Environmental Pollution 262 (2020) 114298

Contents lists available at ScienceDirect

Environmental Pollution
journal homepage: www.elsevier.com/locate/envpol

Microplastic identification and quantification from organic rich

sediments: A validated laboratory protocol*
P. Vermeiren a, b, *, C. Mun
~ oz a, b, K. Ikejima a
a
Laboratory for Coastal Ecology and Conservation, Faculty of Agriculture and Marine Science, Kochi University, Japan
b
Dept. Environmental Science, Faculty of Science, Radboud University, The Netherlands

a r t i c l e i n f o a b s t r a c t

Article history: Plastic pollution presents a global environmental concern with potentially widespread ecological, socio-
Received 16 September 2019 economic and health implications. Methodological advances in microplastic extraction, quantification
Received in revised form and identification from sediments have been made. However, integrating these fragmentary advances
11 February 2020
into a holistic, cost-effective protocol and applying it to organic rich sediments with fine grain size re-
Accepted 28 February 2020
Available online 29 February 2020
mains a challenge. Nonetheless, many hot spots of microplastic contamination such as harbour and
estuarine sediments are characterised by such sediments. We conducted a series of experiments to
integrate methodological advances, and clarify their applicability to organic rich sediments with fine
Keywords:
Fourier transform infrared spectroscopy
grain size. The resulting protocol consisted of three stages. First, pre-treatment with Fenton’s reagent was
Nile red dye found to be efficient in reducing organic matter content, compatible with later Fourier Transform-
Fluorescence microscopy Infrared Spectroscopy (FT-IR) for polymer identification, although it did affect the size of polyethylene
Debris (PE) and polyethylene terephthalate (PET). Secondly, a novel density separation column with a top
Density separation overflow (the OC-T) obtained recovery rates above 90% for microplastics present in a ZnCL2 solution.
Automated particle counting Finally, automated epifluorescence microscopic image analysis of Nile Red stained filters with selected
validation of polymer identities using FT-IR revealed 91.7% of stained particles to be plastics. A case study
on estuarine sediments demonstrated a high extraction efficiency with quantification possible down to
125 mm and detection possible down to 62.5 mm. This makes this protocol suitable for large scale
monitoring of microplastics in sediments of estuarine origin provided polymer specific recovery rates,
background contamination and uncertainty in Nile Red identification is accounted for. Subject to further
validation, the protocol could also offer a solution to similar organic rich sediments with fine grain size,
such as some soils and sludge, to improve our ability to conduct cost-effective, large scale monitoring of
microplastic contamination.
© 2020 The Authors. Published by Elsevier Ltd. This is an open access article under the CC BY license
(http://creativecommons.org/licenses/by/4.0/).

1. Introduction environment directly (termed: primary microplastics) because they

are used, for example, as feed stocks in plastic production or as
Plastic pollution presents a global environmental concern with ingredients in cosmetics and air blasting products (Gregory, 1996;
potentially widespread ecological, socio-economic and health im- Cole et al., 2011). Due to their small size, microplastics become
plications (UNEP, 2014). Despite their many benefits (Andrady and available to a wide range of organisms throughout the food web
Neal, 2009), the shear amount of plastics produced and their poor (Wright et al., 2013). Moreover, their light weight and long
waste management results in large amounts of plastics entering the persistence has led to microplastics being distributed from pole to
environment. Once there, plastics slowly degrade fragmenting into pole across marine, terrestrial and freshwater systems (Moore,
smaller and smaller pieces (termed: secondary microplastics, 2008; Barnes et al., 2009; Wagner et al., 2014).
Andrady, 2015). Additionally, microplastics also enter the Microplastic concentrations are positively correlated with pop-
ulation density (Browne et al., 2010) and with various industrial
activities such as shipping, harbour and port activities, fisheries,
plastic production sites and sewage treatment facilities (Wagner
*
This paper has been recommended for acceptance by Maria Cristina Fossi. et al., 2014). This makes harbours and estuaries particularly sus-
* Corresponding author.
ceptible (Vermeiren et al., 2016) as locations where urban and
E-mail address: [email protected] (P. Vermeiren).

https://doi.org/10.1016/j.envpol.2020.114298
0269-7491/© 2020 The Authors. Published by Elsevier Ltd. This is an open access article under the CC BY license (http://creativecommons.org/licenses/by/4.0/).
2 P. Vermeiren et al. / Environmental Pollution 262 (2020) 114298

industrial development often takes place (Seto, 2011). A distinct can be applied within a typical ecology or environmental science
characteristic of harbour and estuarine sediments is their often fine laboratory. Therefore, we developed and validated a holistic labo-
grain size and high organic matter content. Similar high organic ratory protocol to extract, quantify and identify microplastics from
contents can also be found in matrices such as some soils or sludge. organic rich sediments with fine grain size, starting with samples
The overall process of microplastic extraction, identification and from the field and ending with data on microplastic identity, size
quantification from sediments can be divided in three stages: 1) distribution and concentrations that can be used in statistical
pre-processing, 2) separation of microplastics from sediments and analysis. We integrated fragmentary methodological advances that
3) the identification and quantification of concentrated micro- have previously been identified as showing potential for such
plastics (Fig. 1). A large effort has been directed towards method- sediments. We then conducted a series of experiments to clarify the
ological improvements within each of these stages such as applicability and compatibility of these advances using estuarine
comparing different solutions for the density separation of micro- sediments as a case study.
plastics from sediments (Miller et al., 2017) or improving auto-
mated polymer identification (Primpke et al., 2017; Bergmann et al., 2. Methods
2019). However, there are several challenges in integrating frag-
mentary methodological advances into a cost-effective, holistic Our protocol (Fig. 1) was inspired by the U.S. National Oceanic
protocol applicable to organic rich sediments with fine grain size. and Atmospheric Administration’s (NOAA) manual for the analysis
Firstly, carry over effects are rarely considered in method of microplastics in beach samples (Masura et al., 2015). By contrast
development studies. For instance, the use of Nile Red fluorescent to the NOAA protocol, however, we did not dry sediment samples as
staining for rapid quantification of microplastics was suggested sediments with high organic matter loads turn into a condensed
(Shim et al., 2016; Maes et al., 2017). Yet, its application as part of a agglomerate when drying. This agglomerate can be broken up
full protocol in which microplastics are concentrated on filters, and physically or chemically (Nor and Obbard, 2014; Felsing et al.,
in which also polymer identification is conducted in later steps, is 2018). Yet, this can impact the integrity of microplastics con-
not fully clarified. Secondly, when recovery rates of microplastics tained within the sample. In the current study, samples were taken
are established for individual methodological advances, only a using glass petri dishes with a surface area of 50 cm2 and a volume
snapshot of the whole protocol is being investigated. Hence, re- of 60 mL, allowing quantities of microplastics to be expressed per
covery rates for a holistic protocol are lacking in many cases. Finally, bulk wet sediment, per volume or per surface area. Additionally,
few methods have been specifically tested and tailored towards due to the high organic matter content, we decided to reduce
organic rich sediments with fine grain size. The high organic matter organic matter at the start of our protocol rather than in a later
content can interfere with density separation of microplastics and phase similar to Liu et al. (2019). Organic matter floats in a density
requires cleaning up before identification of microplastics can separation column just like microplastics (step 2, Fig. 1) and sub-
occur. Likewise, fine sediments only settle slowly in density sepa- sequently clogs vacuum filters leading to a slow filtration, often
ration columns. requiring many filters. This not only increases the cost and time for
We aimed to strengthen capacity for action in the context of analysis, but organic matter retained on the filters can also obscure
microplastic pollution, specifically in harbour and estuarine sedi- microplastics from identification and quantification. In the next
ments, by providing a cost-effective and repeatable protocol for sections, we develop individual steps of the protocol in detail using
large scale environmental monitoring of microplastic pollution that a series of experiments. The choice of the different methods tested

