Distinct oxygenation modes of the Gulf of Oman over the past 43 000 years – a multi-proxy approach

Changing climatic conditions can shape the strength and extent of the oxygen minimum zone (OMZ). The presence and variability of the OMZ in the Arabian Sea is of importance to the latter's ecosystem. The state of oxygenation has, for instance, an impact on the pelagic and benthic faunal community or the nitrogen and carbon cycles. It is important to understand the dynamics of the OMZ and related marine environmental conditions because of their climate feedbacks. In this study, we combined three independent proxies to reconstruct the oxygenation state of the water column and bottom water in the Gulf of Oman for the past 43 kyr approximately. This multi-proxy approach is done for the first time at the northeastern Oman margin located in the Gulf of Oman. We used bulk sedimentary nitrogen isotopes (δ¹⁵N) and the alkane ratio (lycopane +n-C₃₅)/n-C₃₁ and benthic foraminiferal faunal analysis to reconstruct the strength of the OMZ in the water column and bottom water oxygenation, respectively. Our results show that the Gulf of Oman experienced strong pronounced OMZ and bottom water deoxygenation during the Holocene. In contrast, during Marine Isotope Stage 2 (MIS 2), including the Last Glacial Maximum (LGM), the Gulf of Oman was very well ventilated, with a highly diverse benthic foraminiferal community. This may have been caused by stronger wind-induced mixing and better ventilation by oxygen-rich water masses. Our results also show moderate oxygenation during MIS 3, with deoxygenation events during most of the warmer Dansgaard–Oeschger (D–O) events. We propose two distinct oxygenation modes for the Gulf of Oman: (1) a stable period of either strongly pronounced water column OMZ and bottom water deoxygenation or well-oxygenated water column and bottom water conditions and (2) an unstable period of oscillating oxygenation states between moderately oxygenated (stadials) and deoxygenated (interstadials, D–O events) conditions. The unstable period may be triggered by an interstadial Atlantic meridional overturning circulation (AMOC) mode, which is required to initiate D–O events.

1 Introduction

Oxygen minimum zones (OMZs) are an important feature of the oceans. Although they cover only about 8 % of the total oceanic surface (Paulmier and Ruiz-Pino, 2009), they have a great impact on the marine environment and the carbon and nitrogen cycles (Friederich et al., 2008; Gruber, 2004; Paulmier et al., 2011; Rixen et al., 2020; Wakeham, 2020). For instance, denitrification plays a major role in the nitrogen cycle of OMZs, and denitrification also acts as the largest sink for nitrogen in the ocean (Gruber, 2004; Rixen et al., 2020). Moreover, OMZs have the potential to act as a source for greenhouse gases like N₂O, CO₂ or CH₄ (Friederich et al., 2008; Naqvi et al., 2010; Paulmier and Ruiz-Pino, 2009). Further, several studies show that the intensification and deoxygenation of OMZs are positively coupled to climate warming (e.g. Bopp et al., 2013; Breitburg et al., 2018; Busecke et al., 2022; Zhou et al., 2022). One of the strongest OMZs is located in the Arabian Sea, comprising about 17 % of the worldwide area of OMZs (Rixen et al., 2020). Over half of the permanently hypoxic shelf sediments are located in the northern Indian Ocean, affecting benthic ecosystems – for instance, the diversity of benthic foraminifera (Helly and Levin, 2004; Levin, 2003). Although the diversity of benthic foraminifera is generally reduced under low-oxygen conditions, the OMZ faunas of the Arabian Sea seem to be more resilient to low-oxygen conditions when compared to other non-OMZ regions. This higher resilience can be attributed to a dominance of nitrate-respiring infaunal benthic foraminifera in OMZ regions (Schmiedl et al., 2023a).

The Gulf of Oman, as part of the Arabian Sea, is a complex and poorly understood region. It is influenced by the intrusion of warm, highly saline and more oxygenated Persian Gulf Water (PGW) when compared to the Arabian Sea Water. Different gyres in the Gulf of Oman and their seasonal variations lead to a patchy distribution of the PGW in this area (Pous et al., 2004; Wang et al., 2013). Little is known about the oceanic environmental conditions in the past and their response to climate change in this region as only a few sedimentary records and modelling studies exist (Cullen et al., 2000; Lachkar et al., 2019; Schmidt et al., 2020; Sirocko et al., 1991; Staubwasser et al., 2003). According to Lachkar et al. (2019), rising sea surface temperatures (SSTs) in the Persian Gulf, especially during the winter months, reduce the ability of the PGW to ventilate the upper OMZ in the Gulf of Oman due to shallower spreading of the PGW. The authors also concluded that a strengthening of the OMZ in the Gulf of Oman can be also induced by lower sea levels. In this scenario, the Persian Gulf has fallen dry, and, therefore, the ventilation via the PGW would stop completely (Lachkar et al., 2019). However, existing Holocene and late Pleistocene sedimentary records (cores M5/3a – 422 and Orgon4-KS8, Sirocko et al., 1991, 2000; Staubwasser et al., 2003) located below the OMZ in the Gulf of Oman do not investigate the oxygenation status in this area. Nevertheless, from sedimentary records in other parts of the Arabian Sea OMZ we know that the OMZ varied in its extend and volume in the past (Altabet et al., 2002; Böll et al., 2014; Burdanowitz et al., 2019; Gaye et al., 2018; Ivanochko et al., 2005; Möbius et al., 2011; Pichevin et al., 2007; Suthhof et al., 2001). Those records have shown a greater expansion of the OMZ during the Holocene (interstadial), also known as Marine Isotope Stage 1 (MIS 1), than during the Last Glacial Maximum (LGM, MIS 2) and stadials of MIS 3. Further, most of the rapid warming events of the Dansgaard–Oeschger (D–O) events during the last glacial period resulted in a strengthening of the OMZ, albeit each single D–O event had a different impact on a local scale (Altabet et al., 2002; Ivanochko et al., 2005; Pichevin et al., 2007). However, due to the lack of continuous high-resolution archives in the Gulf of Oman, little is known about the OMZ and bottom water oxygenation prior to the Holocene, especially without the influence of Persian Gulf water (PGW) in this area.

