O₂ consumption in the Arabian Sea OMZ is affected by various factors. Here we review the role of four major mechanisms, namely: (i) the magnitude of seasonal blooms, (ii) the lateral advection of organic matter and nutrients from the productive western Arabian Sea to the central and eastern sectors, (iii) the depth of remineralization and (iv) the intensity of denitrification ( Figure 1 ).

The summer southwesterly winds drive upwelling along the coasts of Oman and Somalia whereas the winter northeasterly winds induce convective mixing in the northern Arabian Sea. Both mechanisms transport nutrients from the deep layers into the well-lit upper ocean, causing two seasonal phytoplankton blooms (Banse and McClain, 1986; Madhupratap et al., 1996; Wiggert et al., 2005; Lévy et al., 2007; Resplandy et al., 2012; Lachkar et al., 2018). Although previous studies suggest that nitrogen is generally the most limiting nutrient of biological productivity in the Arabian Sea, silicate and more importantly iron have also been suggested to limit productivity locally (Naqvi et al., 2002; Wiggert et al., 2006; Moffett et al., 2007; Wiggert and Murtugudde, 2007; Koné et al., 2009; Naqvi et al., 2010; Resplandy et al., 2011; Moffett et al., 2015; Chinni and Singh, 2022). Indeed, previous works indicate that the region off the coasts of Somalia and Oman is prone to iron limitation during both summer and winter monsoon seasons (Wiggert et al., 2006; Wiggert and Murtugudde, 2007; Moffett and Landry, 2020). A recent modeling study by Guieu et al. (2019) suggests that around half of summer primary production in the Arabian Sea is dependent on iron supply through aeolian dust deposition, confirming the critical role that iron may play in limiting biological production in the Arabian Sea. Iron limitation was also suggested to contribute to the weak and delayed response of diatoms to upwelling (Moffett and Landry, 2020). Finally, interannual variability in dust driven iron fluxes was hypothesized to contribute to interannual variability in Arabian Sea productivity (Moffett and Landry, 2020).

The lateral advection of organic matter and unutilized nutrients from the upwelling zone into the central sectors of the Arabian Sea have been shown to affect the intensity of the OMZ there (McCreary et al., 2013; Moffett and Landry, 2020). For instance, McCreary et al. (2013) investigated the role of lateral advection and transport of organic matter in controlling the oxygen distribution in the Arabian Sea. Their modeling work highlights the importance of lateral transport of organic matter from the western Arabian Sea into the central and eastern sectors of the Arabian Sea, where it enhances remineralization and contributes to lowering O₂ levels to below suboxic thresholds. A recent analysis of back-scatter data in the northern Arabian Sea also supports that cross-shelf transport of organic matter is linked to the intensification of the OMZ in the eastern Arabian Sea (Sarma et al., 2020). This transport is likely dominated by eddies. Indeed, mesoscale eddies and submesoscale filaments have been shown to play a critical role in advecting nutrients from the coastal upwelling region into the oligotrophic central Arabian Sea (e.g. Koné et al., 2009; Resplandy et al., 2011).

The depth of remineralization, defined as the depth at which sinking organic matter is decomposed back to inorganic carbon and nutrients (Kwon et al., 2009; Cavan et al., 2017), is an additional factor modulating the Arabian Sea OMZ intensity. For instance, early observational studies showed evidence that the aggregation of organic matter with lithogenic particles from rivers increases the efficiency of the export fluxes in the Bay of Bengal relative to the Arabian Sea (Nair et al., 1989; Ittekkot et al., 1991; Rao et al., 1994). In a modeling study, Al Azhar et al. (2016) have demonstrated that the shallower remineralization depth in the Arabian Sea relative to the Bay of Bengal (driven by relatively slower particle sinking speeds in the former) contributes to enhancing the OMZ there. Indeed, the absence of a major source of ballast minerals similar to the one associated with the large riverine input in the Bay of Bengal decreases the particle sinking speed, thus increasing the organic matter residence time in the OMZ layer, and hence enhancing oxygen consumption there.

