Jambeck et al. (2015) estimated the global amount of plastic waste that entered the ocean in 2010 from coastal populations living within 50 km of the coastline. Based on a fixed percentage of mismanaged plastic waste entering the oceans (15 % for the low-range estimates and 40 % for the high-range estimates), they estimated that between 4.8 and 12.7×106 t of plastic entered the global oceans. However, it is likely that the estimated amount of plastic waste entering the ocean by Jambeck et al. (2015) for Sri Lanka is incorrect. Jambeck et al. (2015) based their estimate on a reported 5.1 kg of municipal solid waste generated per person per day in Sri Lanka (Hoornweg and Bhada-Tata, 2012). The updated report by Kaza et al. (2018) and dataset available through the World Bank (2021) (“What A Waste Global Database”) indicate that only 0.34 kg of municipal solid waste is generated per person per day in Sri Lanka; this number is also more in line with the amount of waste generated in other developing countries. Using this correction, the amount of plastic waste entering the ocean from Sri Lanka through coastal populations is estimated to be between 0.021 and 0.057×106 t in 2010, instead of between 0.24 and 0.64×106 t as reported in Jambeck et al. (2015). Using this corrected number, around 15 % of global ocean plastic entered the IO directly through coastal sources (Fig. 1a; showing corrected waste input for Sri Lanka).

Lebreton et al. (2017) and Schmidt et al. (2017) estimated the amount of plastic waste entering the oceans through rivers. Similar to Jambeck et al. (2015), their estimates are based on a percentage of mismanaged plastic waste and include the influence of river catchment geography and river discharge. In addition, these estimates were calibrated based on available measurements of river plastic debris globally, ranging between sizes of 0.3 mm to 0.5 m. Lebreton et al. (2017) estimated that between 1.15 and 2.41×106 t of plastic waste enter the global ocean per year, up to 20 % of which enters the IO (Fig. 1b). Schmidt et al. (2017) estimated that between 0.47 and 2.75×106 t of plastic waste enter the global ocean per year, of which around 15 % enters the IO. More recently, Meijer et al. (2021) estimated that between 0.80 and 2.7×106 t of macroplastics (defined by Meijer et al., 2021, as larger than 5 mm) enter the global ocean per year. In this estimate, Meijer et al. (2021) took into account the spatial variability of mismanaged plastic waste generated within a river basin as well as more advanced climate and terrain characteristics than considered in the estimates of Lebreton et al. (2017) and Schmidt et al. (2017). They calibrated their estimates based on visual sampling of macroplastics at river mouths around the world.

The estimates of the amount of plastic waste entering the oceans through rivers by Lebreton et al. (2017), Schmidt et al. (2017), and Meijer et al. (2021) agree relatively well with each other. Jambeck et al. (2015) estimated that the amount of plastic waste entering the oceans through coasts is an order of magnitude higher. In even starker contrast, Weiss et al. (2021) re-evaluated the estimates of Lebreton et al. (2017) and Schmidt et al. (2017) and suggested that only 6100 t of microplastics (defined by Weiss et al., 2021, as smaller than 5 mm) enter the ocean through rivers each year, which is 2 to 3 orders of magnitude smaller than previous estimates. These differences highlight the extreme uncertainty involved in determining the amount of plastic waste entering the ocean from land-based sources. These estimates are based on few measurements of plastics entering the ocean (in the case of Jambeck et al., 2015, only on data from the San Francisco Bay; in the case of Lebreton et al., 2017, Schmidt et al., 2017, Meijer et al., 2021, and Weiss et al., 2021, on 30 to 340 samples from 13 to 89 rivers around the world). None of these samples was from IO rim countries or in rivers that empty into the IO. Expanding on these datasets will likely help improve these estimates, especially for the IO. However, as Weiss et al. (2021) demonstrate, to reduce extreme errors it is essential to use comparable sampling methodologies and collect data on the amount of plastics sampled and their weight. Furthermore, Meijer et al. (2021) emphasize the importance of sampling plastics at river mouths to get a more reliable estimate of the amount of plastic that actually enters the ocean. However, sampling plastics further upstream in addition to the river mouth can help improve models of the probability for plastic to reach the ocean from inland areas.

