High levels of microplastics and microrubber pollution in a remote, protected Mediterranean Cladocora caespitosa coral bed

High levels of microplastics and microrubber pollution in a remote, protected Mediterranean Cladocora caespitosa coral bed Lars Reuning a,* , Lars Hildebrandt b , Diego K. Kersting c , Daniel Pr¨ ofrock b a Institute for Geosciences, CAU Kiel University, Ludewig-Meyn-Str. 10, 24118 Kiel, Germany b Department for Inorganic Environmental Chemistry, Helmholtz-Zentrum Hereon, Max-Planck-Straße 1, 21502 Geesthacht, Germany c Global Change, Conservation and Genetics of Marine Species, Instituto de Acuicultura Torre de la Sal, Consejo Superior de Investigaciones Científicas (CSIC), Torre de la Sal S/N, 12595 Ribera de Cabanes, Spain ABSTRACT Coral reefs are increasingly threatened by anthropogenic stressors, including plastic pollution. This study investigates the abundance and possible ecological impact of microplastics (MPs) and microrubber pollution in sediments from a Cladocora caespitosa coral bed in the north-western Mediterranean. Despite being located in a remote marine protected area with no local plastic pollution sources, our results indicate exceptionally high MP concentrations (mean: 1514 particles/kg dry weight), attributed to long-distance transport of plastics by the Northern Current. Laser Directs Infrared (LDIR) Chemical Imaging and ATR-FTIR spectroscopy were used to characterize the MPs in terms of size, shape and polymer types. Most MPs are fragments (96 %), while fibers contribute only 4 %. The most abundant polymers were polyethylene (PE, 28 %), polyethylene terephthalate (PET, 25 %), and polystyrene (PS, 19 %), with significant contributions from polyurethane (PU) and microrubber. Particle size analysis showed that 92 % of MPs were smaller than 250 μ m, with a median particle size varying by polymer type. Notably, polymers with heteroatoms in their main chain, such as PET and polyurethane, exhibited significantly smaller median sizes compared to polyolefins, possibly suggesting different degradation pathways. The high MP concentrations measured in sediments within coral colonies suggests that MPs could have adverse effects on heterotrophic feeding in C. caespitosa, a critical energy source during stress events. This study underscores the urgent need for targeted research on MP effects on the resilience of C. caespitosa and for increased global and regional efforts to curb plastic pollution mitigation in order to conserve coral populations in the Mediterranean. 1. Introduction Corals form structurally complex habitats that foster biodiversity and provide important ecosystem services essential to human society (Hughes et al., 2002; Pendleton et al., 2016). The monetary value of these ecosystem services is actually higher than that of any other biome on earth (Groot et al., 2012). It is therefore alarming that coral habitats are in decline. The global coverage of tropical coral reefs and their ability to provide ecosystem services has declined by half since the 1950s (Eddy et al., 2021) while coral bioconstructions in mid-latitude and temperate seas, such as the Mediterranean, have been continuously declining in recent decades (Kersting et al., 2013; Kersting et al., 2022). In fact, these ecosystems are globally the most affected by anthropogenic ocean warming (Bindoff et al., 2019; Garrabou et al., 2009; Garrabou et al., 2022; Hughes et al., 2017). Other stress factors such as chemical pollution are known to interact synergistically with climate change and exacerbate coral loss (Donovan et al., 2021). One stress factor that has gained increasing attention in recent years is microplastics (MPs) pollution (Hall et al., 2015; Pantos, 2022; Reichert et al., 2018; Saliu et al., 2019). Plastic particles <5 mm in size are called MPs (Arthur et al., 2009). Primary MPs are intentionally produced small plastic particles, typically for use in manufacturing or consumer products. Secondary MPs, on the other hand, are created when larger plastic items break down into smaller particles due to natural weathering processes. Secondary MPs can come from a variety of sources, including discarded plastic bags, bottles, and packaging, as well as fishing nets and other marine debris. The uncontrolled burning of waste on beaches makes the plastic even more brittle and susceptible to the formation of MPs (Utami et al., 2023). In laboratory experiments, exposure of corals to microplastics has led to adverse effects such as bleaching, necrosis and reduction in growth rates (Hankins et al., 2021). Long-term exposure experiments using realistic microplastic concentrations confirmed first that microplastic pollution can have species specific negative impacts on reef corals (Reichert et al., 2019) and second, that common reef-building corals cannot adopt to long-term microplastic exposure (Rades et al., 2022). In addition, MPs can promote pathogen transmission, further increasing the susceptibility of reef-building corals to disease (Kirstein et al., 2016). MPs can also act as vectors for heavy metals, either as additives in the plastic or absorbed to its surface (Hildebrandt et al., * Corresponding author. E-mail address: [email protected] (L. Reuning). Contents lists available at ScienceDirect Marine Pollution Bulletin journal homepage: www.elsevier.com/locate/marpolbul https://doi.org/10.1016/j.marpolbul.2025.118070 Received 10 March 2025; Received in revised form 28 April 2025; Accepted 28 April 2025 Marine Pollution Bulletin 217 (2025) 118070 Available online 5 May 2025 0025-326X/© 2025 The Authors. Published by Elsevier Ltd. This is an open access article under the CC BY license ( http://creativecommons.org/licenses/by/4.0/ ). 