Physical transport properties of marine microplastic pollution

Biogeosciences Discussions This discussion paper is/has been under review for the journal Biogeosciences (BG). Please refer to the corresponding final paper in BG if available. Abstract Given the complexity of quantitative collection, knowledge of the distribution of mi-croplastic pollution in many regions of the world ocean is patchy, both spatially and temporally, especially for the subsurface environment. However, with knowledge of typical hydrodynamic behavior of waste plastic material, models predicting the dispersal 5 of pelagic and benthic plastics from land sources into the ocean are possible. Here we investigate three aspects of plastic distribution and transport in European waters. Firstly, we assess patterns in the distribution of plastics found in fluvial strandlines of the North Sea and how distribution may be related to flow velocities and distance from source. Second, we model transport of non-buoyant preproduction pellets in the 10 Nazaré Canyon of Portugal using the MOHID system after assessing the density, settling velocity, critical and depositional shear stress characteristics of such waste plastics. Thirdly, we investigate the effect of surface turbulences and high pressures on a range of marine plastic debris categories (various densities, degradation states and shapes tested) in an experimental water column simulator tank and pressure labora-15 tory. Plastics deposited on North Sea strandlines varied greatly spatially, as a function of material composition and distance from source. Model outputs indicated that such dense production pellets are likely transported up and down canyon as a function of tidal forces, with only very minor net down canyon movement. Behaviour of plastic fragments under turbulence varied greatly, with the dimensions of the material, as well as 20 density, playing major determining roles. Pressure was shown to affect hydrodynamic behaviours of only low density foam plastics at pressures ≥ 60 bar.

of pelagic and benthic plastics from land sources into the ocean are possible. Here we investigate three aspects of plastic distribution and transport in European waters. Firstly, we assess patterns in the distribution of plastics found in fluvial strandlines of the North Sea and how distribution may be related to flow velocities and distance from source. Second, we model transport of non-buoyant preproduction pellets in the Nazaré Canyon of Portugal using the MOHID system after assessing the density, settling velocity, critical and depositional shear stress characteristics of such waste plastics. Thirdly, we investigate the effect of surface turbulences and high pressures on a range of marine plastic debris categories (various densities, degradation states and shapes tested) in an experimental water column simulator tank and pressure labora- 15 tory. Plastics deposited on North Sea strandlines varied greatly spatially, as a function of material composition and distance from source. Model outputs indicated that such dense production pellets are likely transported up and down canyon as a function of tidal forces, with only very minor net down canyon movement. Behaviour of plastic fragments under turbulence varied greatly, with the dimensions of the material, as well as High density (HD) microplastics are commonly found on beaches, in river sediments, on continental shelf slopes and in deep sea benthic environments (Cole et al., 2011). They compose approximately half of all manufactured plastics (USEPA, 1992;Morét-Ferguson et al., 2010). Although growing interest in the situation has driven a number of recent studies, data of plastic pollution in the deep-sea is scarce, mainly due to 15 the difficulties of deep-sea sampling (Claessens et al., 2011). One extensive study covering European shelf areas reported spatial densities of 0.064-2.63 plastic pieces (≥ 2 cm diameter) per hectare (Galgani et al., 2000). On the California continental shelf, benthic trawls (net mesh size 333 microns) in the 20 cm above the seafloor at 30 m depth collected microplastics in spatial densities of 6.5 and 1.5 pieces m −3 before and 20 after a storm, respectively (Lattin et al., 2004). Before the storm, plastic density at the seafloor was roughly 60 times the plastic density at the ocean surface (< 1 piece m −3 ). Recently, in a study supported by the HERMIONE programme, Remotely Operated Vehicle (ROV) video surveys of benthic marine litter in the submarine canyons off the coast of Portugal reported highest abundances in canyon heads located off the coast 25 of populated cities (Mordecai et al., 2011). In this study, rather than rely on field observations to determine debris abundance and distribution, we attempt to predict how high density plastics may be transported throughout a submarine canyon ecosystem, by determining the physical transport properties of collected preproduction pellets in the laboratory and applying these results in a hydrodynamic model. 5

