To assess the current state of knowledge on litter in submarine canyons, a literature review was carried out using Google Scholar and the ISI Web of Knowledge. The keywords “litter”, “plastic” and “microplastics” in combination with “canyons”, were used to generate a list of peer-reviewed papers from 1996 till 2022. All papers dealing with litter in surface waters and the water column were excluded. From the selected paper we extracted information about: the geographical area and specific canyon studied; the method used to assess the presence of litter or microplastics; the number of surveyed sites and the depth range explored; litter abundance within the canyons, its spatial distribution and the dominant litter type. In parallel, selected literature on submarine canyons and their transport processes were used to discuss and infer the role of sedimentary and hydrodynamic processes in litter transfer and distribution.

To identify the regions where submarine canyon heads are exposed to the highest input of river-derived litter, we combined a global model of river macroplastic input into the ocean (Meijer et al., 2021) with a global database of seafloor morphology (Harris et al., 2014; Bernhardt and Schwanghart, 2021a; Bernhardt and Schwanghart, 2021b). We calculated the distance from each submarine canyon head to the adjacent river outlets along the world’s coastlines. Then, we divided the modeled floating macroplastic emissions of the closest river outlet (in million metric tons per year (MT/yr)) of Meijer et al. (2021) by the distance from each canyon head to the closest river outlet.

This index is, of course, a simplified metric. It does not account for litter other than macroplastic (e.g., microplastic), the buoyancy of plastic litter and other pathways of litter introduction into the ocean system including coastal inputs, atmospheric deposition, and direct inputs from ships. It omits any oceanographic processes. Moreover, this approach refers to submarine canyons that were mapped on global low-resolution bathymetric data and hence only features large submarine canyons (see details in Harris et al., 2014). Additionally, we restricted this analysis to canyons between 50°N and 50°S and to the main continents including the major islands of the West-Pacific, where data coverage is best. However, we argue that this approach helps to depict highly vulnerable submarine canyons with potential high litter input, canyons with low potential input that may be declared as sanctuaries, and to design future targeted scientific surveys.

To estimate canyon exposure to fishing activity, we used the global compilation of commercial fishing activity (any type) of the NGO “Global Fishing Watch” (available here https://services7.arcgis.com/IyvyFk20mB7Wpc95/arcgis/rest/services/SDG_14_Global_Fishing_Activity_1/FeatureServer). We calculated the mean fishing activity per canyon per year for 4398 canyons of the global canyon database (Harris et al., 2014). All remaining canyons were not covered by fishing activity data. Commercial fishing activity is measured in hours per area (area = grid size = 0.1 x 0.1 degrees) and refers to the annual average of the pre-pandemic year 2019.

Differences in exposure to riverine plastic and to fishing activity between geographical regions have been tested by the non-parametric Kruskal-Wallis test and post-hoc pair-wise comparisons with the Wilcoxon method and Bonferroni correction, using the software R (R Core Team, 2016).

To compare observed litter abundances ( Supplementary Material S1 ) with canyon exposure to riverine plastic and to fishing activity, these indices have been estimated in higher detail for selected canyons where litter data were available. Canyon head locations and canyon areas have been remapped using Emodnet bathymetry for European seas (https://emodnet.ec.europa.eu/) and Gebco 2020 bathymetry (https://www.gebco.net) elsewhere, as these bathymetric data sets are of higher resolution (100 m resolution for Emodnet dataset, 15-arc resolution for Gebco 2020 bathymetry) than the SRTM30_PLUS 30-arc second database (Becker et al., 2009) used in the canyon mapping of Harris et al. (2014).

Canyons hosting a prevalence of plastic litter have been compared to the riverine-plastic input exposure, whereas canyon with dominant fishing-related debris have been compared to the exposure factor to fishing pressure.

