Tracing microplastics in aquatic environments based on sediment analogies

Microplastics (MP) data collection from the aquatic environment is a challenging endeavour that sets apparent limitations to regional and global MP quantification. Expensive data collection causes small sample sizes and oftentimes existing data sets are compared without accounting for natural variability due to hydrodynamic processes governing the distribution of particles. In Warnow estuarine sediments (Germany) we found significant correlations between high-density polymer size fractions (≥500 µm) and sediment grain size. Among potential predictor variables (source and environmental terms) sediment grain size was the critical proxy for MP abundance. The MP sediment relationship can be explained by the force necessary to start particle transport: at the same level of fluid motion, transported sediment grains and MP particles are offset in size by one to two orders of magnitude. Determining grain-size corrected MP abundances by fractionated granulometric normalisation is recommended as a basis for future MP projections and identification of sinks and sources.

contamination, a conservative interpretation requires an entire exclusion of PET fibres from the final results. This includes the data from the sediment trap samples as the surrounding conditions (operators, laboratory facility) did not change and air-born sources are most probable. Although the purification process of the sediment trap samples included less intensive chemical treatment, PET appeared frequently across the samples sets, with more than half (Akrona basin, n PET f ibre = 17) or all (Gotland basin, n PET f ibre = 9) of the polymers identified being assigned to this polymer type.
A second polymer type was eliminated from the results retrospectively; PTFE. This is because the conclusion must be drawn that these MP originate from the treatment within the MPSS, as several parts, such as within the separation chamber and the outlet valve, are made from this polymer. We observed that, without exception, all PTFE particles found have the same appearance (transparent-whitish, within the same size class, morphologically similar). The low but non-correlated and yet widespread occurrence throughout the sediment samples (including Arkona basin sediment sample) and absence in the sediment trap samples which did not undergo the MPSS treatment strengthens an MPSS-associated PTFE contamination. The pH adjustment of the sodium polytungstate solution was done in a Schott flask with a PTFE screw cap which was subsequently refilled into the MPSS. Abrasion of PTFE particles is possible. Although, the sodium polytungstate solution has a lower density than average PTFE (2.1 − 2.3 g cm −3 ), surface tension effects could lead to the flotation, and consequent retrieval, of these small particles 67 .
Contamination prevention and controls -Methods. To comply with requirements of quality assurance, both the application of control samples and best practice contamination prevention measures throughout both lab and field work were ensured. A cellulosic laboratory coat and nitrile gloves were worn at all times. Prior to use, all instruments and vials were sanitised with tap water and rinsed with deionised and micro-filtrated water (MilliQ, 0.2 µm). Where possible prefiltered 70% ethanol was used for decreasing the water drops surface tension in which plastic particles or fibers potentially could remain. Sieves were additionally ultra sonificated (up to three times for 10 minutes). Frequently used metal and glassware were flamed. Filtration measures were undertaken within a laminar flow chamber in order to avoid air-born contamination. During microscopy, petri dishes were kept closed and the Bogorov chamber was covered with aluminum foil while awaiting further analysis.
The sodium polytungstate solution was recycled each time by filtration through a 15 µm gaze, ensuring that the density separation solution was free of contamination within the detection limit. Given the used separation density, flouropolymeric compounds such as polytetrafluorethylen (PTFE, 2.2 gcm −3 ) is not accounted for as the density would need to be increased significantly. The application of PTFEs as coatings of containers, hoses and sealings, bears the potential for unknown systematic contamination, as was the case in the present study. There are multiple reasons for excluding PTFE from particle-based study designs. Production volumes are proportionally low and PTFE MP probably enters the environment preferentially in the lower micrometer or nanometer size range since it has become highly relevant as nano-coatings 68 . The omnipresence of perfluorochemicals (PFCs) in nature along with the ecotoxicological risk have been studied extensively. Specific mass-based detection methods of PFCs already exist 69 .
In general the use of plastic items was avoided where possible. Restrictions are to be noted during sediment trap sampling, as the aperture is partly made out of plastic or coated with paint. The contamination risk for particles > 500 µm is, however, calculable as MP collection bottles closed underwater and were not opened until analysis. It is expected that any system-caused MP contamination would be identifiable by a correspondence between the characteristics of any used and found particles or a repetitive occurrence of the same kind of particles throughout the sample set.
Control steps accounting for contamination during the processing procedures were taken according to the same procedures as the sediment samples or sediment trap samples were processed. In parallel to the sediment samples (n Warnow = 14; n Arkona+Gotland = 2), a total of three controls, where two were covering the MPSS treatment and another one the exposure time during the visual identification process in the Bogorov chamber, were acquired. The degree of contamination during the sediment trap sample processing (n Arkona+Gotland = 2 with 18 -20 subsamples each) was evaluated by means of three controls in total. One separate collection bottle in the sediment trap set-up accounted for possible contamination during this part of the field sampling process. The bottle was filled with MilliQ water while sampling time and bottle lid opening and closing movements resembled the actual sampling. In this way, potential abrasion from the bottles entirely comprised of PE would be detected.

Text S3. Methodological remark on density separation threshold
The crux of MP field studies is the missing standardisation of MP isolation techniques. Density separation is most commonly used, however, setups greatly vary and the critical density is often below that of many HD polymers which prevents a detailed distribution analysis of the full spectrum of synthetic polymers 70 . The authors consider a density separation threshold of 1.8 g cm −3 as scientifically reasonable, as it includes the full spectrum of commonly excepted plastic polymers. It also accounts for the majority of dried paint resins 29 , which were calculated to posses an average specific gravity of approximately 1.6 g cm −3 . The chemical composition of paint resins: polymer base resin, pigments and additives, resembles that of commodity plastics with a larger proportion of heavy pigments. Due to the higher loads of heavy metals, the inclusion of paint resins into MP studies is highly relevant, particularly from an ecotoxicological perspective 44 .