Factors influencing vertical transport of neustonic microplastics
Low density (LD) plastic accumulates in the neustonic zones of the worlds' sub-tropical gyres at spatial densities up to hundreds of thousands of pieces per square kilometer (Moore et al., 2001;Martinez et al., 2009;Andrady, 2011;Cole et al., 2011). The concentrations of plastic found in the underlying mixed layer, and how 10 meteorological and oceanographic conditions may influence these concentrations, are not well known (Doyle et al., 2011). Field studies indicate a general trend between higher wind speed and lower surface plastic counts, suggesting that surface plastics can be drawn down vertically into the water column (Lattin et al., 2004;Thompson et al., 2004;Ryan et al., 2009;Doyle et al., 2011;Kukulka 15 et al., 2012); however, the temporal and spatial extent over which this occurs is not well understood. In this study, two qualitative laboratory experiments were conducted to determine the effect of turbulence on the vertical distribution of various types of LD plastics (fragments, foams, filaments, films and pellets) from the samples collected at the Weser and Elbe Rivers (see Sect. 2.1) and how increasing pressure (depth) may 20 impact on the buoyancy of LD and HD plastics.

Strandline plastics of Weser and Elbe Rivers
Plastic fragments were collected from two strandline sample stations on both the Elbe River (Hamburg) and Weser River ( anthropogenic debris is stranded after a high tide. The top 10-15 cm of flotsam was collected from a 1 m 2 area at each of the four sites, after which plastics were manually sorted from the flotsam using tweezers. Plastic debris from each of these sources was sorted into five categories (film, fragment, mono-filament, preproduction pellet (primary microplastic granule) and foam) and stored in petri dishes. The particles' area,

5
Feret's diameter (or filament length) and minimum Feret's diameter were quantified and measured using the ImageJ (v. 1.45 s) (Rasband, 1997(Rasband, -2012 software application "analyze particles" tool on photographs with color threshold applied (Fig. 1). In ImageJ the Feret's diameter is defined as the "maximum distance between two points on the selection boundary" (Ferreira and Rasband, 2011, p. 123

Settling rates
Settling velocities of the three pellet categories were determined by filming particles sinking through a 1 m still saltwater column (salinity 36 psu). We subtracted JPG images 1 s apart from the video stream, and used the ImageJ software (Rasband, 1997(Rasband, -2012 to determine settling speed after the method described in Pabortsava et 5 al. (2011). The experimental run included ∼ 50 HD black pellets, ∼ 80 HD opaque pellets and ∼ 50 HD transparent pellets.

Deposition and resuspension velocity determination
A 20 cm erosion microcosm similar to that described by Tolhurst et al. (2000) simulates benthic shear environments, and was used to determine the flow velocities at 10 which bedload, resuspension, and deposition of the three categories of pellets occur. Runs were conducted with two groups of pellets: ∼ 300 HD black pellets, ∼ 200 HD opaque/transparent pellets (4 repetitions each) according to a predefined calibration table (Gust, unpublished data) relating rotor angular speed, pump flow and the resultant flow velocity (U * ). Using the water density, the shear velocity values were converted to 15 shear stress values, [N m −2 ], giving τ b , τ cr and τ d (See Appendix A). The experiments were run in a stepwise manner, in which the bottom shear was manually increased over seven, 2-min long steps (Table 1) using a controlling unit to adjust the rotational speed of a plastic disk inside the chamber and an adjustable flow meter attached to the pump discharge tubing (Tolhurst et al., 2000). Experiments were filmed to allow for 20 better analysis of particle behavior in laminar flows. The bedload shear velocity U * b was defined as the shear velocity at which 50 % of the particles rolled, slid or saltated on the chamber floor (percentages determined by direct observation and video analysis). The critical erosion velocity, U * cr , was defined as the shear velocity when 75 % of the particles were suspended in the water column. U Introduction