There is a strong geographic bias to the studies that have been performed to date. Most of the research on seafloor litter and microplastics has focused on Mediterranean canyons (accounting for more than 50% of the canyons where anthropogenic debris has been reported), followed by the northeast and northwest Atlantic canyons ( Figure 2A ). In the Pacific Ocean, few canyons have been studied for litter assessment, mostly offshore California (Watters et al., 2010; Schlining et al., 2013) and in the South China Sea (Peng et al., 2019; Chen et al., 2020; Zhong and Peng, 2021). Canyons in the Indian Ocean are almost absent from published reports, except for a recent study of two canyon systems on the SW Australian margin (Taviani et al., 2023). The striking differences in coverage by studies are also evident when analyzing the sampling effort (i.e., the number of stations surveyed) for individual canyons ( Figure 2B ). On average, studies on litter distribution in Mediterranean canyons have analyzed a larger number of samples than canyons in other regions ( Figure 2B ), especially those in the NW Mediterranean. La Fonera, Cap de Creus and Blanes Canyons on the Catalan margin (Orejas et al., 2009; Ramirez-Llodra et al., 2013; Tubau et al., 2015; Mecho et al., 2018; Dominguez-Carrió et al., 2020) and some canyons of the Gulf of Lyons have been extensively studied (Galgani et al., 1996; Fabri et al., 2014; Fabri et al., 2019; Gerigny et al., 2019), with tens of stations explored per single canyon. Similarly, for a few canyons in the Atlantic Ocean, namely Dangeard, Explorer and Whittard Canyons on the Celtic margin (Pham et al., 2014), Nazaré Canyon off the west coast of Portugal (Mordecai et al., 2011), and Baltimore and Norfolk Canyons off the US coast (Jones et al., 2022) a large number of stations (>10) have been surveyed. However, the assessment of seafloor litter is based in most cases on exploration of only one or two sites of the entire canyon system. This is also true for microplastics assessment, which has been performed mostly based on a single sediment sample, except for a few individual canyons such as Norfolk (Jones et al., 2022) and Blanes Canyons (Sanchez-Vidal et al., 2018) where several samples have been collected. Papers on litter in the Monterey Canyon, one of the most studied canyon systems in the world (Matos et al., 2018), do not report the exact number of sampled sites, although the large number of surveys covering decadal time spans and the availability of a wide database (332 stations for shelf and canyons of Central California as reported in Watters et al. (2010) and 1149 videos in Monterey Bay analyzed by Schlining et al. (2013)) indicate that the sampling effort is likely much higher than all other canyons.

Bathymetric ranges of observations are also highly variable ( Figure 2C ; Supplementary Material S1 ). Overall, Mediterranean canyons have been mainly surveyed in their heads or along their upper reaches, and only a few canyons offshore the Spanish and French coasts have been surveyed in their middle or lower reaches at depths >1500 m (Galgani et al., 2000; Ramirez-Llodra et al., 2013; Tubau et al., 2015; Dominguez-Carrió et al., 2020; Angiolillo et al., 2021). Conversely, in the Atlantic Ocean, wider depth ranges have been covered. Canyons of the Bay of Biscay have been surveyed from 300 to 2300 m depth (Galgani et al., 2000; van den Beld et al., 2017), similar to the canyons of the continental margin off the Northeastern US, with observations from 300 to 2100 m depth (Quattrini et al., 2015; Jones et al., 2022). Explorations at greater depths have been carried out in the canyons offshore of Portugal, from 1600 down to 4500 m depth (Mordecai et al., 2011), as well as in the Whittard (Pham et al., 2014) and Mississippi Canyons (Wei et al., 2012), down to 2600 and 2700 m depth, respectively. In the Pacific Ocean, litter and microplastics have been reported from the tributary canyons of the Xisha Trough at around 2000 m depth, with deeper exploration in the trough down to 3300 m depth (Peng et al., 2019; Chen et al., 2020; Zhong and Peng, 2021). Canyons off SW Australia have been surveyed from 180 m down to 3300 m depth (Taviani et al., 2023) ( Figure 2C ).

Canyons which indent the shelf and are connected directly to river systems may be highly efficient at transferring sediment (and potentially litter) directly to the slope, bypassing the shelf. Dense sediment-laden water generated at the river mouth may plunge to form a hyperpycnal flow that can evolve into turbidity currents and supply sediment directly to the canyon (e.g., Var Canyon, NW Mediterranean; Mulder et al., 2003; Gaoping Canyon, Taiwan; Khripounoff et al., 2009; Carter et al., 2012; La Jolla Fan, Romans et al., 2016; Messina Canyons; Pierdomenico et al., 2019). Alternatively, high discharge from these river mouths may lead to rapid sediment accumulation near the river mouth or at the canyon head; these deposits may become flushed offshore as a buoyant river plume before sinking to the seabed (Tubau et al., 2015; van den Beld et al., 2017), or may collapse, to trigger turbidity currents (Carter et al., 2012; Pope et al., 2017; Hizzett et al., 2018; Hage et al., 2019; Talling et al., 2022).