Most of the records are based on the qualitative reconstruction of the oxygenation status. The most commonly used proxy is the bulk δ¹⁵N of marine sediments. The global average δ¹⁵N of deep-water nitrate is about 4.8 ± 0.2 ‰, but in intermediate and surface water, the δ¹⁵N of nitrate can deviate strongly from this value due to the isotopic fractionation associated with nitrogen-cycling processes (Sigman et al., 2000; Sigman and Fripiat, 2019). In oxygen-depleted environments, heterotrophic bacteria use nitrate as an electron acceptor to oxidize organic matter and reduce the nitrogen, in several steps, to nitrogen gas (Sigman and Fripiat, 2019). Denitrification is associated with a strong isotopic effect of 15 ‰–25 ‰ so that the residual nitrate becomes strongly enriched in ¹⁵N and can thus have δ¹⁵N values > 20 ‰ (Casciotti, 2016). Upward mixing and upwelling can transport this enriched isotopic signal where it is incorporated into phytoplankton. During the nitrate consumption, phytoplankton also use the lighter ¹⁴N with an isotope discrimination factor of about 5 ‰ and incorporate the denitrification signal into the biomass (Montoya, 2008). After the biomass is deceased and sinks to the ocean floor, the denitrification signal is transported to the seafloor and can be stored in the sediments (e.g. Altabet et al., 1995; Gaye-Haake et al., 2005). Therefore, δ¹⁵N values can be used as a proxy for the OMZ strength in the water column (e.g. Altabet et al., 1995; Reichart et al., 1997). However, diagenetic processes can alter sedimentary δ¹⁵N values, especially under low sedimentation rates (Gaye-Haake et al., 2005; Jung et al., 1997; Junium et al., 2015; Möbius et al., 2011; Tesdal et al., 2013). Nevertheless, it was found that, in the Arabian Sea, especially under high sedimentation rates on the shelf and slope, sedimentary δ¹⁵N values are a reliable indicator for past denitrification processes (Möbius et al., 2011).

Figure 1 (a) Map of the study site SL167 (red star); nearby marine records (red circles) MD04-2876 (Pichevin et al., 2007), RC27-14 (Altabet et al., 2002) an NIOP905 (Ivanochko et al., 2005); and stalagmite record (blue circle in inset map) of Sofular Cave (SC) (Fleitmann et al., 2009). The black arrows (SW monsoon and NW winds) show the prevailing wind pattern during the summer, and the dashed black arrows (NE monsoon and subtropical westerly jet or western disturbances, STWJ/WD) indicate the prevailing wind pattern during the winter (after Clift and Plumb, 2008; Hunt et al., 2018). Map was created using ArcGIS v.10.8 (Esri, 2019). The bathymetric data are from the General Bathymetric Chart of the Oceans (GEBCO, 2014; http://www.gebco.net, last access: 4 January 2017). (b) Temperature (blue), oxygen (O₂, black) and salinity (red) depth profiles obtained at the end of the SW monsoon season in September 2007 during M74/1b cruise at the location of SL167.

Another rarely used biomarker is the isoprenoid hydrocarbon lycopane, which is mainly limited to sediments from OMZs or oceanic anoxic events in the past (van Bentum et al., 2009; Dummann et al., 2021; Farrington et al., 1988; Ogihara, 2014; Sabino et al., 2020, 2021; Sinninghe Damsté et al., 2003). It is discussed that photoautotrophic organisms probably produce lycopane, but its origin is not yet fully resolved (Freeman et al., 1994; Sinninghe Damsté et al., 2003; Wakeham et al., 1993). However, these studies have shown that lycopane is well preserved in anoxic environments and is degraded fast under oxic conditions. Lycopane often co-elutes with the n-alkane n-C₃₅ (van Bentum et al., 2009; Naeher et al., 2012; Sabino et al., 2020; Sinninghe Damsté et al., 2003), with the latter mainly being produced by terrestrial plants (e.g. Eglinton and Hamilton, 1967; Meyers and Ishiwatari, 1993). Unusually high abundances of n-C₃₅ and lycopane (if co-eluted) or lycopane alone were detected in some parts of the Arabian Sea, with the highest contents found within the OMZ (Schulte et al., 1999, 2000; Sinninghe Damsté et al., 2003). Therefore the (lycopane + n-C₃₅)/n-C₃₁ ratio can be used as an indicator for bottom water oxygenation, with ratios > 0.5 representing suboxic to anoxic conditions (van Bentum et al., 2009; Sinninghe Damsté et al., 2003). Whereas these two presented proxies are a qualitative way to reconstruct the OMZ strength and bottom water oxygenation, benthic foraminifera can be used to achieve a quantitative oxygenation status reconstruction. The different benthic foraminifera species with their broad specific oxygen level requirements are widely used to reconstruct past bottom water oxygen conditions (e.g. Fontanier et al., 2002; Jorissen et al., 1995; Koho et al., 2008; Kranner et al., 2022; Lu et al., 2023; Schmiedl et al., 2000, 2010).

Here, we present the first high-resolution record reconstructing the OMZ strength and oxygen status of the bottom water at a depth situated within the present OMZ for the Holocene and late Pleistocene (past 43 kyr, approximately) at the Oman margin in the Gulf of Oman using a multi-proxy approach.

2 Modern oceanographic setting in the study area

The Gulf of Oman, as part of the northwestern Arabian Sea, connects the Arabian Sea via the Strait of Hormuz with the Persian Gulf. The northern Arabian Sea and the Gulf of Oman are affected by different water masses. At the surface, the Arabian Sea high-salinity water (ASHSW), with salinities between 35.3 and 36.7, comprises the upper 200 m (Kumar and Prasad, 1999; Shenoi et al., 1993). The highly saline (> 36.5) low-oxygen PGW is responsible for a weak ventilation of the upper OMZ (Fig. 1, 150–300 m water depth), especially in the southern part of the Gulf of Oman (Bower et al., 2000; Kumar and Prasad, 1999; Morrison et al., 1998; Pous et al., 2004; Prasad et al., 2001; Schmidt et al., 2020; Shenoi et al., 1993; Wang et al., 2013). The Red Sea Water (RSW, salinity: 35.1–35.6) enters the Arabian Sea via the Gulf of Aden and is found at intermediate water depths between 600 to 900 m (Bower et al., 2000; Kumar and Prasad, 1999). The RSW is transported northward along the Oman coast, especially during the SW monsoon (Acharya and Panigrahi, 2016; Beal et al., 2000; Kumar and Prasad, 1999; Schmidt et al., 2020). Another intermediate water mass (500–1000 m water depth) is the Indian Central Water (ICW), which is a mixture of the Antarctic Intermediate Water (AAIW) and the Indonesian Intermediate Water (IIW) and which enters the Arabian Sea during the SW monsoon via the Somali current along the Somali and Arabian coasts (Acharya and Panigrahi, 2016; Schott and McCreary, 2001). The initial oxygen-rich ICW becomes less oxygenated during its transport to the Arabian Sea (Acharya and Panigrahi, 2016). However, the oxygen content is still higher than that of the PGW and RSW (Rixen et al., 2014).

Figure 2 Average chlorophyll-a concentration (MODIS Aqua, L3m, 4 km spatial resolution, monthly) for the different seasons of (a) March–April–May (MAM), (b) June–July–August (JJA), (c) September–October–November (SON) and (d) December–January–February (DFJ) over the period from 2002 to 2022 (NASA, 2022). Data were visualized using the Giovanni online data system, developed and maintained by the NASA GES DISC (date of access: 26 January 2023).

