Marine Pollution Bulletin 159 (2020) 111507

Contents lists available at ScienceDirect

Marine Pollution Bulletin

journal homepage: www.elsevier.com/locate/marpolbul

Comparison of microplastic isolation and extraction procedures from marine T

sediments☆,☆☆
⁎
Michaela A. Cashmana,b, , Kay T. Hoa, Thomas B. Bovingb,c, Stephen Russod, Sandra Robinsona,
Robert M. Burgessa
a
U.S. Environmental Protection Agency, ORD/CEMM Atlantic Coastal Environmental Sciences Division, 27 Tarzwell Drive, Narragansett, RI 02882, USA
b
University of Rhode Island, Department of Geosciences, 9 E Alumni Avenue, Kingston, RI 02881, USA
c
University of Rhode Island, Department of Civil Engineering, 9 E Alumni Avenue, Kingston, RI 02881, USA
d
ORAU, c/o U.S. Environmental Protection Agency, ORD/CEMM Atlantic Coastal Environmental Sciences Division, 27 Tarzwell Drive, Narragansett, RI 02882, USA

A R T I C LE I N FO A B S T R A C T

Keywords: Microplastics (MPs) are small (< 5 mm) plastic particles which pose a threat to marine ecosystems. Identifying
Microplastics MPs is crucial for understanding their fate and effects. Many MP extraction methods exist, but procedural dif-
Sediment ferences prevent meaningful comparisons across datasets. This method comparison examines the efficiency of
Method comparison five methods for extracting MPs (40–710 μm) from marine sediments. Known quantities of MPs were spiked into
Plastics
sediments. The MPs were extracted and enumerated to demonstrate percent recovery. Findings determined that
Extraction
sediment matrix, MP properties, and extraction method affect the percent recovery of MPs from sediments.
Average recoveries of spiked microplastics were between 0 and 87.4% and varied greatly by sediment type,
microplastic, and method of extraction. In general, larger particle and lower density MPs were more effectively
recovered. Marine sediments low in organic matter and with larger grain size also had higher percent recoveries
of MPs. These findings support the need for method optimization and unified procedures.

1. Introduction rising to the forefront of emerging contaminant studies due to their

unique quantification challenges and unknown toxic effects. Given this
Oceanic plastic pollution has garnered international attention as an situation and their potential for environmental impacts, it is critical for
example of waste mismanagement. Over 8.3 billion metric tons of researchers to have scientifically robust methods for extracting and
plastic have been produced globally since the 1950s. Plastic consump- isolating MPs from sediments.
tion has surpassed the capacity of modern recycling infrastructure, There is an extensive list of published methods for isolating MPs
leading to mismanaged disposal and environmental pollution. It is es- from sediments (Prata et al., 2019; Mai et al., 2018). The variety of
timated that 8 million tons of plastic enter the oceans from land each published procedures reflects the unique challenges associated with
year (Jambeck et al., 2015). However, floating plastics account for only isolating MPs from marine sediments. Differences in extraction and
1% of the expected 8 million tons of plastic entering oceans annually isolation procedures ultimately determine the ability of various MPs to
(Van Sebille et al., 2015). Much of the 99% of the remaining plastics are be accurately recovered and quantified, resulting in a wide range of
expected to degrade into plastic fragments < 5 mm, known as micro- recovery efficiencies. Therefore, it is difficult to compare MP recovery
plastics (MPs) (Murphy, 2017) through a series of physical, chemical, rates (number of plastic particles per sample) between environmental
and biological processes (Van Cauwenberghe et al., 2013), and ulti- samples using different procedures. Developing methods to address a
mately accumulate in sediments (Fig. 1) (Andrady, 2011; Browne et al., range of sediment and plastic matrices has resulted in a wide variety of
2011; Hidalgo-Ruz et al., 2012; Kowalski et al., 2016). MPs are quickly extraction techniques. Procedural differences include the mass of

☆
This is U.S. EPA ORD-035523.
☆☆
Mention of trade names or commercial products does not constitute endorsement or recommendation for use. This report has been reviewed by the U.S. EPA's
Office of Research and Development Center for Environmental Measurement & Modeling Atlantic Coastal Environmental Sciences Division in Narragansett, RI, and
approved for publication. Approval does not signify that the contents necessarily reflect the views and policies of the Agency.
⁎
Corresponding author at: U.S. Environmental Protection Agency, ORD/CEMM Atlantic Coastal Environmental Sciences Division, 27 Tarzwell Drive, Narragansett,
RI 02882, USA.
E-mail address: [email protected] (M.A. Cashman).

https://doi.org/10.1016/j.marpolbul.2020.111507
Received 11 May 2020; Received in revised form 21 July 2020; Accepted 21 July 2020
Available online 04 August 2020
0025-326X/ Published by Elsevier Ltd.
M.A. Cashman, et al. Marine Pollution Bulletin 159 (2020) 111507

Fig. 1. Suggested pathways for microplastics in marine environments.