Fig. 1. Overview of the protocol for microplastic identification and quantification in estuarine sediments with individual steps in numbered boxes (1e7). Grey boxes with letters A, B
and C indicate the three main stages typically used in laboratory protocols for microplastic identification and quantification.
 P. Vermeiren et al. / Environmental Pollution 262 (2020) 114298 3

in each experiment was based on the current literature and cost amount of organic matter remaining in the freeze-dried samples
effectiveness of these methods. This was a pragmatic strategy to under the two treatments was then established by taking the dif-
obtain a holistic protocol, rather than conducting a comprehensive ference in weight before and after combustion of organic matter in
comparison of all potential methods for a specific step in the pro- a muffle furnace at 750
C for 2.5 h.
tocol (such reviews are available e.g. Ivleva et al., 2017; Miller et al., Effects of the two treatments on the surface area and FT-IR
2017). spectra of microplastics were tested. Five microplastics of each of
Microplastics used in the experiments represent the most the seven plastic polymers tested (i.e. 35 replicate particles) were
commonly encountered plastic polymers in the marine environ- added to either 3 mL of 30% H2O2 or to 1.5 mL of 30% H2O2 com-
ment (Andrady, 2011): polyethylene (PE), polypropylene (PP), bined with 1.5 mL of 0.05 M Fe(II) solution (creating Fenton’s re-
polystyrene (PS), expanded polystyrene (EPS), polyethylene tere- agent) and left for 24 h. The mean 2D surface area of the
phthalate (PET), nylon (PA) and polyvinyl chloride (PVC). The microplastics tested was 0.87 mm2 ± 0.20 mm2 standard error. All
microplastics were cut from larger items such as PE and PP plant five microplastics per polymer and treatment were photographed
pots, EPS and PS food containers, a PET bottle, a PA mesh and a PVC before and after treatment under a binocular microscope at 10x
traffic cone. Characteristics of the microplastics used in the exper- magnification, and their surface area measured using ImageJ
iments are presented in the sections of the respective experiment. (Schneider et al., 2012). FT-IR spectra before and after treatment
Sediment used in the experiments was collected in the Kokubu were taken for each polymer and treatment with attenuated total
estuary within three habitats (Table 1) on 28 and 29 May 2018 reflectance FT-IR (ATR FT-IR) using a JASCO FT/IR-6100 coupled
during spring low tide at a location (33
3401100 North; 133
3401500 with an ATR Pro One unit taking 40 measurements in the wave-
East) in the Northeast of the city of Kochi, Japan. Being in a mixed number range of 550 cm1 e 4000 cm1 in 4 cm1 spectral reso-
urban-industrial area, we suspected that the location would be lution. The similarity in spectra was compared using Pearson’s
subject to microplastic pollution from residential and industrial correlation coefficient.
sources, as was also suggested by the visible presence of large
(mainly plastic) debris. Sediment was characterised for water
2.2. Experiment 2: density separation
content, organic matter content and grain size distribution (Table 1,
Supp. 1).
Following organic matter reduction, we separated microplastics
from the sediment using a density separation column filled with
2.1. Experiment 1: testing methods for organic matter reduction ZnCl2 of density 1.5 g cm3. ZnCl2 is a frequently used and afford-
and their effects on plastics able solution whose density allows separation of the common
microplastic polymers in the marine environment (Coppock et al.,
The oxidant peroxide (H2O2) presents an easy, cost and time 2017; Ivleva et al., 2017; Miller et al., 2017). After density separa-
effective method frequently used in microplastic studies (Miller tion, the top layer of solution containing the microplastics is
et al., 2017). The addition of ferrous iron, Fe(II), to H2O2, known retained for further analysis.
as Fenton’s reagent, was suggested as an improvement to isolate We tested different density separation columns to assess their
microplastics from organic rich wastewater samples (Tagg et al., ability to recover microplastics from spiked solutions onto vacuum
2017) and could potentially also be used for organic rich sedi- filters. Four types were tested (Table 2): (1) a sediment microplastic
ments. We tested the two different treatments, H2O2 and Fenton’s isolation unit (SMI unit, Coppock et al., 2017), containing a ball
reagent, for their ability to reduce organic matter from sediments, valve halfway along the column allowing the upper solution con-
and for the effects they have on the surface area and the Fourier- taining microplastics to be separated from the bottom solution
transform infrared spectroscopic (FT-IR) spectrum of micro- containing the settled sediments; (2) a simple decanting column;
plastics from the seven most common polymer types in the marine (3) a newly designed overflow column with top inflow (OC-T),
environment: PE, PP, PS, EPS, PET, PA and PVC. The aim of this which is a column with an overflow top piece allowing ZnCl2 to be
experiment was to choose the most appropriate method as an in- added by pouring it with a squeeze bottle against the top walls of
termediate step in the protocol development. A complete valida- the column so that the solution starts to overflow; and (4) an
tion of the full protocol, including real field-based samples, is overflow column with mid-level inflow (OC-M), which is a modi-
presented in experiment 3. A full assessment of recovery rates is fication of the OC-T with the addition of a glass tube allowing the
presented in experiment 4. ZnCl2 solution to be added midway in the column. For the SMI unit,
We used sediment from three estuarine habitats (Table 1) to test we used the same dimensions as proposed by Coppock et al. (2017),
the effectiveness of the two treatments. Wet sediment from each who developed this unit. For the other columns, we used slightly
habitat was divided into two samples of 60 g. Then, 20 mL of a 30% larger dimensions (Table 2) to accommodate larger sample
H2O2 solution was added to each sample. Additionally, 20 mL of a volumes.
0.05 M Fe(II) solution was added to one of the samples. The Fe(II) Ten microplastics of each of PP (mean surface area:
solution contained 2.5 g of FeSO4e7H2O, 165 mL water and 1 mL of 0.31 mm2 ± 0.17 mm2 standard deviation), PE (0.43 ± 0.19 mm2),
concentrated sulphuric acid (Masura et al., 2015). After reacting for PET (0.33 ± 0.20 mm2) and PVC (0.30 ± 0.11 mm2) were added to
24 h at room temperature, samples were freeze dried at 80
C each column type. PP and PE represent two of the most common
under vacuum to stop any further reaction at that point. The plastic polymers in the marine environment, with densities lighter

Table 1
Characteristics of the estuarine sediment used in the current study. SD: standard deviation.

Habitat type Water content (%) Organic matter content (%) Grain size (mm)

Mean ± SD Mean ± SD 10th percentile Mean ± SD Mode 90th percentile

Low intertidal mudflat 58.8 ± 23.0 17.3 ± 1.8 2.1 ± 1.6 16.1 ± 2.2 17.2 ± 1.5 128.1 ± 34.3
High intertidal mudflat 33.8 ± 16.2 5.2 ± 2.0 16.3 ± 9.2 137.4 ± 9.8 163.0 ± 3.4 299.3 ± 5.3
Saltmarsh (Phragmites australis) 32.8 ± 20.0 7.4 ± 5.0 7.3 ± 1.0 118.7 ± 89.7 214.0 ± 116.8 425.5 ± 221.5
4 P. Vermeiren et al. / Environmental Pollution 262 (2020) 114298

Table 2
Characteristics (height, inner diameter and material) of the density separation columns tested. SMI: sediment microplastic isolation unit (Coppock et al., 2017), OC-T: overflow
column with top inflow, OC-M: overflow column with mid-level inflow.

