The simulation of the Lagrangian trajectories of microparticles (simulated MPs) released at the sea surface located at the north of the Canary Islands archipelago was performed under Matlab. One thousand massless particles randomly distributed were daily deployed on the region between 30-31N and 19-10W. These particles were driven by the daily velocity field operationally provided by the Atlantic-Iberian Biscay Irish- Ocean Physics Reanalysis (IBI) product available at the Copernicus Marine Service (https://marine.copernicus.eu). The system is based on an application of the eddy-resolving NEMO model run at a horizontal resolution of 1/12 (~8-9 km) (Madec et al., 1998). The velocity data are distributed vertically over 50 z-levels with a resolution decreasing from about 1 m in the upper 10 m to more than 400 m in the deep ocean. The simulations were performed with the data set provided for the years 2020 and 2021 at a depth of 0.5 m. A total of 365000 trajectories per year were produced and each trajectory spanned 90 days. Finally, once a particle reaches a beach, its trajectory concludes, without accounting for potential effects of wave washing.

A similar approach was followed at intermediate levels to produce Lagrangian trajectories related to the spreading of Mediterranean Water in the eastern North Atlantic focusing on the region around the Canary Islands. In this case, the velocity data were also provided by Copernicus though their product GLORYS12V1, which has a regular grid of 1/12 with 50 vertical levels at standard depths. Stokes drift and small-scale turbulence are not included in the models used to produce the Lagrangian trajectories. In this case, 60501 particles were released at 1062 m depth in the vicinity of Cape Sant Vincent and followed for 4 years (2017-2020). In both cases, surface and intermediate levels, the Lagrangian trajectories are defined by the ODE ∂ x → ∂ t = v →   ( x , y , t ) , with the initial condition x(t₀)=x₀ (van Sebille et al., 2018; Nordam and Duran, 2020). An explicit Runge-Kutta scheme of order 4-5 in both space and time was then used to solve the ODE (Tyranowski and Desbrun, 2019; Komen et al., 2020). Windage data has not been considered on this model as MPs do not significantly protrude above the water’s surface (Onink et al., 2019).

World Ocean Atlas data (Reagan, 2023) are used to feature the climatological distribution of salinity and temperature, extracted from the Climate Normal Fields Database. In particular, their objectively analyzed mean on 1/4° and 1° grid of all available decades (1991-2020) was used. Horizontal contours and diagrams were generated with Matlab R2022a.

MPs samples were collected during the oceanographic cruise VULCANA-III-0421 in the southern region of El Hierro island (Canary Islands, Spain). The sampling was conducted onboard R/V Ángeles Alvariño, which belongs to the Spanish Institute of Oceanography (IEO-CSIC). The collection samples were taken at five different stations between April 4^th and 5^th, 2021, specifically at coordinates 18°20’ W and between 26°40’-27°40’ N (see the stations map at Supplementary Figure 1 .

To obtain discrete water samples for MPs analysis, a rosette of 24-12-L Niskin bottles was used to cover the water column from 0 to 1200 m depth. Seawater samples were collected and passed through filters with a mesh size of 100 µm. Each sample consisted of 60-L of seawater, obtained from 5 Niskin bottles (12-L per Niskin). This filter was subsequently washed with Milli-Q water, and the concentrated material was retained on a 47 mm Whatmann GF/F filter for MPs evaluation. The samples were stored at -20°C for later visual examination and counting. Additionally, a blank sample was processed to account for any contamination during the filtration procedure, and it was analyzed onboard.

An Euromex® Nexius Zoom EVO stereo-microscope with a digital camera (model NZ1703S) was used for MPs visual identification, counting and classification.

MPs were analyzed by micro-Fourier Transform Infrared Spectroscopy (µ-FTIR), using a Nicolet™ iN10 infrared microscope (Thermo Fisher Scientific, USA) for a preliminary study to identify the composition of MPs, limited to particles larger than 100 µm. A total of 146 fibers and 114 fragments were analyzed. To carry out the analysis, items were randomly selected and handpicked using a fine stainless-steel tweezer. Then, they were placed in a numbered spot of a reflection slide made of low-e glass slide. The analysis was performed using the OMNIC Picta software (Thermo Fisher Scientific, USA) in reflection mode with a LN-cooled mercury-cadmium-telluride (MCT) detector. Spectral resolution of 4 cm^-1 and 16 scans (5 s of acquisition time) were chosen as working conditions in a spectral range from 4000 to 650 cm^-1.

The spectral libraries ‘Biblioteque Particules’, ‘HR Sprouse Polymers by Transmission’ and ‘Polimeri’ supplied by Thermo Scientific, as well as ‘FLOPP’ and ‘FLOPP-e’ open libraries were used as polymer reference libraries (De Frond et al., 2021). The identification was considered satisfactory if the Pearson’s correlation coefficient between reference spectrum and acquired spectrum was ≥70%. Additionally, if the correlation score for a suspected MP was within the range of 60- 70%, the analysis of that particle was carefully to ensure its identification. Moreover, a double-check of the identification was accomplished using Open Specy databases (Cowger et al., 2021).