The region is influenced by different atmospheric circulations, which result in different seasonal primary-productivity patterns (Fig. 2). During the summer season, strong southwest (SW) winds and the Findlater Jet in the western Arabian Sea induce upwelling and lateral transport of cold nutrient-rich water, resulting in high primary productivity (Fig. 2b) (Andruleit et al., 2003; Haake et al., 1993; Rixen et al., 2020). In contrast, the highest productivity in the northern Arabian Sea and Gulf of Oman is observed during the winter season (Fig. 2d) when high amounts of nutrients are available due to deeper wind-induced surface mixing by the NE winds (Böll et al., 2014; Madhupratap et al., 1996; Rixen et al., 2020). During the summer season, the Ras Al Hadd Jet in the northeastern corner of Oman can transport cold upwelled water into the Gulf of Oman via eddies (Schott and McCreary, 2001). Further, eddies generated in the Gulf of Oman transport PGW trapped in its core into the Arabian Sea (L'Hégaret et al., 2015; de Marez et al., 2019). Both may have an impact on the sea surface temperature, salinity and primary productivity in this region.

3 Material and methods

The 739 cm long gravity core SL167 was collected in the northwestern Arabian Sea from the northeastern Oman margin off northern Oman (Gulf of Oman – 22°37.15^′ N, 59°41.49^′ E; 774 m water depth) during the RV Meteor cruise M74/1b in 2007 (Fig. 1). In total, 371 samples were used for δ¹⁵N analyses, with continuous sampling intervals of 2 cm. For lipid biomarker analyses, 225 samples were used, with an interval of 2 cm in the upper 162 cm of the core and, due to low total organic carbon content (unpublished), an interval of 4 cm below 162 cm. All sediment samples were freeze-dried and homogenized with an agate mortar and pestle prior to chemical treatment for the analyses. For the analyses of benthic foraminifera, 149 samples taken every 5 cm were used.

3.1 Dating and age model

The age model is based on 21 radiocarbon datings by accelerator mass spectrometry (AMS) measurements of surface-dwelling planktonic foraminifera, comprising Globigerinoides spp., Globigerinella spp. and Orbulina universa, at Beta Analytic Inc., Miami, USA. The age–depth model was developed by using the Bayesian model package BACON v.2.5.6 (Blaauw and Christen, 2011) within R statistical software (v.4.3, R Core Team, 2023). We used the default settings, except for accumulation mean (set to 50), memory mean (set to 0.3) and the calibration curve (set to Marine20). We applied a deltaR (ΔR) of 93 ± 61 years. The ΔR is based on the weighted mean of two regional marine reservoir corrections (Muscat) by Southon et al. (2002) using the marine calibration database (Reimer and Reimer, 2001, http://calib.org/marine/, last access: 9 August 2021).

3.1.1 Lipid biomarker analyses

Total lipid extracts (TLEs) of about 3 to 18 g of the sediment samples were obtained by using a Dionex Accelerated Solvent Extractor (ASE 200) at 100 °C and 1000 PSI for 5 min (three cycles) using a solvent mixture of dichloromethane–methanol ($\mathrm{DCM}--\mathrm{MeOH}$, 9:1) following the procedure described in Herrmann et al. (2016). Prior to extraction, a known amount of an internal standard (squalene) was added to the samples. After the extraction, TLEs were concentrated via rotary evaporation and transferred to combusted 4 mL vials. We have used NaSO₄ column chromatography to separate the TLEs into a hexane-soluble and hexane-insoluble fraction. Therefore, a combusted Pasteur pipette was packed with cleaned cotton wool and about 2 cm NaSO₄. The column was then first cleaned with about 8 mL hexane. Then, TLE was transferred with about 1 mL hexane to the column, and the remaining 4 mL vial was cleaned three times with about 1 mL hexane, which was also transferred to the column. For the hexane-insoluble fraction, about 1 mL of DCM was added four times to the column. The hexane-soluble fraction was saponified (85 °C, 2 h) in 5 % potassium hydroxide (KOH) in MeOH solution, and the neutral fraction was extracted with hexane. To obtain the n-alkane-containing apolar fraction from the neutral fraction, a column chromatography packed with deactivated silica gel (5 % H₂O, 60 mesh) was carried out by using hexane as a solvent. Afterwards, the apolar fraction was cleaned with hexane via column chromatography packed with silver-nitrate-coated silica gel (AgNO₃–Si).

For quantification of the n-alkanes, a Thermo Scientific TRACE 1310 gas chromatograph was coupled to a flame ionization detector (GC-FID) equipped with a Thermo Scientific TG-5MS column (30 m, 0.25 mm, 0.25 µm). The carrier gas H₂, with a flow rate of 35 mL min⁻¹, was used. The programmed temperature vaporizing (PTV) injector was operated starting at 50 °C, ramped by 10 °C s⁻¹ to 325 °C in a splitless mode. The initial GC temperature was programmed to be 50 °C (held for 3 min) and was then increased by 6 °C min⁻¹ to a final temperature of 325 °C, which was held for 20 min. For quantification of n-alkanes, an external standard containing the n-C₈–n-C₄₀ alkanes was used at a known concentration. The quantification precision of repeated analyses of the external standard was 8 %.

The mass spectra of two samples (14–16 and 182–186 cm) were investigated with a Thermo Scientific TRACE GC Ultra coupled to a Thermo Scientific DSQ II mass spectrometer (GC-MS) with He (2 mL min⁻¹ flow rate) as the carrier gas. The initial GC temperature of 50 °C was held for 3 min and ramped by 6 °C min⁻¹ to 325 °C, with the final temperature being held for 25 min. To identify the compounds of the apolar fraction, the mass spectra were compared with published mass spectral data.