than seawater (Andrady, 2015). PET and PVC are common plastics in errors and removes bias in visual identification of microplastics
the environment with densities heavier than seawater (Andrady, (Shim et al., 2016; Maes et al., 2017; Tamminga et al., 2017). Nile red
2015). Three replicate recovery experiments were conducted per staining has been used in the quantification of microplastics in
column and polymer type. Therefore, the columns were filled with samples of marine water (Tamminga et al., 2017, 2018), beach
ZnCl2 until 2 cm from the top. The ball valve of the SMI unit was sediment (Hengstmann et al., 2018), and tissues of marine species
closed and opened five times. The solution in all column designs (Fischer, 2019). This method, however, does not allow identification
was then stirred for 5 min followed by 5 min rest and three short of polymer identities although a rough categorisation of polymer
stirring bursts (following Coppock et al., 2017), after which the classes has been proposed (Maes et al., 2017). Alternatively, micro
columns were left for 24 h. The SMI unit was then closed and the Fourier-transform infrared (mFT-IR) spectroscopy is a popular
top solution collected into a beaker for transfer into a vacuum filter technology to verify the polymer identity of microplastics (Lo € der
(¼ 1st wash). The top part of the SMI unit was then rinsed with et al., 2015). Methods for imaging of whole filters using focal
ZnCl2 from a squeeze bottle until no microplastics were visible plane array (FPA)-based mFT-IR systems have been developed,
anymore and the collected solution also filtered (¼ 2nd wash). For allowing identification and quantification (Harrison et al., 2012;
the other column designs, 200 mL of ZnCl2 was either decanted or Lo€der et al., 2015; Tagg et al., 2015; Bergmann et al., 2019; Liu et al.,
overflown into a beaker and then transferred to a vacuum filter (¼ 2019). Yet, such imaging techniques, albeit rapidly advancing, still
1st wash). After 10 min rest, this step was repeated a second time require long processing times. For example, a filter area of
(¼ 2nd wash). The number of microplastics from each polymer 10.54 mm  10.54 mm required 10.75 h for analysis (Lo € der et al.,
recovered on the first and second wash filters was counted under a 2015). More recently, Bergmann et al. (2019) analysed an area of
binocular microscope at 40x magnification. This recovery experi- 14.1 mm by 14.1 mm on a 25 mm vacuum filter in 4.5 h. While only
ment was conducted to assist in the selection of the most appro- small sample volumes could be analysed (e.g. Liu et al., 2019 added
priate density separation column. A recovery of 90% across all four subsamples in 100 ml increments until saturation of filters to be
polymer types was set as a required target for a density separation imaged), this shows the future potential of such technologies.
column to be considered suitable. A detailed recovery experiment Nonetheless, at the current time, the application of FPA-based
of the full protocol, including the most suitable density separation spectroscopic imaging techniques for microplastic quantification
column, is presented in experiment 4. within large scale monitoring programs remains expensive and
time consuming.
Taking advantage of the benefits of both methods, we test a
2.3. Experiment 3: microplastic quantification and identification
combination of quantification of microplastics using Nile Red,
case study with estuarine sediments
combined with identification of selected particles using mFT-IR as a
solution for cost-effective, large scale monitoring of microplastics
Following density separation and overflow, microplastics are
in sediment samples. In a case study to demonstrate our approach,
concentrated on vacuum filters (47 mm glass fibre filters with
we used sediment from two estuarine habitats, high intertidal
0.7 mm pore size, further referred to as GFF filters) on which they
mudflat and salt marsh (Table 1). Three replicate samples, a mini-
are subsequently counted and identified. Nile red dye is sol-
mum of 5 m apart, were taken in each habitat with a metal corer of
vatochromic based on the polarity of its environment, and has been
10 cm diameter, and subsamples with a volume of 60 mL were
demonstrated to fluoresce when sorbed to plastics (Shim et al.,
collected at depths of 0e1 cm, 5e6 cm, 10e11 cm and 15e16 cm.
2016; Maes et al., 2017). Nile Red staining therefore presents a
Sediment wet weight of each 60 mL subsample was recorded
cheap and relatively rapid quantification method that reduces
 P. Vermeiren et al. / Environmental Pollution 262 (2020) 114298 5

and microplastics separated from the sediment following the pro- applied which selected particles in the following ranges of Hue
tocol developed in the current study (Fig. 1). First, to reduce organic [0e43], Saturation [0e255], and brightness [160e255] on a scale of
matter content, samples were treated with Fenton’s reagent by 0e255. Selected particles were automatically counted and the
adding 20 mL 30% H2O2 followed by 20 mL 0.05 M Fe(II) solution length along their longest axis measured using the “Analyse Par-
and leaving the mixture at room temperature for 24 h. Secondly, ticles” function of ImageJ. Microplastics were then classified based
density separation and filtration were conducted using an overflow on the particles’ longest axis lengths into large microplastics
density separation column with top inflow (OC-T, Table 2) and a [<5 mm and >¼0.5 mm], based on the definition used by Hidalgo-
1.5 g cm3 zinc chloride (ZnCl2) solution. The mixture was stirred in Ruz et al. (2012), and small microplastics in several classes based on
a sequence of 5 min stirring, 5 min rest and three short, 2e3 s, the Wentworth (1922) classification of sediment grain sizes: class A
stirring bursts (following Coppock et al., 2017) after which the so- [<0.5 mm and >¼0.250 mm], class B [<0.250 mm and
lution was left to settle overnight. A stirring magnet in a glass >¼0.125 mm], class C [<0.125 mm and >¼0.0625 mm] and class D
casing was used and a removable flat glass disc (60 mm diameter) [<0.0625 mm and >¼0.05 mm]. The lower size limit in class D was
was placed at the bottom inside the OC-T column to avoid abrasion based on the pixel size of the image, and represents 4.5 pixels.
of the plastic base by the stirring magnet, which could lead to the Among the particles that were suggested by the “Analyse Par-
creation of plastics alien to the sample. Sometimes it was difficult to ticles” function of ImageJ as potential microplastics, a subset of 120
get the magnet spinning. We then used a glass stick to manually stir particles was subsequently randomly selected for further analyses
and homogenise the OCT-T column. of polymer identity using mFT-IR spectroscopy. Additionally, 50
Once settled, ZnCl2 was added until the solution reached the top particles that lightly fluoresced, but did not meet the Nile Red
of the OC-T column, and subsequently overflowed into a glass threshold values to be considered as a potential plastic using the
beaker. ZnCl2 was carefully directed with a squeeze bottle against “Analyse Particles” function, were also subsequently randomly
the walls of the OC-T column to minimize disturbance during the selected and analysed. Therefore, the selected particles were care-
overflow process. Once 200 mL of overflow was collected, the fully transferred from the filters to a 3  3  0.5 mm mini-KBr plate
column was let to rest for at least 10 min and then another 200 mL (JASCO, model 2016-1). The sample was then pressed between two
was overflown. After this second overflow, a little extra solution mini-KBr plates, allowing measurement using transmission mFT-IR
was carefully decanted into the beaker until the level of ZnCl2 in the on a JASCO FT/IR-6600 interfaced with an IRT-7200 microscope unit
OC-T column was about 1 cm below the top. The OC-T column was taking 100 readings in the range of 400e6000 cm1 with a reso-
then completely homogenized again using the stirring sequence lution of 4.0 cm1. Obtained spectra were then compared using the
described above and left to settle overnight. The next day, another Spectra Manager software against reference spectra in the KnowI-
two times 200 mL was overflown to capture any microplastics that tAll databases of BioRad Stadtler (1: IR-Demo database incl. 4000
might not have been retrieved the first time. The two overflow polymer spectra, 2: IR-polymers, Hummel, database incl. 1900
fractions were then filtered over a vacuum column containing first spectra of polymers, copolymers and additives, and 3: IR-standards
a 500 mm sieve and then a 47 mm GFF filter. The 500 mm sieve was organics and polymeric compounds subset database incl. 10,000
put on top of the column to capture larger pieces such as sticks and spectra of common organic and polymeric compounds), the JASCO
leaves which had not been completely disintegrated with the standard database supplied with the FT/IR 6600 incl. 400 spectra of
Fenton’s reagent as well as larger shells and plastics, which might common organic and inorganic compounds and polymers, and a
otherwise obscure microplastics on the GFF filters. The sieve was reference database collected by the Kochi Prefectural Industrial
visually checked for potential plastics among such larger items Technology Centre containing a variety of materials (many found in
under a binocular microscope at 10x magnification using the the local environment and industry). These databases thus include
n (2007). Any suspected microplastics were picked
criteria of Nore spectra enabling identification of plastic polymers as well as other
with tweezers and placed on a GFF filter. This filter with larger materials present in estuaries such as plant and animal tissues (incl.
microplastics was then processed just as any other filter in the protein, cellulose and chitin).
following steps. All filters were dried at 50
C for 24 h.
Microplastic abundance on the dried filters was established by 2.4. Experiment 4: recovery rates with full protocol
fluorescent dying. Therefore, drops of a Nile Red solution prepared
with 1 mg Nile Red dye in 1 mL 99.5% acetone and diluted with This experiment aimed to establish the recovery rates of the full
100 mL distilled water (based on Maes et al., 2017) were spread protocol for microplastic particles of polymers commonly
until they covered the filter. Stained filters were then photographed encountered in the environment, across their size range. Therefore,
under a Optocode LED 530MS-EPI green light with peak emission in sediments from the high intertidal mudflat (Table 1) were com-
wavelengths between 490 and 570 nm (which includes the 553 nm busted at 750
C for 2.5 h to remove any microplastics present
absorption peak of Nile Red dye) using a binocular microscope within it. A volume of 60 mL of cleaned sediment was then spiked
fitted with an orange filter which cuts out light below 600 nm with different sized microplastics of one of the following polymers
wavelength, allowing visualisation of the fluorescence emitted by (Table 3): coloured microplastics of blue PE, red PP and red PVC, or
Nile Red dye at 637 nm. A Sony NEX-5K camera body (14.2 mega- transparent Nile Red stained microplastics of PS, PA and PET,
pixels) was mounted on the microscope and pictures were taken obtaining six different spiked sediment samples. The spiked sedi-
with manual focus and automatic light metering across the whole ment samples were then put through the full protocol (Fig. 1). The
image. A magnification of 6.7x was used allowing the full filter to be obtained GFF filters were then checked under white (for coloured
photographed in one image. plastics) or green (for Nile Red stained plastics) light and the
Filter images were subsequently analysed with ImageJ number of recovered microplastics >¼0.5 mm and <0.5 mm
(Schneider et al., 2012) to establish abundance and sizes. A scale bar counted for each polymer type corresponding to the division in
was set using a 20 mm section of millimetre paper photographed large and small microplastics proposed by Hidalgo-Ruz et al. (2012).
under the same setup as the filters to allow measurements to be
made. Additionally, a colour threshold (developed during pilot 2.5. Quality control
work where Nile Red stained filters with known composition of
microplastics and other materials were analysed with ImageJ) was For the protocol to be applied under real life conditions, it needs
6 P. Vermeiren et al. / Environmental Pollution 262 (2020) 114298

