In addition to the perennial ODZ in the open ocean, anoxic conditions also develop seasonally over the western Indian continental shelf. This phenomenon was first discovered in the 1950s by Karl Banse who came to India in 1958 to pursue research on tropical plankton. He was then a young man, having obtained a Ph.D. in Oceanography/Zoology from University of Kiel in 1955 followed by a post-doc at Institut für Meereskunde. At that time Cochin (now Kochi) was the most important centre for marine research in India, and Karl was based at what was then a sub-station of the Central Marine Fisheries Research Institute (CMFRI) in Ernakulam. It must have been quite a change for Karl moving from Kiel to Cochin. In addition to managing with a rudimentary infrastructure for marine research, which India had at that time, the hot and humid climate of Cochin must have been unbearable to him (there was no air conditioning then!). What was worse is that he soon realised that the sea off Cochin was not what he had planned to work on; it was not a typical tropical region: We now know that the seasonally reversing West India Coast Current flows poleward (Fig. 2.1), opposing the wind, during the winter, when the sea surface temperature is the highest; the surface temperature is at its minimum in summer. Karl learnt this only after he had settled in Cochin. Anyway, now that he was in India, he tried to make the most out of the opportunity.

Karl designed a time series study off Cochin, carrying out monthly observations at two stations, including temperature, salinity and oxygen in addition to the plankton collections he was mainly interested in. His study spanned well over a year. This was, in fact, the first systematic time series study of Indian coastal waters, which provided the baseline data that have now become even more valuable as we are struggling to evaluate the changes brought about by human activities (Gupta et al., 2016). Karl’s seminal work not only provided the first account of seasonal upwelling along the west coast of India, but it also revealed the associated low oxygen conditions. Unfortunately, he was ill advised to publish his results in a local biology journal (Banse, 1959), and in the journal’s very first issue! For this reason, his work went largely un-noticed (in fact, for many years after joining NIO, I was not aware of it). Another paper published in Nature (Carruthers et al., 1959) at about the same time with fewer observations reporting prevalence of low oxygen conditions in coastal waters off Bombay (now Mumbai) drew a lot more attention.

After staying in India for two years, Karl took up a faculty position in University of Washington. However, he remains interested in the Arabian Sea to this day, continuing to write some outstanding papers (Banse, 1968, 1984; Banse and McClain, 1986; Banse et al., 2014). Karl has mentored many younger Indian researchers, and I am privileged to be one of them (Fig. 5.1). Before the advent of the Internet, he used to regularly update me by post about new publications that he thought would be of interest to me.

Karl takes strong exception to the fact that many people do not count the southwest coast of India among the eastern boundary upwelling systems. During the SWM, the region does behave like an eastern boundary upwelling system with an equatorward surface flow, a poleward undercurrent and coastal upwelling (Shetye et al., 1990). One major characteristic of this region, however, is that upwelling is not entirely forced by local winds. It is far less intense than upwelling in the western Arabian Sea. It is also confined to belt that is just a few tens of kilometres wide, except in the south (off Sri Lanka and Kerala) where the winds are upwelling favourable and the process is more intense. Another unique feature of this region is that, unlike the four major eastern boundary upwelling systems in the Pacific and the Atlantic oceans, the coastal zone off western India receives heavy rainfall during the SWM that is orographically forced by the Western Ghats (a chain of mountain ranges bordering the western coastal plain on its east). The cold, saline upwelled water is usually not allowed to break to the surface as it gets capped by a thin (5–10 m) warm, low salinity lens produced by freshwater inputs. This creates extremely strong thermohaline stratification very close to the sea surface (Banse, 1968; Naqvi et al., 2000, 2006a,b,c, 2009). As a result, although primary production is not very high, oxygen depletion below a shallow pycnocline is extremely severe.