Rivers, globally recognized as the main conveyers and transporting agents of land-based litter (Rech et al., 2014; Lebreton et al., 2017; Meijer et al., 2021), can be responsible for significant inputs of mismanaged anthropogenic waste to the sea, although many uncertainties still exist about the proportion of litter rapidly sinking close to the river mouth versus being transported offshore, and consequently about how riverine outflows may enter and interact with the sedimentary regime in canyons. While the largest discharging rivers have been traditionally considered to contribute almost entirely to the global river emissions (Lebreton et al., 2017; Schmidt et al., 2017), the important role of small (low discharge), heavily polluted urban rivers has been recently reconsidered (Meijer et al., 2021). Direct input from a small river affected by flash-floods and draining a heavily populated area explains the high concentration of litter in the Gioia-Petrace Canyon system, where up to 560 items per linear km of ROV track were observed (Pierdomenico et al., 2020). Even higher densities have been reported from the Messina Strait, ranging from 121,000 to up to 1.3 million items/km², making it the most litter-affected canyon and deep-marine environment recorded to date globally ( Figure 3 , Pierdomenico et al., 2019). These canyons are connected to subaerial drainage networks and are frequently affected by sedimentary gravity flows triggered by river flash-floods, which are able to funnel huge amounts of anthropogenic waste into adjacent canyons. In this area, the high coastal population densities coupled with poor disposal practices likely contribute to the extreme density values observed (Pierdomenico et al., 2019).

The presence of large rivers may influence litter abundance even in canyons separated from land by wide continental shelves, such as observed in the Bay of Biscay, where the higher litter abundances found in the Belle-île and Arcachon Canyons (with maximum densities up to 59,000 items/km², comparable to those observed in other Mediterranean coastal canyons), with respect to the other canyons of the margin, have been attributed to the influence of the Loire and Gironde rivers, respectively (van den Beld et al., 2017). In the Gulf of Lyon, despite low litter densities being observed off the mouth of Rhône River (Galgani et al., 1996) and in the upper reach of the Petit Rhône Canyon (Fabri et al., 2014; Gerigny et al., 2019), trawling on the deep-water Rhône fan at 2200 m depth, more than 150 km from the river mouth, revealed litter concentrations of 52,000 items/km² (Galgani et al., 2000), suggesting efficient transfer of river-derived waste.

Submarine canyons that indent the shelf, but that do not have a direct connection to a river mouth, are fed by along-shelf currents and wave and tide action, which redistributes and disperses sediments and litter (Mulder et al., 2012; Schlining et al., 2013; Eidam et al., 2019; Pierdomenico et al., 2020). Natural and anthropogenic debris may be transported along the shelf, until the load is diminished through wave and storm action, or until it meets an intersecting canyon head (e.g., La Jolla or Monterey Canyons, California; Xu et al., 2002; Covault et al., 2007). Particularly, high energy seasonal events that might enhance cross- and along-shelf transport such as storms, dense water shelf cascading, hurricanes or typhoons (e.g., Flemming, 1980; Canals et al., 2006; Harris and Heap, 2009), can be responsible for the transfer of heavy litter from the shelf to canyons (Wei et al., 2012; Tubau et al., 2015; Zhong and Peng, 2021).

Submarine canyons deeply incising the shelf offshore densely populated areas were observed to accumulate substantial amounts of litter, such as the Paillon Canyon offshore Nice (>80 items/km, Galgani et al., 1996), La Fonera and Cap de Creus Canyons (>25,000 items/km² Tubau et al., 2015; Dominguez-Carrió et al., 2020) and the Dohrn Canyon, offshore Naples (50 items/km, Taviani et al., 2019) ( Figure 3A ). Reduced distances between the canyon head and the coastline may also in part explain the higher abundance of anthropogenic waste observed in the Ligurian and Corsica canyons with respect to the offshore Gulf of Lyon canyons (Gerigny et al., 2019). The higher litter abundance reported from the Lisbon Canyon compared to other canyons of the Portuguese margin was also interpreted as due to the proximity to a large population center (Mordecai et al., 2011). On the other hand, the low litter density observed in the canyon systems off southwestern Australia (1 to 1.7 items/km), has been attributed to reduced coastal urbanization in this region compared to other regions worldwide, combined with the national status as marine parks which further limits human activities in these areas (Taviani et al., 2023). Exploration of north-western Atlantic canyons, which are separated from land by a wide continental shelf, revealed an overall low litter abundance, despite anthropogenic debris being observed in most dives (Quattrini et al., 2015; Jones et al., 2022) ( Figure 3 ). Surprisingly, the highest litter densities >10,000 items/km² were found in an un-named slope-confined canyon at 1000-1100 m depth (Quattrini et al., 2015).

While the proximity of a canyon’s head to land may facilitate the transfer of litter of terrestrial origin from the coast and shelf environments into the canyons, one of the most littered canyons observed to date is the SY82 Canyon, a tributary of the Xisha Trough in the southern China Sea (Peng et al., 2019). This canyon has its head at 350 m depth, being located ~150 km from the coast. Here, large litter accumulations reaching densities >50,000 items/km² were found in the middle course of the canyon at 1700-1800 m depth, indicating that the massive transfer of anthropogenic debris to the deep sea via canyons does not necessarily require close proximity of the canyon head to land. Even though it has been proposed that most of the debris came from fishery and navigation activities, the uneven and focused distribution of the litter accumulations indicates subsequent reworking and down-canyon transport of litter (Zhong and Peng, 2021), showing that even canyons indenting wide shelves may become hotspots of litter accumulation.