Effect of turbulence
Under standard conditions, turbulence dissipation rates (ε) at the surface boundary 15 layer and thermocline range between 10 −3 -10 −1 W m −3 (Sanford, 1997;Petersen et al., 2009). In the stratified interior of the open ocean ε is commonly ∼ 10 −7 W m −3 (Petersen et al., 2009). To reproduce these turbulence intensities in a controlled environment, a Multiscale Experimental Ecosystem Research Center pelagic/benthic (MEERC P/B) type C tank as developed by Crawford and Sanford (2001) was used. The cylin-20 drical 1 m 3 tank is fitted with horizontally rotating paddles ( Fig. 3a) capable of replicating turbulence intensities of the dimension and structure shown in Fig. 3b (Sanford, 1997;Crawford and Sanford, 2001;Stiansen and Sundby, 2001;Petersen et al., 2009;Porter et al., 2010). The average turbulence intensity within the tank is directly related to the rotation speed of the paddles ( Fig. 4 and mono-filament, and fragments of three size classes (≥ 10 mm, 5-10 mm and ≤ 5 mm). The largest fragment category (LD fragment, LA beach, ρ ∼ 0.698 g cm −3 ) consisted of irregular but ∼ 3 mm thick flattened pieces all of the same plastic composition. The medium and small fragments and remaining subsamples were collected from the Elbe and Weser sites (see Sect. 2.1) and varied in shape, size and type (Table 3).

5
To study how the buoyancy force of the plastics and the turbulent forces of the water interacted, the turbulence intensity within the water column was increased step-wise, with each step consisting of a minimum of 4 right-left cycles (5 forward rotations and 5 backward rotations), starting at 3 RPM and increasing to 27 RPM in increments of 3 RPM, turbulence dissipation rates ε ranging from 1.58 × 10 −4 to 2.21 W m −3 .

Effect of hydrostatic pressure
A pressure laboratory was used to observe the effect of increasing pressure on the buoyancy of microplastic debris. Direct visual observation was achieved using a digital camera in a 200 bar-proof Plexiglas pressure housing (Meerestechnik Bremen GmbH) inside a pressure laboratory. The plastics were kept in a saltwater (36 psu) environment 15 using a 10 cm diameter Plexiglas cylindrical container with a rubber sealed lid and rubber bladder suspended within the pressure chamber. A manual hydraulic pump was used to slowly increase the pressure from 1 to 200 bar over 20 min and a valve released the pressure again. Sub-samples of HD opaque pellets, HD transparent pellets, HD fragments, LD fragments, LD pellets, and foams (both clean and biofilmed from Elbe 20 Site 2) were tested.

Strandline plastics of Weser and Elbe Rivers
The Elbe strandline contained the largest number of plastic pieces, mostly fragments and foams. Histograms of the samples depict the size and type distributions (Fig. 5). 9,2012 Physical transport properties of marine microplastic pollution When comparing the urban Elbe sites to the rural Weser sites consistently higher counts are found in urban areas; ∼ 5000-7000 pieces m −3 and ∼ 150-700 pieces m −3 , respectively as extrapolated from sample counts. Pellets were most consistently abundant across sites, whereas films were almost exclusively found at Elbe Site 1. Fragments, pellets and foams were the three types of plastic found at each site. As de-5 termined from the measured Feret's diameter, pieces smaller than 5 mm were most abundant, followed by pieces between 5 and 10 mm and then pieces 10-15 mm. Table 3 lists size and count data for all samples according to piece type. The area and minimum Feret's diameter of filaments were not measured.