sediment samples, sample preparation, and sample handling. With no sediments involve density separation (i.e., floatation) by agitating se-
sediment standard reference material for MPs, methods are developed diment samples with aqueous salt solutions (Thompson et al., 2004).
with an array of sediment and plastic matrices. Sediments may go Methods that rely on floatation separations are restricted by the density
through pretreatment steps including oven drying (Su et al., 2018; of their respective salt solutions. Common plastics range in density from
Gilbreath et al., 2019; Nuelle et al., 2014), pre-sieving of coarse or fine 0.8–2.35 g/cm3 (Hidalgo-Ruz et al., 2012). Low density salt solutions
materials (Gilbreath et al., 2019; Zobkov and Esiukova, 2017), or such as sodium chloride may be insufficient to separate higher density
chemical oxidation (Zobkov and Esiukova, 2017; Masura et al., 2015; plastics from sediment (Hidalgo-Ruz et al., 2012; Coppock et al., 2017).
Maes et al., 2017; Hurley et al., 2018). The method efficacy is often However, high density salts (e.g., NaBr, NaI, ZnCl2) may not allow
dependent on sediment composition. Sediment properties such as differentiation among plastics and other sediment components making
grainsize, organic matter content, and minerology largely affect results separation from sediment particles difficult. In addition, the various
and method complexity. Differentiating between plastic and non-plastic salts used in density separation methods vary greatly in price, toxicity,
particles in environmental samples is another major obstacle in iso- reactivity, and waste disposal. These considerations can be restrictive or
lating and visually identifying microplastics (Shaw and Day, 1994; prohibitive to laboratories seeking to use higher density salts.
Tamminga et al., 2017). Recent studies indicate that using selective Many extraction methods favor low density plastic particles, but
fluorescent stains, such as Nile Red, may improve the detection of MPs environmental microplastics include a myriad of high-density plastic
in environmental samples (Maes et al., 2017; Shim et al., 2016). Nile polymers, including polyester (Browne et al., 2011; Lusher et al., 2014;
Red is a fluorescent stain that adheres to hydrophobic substances in- Lusher et al., 2013; Nor and Obbard, 2014), polyethylene terephthalate
cluding lipids and plastic. Hypothetically, staining environmental (PET) (Nor and Obbard, 2014; Peng et al., 2017), and polyethylene (PE)
samples with Nile Red reduces the likelihood of false positive identifi- (Rios et al., 2007). Small variations in plastic chemical composition
cation (Vianello et al., 2013). lead to large differences in polymer properties (Brydson, 1999). It is
This research assessed five current methods for the extraction and important to consider that many methods may inadvertently select for
isolation of MPs from marine sediments. Two sediments (one sandy and specific polymer fragments (i.e., microplastic spheres versus fibers)
one silty) were used as representative matrixes for amending known based on their physical properties (Nel et al., 2018). Another compli-
quantities of five common types of MPs. These matrices are considered cation is microplastics stimulate biofilm formation. MPs with biofilm
representative of the types of sediments found along the coasts of the will increase their particle mass and complicate density separation
United States including sand occurring at beaches and typical subtidal (Rummel et al., 2017; Zettler et al., 2013). Surficial biofilms often mask
temperate silty muds. Microplastics are known to accumulate in these the polymer type from spectrometers and can effectively camouflage
types of sediments. Consequently, methods for isolating and extracting plastic particles embedded within sediments. While this study does not
MPs from sediments would need to be functional with these types of address the complications of polymer spectrometry, many isolation
matrices. The MPs were chosen to represent a range of MP polymer methods use chemical oxidation as a means of further separating or-
types, shapes, sizes, densities, and colors. Nile Red was also evaluated ganic material and removing biofilms from the microplastic-sediment
for improving MP visibility during extraction and isolation. We present matrix (Gilbreath et al., 2019; Zobkov and Esiukova, 2017; Masura
a comparison of the efficiency of commonly used methods to char- et al., 2015; Coppock et al., 2017). In summary, several factors greatly
acterize the number and types of MPs in marine sediments. affect the overall method performance for recovering MPs from marine
Most techniques used to isolate MPs from marine and estuarine sediments. Many isolation and extraction methods exist, but there is no