The global maps in Figure 4 and the boxplots in Figure 5A show a highly spatially-variable exposure of canyons to river-supplied plastic fluxes, as confirmed by Kruskal-Wallis test (p<0.001) and Wilcoxon pair-wise comparisons ( Supplementary Material S2 ). Overall, the Mediterranean Sea, the North Pacific Ocean and the South Atlantic Ocean show a significantly higher proportion of submarine canyons that are most exposed to plastic contamination when compared to other ocean basins ( Figure 5A ). Specifically, we infer high exposure of canyons to litter in areas where high macroplastic emission from rivers (Meijer et al., 2021) coincides with short distances between the river outlets and submarine canyon heads. For instance, in the Philippines and India, which are recognized as the largest contributing countries to global plastic emissions (Meijer et al., 2021), tectonics and high rainfall result in high sediment supply (and therefore litter export) to the oceans (Milliman and Meade, 1983). Narrow shelves and the higher percentage of shelf-indenting canyons compared to other margins (Harris and Whiteway, 2011) makes these areas more prone to transfer plastic from land to the deep sea via canyons. Similarly, canyons off West Africa, Central America and Brazil as well as those in the Mediterranean and Black seas are also susceptible to receiving large amounts of river-derived litter, mostly because of rivers draining highly populated coastal areas. They are therefore identified as hotspots for plastic emissions (Meijer et al., 2021).

Our global assessment highlights that most of the canyons highly vulnerable to plastic input are located in areas that have not been yet surveyed for litter (see Figure 1 for comparison). Hence, we are likely far from a comprehensive knowledge of the real magnitude of this worldwide impact. Moreover, it also highlights areas with canyons that are potentially less affected by riverine plastic emissions, such as those located in the North Atlantic Ocean, on the Australian coasts and in the NE Pacific ( Figure 4 ). These canyons, theoretically less impacted by plastic inputs along their course, may host well-preserved and relatively pristine ecosystems, and could be considered as priority areas to concentrate conservation efforts (Fernandez-Arcaya et al., 2017).

Notwithstanding the limitations of this approach (as explained in the methods section), the comparison of the estimated exposure values with litter densities observed in canyons shows an overall increase in litter abundance with increasing exposure values ( Figures 5B, C ). This supports the validity of the global assessment and stresses the strong role of rivers as main inputs of litter to the Global Ocean (Rech et al., 2014; Jambeck et al., 2015). The increase is more evident for Mediterranean canyons ( Figure 5B ), which have been more extensively surveyed, especially in their upper reaches, than other canyons worldwide. In canyon systems developed on the edge of wide continental shelves, as in the case of the North Atlantic passive margin, that show highly variable litter abundance for similar exposure values ( Figure 5C ), complex hydrographic regimes acting over large areas may concur to generate more complex dispersal patterns of litter. Departures from the overall trend could also be related to the fact that river plastic discharge from Meijer et al. (2021) may be not representative of local scenarios. The mismanagement of litter dumped in small torrential rivers may for instance explain the unpredicted higher abundances in the Messina Canyons with respect to other ones with comparable exposure (Pierdomenico et al., 2019).

Although overlooked by many global studies on plastic transport to the ocean, fishing activities can represent an important source of litter in canyons, which can be the dominant source of plastic pollution in some systems. Many canyons are the sites of intense commercial and artisanal fishery activities (due to the rich commercially-important fauna they host), including bottom trawling along the rims and long-line fisheries on the rocky substrates not accessible to trawling (Puig et al., 2012; Fernandez-Arcaya et al., 2017). Small-scale artisanal fishing activities may have a deep impact on littering in canyons, as indicated by the very high litter densities of 280,000 items/km² observed in the East Sardinian canyons (Cau et al., 2017), with abandoned or discarded fishing gears comprising 84 to 100% of litter, approx. three times higher than other geomorphological settings (Cau et al., 2017). Similarly, concentrations of benthic long-line fishing gears of up to 220,000 items/km² were observed on rocky outcrops along the flanks of Cap de Creus Canyon (Orejas et al., 2009). During submersible explorations off the Californian margin, hotspots of litter, with densities up to 380 items per linear km of ROV track, were found in Monterey Canyon and the south-western edge of Soquel Canyon at places that are traditionally fished (Watters et al., 2010).