Table 3
Number of particles (N) and their size distribution (length of longest axis) for the different polymers tested in the recovery experiment. Mean and median sizes are presented
for the two size classes: large (L-MP >¼0.5 mm) and small (S-MP <0.5 mm) microplastics. SD ¼ standard deviation.

Polymer N Min-max size (mm) Mean size (mm) ± SD Median size (mm)

L-MP S-MP L-MP S-MP L-MP S-MP

PE 53 66 0.16 e 12.13 1.77 ± 2.64 0.33 ± 0.08 0.67 0.33

PP 73 75 0.16 e 6.80 1.12 ± 1.13 0.40 ± 0.07 0.76 0.41
PVC 64 70 0.16 e 6.89 0.92 ± 0.82 0.39 ± 0.06 0.73 0.40
PS 33 47 0.17 e 4.42 1.03 ± 0.91 0.41 ± 0.04 0.78 0.39
PA 84 54 0.16 e 6.19 2.46 ± 1.52 0.26 ± 0.08 2.04 0.23
PET 51 30 0.18 e 5.36 1.40 ± 0.98 0.25 ± 0.13 1.10 0.20

to be able to produce results with limited background contamina- tested (Fig. 3). The SMI unit also performed well, although recovery
tion. Consequently, simple preventive measures that can be applied of PET was sometimes below 90%. The decanter and the OC-M
in most laboratories were used in the current study (Supp. 2) and performed relatively poorer. For the following experiments and
potential sources of contamination were identified using a series of our final protocol, we used the OC-T considering the good perfor-
blank samples. First, blanks of the three solutions used, 200 mL mance and greater volume of sediment that could be handled by
ZnCl2, 5 mL Nile Red solution and 200 mL water were analysed. For this column.
the latter, unfiltered tap water, unfiltered deionized water, and tap
water vacuum filtered through a GFF filter were tested. Addition-
3.3. Experiment 3: microplastic quantification and identification
ally, filtered tap water in a plastic squeeze bottle was also tested.
case study with estuarine sediments
Secondly, a laboratory blank in which the whole protocol was
executed without adding any sediment was conducted.
A large number of microplastics were extracted from the sedi-
ment sample (representing different habitats and depth layers)
3. Results across size classes (Fig. 4, summary statistics per habitat, depth
layer and size class are presented in Supp. 5). For example, the
3.1. Experiment 1: testing methods for organic matter reduction mean number of microplastics extracted ranged from 111
and their effects on plastics (mean ± 62 SE) large microplastics L1 wet sediment at the surface
of the high mudflat to 8128 (mean ± 4181 SE) small microplastics of
Fenton’s reagent was on average 25.3% (± 18.5% standard error, size class B L1 wet sediment at 5 cm depth in the saltmarsh. The
SE) more efficient than H2O2 (Supp. 3). However, results were mean number of microplastics L1 wet sediment in each habitat
dependent on the original organic matter content of the sample. For (integrating all depth layers) increased from large microplastics
example, Fenton’s reagent was 59.3% more efficient than H2O2 in towards smaller microplastics in size classes A and B (covering sizes
the low intertidal sample, which only contained 2.48% organic from 5 to 0.125 mm, Fig. 4). The concentration then remained stable
matter. By contrast, in the high intertidal mudflat and saltmarsh across classes B and C (covering sizes from 0.250 to 0.0625 mm),
samples, which were characterised by higher organic matter loads and then declined again towards small microplastics below
(Table 1), the difference in efficiency was only 20.8% and 0.04% 0.0625 mm (size class D, Fig. 4).
respectively. The variability in the concentration of microplastics L1 wet
Microplastics of most polymer types showed only minor (up to sediment in a sample remained comparable across the large
10%) increase or decrease in surface area after treatment with H2O2 microplastics and the small microplastics of size class A and B
and Fenton’s reagent (Fig. 2, Supp. 4). The fact that there was both (Fig. 4). Variability, however, became high in some sediment sam-
an increase and a decrease suggests that this 10% is the observa- ples for size class C and for most samples in size class D (Fig. 4).
tional error, rather than a treatment effect. Notable exceptions were Indeed, the mean coefficient of variation (¼ ratio of the standard
PE microplastics which showed a decrease of 31.1% ± 2.4% SE and deviation to the mean) around the mean number of microplastics
21.7% ± 1.8% SE in surface area for H2O2 and Fenton’s reagent L1 wet sediment in each habitat remained largely comparable
respectively. Additionally, PET microplastic surface areas decreased within the saltmarsh habitat across size classes, but increased
by 42.5% ± 2.7% SE with H2O2 but showed no substantial change strongly for small microplastics of size class D (Supp. 6).
(3.3% ± 3.9% SE) with Fenton’s reagent. Pearson’s correlations in The majority of particles (91.7%) selected using automated im-
ATR FT-IR spectra before and after Fenton’s reagent or H2O2 treat- age analysis of Nile Red stained filters were found to be of plastic
ment were on average 0.930 (± 0.015 SE) and 0.931 (± 0.015 SE) origin, dominated by polyolefins (PE, PP and mixtures of these two)
respectively across the seven plastic polymers. All polymers had after validation using mFT-IR (Fig. 5). Non-plastic items were mainly
correlations above 0.900, except for EPS treated with Fenton’s re- of organic origin. The 50 particles that did not get selected as po-
agent (correlation: 0.894), and PS treated with Fenton’s reagent tential microplastics using the Nile Red threshold method were all
(correlation: 0.869) or H2O2 (correlation: 0.870). Spectra of poly- confirmed to be non-plastics. Example spectra for some particles
mers before and after Fenton’s reagent treatment are provided in identified as potential microplastics using Nile Red staning, and
supplement 4. corresponding mFT-IR reference spectra, are available in Supp. 7.