2
M.A. Cashman, et al. Marine Pollution Bulletin 159 (2020) 111507

information on their relative performance. covered in aluminum foil. The foil cover was removed whenever the
working samples were exposed to air to assess possible air-born con-
2. Materials and methods tamination. Each filter was inspected under the microscope with both
normal light and UV excitation using a NightSea (Lexington, MA, USA)
2.1. Experimental set-up fluorescence filter (Excitation 360–380 nm, emission 415 nm long pass)
to quantify the number of particles adhered to the filter. Further in-
Methods were chosen to represent a wide range of commonly used, formation on clean laboratory setup and quality control can be found in
literature documented procedures, and for the ability to be performed the Supplemental information section.
easily and inexpensively. Other considerations included minimal waste
generation, low start-up costs, simple equipment and instrumentation 2.3. Methods compared
set-up, and overall quick processing time. Each method was assessed
using two model sediments, silty sediment from Long Island Sound, The following section outlines the general approach of each method.
New York (USA) and beach sand from Narragansett Beach, Rhode Each method was explicitly followed as described by the authors unless
Island (USA). Long Island Sound sediment (LIS) is a well characterized, stated below. Detailed extraction steps can be found in published
fine-grained sediment, collected using a Smith MacIntyre grab sampler methods. Methods will be referred to by the last name of the first author
(0.1 m2) in September 2010 (Ho et al., 2000). Narragansett Beach sand for the remainder of this manuscript.
(NAR) was collected by hand from the intertidal zone using a metal Fries et al. (2013): This method is a density separation approach
shovel in January of 2018. Sediments were press sieved through a 2 mm using sodium chloride (NaCl) solution (ϱ = 1.2 g/cm3). Wet sediment
sieve prior to analysis to remove any coarse fragments, and the NAR samples (175 g) underwent extraction in 2 L glass separatory funnels
was heated in a muffle oven at 550 °C for 6 h to remove organic ma- with the NaCl solution. Samples were vigorously shaken to float mi-
terial. Representative samples were analyzed for particle size distribu- croplastics to the NaCl solution surface. After a settling period, sedi-
tions using a Restech CamSizer P4 (Haan, Germany) (Table 1). ment was removed through the bottom port of the separatory funnel.
Five representative MPs were amended in known quantities into Suspended MPs in NaCl solution were filtered onto a 20 μm PCTE
each sediment sample to evaluate the recovery efficiency of the selected membrane filter and visually inspected using a Nikon SMZ745-T mi-
microplastic extraction methods. The representative microplastics re- croscope.
flect a wide variety of polymer type, fragment shape, and particle size. Gilbreath et al. (2019): This manuscript extracts MPs from bior-
For this study, we analyzed MPs ranging from 40 to 710 μm. The etention ponds, but uses a methodology commonly used to extract MPs
plastics used for spiking included polystyrene (PS), PE, polyvinyl from sediments. This method is a modified version of Stolte et al.
chloride (PVC), PET, and polypropylene (PP) (Table 2). Fluorescent (2015) that extracts MPs > 45 μm using a calcium chloride (CaCl2)
colored MPs were chosen for their ability to be easily enumerated as solution (ϱ = 1.4 g/cm3). Sieved (> 45 μm) sediment samples (150 g)
spiked reference materials. Both the PS and PE microbeads were pur- were split into size fractions (45–500 μm, 501–1000 μm) and placed in
chased from Cospheric LLC (Santa Barbara, CA, USA). The other three 600 mL glass beakers with CaCl2 solution. Samples were stirred vigor-
microplastics purchased were PVC pipe (Home Depot, GA, USA), PET ously and left to settle. All floating materials were transferred using a
embroidery floss (J&P Coats, Middlesex, UK), and PP rope (SeaChoice, metal spoon to 1-L glass separatory funnels filled with CaCl2. From
Pompano Beach, FL, USA). These three plastics were ground or cut into there, separatory funnels were shaken, and the suspension allowed to
small pieces and sieved through a series of stacked mesh sieves to ob- settle. After settling, floating materials were filtered onto a 20 μm PCTE
tain desired size classes (Table 2). MPs were stored in a glass jar con- membrane filter and visually inspected using a Nikon SMZ745-T mi-
taining filtered seawater (20 μm) from Narragansett Bay (Narragansett, croscope.
RI, USA) for a minimum of two weeks at 20 °C to develop a biofilm. Nuelle et al. (2014): This method is a density separation approach
Prior to sediment addition, each MP particle was individually inspected that uses both NaCl (ϱ = 1.2 g/cm3) and sodium iodide (NaI)
microscopically (Nikon SMZ745-T, Nikon, Minato, Tokyo, Japan) for (ϱ = 1.8 g/cm3) solutions. Sieved (< 1 mm) and dried (60 °C) sediment
shape abnormalities or fragmentation by two analysts. After inspection, samples (1 kg) were initially separated with air induced overflow (AIO),
a minimum of twenty plastic pieces per polymer type were carefully which uses an aerated NaCl solution for density separation. Sediment
transferred to a sediment sample (20 pieces ∗ 5 plastic types = 100 samples were fluidized using the AIO method, which floats the MPs out
pieces of plastic/sample). The plastic-amended sediments were mixed of the fluidized sample causing them to overflow into secondary con-
on a roller mill (4 RPM) at 4 °C for a minimum of 48 h. tainment. All materials in secondary containment were transferred to
500 mL glass volumetric flasks filled with NaI solution. The volumetric
2.2. Quality control flasks were shaken and decanted after a settling period. All decanted
materials were filtered onto a 20 μm PCTE membrane filter and visually
Each method was evaluated with a total of 12 spiked sediment inspected using a Nikon SMZ745-T microscope.
samples. An additional two sediment blanks (one sediment blank per Coppock et al. (2017): This method is a density separation approach
sediment type) and a water blank were used per method to assess that uses zinc chloride (ZnCl2) solution (ϱ = 1.5 g/cm3) and a sediment
background and cross contamination during extraction. Airborne microplastics isolation (SMI) unit. The SMI unit was constructed in our
background contamination was assessed with one air blank per sample laboratory. Sediment samples (70 g) were placed in the SMI unit with
extraction. Air blanks were collected by wetting a 20 μm polycarbonate ZnCl2 and a stir bar. Plastics were separated through density separation
track-etched (PCTE) filter (Poretics, GVS North America, Sanford, ME, driven by mixing with the stir bar. After settling, all floating materials
USA) with deionized water and placing the filter into a glass petri dish were filtered onto 20 μm PCTE filters and underwent oxidation (30%

Table 1
Physical properties and sampling locations for representative sediments: Long Island Sound and Narragansett Beach. Sediment sizes classified using grainsize
diameter 10, 50, and 90% cumulative percentile value.
D10 (μm) D50 (μm) D90 (μm) Water wt/wt% Organic carbon % GPS coordinates of collection location

Long Island Sound (LIS) 4.1 13.7 62.6 43 2 41° 7′N 72° 52′W
Narragansett Beach (NAR) 179.1 251.6 345.2 <1 0 41° 26′N 72° 27′W

3
M.A. Cashman, et al. Marine Pollution Bulletin 159 (2020) 111507

Table 2
Properties of microplastics used in this investigation.
Size (μm) Density (g/cm3) Shape Color Source

Polystyrene (PS) 40 0.96 Sphere Transparent Cospheric

Polyethylene (PE) 96–106 1.13 Sphere Blue Cospheric
Polyvinyl chloride (PVC) 500–710 1.35 Fragment Orange PVC pipe
Polyethylene terephthalate (PET) 250–500 1.38 Fiber Pink Embroidery floss
Polypropylene (PP) 500–710 0.91 Fiber Yellow Rope