2021; Patterson et al., 2020). Recent studies on the accelerated weathering of different consumer plastics revealed several 1000 chemical features during non-target chemical analysis originating from the UV degradation of the different organic additives present in the tested materials (Menger et al., 2024). Recent field studies have further shown that microplastic concentrations in the coastal waters can be high enough to be ecologically relevant for coral reef health (Tang et al., 2021). They observed that microplastic concentrations in reef corals correlate negatively to their symbiont density, indicating a negative effect on the coralSymbiodiniaceae symbiosis. In the long term, MPs therefore pose a threat to reef corals and their ability to function as framework builders in coral reef systems. The United Nations Environment Program, therefore considers understanding plastic pollution around coral reefs a key knowledge gap in reef ecosystem research (Sweet et al., 2019). This has sparked a growing interest in the role of plastic, and especially microplastic pollution, as a stressor for coral reefs (Biswas et al., 2024; Huang et al., 2021; John et al., 2022; Utami et al., 2021; Utami et al., 2023). Coral reef systems are an iconic coastal ecosystem of the tropics, but also occur at higher latitudes. Cladocora caespitosa is the only scleractinian, zooxanthellate coral with reef-building capacity in the temperate Mediterranean Sea (Morri et al., 1994; Peirano et al., 2001). This coral plays a key role in forming the benthic habitat in the coastal zone of the Mediterranean and Adriatic Sea. Peirano et al. (1998) described two types of C. caespitosa colony distributions: beds and banks. Beds are composed of a large number of distinct subspherical colonies (e. g., Gulf of Trieste, Schiller (1993)), while banks are made up of large colonies, reaching several decimeters in height and covering several square meters in surface area (e.g. Mjlet National Park, Kruˇ zi´ c and Benkovi´ c (2008)). Mixed distributions of beds and banks can also occur (e.g. Columbretes Islands, Kersting and Linares (2012)). In addition, C. caespitosa can also occur as free-living coral nodules or coralliths, usually smaller than 10 cm in diameter (Kersting et al., 2017a, 2017b). Most reef-building corals acquire their energy mainly through photosynthesis. This process is facilitated by symbiotic algae known as zooxanthellae, which convert sunlight into nutrients to maintain the corals metabolism. Heterotrophic feeding on e.g. planktonic zooplankton is used as a supplementary energy source by many tropical coral species (Houlbr` eque and Ferrier-Pag` es, 2009), particularly during stress events when photosynthesis may be compromised (Ferrier-Pag` es et al., 2010; Grottoli et al., 2006). The ability to use both feeding strategies is called heterotrophic plasticity. C. caespitosa is particularly dependent on heterotrophic plasticity, as it covers a large part of its energy demands in winter through heterotrophic feeding (Ferrier-Pag` es et al., 2011). Similar to its tropical counterparts, warming-induced mortality from marine heatwaves is the main threat to C. caespitosa (Kersting et al., 2013). Several tropical coral reef species respond to thermal stress by increasing their food intake, which can make them more resilient to bleaching (Grottoli et al., 2006). This has also been hypothesized for C. caespitosa (Quintano Fernandez et al., 2024). On the other hand, a higher feeding rate could also increase the risk of unwanted interactions with MPs. Corals can trap and ingest MPs due to its similarity in size to plankton (Hall et al., 2015; Reichert et al., 2018). A laboratory study suggests that heterotrophic feeding is a major factor contributing to the impact of MPs, with high reaction and ingestion rates in highly heterotrophic species (Reichert et al., 2024a). In fact, heterotrophic feeding has been suggested as the pathway of incorporation of other kind of anthropogenic particles (fly ash) in C. caespitosa skeletons (Roberts et al., 2024). Tremblay et al. (2011) reported that C. caespitosa has a greater capacity for heterotrophy than other scleractinian symbiotic corals and Roberts et al. (2024) suggested that these traits may favor the uptake of micro-particles by C. caespitosa. Ingesting MPs instead of food could impact coral resilience by reducing the effectiveness of feeding as an alternative energy source during heat-induced bleaching. Additionally, plastic associated contaminants (e.g. phthalic acid esters) with potential adverse effects on coral health (Saliu et al., 2019), have been detected in Mediterranean anthozoans, including the highly heterotrophic C. caespitosa (Gobbato et al., 2024). Mediterranean coral bioconstructions formed by C. caespitosa therefore likely face multiple stressors not only from anthropogenic warming but also plastic pollution. The north-western Mediterranean is a hotspot for marine litter (SotoNavarro et al., 2020). The concentration of floating plastic debris in the area is one of the highest in the world rivaling the famous plastic garbage patches in the subtropical gyres (C´ ozar et al., 2015). This region harbors one of the largest C. caespitosa bioconstructions in the Mediterranean Sea, located in the Columbretes Islands Marine Reserve. In this site, C. caespitosa forms a mixed distribution of beds and banks with a total coral cover of 2900 m 2 (Kersting and Linares, 2012). This remote archipelago is considered a global change sentinel site, hosting decades long monitoring on coral health and temperature (Kersting et al., 2013; Kersting and Linares, 2019). We studied the sediments from this C. caespitosa population to analyze how strongly their habitat is affected by microplastic pollution. The samples were taken from the seafloor adjacent to C. caespitosa colonies and from sediments trapped within the colonies. In particular, the study aims not only to analyze the concentration of microplastics in the sediment, but also to characterize the size and polymer type of all microplastics found in the samples. The relative abundance of the polymer types and their concentration in the sediment are used to calculate pollution risk indices that help to evaluate their environmental impact. In the study, the particle-size distribution of the microplastics is examined in particular, as it has been shown that the size of the microplastics can influence their interaction with the corals (Hankins et al., 2022; Hankins et al., 2021). In combination these data provide important insights into the role of microplastics as a potential stressor for one of the largest bioconstructions of the scleractinian symbiotic coral C. caespitosa. 