3.2. Experiment 2: density separation columns 3.4. Experiment 4: recovery rates

The overflow column with overflow at the top (OC-T) performed Mean recovery rates across polymer types were 90.7% (± 7.7
well with a mean recovery rate above 90% for all four polymers standard deviation, SD) for large microplastics (>¼ 0.5 mm) and
 P. Vermeiren et al. / Environmental Pollution 262 (2020) 114298 7

Fig. 2. Changes in the surface area of microplastics of seven common polymers encountered in the marine environment, before and after treatment with two different treatments to
remove organic matter from sediment samples: H2O2 (in blue) and Fenton’s reagent (in red). For interpretation of the references to colour in this figure legend, the reader is referred
to the Web version of this article.

both size classes and PE in the large microplastic size class. The only
polymer with more than 100% recovery is PA for the small sized
microplastics.

3.5. Quality control

No plastics were detected in the Nile Red solution. The ZnCl2

solution contained five pieces (mean longest axis ± SD:
293.6 mm ± 93.0 mm) in 200 mL. Deionized water contained 129
microplastics (mean longest axis ± SD: 145.1 mm ± 165.9 mm) in
200 mL, while 200 mL tap water only contained 1 fibre (0.9 mm
long) and three pieces (mean longest axis ± SD:
222.0 mm ± 63.3 mm). 200 mL tap water checked directly after
vacuum filtering through a GFF filter did not contain any micro-
plastics, but the same volume of filtered tap water coming from a
squeeze bottle contained 15 pieces (mean longest axis ± SD:
119.1 mm ± 81.3 mm).
The blanks for the whole protocol revealed a mean of 2 (± 1 SD)
Fig. 3. Recovery rates in percentage of microplastics from four polymer types across large microplastics (<5 mm and >¼0.5 mm) and 10 (± 2), 23 (± 4),
four different density separation column types. SMI: sediment microplastic isolation
44 (± 8) and 33 (± 3) small microplastics of class A (<0.5 mm and
unit (Coppock et al., 2017), OC-T: overflow column with top inflow, OC-M: overflow
column with mid-level inflow, see Table 2 for details of column types.
>¼0.250 mm), class B (<0.250 mm and >¼0.125 mm), class C
(<0.125 mm and >¼0.0625 mm) and class D (<0.0625 mm and
>¼0.05 mm) respectively per sample of 60 mL, which were sub-
95.0% (± 12.5 SD) for small microplastics (<0.5 mm, Table 4). All tracted from the presented results in the case study to correct for
polymer types have recoveries above 90% except PVC and PET in background contamination.
8 P. Vermeiren et al. / Environmental Pollution 262 (2020) 114298

Fig. 4. Mean (± SE) number of microplastics (MPs) L1 wet sediment across three replicate core samples in two habitat types (saltmarsh and high tidal mudflat) and four different
depth layers within size ranges (in mm): large microplastics (L-MPs) and small microplastics (S-MPs). Note that in the saltmarsh habitat at 15 cm only two replicate cores were
taken. The bottom right panel shows the mean number of microplastics L1 wet sediment in each habitat (across three replicate cores and four depth layers, except for the saltmarsh
habitat at 15 cm where only two replicate cores were taken), with saltmarsh in green and high intertidal mudflat in brown. For interpretation of the references to colour in this
figure legend, the reader is referred to the Web version of this article.

4. Discussion when comparing results to other studies. Such an adjustment factor

is subject to further validation and likely varies among polymers.
The current study developed and validated a holistic protocol to For example, 50 particles not meeting the Nile Red threshold were
extract, quantify and identify microplastics from organic rich sed- analysed in the current study. However, a full analysis of all parti-
iments with fine grain size. The protocol targeted large scale cles on a filter using mFT-IR scanning techniques would be the most
environmental monitoring of microplastic pollution that can be reliable method to establish under- or over-estimation for the
conducted within a typical ecology or environmental science lab- protocol. Such scanning methods, however, were not available for
oratory. Therefore, we opted to combine Nile Red staining the current study. When such methods become more generally
(including specific thresholds to separate plastics from non- available, it would be an important further research direction to
plastics) with selected validation using mFT-IR as a relatively cost- reduce uncertainty in concentration estimates obtained from Nile
effective and repeatable solution. Red stained particles.
High recovery rates, on average 91.9% (±7.7% SD), were detected Recovery rates varied among polymers. For example, the heavier
in an estuarine case study across size classes of the six most com- polymers, PET and PVC, had lower recovery rates, despite the fact
mon polymers. Meanwhile, none of the unstained particles (i.e. not that their densities (1.37 g cm3 and 1.38 g cm3 respectively)
meeting the Nile Red threshold) analysed using mFT-IR were iden- should promote floating in the 1.5 g cm3 density ZnCl2 solution
tified as plastics. Hence, the protocol currently overestimates the used in the current protocol. The lower recovery of PVC and PET
number of microplastics, requiring an adjustment factor to be used could lead to an underestimation of their concentrations in
 P. Vermeiren et al. / Environmental Pollution 262 (2020) 114298 9