Laboratory experimentation
The LA beach sample consisted of preproduction pellets and fragments of both high and low density which could be divided into six groups: HD black pellets with homogenous appearance, HD opaque pellets, HD transparent pellets, LD translucent pellets containing small entrapped bubbles, HD fragments and LD fragments. The average 15 Feret's diameter of the pellets was 4.8 mm, while the flattened fragments of the LA beach sample measured ∼ 12.7 mm. The density of the pellets ranged across values above and below the density of sea water (1.03 g cm −3 ); black, opaque and transparent pellets were roughly 10 % more dense than sea water whereas the buoyant translucent pellets were about 20 % less 20 dense (Table 4). The velocity at which the HD pellets from the LA Beach site settled varied from 20-60 mm s −1 ( The two groups of pellets used in the resuspension chamber investigations behaved as expected given their densities and settling velocities. Black pellets began bedload transport in the first time-step (1.4 × 10 −2 N m −2 ) and almost all pieces were in rolling or saltating motion before any pellets were suspended. Critical erosional shear stress was determined to be ∼ 0.14 N m −2 . At the highest possible shear stress achieved in 5 the chamber (∼ 0.2 N m −2 ) almost all black pellets were in suspension (Fig. 6a). In contrast, the majority of the opaque/transparent group was not significantly suspended at the highest shear stress. In general, this group showed more resistance to erosion and less uniform behavior; opaque pellets eroded most easily and transparent pellets least. Approximately 85 % of all pellets were transported by saltation (nearing critical 10 erosion threshold) at a shear stress of 0.2 N m −2 (Fig. 6b) and the last pieces settled at ∼ 4.6 × 10 −2 N m −2 . All shear stress values in Table 6 are approximated from direct observation and video analysis and are averaged across replicates. Accuracy of the erosion experiment is low due to reliance on observation to determine the exact stage during the time-step experiment when the pellets alter their 15 behavior. Pellets had a slow response time to changes in flow velocity and did not behave uniformly within or between replicates, possibly due to slight differences in particle properties (i.e. shape, size, density, degree of bio-fouling). It was also often difficult to determine whether a pellet was in suspension or bedload transport, and therefore define the shear stress threshold separating the two behaviors. Additionally, the pump 20 had a large influence on the instantaneous shear within the chamber and the resulting transport behavior of the pellets. The pump flow fluctuated slightly on high frequencies and between replicate runs, sometimes causing suspended pellets to fall back to bedload transport due to an abrupt loss in pumping power. The restricted power of the pump also limited the maximum shear stress generated in the chamber. Despite these BGD 9, 2012 Physical transport properties of marine microplastic pollution

Transport modelling
In the Nazaré Canyon simulation model, HD black pellets showed little displacement from the monitor boxes. Three output parameters were used to characterize the pellet transport behavior: distance (the total distance a pellet travelled [km]), displacement (the net distance a pellet was transported [km]), and velocity [km yr −1 ]. As depicted in 5 Fig. 7, averages for each parameter were calculated twice for each box; once for (a) all pellets (n = 2004) and once for (b) those pellets which were transported out of the monitor box during the model simulation period (1 pellet from Box 1, 13 pellets from Box 2, 53 pellets from Box 3 and 39 pellets from Box 4). In general, pellets traveled greater distances than they were displaced, indicating the pellets were transported in an os-10 cillating manner, up and down canyon repeatedly. The average transport distances for all pellets was around 0.1 km, with an average displacement of 0.04 km. Average velocities ranged from 0.1 to 0.9 km yr −1 ; however, the maximum velocity of an escaped pellet was 7.03 km yr −1 . Pellets in Monitor box 1 were transported the least distance while pellets from Monitor box 2 were, on average, displaced the furthest. Pellets from 15 Monitor box 4 traveled on average the largest distances (in oscillatory up and down canyon motion). The residence time of the HD black pellets in each monitor box is depicted in Fig. 8 as the fraction of pellets over time. The number of pellets in Box 1 fluctuates slightly in a sinusoidal manner around 100 % of initial number of pellets for the entirety of the 20 model simulation period in a manner suggestive of a tidal rhythm. In Boxes 2, 3 and 4 tracer fractions change in a similar pattern, although to a lesser degree, but also change abruptly at certain points of time, indicating forces additional to tides act upon the pellets' transport at greater depths. These events occur simultaneously in each of the three deepest boxes signifying that the pellets' movements are forced by a large which gives 315, 24, 6 and 8 yr for pellets in Monitor boxes 1-4, respectively.

Turbulence assay 5
Studying the random nature of turbulent flows required a qualitative approach and heavy reliance on observation. Figure 9 shows the percentage of pieces, as determined by observation, of each plastic type submerged at each step of increasing turbulence intensity. Turbulence dissipation rate, ε, is plotted to show increasing turbulence levels associated with increasing wind shear. Overall, plastic composition (density) appeared 10 to have the largest influence on vertical displacement, followed by size and flatness. Large, irregularly shaped pieces (e.g. films and filaments) were most susceptible to turbulence and were drawn below the surface at the lowest turbulence intensities of 2.5 cm s −1 . The plastics most resistant to surface turbulence were the round LD pellets. The paddle rotation often generated visible vortices extending from the surface 15 and dissipating within 1-5 s of formation. Some of these were forceful enough to pull large, buoyant pieces below the surface, where they would disperse within the water column before returning to the surface. Pieces with irregular shape or densities close to that of the seawater were often submerged by small scale turbulence invisible to the observer.