Based on the currently available estimates, the largest coastal and riverine plastic sources in the IO are in the Northern Hemisphere around the Bay of Bengal and on the eastern side of the Arabian Sea (Fig. 1). The input of plastic waste from rivers mainly depends on the river discharge, which varies seasonally. The highest river discharges occur during the wet season, during the boreal summer in the northern IO. Lebreton et al. (2017) estimated that plastic waste input from rivers in the IO peaks in August. In the Southern Hemisphere, the largest coastal and riverine sources of IO plastic waste are from Indonesia (Fig. 1).

Plastic waste can also enter the ocean directly from ocean-based sources such as the fishing industry, commercial and recreational shipping, and offshore platforms. In 1988, the International Convention for the Prevention of Pollution from Ships (MARPOL) prohibited waste dumping from vessels. However, accidental losses and illegal dumping still contribute to plastic debris (Richardson et al., 2021; Stelfox et al., 2020a; see Sect. 7 for a case study of a recent incident off the coast of Sri Lanka). Abandoned, lost, or discarded fishing gear (ALDFG) by the fishing industry can produce large quantities of plastic debris, i.e. monofilament lines and nets primarily made from synthetic materials (Sheavly and Register, 2007; Bond et al., 2012). Since 2000, around 13 % of the global marine capture of fishes originated from the IO region (Pauly and Zeller, 2016). It is, therefore, likely that ALDFG is a significant source of plastic debris in the IO. Estimates of plastic waste entering the ocean from fishing vessels (e.g. ghost nets) or offshore platforms do not currently exist.

Finally, plastic waste originating from southeast Asia can also be transported to the IO by the Indonesian Throughflow (Sect. 4.1). However, these sources are currently undocumented and need to be further investigated.

Plastics entering the ocean from land-based sources are subject to many physical processes on the continental shelf before they are transported into the deeper ocean or beach on coastlines (Fig. 3). Several studies investigating plastic debris in the IO have mainly focussed on the biology and biogeochemistry (Roy et al., 2015; Sarma et al., 2015; Sarkar et al., 2018), and only a few studies have addressed the transport of plastics between inshore and offshore regions. Physical processes that lead to convergent flows, where ocean currents flow towards each other, are one of the most important features for the transport of buoyant plastics. Convergent flows promote downwelling, which causes the accumulation of buoyant plastic debris along the convergent flow boundary defined as the front (Fig. 3b).

Convergent flows occur along ocean fronts, defined as the boundary between two distinct water masses or a region where the rate of change in selected physical properties is much greater than the surrounding areas (Bowman and Esaias, 1977; Belkin and Cornillon, 2007; D'Asaro et al., 2018). Fronts occur from hundreds of metres to many thousand kilometres, and some are short-lived, but most are quasi-stationary and emerge at the same location on seasonal timescales (Belkin and Cornillon, 2007). Ocean fronts are considered hotspots of marine life (Belkin et al., 2009) because flow convergence at fronts channels nutrients towards the fronts and stimulates increased production at different trophic levels (Owen, 1981; Baltar et al., 2016; Sarma et al., 2015; Woodson and Litvin, 2015; Sarkar et al., 2018). Aggregations of plankton, larvae, and eggs are often found on the surface (Fig. 3b). Here, as the water sinks at the front due to convergent flow, buoyant material will remain at the surface (Miyao and Isobe, 2016). Predators such as fish and higher-order biota are found above and beneath the front. Due to the accumulation of buoyant debris, including plastics, along fronts, the risk of marine organisms interacting with plastics is high.

There are many types of fronts that are formed through different physical processes. They include river plumes (Luketina and Imberger, 1989; O'Donnell et al., 1998; Karati et al., 2018; Cole et al., 2020), shelf-sea tidal fronts (Simpson and Hunter, 1974; Nahas et al., 2005; Sharples and Simpson, 2020), shelf break fronts (Sharples and Simpson, 2020), upwelling fronts (Brink, 1987), and fronts formed through interaction between flow and topography (Wolanski and Hamner, 1988; Pattiaratchi et al., 1987). Fronts appear as a visible band along the sea surface with differences in temperature and salinity (and, as a result, density) on either side of the front. All these different types of fronts occur in the IO and are extremely important for the accumulation of plastics.