For quite some time after Banse’s seminal work, not much attention was paid to coastal upwelling and associated oxygen deficiency over the Indian shelf, with the exception of intensive surveys carried out during 1971–75 under the UNDP/FAO supported Integrated Fisheries Project (IFP). These surveys involved repeated occupation of several coast-perpendicular sections (Naqvi, 2006). The scarcity of data during the SWM period was not only because of the difficulty in making observations due to rough seas; it was also due to a lack of interest: frankly we prided ourselves to be deep sea oceanographers after we had acquired an ocean going vessel. It was not until the 1990s that interest in coastal upwelling was rekindled. In the summer of 1995, we secured ship time to investigate this phenomenon in some detail, and the results were so interesting that our focus shifted to shelf anoxia. In 1997, we set up a coastal quasi-time series station off Goa. We named it CaTS (Candolim Time Series) as it was located off Candolim, a coastal village in North Goa that is famous for its beach. Over the past 25 years, my colleagues at NIO have been monitoring this site and the data generated have provided very useful insights into coastal processes, the development of shelf anoxia, and how they are being impacted by climate change (e.g., Naqvi et al., 2006b,c, 2009).

Our work established the seasonal hypoxic5 zone over the Indian shelf (Fig. 5.2) to be the largest shallow water hypoxic system in the world, covering an area about 10 times the area of the famous “dead zone” in the Gulf of Mexico (Rabalais et al., 2002). The sulfidic conditions observed by us in the bottom water over the inner shelf had not been documented previously, including during the above mentioned sustained surveys conducted in the 1970s under the IFP, even though near zero oxygen concentrations have been known to persist since the 1970s, as stated earlier. Therefore, it was suggested that such conditions could arise from human induced intensification of oxygen deficiency (Naqvi et al., 2000; 2006c). Subsequently, CaTS data confirmed frequent occurrence of sulfidic conditions in late summer/early autumn, including the most recent observations in 2022 (Damodar Shenoy, personal communication). However, the data do not show a clear increase in H2S since the late 1990s. The inter-annual variations in the intensity of oxygen deficiency are large, but irregular (Naqvi et al., 2009). For example, conditions were extremely severe in 2001, when the entire shelf off Goa was covered by sulfidic bottom water, drastically (negatively) impacting demersal fish landings all along the Indian west coast. The exact driver of these changes has not been identified, but it is probably a basin scale phenomenon. From a coupled physical-biogeochemical regional simulation with 1/4° resolution over the 1960–2012 period, Vallivattathillam et al. (2017) found oxycline fluctuations over the shelf to be strongly influenced by the Indian Ocean Dipole with positive dipole events associated with weaker anoxia.

Despite strongly reducing conditions prevailing in bottom waters over the Indian shelf, this system exhibits some distinct differences from the two other coastal eastern boundary environments where sulfate reduction has also been found to occur in the water column; off Namibia in the Benguela Current system (Brüchert et al., 2006; Lavik et al., 2009) and off Peru in the Humboldt Current system (Schunck et al., 2013). These differences are as follows.

(1) Seasonal ODZ over the Indian shelf is not contiguous with the open ocean ODZ, unlike in the other eastern boundary systems (Peru and Namibia), with the undercurrent keeping the waters just off the shelf slightly oxygenated, enough to prevent the onset of nitrate reduction (Naqvi et al., 2006b). However, the undercurrent, which is invariably rich in oxygen to begin with, also occurs in other eastern boundary upwelling systems, but in these other systems, dissolved oxygen gets quickly exhausted due to high demand such that in areas like the ETSP, the most intense nitrogen loss actually occurs within the undercurrent itself (Codispoti et al., 1989).

(2) While H2S builds up to fairly high levels (up to 14.1 µM) at very shallow depths over the Indian Shelf (in 2000 it occurred in concentrations ~5 μM just 5 m below the surface off Mangalore), its accumulation in porewaters of sediments of the inner shelf is not very large (<3 μM; Naik et al., 2017). This may in part be due to the removal of sulfide as insoluble FeS in sediments owing to large Fe supply from land. Moreover, sedimentary sulfate reduction rates in the region (0.066–0.46 mol m−2 yr−1) are much lower than those reported from other coastal environments experiencing upwelling (Naik et al., 2017). The lack of free sulfide build-up in sediments explains the absence of sulfur bacteria such as Thiomargarita, Thioploca and Beggiatoa that are known to oxidise sulfide with nitrate and are commonly found off Peru-Chile and Namibia (Fossing et al. 1995; Schulz et al., 1999).