number of microplastics L1 was much lower in the smallest size

class (class D, <0.0625 mm and >¼0.05 mm) compared to the other
size classes. Size class D only included a narrow size range
compared to the other size classes. One would generally expect the
number of microplastics to increase with increasing fragmentation
(Wright et al., 2013). For these reasons, we suggest the quantifi-
cation limit of our method to be currently at 125 mm. The detection
of small sized microplastics within class C and D (i.e. <0.125 mm), in
absence of their quantification, should not automatically lead to
these data being discarded. In fact, with decreasing size, micro-
plastics become more easily ingested by a wide range of organisms
(Wright et al., 2013) and integrated into tissues (Farrell and Nelson,
2013). Further methodological improvements could assist in
bringing the size limit down. The current 50 mm lower size limit
represented 4.5 pixels with the present camera setup and magni-
fication, which was taken as a minimum number of pixels to
Fig. 5. Percentage of particles (n ¼ 120) stained with Nile Red, identified as different represent a distinct particle. Alternatively, albeit small micro-
plastic polymer families (with main polymers within the family in between brackets)
plastics (<0.5 mm) were recovered at a higher rate than large
and non-plastic materials after validation with mFT-IR.
microplastics (Table 4), it is possible that the extraction process
itself is less efficient for the very small microplastics in size classes
sediment samples. By contrast, small microplastics (<0.5 mm) of PA C and D (i.e. <0.125 mm). The recovery rates could currently only be
were recovered in excess of 100% (Table 4), leading to their over- tested with microplastics down to about 0.160 mm (Table 3).
estimation and a shift in size distribution of PA particles towards The recovery rates (experiment 4) and results of the case study
smaller sizes. This likely resulted from the fragmentation of the PA (experiment 3) were based on sediment samples with a volume of
mesh that we used in our experiments. The edges of this mesh were 60 mL. Given the high number of microplastic detected within each
often frayed after going through the protocol, when inspected un- sample (see e.g. Fig. 4), a sample volume of 60 mL seems appro-
der a microscope. Recovery rates of large PE microplastics priate to detect trends in microplastic concentrations within estu-
(>¼0.5 mm) were less than recovery rates of other low-density arine sediment matrices. Other studies have reported
polymers such as PP and PS. Potentially, the shrinking effect of concentrations per litre or per kilogram. Further standardisation of
Fenton’s reagent pre-treatment to reduce organic matter on PE monitoring protocols is needed to improve comparability among
pieces (Fig. 2) resulted in their relative lower recovery in the larger studies. In this context, an assessment of the potential uncertainty
microplastics size class. We did not monitor the temperature dur- introduced when extrapolating from a 60 mL sample towards other
ing the Fenton’s reaction. Recently, Liu et al. (2019) kept their units should be conducted.
samples on ice while applying Fenton’s reagent to control the Techniques such as FT-IR and Raman spectroscopy are able to
temperature during this exothermic reaction to between 15
C and detect microplastics down to 5e20 mm (Harrison et al., 2012; Lo € der
19
C. Although plastic polymers generally have high melting et al., 2015; Tagg et al., 2015; Bergmann et al., 2019) and 5e10 mm
points, further analyses at different temperatures could shed light (Ka€ppler et al., 2015) respectively. However, major issues in current
into the causes of particle shrinking for specific polymers. We available methodologies are the high cost and extended time spans
suggest that results be interpreted in light of polymer-specific re- required for automated sample processing, preventing large scale
covery rates, and that comparison among studies should be done monitoring. Recent advancements have reduced the time require-
using an adjustment factor. It should be noted that many of the ment (e.g. Bergmann et al., 2019 analysed polymer composition on
current used protocols would require such an adjustment factor as an area of 14.1 mm by 14.1 mm on a 25 mm vacuum filter in 4.5 h
almost all current protocols show bias. However, the current pro- using array based mFT-IR imaging). Yet, the high cost of these
tocol, by defining clear thresholds, produces repeatable (and technologies will likely remain an obstacle towards their general
adjustable in the future) results compared to pure visual observa- implementation across multiple laboratories in the near future. We
tion methods. opted to combine Nile Red staining with selected validation using
A case study with estuarine sediments across different depths mFT-IR with the aim of developing a relatively low cost and time
within two habitats demonstrated the ability of the protocol to effective alternative (Tamminga et al., 2017, 2018). There are 24 h
detect microplastics down to 62.5 mm (Fig. 4). However, quantifi- resting periods after organic matter reduction and density sepa-
cation is likely only possible down to 125 mm. Sample variation ration, and for drying of the filters (steps 1 to 5, Fig. 1). Conse-
increased in the size classes below 125 mm (class C and D), and the quently, the whole protocol still takes about 1 week for a set of

Table 4
Recovery rates of different sized microplastics from spiked sediments undergoing the full protocol developed in the current study. Results are displayed for microplastics of
polymers commonly detected in the marine environment. Data on the density of plastic polymers was derived from Table 1 in Andrady (2011), except the density of PA, which
was obtained from the British Plastics Federation (https://www.bpf.co.uk/plastipedia/polymers/Polyamides.aspx, accessed 05 June 2019). SD ¼ standard deviation.

Polymer Total recovery (%) L-MP (>¼0.5 mm) recovery (%) S-MP (<0.5 mm) recovery (%) density (g cm3)

PE 89.1 83.6 93.8 0.91e0.94

PP 98.6 100 96.6 0.83e0.85
PS 96.3 97.5 95.0 1.05
PA 100.7 95.2 117.6 1.1e1.4
PET 84.0 83.6 85.7 1.37
PVC 82.8 84.1 81.5 1.38
Mean 91.9 90.7 95.0
SD 7.7 7.7 12.5
10 P. Vermeiren et al. / Environmental Pollution 262 (2020) 114298

samples. In the current study, we processed three samples at the of the SMI unit, but obtaining a quality ball valve needed to avoid
same time. However, techniques such as automated mFT-IR and microplastics getting stuck in the mechanism, thereby cross
mRaman spectroscopic imaging also require extraction of micro- contaminating between samples, became expensive for larger sized
plastics from the sediment and preparation steps (e.g. removal of columns, and thus not in line with our general objective of devel-
organic matter). These extraction and preparation steps are the oping a cost-effective protocol. Pieces of lighter polymers, PP and
main time consuming steps, not the actual counting and identifi- PE, stuck to the decanting column during decanting, thereby