The arrival of MPs at different beaches and coastal regions in the Canary Islands is not homogeneous, as previous results have shown ( Table 1 ). A high variability is observed between the different islands, which are located less than 100 km from each other, and also within each island (Baztan et al., 2014; Álvarez-Hernández et al., 2019). A similar pattern has also been observed in the Azores, with large variations in MPs concentrations from one island to another, despite having similar orientations and closer distances (Pham et al., 2020).

Several authors have reported that MPs arrive to the Canary Islands coasts as a result of their position under the influence of the North Atlantic Subtropical Gyre (Herrera et al., 2018; Reinold et al., 2020), so MPs concentration is higher at beaches facing north aligned with the predominant current (Baztan et al., 2014; Herrera et al., 2018). However, field data reveal that some of the main MPs hot spots in the Canary Islands face somewhere else ( Figure 1 ) (González-Hernández et al., 2020), which suggest that mesoscale processes as eddies (scales between 10-100 km) might play a dominant role on the transport of buoyant MPs at the ocean’s surface, distorting the classical pattern of MPs being found at facing north beaches.

In this regard, the lagrangian trajectories obtained for 1,000 virtual particles launched between 30-31°N along 2020 are presented in Figure 2 . This same analysis was repeated for each day of 2020 and 2021, producing 730,000 trajectories lasting 90 days from the release date. All these trajectories are integrated according to their frequency of occurrence, hence producing areas of potential accumulation of particles. These integrations cover 3 months of data, obtaining four 2D histograms per year: January-March, April-June, July-September, October-December ( Figure 3 ).

The areas with a higher potential for the accumulation of particles are highlighted in yellow, while the areas with a lower potential to accumulate particles are indicated in blue. These distribution maps are useful to highlight the preferred trajectories in the open ocean but also to reveal the locations in the coast with higher potential to receive and accumulate MPs. These locations with highest potential present indeed a good agreement with four of the previously reported “hot spots” of MPs arrival in the Canary Islands ( Figure 1 ): Famara (Lanzarote Island), Lambra (La Graciosa Island), Playa Grande (Tenerife Island) and Caletillas (Fuerteventura Island). This last hot spot was recently discovered as part of the IMPLAMAC project and the Marine Litter Observatory (IMPLAMAC, 2022; OBAM, 2022). However, according to the model, no apparent accumulation of MPs take place in the westernmost hot spot located at El Hierro island. In this case, we note that it might be insightful to expand the release of particles further west on future simulations to elucidate whether another location with higher potential for the accumulation of MPs is also revealed.

Large-scale features such as the North Atlantic Subtropical Gyre and the eastern boundary represented by the Canary Current play a significant role in the overall transport of MP particles across the major oceans. However, our findings highlight the important role played by mesoscale structures in the vicinity of the Canary Islands, as there seems to be a direct correlation between surface currents at the mesoscale level and the areas with high potential for MPs accumulation. These hot spots exhibit a complex distribution pattern that is not simply oriented from north to south or northeast to southwest. Instead, they display considerable temporal and spatial variability influenced by the presence of eddies, which make these trajectories much more complex than the previously described. In order to accurately predict MP pollution in oceanic island regions like the Canary Islands, it is essential to consider these mesoscale phenomena.

Marine MPs were sampled in the same region up to 1,200 m depth during an oceanographic cruise in 2021. Every single sample presented MPs, as it was also reported previously in the same region with samples from 2019 (Vega-Moreno et al., 2021). Both fibers and fragments were found, with an average concentration at all depths between 1 and 6 items/L per fibers, and between 3 to 220 items/L, being small fragments (<200 µm) the most abundant. This data agrees with the previous data analyzed in 2019.

We note that, in future studies, it would be recommended to expand our knowledge regarding the specific composition of these MPs. Nonetheless, a preliminary assessment of sample composition has been conducted using µ-FTIR, revealing that around 45% of the analyzed fibers are of synthetic origin (67 fibers) and 55% cellulosic (79 fibers). Interestingly, all the analyzed fragments (114 MPs) are of synthetic origin with diverse compositions, such as polyethylene, polypropylene and polyvinyl, among others.

MPs concentration (fibers and fragments) is not homogeneous in the water column. As can be seen in Figure 4 , which shows MPs concentrations with depth, the concentration of small MPs (<200 µm) is much higher (30-220 items/L) than the bigger ones (3-20 items/L); this observation agrees with previous results that indicate that MPs concentration increases as the size decreases (Poulain et al., 2019). Furthermore, MPs concentrations usually present a maximum value around 200 m for both fragments and fibers. Third, a consistent minimum is found for MPs concentrations sampled at 600 m. Finally, a striking second maximum for small MPs fragments is generally observed at 1,100 m.