(3) As compared to other systems, especially the Namibian shelf where methane (CH4) concentrations exceeding 5 μM have been observed (Brüchert et al., 2006), CH4 accumulation in the anoxic bottom waters over the Indian shelf is relatively modest: concentrations exceeding 100 nM are extremely rare (Jayakumar et al., 2001; Shirodkar et al., 2018; Sudheesh et al., 2020). Anaerobic oxidation of CH4 (e.g., by sulfate and nitrite) has been invoked to explain the relatively low CH4 values in anoxic waters over the Indian shelf (Sudheesh et al., 2020). There is no information on the extent of CH4 oxidation by sulfate, but its oxidation by nitrite does not appear to be significant in anoxic waters over the Indian shelf, as will be discussed later. It is likely that a more subdued CH4 production through non-competitive methanogenesis rather than enhanced methanotrophy is mainly responsible for the lack of large CH4 accumulation. Moderate CH4 levels (5–62 μM in the upper 17 cm) in porewaters at the CaTS location are consistent with this view (Araujo, 2018).

The above mentioned differences between the Indian shelf with similar systems off Peru-Chile and Namibia can be explained by a smaller organic loading due to a much lower primary production over the Indian shelf (Pitcher et al., 2021). This is also reflected by the lower organic carbon content of sediments over the inner Indian continental shelf (<4 %; Shirodkar et al., 2018) than off Namibia and Chile (10–40 %; Fossing, 1990; Fossing et al., 1995; Brüchert et al., 2006). It would thus appear that the relative importance of biology and physics in the formation of shallow water anoxic zone over the western Indian shelf may be different from that in the other two eastern boundary coastal anoxic zones. That is, large organic loading may be the main driver of anoxia off Namibia and Peru/Chile. Over, the Indian shelf, on the other hand, where the lower productivity allows the undercurrent to remain slightly oxygenated, a longer residence time of upwelled water may be a more important contributor to the development of anoxic conditions (Naik et al., 2017; Pitcher et al. 2021).

Perhaps the most important aspect of biogeochemistry of the coastal ODZ is an accumulation of N2O at hundreds of nM level. Such high concentrations were unprecedented at that time (Naqvi et al., 2000), but similar build-up of N2O has since been observed also in the ETSP off Peru (Arévalo-Martínez et al., 2015).

An example of this phenomenon is provided in Figure 5.2, which shows vertical sections of several physico-chemical variables including N2O off Goa in October 1999. Clear zonation in redox conditions over the shelf is usually seen around this time off the Konkan and northern Malabar coasts: slightly oxic conditions prevail over the outer shelf; denitrifying conditions are seen over the mid-shelf and sulfidic conditions develop over the inner shelf. The highest N2O concentrations mostly occur at mid-depths in association with very high nitrite concentrations (reaching up to 16 µM) and very low nitrate concentrations (Naqvi et al., 2000, 2006a,b,c, 2009). N2O concentrations decrease in sulfidic waters, apparently due to N2O reduction to N2. Transient N2O accumulation to micromolar levels was also seen in anaerobic incubations of water samples (Naqvi et al., 2000). These results strongly point to N2O production through denitrification over the Indian shelf, in contrast to the net consumption within the SNM in the open ocean, as discussed above. The greatly enhanced N2O production was attributed to the regulation of N2O reductase activity by frequent oxygenation of water arising from turbulence in shallow, rapidly denitrifying systems (Naqvi et al., 2000). The large production of N2O at shallow depths maintains high N2O concentrations in the surface layer (5–436 nM, mean 37.3 nM), in turn resulting in a large atmospheric efflux (0.05–0.38 Tg N2O yr−1) (Naqvi et al., 2006c).

As already mentioned in Section 3, N2O over the shelf exhibited a different stable isotopic composition in the coastal ODZ than in the open ocean, which was attributed to greater exchange of common intermediates, especially nitrite, having variable isotopic composition between nitrification and denitrification over the shelf. An alternative explanation could be that fractionation patterns of isotopes in the two systems may be different. Measurements of isotopic composition of nitrate also revealed different degrees of heavy isotope enrichment between the coastal and open ocean ODZs (Naqvi et al., 2006b; Bardhan and Naqvi, 2020). The relatively limited data used by Naqvi et al. (2006b) from two stations (one each off Goa and Mangalore) provided the δ15N of the combined nitrate and nitrite pool. The overall range of measured values was 3.43–22.5 ‰, but with a large geographical variability. The δ15N varied within a surprisingly narrow range off Mangalore (3.43–7.41 ‰) despite large nitrate deficits occurring below the shallow pycnocline (up to ~15 µM). The isotopic fractionation factor computed assuming Raleigh distillation was 7.21–7.7 ‰, much lower than estimates for the open ocean ODZs: 27 ‰ by Brandes et al. (1998) for the ETNP and the Arabian Sea and 25.6 by Gaye et al. (2013) for nitrate + nitrite for the Arabian Sea. Relatively low fractionation factors (10–15 ‰) have been reported for denitrification at the cellular level based on experiments with cultured denitrifying bacteria (Kritee et al., 2012). However, even these values are significantly higher than those derived by us from the Raleigh model over the Indian shelf.