Finally, the biological consumption of O₂ is reduced at low-O₂ levels as anaerobic respiration (canonical denitrification and anaerobic ammonium oxidation) is favored over aerobic respiration (Oschlies et al., 2019; Rixen et al., 2020). Thus, increasing denitrification (in case of the expansion of the OMZ suboxic core) can act as a relative source of O₂ that stabilizes the OMZ and limits its intensification (negative feedback) (Lachkar et al., 2016; Lachkar et al., 2019). An additional mechanism through which denitrification acts on O₂ is via its impact on nitrogen inventory. Indeed, denitrification generates deficits in bio-available nitrogen that may not be compensated by N₂ fixation on short timescales (Naqvi, 2008; Oschlies et al., 2019). This can lead to a reduction in biological productivity and respiration, thus limiting O₂ consumption (Altabet et al., 1995). Under such conditions, an eventual expansion of the volume of suboxic waters in the OMZ core can be accompanied by a reduction of the volume of hypoxic waters (Deutsch et al., 2011; Lachkar et al., 2016).

The ventilation of the Arabian Sea OMZ is sensitive to both basin-scale circulation as well as to local mixing processes ( Figure 1 ). Previous studies point to the Indian Ocean Central Water (IOCW) and the Persian (Arabian) Gulf Water (PGW) as two major sources of ventilation of the OMZ (Wyrtki et al., 1971; Gupta and Naqvi, 1984; Resplandy et al., 2012; McCreary et al., 2013; Acharya and Panigrahi, 2016; Lachkar et al., 2019). IOCW forms through convective mixing as Subantarctic Mode Water in the southern Indian Ocean and is advected northward into the OMZ as a part of the Somali Current. The relatively oxygen-rich PGW subducts in the northern Arabian Sea after its outflow from the Gulf and is exported into the Arabian Sea OMZ (Rixen et al., 2005; Lachkar et al., 2019; Schmidt et al., 2020; Schmidt et al., 2021). Moreover, waters from both the Red Sea and the Indonesian throughflow have been suggested to influence properties of the Arabian Sea thermocline (McCreary et al., 2013; Lachkar et al., 2019). Finally, winter convective mixing in the northern Arabian Sea is a well-known mechanism of local ventilation of the upper OMZ (Resplandy et al., 2012; McCreary et al., 2013).

Here we present the results of a quantitative analysis of the sources and timescales of the Arabian Sea OMZ ventilation based on a Lagrangian particle tracking approach (Supplementary Information). OMZ water particles are traced back to five ventilation source entries: namely the southern entry at 5°S (south), the south-eastern entry at 78°E (southeast), the Bab El Mendeb (Red Sea), the Strait of Hormuz (Arabian Gulf) and the base of the winter mixed layer (local ventilation) ( Figure 2 ). This analysis confirms that the ventilation of the OMZ is dominated by the inflow of IOCW from the south as nearly 3 4 of OMZ water parcels originate from the southern source ( Figure 2A ). The analysis of the east-west distribution of particles at the southern entry section (5°S) further indicates that they are mostly associated with waters advected along the western Arabian Sea (west of 42°E) ( Figure S1 in Supplementary Information). The remaining sources (southeastern, local ventilation and marginal seas) play a substantially smaller role. However, when considering the upper OMZ only (100-300 m), the contribution of the southwestern source to the OMZ ventilation becomes considerably smaller (45%). Conversely, the contributions of the remaining ventilation sources are twice to three times larger ( Figure 2A ). The analysis of the depth profiles of the source waters at their original locations as well as when they enter the OMZ reveals that the dominance of the southwestern source is established mostly for the deeper layers of the OMZ (300-1500 m), whereas in the upper layer (100-300m) the contribution of the remaining sources is of comparable magnitude ( Figures 2B, C ). Furthermore, the analysis of ventilation timescales associated with the different sources reveal that despite its relatively smaller contribution in terms of water volume, the Arabian Gulf has a ventilation timescale (a few months) that is around two orders of magnitude shorter than that associated with the inflow of IOCW from the south (a few decades) ( Figure 2D ). Thus, despite its modest contribution to the ventilation on decadal timescales, the Arabian Gulf is a more efficient source of ventilation on shorter timescales (months to years). This implies that the Arabian Sea OMZ is likely to respond more rapidly to changes in the Arabian Gulf water properties (e.g., changes in O₂ or water density) than to changes involving the transport of IOCW or its O₂ content ( Table 1 ).