2. Materials and methods 2.1. Study site Illa Grossa (39◦53.825 ′ N, 0◦41.214 ′ E), the largest of the Columbretes Islands (0.14 km 2 ), is a submerged Quaternary volcanic caldera (Mu˜ noz et al., 2005). It is located on the Spanish outer Mediterranean shelf, approximately 55 km off the Spanish coast (Fig. 1). The C-shaped island is open to the northeast in the main direction of winter storm waves (Kersting and Linares, 2012) and the Northern Current (Ourmieres et al., 2023), which flows parallel to the slope along the continental margin of the Iberian Peninsula (Fig. 1). There are no beaches; the rocky slopes of the islet drop steeply into the bay to a minimum water depth of 5 m. The average depth of the bay is 15 m, with a maximum depth at 30 m. The coral colonies form beds and banks in the rocky bottoms of this semienclosed bay (Fig. 2), with highest coral cover in a depth between 10 and 20 m. The coral colony size varies between 5 and 150 cm, with a mean colony age of ~50 years, based on average growth rates (Kersting and Linares, 2012). Coral cover exceeds 10 % in the northwestern and southeastern parts of the bay (Fig. 2, Fig. S1), where the corals are relative protected from the strong NE storms in autumn and winter. These two areas are separated by a NE–SW running central channel where coral cover is generally below 2 % (Fig. 2). These central and deeper parts of the bay are covered with sand and gravel from alkaline volcanic rock and bioclasts (Kersting and Linares, 2012). The studied bay is part of the Columbretes Island Marine Reserve, a 5400-ha marine protected area which was established in 1990 to protect the area from direct anthropogenic influences. The only recreational activity permitted in Illa Grossa Bay is scuba diving, but it requires authorization from the marine reserve management authorities. The study area therefore is ideal to test for the first time the level of microplastic pollution on a remote, protected Mediterranean coral ecosystem. L. Reuning et al. Marine Pollution Bulletin 217 (2025) 118070 2 2.2. Contamination mitigation A strict protocol was followed in order to prevent contamination from the sampling equipment, laboratory equipment, reagents, clothing and airborne sources. This included the rigorous use of laminar flow benches class II, filtration of all reagents, the use of cleaned metal or glass laboratory equipment and the conduction of method blanks. However, no MPs in the investigated size range were found in the blank test that was performed to evaluate the background values of MPs. A more detailed description of the conducted contamination mitigation procedures can be found in Hildebrandt et al. (2022b). 2.3. Sample collection Five sediment samples were collected by scuba diving in the bay of Illa Grossa between 2017 and October 2022. Two sediment samples were recovered from ~10 cm high, living C. caespitosa colonies from a water depth of 15 m (Fig. 2, Table 1). Based on an average growth rate of 3.2 mm/year (Vergotti et al., 2025), the coral colonies represent a growth history of ~30 years starting in the mid-1980s.The relatively open, phaceloid growth form of C. caespitosa colonies (Fig. 2) is favorable for the accumulation of sediment between the individual corallites. The sample size was limited by the amount of sediment contained in the coral colonies (11 and 61 g dry weight (dw)). Three additional sediment samples of about 200 g dw each were collected from the seafloor (0–5 cm depth in sediment) in water depth between 17 and 21 m (Table 1). Two of these samples were taken adjacent to C. caespitosa colonies in areas with a total coral cover between 10 and 11 %, while one sample was taken in the area of sparse coral cover (1 to 2 %) in the NE–SW central channel of the bay (Fig. 2, Table 1). All samples were taken to the laboratory at CAU Kiel and freeze-dried to constant weight. 2.4. Grain-size analysis Dried samples were sieved by hand through 1 mm, 2 mm, 4 mm and 8 mm sieves (Retsch GmbH, Germany) at CAU Kiel. All individual grainsize fractions were weighed. Mean grain size and sorting were calculated using the software Gradistat (Blott and Pye, 2001). The sorting of samples was quantified as graphic standard deviation using the equation of Folk and Ward (1957). After grain-size analysis the samples were split in two fractions (<1 mm and >1 mm) for further processing. 2.5. Statistical analysis Two non-parametric statistical tests were used to analyze the particle-size distributions of all polymer types. The Mann-Whitney pairwise U test was used to test if the median particle size of different polymers is the same. The Kolmogorov-Smirnov two-sample test was used to test if the particles from different polymers show the same size distribution. A Monte Carlo Simulation was used to evaluate the likely Fig. 1. Location of Illa Grossa (yellow star), the largest of the Columbretes Islands in the north-western Mediterranean Sea. Illa Grossa is located in the pathway of the Northern Current, the most important boundary current of the north-western Mediterranean. It flows parallel to the slope along the continental margin of the Iberian Peninsula. Northward flowing, secondary currents connect the Balearic Sea with the Algerian Subbasin (Amores et al., 2013). Interannual variations in the regional oceanography are caused by mesoscale features (dashed line), such as eddies (Ourmieres et al., 2023). Base map created with GeoMapApp (www.geom apapp.org)/CC BY using the Global Multi-Resolution Topography (GMRT) synthesis (Ryan et al., 2009). The inset shows a satellite image of Illa Grossa, a C-shaped drowned volcanic caldera open to the northeast in the main direction of winter storm waves and the Northern Current. Satellite image from Google Earth©. (For interpretation of the references to colour in this figure legend, the reader is referred to the web version of this article.) L. Reuning et al. Marine Pollution Bulletin 217 (2025) 118070 3 range of hazard scores for the polymer group PU/acrylate/varnish (see below). The PAST 4.0 software (Hammer et al., 2001) was used for all statistical analyses, with the exception of the Monte Carlo Simulation, which was carried out in Microsoft Excel 2019. 