A larger data set on dual isotopic composition of nitrate with better geographical coverage (from Calicut in the south to Mumbai in the north) has been generated more recently by Bardhan and Naqvi (2020). The range of values reported by these authors is wider (1.9–30.7 ‰ for δ15N and 1.5–36.2 ‰ for δ18O). The slope (0.9) of the linear relationship between δ18O and δ15N deviated significantly from the expected 1:1 trend, implying that other processes such as nitrite oxidation to nitrate were important in determining the isotopic composition of nitrate. However, it must be noted that nitrite present in the samples was removed in this study before isotopic analysis. Since nitrite in the ODZs is highly 15N-depleted due to the inverse isotope effect associated with nitrite oxidation to nitrate (Casciotti, 2009; Casciotti et al., 2013; Gaye et al., 2013), the δ15N values reported by Bardhan and Naqvi (2020) should be higher than those reported by Naqvi et al. (2006b). Nonetheless, the fraction factor estimated by these authors was also quite low (6.1 ‰). Assuming that the low fractionation value arose from denitrification in the sediments that has only a small isotope effect, if any (Brandes and Devol, 2002; Zhang et al., 2020), Bardhan and Naqvi (2020) estimated that sedimentary denitrification could account for about half of the nitrogen loss over the Indian shelf. In addition to the isotopic influence of sediment denitrification, Naqvi et al. (2006b) mentioned other possible reasons for the apparently low fractionation factor for denitrification in the coastal ODZs (not only in the Arabian Sea but in many several other areas as well; Bardhan and Naqvi, 2020, and references therein). These are (1) the extent of nitrogen isotope fractionation in coastal ODZs may actually be lower than in the open oceanic ODZs, (2) low fractionation values could be an artefact of mixing between anoxic waters (that would have lost all nitrate and nitrite) and freshly upwelled waters, and (3) processes other than heterotrophic denitrification (e.g., anammox, DNRA coupled to anammox, and autotrophic denitrification) may also be important in the reduction of nitrate.

More recent work has enabled us to estimate the rates and evaluate relative importance of various processes that reduce nitrate in anoxic waters (Sarkar et al., 2020). Rates of denitrification, anammox and DNRA were measured using the isotope dilution technique (Holtappels et al., 2011) at a number of stations located off Goa, Karwar and Mangalore during the anoxia season (late August to late September) in three consecutive years (2008, 2009 and 2010). Addition of 15NH4+ resulted in low to moderate production of 29N2, indicating anammox in about half of the incubations of anoxic waters. In contrast, incubations with 15NO2 led to production of 30N2 in almost all incubation experiments (e.g., Fig. 5.3), yielding rates of denitrification (up to 10.03 ± 1.70 µmol N2 l-1 d-1) that are among the highest reported from any aquatic environment. As stated earlier, the production of 30N2 indicates denitrification by microbes. These microbes could be either heterotrophic (those which oxidise organic matter using nitrate/nitrite as electron acceptors) or autotrophic (those which oxidise reduced inorganic species such as H2S with nitrate/nitrite to synthesise organic matter (Lavik et al., 2009). Our results cannot differentiate between these pathways. However, since nitrate was often present in anoxic waters, as also evident by substantial production of 29N2 through pairing of 15N and 14N of nitrate/nitrite, it appears that heterotrophic denitrification should be more important. When detected through production of 15NH4+ in incubations involving 15NO2-, DNRA rates were much lower than denitrification rates. Time series data from fixed coastal sites show that, following the onset of dentrification, it takes about a month for nitrate with an initial concentration of ~24 µM to be fully consumed (Naik, 2003). The measured denitrification rates are even higher than derived from these data, requiring source(s) of nitrate in addition to inputs by upwelling. These results clearly show that denitrification is the dominant pathway of combined nitrogen loss in the seasonal ODZ over the Indian shelf.