cation of microplastics on vacuum filters. For example, the prepa- potentially explaining the low recovery for these polymers with
ration (i.e. picking up the 120 fluorescing and 50 lightly fluorescing this column. The OC-M column obtained lowest recovery rates,
particles and placing them in mini-KBr plates) and analysis of the possible as the mid column inflow caused too much turbulence to
particles for selected validation using mFT-IR in the current study allow a clean separation using density flotation.
was conducted in about 3 days of preparation and 2 days of anal- A range of alkaline, acidic, oxidative and enzymatic methods to
ysis. The photographing and image analysis of the filters of the case digest organic matter are available with enzymatic methods likely
study (experiment 3), likewise took about 1 week. most effective and specific, but also most expensive. (Ivleva et al.,
Potentially, some steps of the protocol can be sped up, for 2017; Miller et al., 2017). All methods are likely to have some im-
instance using centrifugation to assist in density separation. How- pacts on microplastics (Miller et al., 2017). We found Fenton’s re-
ever, we wanted to keep the number of manipulations to a mini- agent to be more effective compared to H2O2 supporting the
mum to prevent background contamination. Similarly, organic findings of Tagg et al. (2017). Fenton’s reagent resulted in the
matter reduction can be sped up, and potentially increased in ef- shrinking of PE and on PET pieces, similar to effects of H2O2 (Masura
ficacy, by higher temperatures. We tested Fenton’s reagent in a hot et al., 2015; Tagg et al., 2017).
bath at 40
C and 60
C after 30 min at room temperature. However, ZnCl2 has some environmental impact, therefore we vacuum
the reaction often became violent. We opted not to use this, given filtered this solution for re-use (after checking the density, Lo € der
the already high efficiency, potential safety issues, and the prob- and Gerdts, 2015). An often used alternative, NaCL, is inexpensive
lems of having to individually tailor the heating time and temper- and environmentally friendly. At a density of 1.2 g cm3, not all
ature to each sample. plastics are sampled with NaCL although techniques such as Air
Nile Red staining produced good results with 91.7% of particles Induced Overflow or Elutriation (Claessens et al., 2013; Nuelle et al.,
identified as plastics. These results likely vary with the exact light 2014) can offer a solution. A main concern, however, is that the
and camera configuration. Currently, the parameters of the camera- more involved manipulations could also increase background
microscope setup should be checked in each individual laboratory contamination. We opted to use regularly available laboratory
where the protocol is applied. Standardisation of the camera- equipment as this would represent the likely scenario in which the
microscope setup is therefore an important future research direc- protocol would be used, rather than assuming top level equipment
tion. Some of the microplastics that we tested using mFT-IR were to be available everywhere. Nonetheless, contamination could be
very small and difficult to pick up from the filter. Potentially, some reduced with improved facilities (Supp. 2).
of the measurements that came out as non-plastic might have been
misreadings were other materials were picked up together with, or
instead of, the actual microplastic. Equipping the mFT-IR microscope 5. Conclusion
with a fluorescent light could be a methodological improvement.
Some organic materials, such as chitin, can show a dim glow with Many hot spots of microplastic contamination such as harbours
red fluorescence (Shim et al., 2016). The red fluorescence threshold and estuaries are characterised by organic rich sediments with fine
developed in the current study likely avoided false identifications of grain size. Consequently, protocols applicable to such sediments are
chitin as plastics as none of the mFT-IR validated particles came out critically needed. We developed and validated a holistic protocol
as chitin. Nonetheless, an alternative option might be to also test pooling together fragmentary methodological advances and
green fluorescence in future studies (Shim et al., 2016). The use of applying them to organic rich sediments with fine grain size, spe-
visual identification of the items above 500 mm, although subse- cifically estuarine sediments. The overall recovery of the protocol
quently validated with Nile Red staining, can create uncertainty. across polymer types was on average 90.7% (± 7.7% SD) for large
Ultrasonication of the overflow liquid could provide an alternative microplastics (>¼0.5 mm) and 95.0% (± 12.5% SD) for small
to further disintegrate larger organic matter pieces (Wagner et al., microplastics (<0.5 mm e 0.160 mm). This makes the protocol
2017). An alternative is to use a larger sieve size of 1 or 2 mm, suitable for large scale monitoring of microplastics in sediments of
which would retain pieces that are easier identified visually estuarine origin provided polymer specific recovery rates, back-
(Hidalgo-Ruz et al., 2012). ground contamination, and uncertainty in Nile Red identification is
Among the different density separation columns used, the accounted for. The protocol currently allows for quantification of
newly designed OC-T column performed best. Similar to the prin- microplastics down to 125 mm, but could likely be extended using
ciple of the OC-T, Mani et al. (2019) conducted an overflow with a photographs of increased magnification. Given the demonstrated
1000 mL Erlenmeyer flask. During initial pilot testing of different high contamination of harbour and estuarine locations by micro-
options, we found the overflow with an Erlenmeyer flask messier as plastics, the protocol provides a useful tool for assessing the spatial
it overflows on all sides. This then necessitates an extra step of and temporal variability of this contamination, which is a necessary
having to clean the whole outside of the flask, and collecting the basis for its scientifically based environmental management.
overflow liquid becomes harder for large volumes. Nonetheless, an Although specifically tested for estuarine sediments in the current
Erlenmeyer made of glass is easily available, while the OC-T was study, the protocol could also offer a solution to similar organic rich
custom made. We have been able to produce a new OC-T with a sediments with fine grain size, such as some soils and sludge, to
glass column, but the overflow piece is currently still plastic. The improve our ability to conduct large scale monitoring of micro-
poorer performance of the SMI unit might have resulted from lack plastic contamination.
of experience recovering transparent pieces from the vacuum filter
mouth rather than from the SMI unit itself. Alternatively, the SMI Declaration of competing interest
unit had smaller dimensions, which might have affected the
settling behaviour of the plastics. We considered increasing the size None.
 P. Vermeiren et al. / Environmental Pollution 262 (2020) 114298 11