H2O2) for 1 week. Oxidized samples were filtered onto new 20 μm PCTE temporal duration to effectively stain the five model plastics. The best
filters and visually inspected using a Nikon SMZ745-T microscope. results were obtained with a concentration of 0.025 g/L NR for a
The referenced ball valve used by Coppock et al. (2017) to construct staining duration of 10 min.
the SMI was not commercially available in the United States. Therefore, As described previously, post-processed filters from each method
our laboratory opted for a PVC ball valve constructed from a 6.4 cm slo- were analyzed to determine percent recovery of MPs. After analyzing
close valve made by Colonial Engineering Inc. (Portage, MI, USA). The percent recovery, a subset of these filters were stained with NR and
unit was constructed using 63 mm outer diameter PVC piping with the recounted to determine if NR staining affects percent recovery based on
ISO ball valve fixture adhered to a PVC plate (SI Fig. 1). Details for visual observation. Two samples (one sandy and one silty) from each
construction and operation can be found in Coppock et al. (2017). method were randomly selected for staining. Samples were mounted
Zobkov and Esiukova (2017): This method is an adaptation of the onto a vacuum filter apparatus and stained with 10 mL of 0.025 g/L NR
laboratory method published by the National Oceanic and Atmospheric solution for 10 min, ensuring the entire filter was covered with stain.
Administration (NOAA) (Masura et al., 2015) that uses ZnCl2 After 10 min, the samples were filtered and thoroughly rinsed with DI
(ϱ = 1.6 g/cm3) for density separation followed by oxidation catalyzed water to remove all NR stain. The filters were then inspected under on a
with a heated water bath. Sediment samples (400 g) were added to glass Nikon SMZ745-T microscope equipped with NightSea fluorescence
beakers containing aqueous ZnCl2 solution. After stirring with stainless filter (excitation 490–515 nm, emission 550 nm long pass) to recount
steel spoons and settling, floating debris and supernatant were filtered microplastics. MP counts were compared for each filter pre- and post-
through a 170 μm stainless steel sieve. Debris retained on the sieve were NR staining to determine whether NR affected percent recovery.
rinsed into clean glass beakers by tilting the sieve and rinsing into a
clean beaker with the addition of a 30% H2O2 and Fe (II) catalyst so- 2.6. Statistical analysis
lution. Beakers were covered with aluminum foil and placed in a hot
water bath (75 °C) for 15 h. A solution containing 4.5% hydrochloric Mean percent recovery of microplastics achieved by each method
acid (4.5%) was then added to each beaker. Samples underwent an- was determined as a function of polymer and sediment type using
other round of density separation with ZnCl2 and then were filtered Microsoft Excel (2016). All analyses of variance (ANOVA) were per-
onto 20 μm PCTE filters and visually inspected using a Nikon SMZ745-T formed using the SAS statistical software (SAS Institute Inc., Cary, NC,
microscope. USA; Version 9.4). Statistically significant differences (p < 0.05)
among methods were determined for each polymer and sediment type
2.4. Plastic characterization using ANOVA. Significant differences identified by the ANOVA were
further analyzed with a Bonferroni f-test to identify significant differ-
Each sample was ultimately filtered onto a 20 μm PCTE membrane ences among recovery rates. A recovery threshold of 70% was de-
filter for visual inspection. Samples high in organic matter and sedi- termined before the start of experiments as a desirable recovery rate to
ment were often filtered onto several separate filters to more evenly evaluate the effectiveness of each method. A one-way t-test was used to
distribute the debris including MPs. Filters were visually inspected compare each mean recovery to the 70% threshold between method,
under the microscope (Nikon SMZ745-T) using 2× magnification and sediment type, and plastic. Samples with average percent recoveries of
identified as spiked MPs with white and fluorescent light. Physical 0% were excluded from the t-test. Mean recoveries of MPs pre- and
properties of spiked MPs (color, fluorescence, shape and size) made post-Nile Red staining were determined for each polymer type and then
them easily identifiable with microscopy. Two people verified each MP analyzed using a one-way t-test to determine significant (p < 0.05)
count using both normal light and cyan excitation with a NightSea differences.
(Lexington, MA, USA) fluorescence filter (excitation 490–515 nm,
emission 550 nm long pass). Samples were recounted if there were 3. Results and discussion
discrepancies between the MP counts by both analysts. Spiked micro-
plastics were counted on each filter and tallied by polymer type. Filters 3.1. Overall trends
from sediment blank samples and water blank samples were visually
inspected in the same manner. All filters were stored at 20 °C in glass Recovered MPs were compared against known spiked MP quantities
petri dishes with foil lids after identification. to determine percent recovery for each sample (SI Tables 1; 2). Overall,
mean recoveries were slightly better in the sandy sediment with non-
2.5. Nile Red addition zero values ranging from 5% to 87% in the NAR sand compared to 2%
to 77% in the silty LIS sediment (Fig. 2). In addition, 36% of the re-
A secondary objective of this study was to determine the effects of coveries in the NAR exceeded 50% while only 20% of the recoveries
Nile Red (NR) staining on the observational counting of microplastics from the silty sediment exceeded 50% (Fig. 2). Mean recoveries for PVC
from each sample. Lipophilic dyes such as Nile Red help differentiate ranged from 33–86% for sand and 11–68% for silt. PE recoveries ranged
microplastics from their environmental matrices during visual ob- from 32–61% for sand and 0–52% for silt. For both types of sediments,
servation. Nile Red was purchased from Thomas Scientific (MP recoveries of PS were very low ranging from 0% to less than 20%. Mean
Biomedicals, Solon, OH, USA). A NR stock solution was prepared at recovery of PET ranged from 5–68% for sand and 2–58% for silt. Fi-
0.05 g/L in acetone according to methods developed by Maes et al. nally, for PP, recoveries ranged from 23–87% for sand and 0–77% for
(2017). Prior to staining experimental samples, a laboratory trial was silt.
performed to determine an appropriate staining concentration and Across all methods, the hierarchical ranking of mean recovery of

4
M.A. Cashman, et al. Marine Pollution Bulletin 159 (2020) 111507

Fig. 2. Mean percent recoveries of microplastics (error bars are the standard deviations). Different letters represent statistical differences between polymer recovery
(p < 0.05) per sediment type and method. Group “a” mean percent recovery is significantly greater than group “b”, which is significantly greater than group “c”.
Bars with two letters are not significantly different from either group. Orange bar color signifies samples with significantly greater (p < 0.05) than 70% recovery. X-
axis is organized first by plastic type (PVC = polyvinyl chloride, PE = polyethylene, PS = polystyrene, PET = polyethylene terephthalate, and PP = polypropylene),
and further subdivided by extraction method (C = Coppock, F = Fries, N = Nuelle, G = Gilbreath, and Z = Zobkov).