2.6. Processing of >1 mm fraction The >1 mm fraction was checked for MPs under a binocular microscope (Zeiss, Discovery.V8, Germany). Large MPs were identified following protocols developed by Lusher et al. (2017). Unnatural colors and/or shininess and unnatural forms/structures were used as indicators of potential MPs. Particles with potentially cellular or organic structures and translucent fibers were rejected as MPs. Translucent fibers and fibers that were not characterized by three-dimensional bending and uniform thickness were also rejected (Martin et al., 2017). All suspected MPs in the >1 mm fraction were photographed and their size measured on the digital photographs using the software Fiji (Schindelin et al., 2012). Fig. 2. Map and photos of C. caespitosa coral banks in Illa Grossa Bay. A) Map of coral cover (%) in Illa Grossa Bay (modified from Kersting and Linares (2012), note that the isolines in this map reflects the coral cover values, not the sea-floor morphology). The inset at the top right shows an enlargement of the sampling area near the largest coral banks in the south-west of the bay. The sampling sites are marked as yellow stars. B) Large (~ 100 cm in diameter) C. caespitosa colony near sampling sites IG4 and 5. C) The seafloor at a water depth of ~17 m near sampling site IG3 consists mainly of sandy gravel from alkaline volcanic rock (black) and bioclasts (white). The width of the field of view in the foreground of the image is ~20 cm. D) Close-up of a C. caespitosa colony showing a relatively open, phaceloid growth form with sediment trapped between the individual corallites (blue arrow). The typical diameter of C. caespitosa corallites in this image is 4–5 mm. (For interpretation of the references to colour in this figure legend, the reader is referred to the web version of this article.) Table 1 Sampling date, location, environmental parameters and sediment texture at sampling sites. Sample Date Coordinates Water depth (m) Coral cover Sediment texture Sorting Grain size (mm) IG1 Oct. 2022 39.89635◦; 0.68643◦21 1–2 % Sandy Gravel poor 1.5 IG2 Oct. 2022 39.89585◦; 0.68668◦17 10–11 % Gravelly sand moderate 1.3 IG3 Apr. 2019 39.89579◦; 0.68674◦17 10–11 % Gravelly sand poor 1.2 IG4 Aug. 2017 39.89574◦; 0.68667◦15 coral colony Gravelly sand poor 0.9 IG5 Aug. 2017 39.89572◦; 0.68671◦15 coral colony Sandy Gravel poor 1.7 L. Reuning et al. Marine Pollution Bulletin 217 (2025) 118070 4 2.7. Polymer characterization (>1 mm fraction) Potential large MPs (>1 mm) were analyzed at the HelmholtzZentrum Hereon by attenuated total reflectance - Fourier transform infrared (ATR-FTIR) spectroscopy (Alpha I, diamond ATR crystal, Bruker Daltonik GmbH, Bremen, Germany). Measurements were performed three times with 32 scans and a resolution of 4 cm −1 (wavenumber range: 4000 cm −1 –400 cm −1 ). Spectra were compared with reference spectra from the siMPle library developed by Primpke et al. (2020b, 2020a). Spectral assignments (vector-normalized first derivatives) with a hit quality index (HQI) ≥700/1000 were accepted. 2.8. Processing of <1 mm fraction MPs were processed at CAU Kiel by carbonate dissolution followed by density separation and organic matter digestion. To prevent CO 2 outgassing in the acidic flotation media, the carbonate clasts in the <1 mm fraction were dissolved using hydrochloric acid (10 % v/v) until the reaction stopped. Subsequently, up to 30 g of dried sediment was mixed with 80 ml CaCl 2 (density: 1.5 g/cm 3 ) salt solution in a beaker. CaCl 2 - solution has the advantage of being non-toxic and relatively cheap. Sediment and solution were stirred three times for 10 min and afterwards left to settle for 1 h until the supernatant was clear. Floating solids were separated via overflow from the supernatant, collected and transferred to a cellulose nitrate filter with a pore size of 0.45 μ m. These separation steps were repeated for 30 g batches of sediment until the entire sample was processed. The recovery rate of this method was monitored in the laboratory at CAU Kiel using artificially spiked sediment samples. Particles of common polymer types (PP, HDPE, LDPE, PS, PVC) covering a large density (0.9–1.38 g/cm 3 ) and size (100 μ m to 1 mm) range were mixed with pre-cleaned sand. Using the method described above, the recovery rate of MPs was 97 ±5 %. At HelmholtzZentrum Hereon, two samples were subjected to an additional treatment with H 2 O 2 solution (30 %, v/v) at 40 ◦C for 24 h, to oxidize excess organic matter. The suspensions (V (50 % ethanol) =80–100 ml) were pre-concentrated (T heater cover =40 ◦C, T heater base =40 ◦C) to a final volume of 1 ml using a Syncore-Plus® automated evaporation system in conjunction with a EasyFill Rack R-12 Polyvap, a Vacuum Cover R-12 with PTFE sealing disks, a set of 12 graduated glass tubes for EasyFill rack R-12 Analyst, residual volume 1.0 ml, and a Flushback Module R-12 (Büchi Labortechnik AG, Flawil, Schwitzerland). Subsequently, all particles were transferred onto MirrIR slides (Kevley Technologies, USA) for the final laser direct infrared analysis using a glass pipette. L´ opezRosales et al. (2022) achieved high recoveries (88 %) and high precision (RSD =4 %) for microplastic particle transfer using the Syncore® automatic evaporation system. 2.9. Polymer characterization (<1 mm fraction) The <1 mm sample fraction was analyzed at the Helmholtz-Zentrum Hereon using the Agilent 8700 Laser Directs Infrared (LDIR) Chemical Imaging system (Agilent Technologies) in transflection mode. The instrument’s functional principles are described in more detail in previous publications (Da Costa Filho et al., 2020; Dong et al., 2022; Hildebrandt et al., 2020; Scircle et al., 2020). The particle analysis workflow of the Agilent Clarity software (version 1.1.2) was used for the automated analysis of the entire sample set. Hereby, the sensitivity was set to the maximum (6/6). The particle analysis workflow includes a complete analysis of the size and shape of all particles. Fibers were distinguished from fragments based on their elongation factor (aspect ratio) of >3 (Hildebrandt et al., 2020). In the following, we use the term particles to refer to both shape classes, fragments and fibers. Spectra were acquired with a spectral resolution of 8 cm −1 . The particle size range of the LDIR imaging system was set to 50 μ m–5000 μ m. The automatic workflow of the LDIR technically enables MP detection down to 10 μ m. However, the practically achievable size detection limit highly depends on the analyzed matrix and the level of cleanliness of the sample. The used spectral library (Microplastic starter 1.0, Agilent Technologies) was expanded by spectra of in-house reference MPs and of relevant environmental particles (Hildebrandt et al., 2022a). The automated workflow of the Clarity software acquired IR spectra for all particles. The hit quality thresholds for a positive assignment were adapted according to the preset values. MP identifications were either accepted, manually assigned to another polymer class, or not accepted. Only spectra in conjunction with high hit quality values related to reference spectra (>0.80) were considered for the final statistics without further manual confirmation. In order to prevent any overestimation of the MP concentrations, all analyses were thoroughly re-evaluated manually in transflection mode and, if necessary, also by the LDIR’s μ -ATR function. If unambiguous confirmation of the assignment was not