Apart from fishing gear, other maritime activities are also attributed as the source for heavy items found in canyons such as glass, clinker or large metal objects, whose distribution has been tentatively linked to direct dumping from ships and therefore associated with marine traffic and shipping routes (e.g., Ramirez-Llodra et al., 2013; van den Beld et al., 2017). Other local factors may determine the accumulation of specific litter types, such as the presence of naval bases or military dumping sites, which are thought to be responsible for the large proportion of metal objects observed in some canyons of the Gulf of Lyon (Fabri et al., 2014; Gerigny et al., 2019) and the Gulf of Cadiz (Mecho et al., 2020).

The exposure of global canyons to commercial fishing activity was calculated for 4398 large submarine canyons of the 9477 canyons mapped by Harris et al. (2014) ( Figure 6 ). The remaining canyons do not overlap with fishing activity data as such data do not exist for large regions of the global ocean, such as the South and Southern Asia, the Central and South-Eastern America and the Southern Mediterranean Sea ( Supplementary Material S3 ). However, the available results show that the canyons of the Mediterranean Sea are more heavily affected by fishing pressure along their course than canyons in other regions in the world (p<0.001). No striking differences in canyon exposure to fishing pressure were observed across the main oceans, except for a slightly higher exposure of canyons in the northern hemisphere, which is significant only for the Mediterranean Sea and the North Atlantic Ocean ( Figure 7A ; Supplementary Material S2 ).

Comparison of the observed litter densities with mean commercial fishing activities at submarine canyons, carried out for canyons where the dominant litter type was fishing-related debris, does not show clear trends ( Figure 7B ). However, along with the data gaps, it has to be considered that the calculated mean commercial fishing activity may not recognize the artisanal fishery activities, which are regarded as an important source of litter in specific cases (Cau et al., 2017). This may explain, for instance, the high densities of fishing gears in Sardinian canyons (Mediterranean Sea) despite variable fishing activity ( Figure 7B ). In addition, as the distribution of fishing debris may be largely controlled by the substrate topography and the occurrence of rocky outcrops that favor entanglement, the estimation of fishing-debris concentration may be strongly influenced by the sampling location along the canyon course.

Plastic and other land-based litter that is funneled into canyon heads can be remobilized and redistributed downslope toward deep-sea areas by a variety of oceanographic and sedimentary processes, including enhanced bottom currents, dense water cascading and sediment gravity flows (Schlining et al., 2013; Tubau et al., 2015; Dominguez-Carrió et al., 2020; Pierdomenico et al., 2019; Pierdomenico et al., 2020; Zhong and Peng, 2021). These processes act at different temporal and spatial scales from the continental margin down to individual canyons and on to small-scale features along the canyon course and can be responsible for the uneven distribution of litter among and within individual canyons. Furthermore, various sediment-transport mechanisms (and by extension litter transport mechanisms) often coexist in a given submarine canyon and along-canyon transport is not a constant or unidirectional process (Puig et al., 2014; Amaro et al., 2016). This means that understanding the magnitude of transport activity and the role of canyons in funneling and accumulating litter in the deep sea is not a trivial task, as it is driven by a complex interplay of natural and anthropogenic factors linked to sources, transport, and depositional mechanisms (Ramirez-Llodra et al., 2013; Maier et al., 2019; Pearman et al., 2020; Pierdomenico et al., 2020).

Local topography and spatial variations in flow energy seem to play an important role in the deposition of litter in canyons. Litter has been observed frequently within the troughs of furrows, behind rock slabs, tree trunks or other obstacles ( Figures 8A, B ) (Galgani et al., 1996; Tubau et al., 2015; Pierdomenico et al., 2020). Specific types of litter, such as fishing gear, or cables and pipelines lying on the seafloor, can nucleate the formation of small litter accumulation points, where mostly plastic items get trapped ( Figure 8C ) (Mordecai et al., 2011; Tubau et al., 2015; Biede et al., 2022). On a larger scale, morphological features along the canyon axis such as scours, knickpoints, or changes in thalweg slope, which may interact with flows promoting deposition, can represent important accumulation areas for litter (Pierdomenico et al., 2019; Angiolillo et al., 2021; Zhong and Peng, 2021).