MPs by polymer was the same for both sediment types. PVC had the ranking was often a vanishingly small margin, and this ranking does not
highest recovery (i.e., 59 ± 25% sand; 43 ± 35% silt), followed by PP consider the method's efficacy or recovery rate variability.
(53 ± 27% sand; 40 ± 29% silt), PE (48 ± 29% sand; 34 ± 29% The Zobkov method consistently ranked the lowest in recovery per
silt), PET (43 ± 29% sand; 23 ± 25% silt), and PS (6 ± 9% sand; polymer and sediment type, but this is likely due to method constraints
8 ± 10% silt). Mean recoveries across methods were higher in sandier from size cutoffs. Many of the MPs tested for this study were smaller
sediment than silty sediments for each polymer, except for PS as the than the detection limit for this method (i.e. < 175 μm). Overall, the
small size of PS (40 μm) prevented high recoveries for all methods. In Fries method and Nuelle method had higher recoveries for most plastic
general, the quantitative ranking of polymer recovery followed the size- polymers in sand. The Fries method yielded the highest mean recoveries
ranking of each MP. That is, PVC and PP were the largest MPs for PET (59 ± 25%), PP (87 ± 9%) and PE (59 ± 35%) while the
(500–710 μm), and the most highly recovered from both sediments. PE Nuelle method ranked the highest for mean recoveries of PE
(96–106 μm) was the next highest recovered, followed by PET (62 ± 27%), PET (68 ± 20%), and PS (13 ± 12%). However, the
(250–500 μm). Although PET was classed as a larger particle than PE, Coppock method achieved a mean 86% recovery of PVC in sand, the
the fiber diameter (20 μm) in contrast to their long length made their second highest recovery of any polymer by any method. The Coppock
recovery more difficult. PS was generally the most difficult MP to re- method and Gilbreath method generally have the highest mean re-
cover and was also the smallest plastic studied (40 μm). There was no coveries of plastic polymers in silty sediments. In addition, the Coppock
recovery of PS from either the Zobkov (170 μm) or Gilbreath (45 μm) method had the highest mean recoveries for PE (53 ± 25%), PP
methods, as the initial sieve step for both methods removed smaller- (77 ± 16%), and PS (17 ± 12%). The Gilbreath method had highest
sized particles. mean recovery for PE (tied with Coppock, 53 ± 25%), PET
There was no statistically significant trend of quantitative ranking of (55 ± 22%), and PVC (76 ± 25%).
polymer recovery based on MP density. The ranking of density from
greatest to least (Table 2) was PET, PVC, PE, PS, and PP, whereas the 3.2. Comparison of recovery to a standard
ranking of mean recovery from greatest to least was PVC, PP, PE, PET,
and PS. However, it is challenging to draw comparisons among MPs When performing relative comparisons, we established a target goal
based on properties without noting that MP color and shape may also of ≥70% recovery as achievable and desirable. No singular method
affect recovery efficacy (i.e., colorful plastics are easier to see micro- effectively or consistently recovered > 70% of each polymer in either
scopically). There was no consistent pattern of quantitative ranking of sediment. The mean recovery was significantly greater than 70% in
polymer recovery based on method for either sediment type. More only two extractions from NAR sand. PP plastic was extracted with a
specifically, there was no recovery of PE from the Zobkov method, and mean efficiency of 87% ( ± 9%) using the Fries method and PVC was
no recovery of PS from the Zobkov or Gilbreath methods. extracted with a mean efficiency of 86% ( ± 11%) using the Coppock
In a quantitative ranking of methods based on mean percent re- method. This analysis indicates that less than 10% of the isolation and
covery, the Gilbreath method was the most successful at recovering extraction procedures meet the sandy sediment 70% standard and none
dense plastics (PVC and PET) from silty sediments. The Coppock of the procedures met the silty sediment standard. Had the standard
method was the most effective method for recovering light plastics (PP been set at the low value of ≥50% recovery, 40% of the isolation and
and PE) from silty sediments. The Nuelle method recovered the most extraction procedures have met or exceeded the standard for sandy
PET, PS, and PE from sandy sediments, as well as PS from silty sedi- sediment. For the silty sediment, the procedures meeting the standard
ments. Overall, the Zobkov method was found to be the least effective were approximately 30%.
for the isolation and extraction of our preselected microplastics. This is
likely due to the higher size fraction cutoff of their samples (45 μm 3.3. Variability associated with the methods
Gilbreath and 175 μm Zobkov). It should be stressed that these quan-
titative rankings are not method recommendations. The difference in Coefficients of variance (CV = (standard deviation/mean) ∗ 100)