20
LD pellets submerged to maximum 20 cm depth during the formation and dissipation of vortices at the fastest rotation speed (27 RPM). At lower turbulence intensities, pellets would occasionally submerge 1-2 cm before popping back to the surface. LD fragments 10-15 mm in size were observed separately from those < 10 mm. Similar to the case of LD pellets, only ∼ 10 % of large fragments were submerged at the highest turbulence dissipation rates (ε = 2.2 W m −3 ). The smallest fragments were affected at much lower turbulence intensities; approximately 70-80 % remained submerged for longer than 1 min at turbulence intensities (U RMS ) of 7.0 cm s −1 . Foams behaved similarly to LD pellets; both plastic types would occasionally become slightly submerged 5 with strong vortex events at turbulence intensities of 5.5 cm s −1 or greater. Films and filaments were most susceptible to turbulence, several pieces submerging during the third rotation step (9 RPM) where turbulence intensities within the chamber averaged at ∼ 2.5 cm s −1 , and almost all pieces submerged before the final rotation step (27 RPM; U RMS ∼ 7.0 cm s −1 ). The film and filament pieces with the most irregular shape were 10 more readily submerged than thick, rigid, or flat pieces, though interestingly, flatness had a two-sided effect. Flat pieces of plastic had higher resistance to submerging (likely due to larger area affected by surface tension) but once within the water column they remained below the surface longer due to a slower rising speed and susceptibility to turbulences. In general, microplastics which were submerged at lower turbulence intensi- 15 ties also remained within the water column for longer time periods than those less easily submerged. Some fragments, films and filaments remained submerged for the duration of the experiment only returning slowly to the surface (at rates ∼ 0.1-0.003 m s −1 ) after paddle rotation was stopped and flow velocities significantly reduced.
When the turbulence dissipation rates produced in the MEERC p/b water column 20 simulation tank are correlated to wind speed, the effect of increased wind shear on surface counts of various debris types can be visualized. Equation (2)

Pressure assay
The only plastics affected by pressures ≤ 200 bar were some partially degraded/biofilm covered Styrofoam pieces (Fig. 10a-d). The most biofilm rich pieces began to sink at 60 bar corresponding to 600 m depth, and four relatively large (> 10 mm), less degraded pieces lost buoyancy at pressures of 140, 150, 170 and 180 bar, sinking slowly 5 to the bottom of the chamber. When pressure was slowly released pieces regained buoyancy; however, when removed from the chamber, most pieces remained partially compressed. The least degraded/biofilm covered pieces were the least compressed.
In a second and third run, subsamples of LD fragments and HD opaque/transparent pellets (Fig. 10e, f) and LD pellets (Fig. 10g, h) were tested, with no visible pressure ef-10 fects apparent. The buoyancies of these microplastics were not affected by an increase in pressure up to 200 bar.

Strandline plastics
Plastic quantities and types found in near-shore marine environments depend on the 15 size and distance of plastic sources. The Elbe sites, located near the city center of Hamburg, contained significantly higher concentrations of debris than the Weser sites located on the outskirts of Bremen. This can be explained by assuming that more debris was released into the environment from the more densely populated city of Hamburg than from the smaller city of Bremen. Similar findings have been reported by Galgani et The type of sediment and flotsam on the bank may affect the degree to which debris is "filtered" from the water. For example, rocky shores should have a lower capacity to retard microplastic transport but may more effectively remove larger debris from the 10 water. A general trend between the size of organic material and the size of microplastics was observed (data not shown). The Elbe strandline contained much smaller sized debris than the Weser strandline. Wood chips, degrading leaf pieces, small twigs and seeds made up the majority of the Elbe material, whereas that found at the Weser strandline comprised primarily of large reeds and drift wood. This compositional differ-15 ence is another possible explanation of why larger counts of microplastics were found at the Elbe than Weser sites.
Current and tidal flow velocities may also affect the removal rates of plastics from river and coastal waters. Strong currents would have a higher eroding capacity for high density debris, transporting such material down-current or to areas of slower flow ve-20 locity. Flow velocities and sediment loads correlate well with debris transport capacities of rivers (Galgani et al., 2000). Moreover, rivers and estuaries in regions of strong tidal variation or large water level fluctuations may deposit greater amounts of microplastics due to the frequent occurrence of low flow velocities and the regular occurrence of high tides. Similarly, areas with greatly reduced flow velocity, estuaries, deltas, bays 25 and harbors accumulate more debris as compared to high discharge river plumes where debris is more easily dispersed (Galgani et al., 1996(Galgani et al., , 2000Browne et al., 2010;Claessens et al., 2011). Benthic and coastal topography (e.g. ripples or indentations in the sediment, canyons, slopes, bioturbation mounds, rocks etc.) which affect local BGD 9, 2012 Physical transport properties of marine microplastic pollution flow velocities, pressure gradients and inhibit flow may also influence whether or not microplastic pieces are deposited or eroded.