Buoyant plastics entering the IO from land-based sources such as rivers may accumulate at many different frontal systems before being transported offshore (Fig. 3). During the southwest monsoon season (Sect. 4.2), coastal fronts are generally formed by the interaction of two different water masses moving in opposite directions, where the low-saline runoff waters from rivers are trapped as a salinity front by the high-saline upwelling waters. Examples of such fronts are along the southern coast of Sri Lanka (de Vos et al., 2014). As a result, buoyant plastic debris is trapped in the frontal zone along these water masses (Naidu et al., 2021), referring to locations of future investigations. Naidu et al. (2021) also found that the abundance of plastics, mainly fibres and fragments, was an order of magnitude higher in the frontal zone compared to outside of it off the west coast of India. Similarly, Hajbane et al. (2021) found that close to an offshore reef located in the eastern IO, concentrations of plastics along a coastal front were up to 2 orders of magnitude higher than in surrounding waters. Frontal systems can also be eroded under storm conditions (e.g. Chen et al., 2020), and as a result, plastics accumulated along coastal fronts may beach on coastlines.

Compared to other ocean basins, the IO has several unique topographic, atmospheric, and oceanic features that influence the transport of buoyant plastics. To the north, the IO is bounded by the Indian subcontinent, to the west by Africa and the Middle East, to the east by Indonesia and Australia, and in the south by the Southern Ocean. There are two distinctive tropical basins in the northern IO: the Arabian Sea in the west and the Bay of Bengal in the east. The IO is connected to the Pacific Ocean through the Indonesian Archipelago that allows tropical water inflow from the Pacific Ocean. In the southern IO, the eastern boundary formed by the African continent does not extend beyond 35∘ S, which allows for a connection between the southern IO and the South Atlantic Ocean (Gordon, 2003; Lutjeharms, 2006) and facilitates plastic transport between these oceans leading to debris exchange between the garbage patches (van der Mheen et al., 2019; Sect. 4.3). Because the subtropics in the northern IO are covered by land mass, there is no subtropical gyre. Instead, temperature differences between the northern land mass and the ocean drive the monsoon system, which dominates the atmospheric and oceanic dynamics in the region. These dynamics determine the transport of buoyant plastics in the IO (van der Mheen, 2020) and are described in more detail here.

Approximately 100×106 t of marine fishes have been landed globally each year since 2000, approximately 13 % of which originated from the IO (Pauly and Zeller 2016). The IO supports a range of marine ecosystem services including numerous and expansive fishing ventures with 7 %, or 1×106 t, of IO landings considered to be large commercial pelagic fishes, e.g. tuna and billfishes. The commercial pelagic fisheries of the IO alone are worth over USD 1300 million annually. ALDFG becomes “ghost gear”, which continues to entangle and injure or catch wildlife, for decades or centuries as it slowly degrades, increasing the pressure on marine wildlife. Entangled and caught fauna of ALDFG are initially sea turtles, marine mammals, sharks, and large predators for the first few months, later including crustaceans when gear fragments into smaller parts (Macfadyen et al., 2009; Wilcox et al., 2015; Stelfox et al., 2016).

ALDFG affects the tourism industry through a multitude of factors, including the removal of iconic marine species and wildlife in general and contributes up to 90 % of shoreline debris and degrading of the perceived beauty of an area (Gunn et al., 2010; Lachmann et al., 2017). Ghost gear is also considered to be destructive to both natural habitats, i.e. coral reefs, and manufactured objects, i.e. boats (Laist and Liffman, 2000; Gunn et al., 2010). These devaluations decrease benefits from coral reef tourism (USD 36 billion) (Spalding et al., 2017) and global exports of aquarium fish (USD 15 billion) (Raja et al., 2019). Data from genetic analyses of Olive Ridley turtles entangled in ghost nets in the Maldives showed that the individual turtles originated from populations nesting in India and Sri Lanka (Stelfox et al., 2020b). This shows that impacts on charismatic marine species that drive tourism can simultaneously impact multiple economies in the IO rim. Whilst regional data are difficult to obtain, 60 % of aquarium fish are exported from developing nations (Raja et al., 2019). Thus, the adverse cultural, societal, and sustenance impacts of ghost fishing may disproportionately affect IO rim developing nations (Wong, 2011; Guillotreau et al., 2012; Lachmann et al., 2017).