5
M.A. Cashman, et al. Marine Pollution Bulletin 159 (2020) 111507

were calculated to measure the relative variability of the recoveries with its large grainsize and lack of organic matter consistently gener-
(Supplemental information Table 1). CVs ranged from 6 to 141%, in- ated better mean recoveries than the silty sediment. Silty sediment
dicating large variation in recovery of MPs. Mean recovery values of 0 samples consistently took longer to extract and had lower percent re-
were excluded from this analysis. For PET extracted using the Zobkov coveries. High sediment cohesion complicated procedural steps invol-
method, the CVs of 141% for both sand and silt indicated higher ving bulk sediment transfer, sediment suspension, and/or oven drying.
variability of percent recoveries compared with other polymers and As noted regarding the NR stain, moderate levels of organic matter add
methods. CVs for PVC, PET and PP were consistently lower in sand. CVs complexity to the plastic identification in the silty sediment. It is im-
for PS were consistently lower in silty sediment, and CVs for PE ex- portant to note the challenges of working with fine grained sediments
tractions were method dependent. In general, CV values for PP and PVC because they represent a large fraction of global sediment inventories
were lower than PE, PET, and PS. This suggests that the recoveries of PP (Thorp, 1937), particularly in low energy depositional environments
and PVC were more consistent with variability. Consequently, PVC and such as estuaries and protected bays where MPs and other anthro-
PP were on average, the most highly recovered MPs independent of pogenic contaminants will likely settle (Hume and Herdendorf, 1992;
sediment and method. Reineck and Wunderlich, 1968; Pettijohn and Ridge, 1932). In addition,
Several of the individual recovery replicates were greater than sediment property variation can inadvertently influence microplastic
100%. This highlights the important issue of MP fragmentation during visual identification. Plastics that mimic or are masked by sediment
isolation and extraction. Both the ground PVC and manufactured PE composition may be under-reported depending on the isolation and
beads were noted as highly friable. Methods that used abrasive mea- extraction method. Sandy beaches are distributed globally and re-
sures such as dry sieving likely caused these plastics to break down present an important aesthetic, recreational, economic and ecological
further and resulted in artificially high recoveries (i.e., > 100%). resource. Some of the methods compared here demonstrated con-
Several methods had consistent recoveries of 0%, especially for smaller siderable promise with sandy sediments.
sized microplastics. As previously noted, Gilbreath and Zobkov methods
had higher size cut-off ranges (45 μm and 175 μm, respectively) that 3.6. Recommendations
caused the loss of small MPs from sediments. None of the plastics tested
in this evaluation were greater than the 1 mm upper size threshold used The diversity of MPs and range of sediment matrixes may be too
by several methods. broad to standardize recoveries for isolation and extraction (e.g., 70%)
The variability in mean recovery is much larger in this evaluation with a single extraction procedure. While we saw positive aspects of
than the variability reported by each author's individual methodology each extraction method, we cannot make a recommendation for a single
validation in the scientific literature. For the four published methods, method that functions best for all sediment and microplastic types.
reported MP mean recoveries ranged from 70 to 100% (Nuelle Therefore, it is imperative that researchers first define what types
(91–99%), Fries (80–100%), Zobkov (85–99%), and Coppock (particularly size) of plastics they would like to quantify and how their
(70–100%)). The discrepancies between published recoveries and our environmental samples may affect the extraction process. In terms of
laboratory trials clearly highlight the influence of sediment matrix, and method efficacy, the Gilbreath method was the easiest to use with
MP properties of size, shape and density when reporting microplastic sandier sediments, whereas the Coppock method and Fries method
abundance in environmental samples. Standardization of isolation and were the easiest to apply with silty sediments. However, efficacy does
extraction techniques needs to be paired with explicit limitations of not reflect best percent recovery, nor does it describe the total number
recovery. Based on this comparison, it is unreasonable to assume that of nonpolymer particles (e.g., sediment, natural organic carbon) that
one method will extract all MPs from all matrices with the same level of remain on the final filters for polymer analysis. In these instances, la-
efficiency. As discussed above, statistical analyses indicated PP ex- boratories conducting polymer spectral analysis (e.g., Raman, FT-IR)
tracted from sandy sediment by the Fries method and PVC extracted after MP extraction need to perform organic matter oxidation to reduce
from silt by the Coppock method were the only two mean recoveries the number of particles on each filter. Oxidation steps do lengthen the
significantly greater than 70%. processing time, but they were crucial for sediment high in organic
matter. The biofilms were often removed in oxidation, and some MP
3.4. Effectiveness of Nile Red particles with surficial dyes lost some of their coloration. These points
may be notable if researchers are looking to identify MP surface char-
There was no statistically significant benefit to using NR to identify acteristics. The Zobkov method is an adaptation of the NOAA sediment
MPs on filters (Fig. 3). Initial investigations from Maes et al. used NR to method and would work well on larger MPs (Masura et al., 2015).
recover an average of 96.6% spiked MPs from various sediments (Maes Given our laboratory setup, we found the Nuelle method most chal-
et al., 2017). In our study, mean recovery was higher before NR staining lenging to replicate; however, the Nuelle method is the only one that
for PE (50.6 vs. 44.9%), PET (42.6 vs. 38.7%), and PP (52.9 vs. 44.9%). examined sediment sample masses in the1 kg range. The other four
Mean recovery of PVC (60.7 vs. 67.9%) and PS (14.8 vs. 15.5%) was methods were easier to perform partly due to their smaller sediment
higher after staining with NR (SI Tables 3; 4). NR did not uniformly mass. This method is advantageous for larger sediment samples.
stain the spiked plastics on each filter. This suggests the potential to Even though these five methods are a small fraction of the existing
miss certain MPs due to low stain uptake. Another major difficulty in methods in use, they represent distinct processes common to many
using NR to stain MPs came from the incidental false-positive staining methods. Readers may also find these results helpful in developing their
of organic debris such as benthic organisms and diatoms also present in own extraction methods, but we emphasize that these are only opinions
the final filter samples. The silty sediment's high organic carbon content of the laboratory researchers. MPs were frequently lost from the sam-
made identifying stained plastics particularly difficult, especially when ples in extraction steps that involved transferring the sample from one
differentiating smaller plastics such as PS and PE. These results suggest container to another. In addition, methods that limit the amount of
that the use of NR may confuse MP identification in high organic carbon sediment transfer are easier to perform. Many MPs were also observed
sediments rather than providing improved identification. as sticking to the walls of containers during density separation.
Therefore, extraction methods need multiple rinsing steps to ensure
3.5. Factors affecting method efficacy complete transfer. The fine grained sediment was consistently more
difficult to work with. Methods that included sediment drying are not
Based on this investigation, we suspect physical properties (i.e., recommended for silty sediments, as this resulted in sediment “bricks”
grain size and distribution, minerology and % carbon) play a significant that were difficult to sieve and further process. Wet sediment matrixes
role in microplastic extraction efficacy. For example, the beach sand with high amounts of fine silts and clay are often difficult to sieve and

6
M.A. Cashman, et al. Marine Pollution Bulletin 159 (2020) 111507

Fig. 3. Mean recovery of microplastics using visual spectroscopy prior to staining (“No Stain”) vs. after staining (“Nile Red”). “*” = p < 0.05 t-test. Error bars are
the standard deviations.