The average chain length (ACL) of plant-wax-derived n-alkanes is commonly used to identify plant functional types and environmental conditions (e.g. Collister et al., 1994; Cooper et al., 2015; Eglinton and Eglinton, 2008; Rommerskirchen et al., 2006). For instance, plants located in arid environments tend to have a longer ACL than plants in humid environments (e.g. Carr et al., 2014; Rommerskirchen et al., 2006; Vogts et al., 2009). But the ACL also differs with the plant functional type; for instance, C₄ grasses have, in general, a higher ACL than woody gymnosperms (e.g. Bush and McInerney, 2013; Carr et al., 2014; Cooper et al., 2015). However, the validity of the ACL is limited and, if possible, should be combined with compound-specific isotope measurements of the n-alkanes (e.g. Eglinton and Eglinton, 2008; Rommerskirchen et al., 2006; Vogts et al., 2009). We have calculated the ACL of n-alkanes using the following equation:

$$\begin{array}{}\text{(1)}& {\mathrm{ACL}}_{\mathrm{27}\text{–}\mathrm{33}}={\displaystyle \frac{\mathrm{27}\times {\mathrm{C}}_{\mathrm{27}}+\mathrm{29}\times {\mathrm{C}}_{\mathrm{29}}+\mathrm{31}\times {\mathrm{C}}_{\mathrm{31}}+\mathrm{33}\times {\mathrm{C}}_{\mathrm{33}}}{{\mathrm{C}}_{\mathrm{27}}+{\mathrm{C}}_{\mathrm{29}}+{\mathrm{C}}_{\mathrm{31}}+{\mathrm{C}}_{\mathrm{33}}}}.\end{array}$$

The carbon preference index (CPI), an indicator of the odd-over-even predominance, is usable to distinguish between terrestrial plant and petroleum sources (e.g. Bray and Evans, 1961; Cranwell, 1978, 1981; Pancost and Boot, 2004). Terrestrial-plant-derived n-alkanes and recent sediments with unaltered organic material have an odd-over-even predominance, with a CPI higher than 3 (e.g. Bray and Evans, 1961). In contrast, petroleum sources have higher abundance of even n-alkanes, with CPIs < 1, and are an indication of higher degradation (e.g. Bray and Evans, 1961). The CPI was calculated by using the following equation:

$$\begin{array}{}\text{(2)}& \begin{array}{rl}{\mathrm{CPI}}_{\mathrm{27}\text{–}\mathrm{33}}& =\mathrm{0.5}\times ({\displaystyle \frac{{\mathrm{C}}_{\mathrm{27}}+{\mathrm{C}}_{\mathrm{29}}+{\mathrm{C}}_{\mathrm{31}}+{\mathrm{C}}_{\mathrm{33}}}{{\mathrm{C}}_{\mathrm{26}}+{\mathrm{C}}_{\mathrm{28}}+{\mathrm{C}}_{\mathrm{30}}+{\mathrm{C}}_{\mathrm{32}}}}\\ & +{\displaystyle \frac{{\mathrm{C}}_{\mathrm{27}}+{\mathrm{C}}_{\mathrm{29}}+{\mathrm{C}}_{\mathrm{31}}+{\mathrm{C}}_{\mathrm{33}}}{{\mathrm{C}}_{\mathrm{28}}+{\mathrm{C}}_{\mathrm{30}}+{\mathrm{C}}_{\mathrm{32}}+{\mathrm{C}}_{\mathrm{34}}}}),\end{array}\end{array}$$

where C_x is the concentration of the n-alkane with x atoms.

3.2 Bulk nitrogen isotope (δ¹⁵N) analyses

For the δ¹⁵N analyses, following the procedure described in Menzel et al. (2014), 6 to 61 mg of the freeze-dried and homogenized sediment was weighted into Sn capsules. δ¹⁵N values were obtained by combusting the samples at 1050 °C in a Thermo Scientific Flash EA 1112 elemental analyser coupled to a Finnigan MAT 252 isotope ratio mass spectrometer. Nitrogen was calibrated against the International Atomic Energy Agency (IAEA) reference standards IAEA-1 and IAEA-2. In addition to being internal standards, both (IAEA-1 and IAEA-2) were used as working standards. Replicate measurements of these standards yielded a precision better than 0.2 ‰. The samples were measured in duplicate, with a mean standard deviation of 0.07 ‰.

Figure 3 (a) Eutrophic to mesotrophic section of the trophic-oxygen (TROX) model of Jorissen et al. (1995), with the addition of Gooday (2003) describing the general dependence of benthic foraminiferal microhabitat structure and diversity from oxygen concentration and food supply. Figure modified from Schmiedl et al. (2023a). (b) Reconstructed dissolved oxygen concentration of bottom and pore waters versus benthic foraminiferal diversity H(S) for core SL167. The stippled line represents a polynomial of the second-degree model with p < 0.001. The observed significant correlation suggests a dominant oxygen control of benthic foraminiferal diversity at site SL167.

3.3 Benthic foraminifera analyses

For faunal analyses, between 198 and 558 (average of 311) individuals were at each depth, counted from representative splits of the size fraction > 125 µm. Species assignment is mainly based on Jones (1994), Den Dulk (2000), Szarek (2001), Schumacher et al. (2007) and Debenay (2012). Allochthonous shelf taxa, including larger benthic foraminifera (e.g. genera Amphistegina, Heterostegina, Borelis, Peneroplis), typical neritic taxa (e.g. Ammonia spp., Cibicides refulgens, Elphidium spp., Lobatula lobatula) and taxa with floating chambers (genera Cymbaloporetta, Millettiana, Tretomphalus, Tretomphaloides), were removed from the census data set (Murray, 1991). For benthic foraminiferal diversity, the Shannon Index H(S) was calculated according to Buzas and Gibson (1969). The H(S) considers the number of species and their relative proportion in the sample. The H(S) value is at a maximum when all species have equal proportions, while species with low abundances contribute little to it. In eutrophic to mesotrophic ecosystems, the diversity and microhabitat structure are oxygen controlled, as predicted by the TROX (trophic-oxygen) model (Gooday, 2003; Jorissen et al., 1995) (Fig. 3). For the estimation of bottom water oxygenation, the different taxa were classified into oxic, suboxic and dysoxic taxa based on their modern microhabitat preferences (Table S1 in the Supplement) (Schmiedl et al., 2023a). Bottom water oxygen concentrations were then calculated based on the enhanced benthic foraminiferal oxygen index and associated transfer function of Kranner et al. (2022), with the modification of Schmiedl et al. (2023a). The mean standard deviation (SD) across the entire oxygen range (0–6 mL L⁻¹) is ± 0.61 mL L⁻¹. However, the SD is lower in suboxic to dysoxic conditions, with SDs of ± 0.49 and ± 0.08 mL L⁻¹ across oxygen ranges of 1–2 and 0–1 mL L⁻¹, respectively. The expected relation between oxygen concentration and benthic foraminiferal diversity is clearly expressed at core site SL167 since it is influenced by comparatively high organic matter fluxes and low-oxygen conditions (Fig. 3).

3.4 Statistics

To identify periodicities in our data sets we performed spectral and wavelet analyses in R (v.4.3, R Core Team, 2023). For the spectral analysis, we used the REDFIT function of the package dplR v.1.7.4 (Bunn et al., 2022; Bunn, 2008, 2010), which is based on the Fortran 90 REDFIT source code by Schulz and Mudelsee (2002). For the wavelet analysis the data sets were first interpolated to an evenly spaced data set by using the package ncdf4.helpers v.0.3-6 (Bronough, 2021) and the approx. function. The wavelet analyses were performed with the package biwavelet v.0.20.21 (Gouhier et al., 2021) using the Morlet wavelet function and bias-corrected power spectrum, which is based on Torrence and Compo (1998).