Sediment gravity flows, principally turbidity currents, are the main agents for sediment transport down canyons and consequently for litter transfer to the deep sea (Pierdomenico et al., 2019; Zhong and Peng, 2021). There are several processes able to trigger sediment gravity flows within canyons (Piper and Normark, 2009), including the evolution of mass-failures along canyon walls (Paull et al., 2010), advection of resuspended shelf sediments by oceanographic currents or storm waves (Xu et al., 2010; Martín et al., 2011) and hyperpycnal flows during pulses of river discharge or ice melting (Mulder et al., 2003; Khripounoff et al., 2012; Hizzett et al., 2018; Bailey et al., 2021; Talling et al., 2022). Sediment resuspension induced by bottom trawling can be an additional trigger for the development of turbidity currents along canyons (Puig et al., 2012). However, it is increasingly apparent that major external triggers are not always required for the inception of gravity flows; the key driver appears to be the availability and supply of sediment, which may be sustained or ephemeral in its delivery to canyons (Bailey et al., 2021). While it has been long assumed that during the modern high-stand of sea level, the level of activity of canyons whose heads lie at large distances from terrestrial sediment sources is significantly reduced, recent studies have shown that turbidity flows with high frequencies (i.e., sub-annual) and velocities comparable with highly active coastal canyons, may also affect such land-detached canyons (Heijnen et al., 2022a), with potential implication for the transfer of litter and microplastics in the deep-sea realm via submarine canyons.

Due to the high flow velocities, transport capacity and the potential to run out to significant distances, turbidity currents are able to carry and redistribute significant amount of litter at high depths. In Monterey Canyon, which is frequently affected by powerful turbidity currents (Paull et al., 2010), surveys for litter from 25 to 4000 m depth found the greatest abundance of plastic between 2000 and 4000 m depth (Schlining et al., 2013). The activity of turbidity flows can also lead to the deposition of large litter accumulation hotspots, such as those found in the upper reach of the Messina Strait Canyons (200-600 m depth, Figures 8E, F ) (Pierdomenico et al., 2019), in the middle reach of the SY82 Canyon at 1800-1900 m depth (Zhong and Peng, 2021) and at the base of the Monaco Canyon at 2100 m depth (Angiolillo et al., 2021). These accumulation hotspots are a few to tens of meters wide and formed by a mixture of items, whose chaotic arrangement evidences remobilization and emplacement by sediment transport processes (Pierdomenico et al., 2019; Zhong and Peng, 2021). Their composition, dominated by litter of urban origin, primarily plastic, the presence of terrestrial plant debris and the paucity of fishing-related debris further support their link with the action of turbidity flows carrying litter from shallower depths (Angiolillo et al., 2021). Furthermore, the occurrence of large rocky boulders and heavy litter items, such as pieces of furniture, metals sheets and cars, within litter piles testifies to the high transport competence of turbidity currents (Pierdomenico et al., 2019).

Even though it might be expected that turbidity flows (that can travel for hundreds of kilometers), should enable transport of anthropogenic debris to the lowest reaches of canyons, very little is known about potential accumulation at the base of canyons and on deep-sea fans, as these zones have not been extensively surveyed for litter. However, the high concentration of litter found in the Rhône deep-sea fan (Galgani et al., 2000) suggests that litter can be transferred throughout the entire canyon course in systems actively carrying sediment toward deep-sea fans, where it can eventually be prone to entrainment by bottom currents.

Sediment-gravity flows may also bury litter within the sediment, as indicated by reports of partially buried items in canyon sediments ( Figure 8D ) (Pierdomenico et al., 2019; Angiolillo et al., 2021). The ability of gravity flows to transport and bury litter is largely unstudied, although flume experiments evidenced that microplastic can be transported and buried by turbidity currents (Pohl et al., 2020). Furthermore, macroplastics enclosed within coarse grained turbidite deposits buried 2.5 m below the seafloor have been recovered from a sediment core in a prodelta channel, further demonstrating the potential of gravity flows in burying litter deeply beneath the sea floor (Pierdomenico et al., 2022).

Oceanographic processes can also be involved in the transport of sediment and litter within canyons. The accumulation of plastic below 1000 m depth in the Cap de Creus Canyon has been linked to the action of dense shelf water cascading that generates strong bottom flows able to erode and entrain seafloor sediment in the upper reach of the canyon (Canals et al., 2006). These flows transport litter from the continental shelf and upper canyon, favoring its settling once the current speed has slowed down (Tubau et al., 2015; Dominguez-Carrió et al., 2020).

Bottom currents are recognized as important agents for the dispersal of microplastics in the deep sea (Kane et al., 2020) and can be steered down- and up-slope through canyons, where they can also transport larger macrodebris. In the Gioia Canyon, almost 40% of total macroplastic found between 50 and 490 m depth was observed drifting above the seafloor, demonstrating the important role of bottom currents in the downward flux of plastic (Pierdomenico et al., 2020). The decrease in the amount of drifting litter downslope suggests trapping of debris in the upper parts of canyons for some period of time.

Internal waves and tidal currents, which are usually amplified by the topographic constraint of submarine canyons, can contribute to sediment resuspension and advection along the canyon axis, sometimes favoring sediment and litter accumulation in specific regions (Hotchkiss and Wunsch, 1982; de Stigter et al., 2011; Pearman et al., 2020). Changes in strength and direction of currents driven by internal tides (Mulder et al., 2012), combined with the occurrence of biologically and geologically complex habitats promoting trapping of litter, are thought to be responsible for the higher abundances observed in specific mid-depth ranges (800-1100 m) in the Bay of Biscay canyons (van den Beld et al., 2017).