filter. We recommend removing the fine fraction of MPs, sediments, and MPs. The isolation and extraction of MPs from sediments is a crucial
organic matter from samples (< 45 μm) prior to analysis to greatly first step in the identification of MPs by polymer. Differences in MP
improve method efficacy. MPs smaller than 45 μm in size cannot be extraction procedures prevent meaningful comparisons across field
easily seen under current stereomicroscopes. While this investigation analyses. Further, differences in sediment matrix and MP properties can
did not look at polymer spectral analysis, many methods implore re- substantially affect extraction efficacy of MPs from sediments. The de-
searchers to transfer suspected MPs with tweezers off of filters and onto velopment of an internal standard composed of multiple types of MPs is
clean surfaces prior to spectral analysis. Researchers interested in urgently needed to allow standardization of MP extractions in marine
MPs < 45 μm might consider further delineating size fractions to help sediments.
with visual observations. A positive aspect of all five methods is that
salt solutions for density separation can be reused. For these studies,
CRediT authorship contribution statement
used salt solutions were filtered, reconstituted up to appropriate den-
sity, and re-filtered to cutdown on the purchasing of salts and genera-
Michaela Cashman:Conceptualization, Methodology,
tion of waste.
Investigation, Visualization, Validation, Writing - original draft.Kay
Potential observation bias must be documented when reporting data
T. Ho:Conceptualization, Investigation, Validation, Resources,
on MP abundance in sediments. It may be ultimately necessary to move
Writing - review & editing, Visualization, Project administration,
towards developing a MP internal standard for sediments to help
Funding acquisition.Thomas B. Boving:Validation, Formal analysis,
identify bias in MP isolation. Our recommendation is to develop a suite
Resources, Data curation, Writing - review & editing.Stephen
of MPs that are representative of, and span the types of MP particles
Russo:Validation, Investigation, Writing - review & editing.Sandra
(size, shape) of interest. This suite could be amended into environ-
Robinson:Validation, Investigation, Data curation, Writing - review
mental matrixes and extracted along with environmental MPs to esti-
& editing.Robert M. Burgess:Conceptualization, Validation, Formal
mate efficiency. This suite of MPs should be determined by the research
analysis, Data curation, Writing - review & editing, Visualization,
project objectives. Recovery rates for the internal standard MPs should
Project administration, Funding acquisition.
extrapolate the estimated recovery of environmental MPs isolated from
environmental samples. An internal standard would allow for better
standardization of data across environmental sampling and a better Declaration of competing interest
understanding of challenges posed by sediment matrices.
The authors declare that they have no known competing financial
interests or personal relationships that could have appeared to influ-
4. Conclusions
ence the work reported in this paper.
The comparison of five methods to extract varying MPs from two
sediment types indicate that method, sediment matrix, and plastic Acknowledgments
properties play substantial roles in the isolation of MPs from environ-
mental sediment matrixes. Sediments high in organic matter and with The authors appreciate the insightful comments on the draft
smaller grain sizes were generally more difficult to extract MPs from manuscript by the internal reviewers Dr. James Lake, Charlie Strobel,
and had lower mean recoveries when compared to MP recoveries from and Troy Langknecht. We also thank Ken Miller (General Dynamics
sand. In addition, most methods reviewed had higher mean recoveries Information Technology, Alexandra, VA, USA) for providing their sta-
for larger and low-density plastics. These findings highlight potential tistical analysis expertise. We would also like to acknowledge Dr.
biases in the current approximations of MP distribution in sediments Chelsea Rochman, Edie Guo, and Dr. Elke Fries for their technical ex-
worldwide. Further, the variability associated with each method was pertise. This work was performed while MAC was a participant of the
elevated with CVs ranging from 8% to 140% and 6% to 110% for the Oak Ridge Institute for Science Education program at the U.S. EPA's
silt and sand, respectively. These CVs suggest that larger MPs ORD/CEMM Atlantic Coastal Environmental Sciences Division
(> 500 μm) are easier and more consistently recovered than smaller (Narragansett, RI, USA).