Table 1 AMS ¹⁴C dates of planktic foraminifera of sediment core SL167 and calibrated ages using Calib 8.2 with Marine20 calibration curve and ΔR of 93 ± 61 years.

Figure 4 Age–depth model of SL167 using the R package BACON v. 2.5.6 (Blaauw and Christen, 2011). The upper panels (a–c) show the Markov chain Monte Carlo iterations (a), the distributions of the prior (green curve) and posterior (grey area) accumulation rates (b), and memory (c). The lower panel (d) shows the age–depth model of SL167. The calibrated ¹⁴C dates are shown in blue. The red line shows the modelled mean age of SL167 with the 95 % confidence interval (dotted black lines). A ΔR of 93 ± 61 years was applied.

4 Results

4.1 Age model of SL167

The results of the AMS radiocarbon measurements are shown in Table 1. The core SL167 comprises the last 42.6 kyr, approximately (Fig. 4). However, the last 3 kyr are missing in the core as the core top represents ca. 3.1 ka. The mean sedimentation rates range between 0.10 and 0.49 mm yr⁻¹, with higher sedimentation rates during the Holocene compared to the Pleistocene.

Figure 5 GC-MS measurement of the apolar fraction of sample from 14 to 16 cm sediment depth with retention time (a) and mass spectra of the n-alkane C₃₅ + lycopane (b).

4.2 Lipid biomarker and δ¹⁵N reconstructions

The samples of SL167 show a strong odd-over-even carbon number predominance with CPI values ranging between 2 and 13.8, and they show a dominance of land-plant-derived long-chain n-alkanes, where the ACL_27–33 varies between 30 and 31.2 (Fig. A1 in the Appendix). In some samples, for land-plant-derived n-alkanes, unusually high contents of the n-alkane C₃₅ were detected in the GC-FID chromatograms. We chose one exemplary sample with a very high content of n-C₃₅ and one sample with an usual distribution of land-plant-derived n-alkanes for mass spectral analyses. For the latter one, we found only the mass spectra of the n-alkane C₃₅ at the expected retention time. However, for the sample with the high n-C₃₅ content and unusual n-alkane distribution pattern, we found the mass spectra of both lycopane (2,6,10,14,19,23,27,31-octamethyldotriacontane) and n-C₃₅ (Fig. 5). The characteristic fragments of lycopane are the ions of m/z 113, 183, 253, 309, 337, 407 and 447 (Sinninghe Damsté et al., 2003), which are also detected in our sample with the co-elution of n-C₃₅ and lycopane (Fig. 5b). It is known from other studies that n-C₃₅ and lycopane can co-elute during gas chromatography measurements (van Bentum et al., 2009; Sabino et al., 2020; Sinninghe Damsté et al., 2003).

Figure 6 Oxygen proxies of SL167. Shannon Wiener (H(S)) diversity index for benthic foraminifera (BF) (a), nitrogen isotope values (δ¹⁵N) as denitrification and oxygen minimum zone (OMZ) strength indicator for the water column (b), reconstructed bottom water oxygen content by benthic foraminifera (c), and ratio of (lycopane +n-C₃₅)/n-C₃₁ as indicator for bottom water oxygen (d). The red line in panel (d) indicates the threshold of oxic to suboxic to anoxic conditions (Sinninghe Damsté et al., 2003). Grey bars indicate Bølling–Allerød (B–A) and Dansgaard–Oeschger events (D–O, numbers), mostly based on Fleitmann et al. (2009). Blue bars indicate Younger Dryas event based on Fleitmann et al. (2009) and Heinrich (H) events based on a compilation by Allard et al. (2021). Diamonds show dated ages of SL167. Note the reversed axes for H(S) and BF reconstructed oxygen. The grey bars at the bottom represent the different Marine Isotope Stages (MISs).

The (lycopane + n-C₃₅)/n-C₃₁ varies between 0.1 and 1.7 throughout the core (Fig. 6d). The highest ratios occur during the late Holocene (3–4 ka) and warmer Pleistocene periods reconstructed from the NGRIP ice core δ¹⁸O (North Greenland Ice Core Project members, 2004) and Sofular Cave δ¹³C records (Fleitmann et al., 2009), respectively.

The δ¹⁵N values in our Gulf of Oman record vary between 4.7 and 9.2 ‰ (Fig. 6b), with constantly high values throughout the Holocene (> 7.2 ‰). The variability of δ¹⁵N is high (4.5 ‰) during the Pleistocene, with higher values > 7 ‰ during warmer phases.

4.3 Benthic foraminiferal diversity and reconstructed bottom water oxygenation

Overall, the benthic foraminiferal diversity is high and ranges between H(S) = 1.94 and 4.27 (average of 3.31 ± 0.64 SD). Significant fluctuations occur in the older part of the core, with minima during interstadials, while Heinrich events H2 and H1 and the Last Glacial Maximum are characterized by comparatively high H(S) values (Fig. 6). During the Holocene, the H(S) values exhibit a gradual decrease. The reconstructed oxygen values range between 0.27 and 3.25 (average of 1.46 ± 0.74 SD) and are negatively correlated to the benthic foraminiferal diversity and δ¹⁵N. D–O and B–A events and the Holocene are characterized by relatively low oxygen concentrations, while cold intervals, particularly the interval around H1, exhibit higher values (Fig. 6).

5 Discussion

The δ¹⁵N is often used as an indicator for past denitrification processes and, in the Arabian Sea, the strength of the OMZ (e.g. Altabet et al., 2002; Gaye et al., 2018; Ivanochko et al., 2005; Möbius et al., 2011; Pichevin et al., 2007). However, the δ¹⁵N signal can be biased by different processes, which need to be considered. One factor is a diagenetic overprint of δ¹⁵N leading to an enrichment of the heavier ¹⁵N isotope, especially under low sedimentation rates (3–4 cm kyr⁻¹) (Altabet et al., 1999; Gaye-Haake et al., 2005; Jung et al., 1997; Junium et al., 2015; Möbius et al., 2011; Tesdal et al., 2013). In contrast, shelf and slope sediments deposited under high sedimentation rates in the Arabian Sea are considered to have unaltered δ¹⁵N values and, therefore, are considered to be a reliable indicator for past denitrification processes (Möbius et al., 2011). The sedimentation rates are high in SL167 and vary between 9.7 and 49 cm kyr⁻¹. Further, they correlate with the nitrogen content but not with δ¹⁵N values (Fig. A2). The correlation of the nitrogen content and δ¹⁵N may due to higher productivity and/or better preservation of organic matter under sub-anoxic conditions. Therefore, we argue that sedimentary δ¹⁵N values in SL167 are a reliable indicator for past denitrification processes. However, it may be possible that the δ¹⁵N signal in SL 167 is affected by changes in the water mass transport, mixing or lateral transport from e.g. the Oman upwelling area.