Modelling transports of benthic microplastics
The Nazaré Canyon model reveals that HD black pellets are dispersed extremely slowly and only 0.05-2.6 % of pellets escape within a 56 day period from their original location.

5
Pellets near shore were transported regularly by tidal forces but were not dispersed. Sudden displacement events occurred more frequently at depth; possibly as a result of increased turbulence following storms, breaking internal waves, canyon turbidity flows, etc. Considering that the majority of plastic waste comes from land (Cole et al., 2011), this data may indicate that the majority of small HD plastics remain in coastal areas.

10
This is supported by Mordecai et al. (2011) who found a correlation between macro debris in the Lisbon and Setubal submarine canyons off Portugal and distance from the coastline. The model results also suggest that debris which is displaced from the coast becomes well dispersed within the shelf, slope and abyssal ecosystems, potentially sparing them from high exposures, but intensifying consequences for more 15 shallow regions. However, this does not mean that benthic debris may not accumulate in certain zones where further transport is inhibited by benthic topography as reported by Galgani et al. (1996) and Mordecai et al. (2011). Models could be used to locate and identify these areas from high resolution physical oceanography and topography data. For higher accuracy, additional local field data (e.g. particle counts from box core 20 sediment samples and near bottom sediment trap samples) should be used with such models.
Critical erosion values in this model were determined in a laminar flow environment by simulating the logarithmic benthic boundary layer velocities found in deep sea environments (Thomsen et al., 2002). However, turbulent flows generated by tides, waves 25 and uneven bottom surfaces also play a role in the resuspension of benthic particles. Using both laminar-and turbulence-induced critical erosion shears would improve future models. The pelagic-benthic water column simulation tanks used in the turbulence 18771 Introduction  1.) can also accurately mimic turbulence-induced benthic shears (Crawford and Sanford, 2001). When modelling benthic microplastic transport in various locations it must be considered that critical erosion shear values for microplastics and the forces required for resuspension may vary between shallow coastal areas and the deep seafloor. Wave action, tidal flows and the depth of the mixed layer may 5 well have greater influence in shallow areas, with bottom currents, turbidity flows and storm events of greater significance in deeper areas. In Table 7, a comparison between the HD black pellets from this study and similarly sized organo-mineral aggregates of the Iberian continental margin studied by Thomsen et al. (2002) is made for average diameter (d 50 ), density, settling velocity and critical shear stress. In Fig. 11, the difference is illustrated by plotting each of these on the quartz erosion curve as taken from Thomsen et al. (2002). The pellets have a relatively high settling velocity and erosional shear stresses approximately 5 times greater than 4 mm aggregates; overall, they behave more similarly to large sand particles or gravel than benthic boundary layer aggregates of similar size.

