The streaming of plastic in the Mediterranean Sea

The study addresses why the highly polluted, semi-enclosed Mediterranean Sea lacks persistent surface accumulation zones typical of subtropical gyres, and proposes a new framework: instead of seeking accumulation “stocks,” identify crossroad regions that carry large fluxes of plastic. Motivated by high mesoscale variability, complex coastline, riverine inputs, and winds that likely prevent stable accumulation, the research combines basin-wide observations and Lagrangian modeling to detect and quantify plastic crossroads and to assess coastal retention, beaching, and surface sinking processes. The central hypothesis is that despite the absence of steady accumulation zones, the Mediterranean circulation organizes plastic transport into persistent transit pathways where large fractions of debris flow through limited areas.

Prior work has documented global ocean plastic contamination and its ecological and socio-economic impacts, with floating debris transported by currents and accumulating in subtropical gyres. The Mediterranean has plastic levels comparable to major accumulation zones but appears to lack persistent surface convergence features, likely due to strong spatio-temporal variability, mesoscale activity, river inputs, winds, and complex bathymetry. Previous Mediterranean modeling emphasized sources and sinks, coastal retention, and the role of boundary currents, and estimated annual inputs and mass budgets. Observational efforts, including the Tara Expedition, provided basin-scale, methodologically consistent surface concentrations. Processes such as beaching and biofouling-driven sinking are recognized as key sinks, though their rates and timescales vary and remain uncertain across particle sizes and environmental conditions. This background motivates an alternative diagnostic focusing on flux pathways (crossroads) rather than accumulation zones.

Data and observation: The study uses the largest Mediterranean surface microplastic dataset to date from the Tara Expedition (2014), with 122 manta-net stations measuring multiple plastic metrics (by size, mass, items, and surface area). Station data were aggregated by sub-basin (western, central, eastern Mediterranean, Tyrrhenian, Aegean).

Numerical model: The TrackMPD Lagrangian model simulated surface transport of virtual plastic particles released every minute from 01/01/2013 to 12/31/2016. Sources: (a) 185 coastal cities (50% of particles) proportional to city population times country mismanaged waste index; (b) 200 river mouths (30%) proportional to monthly plastic discharge; (c) vessel-related at-sea releases (20%) proportional to 2015 vessel density. Particles are non-inertial passive tracers advected by hourly surface currents and isotropic horizontal diffusion (K_h tested at 0, 5, 10, 15 m²/s). The velocity field combines CMEMS MEDSEA reanalysis surface currents (geostrophic + Ekman, 1/16°, daily) temporally interpolated to hourly and spatially onto the MEDSEA wave hindcast grid (1/24°, hourly) and added to Stokes drift from waves.

Beaching and washing-off: Beaching occurs if a particle reaches shoreline segments with slope <40%; otherwise it returns to water. Shoreline slope maps (25 m resolution) were compiled (EEA/USGS); about 15% of coasts are too steep for beaching. Washed-off probability during storm events follows P = 0.5 exp(−t_a/(2 T_w)), where T_w (25–100 days across scenarios) controls beach residence half-life; storms defined by significant wave height exceeding 95th percentile (1.69 m) at a location, derived from CMEMS wave product. Storms vary seasonally.

Biofouling and surface sinking: Offline, a logistic probability of sinking P_s(t_w) = 1/(1+e^{r(T_gp−t_w)}) was applied to represent biofouling-induced loss from surface, with four biofouling times (T_gp = 50, 100, 150, 200 days) and r = 1/3. Surface sinking rate (g/km²/day) and net beaching rate (kg/km/day) were computed assuming 100,000 tons/year input to scale particle mass; patterns are independent of this scaling.

Simulation design: Sixteen scenarios from combinations of K_h (0,5,10,15 m²/s) and T_w (25,50,75,100 days). A no-shore-slope sensitivity (NSS) was also run (excluded from ensemble). Approximately 1.564×10^8 particles were released; N = 1.472×10^8 used in the ensemble (Scenario M). Particles were typically advected ~380 days; simulation ended 12/31/2017.

Crossroadness diagnostic: A grid (10,255 points; 15.7 km spacing) was used. For each grid point, a circle of radius r = 11 km was defined and the number of unique trajectories N_i passing at least once during 2013–2017 was counted. Crossroadness CR_i = 100 * N_i / N (percentage of total particles passing through). To exclude trivial near-source interceptions, a ~11 km coastal buffer around all land sources (cities/rivers) was imposed; overlapping grid points were removed (9193 candidate locations remain). Crossroads were ranked greedily: the top CR point is first; subsequent points are selected after removing particles already intercepted by higher-ranked crossroads.

Validation and linkage to observations: For each Tara station, a stadium-shaped area around the transect (radius ~5 km) was used to count coincident model particles on sampling day; concentrations were aggregated by sub-basin for comparison. Sensitivity analyses assessed robustness to source proportions (p_c, p_r, p_v), buffer distance, neighborhood radius, diffusion, washing-off, and biofouling.

Outputs: Crossroadness maps and ranked crossroads; cumulative interception percentages; spatial patterns of net beaching and surface sinking rates (2014–2016).

