⁸O_ruber. The sedimentation rate is very high on the slope and decreases in the deeper regions of both the Arabian Sea (Singh et al., 2017) and BoB (e.g. Bhonsale and Saraswat, 2012; Suokhrie et al., 2022). However, it should be noted here that the sedimentation rate in large parts of the BoB is still high enough to prevent availability of deglacial or last glacial maximum shells within the bioturbation-induced mixing limits.

The large influence of the terrestrial freshwater influx in the shallower region, as compared to the deeper parts of the northern Indian Ocean, is also likely to contribute to the observed increase in δ¹⁸O_ruber with depth. The freshwater is depleted in heavier oxygen isotopes as compared to the seawater (Bhattacharya et al., 1985; Ramesh and Sarin, 1992). Thus, the foraminiferal shells secreted in the shallow waters are likely to be enriched in the lighter oxygen isotope, resulting in a depth-related bias. Therefore, to delineate the influence of depth-related diagenetic alteration and secondary calcification in δ¹⁸O_ruber, we subtracted the expected δ¹⁸O_calcite from the measured δ¹⁸O_ruber. The difference between the measured δ¹⁸O_ruber and expected δ¹⁸O_calcite was plotted with water depth (Fig. 12). The difference (measured δ¹⁸O_ruber minus expected δ¹⁸O_calcite) increased with depth, suggesting a strong influence of the depth-related processes in δ¹⁸O_ruber.

Figure 12 The relationship of the difference between measured δ¹⁸O_ruber and expected δ¹⁸O_calcite with the water depth from which the surface samples were collected in the northern Indian Ocean. The δ¹⁸O_ruber−δ¹⁸O_calcite increased with increasing water depth.

6.4 Salinity contribution to δ¹⁸O_ruber

We report a strong influence of the freshwater-influx-induced salinity on δ¹⁸O_ruber (R²=0.63). As expected, δ¹⁸O_ruber has a direct positive relationship with the ambient salinity. The δ¹⁸O_ruber increased by 0.29 ‰ for every practical salinity unit (psu) increase in salinity. The northern Indian Ocean has a large salinity gradient (∼10 psu) from the lowest in the northern BoB to the highest in the northwestern Arabian Sea, which is mainly driven by the freshwater input. The river water and direct precipitation is enriched in the lighter oxygen isotope (Kumar et al., 2010; Kathayat et al., 2021). Thus, the increased riverine influx and precipitation contributes isotopically lighter water to the surface ocean (Rai et al., 2021) and decreases the δ¹⁸O_ruber. From the surface seawater samples collected during the winter monsoon season (January–February 1994), a δ¹⁸O–salinity slope of 0.26 ‰ was deduced for the Arabian Sea and 0.18 ‰ for the BoB (Delaygue et al., 2001). However, the δ¹⁸O–salinity slope also varies regionally during different seasons (Singh et al., 2010; Achyuthan et al., 2013). The δ¹⁸O–salinity slope varied from as low as 0.10 for the coastal BoB samples collected during the months of April–May to as high as 0.51 for the samples collected from the western BoB during the peak southwestern monsoon season (August–September 1988; Singh et al., 2010). The large seasonal variation implies limitations in the δ¹⁸O–salinity slope deduced from snapshot surface seawater samples. Additionally, G. ruber flux is reported throughout the year (Guptha et al., 1997), suggesting that the fossil population represents annual average conditions (Thirumalai et al., 2014).

A different δ¹⁸O_ruber–salinity slope for the Arabian Sea (0.28) and BoB (0.19) is attributed to the different hydrographic regimes of these two basins. The runoff and precipitation excess in the BoB results in a comparatively lower salinity as compared to the evaporation-dominated Arabian Sea. However, it should be noted here that the relationship between δ¹⁸O_ruber and salinity was very robust for all of the northern Indian Ocean samples plotted together. Interestingly, the slope of δ¹⁸O–salinity for the entire northern Indian Ocean samples is much lower than that for the Atlantic Ocean (0.59 for the North Atlantic and 0.52 for the South Atlantic; Delaygue et al., 2000), despite the large meltwater influx into the North Atlantic. The dissimilar δ¹⁸O–salinity slope in different basins and also during different seasons in the same basin is mainly attributed to the variation in the end-member composition and the relative amount of freshwater (riverine/precipitation/sub-marine groundwater discharge) input from various sources during different seasons (Achyuthan et al., 2013; Tiwari et al., 2013). The heavier oxygen-isotope-depleted precipitation/freshwater influx in the higher latitudes ($\sim -\mathrm{35}$ ‰) as compared to the tropical areas ($\sim -\mathrm{5}$ ‰) also results in a higher slope of the δ¹⁸O–salinity relationship in the North Atlantic Ocean (Rozanski et al., 1993). Additionally, the difference in the δ¹⁸O–salinity slope despite the huge freshwater input into both the basins is also because a large fraction of the riverine freshwater spreads across the surface of the northern Indian Ocean, while the meltwater sinks to deeper depths in the North Atlantic Ocean. A consistent systematic difference has previously been observed between planktic foraminiferal shells collected in plankton tows and surface sediments, with shells from the sediments being comparatively enriched in ¹⁸O (Vergnaud-Grazzini, 1976).