Figure 7 Comparison of high-resolution nitrogen isotope (δ¹⁵N) records in the Arabian Sea. SL167 (this study) from the Gulf of Oman (a), MD04-2876 (Pichevin et al., 2007) from the Pakistan margin in the northern Arabian Sea (b), RC27-14 (Altabet et al., 2002) from the Oman upwelling region in the western Arabian Sea (c) and NIOP905 (Ivanochko et al., 2005) from the Somali upwelling in the southwestern Arabian Sea (d). Grey bars indicate Bølling–Allerød (B–A) and Dansgaard–Oeschger events (D–O, numbers), mostly based on the δ¹³C Sofular Cave record (e) by Fleitmann et al. (2009). Combined Red Sea sea level reconstruction of sediment core KL11 (f) by Siddall et al. (2003) and Rohling et al. (2008) for the Holocene and Pleistocene, respectively. The δ¹⁸O records of NGRIP (North Greenland Ice Core Project members, 2004) and SPICE (Steig et al., 2021) are shown in (g). Blue bars indicate Younger Dryas event based on Fleitmann et al. (2009) and Heinrich (H) events based on a compilation by Allard et al. (2021). The grey bars at the bottom represent the different Marine Isotope Stages (MISs).

5.1 Strength of the OMZ and bottom water oxygenation in the Gulf of Oman

5.1.1 Pleistocene

Both independent proxies for bottom water oxygenation – (lycopane + n-C₃₅)/n-C₃₁ ratio and the O₂ reconstruction by benthic foraminifera – and the Shannon diversity index are in very good agreement, at least for the Pleistocene part of the core (Fig. 6a, c and d). Phases of sub-anoxic bottom water match well with observed interstadials, the D–O events, in the Greenland ice core records, e.g. NGRIP (North Greenland Ice Core Project members, 2004), except for D–O 6 and 7. The highest reconstructed bottom water oxygen content (up to 3.2 mL L⁻¹) occurred from about 14.3 to 19.6 ka and during the Younger Dryas between 11.2 and 12.3 ka (Fig. 6c). Further, MIS 3 (about 25–43 ka), is characterized by several distinct peaks of a strong OMZ (Fig. 6b) in the water column. Note that we used the Sofular Cave δ¹³C record from Turkey (Fleitmann et al., 2009) instead of the NGRIP to allocate the distinct D–O events as this seems to be more appropriate (Fig. 7). Fleitmann et al. (2009) found a systematic age offset for most D–O events between the NGRIP and the more regional Sofular and Hulu cave (China) records of several centennial years, with younger ages for the D–O events in the NGRIP record. The D–O events 3–5 and 8–11 are well pronounced as strong abrupt OMZ intervals in the Gulf of Oman. In contrast, in the nearby Oman upwelling (core RC27-14), all D–O events, except 11, are expressed as strong OMZ intervals in the water column (Fig. 7c) (Altabet et al., 2002). A high-resolution record (MD04-2876) in the northern Arabian Sea shows a similar pattern as core RC27-14 (Pichevin et al., 2007), whereas a record (NIOP905) from the more southern Somali upwelling area (Ivanochko et al., 2005) does not show a strong OMZ in the water column during the first four D–O events (Fig. 7b and d). Note that shifts in OMZ peaks of the different records in the Arabian Sea may be due to age uncertainty, as well as peak tuning of the other records with the Greenland ice core (GISP2) δ¹⁸O record.

Bottom water oxygenation may be driven not only by processes in the upper water column but also by the occurrence of different water masses. The RSW is transported northward, especially during the SW monsoon (Acharya and Panigrahi, 2016; Beal et al., 2000; Kumar and Prasad, 1999), and may undergo oxygen depletion due to higher organic matter supply and its decay on its way to the Gulf of Oman (Pathak et al., 2021). Acharya and Panigrahi (2016) have shown that today's RSW core extends between 600 and 660 m, nearly reaching the seafloor at the core site (∼ 740 m). Thus, phases with stronger SW monsoons, such as the D–O interstadials, are characterized by more inputs of low-oxygen RSW and, therefore, sub-anoxic conditions of bottom water in the Gulf of Oman. This corroborates findings from the Oman upwelling area, where oxygen bottom water reconstructions using benthic foraminifera show oxygen-depleted conditions during phases with higher outflows of RSW into the northwestern Arabian Sea (Pathak et al., 2021).

The missing D–O 6 and weak D–O 7 events might be a result of the combination of local or regional and global conditions. First, the strength of the Indian and Asian monsoon systems are strongly coupled to high-latitude climate (e.g. Deplazes et al., 2013, 2014; Overpeck et al., 1996; Schulz et al., 1998; Singh et al., 2011; Wang et al., 2001). Phases with strong monsoon intervals are coupled to the D–O events found in Greenland ice cores. However, the duration and strength of both varied through the time. In some records from the Asian region, D–O 6 had a weaker environmental impact than other D–O events (e.g. Deplazes et al., 2014; Ivanochko et al., 2005; Wang et al., 2001). Second, the fluctuation of the Red Sea sea level (Fig. 7f) could lead to variations in the exchange of the RSW and the Indian Ocean or Arabian Sea, with a stronger influence during higher sea levels and vice versa (Arz et al., 2007; Siddall et al., 2003). However, the sea level was, in general, lower during MIS 3 than during the Holocene, which might have led to a weaker influence of the RSW compared to recent conditions (Rohling et al., 2008; Siddall et al., 2003). Third, the degree of northward extension of the oxygen-rich AAIW is linked to the North Atlantic climate (Jung et al., 2009; Pahnke and Zahn, 2005). The maxima in the northward extension and, therefore, ventilation of the Arabian Sea water column are found during the Heinrich stadials. In contrast, minima extension occurred during interstadials with a stronger monsoon (Jung et al., 2009). In total, the interplay of these three mentioned factors, the bipolar seesaw structure of the northern and southern hemispheric climate but also the climate and oceanic patterns on regional scale (e.g. Lemieux-Dudon et al., 2010; Stocker and Johnsen, 2003), may lead to the feature that some of the D–O events, as well as some of the Heinrich events (H4), are not represented in the Gulf of Oman record. Further, stronger northwesterly and northeasterly winds during stadials (Leuschner and Sirocko, 2000) could have induced stronger deep winter mixing compared to present and, therefore, better ventilation (Lu et al., 2023; Reichart et al., 1998).