Because of the different entry points and transport mechanisms, litter of terrestrial and maritime origins often shows a different distribution in the different morphological zones of canyons, as observed in the Cap de Creus Canyon (Dominguez-Carrió et al., 2020), the Dohrn Canyon (Taviani et al., 2019; Angiolillo et al., 2023) and the Gioia Canyon (Pierdomenico et al., 2020). In these canyons, fishing-related debris was observed mainly on rocky substrates at the canyon head or along the walls as opposed to general plastic items, which were concentrated within the thalweg. This may be related to the fact that accumulation of fishing debris is linked to the spatial distribution of fishing activities, which are often focused in specific sectors of canyons such as the canyon head and rims (Puig et al., 2012; Tubau et al., 2015). In addition, fishing gear may be too large or heavy to be easily displaced by bottom currents, compared to plastics (van den Beld et al., 2017), and due to their shape can be easily entangled in complex terrains such as rocky outcrops, coral frameworks or other biogenic structures that are common in canyons (Angiolillo et al., 2015). A dominance of fishing gear, especially lines and trammel nets, has been reported in several studies that surveyed the rocky habitats at the canyon head or flanks in the upper canyon reach (e.g., Watters et al., 2010; Oliveira et al., 2015; Cau et al., 2017; Enrichetti et al., 2020), often as additional observations within the assessment of coral communities that thrive in these habitats (Orejas et al., 2009; Cau et al., 2017, Giusti et al., 2019; Fabri et al., 2019; Angiolillo et al., 2023). The presence of fishing gear within the thalweg in the Nazaré Canyon down to 4300 m depth (Mordecai et al., 2011) suggests that fishing-related debris can be also funneled to the canyon thalweg and transported to great depths, although direct dumping by fishermen cannot be excluded. However, the greater abundance of fishing debris at shallower depths, compared to general waste (e.g., Schlining et al., 2013; Tubau et al., 2015; Angiolillo et al., 2021; Taviani et al., 2017; Dominguez-Carrió et al., 2020) and the paucity of fishing gear within the large litter piles found in the middle and lower reach of canyons (Angiolillo et al., 2021; Zhong and Peng, 2021), suggest a relatively lower mobility for this type of litter compared to general plastic items.

Recent literature addresses transport and burial of microplastics by sediment gravity flows (Ballent et al., 2013; Pohl et al., 2020; Bell et al., 2021), with some attention being paid to interaction with bottom current systems (Kane et al., 2020). The resemblance of microplastics size to the natural sediment present in gravity-driven flows, has motivated application of existing sediment transport hydrodynamics of sediment gravity flows to microplastics transport, specifically the Rouse perspective (Kane and Clare, 2019; Waldschläger et al., 2022). Suspension of dense particles in turbulent environmental fluid flow is understood to result from a balance between downward settling due to gravity and upward mixing due to turbulence. This process was resolved by Rouse nearly a century ago for rivers (Rouse, 1937). In rivers, turbulence is exclusively generated through friction with the floor. Mixing of air into the top of rivers can be neglected. The non-dimensional Rouse number characterizes the ability of river-like flows to suspend particles of different size and density. It can be used to calculate vertical distributions of each particle type suspended in a flow (the Rouse equations). There are fundamental differences between rivers and turbidity currents due to the presence of ambient sea water above turbidity currents rather than air above rivers: a) the density difference between the flow and its surroundings is 1 to 2 orders of magnitude smaller; b) there is no viscosity discontinuity at the top of turbidity currents, and this generates friction and turbulence at the top of the flow as well as at the bottom; c) as a result of a) and b) ambient water can be mixed into the top of turbidity currents, which dilutes the particle concentration in the top of the flow. These fundamental differences from rivers are widely acknowledged to be problematic for the application of the Rouse equations to turbidity currents, yet Rouse still dominates quantification of particle suspension in turbidity currents (Hiscott, 1994; Hiscott et al., 1997; Straub and Mohrig, 2008; Bolla Pittaluga and Imran, 2014; Jobe et al., 2017; Bolla Pittaluga et al., 2018; Eggenhuisen et al., 2020). Indeed, Rouse equations satisfy observations for dense mineral sand with a Rouse number as low as ≈1/2 in turbidity currents (Eggenhuisen et al., 2020). However, here we test the Rouse equations with the experiments with plastic microfibers of Pohl et al. (2020) to reveal a critical shortcoming when applied to sediment gravity flows: the Rouse number of the microfibers is ≈1/50; yet the observed vertical distribution in the turbidity currents is identical to that of dense mineral sand with a Rouse number of ≈1/2 ( Figure 9 ). This violation of the Rouse equations is a critical barrier to using the Rousean perspective to quantify fluxes of low-density particles such as plastics in sediment gravity flows. Plastic particles that are virtually neutrally buoyant should hardly settle downwards due to gravity, should be easily mixed homogenously throughout the water column, and are therefore unlikely to ever be deposited with dense mineral sand at the base of the flow. Observations of plastic floating in the water column above the seabed (Pierdomenico et al., 2020) are indeed intuitively explained by their approximately neutral buoyancy in sea water. However, the burial of litter beneath the seafloor together with dense mineral sand in both observations (Pierdomenico et al., 2022) and experiments (Pohl et al., 2020; Bell et al., 2021) is not. It appears that a conventional buoyancy and turbulent diffusion treatment does not suffice to treat transport and burial of plastic litter in sediment gravity flows in submarine canyons. Pohl et al. (2020) explain the preferential burial of microplastic fibres relative to microplastic fragments with shape effects. Burial of the fibres is promoted by their larger surface area and length. As one section of the fibre is pushed downwards by settling sediment particles it gets trapped in the seabed, and the rest of the fibre is subsequently buried, despite it still sticking upwards partly above the bed into the ongoing flow. This shape effect does explain predominance of burial of plastic fibres over plastic fragments, and this could extend to explain preferential burial of macrolitter films, sheets, nets, and wires. However, it cannot explain the similarity in transport to natural dense sediment with Rouse numbers of ≈1/2. Furthermore, biological processes such as fouling, consumption and incorporation in fecal material (Choy et al., 2019) could increase density and settling velocities of microparticles. Though further research is needed, provisional treatment of plastic litter in similarity to natural sediment of Ro ≈ ½ is advocated here based on the empiric evidence of our analyses ( Figure 9 ) of the experiments by Pohl et al. (2020).