possible, the respective particles were assigned to natural material classes or marked as “unknown”. Particles with hit qualities <0.60 were automatically classified as “unknown”. Acrylates, polyurethane (PU) and varnish are polymers that are difficult to distinguish using FTIR spectroscopy (Primpke et al., 2018) and LDIR and were therefore analyzed as one class of polymers (Hildebrandt et al., 2022b) referred to as PU in the following. The wavenumber range of the LDIR (1800 cm −1 –975 cm −1 ) hampers the accurate differentiation between natural polyamide (PA) and synthetic PA in environmental samples (Hildebrandt et al., 2022b). Particles assigned to PA by the automated workflow of the Clarity software were therefore excluded from further analysis. The few polytetrafluoroethylene (PTFE) MPs that were detected in the samples were also discarded from further analysis, since laboratory equipment containing PTFE were used during sample processing. Synthetic rubber differs from conventional plastics in that it is an elastomer formed by the connection of polymer chains with sulfur bridges. The majority of rubber elastomers in use are a blend of natural and synthetic rubber. Microrubber therefore is not identical to MPs in a strict sense, but was included in the analysis since it is an important pollutant with a similar effect on corals (Reichert et al., 2024b). 2.10. Calculation of risk indices The health hazard of different polymers varies widely (Lithner et al., 2011). The environmental risk of MPs pollution therefore depends not only on the concentration of MPs, but also on the relative abundance of different polymers (Xu et al., 2018). Most environmental risk assessments therefore are based on these two parameters, even if the calculation of environment risks differs in detail in the individual studies. We calculate the Contamination Factor (CF) to compare the relative abundance of MPs between different sampling stations and with the literature (Tomlinson et al., 1980; Xu et al., 2018): CFi=Ci/Co(1) Where i denotes the sampling sites and CF i is the Contamination Factor for the i th sampling site. C i represents the abundances of MPs at the i th site and C 0 denotes the baseline concentration of MPs in sediments. The lowest MPs concentration found in this study was chosen as baseline concentration, since it is nearly identical to the lowest concentration reported for sediments from the Spanish Shelf (Filgueiras et al., 2019). A Polymeric Risk Index (H) expresses the combined environmental hazard of all polymers present in a sample and was adopted from previous studies (Rakib et al., 2022; Xu et al., 2018): Hi=∑(Pji/Ci*Sj)(2) Where H i is the Polymeric Risk Index at sampling site i. Pji is the abundance of the specific polymer j at the sampling site i, C i is the total abundance of MPs at the sampling site i and S j is the hazard score of the specific polymer j. The values for all polymer types present at the site are summed up to calculate H i . The hazard scores of the polymers are based L. Reuning et al. Marine Pollution Bulletin 217 (2025) 118070 5 on Lithner et al. (2011), with PP =1, PET =4, PE =11 and PS =30. Acrylates/polyurethane/varnish is a group of polymers which are difficult to distinguish using FTIR or LDIR (Primpke et al., 2018; Hildebrandt et al., 2020). The polymers within these group have different hazard scores, but include many polymers that are categorized as particularly hazardous by Lithner et al. (2011). The hazard scores for nine different polymers from this group are reported by Lithner et al. (2011) and range from 230 to 13,844. A Monte Carlo Simulation was employed to estimate the potential distribution of total hazard scores for a sample composed of these nine different polymers. We simulated their unknown relative abundances using 1000 iterations to represent possible compositions. For each simulated composition, a weighted average hazard score was computed. The resulting distribution of hazard scores was used to assess the likely range of hazard scores of this polymer group statistically. The average hazard score of the distribution is 7197, while 5386 represents the 5th percentile and 8912 the 95th percentile of the distribution. These values can be used as input for the hazard score of the acrylates/polyurethane/varnish group in the calculation of the Polymer Risk and the Pollution Risk indices. This results in a range of likely Polymer Risk and Pollution Risk indices for each sample containing polymers of the acrylates/polyurethane/varnish group. The polymer PVDC is not directly covered in Lithner et al. (2011). However, vinylidene chloride with a hazard core of 111 (Lithner et al., 2011) is the only monomer used in the production of PVDC. A hazard core of 111 is therefore also assigned to PVDC. No hazard score can be given for rubber, which is therefore not considered for the calculation of the Polymeric Risk Index. For a given sampling site (i) the Polymeric Risk Index (H i ) is combined with the Contamination Factor (CF i ) to calculate the Pollution Risk Index (PRI i ) following the equation: PRIi=Hi*CFi(3) The Polymeric Risk Index (H) and the Pollution Risk Index (PRI) for the entire study area are expressed as geometric means. Risk categories for each of the indices (CF, H and PRI) were defined from low (I) to very high (V) and reported in Table S1. 3. Results The most important sedimentary components are volcanic-rock fragments, minerals and bioclasts. The most frequently identified bioclasts originate from echinoderms, bivalves, corals and barnacles. Bioclasts from calcareous red algae, gastropods and polyplacophorans are less abundant. The most important information on the texture of the sediment samples is summarized in Table 1. Mean grain sizes vary between coarse sand (0.9 mm) and very coarse sand (1.2 to 1.7 mm), while the sediment texture can be classified as sandy gravel (IG1 and 5) or gravelly sand (IG2 to 4). The sorting of the sediment is generally poor, except for sample IG2 which shows a slightly better moderate sorting (Table 1). The grain-size distributions of all samples show a slightly negative skewness, i.e. an asymmetric grain size distribution towards coarser grain sizes. The grain-size distribution is unimodal in all samples except IG4, which is bimodal. The bimodal distribution in this sample from a coral colony is due to the contribution of relatively large coral fragments. This leads to a second peak at larger grain sizes in the distribution. Relatively large coral fragments also