A striking feature of our Gulf of Oman δ¹⁵N record is the prominent triple peak of strong water column deoxygenation from D–O event 3 to D–O event 5 (Fig. 6b). This is in contrast to other Arabian Sea records that do not show a prominent triple peak from D–O event 3 to D–O event 5 but from 5 to 7, with a broader D–O 8 event beforehand (Schulz et al., 1998). Interestingly, at the core site, suboxic to anoxic bottom water conditions occur during D–O 2, whereas the OMZ in the water column was not developed. We speculate that the inflow of oxygen-depleted RSW into the Gulf of Oman was still strong and/or that the maxima northward extension of the ventilating AAIW and/or ICW was further south, while denitrification processes were not as strong as during other pronounced D–O events. This is also the case for a record from the southern Somali upwelling area (Fig. 7d). However, along the Oman upwelling area (Altabet et al., 2002), a strong SW monsoon could have favoured upwelling-induced denitrification and therefore a strengthening of the OMZ in this region only, albeit not as strongly as during former D–O events.

In contrast to other parts of the Arabian Sea, the water column in the Gulf of Oman was well ventilated during the whole LGM, indicated by low δ¹⁵N values (Fig. 7) and by high benthic foraminiferal diversity (Fig. 6). The northern and northeastern Arabian Sea experienced a stronger and/or longer NE monsoon season during the LGM (Pichevin et al., 2007; Suthhof et al., 2001). Thus, the stronger wind-induced mixing may have resulted in a better-ventilated OMZ compared to the interstadials. A recent study from the Oman upwelling area also suggests slightly increased oxygenation in the upper water column (∼ 600–820 m) during the LGM compared to during the modern interstadial (Lu et al., 2023). Further, less lateral transport of nutrient- and oxygen-poor waters from the Oman upwelling due to weakened SW winds was supposed to mitigate denitrification in the northern and northeastern Arabian Sea (Suthhof et al., 2001). In addition, a northward expansion of the well-oxygenated AAIW during stadials results in better-ventilated intermediate waters in the Arabian Sea, reducing the strength of the OMZ (Jung et al., 2009; Pichevin et al., 2007). However, the productivity was still high enough to maintain an OMZ. The absence of the OMZ at the core location from the LGM to the beginning of the Bølling–Allerød (B–A) events may be due to more intense northwesterly winds in the Gulf of Oman and the NE monsoon (Leuschner and Sirocko, 2000; Sirocko et al., 2000) leading to stronger wind-induced mixing and ventilation of the water column. Weak Indian summer monsoon phases also occurred during the LGM in NE India (Dutt et al., 2015) and in the eastern Arabian Sea (Saravanan et al., 2020). In addition, Red Sea sea level was too low during that time (Sergiou et al., 2022; Siddall et al., 2003) to induce significant RSW influence on bottom water deoxygenation in the Gulf of Oman. The Persian Gulf plays no role in contributing water masses to the Gulf of Oman as it was nearly completely dry until the LGM and had no connection to the Gulf of Oman (Lambeck, 1996; Stoffers and Ross, 1979).

During the B–A interstadial, a strong OMZ and bottom water oxygenation occurred in the Gulf of Oman and was also observed for the whole Arabian Sea (Altabet et al., 2002; Ivanochko et al., 2005; Kessarkar et al., 2013; Orsi et al., 2017; Pichevin et al., 2007; Suthhof et al., 2001). Contrarily to the O₂ reconstruction of the benthic foraminifera, we see no distinct high ratios of (lycopane + n-C₃₅)/n-C₃₁ during the B–A interstadial. Here, it is possible that the threshold of oxygen depletion was not reached, preserving lycopane in the sediment.

Figure 8 Spectral analysis (a) and wavelet power spectrum (b) of δ¹⁵N of SL167. Red and blue lines in (a) indicate the red noise level and 95 % confidence interval (CI), respectively. Black numbers show significant periodicities on the 95 % CI. The black lines in (b) indicate the 95 % significance level. The colours in (b) represent the power of the wavelet power spectrum, with red (blue) showing high (low) power. The grey-shaded area represents the cone of influence.

5.1.2 Holocene

The observed low reconstructed O₂ contents and high δ¹⁵N values with minor fluctuations during the Holocene indicate a continuously strong OMZ and sub-anoxic bottom water at the Oman margin area in the Gulf of Oman. This is in line with other studies for most parts of the Arabian Sea (Fig. 7, Altabet et al., 2002; Burdanowitz et al., 2019; Gaye et al., 2018; Kessarkar et al., 2013; Pichevin et al., 2007; Suthhof et al., 2001). In the course of the post-LGM sea level rise, the Persian Gulf became connected to the Gulf of Oman via the Strait of Hormuz at around 13 ka, with flooding of the central part of the Persian Gulf until about 11.5 ka (Lambeck, 1996). Since at least the mid-Holocene, the PGW has been an important factor in ventilating the upper OMZ (150–300 m) in the northern Arabian Sea (Lachkar et al., 2019). However, modern data show a relatively low increase in O₂ content (about 0.5 mol L⁻¹) due to PGW at (Fig. 1) and near the core location (Munz et al., 2017).

The highest ratios of (lycopane + n-C₃₅)/n-C₃₁ (> 1.0) occurred during the late Holocene, which may be due to relatively high total organic carbon mass accumulation rates during that time (Fig. A1) and thus is an effect of fast burial. This assumption is also corroborated by the increasing dominance of the opportunistic food indicator Uvigerina peregrina (Koho et al., 2008; Schmiedl et al., 2010), which favours a high supply of organic matter, leading to decreasing benthic foraminiferal diversity.

Overall, our multi-proxy oxygen record for the Gulf of Oman shows striking features. First, during most interstadial periods (D–O events and the Holocene) the denitrification and OMZ in the water column, as well as bottom water deoxygenation, were strong. Second, during MIS 3, oxygen conditions oscillated between moderately oxygenated (glacial) and deoxygenated or denitrification conditions (interglacial, D–O events), marking this period as an environmentally unstable period. Here, the oxygen concentrations were close to the threshold of suboxic and anoxic conditions and therefore fluctuated between oxia and suboxia, driven by changes in monsoon strength. In contrast, LGM or MIS 2 and the Holocene period show quite constant oxygen and deoxygenation and denitrification conditions, respectively. We argue that these two periods are stable enough to suppress strong oscillations of oxygenation versus deoxygenation.

Figure 9 Spectral analysis (a) and wavelet power spectrum (b) of the reconstructed oxygen content of SL167 using benthic foraminifera assemblage (O₂ BF). Red and blue lines in (a) indicate the red noise level and 95 % confidence interval (CI), respectively. Black numbers show significant periodicities on the 95 % CI, and the grey number (3.17 ka) shows the significant periodicity at the 90 % CI. The black lines in (b) indicate the 95 % significance level. The colours in (b) represent the power of the wavelet power spectrum, with red (blue) showing high (low) power. The grey-shaded area represents the cone of influence.