Figure 5 compiles some photographs illustrating the different types of fibers and fragments observed in such oceanic samples, with predominance of small fragments.

As previously emphasized, there is a notable concentration of small MPs at a depth of 1100 m, which is surprisingly comparable to, or even exceeds, concentrations found at shallower levels at the same latitude (28°N). This leads us to suggest that these plastic fragments < 100-200 µm have not sunk directly from the surface to 1100 m at 28°N but are being transported to the Canary Islands domain and are finally resident at this depth. The proposed process would be as follows: MW leaves the Mediterranean Sea towards the Atlantic Ocean at 36°N through the Strait of Gibraltar (Iorga and Lozier, 1999). Its primary flow is westward and northward, sinking below the North Atlantic Central Water (NACW) down to 1000-1200 m where it stabilizes according to its density. However, MW is also present at latitudes further south around the Canary Islands (Machín et al., 2006; Fraile-Nuez et al., 2010). This near-horizontal distribution of MW in the eastern North Atlantic is summarized in Figure 6 ; Supplementary Figure 3 with a climatological approach, presenting the salinity on the Atlantic Ocean at 1100 m (data from the World Ocean Atlas). Salinity is chosen due to its higher value for this water mass compared to other surrounding water masses at this latitude [actually, the variable contoured in Figure 6 is the salinity anomaly, as defined by Richardson et al. (2000)].

The climatological MW footprint suggests that the Canary Islands region is clearly influenced by MW, retaining some of its properties from its origin at 36°N. In the vicinity of the Canary Islands, salinity clearly decreases latitudinally from approximately 32°N to 29°N at intermediate levels, revealing a reduction of the contribution from Mediterranean Water to the observations in the range from 900 down to some 1400 m ( Supplementary Figure 5 ). To test the feasibility of this hypothesis, we performed a 4-year numerical modelling simulation (2017-2020), where virtual particles were released near Cape St. Vincent at 1062 m depth. The motivation behind releasing particles at this particular location is supported by the recurrent pathway followed by the MW entering into the Atlantic as displayed in Supplementary Figure 4 (see the spread of the high salinity signal); contouring the southwestern slope of the Iberian Peninsula and shedding meddies at Cape St. Vincent (Serra et al., 2002). Thus, the simulation results are also presented in Figure 6 to support the likelihood that marine pollution transported by MW is expected to be found across the Atlantic Ocean wherever MW mixes with other water masses. These results underscore the dynamic nature of the influence of MW in the Canary Islands region. The latitudinal reduction in salinity and the recurrent path of MW indicate a changing pattern, supported by the 4-year numerical simulation. The release of particles near Cape St. Vincent and their dispersion along the southwestern slope illustrate the dynamics related to the MW, highlighting the possibility of finding marine pollution in various areas of the Atlantic where this water mass is observed.

In this regard, the high content of MPs at 1100 m also suggests that MW would leave the Strait of Gibraltar with a large amount of MPs of various kinds and sizes; the larger MPs (LMPs: 1-5 mm) mostly remain close to the surface (if it is buoyant plastic) or sink to the seafloor (for non-buoyant MPs), but the smallest ones (small MPs, SMPs,< 1 mm) sink along within the water mass down to 1100 m and are then transported beyond 36°N, in particular to the latitude of the Canary Islands (28°N) at this depth.

The main reason to observe SMPs at intermediate waters is therefore not due to the sinking process of the plastic particles by themselves, nor to density equilibrium processes associated with the composition of the plastic in relation to the density of seawater (weight-thrust forces), nor it is even due to physical-chemical changes in the plastic by degradation, weathering, or biological processes as biofouling, among others. On the contrary, SMPs sink through the water column behaving like passive drifters. This is related to the drag by subduction processes of the water mass itself, forced by the ocean dynamics. Thus, the SMPs vertical transport (or sinking) is not dictated by the chemical or biological properties of the MPs particles, because their size is so small to counteract ocean transport processes governed by the physical properties of the water masses.

In summary, if MPs are somewhat larger in size (generally > 1 mm), particle density will be a determining factor and the weight-thrust forces will condition the driving of MPs (close to the surface in the case of a buoyant MPs). However, the SMPs cannot counteract the ocean dynamics regardless of particle composition or density and will subduct down to 1100 m where MW stabilizes. For intermediate sizes or near to the limit of this classification between LMPs & SMPs (sizes in the range 0.5 – 1 mm), it will be decisive for its behavior not only their composition, but also their shape, state of degradation, or biofouling. Therefore, the larger and more spherical particles can stay at the surface more easily (for buoyant MPs) (Kooi et al., 2017; Eo et al., 2021). Despite the sinking of the water mass containing them, the larger (0.5 – 1 mm) and spherical particles will persist at the surface, while those that are more flattened, degraded, and smaller will sink with the water mass.