contribute to the relatively large average grain size of the sediments in the other coral colony sample (IG5). Otherwise the grain size tends to increase with water depth and decrease with coral cover (Table 1). The automated particle analysis workflow of the Agilent Clarity software identified a total of 3157 particles from all sampling sites, which were analyzed using LDIR. Two hundred three (203) of these particles were characterized as MPs, including rubber. The majority of the MPs are fragments (96 %), while fibers contribute only 4 % (Table S2). The concentration of MPs varies strongly between 41 and 6345 particles/kg dw (Table 2), with a mean of 1514 particles/kg dw and an interquartile range of 530 particles/kg dw. The lowest value was found at site IG1 located in the central channel without significant coral cover, while the highest concentration was found in sediments trapped in one of the coral colonies (IG5, Table 2, Fig. 2). All, except two MPs, are smaller than 500 μ m (Fig. 3) and 92 % of all MPs are smaller than 250 μ m. The two larger outliers are 736 and 2009 μ m in size. The mean particle size of all MPs excluding the outliers is 123 ±88 μ m (Table 2). The size distribution of MPs shows a general increase in abundance towards the detection limit at 50 μ m. There is no clear relationship between the mean grain size of sediment in a sample and the size of MPs. This lack of correlation appears to be primarily due to the abundance of relatively large coral bioclasts in the coral colony samples. Instead there is a general tendency for the size of MPs to decrease with the density of the coral canopy (Tables 1 and 2). PE and PET are the most common polymers in the study area and together account for >50 % of all MPs (Table 2, Fig. 3). The fibers in particular, which only account for a relatively small proportion of all particles (4 %), consist mainly of PET (75 %). PP, rubber and PS each contribute between 10 and 20 % of all polymers, while PU and PVDC are less abundant (<5 % each). The particle size of MPs varies strongly with polymer type (Fig. 3), e. g. PE and PS are the only polymer types present in the size range >250 μ m. In contrast, all PU and PVDC particles and >90 % of all PET particles are smaller than 100 μ m. This results in median size values of <100 μ m for PP, PU, rubber, PET and PVDC, while PE and PS have median size values of 141 and 134 μ m respectively (Table 3). The Mann-Whitney U test confirms that the median size values of most polymer types are different from each other. This is supported by the Kolmogorov–Smirnov test, which shows that the grain size distributions of the polymer types, which few exceptions, are different from each other (Table 3). The CF values show a high variability (Table 4) and range from 1 to 115.5, indicating moderate contamination in two samples (IG1 and 2) and very high contamination in three samples (IG3–5). In average the contamination in the study area is very high, as indicated by an average CF of 28.8 (Table 4, Fig. 4). The Polymeric Risk Index for the individual samples, varies between low and very high risk. This variability primarily reflects the relative abundance of PU (Table 2), the polymer group with by far the highest hazard score. The Polymeric Risk Index (H) for the entire study area generally indicates a considerable risk (Table 4). A moderate risk only applies under the assumption of a low hazard score for PU (5th percentile of the Monte Carlo Simulation). The Pollution Risk Index (PRI) indicates a high risk for the study area, while the risk for individual samples ranges from low to very high (Table 4, Fig. 4). With the exception of sample IG5, the Pollution Risk Index is mainly determined by the Polymeric Risk Index (H) rather than the concentration factor (CF). This is because the hazard scores of the polymers vary more strongly compared to the MP exposure level expressed as concentration factor (Table 4). Only the Pollution Risk Category of sample IG1 is influenced by the chosen hazard score for PU, indicating high risks for high and average hazard scores but only considerable risk for low hazard scores (5th percentile of the Monte Carlo Simulation). Overall, this shows that the risk assessment does not depend significantly on the exact hazard score selected for PU. 4. Discussion The concentration of floating plastic debris in the north-western Mediterranean is one of the highest in the world and similar to the plastic garbage patches in the subtropical gyres (C´ ozar et al., 2015). Fagiano et al. (2023) recently confirmed that this area, including the Columbretes Islands, receives high amounts of floating plastics. Therefore, it is not surprising that the absolute concentration of MPs in the sediments of Illa Grossa is very high (Tables 2 and 4). However, the average pollution expressed as Contamination Factor (CF) is also much L. Reuning et al. Marine Pollution Bulletin 217 (2025) 118070 6 higher compared to nearly all other shelf sediments in the western Mediterranean (Fig. 4). This even holds if the very highly contaminated sample (IG5) is not considered in the calculation of the average. 4.1. Coral habitats as plastic sinks The higher MPs concentration in the bay of Illa Grossa compared to other regions of the western Mediterranean could be partially related to the presence of the C. caespitosa coral colonies. Sediments in coral reef environments have previously been identified as microplastic sinks (Huang et al., 2021; Lin et al., 2024; Utami et al., 2021). Habitat-forming species such as seagrass, macroalgae and corals, are known to trap MPs into sediments (Feng et al., 2020; Mendrik et al., 2024; Sanchez-Vidal et al., 2021; Smit et al., 2021). These habitat forming species effectively reduce the flow velocity and turbulent kinetic energy on the seafloor through energy dissipation (Hendriks et al., 2010) and thus promote the settling and retention of MPs on the sediment (Smit et al., 2021; Yen et al., 2024). Seagrass is absent from Illa Grossa Bay (Templado and Calvo, 2002), but macroalgae and corals are abundant (Kersting and Linares, 2012; Kersting et al., 2014; Pons-Fita et al., 2019) and likely contribute to trapping of MPs. However, the effect of macroalgae is likely limited to their growth period in spring and summer (Pinedo et al., 2015), whereas corals are present all year around. Different trapping mechanisms of MPs in coral canopies have been identified (Mendrik et al., 2024): corals can act as an obstacle for fluid flow, leading to deposition of