Many of the countries bordering the IO are small island developing states (SIDSs) where more than 70 % of the population lives along the coast and is highly dependent on marine ecosystem services (Canales et al., 2017). Recent interviews of fishers by Richardson et al. (2021), which included fishers from Indonesia along the IO rim, showed that the main reasons for gear loss reported were bad weather and interactions with wild life, respectively. Illegal and deliberate gear discard on the other hand was reportedly low. Furthermore, over half of the fishermen interviewed across the world reported being “concerned” or “very concerned” about ALDFG, whereby economic losses scored highest (54 %) as an issue of concern followed by environmental harm (41 %). The reported loss prevention strategies that scored highest were gear maintenance and training crew in gear management, which provide clear avenues for targeted programmes to educate and raise awareness around ALDFG in low-income fisheries, such as in many IO rim economies (Richardson et al., 2021). A single ghost net may decrease fishing revenues by USD 20 000 annually (Lachmann et al., 2017) and can consume as much as 30 % of landings within a single fishery (Laist and Liffman, 2000). Although ALDFG is thought to diminish in catching efficiency over time, the catching potential of gill nets was estimated to be 20 %–30 % of original levels for long periods and at least 5 % after more than 2 years of abandonment (Macfadyen et al., 2009; Stelfox et al., 2016). Some catch rates were estimated to be 81 kg d-1 for tangle nets over 1.5 years, an average of 64 kg d-1 for deep-water gill nets over 45 d and 0.6 kg d-1 for traps left for 3–6 months (Macfadyen et al., 2009).

As described in Sect. 4.3.2, multiple studies in the IO have found ingested plastic in different marine species. However, as described in the review by Law (2017), confirmed cases of ingestion by themselves do not determine the impact of plastic debris ingestion. Even confirmed cases of individual deaths are not readily translated to impact on species or ecosystem levels. Recently, Roman et al. (2021) determined which types of plastic debris contribute the most to mortality in marine megafauna (cetaceans, pinnipeds, sea turtles, and seabirds). In the IO plastic ingestion has been studied in sea turtles (Hoaru et al., 2014) and seabirds (Cherel et al., 2017; Cartraud et al., 2019).

For turtles, Roman et al. (2021) concluded that film-like plastics, plastic fragments, and hard plastics were most likely to cause death. Although Hoarau et al. (2014) did not find a relationship between the mortality of loggerhead turtles in the southwest IO and the type of plastic debris they had ingested, they did find a correlation between the length of debris and mortality. They also suggested that loggerhead turtles are have a relatively high tolerance to the ingestion of plastic debris, only 9 % of approximately 450 dead turtles were thought to have been killed directly by the ingestion of plastic.

Roman et al. (2021) found that ingestion of hard plastics was most likely to cause death for sea birds. Cherel et al. (2017) sampled the stomach contents of live wandering albatross chicks on the Kerguelen and Crozet Islands in the IO. They found that approximately 50 % of chicks had ingested plastic fragments. However, the plastic content was relatively low and Cherel et al. (2017) suggested that this was unlikely to cause any significant deleterious effects. On Réunion and Juan de Nova, Cartraud et al. (2019) determined the plastic ingestion of deceased petrels and shearwaters. They could find no correlation between the muscular condition of the birds and the amount of plastic ingested. However, all of the birds likely died as a result of injury after being attracted and disoriented by urban light.

Ocean currents around Sri Lanka during May are such that they flow south along the west coast of India (West Indian coastal current; Fig. 4), cross the Gulf of Mannar, and flow south and east along the west and south coasts of Sri Lanka, respectively (see also de Vos et al., 2014; Su et al., 2021). There is a recirculation, the SLD, to the east of the island (Figs. 4 and 7). The SLD connects the water from the west coast to the east coast, with currents flowing south along both coasts (Fig. 7). During the initial period of the incident, there were strong onshore winds (>10 m s-1) and swell waves (height ∼2 m). This was associated with the beginning of the southwest monsoon and the currents were flowing to the south along the west coast of Sri Lanka at the location of the accident. On 22 May, tropical cyclone Yaas formed in the northern Bay of Bengal and propagated to the north and crossing the coast to the north of Visakhapatnam (India). Examination of the water level records along the east coast of India and at Trincomalee (Sri Lanka) indicated the formation of a continental shelf wave (Huyer, 1990) subsequent to the cyclone making landfall (Eliot and Pattiaratchi, 2010) and propagating southward and clockwise around Sri Lanka. As a consequence of the continental shelf wave, the sub-tidal water level at Trincomalee started to increase whilst the mean water level at Colombo was decreasing. This resulted in the currents reversing from that due to the monsoon and flowing east and north along the south and west coasts of Sri Lanka, respectively, after 30 May (Fig. 2). The winds decreased to <5 m s-1 after 2 June.