• Existence and efficiency of crossroads: Despite lacking persistent accumulation zones, the Mediterranean exhibits crossroad regions through which large fluxes of plastic transit. The top 20 crossroads collectively intercepted about 13% of all simulated particles while covering only ~0.3% of the Mediterranean surface. The top 60 crossroads intercepted ~21% within less than 1% of the surface. Overall, around 20% of plastic released annually passes through about 1% of basin area. Many crossroads lie near coasts and are associated with boundary currents.
• Example crossroad: A sixth-ranked crossroad north of Mallorca intercepted particles from multiple distant sources, including vessel-released plastics and city-derived debris from Barcelona (~380 km), Valencia (~150 km), and Algiers (~250 km), illustrating mixing of origins and long-range transport.
• Mass flux interpretation: Assuming 100,000 tons/year input and 4 years of release, 1% crossroadness corresponds to roughly 4000 tons transiting during the simulation period.
• Coastal retention and buffer interception: By construction, all particles that left land-source buffer zones and entered the open domain were available for interception; 33% of all released particles crossed the buffer into the domain and were intercepted by crossroads; the remaining 67% stayed within buffers, indicating strong coastal retention consistent with previous studies.
• Beaching and biofouling budgets: By the end of the release period, ~87% of particles were beached and ~12% fully biofouled (sank from the surface). These magnitudes align with prior estimates.
• Spatial patterns of net beaching rates (kg/km/day, 2014–2016): Highest along Egyptian and central Algerian coasts (43–47 kg/km/day); elevated in the Cilician basin, Syrian coast, and Po Delta (10–14 kg/km/day). Lowest rates around southern Crete, Gulfs of Taranto and Lion, and eastern Sardinia (~0.3 kg/km/day). Modelled rates on Corfu matched observations (~1.9 ± 2.3 kg/km/day vs. observed 1.9 ± 2.2 kg/km/day).
• Spatial patterns of surface sinking rates (g/km²/day): Highest (>40 g/km²/day) in the Cilician basin (Mersin Bay, Gulf of Antalya); also elevated in the Adriatic and western Mediterranean (Balearic sector). Lowest in the Gulf of Lion, Gulf of Taranto, and Aegean Sea.
• Robustness: Results were consistent across 16 scenarios and insensitive to variations in source proportions, coastal buffer size, and neighborhood radius. Crossroadness was largely unaffected by biofouling because most interceptions occur within the first days at sea. Including shoreline slope improved correspondence with in situ data; neglecting it degraded agreement.

The findings demonstrate that, although the Mediterranean lacks stable surface accumulation gyres, its circulation organizes plastic transport into narrow transit regions (crossroads) that carry disproportionately large fluxes over very limited areas. This answers the research question by shifting focus from identifying accumulation stocks to locating dynamic flux pathways. The location of these crossroads near coasts and along boundary currents indicates that anthropogenic source distributions and coastal circulation features jointly funnel debris, enabling targeted monitoring and mitigation. Quantified interception efficiencies (e.g., 21% of particles via <1% surface) suggest that strategically placed monitoring or cleanup systems could intercept a substantial fraction of transiting plastics even without persistent accumulation zones. The coastal retention and high beaching rates highlight the dominant role of nearshore processes, with spatially variable exposure risks and management implications. The agreement in patterns with limited observations (e.g., Corfu beaching, qualitative seabed debris distributions) and robustness across parameters increase confidence in the approach. Beyond plastics, the crossroadness framework can generalize to other pollutants and passive tracers, supporting the design of monitoring networks and informing conservation by identifying regions of cumulative exposure for marine life, including nursery and protected areas.

This work introduces and applies the crossroadness diagnostic to reveal plastic flux pathways in the Mediterranean Sea, integrating the largest regional in situ dataset with a high-resolution Lagrangian model. A small set of coastal crossroads intercept a substantial share of basin-wide plastic transit (e.g., ~21% via <1% of surface), reconciling high pollution levels with the absence of persistent accumulation zones. The study quantifies spatial patterns of net beaching and biofouling-driven surface sinking and underscores the primacy of coastal retention. These results provide actionable targets for monitoring and mitigation (e.g., fixed interception stations) and a transferable methodology for other regions and pollutants. Future work should incorporate fragmentation processes, refine and temporally resolve source terms, enhance coastal hydrodynamics with higher-resolution and three-dimensional modeling, and obtain time series of beaching and benthic debris to validate fluxes. Assessing biofouling spatial variability and ecological impacts in crossroads, especially within sensitive habitats and MPAs, will improve risk assessments and management strategies.

• Fragmentation of plastics into microplastics was not modeled due to lack of quantitative descriptions, potentially affecting particle sizes, transport, and sinking behavior.
• Biofouling timescales are uncertain and variable across particle properties and environmental conditions; sinking was handled offline with assumed logistic functions and four timescales.
• Model is primarily surface and two-dimensional for advection; three-dimensional processes (vertical mixing, subduction) are not explicitly resolved, and 3D modeling is recommended.
• Limited observational datasets constrained quantitative validation, particularly for surface sinking and seabed deposition; comparison was largely qualitative except for select beaching data.
• Assumed annual plastic input (100,000 tons/year) sets mass scaling; real inputs may vary spatially and temporally (e.g., tourism seasonality), influencing flux magnitudes.
• Vessel-source representation used shipping density but omitted explicit fishing activity releases that may contribute to seabed litter; shoreline slope and storm threshold choices introduce additional uncertainties.
• Crossroadness measures flux transit rather than standing stock; interceptions are sensitive to definition of buffers and neighborhood radii, though sensitivity tests indicated robustness.