MPs on the up-current side of colonies. MPs can get trapped within the coral colonies themselves, or accumulate in the wake zone on the down-current side of corals. Even sparse coral canopies lead to strongly enhanced microplastic trapping compared to Table 2 MPs and microrubber concentration, size information and relative abundance of different polymers. Relative abundance of polymers (%) Site MPs particles/kg dw MP +rubber particles/kg dw MPs +rub. Mean size ( μ m) >250 μ m (%) PE PP PU Rubber PS PET PVDC IG1 41 41 172 20 50 10 10 0 30 0 0 IG2 49 49 116*20 20 0 20 0 20 20 20 IG3 549 579 152*15 51 0 0 5 40 4 0 IG4 522 555 109 3 29 59 0 6 6 0 0 IG5 4736 6345 83 0 0 3 7 25 0 65 0 Study area 1179 1514 123 8 28 11 4 12 19 25 1 * Mean was calculated without the two outliers in samples IG2 (2009 μ m) and IG3 (736 μ m). Fig. 3. Relative abundance of different polymer types in the study area for the entire size spectrum (A) and only MPs larger than 250 μ m (B). (C) shows the particle size distribution of all polymer types combined (violin plot) and for each individual polymer type (jitter plot). Not included in (C) are two outliers (PS, 736 μ m and PE, 2009 μ m). (D) Particle size distribution plot for all polymer types except PVDC, with the 50 % line indicating the median particle size. Table 3 Median particle sizes of different polymers (A) and statistical parameters for the equality of distributions (B and C). B: The p-values of the two-tailed MannWhitney test indicates whether the median values of two polymer types are equal. The medians can be regarded as statistically different, when the p-value is <0.05. C: The Kolmogorov-Smirnov test indicates whether two distributions are equal. The D-values are a measure for the maximum absolute difference between the two tested distributions. The difference can be regarded as statistically distinct, when the p-value is <0.05. (a) PE PP PU Rubber PS PET PVDC Median size ( μ m) 141 75 65 89 134 61 81 (b) Mann-Whitney pairwise test (p value) PE PP PU Rubber PS PET PE 0.0006 0.0002 0.0144 0.4348 3.69E-12 PP 0.0675 0.2639 0.0020 0.0024 PU 0.0109 8.37E-05 0.7149 Rubber 0.0367 3.39E-05 PS 3.95E-11 PET (c) Two-sample Kolmogorov–Smirnov test (D value) PE PP PU Rubber PS PET PE 0.40528 0.82143 0.32738 0.16892 0.68768 PP 0.47826 0.24457 0.44066 0.44075 PU 0.625 0.83784 0.23529 Rubber 0.29842 0.55882 PS 0.68362 PET L. Reuning et al. Marine Pollution Bulletin 217 (2025) 118070 7 Table 4 Contamination, Polymeric and Pollution Risk Indices and risk categories for individual samples and the entire study area. A range of values for H and PRI were calculate for samples containing PU. The average value and upper and lower bounds of these risk indices were calculated using a Monte Carlo Simulation of the likely hazard score distribution of PU. H and PRI for the study area are expressed as geometric mean. The risk categories are low (I), moderate (II), considerable (III), high (IV) and very high (V). For details on the association of risk indices with categories see Table S1. Site Contamination Factor (CF) Contamination Risk Category Polymeric Risk Index (H) Polymer Risk Category Pollution Risk Index (PRI) Pollution Risk Category Range Range Range Range low average high low average high low average high low average high IG1 1.0 II 544 725 897 III III III 544 725 897 III IV IV IG2 1.2 II 1106 1469 1812 IV IV IV 1322 1755 2165 V V V IG3 13.4 V 8 I 107 I IG4 12.7 V 4 I 51 I IG5 115.5 V 485 648 802 III III III 55,998 74,825 92,665 V V V Study area (mean) 28.8 V 99 117 133 II III III 739 878 997 IV IV IV Fig. 4. (A) Map of coral cover (%) (modified from Kersting and Linares (2012)) in Illa Grossa Bay with the category of the Pollutant Risk Index (PRI) for individual samples and the study area. The PRI in the figure was calculated assuming an average hazard score for PU. Details on the relationship between risk categories and pollution indices can be found in Table 4 and Table S1. (B) Map of the western Mediterranean showing the Contamination Factor (CF) risk category for subtidal shelf sediments. The CF was calculated from MP concentrations from this study (red box) and 1: Filgueiras et al., 2019; 2: Alomar et al., 2016 and 3: Fagiano et al., 2023. (For interpretation of the references to colour in this figure legend, the reader is referred to the web version of this article.) L. Reuning et al. Marine Pollution Bulletin 217 (2025) 118070 8 bare seafloors, but the effect increases further with coral canopy density (Mendrik et al., 2024). This is broadly consistent with the results from this study. Very high MP contamination in the study area is limited to samples (IG3–5) from coral colonies and with high coral cover (>10 % coral). The sample from the central area (IG1) that is nearly devoid of corals in contrast shows the lowest MPs concentration (Fig. 2; Tables 1 and 2.). Hydrodynamic studies suggest that the coral canopy traps smaller MPs more efficiently than larger MPs (Smit et al., 2021). This is reflected in the smaller grain size of MPs in coral colonies and areas with higher coral canopy density compared to the sample with <2 % coral cover (Fig. 2; Tables 1 and 2.). The concentration and particle size of MPs in the analyzed samples therefore appear to be consistent with hydrodynamic considerations, although the relatively small number of samples precludes a more rigorous analysis. The differences in sampling years (Table 1) and thus the age of the sediment cannot explain the differences in the MPs concentration between the sites. Samples IG4 and 5 were taken within coral colonies. Based on the growth rates of the corals, these sediments have likely been deposited since the mid-1980s (see section “Method”) and in average are therefore likely older compared to the sediments from the seafloor (IG1 to 3). Due to the increase in MP contamination over the last decades, one could assume that older sediments might contain lower MP concentrations. This does not seem to be the case. In fact, the “old” sediments within the coral colonies have the highest MP concentrations (Table 2), suggesting that the trapping potential of corals seems to cancel out the effect of sediment age. In addition to the hydrodynamic effects of the coral canopy, it is assumed that the specific properties of coral reef sediments favor the trapping of MPs (Utami et al., 2021): the biogenic grains in reef environments are often porous and irregularly shaped which can favor entanglement with MPs, thereby trapping them in the sediment. MP can also be protected against resuspension and entrainment by the often comparatively large biogenic grains in these environments. This could also be the case in our study with MPs of considerably smaller size compared to the mean sediment grain sizes (Tables 1 and 2). 