CRediT authorship contribution statement j.marpolbul.2017.11.010.

Hidalgo-Ruz, V., Gutow, L., Thompson, R.C., Thiel, M., 2012. Microplastics in the
marine environment: a review of the methods used for identification and
P. Vermeiren: Conceptualization, Methodology, Validation, quantification. Environ. Sci. Technol. 46, 3060e3075.
Formal analysis, Data curation, Writing - original draft, Writing - Ivleva, N.P., Wiesheu, A.C., Niessner, R., 2017. Microplastic in aquatic ecosystems.
review & editing, Visualization, Supervision, Funding acquisition. C. Angew. Chem. Int. Ed. 56, 1720e1739.
Ka€ppler, A., Windrich, F., Lo € der, M.G.J., Malanin, M., Fischer, D., Labrenz, M.,
Mun ~ oz: Conceptualization, Methodology, Validation, Data cura- Eichhorn, K.J., Voit, B., 2015. Identification of microplastics by FTIR and Raman
tion, Writing - review & editing. K. Ikejima: Conceptualization, microscopy: a novel silicon filter substrate opens the important spectral range
Methodology, Writing - review & editing, Funding acquisition. below 1300 cm(-1) for FTIR transmission measurements. Anal. Bioanal. Chem.
407, 6791e6801. https://doi.org/10.1007/s00216-015-8850-8.
Liu, F., Olesen, K.B., Borregaard, A.R., Vollertsen, J., 2019. Microplastics in urban and
Acknowledgements highway stormwater retention ponds. STOTEN 671, 992e1000. https://doi.org/
10.1016/j.envpol.2019.113335.
€der, M.G.J., Kuczera, M., Mintenig, S., Lorenz, C., Gerdts, G., 2015. Focal plane array
Lo
P. Vermeiren was supported by an international postdoctoral detector-based micro-Fourier-transform infrared imaging for the analysis of
fellowship from the Japan Society for the Promotion of Science microplastics in environmental samples. Environ. Chem. 12, 563e581. https://
(KAKENHI grant nr. JP 18F17405). The authors acknowledge the doi.org/10.1071/EN14205.
€der, M.G.J., Gerdts, G., 2015. Chapter 8, Methodology used for the detection and
Lo
valuable assistance of the students, particularly Hotta Kousuke, of identification of microplastics - a critical appraisal. In: Bergmann, M., Gutow, L.,
the Laboratory for Coastal Ecology and Conservation in sample Klages, M. (Eds.), Marine Anthropogenic Litter, pp. 201e227. https://doi.org/
collection and processing. The authors are grateful to Prof. Ichiura 10.1007/978-3-319-16510-3.
Maes, T., Jessop, R., Wellner, N., Haupt, K., Mayes, A.G., 2017. A rapid-screening
of the Faculty of Agriculture and Marine Science, Kochi University, approach to detect and quantify microplastics based on fluorescent tagging
for the use of the ATR FT-IR and to Dr. Horikawa from the Kochi with Nile Red. Sci. Rep. 7, 1e10. https://doi.org/10.1038/srep44501.
Prefectural Industrial Technology Centre for use and technical Mani, T., Primpke, S., Lorenz, C., Gerdts, G., Burkhardt-Holm, P., 2019. Microplastic
support of the mFT-IR. The authors also thank Dr. Arai and Dr. pollution in benthic midstream sediments of the Rhine river. Environ. Sci.
Technol. 53, 6053e6062.
Uramoto, from the Centre for Advanced Marine Core Research, Masura, J., Baker, J., Foster, G., Arthur, C., 2015. Laboratory methods for the analysis
Kochi University, for their technical support in using the laser of microplastics in the marine environment: recommendations for auantifying
diffraction particle sizer for sediment grain size analysis. The au- synthetic particles in waters and sediments. NOAA technical memorandu, NOS-
OR&R 48.
thors are grateful for the comments of three anonymous reviewers. Miller, M.E., Kroon, F.J., Motti, C.A., 2017. Recovering microplastics from marine
samples: a review of current practices. Mar. Pollut. Bull. 123, 6e18. https://
Supplementary data. Supplementary data doi.org/10.1016/j.marpolbul.2017.08.058.
Moore, C.J., 2008. Synthetic polymers in the marine environment: a rapidly
increasing, long-term threat. Environ. Res. 108, 131e139. https://doi.org/
Supplementary data to this article can be found online at 10.1016/j.envres.2008.07.025.
https://doi.org/10.1016/j.envpol.2020.114298. Nor, N.H.M., Obbard, J.P., 2014. Microplastics in Singapore’s coastal mangrove eco-
systems. Mar. Pollut. Bull. 79, 278e283.
Noren, F., 2007. Small plastic particles in coastal Swedish waters. KIMO Rep. 11.
References Nuelle, M.T., Dekiff, J.H., Remy, D., Fries, E., 2014. A new analytical approach for
monitoring microplastics in marine sediments. Environ. Pol. 184, 161e169.
Andrady, A.L., 2011. Microplastics in the marine environment. Mar. Pollut. Bull. 62, https://doi.org/10.1016/j.envpol.2013.07.027.
1596e1605. Primpke, S., Lorenz, C., Rascher-Friesenhausen, R., Gerdts, G., 2017. An automated
Andrady, A.L., 2015. Persistence of plastics litter in the oceans. In: Bergmann, M., approach for microplastics analysis using focal plane array (FPA) FTIR miscro-
Gutow, L., Klages, M. (Eds.), Marine Anthropogenic Litter, pp. 57e72. scopy and image analysis. Anal. Methods 9, 1499e1511. https://doi.org/10.1039/
Andrady, A.L., Neal, M., 2009. Applications and societal benefits of plastics. Philos. c6ay02476a.
Trans. Royal Soc. B 364. https://doi.org/10.1098/rstb.2008.0304, 1977e84. Schneider, C.A., Rasband, W.S., Eliceiri, K.W., 2012. NIH Image to ImageJ: 25 years of
Barnes, D.K.A., Galgani, F., Thompson, R.C., Barlaz, M., 2009. Accumulation and image analysis. Nat. Methods 9, 671e675.
fragmentation of plastic debris in global environments. Phil. Trans. Roy. Soc. Seto, K.C., 2011. Exploring the dynamics of migration to mega-delta cities in Asia
Lond. 364, 1985e1998. and Africa: contemporary drivers and future scenarios. Global Environ. Change
Bergmann, M., Mützel, S., Primpke, S., Tekman, M.B., Trachsel, J., Gerdts, G., 2019. 21S, 94e107.
White and wonderful? Microplastics prevail in snow from the alps to the arctic. Shim, W.J., Song, Y.K., Hong, S.H., Jang, M., 2016. Identification and quantification of
Sci. Adv. 5, eaax1157. microplastics using Nile Red staining. Mar. Pollut. Bull. 113, 469e476. https://
Browne, M.A., Galloway, T.S., Thompson, R.C., 2010. Spatial patterns of plastic debris doi.org/10.1016/j.marpolbul.2016.10.049.
along estuarine shorelines. Environ. Sci. Technol. 44, 3404e3409. Tagg, A.S., Sapp, M., Harrison, J.P., Ojeda, J.J., 2015. Identification and quantification
Claessens, M., Van Cauwenberghe, L., Vandegehuchte, M.B., Janssen, C.R., 2013. New of microplastics in wastewater using focal plane array-based reflectance micro-
techniques for the detection of microplastics in sediments and field collected FT-IR imaging. Anal. Chem. 87, 6032e6040. https://doi.org/10.1021/
organisms. Mar. Pollut. Bull. 70, 227e233. acs.analchem.5b00495.
Cole, M., Lindeque, P., Halsband, C., Galloway, T.S., 2011. Microplastics as contami- Tagg, A.S., Harrison, J., Ju-Nam, Y.K., Sapp, M., Bradley, E.L., Sinclair, C.J., Ojeda, J.J.,
nants in the marine environment: a review. Mar. Pollut. Bull. 62, 2588e2597. 2017. Fenton’s reagent for the rapid and efficient isolation of microplastics from
Coppock, R.L., Cole, M., Lindeque, P.K., Queir, A.M., Galloway, T.S., 2017. A small-scale wastewater. Chem. Commun. 53, 372e375. https://doi.org/10.1039/
, portable method for extracting microplastics from marine sediments. Environ. C6CC08798A.
Pollut. 230, 829e837. https://doi.org/10.1016/j.envpol.2017.07.017. Tamminga, M., Hengstmann, E., Kerstin Fisher, E.K., 2017. Nile red staining as a
Farrell, P., Nelson, K., 2013. Trophic level transfer of microplastic: Mytilus edulis (L.) subsidiary method for microplastic quantification: a comparison of three sol-
to Carcinus maenas (L.). Environ. Pol. 177, 1e3. vents and factors influencing application reliability. JESES 2. https://doi.org/
Felsing, S., Kochleus, C., Buchinger, S., Brennholt, N., Stock, F., Reifferscheid, G., 2018. 10.15436/JESES.2.2.1.
A new approach in separating microplastics from environmental samples based Tamminga, M., Hengstmann, E., Kerstin Fisher, E.K., 2018. Microplastic analysis in
on their electrostatic behavior. Environ. Pol. 234, 20e28. https://doi.org/ the south Funen Archipelago, Baltic Sea, implementing manta trawling and bulk
10.1016/j.envpol.2017.11.013. sampling. Mar. Pollut. Bull. 128, 601e608. https://doi.org/10.1016/
Fischer, E., 2019. Distribution of Microplastics in Marine Species of the Wadden Sea j.marpolbul.2018.01.066.
along the Coastline of Schleswig-Holstein. Technical Report CEN, University of UNEP, 2014. UNEP Year Book 2014: Emerging Issues in Our Global Environment.
Hamburg, Germany. Nairobi, Kenya.
Gregory, M.R., 1996. Plastic ‘scrubbers’ in hand cleansers: a further (and minor) Vermeiren, P., Mun ~ oz, C.C., Ikejima, K., 2016. Sources and sinks of plastic debris in
source for marine pollution identified. Mar. Pollut. Bull. 32, 867e871. https:// estuaries: a conceptual model integrating biological, physical and chemical
doi.org/10.1016/S0025-326X(96)00047-1. distribution mechanisms. Mar. Pollut. Bull. 113, 7e16. https://doi.org/10.1016/
Harrison, J.P., Ojeda, J.J., Romero-Gonz alez, M.E., 2012. The applicability of reflec- j.marpolbul.2016.10.002.
tance micro-Fourier-transform infrared spectroscopy for the detection of syn- Wagner, J., Wang, Z., Ghosal, S., Rochman, C., Gassel, M., Wall, S., 2017. Novel method
thetic microplastics in marine sediments. Sci. Total Environ. 416, 455e463. for the extraction and identification of microplastics in ocean trawl and fish gut
https://doi.org/10.1016/j.scitotenv.2011.11.078. matrices. Anal. Methods 62, 2588e2597. https://doi.org/10.1039/C6AY02396G.
Hengstmann, E., Tamminga, M., vom Bruch, C., Kerstin, F., 2018. Microplastic in Wagner, M., Scherer, C., Alvarez-Munoz, D., Brennholt, N., Bourrain, X., Buchinger, S.,
beach sediments of the Isle of Rügen (Baltic Sea) e implementing a novel glass Fries, E., Grosbois, C., Klasmeier, J., Marti, R., Rodriguez-Mozaz, S., Urbatzka, R.,
elutriation column. Mar. Pollut. Bull. 126, 263e274. https://doi.org/10.1016/ Vethaak, A.D., Winther-Nielsen, M., Reifferscheid, G., 2014. Microplastics in
12 P. Vermeiren et al. / Environmental Pollution 262 (2020) 114298

freshwater ecosystems: what we know and what we need to know. ESEU 26, 12. Wright, S.L., Thompson, R.C., Galloway, T.S., 2013. The physical impacts of micro-
Wentworth, C.K., 1922. A scale of grade and class rerms for clastic sediments. J. Geol. plastics on marine organisms: a review. Environ. Pol. 148, 483e492. https://
30, 377e392. doi.org/10.1016/j.envpol.2013.02.031.