Another possible pathway of N2 production is the nitrite dependent anaerobic methane oxidation (n-damo), the significance of which in nitrogen loss from marine environments is not fully known (Thamdrup et al., 2019; Rogener et al., 2021). In order to investigate such a linkage between denitrification and methanotrophy, we also carried out incubations with 15N-labelled NO2- in the presence of CH4. Addition of CH4 had no discernible effect on labelled N2 production (Fig. 5.3). This may be because of low ambient CH4 levels as a result of which microbes that mediate CH4 oxidation by nitrite, detected elsewhere in the ocean (e.g,.Thamdrup et al., 2019), may not be present over the Indian shelf.

Large spatio-temporal variations (including interannual changes) in redox conditions seem to occur in this region, controlling rates of different nitrogen transformations. Higher denitrification rates were recorded off Mangalore and Karwar as compared to Goa. The rates also decreased rapidly offshore from the inner shelf (Fig. 5.3). Extrapolation of the average N2 production rates (2.45 ± 0.6 µmol N2 l–1 d–1 through denitrification and 0.06 ± 0.004 µmol N2 l–1 d–1 through anammox) over the volume of anoxic waters (1.2–3.6 × 1012 m3) yielded estimates of annual nitrogen loss ranging from 3.70 to 11.1 Tg N, roughly three times the previously reported estimate (1.3–3.8 Tg yr–1). Thus, the shallow ODZ may account for a significant fraction of the total nitrogen loss from the Arabian Sea (Sarkar et al., 2020).

Similar anoxic conditions are not known to develop over the continental shelves in the Bay of Bengal (Naqvi et al., 2006a). This is surprising as this region receives enormous amounts of freshwater by numerous rivers that drain thickly populated and intensely cultivated lands with large utilisation of synthetic fertilisers (Naqvi, 2008). So, the riverine inputs of nutrients (nitrogen and phosphorus) into coastal waters of Bay of Bengal are expected to be very high. Employing a model called “Global NEWS” (Global Nutrient Export from WaterSheds; Mayorga et al., 2010), Pedde et al. (2017) estimated that 7.1 Tg of N (4.1 Tg of DIN) and 1.5 Tg of P were brought to the Bay of Bengal by rivers in the year 2000. These high fluxes constitute significant fractions of the total amounts of chemical fertilisers used globally for agriculture (~82 and 14 Tg yr-1 in the same year)6. Future projections showed that while phosphorus fluxes may not change much, those of nitrogen will rise to 8.6 Tg N yr-1 (6 Tg DIN yr-1) by 2050. The Ganges-Brahmaputra and Irrawaddy river systems are the main contributors to these fluxes, and both these river systems discharge onto wide continental shelves. Thus, coastal “dead zones” such as those found in the Gulf of Mexico and Black Sea (Rabalais et al., 2010) are expected to form off the mouths of these rivers, which is not the case (Naqvi et al., 2006a). There is a report by Satpathy et al. (2013) on the occurrence of nearly anoxic water (Winkler O2 ~4.5 µM) over the shelf (at 59 m depth) off the southeast coast of India in September 2010 (the upwelling season), which was attributed to land based pollution; however, an examination of data presented by these authors points to an offshore origin of these waters (as discussed later, subsurface waters in the Bay of Bengal are also severely oxygen-depleted, although not anoxic). Thus, the absence of naturally formed coastal ODZs is most likely due to weaker upwelling in the Bay of Bengal as compared to the Arabian Sea. The absence of large anthropogenically driven coastal dead zones in the Bay of Bengal strongly indicates that the model-derived nutrient inputs may be too high (Naqvi, 2008; Krishna et al., 2016). It seems that the terrestrial ecosystems efficiently remove these nutrients, particularly nitrogen (Naqvi et al., 2018). This will be discussed in some detail in Section 9.

Hypoxia is traditionally defined by the threshold oxygen concentration of 2 mg l-1 (~1.4 ml L-1, 62.5 µM), below which animal behaviour is believed to be affected. However, this threshold is known to vary greatly among various groups of marine organisms and even within the same group (Vaquer-Sunyer and Duarte, 2008).