Because of the enhanced ventilation of the western Arabian Sea from the IOCW and the PGW, the Arabian Sea OMZ is more intense in the eastern portion of the basin despite the productivity being highest along the western side (Morrison et al., 1999; Resplandy et al., 2012; McCreary et al., 2013; Rixen et al., 2014; Acharya and Panigrahi, 2016; Zhang et al., 2022). This feature, previously referred to as the eastward shift of the AS OMZ is also presumably aided by eddy-driven lateral transport of organic matter and unutilized nutrients from the western to the central Arabian Sea that contributes to enhancing respiration and lowering O₂ concentrations there below suboxic thresholds (McCreary et al., 2013). Moreover, intense vertical and lateral eddy mixing in the western Arabian Sea (Resplandy et al., 2012; McCreary et al., 2013; Lachkar et al., 2016) play a crucial role in supplying O₂ to the western Arabian Sea.

Finally, isopcynal and diapycnal mixing have been shown to strongly contribute to OMZ ventilation (Gnanadesikan et al., 2012; Lévy et al., 2022). In the Arabian Sea, model simulations suggest that in the absence of oxygen supplied by eddy mixing, the volume of the Arabian Sea OMZ would double (Lachkar et al., 2016). Changes in mixing have also been shown to dominate future projected deoxygenation in global models (Couespel et al., 2019; Lévy et al., 2022).

Paleo-reconstructions suggest important past changes in the intensity of the Arabian Sea OMZ over different periods and on timescales ranging from decades to millennia (Altabet et al., 1995; Altabet et al., 1999; Altabet et al., 2002; Gupta et al., 2003; Singh et al., 2011). For instance, Altabet et al. (1999) suggested that denitrification in the Arabian Sea was greatest during interglacial periods and probably ceased during most glacial phases. Additionally, evidence from paleo-records suggests an intensification of the OMZ core from the mid to late Holocene (Das et al., 2017). Previous studies attributed this to both large-scale ventilation changes as well as increased monsoon intensity and export fluxes (Schulz et al., 1998; Pichevin et al., 2007; Böning and Bard, 2009; Rixen et al., 2014; Das et al., 2017; Gaye et al., 2018; Joshi et al., 2021). On shorter timescales, Altabet et al. (2002) established a strong correspondence between changes in denitrification (and OMZ intensity) in the Arabian Sea and century-scale Dansgaard–Oeschger events during the last glacial period, with denitrification increasing during warm periods concurrently with the summer monsoon and productivity intensification, and decreases during cold phases. Other studies linked past OMZ intensity changes to changes in winter monsoon wind intensity y (Reichart et al., 1998; Reichart et al., 2004; Klöcker and Henrich, 2006). For instance, Reichart et al. (1998) reconstructed the fluctuations of the Arabian Sea OMZ over the past 225,000 years using multiple paleo-proxies and found the lowest O₂ concentrations to correlate with maxima in productivity and weak winter mixing. On the basis of an analysis of sediment samples collected off Pakistan in the northern Arabian Sea, Klöcker and Henrich (2006) attributed the weakening (strengthening) of the OMZ during the stadials (interstadials, Late Holocene) to enhanced (reduced) winter monsoons resulting in an invigorated (weakened) ventilation at intermediate depths. Similarly, Reichart et al. (2004) linked the intensification of the Arabian Sea OMZ during interstadials to enhanced stratification in the northern Arabian Sea during these warm periods leading to reduced winter mixing.