Modeling and quantification of physical transport processes of litter in submarine canyons can substantiate extrapolations of spatial occurrence and abundance of litter from sparse observations, and even provide justifications for predictions in systems that are designated as highly vulnerable to litter input, but that have not been surveyed yet. As an example of the potential of integrated studies in inferring litter distribution in canyons we apply the Sediment Budget Estimator (SBE) process model (Eggenhuisen et al., 2022) to the Congo Canyon, which we identified as a potential hotspot for plastic litter based on the canyon head exposure parameter to land-derived plastic litter ( Figure 4 ). We assume that plastic is transported in association with natural sediment ( Figure 9 ) in sediment gravity flows. Unfortunately, no direct survey information is available to quantitatively inform this association yet. However, such quantifications are available for the association between natural sediment and Particulate Organic Carbon (POC; Azpiroz-Zabala et al., 2017; Simmons et al., 2020). POC overlaps in density, sizes, and shapes with at least some plastic litter. Furthermore, we suggest that there is a physical similarity between transport of this litter and POC that can be leveraged. The potential for successful process modeling of litter fluxes is illustrated by populating the process model with input conditions ( Supplementary Material S4 ) based on published monitoring studies at 2000 m water depth in the Congo Canyon (Cooper et al., 2013; Azpiroz-Zabala et al., 2017; Simmons et al., 2020), and comparing the simulated POC budgets to the estimates of those studies ( Figure 10 ). The flow-velocity and sediment-concentration profiles of the sediment gravity flows compare well to the ADCP results. More importantly, resulting sediment fluxes closely match the variability observed in the 4 months monitoring deployment ( Figure 10C ). The simulated estimates for sediment-associated POC fluxes cover the range estimated from ADCP monitoring, though the histogram reveals an overestimation of the top end of the range reported by Azpiroz-Zabala et al. (2017). This success highlights that, when the association between litter and natural sediment is clarified, existing knowledge of sediment gravity flows suffices to quantitatively simulate fluxes and budgets of litter transport and burial down submarine canyons. Various approaches exist in process modeling of sediment gravity flows. The SBE demonstrated here uses a simplified stochastic approach. The benefits of this approach make it uniquely suited to integration with global estimation of canyon-head exposure to river-derived litter ( Figure 4 ) and sparse modeling: input conditions can be based on robust estimates from bathymetric surveys and sparse monitoring results; and ensemble simulations can rapidly be performed for tens of thousands of events in one or more canyons. The modeling results are still critically dependent on direct flow monitoring results, specifically sediment gravity flow thickness and natural sediment concentration. This means that at present, process modeling is only feasible in parallel to continued efforts in sea going research. Specifically, there is an urgent need to establish the association of litter and natural sediment in sediment gravity flows and buried deposits.