5.2 Periodic changes and potential global drivers for OMZ strength and bottom water oxygenation

In order to identify specific periodicities, we have performed spectral and wavelet analyses for the δ¹⁵N record (Fig. 8) and the oxygen (O₂) reconstruction of the benthic foraminifera (Fig. 9), respectively. For the strength of the OMZ in the water column based on δ¹⁵N, we have found significant (95 % confidence interval, CI) periods of 3.8, 3.2, 2.7, 1.3 and 1.1 kyr, as well as 690 and 570 years (Fig. 8a). For the reconstructed O₂ bottom water variations, we have found significant periods (95 % CI) of 1.1 kyr and 710, 670, 600 and 570 years, as well as 3.2 kyr (90 % CI, Fig. 9a). These periods are mainly present between about 25–43 ka (MIS 3) and 8–16 ka (transition of LGM or MIS 2 to the Holocene, Figs. 8b and 9b) for both proxies. We attribute the periods of about 1.1–1.3 ka and around 3.2 ka to the D–O events as they occur within the periods of 1.5 and 3.0 ka approximately (Kuniyoshi et al., 2022; Schulz, 2002). In particular, the period of around 1.5 ka is a widespread feature in climate records (e.g. Bond et al., 2001; Jaglan et al., 2021; Lauterbach et al., 2014; Leuschner and Sirocko, 2000; Saravanan et al., 2020; Thamban et al., 2007; P. Wang et al., 2005). However, Obrochta et al. (2012) question the general idea of a 1.5 ka oscillation as they propose a superposition of the 1.0 and 2.0 ka cycles, which is also evident in our record with the 1.1 to 1.3 ka periods. Further strong evidence for D–O events as triggers of these periods is that the D–O events are also absent during the LGM or MIS 2 period (Buizert and Schmittner, 2015; North Greenland Ice Core Project members, 2004). The cold phase and its high global ice volume during the LGM inhibited an interstadial AMOC mode (strong AMOC and/or warm North Atlantic), which is required to initiate D–O events (Buizert and Schmittner, 2015). An interesting feature is the successive decrease in the duration and magnitude of the D–O events between two Heinrich events (Buizert and Schmittner, 2015). However, this pattern may or may not be the reason for the missing D–O 6 signal in our record as D–O 5 is strong and the last event between H4 and H3.

The periods of 570 to 710 years can be related to solar cycles (Liu et al., 2012; Stuiver and Braziunas, 1993; P. Wang et al., 2005). These cycles are prominent worldwide and are also found in the east Asian (Liu et al., 2012; Wang et al., 1999; P. Wang et al., 2005; Xu et al., 2014) and Indian monsoon (Burdanowitz et al., 2021; Neff et al., 2001; von Rad et al., 1999; Saravanan et al., 2020; Sarkar et al., 2000; Thamban et al., 2007) realms. Some authors attribute a 725–775-year periodicity to changes in the Indian summer monsoon (Saravanan et al., 2020) and in El Niño–Southern Oscillation (ENSO)-like cycles (Russell et al., 2003) or a harmonic cycle of the 1500-year cycle (von Rad et al., 1999), linked to the thermohaline circulation (Wang et al., 1999) or to the Intertropical Convergence Zone (ITCZ) movement in the tropics (Russell and Johnson, 2005). Most of the high-frequency periods (< 1000 years) are reported for the Holocene (Burdanowitz et al., 2021; Liu et al., 2012; Neff et al., 2001; von Rad et al., 1999; Russell et al., 2003; Russell and Johnson, 2005; Stuiver and Braziunas, 1993; Thamban et al., 2007; Wang et al., 1999; Y. Wang et al., 2005; Xu et al., 2014) or the LGM (Saravanan et al., 2020). Here, we show that these high-frequency periods are also present during MIS 3. The intensity and position of the monsoon system are driven by the summer insolation (precession signal) and the latitudinal temperature gradient (obliquity signal), respectively, with the latter being influenced by the latitudinal insolation gradient and ice cover (Davis and Brewer, 2009). The authors postulate that a strong latitudinal temperature gradient moves the monsoon system towards the Equator, which was the case during the LGM or MIS 2 for the summer (ca. 20–28 ka) and winter (ca. 18–25 ka) latitudinal temperature gradients (Davis and Brewer, 2009). This coincides with the well-ventilated water column and bottom water in the Gulf of Oman during that time, suggesting a response to the obliquity signal.

6 Conclusion

Our study presents a multi-proxy approach to reconstructing the water column and bottom water oxygenation in the Gulf of Oman for the past 43 kyr, approximately, for the first time. The three independent proxies based on bulk sediment (δ¹⁵N), lipid biomarker analysis (ratio of (lycopane + n-C₃₅)/n-C₃₁) and benthic foraminifera taxa (enhanced benthic foraminiferal oxygen index, including transfer function of Kranner et al., 2022) show a robust and mostly consistent pattern in our record. The Holocene is characterized by strong OMZ conditions and bottom water deoxygenation, which is consistent with other studies from the Arabian Sea. In contrast to other regions in the Arabian Sea, the water column and the bottom water in the Gulf of Oman were very well ventilated during the LGM or MIS 2. The well-oxygenated conditions of the Gulf of Oman lead to a highly diverse benthic foraminiferal assemblage. We attribute the good ventilation to stronger wind-induced mixing of the water column and better ventilation by oxygen-richer oceanic currents leading to stronger northward intrusion of the AAIW, corroborating other recent studies. The OMZ strength and bottom water oxygenation reconstruction during MIS 3 reveal oscillating conditions of moderately oxygenated (stadials) and deoxygenated conditions (interstadials, D–O events). Besides the prominent D–O cycles, we also found prominent high-frequency periods (570 to 710 years) during MIS 3 by using spectral and wavelet analyses.

In consideration of our results, we propose two different modes of the oxygenation status for the past 43 kyr, approximately. The first mode is a stable period of either strong OMZ and bottom water deoxygenation (Holocene) or a well-oxygenated water column and bottom water (MIS 2 including LGM). The second mode is an unstable period with oscillations between moderately oxygenated and deoxygenated conditions. This is visible in the MIS 3 period of our record with fluctuating high- and low-oxygen conditions.

Appendix A

Figure A1 Bulk and plant-wax-derived n-alkane data of SL167. (a) Total mass accumulation rates of total organic carbon (TOC MAR), (b) total mass accumulation rates of total nitrogen (TN MAR), (c) carbon preference index (CPI) of the n-alkanes C₂₇ to C₃₃ and (d) average chain length (ACL) of the n-alkanes C₂₇ to C₃₃. Diamonds show dated ages of SL167.

Figure A2 Cross plots of (a) total nitrogen (TN) and δ¹⁵N, as well as (b) the sedimentation rate and δ¹⁵N of sediment core SL167.