More recently, the analysis of historical O₂ observations in the Arabian Sea over the last few decades documents trends dominated by a decline in most of its northern sector and inconsistent changes in the central and southern parts (Rixen et al., 2020) ( Figure 3 ). For example, the analysis of global historical oxygen observations by Schmidtko et al. (2017) reveals a moderate O₂ decline in the subsurface of the Arabian Sea between 1960 and 2010. Similarly, O₂ trend analysis by (Schmidtko et al., 2017) indicates a drop in oxygen in the northern and western Arabian Sea as well as along the west coast of India over the same period. Using over 2000 O₂ profiles collected between 1960 and 2008 off the coast of Oman, Piontkovski and Al-Oufi (2015) reported a decline in O₂ in the upper 300 m in the northern and northwestern Arabian Sea, attributed to increased thermal stratification and a shoaling of the oxycline between the 1960s and 2000s. In an analysis of sea glider data and historical profiles, Queste et al. (2018) also reported an intensification of the suboxic conditions at depth in the Sea of Oman over the recent decades. Banse et al. (2014) analyzed historical O₂ measurements collected in the Arabian Sea in the 150-500 m layer between 1959 and 2004. They found no clear systematic trend across the entire basin, although O₂ was found to decline in most of the central Arabian Sea and slightly increase in the northeastern Arabian Sea and in the southern Arabian Sea ( Figure 3 ). Nitrite (NO₂ ^-) is an intermediate product formed during nitrification and denitrification. Due to the lack of oxygen required for nitrification in the upper part of the Arabian Sea OMZ, the accumulation of nitrite known as the secondary nitrite maximum (SNM) is assumed to be caused primarily by denitrification (Gupta et al., 1976; Naqvi, 1991). Banse et al. (2014) also analyzed historical trends in nitrite (NO₂ ^-) concentrations in the subsurface. They found inconsistent trends in different locations with a dominance of profiles indicating an increase in nitrite over time, suggesting a potential intensification of denitrification over the observation period. Similarly, Rixen et al. (2014) analyzed trends in nitrite in the Arabian Sea from the Joint Global Ocean Flux Study (JGOFS) measurements in 1995 and from published data prior to 1993. They found an expansion of the SNM in 1995 relative to the pre-JGOFS data with an increase in western and southern sectors, indicating a potential increase in denitrification. However, these changes could also be linked to interannual variability characterizing denitrification in the region (Lachkar et al., 2021), rather than reflecting long-term trends (Naqvi, 2019). A recent analysis of historical observations by Zhou et al. (2022) has revealed an expansion of the area occupied by all major OMZs, including the Arabian Sea OMZ, over the last 3 to 4 decades. Goes et al. (2020) reported evidence of increased winter stratification (and reduced winter convection) in the northern Arabian Sea, together with a decrease in the N:P ratios, indicative of an increase in denitrification over the recent decades. These changes were suggested to create a niche favorable for the mixotroph Noctiluca scintillans in the northern Arabian Sea (do Rosário Gomes et al., 2014; Goes et al., 2020). Finally, preliminary observations suggest recent O₂ decline in the Arabian marginal seas (i.e., the Red Sea and the Gulf) with the emergence of summertime hypoxia in the Gulf (Al-Ansari et al., 2015; Al-Yamani and Naqvi, 2019; Saleh et al., 2021; Lachkar et al., 2022) and ongoing deoxygenation in the northern Red Sea (Naqvi, 2019).

Recent modelling works have reproduced declining O₂ trends in the Arabian Sea region and investigated their potential drivers. For instance, historical deoxygenation trends were simulated in the mesopelagic zone (200-1000 m) in the northern and northwestern Arabian Sea in a global hindcast simulation covering the period between 1958 and 2014 (Buchanan and Tagliabue, 2021). The study further suggests that this O₂ loss is driven by changes in apparent oxygen utilization (AOU), likely associated with ventilation changes. These conclusions were also supported by the analysis of O₂ trends in the Arabian Sea region over the period from 1975 and 2014 using the latest Coupled Model Intercomparison Project (CMIP6) model simulations (Buchanan and Tagliabue, 2021). Finally, a strong deoxygenation in the northern Arabian Sea in the 100-1000 m subsurface layer was reported in a recent regional modeling study of the Indian Ocean covering the period from 1982 to 2010 (Lachkar et al., 2021). Causing an important intensification of the OMZ core and a significant increase in denitrification there, this O₂ loss was attributed to increased stratification and reduced ventilation as a result of the recent fast warming in the northern Arabian Sea and the Gulf (Lachkar et al., 2021).