4.2. Oceanographic and topographic influences In our study area the contamination by MPs is relatively high even compared to most other reef systems (Huang et al., 2021; Patterson et al., 2022; Pradhap et al., 2023; Utami et al., 2021). The high MPs measured can therefore not solely be explained by the hydrodynamic effects of the coral canopy and the associated sediments. The high MP concentrations (Table 2) are more similar to the ones found in reef systems close to highly populated areas (Patti et al., 2020; Zhou et al., 2023) and tourism hotspots (Lim et al., 2022; Lin et al., 2024). However, pollution from local sources seems unlikely for our study area, which is located in a remote Marine Reserve without direct land-based sources of plastic contamination. It therefore seems likely that long distance transport plays a role for the accumulation of MPs in Illa Grossa Bay, as it was proposed for similarly high MP concentrations in sediments of the southern Xisha Islands in the South China Sea (Lin et al., 2024). This is also in line with previous observations (Alomar et al., 2016; Fagiano et al., 2022) and modelling results (Hatzonikolakis et al., 2022) from the western Mediterranean, which show that the concentrations of MPs in Marine Protected Areas (MPAs), where local sources are scarce or nonexistent, are similar to or even higher than in touristic hotspots (Alomar et al., 2016; Fagiano et al., 2023). Modelling studies suggests that the Northern Current (Ourmieres et al., 2023; Soto-Navarro et al., 2020), the major current of the northwestern Mediterranean zone (Fig. 1), is the main vector for MPs to the Balearic Sea. The Northern Current originates in the Ligurian Sea and flows southwest-wards parallel to the continental slope into the study area. Some of the most important point sources for plastic marine litter in the Mediterranean, such as the city of Barcelona, the Rhone river (Liubartseva et al., 2018), and the Ebro river (Simon-S´ anchez et al., 2019) are situated along its path. It therefore can transport floating MPs from the densely populated coastal or inland areas of Italy, France and Spain into the study area (Ourmieres et al., 2023). In addition, particle tracking models show that, at least seasonally, particles from even more distant sources such as the Algerian coast can reach the Columbretes Islands through northward directed currents (Kersting et al., 2020). This is consistent with the observation that floating plastics in the area of Columbretes are characterized by relatively small items composed mainly of fragments (Fig. S2), which was interpreted to indicate more aged plastic from distant sources (Fagiano et al., 2022). In addition, the C-shaped island of Illa Grossa is open towards the northwest (Fig. 1), the main direction of storms and waves (Kersting and Linares, 2012), and therefore could act as a trap for marine litter. Floating plastics, brought to the region from distant sources, could therefore been trapped in the inner bay of the island and eventually retained in its sediment. Similar shape dependent trapping effects have been previously observed for concave beaches that trap marine debris much more efficiently than other shaped shorelines (Brennan et al., 2018). The combination of this topographic feature with the trapping potential of the coral canopy likely is the reason why the CF at Illa Grossa is higher compared to almost all other sites in the western Mediterranean (Fig. 4), despite the fact that the island is situated in a Marine Protected Area. This confirms previous studies that have shown that Marine Protected Areas can be hotspots of plastic pollution (Fagiano et al., 2023). Local protection alone is therefore not effective in reducing the pollution load. 4.3. Polymer composition The polymer PE, the plastic type with the highest global production rate (Plastics Europe, 2022), is also the most common MP type found in the study area (28 %). The abundance of PU also roughly corresponds to its importance in global production (~ 5 %). The low abundance of the polymer type PVDC (1 %), which only occurs in one sample, is consistent with its relatively low productions rates compared to other plastic types. To our knowledge, it was previously reported in only one other study on MPs in reef systems, where it was identified as a copolymer (Ding et al., 2019). In contrast, the polymer types PET (25 %) and PS (19 %) are highly overrepresented in our study area compared to their share of global production, which is between 5 and 7 % each (Plastics Europe, 2022). A study on floating plastic litter (>355 μ m), from the coastal regions of several islands in the NW Mediterranean shows that the polymer composition off the Columbretes Islands, including Illa Grossa, differs from other islands in the region (Fagiano et al., 2022). The coastal waters of the Columbretes Islands show higher abundances of PC and PS and the lowest abundance of PP compared to the other islands in the area (Fagiano et al., 2022) and other areas of the central western Mediterranean (Suaria et al., 2016). This is consistent with relatively low PP (11 %) and high PS (19 %) concentrations in the sediments of Illa Grossa Bay (Table 2). Fagiano et al. (2022) suggested that these high levels of PS in the region could be related to its use for containers and packaging material in the fishing industry. PET fragments by far predominate over PET fibers in our study area, indicating that PET-MPs likely originate from the degradation of larger PET debris such as bottles (Ioakeimidis et al., 2016) rather than from textiles (Oliveira et al., 2023). Little is known about the abundance of microrubber in the Mediterranean Sea. To the best of our knowledge, microrubber abundance has not been quantified for sub-tidal sediments in the Mediterranean to date. The analysis shows a relatively high contribution (12 %) of rubber to the total MP concentration, with very high concentrations (25 %) in sediments from within one of the coral colonies. Previous studies on microplastic pollution in the Mediterranean Sea indicate that rubber contributes only very small amount (typically <1 %) to the total floating plastic load (Kedzierski et al., 2022; Suaria et al., 2016). L. Reuning et al. Marine Pollution Bulletin 217 (2025) 118070 9