Nurdles are small plastic pellets used as raw materials for the manufacture of virtually anything that is plastic (e.g. from plastic bags and bottles to automobile parts) and are by definition classified as microplastics as their size is <5 mm. For comparison, the potential discharge of 78 t of nurdles released by the X-Press Pearl is the second-worst release of nurdles from shipping accidents. This includes the 150 t released into Hong Kong harbour in 2012 during a typhoon, and 49 t released into Durban harbour in South Africa in 2017 (Schumann et al., 2019). On 25 May, large quantities of nurdles were washed up on the beaches closest to where the X-Press Pearl was anchored (Fig. 8). These were transported onshore because of the strong winds and high waves present at that time. Subsequently nurdles were found along the whole of the west of Sri Lanka (https://storymaps.arcgis.com/stories/054f6746857242128298d53d220d76f0, last access: 28 December 2021).

We used the Regional Ocean Modelling System (ROMS) configured to the oceans around Sri Lanka (de Vos et al., 2014; Su et al., 2021) to provide surface currents together with the Lagrangian particle tracking model Ichthyop (Lett et al., 2008) to simulate the transport of buoyant nurdles along the coast of Sri Lanka (Fig. 9). The model results indicated movement of the nurdle plume southwards with the prevailing currents and by 27 May the plume had extended to the southwest corner of the Island (Fig. 9b). On 29–31 May, the nurdle plume had detached from the coast and was moving offshore to the southeast (Fig. 9c and d). The currents off the southeast of the island was getting stronger and were flowing to the south. By 2 June, this current has further strengthened and has curved around the island flowing to the east (Fig. 9e). By 4 June the currents along the west coast has reversed with the nurdle plume closest to the coast moving north (Fig. 9f). The nurdles that were detached from the coast on 29 May continued to be transported offshore and subsequently to the east. A budget of the simulated nurdles indicated that on 11 June 32 % of the nurdles released would have beached, 28 % would have been within the model domain (Fig. 9), and 40 % would have exited the model domain. A feature of the simulations is that the nurdles were contained at distinct boundaries representing the fronts separating cold and warm water as described in Sect. 4.2. Also the reversal of the surface currents on 2 June and transport northwards along the Sri Lanka coastline was associated with the continental shelf wave that was due to tropical cyclone Yaas (see Sect. 7.1). With time, the nurdles that leave the model domain will make landfall along the countries in the northern Indian Ocean (e.g. Indonesia, India, Maldives, Somalia) because of the reversing monsoon currents in the region (see Fig. 5) and will be a visible pollutant on the beaches for many decades to come.

The main gap in knowledge for the IO is the scarcity of data in both the surface and deeper ocean. Only a few data points are available for the whole IO including along coastlines and the deeper ocean (Fig. 2). This is a very large constraint, which needs addressing. As demonstrated in this paper, we can simulate the transport of plastics using numerical models, but confirmation of these results is of paramount importance. However, the simulations pinpoint regions that require attention in future investigations. For example, it is unclear which type of simulated garbage patches (leaky with low concentrations and on the western side of the IO basin or stable with high concentrations spanning the entire width of the IO basin) best represents the behaviour of the garbage patch in the subtropical southern IO. For the same reasons, it is also unknown how much plastic is contained in the subtropical IO garbage patch as well as how long it is likely to remain in this garbage patch (both as a result of escaping the IO into the South Atlantic Ocean and as a result of other factors such as sinking and ingestion of plastics).

This paper's main sources of plastics were derived from rivers mainly in the northern Indian Ocean. There is most likely transport of plastics from southeast Asia through the Indonesian Throughflow (Sect. 4.1). However, these sources are currently undocumented and need to be investigated. Individual events such as the X-Press Pearl ship incident off Sri Lanka are also responsible for large input of plastics.

Considering the importance of increasing plastic pollution, the dynamic pathways and their fate in the marine environment of the Indian Ocean needs further attention. Further, studies on the ingestion of plastic particles by marine biota and their residence time in seafood in the marine environment will also be useful for food quality and the ecosystem's overall health.

The understanding of the exact processes of biofouling, fragmentation, and sinking as well as the timescales on which these occur is limited. As far as we are aware, no studies have focussed on this in relation to the IO.