The Arabian Sea OMZ is vulnerable to various potential future changes affecting O₂ demand and supply terms ( Table 1 and Figure 4 ). More specifically, increases in stratification are expected to lead to a significant reduction in productivity and export under future climate change (Bopp et al., 2013; Long et al., 2019; Kwiatkowski et al., 2020). While a few previous works suggested a potential future intensification of summer upwelling-favorable winds in the northern Arabian Sea (Sandeep and Ajayamohan, 2015; deCastro et al., 2016; Praveen et al., 2016), other studies indicate a future weakening of winter monsoon winds that could cause a reduction in convective mixing in the northern Arabian Sea and decrease productivity there (Vallivattathillam et al., 2017). Additionally, a weakening of the summer Walker circulation was also suggested to lead to summer productivity decline in the southern Arabian Sea (e.g., Vallivattathillam et al., submitted). Lachkar et al. (2018) highlighted that changes in OMZ intensity associated with productivity changes can develop over relatively short timescales ( Table 1 ). Although both CMIP5 and CMIP6 multi-model averages point to a future reduction in productivity in the region, individual models show important discrepancies and project changes that go in both directions (Bopp et al., 2013; Kwiatkowski et al., 2020). Additionally, these models generally do not take into account potential changes in dust deposition or effects of changes in denitrification on productivity. For instance, Guieu et al. (2019) have shown that almost half of the Arabian Sea productivity is dependent on iron supply via dust deposition. Thus, changes in this source of iron due to wind changes for instance can have a dramatic impact on productivity and OMZ (Moffett and Landry, 2020). Lachkar et al., (2016) demonstrated that a weakening of denitrification can lead to a significant increase in productivity and export, with implications for the OMZ volume.

Additional mechanisms can affect the O₂ demand in the Arabian Sea region such as the increase in nutrient loading associated with major river runoff and changes in the plankton community composition ( Table 1 ). Early studies (e.g., Naqvi et al., 2000) suggested that eutrophication associated with increased nutrient loading resulting from agricultural and urban coastal pollution may have contributed to reported intensification of the hypoxic zone along the western continental shelf of India during summer and autumn. Nevertheless, more recent work by Gupta et al. (2021) tends to imply that the anoxic zone along the west coast of India is essentially formed through the upwelling of deoxygenated waters during the summer season and is little affected by riverine input of nutrients or sewage discharge from Indian megacities. These authors argue that acute oxygen deficiency was reported in areas where no major rivers exist, and far from the major cities on the west coast of India. They also noted that hypoxia intensity has remained relatively stable in the region despite a substantial increase in anthropogenic nutrient input over the last five decades (Gupta et al., 2016; Gupta et al., 2021). Finally, changes in the size and composition of phytoplankton dominant groups can affect the efficiency of particle export and hence alter the remineralization depth and the intensity of the OMZ. For instance, the reported recent shift in the composition of winter phytoplankton blooms in the northern Arabian Sea from diatoms to Noctiluca scintillans can have repercussions on the export fluxes, as these mixotrophs are essentially grazed on by salps, which typically produce large and rapidly sinking fecal pellets (do Rosário Gomes et al., 2014). Yet, predicting future changes in plankton community structure is challenging because of the complexity and the uncertainty around the mechanisms controlling the diversity within the plankton functional groups (Henson et al., 2021). Nevertheless, recent model projections suggest a future increase in the dominance of smaller phytoplankton groups, which could imply a reduced export efficiency and hence a shallower remineralization depth under warmer climate (Henson et al., 2021).

Potential future changes in the O₂ supply terms entail changes in solubility or ventilation terms. Surface warming is expected to increase outgassing, causing a loss of O₂ that is particularly important near the surface ( Figure 4 ). In the ocean interior, changes in the intensity of ventilation can lead to changes in the OMZ intensity. For instance, an increase in winter stratification is predicted to inhibit convection and hence limit winter ventilation of the upper OMZ in the northern Arabian Sea ( Figure 4 ). This mechanism has been suggested to contribute to recent OMZ intensification in the region (Goes et al., 2020; Lachkar et al., 2021). Furthermore, the warming of the Arabian Gulf is predicted to lead to a reduction in the subduction of the Gulf water in the northern Arabian Sea (Lachkar et al., 2019; Lachkar et al., 2021). Additionally, declining levels of O₂ in the Gulf waters (Saleh et al., 2021; Lachkar et al., 2022) and potentially increasing semi-labile total organic carbon (TOC) in the Gulf may further contribute to lowering O₂ in the upper layers of the OMZ in the northern Arabian Sea (Al-Yamani and Naqvi, 2019).