6.5 Temperature control on δ¹⁸O_ruber

A first-order comparison of the uncorrected δ¹⁸O_ruber with ambient temperature of the top 30 m of the water column at respective stations showed 0.32 ‰ decrease with every 1 ^∘C warming. The change in δ¹⁸O_ruber, as inferred from the core-top sediments of the northern Indian Ocean is higher than that estimated from the plankton tows (0.22 ‰ per 1 ^∘C change in temperature; Mulitza et al., 2003). The seawater temperature was amongst the primary factors identified to affect δ¹⁸O_ruber (Emiliani, 1954; Mulitza et al., 2003). The low correlation between δ¹⁸O_ruber and temperature in this data set is attributed to the limited temperature variability (1, 28–29 ^∘C) at a majority of the stations. The large salinity difference (∼6.5 psu) between stations further obscures any significant correlation between uncorrected δ¹⁸O_ruber and temperature. The temperature influence on δ¹⁸O_ruber was thus assessed by comparing the ambient temperature with the δ¹⁸O_ruber corrected for δ¹⁸O_sw (δ¹⁸O_ruber−δ¹⁸O_sw). The pH of the seawater has also been identified as a factor affecting the stable oxygen isotopic composition of planktic foraminifera (Bijma et al., 1999). However, as argued by Mulitza et al. (2003), the limited modern surface seawater pH variability (Chakraborty et al., 2021) and its close dependence on temperature implies that the pH contribution to δ¹⁸O_ruber is well within the error associated with the measurements. The seawater pH in the immediate vicinity of the foraminiferal shell is strongly influenced by the light intensity in the presence of symbionts (Jørgensen et al., 1985). The riverine influx in the northern Indian Ocean makes the surface waters turbid, reducing the light penetration depths (Prasanna Kumar et al., 2010). Therefore, riverine-influx-induced variations in turbidity in the northern Indian Ocean can influence the δ¹⁸O_ruber via the pH effect.

The comparison of δ¹⁸O_sw-corrected δ¹⁸O_ruber with the ambient temperature also confirms the enrichment of (δ¹⁸O_ruber−δ¹⁸O_sw) in heavier oxygen isotope with the decrease in temperature (Fig. 13). We obtained the following relationship between temperature and (δ¹⁸O_ruber−δ¹⁸O_sw) in the northern Indian Ocean.

$$\begin{array}{}\text{(1)}& \text{Temperature}=-\mathrm{0.59}\cdot ({\mathit{\delta }}^{\mathrm{18}}{\mathrm{O}}_{\mathrm{ruber}}-{\mathit{\delta }}^{\mathrm{18}}{\mathrm{O}}_{\mathrm{sw}})+\mathrm{26.40}.\end{array}$$

The slope of the temperature and (δ¹⁸O_ruber−δ¹⁸O_sw) was somewhat different (0.17 ‰ 1 ^∘C change in temperature) than that of the temperature versus uncorrected δ¹⁸O_ruber but similar to that reported for the plankton tows (0.22 ‰ 1 ^∘C change in temperature; Mulitza et al., 2003).

Figure 13 The relationship between ambient temperature and (δ¹⁸O_ruber−δ¹⁸O_sw) for the northern Indian Ocean. As expected, the (δ¹⁸O_ruber−δ¹⁸O_sw) becomes enriched in heavier isotopes with decreasing ambient temperature.

7 Data availability

The newly generated data and the data compiled from previous studies from the northern Indian Ocean have been submitted to PANGAEA and are available at https://doi.org/10.1594/PANGAEA.945401 (Saraswat et al., 2022). These data are also available in the Supplement of this article.

8 Conclusions

We measured the stable oxygen isotopic ratio of the surface-dwelling planktic foraminifera Globigerinoides ruber (white variety) from the surface sediments of the northern Indian Ocean. A comparison of the δ¹⁸O_ruber with the depth suggests a strong diagenetic alteration of the isotopic ratio. The freshwater-influx-induced changes in the ambient salinity exert the maximum influence on the δ¹⁸O_ruber, suggesting its robust application for reconstructing the past salinity in the northern Indian Ocean. The large east–west salinity gradient in the northern Indian Ocean results in a strong longitudinal variation in δ¹⁸O_ruber. The temperature influence on δ¹⁸O_ruber is subdued as compared to the effect of large salinity variation in the northern Indian Ocean. We report a relatively smaller change in δ¹⁸O_ruber with a unit increase in ambient temperature in case of specimens retrieved from the surface sediments as compared to those collected live from the water column.