Factors influencing vertical transport of neustonic microplastics
The vertical transport of buoyant microplastics may be as important as horizontal transport when determining the extent of debris in regions of the ocean and consequences for marine ecosystems. For example, microplastics which reside, even temporarily, within the water column may be consumed by pelagic organisms, not only those feed- 20 ing at the surface.
The turbulence assay was used to simulate wind shear-and wave-generated turbulence within the upper meter of the open ocean (MacKenzie and Leggett, 1993) with the aim of correlating wind speeds with the percentage of each investigated plastic type vertically displaced. Wind is a good proxy for estimating turbulence intensities of 25 the upper mixed layer (MacKenzie and Leggett, 1993); however, comparing laboratory turbulence levels to those found in natural systems is difficult due to scale differences and methods of turbulence generation. In natural systems vertical displacement may 18772 Introduction  (Sanford, 1997;Petersen et al., 2009) and observations were compatible with field observations of trends relating surface trawl counts and wind speed Kukulka et al., 2012). For example, Moore et al. (2001) reported that subsurface trawls in the North Pacific gyre contained mostly biofilmed filament type plastics, which were also the most readily submerged in the turbulence assay of this study. At the ocean surface, turbulence intensities decrease rapidly with depth (MacKenzie and Leggett, 1993); thus, turbulence intensities within the simulation tank could be used to imitate a certain depth within the mixed layer. From this point of view, further studies 15 may reveal insights into how much time a particular particle remains below the surface under particular wind conditions and how quickly the vertical distribution of neustonic plastic adjusts to changes in wind speed.
The pressure assay demonstrates the potential for very low density plastics to be trapped at greater depths within the ocean basins in the event that it is brought to 20 those depths via e.g. a sinking animal carcass which had ingested the plastics while alive. While this is not a likely sink pathway for large quantities of plastics it may present unknown consequences to abyssal ecosystems.

Conclusions
This investigation was an attempt to gauge the degree to which the intrinsic proper- 25 ties of plastic debris fragments affect their transport within the marine environment. From these results, it can be clearly seen that the density, shape and size of a piece of plastic influence its transport, in addition to the external forcing parameters such as seawater density, seabed topography, flow velocity, turbulence, and pressure. It is crucial to understand how microplastics are transported to effectively estimate their global distribution, residence times, convergence zones and ecological consequences using hydrodynamic models. The model runs presented here indicated the slow transport of benthic microplastic in the Nazaré Canyon, which suggests an intensified long term exposure to plastics for those benthic ecosystems. Future research should focus on the ecological consequences of such exposures, particularly in critical areas such as biodiversity hotspots, to allow the development of preventative measures and policy/legislation changes if required. Decreasing the amount of plastic debris originating 10 from urban consumers would greatly reduce exposure levels in many deep sea regions close to shore, such as the Atlantic canyon ecosystems focused on in the current study. Vertical transport of microplastics leads to questions such as (1) to what degree is plastic ingestion by plankton facilitated by the increased encounter rates resulting from turbulence and mixing (Doyle et al., 2011) and (2) how may this affect the rate of 15 persistent organic pollutants (POPs) entering food chain. Further research is needed to be able to more accurately estimate the amount of plastic residing in the oceans and to better understand the behavior of the smallest microplastics and their sinks within the natural environment.
Further investigation of the physical transport properties of marine microplastic pol-20 lution should include the use of models simulating a variety of benthic environments and should incorporate improved simulation techniques of wind induced turbulence, the effect of surface tension on neustonic plastics, how size affects the degree of consequence and risk for organisms and how pressures greater than 200 bar affect plastic buoyancy. 9,2012 Physical transport properties of marine microplastic pollution

5
Appendix B

Relating paddle rotation to turbulence intensity in water column simulation tank
The mean turbulence intensity (root mean square velocity with the mean not removed), (B3) for x, the relationship between rotation rate and turbulence intensity is produced as shown in Eq. (B4): From U RMS the turbulent dissipation rate can be estimated using Eq. (B5), where l is the integral length scale (largest eddy size) (Sanford, 1997;Petersen et al., 2009). In another investigation using the same MEERC benthic/pelagic type C tank, Porter et al. (2000) measured average volume weighted turbulence dissipation rates of 0.08 cm 2 s −3 and average volume weighted U RMS of 1.08 cm s −1 , from which a bulk estimate of the length scale can be approximated to l = 15.7464 cm. In culmination, the paddle rotation rate can be used to approximate ε within the mixing chamber ( Fig. 4 and Table 2), and also to compare the turbulence in the chamber with turbulence in the surface layer of the ocean under various wind conditions.

Tables Figures
Back Close

Full Screen / Esc
Printer-friendly Version
Step       Fig. 11. The critical bed shear stress erosion curve for quartz relates particle (sediment) size to critical shear stress, τ cr , and includes average diameter (d 50 ) 4 mm benthic boundary layer aggregate data point (Thomsen et al., 2002). The mean HD black pellet size (d 50 ∼ 4 mm) and τ cr is plotted over the curve for comparison of aggregate and plastic erosional behavior.