In the central and southern Arabian Sea, as well as in the deeper layers of the northern Arabian Sea, water ventilation is mostly associated with the inflow of the IOCW from the south along the western Arabian Sea ( Figure 2 ; Schmidt et al., 2020; Schmidt et al., 2021). Thus, an increase in the O₂ content of these water masses is expected to lead to a weakening of the OMZ, particularly in the southern Arabian Sea (Vallivattathillam et al., in review) ( Table 1 and Figure 4 ). The enhanced O₂ content of the equatorial Indian Ocean thermocline is relatively robust in most CMIP5 and CMIP6 models (Bopp et al., 2013; Kwiatkowski et al., 2020) and is likely associated with an increase in the ventilation of the tropical thermocline (Bopp et al., 2017; Oschlies et al., 2017; Oschlies et al., 2018) aided with a decrease in O₂ biological consumption.

Finally, changes that affect O₂ gradients without changing the regional O₂ inventory can also lead to changes in the volume and intensity of the Arabian Sea OMZ. For instance, a change in mixing driven by changes in eddy activity (e.g., Beech et al., 2022) in the region could alter the intensity of the OMZ (Lachkar et al., 2016; Lévy et al., 2022). Furthermore, changes in the remineralization depth associated with changes in the particle sinking speed or remineralization rate (for instance due to warming or changes in the community composition) can also affect the depth and intensity of the Arabian Sea OMZ as has been demonstrated by previous studies (e.g., Al Azhar et al., 2016) ( Table 1 ).

The northern sector of the Arabian Sea is among the few rare oceanic regions where forced O₂ trends emerge from noise already in the present-day (2005-2014) (Rodgers et al., 2015). This is particularly true for the upper layers (200-300 m) of the northern Arabian Sea where forced negative O₂ trends, associated with an increase in water age, can already be detected (Long et al., 2016). The time of emergence of the climate change driven signal varies spatially and for different depth layers and occurs later in the central Arabian Sea (Long et al., 2016; Gong et al., 2021). Furthermore, as hypoxic and suboxic water volumes display stronger interannual variability than O₂ concentrations (Deutsch et al., 2011), the forced climate change signal is likely to emerge later for the OMZ volume and intensity.

Global model future projections suggest that while the northern Arabian Sea is likely to undergo significant deoxygenation, the central and southern parts will be affected by a weak oxygenation signal that has low detectability due to strong natural variability (Bopp et al., 2017). The deoxygenation in the northern Arabian Sea appears to be driven by an increase in AOU, induced by an important increase in the water age (Long et al., 2019). In contrast, a decrease in water age is expected in the southern Arabian Sea (as in most of the equatorial Indian Ocean). This factor, together with reduced export, is predicted to lead to an increase in O₂ levels (Long et al., 2019) ( Figure 4 ).

Nevertheless, these projections come with important uncertainties stemming from global models’ persistent misrepresentation of oxygen in the tropics in general (Oschlies, 2019), and in the Arabian Sea in particular (Schmidt et al., 2021). For instance, observations suggest a decline of O₂ in the tropics over the last 50 years, whereas global models simulate non-significant (or even positive) oxygen trends (Stramma et al., 2012a; Oschlies et al., 2017). In addition to poorly constrained biogeochemical processes, different hypotheses were proposed to explain this mismatch such as excessive diapycnal mixing, ill-represented equatorial jets and natural variability in winds not properly represented in models (Stramma et al., 2012a; Oschlies et al., 2017). In the Arabian Sea, the inaccurate representation of the OMZ in global models was linked to several factors, including an imprecise representation of marginal seas (the Arabian Gulf and the Red Sea) and an overestimation of O₂ in the IOCW source regions in the Southern Ocean (Schmidt et al., 2021). Despite an improvement in resolution and biogeochemical complexity, the biases reported in CMIP5 models are also present in CMIP6 models with the exception of two models that show significant improvement (Séférian et al., 2020). Furthermore, the projected O₂ trends in the Arabian Sea show a strong sensitivity to the employed lateral mixing coefficient with low mixing models showing a strong O₂ decline and strong mixing models displaying weak O₂ trends (Bahl et al., 2019).