7
M.A. Cashman, et al. Marine Pollution Bulletin 159 (2020) 111507

Appendix A. Supplementary data Quantifying Synthetic Particles in Waters and Sediments. US Department of
Commerce, National Oceanic and Atmospheric Administration,[National Ocean
Service], NOAA Marine Debris Program.
Supplementary data to this article can be found online at https:// Murphy, M., 2017. Microplastics Expert Workshop Report. US Environmental Protection
doi.org/10.1016/j.marpolbul.2020.111507. Agency.
Nel, H.A., Dalu, T., Wasserman, R.J., 2018. Sinks and sources: assessing microplastic
abundance in river sediment and deposit feeders in an Austral temperate urban river
References system. Sci. Total Environ. 612, 950–956.
Nor, N.H.M., Obbard, J.P., 2014. Microplastics in Singapore’s coastal mangrove ecosys-
Andrady, A.L., 2011. Microplastics in the marine environment. Mar. Pollut. Bull. 62 (8), tems. Mar. Pollut. Bull. 79 (1–2), 278–283.
1596–1605. Nuelle, M.-T., Dekiff, J.H., Remy, D., Fries, E., 2014. A new analytical approach for
Browne, M.A., Crump, P., Niven, S.J., Teuten, E., Tonkin, A., Galloway, T., Thompson, R., monitoring microplastics in marine sediments. Environ. Pollut. 184, 161–169.
2011. Accumulation of microplastic on shorelines woldwide: sources and sinks. Peng, G., Zhu, B., Yang, D., Su, L., Shi, H., Li, D., 2017. Microplastics in sediments of the
Environ. Sci. Technol. 45 (21), 9175–9179. Changjiang Estuary, China. Environ. Pollut. 225, 283–290.
Brydson, J.A., 1999. Plastics Materials. Elsevier. Pettijohn, F.J., Ridge, J.D., 1932. A textural variation series of beach sands from Cedar
Coppock, R.L., Cole, M., Lindeque, P.K., Queirós, A.M., Galloway, T.S., 2017. A small- Point, Ohio. J. Sediment. Res. 2 (2), 76–88.
scale, portable method for extracting microplastics from marine sediments. Environ. Prata, J.C., da Costa, J.P., Duarte, A.C., Rocha-Santos, T., 2019. Methods for sampling and
Pollut. 230, 829–837. detection of microplastics in water and sediment: a critical review. TrAC Trends Anal.
Fries, E., Dekiff, J.H., Willmeyer, J., Nuelle, M.-T., Ebert, M., Remy, D., 2013. Chem. 110, 150–159.
Identification of polymer types and additives in marine microplastic particles using Reineck, H.E., Wunderlich, F., 1968. Classification and origin of flaser and lenticular
pyrolysis-GC/MS and scanning electron microscopy. Environ. Sci. Process. Impacts 15 bedding. Sedimentology 11 (1–2), 99–104.
(10), 1949–1956. Rios, L.M., Moore, C., Jones, P.R., 2007. Persistent organic pollutants carried by synthetic
Gilbreath, A., McKee, L., Shimabuku, I., Lin, D., Werbowski, L.M., Zhu, X., Grbic, J., polymers in the ocean environment. Mar. Pollut. Bull. 54 (8), 1230–1237.
Rochman, C., 2019. Multiyear water quality performance and mass accumulation of Rummel, C.D., Jahnke, A., Gorokhova, E., Kühnel, D., Schmitt-Jansen, M., 2017. Impacts
PCBs, mercury, methylmercury, copper, and microplastics in a bioretention rain of biofilm formation on the fate and potential effects of microplastic in the aquatic
garden. J. Sustain. Water Built Environ. 5 (4), 04019004. environment. Environ. Sci. Technol. Lett. 4 (7), 258–267.
Hidalgo-Ruz, V., Gutow, L., Thompson, R.C., Thiel, M., 2012. Microplastics in the marine Shaw, D.G., Day, R.H., 1994. Colour-and form-dependent loss of plastic micro-debris from
environment: a review of the methods used for identification and quantification. the North Pacific Ocean. Mar. Pollut. Bull. 28 (1), 39–43.
Environ. Sci. Technol. 46 (6), 3060–3075. Shim, W.J., Song, Y.K., Hong, S.H., Jang, M., 2016. Identification and quantification of
Ho, K.T., Kuhn, A., Pelletier, M., McGee, F., Burgess, R.M., Serbst, J., 2000. Sediment microplastics using Nile Red staining. Mar. Pollut. Bull. 113 (1–2), 469–476.
toxicity assessment: comparison of standard and new testing designs. Arch. Environ. Stolte, A., Forster, S., Gerdts, G., Schubert, H., 2015. Microplastic concentrations in beach
Contam. Toxicol. 39 (4), 462–468. sediments along the German Baltic coast. Mar. Pollut. Bull. 99 (1–2), 216–229.
Hume, T.M., Herdendorf, C.E., 1992. Factors controlling tidal inlet characteristics on low Su, L., Cai, H., Kolandhasamy, P., Wu, C., Rochman, C.M., Shi, H., 2018. Using the Asian
drift coasts. J. Coast. Res. 355–375. clam as an indicator of microplastic pollution in freshwater ecosystems. Environ.
Hurley, R.R., Lusher, A.L., Olsen, M., Nizzetto, L., 2018. Validation of a method for ex- Pollut. 234, 347–355.
tracting microplastics from complex, organic-rich, environmental matrices. Environ. Tamminga, M., Hengstmann, E., Fischer, E., 2017. Nile Red staining as a subsidiary
Sci. Technol. 52 (13), 7409–7417. method for microplastic quantification: a comparison of three solvents and factors
Jambeck, J.R., Geyer, R., Wilcox, C., Siegler, T.R., Perryman, M., Andrady, A., Narayan, influencing application reliability. SDRP J. Earth Sci. Environ. Stud. 2 (2), 165–172.
R., Law, K.L., 2015. Plastic waste inputs from land into the ocean. Science 347 Thompson, R.C., Olsen, Y., Mitchell, R.P., Davis, A., Rowland, S.J., John, A.W.,
(6223), 768–771. McGonigle, D., Russell, A.E., 2004. Lost at sea: where is all the plastic? Science 304
Kowalski, N., Reichardt, A.M., Waniek, J.J., 2016. Sinking rates of microplastics and (5672), 838.
potential implications of their alteration by physical, biological, and chemical factors. Thorp, E.M., 1937. Description of Deep-Sea Bottom Samples From the Western North
Mar. Pollut. Bull. 109 (1), 310–319. Atlantic and the Caribbean Sea. vol. 3 University of California Press.
Lusher, A., Mchugh, M., Thompson, R., 2013. Occurrence of microplastics in the gas- Van Cauwenberghe, L., Claessens, M., Vandegehuchte, M.B., Mees, J., Janssen, C.R.,
trointestinal tract of pelagic and demersal fish from the English Channel. Mar. Pollut. 2013. Assessment of marine debris on the Belgian Continental Shelf. Mar. Pollut. Bull.
Bull. 67 (1–2), 94–99. 73 (1), 161–169.
Lusher, A.L., Burke, A., O’Connor, I., Officer, R., 2014. Microplastic pollution in the Van Sebille, E., Wilcox, C., Lebreton, L., Maximenko, N., Hardesty, B.D., Van Franeker,
Northeast Atlantic Ocean: validated and opportunistic sampling. Mar. Pollut. Bull. 88 J.A., Eriksen, M., Siegel, D., Galgani, F., Law, K.L., 2015. A global inventory of small
(1–2), 325–333. floating plastic debris. Environ. Res. Lett. 10 (12), 124006.
Maes, T., Jessop, R., Wellner, N., Haupt, K., Mayes, A.G., 2017. A rapid-screening ap- Vianello, A., Boldrin, A., Guerriero, P., Moschino, V., Rella, R., Sturaro, A., Da Ros, L.,
proach to detect and quantify microplastics based on fluorescent tagging with Nile 2013. Microplastic particles in sediments of Lagoon of Venice, Italy: first observations
Red. Sci. Rep. 7. on occurrence, spatial patterns and identification. Estuar. Coast. Shelf Sci. 130,
Mai, L., Bao, L.-J., Shi, L., Wong, C.S., Zeng, E.Y., 2018. A review of methods for mea- 54–61.
suring microplastics in aquatic environments. Environ. Sci. Pollut. Res. 25 (12), Zettler, E.R., Mincer, T.J., Amaral-Zettler, L.A., 2013. Life in the “plastisphere”: microbial
11319–11332. communities on plastic marine debris. Environ. Sci. Technol. 47 (13), 7137–7146.
Masura, J., Baker, J.E., Foster, G., Arthur, C., Herring, C., 2015. Laboratory Methods for Zobkov, M., Esiukova, E., 2017. Microplastics in Baltic bottom sediments: quantification
the Analysis of Microplastics in the Marine Environment: Recommendations for procedures and first results. Mar. Pollut. Bull. 114 (2), 724–732.