Marine Pollution Bulletin 122 (2017) 166–175

Contents lists available at ScienceDirect

Marine Pollution Bulletin

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

OSPAR standard method and software for statistical analysis of beach litter MARK
data
Marcus Schulza,⁎, Willem van Loonb, David M. Fleetc, Paul Baggelaard, Eit van der Meulene
a
AquaEcology GmbH & Co. KG, Marie-Curie-Str. 1, 26129 Oldenburg, Germany
b
Rijkswaterstaat, Ministry of Infrastructure and Environment, The Netherlands
c
The Schleswig-Holstein Agency for Coastal Defence, National Park and Marine Conservation, Schlossgarten 1, 25832 Tönning, Germany
d
AMO-Icastat, Niagara 18, 1186 JP Amstelveen, The Netherlands
e
AMO-Icastat, Deldenerstraat 81, 7551 AC Hengelo, The Netherlands

A R T I C L E I N F O A B S T R A C T

Keywords: The aim of this study is to develop standard statistical methods and software for the analysis of beach litter data.
Beach litter monitoring The optimal ensemble of statistical methods comprises the Mann-Kendall trend test, the Theil-Sen slope esti-
Assessment mation, the Wilcoxon step trend test and basic descriptive statistics. The application of Litter Analyst, a tailor-
OSPAR made software for analysing the results of beach litter surveys, to OSPAR beach litter data from seven beaches
MSFD
bordering on the south-eastern North Sea, revealed 23 significant trends in the abundances of beach litter types
Litter Analyst
Statistical software
for the period 2009–2014. Litter Analyst revealed a large variation in the abundance of litter types between
beaches. To reduce the effects of spatial variation, trend analysis of beach litter data can most effectively be
performed at the beach or national level. Spatial aggregation of beach litter data within a region is possible, but
resulted in a considerable reduction in the number of significant trends.

1. Introduction
Marine litter, here defined as anthropogenic waste dispersed in the
marine environment, has become a global threat to marine ecosystems.
Depending on its density, it occurs floating on the surface and in the
water column or settled on the seafloor and on beaches (Browne et al.,
2010). Litter can be washed onto a beach by the sea or directly de-
posited there by humans. Once on the beach, marine litter can, in part,
be moved away again by wave action and wind. It generally accumu-
lates as beach litter and a steady state between input and erosion is
reached within at least four weeks (Ribic et al., 2010, 2012; Schulz
et al., 2015). Most of the beach litter consists of plastic polymers, which
break down over a time span of several hundred years (Barnes et al.,
2009). Marine litter embodies a clear risk of harm to marine animals via
ingestion of mesolitter (size range 0.5–2.5 cm) and microlitter
(< 0.5 cm, van Franeker et al., 2011), and by entanglement in plastic
items, especially filamentous macrolitter (> 2.5 cm, OSPAR, 2010;
Derraik, 2002).
In 2008, the Marine Strategy Framework Directive (MSFD) of the
European Union (EU) was implemented (EU, 2008, 2010). According to
descriptor 10 ‘Marine Litter’ of the MSFD, the properties and quantities
of marine litter should not cause harm to the coastal and marine
environment. In order to evaluate marine litter pollution for the MSFD,
indicators have to be defined. In the North East Atlantic, at the OSPAR
and national levels, the amount of litter on beaches, “beach litter”, is
one of the indicators developed for this purpose. The level of litter
pollution and trends in abundance of litter items are monitored on
beaches by applying a standard OSPAR monitoring protocol (OSPAR,
2010) and are analysed using the statistical methods defined in this
study.
Worldwide, a number of different monitoring methods have been
used to assess beach litter, such as those by Alkalay et al. (2007), Bravo
et al. (2009), Cheshire et al. (2009), Opfer et al. (2012) and OSPAR
(2010). Alkalay et al. (2007) only recorded plastic items and monitored
transects perpendicular to the strandline. Abundances of plastic parti-
cles larger than 2 cm were chosen as basis for a simple beach litter
index. Bravo et al. (2009) randomly selected beaches along the Chilean
coast. These authors used densities per square meter as a litter indicator
and found plastic items and cigarette butts to be the most abundant
types of beach litter. Similar to the OSPAR method (2010), Cheshire
et al. (2009) selected survey beaches according to criteria, such as their
accessibility, a lack of beach cleaning activities and their neighbour-
hood to potential sources. Cheshire et al. (2009) recommend using flux
rates of 77 individual litter types as a measure of pollution, rather than

⁎
Corresponding author.
E-mail addresses: mschulz@uos.de (M. Schulz), willem.van.loon@rws.nl (W. van Loon), David.Fleet@lkn.landsh.de (D.M. Fleet), paul.baggelaar@planet.nl (P. Baggelaar),
amo@home.nl (E. van der Meulen).

http://dx.doi.org/10.1016/j.marpolbul.2017.06.045
Received 23 April 2017; Received in revised form 14 June 2017; Accepted 15 June 2017
Available online 21 June 2017
0025-326X/ © 2017 The Authors. Published by Elsevier Ltd. This is an open access article under the CC BY-NC-ND license (http://creativecommons.org/licenses/BY-NC-ND/4.0/).
M. Schulz et al.
Fig. 1. Map of the OSPAR region. Blue lines confine sub-regions. Red dots mark position of beaches monitored for beach litter. (For interpretation of the references to colour in this figure
legend, the reader is referred to the web version of this article.)
their abundances. According to their monitoring protocol, beaches are
surveyed every three months, while there is no clear definition of the
length of the beach, which is surveyed. The method of Opfer et al.
(2012) is based on a random selection of beach transects and the
abundances of 43 individual litter types. This method defines several
criteria for beach selection, such as substrate, accessibility, minimum
length and cleaning activities.
The most detailed protocol for monitoring beach litter is used by
OSPAR (2010). Since 2001, beach litter surveys have been carried out
on 100 m stretches of coastline in a total of 129 beach survey sites
bordering on the North Sea and the North-East Atlantic (Fig. 1). Surveys
are carried out at regular intervals of three months and abundances of
112 different litter types are recorded. All data are entered into a
central beach litter database. A selection of data from this database was
analysed by Schulz et al. (2013). These authors made a first proposal for
an evaluation system based on abundances and trends of individual
litter types and of general categories, such as total plastic litter. How-
ever, the necessity of establishing a standard method for the analysis
and assessment of beach litter data for the whole of the OSPAR region
and for the MSFD, required further development. An assessment system
for OSPAR beach litter data has to be robust against outliers, it must be
applicable to time series of data, which are partly non-normally dis-
tributed and it has to use an objective and unbiased selection of litter
types as input data. In addition, the length of the time series to be
analysed and descriptive statistical parameters, need to be defined.
Standard data analysis and assessment methods were proposed by the
OSPAR Intersessional Correspondence Group Marine Litter (ICG-ML).
OSPAR has agreed to use the proposed methods, which are described
here, as the standard for the analysis of the OSPAR beach litter dataset.
167
Marine Pollution Bulletin 122 (2017) 166–175
Software has been developed, which fulfills the requirements men-
tioned above, and which can be applied easily to the OSPAR beach litter
data format. The overall aims of this study were:
• To develop a transparent, effective, practical and efficient beach
litter assessment method for state and trend analysis of total abun-
dance of beach litter and the abundance of individual litter types,
for OSPAR and MSFD application,
• to use the OSPAR dataset for the further development of analytic
methods for beach litter data,
• to present a case study using German and Dutch beach litter data,
• to test different spatial aggregation levels for trend analysis,
• to select an optimized set of state and trend analysis methods and to
implement these statistical methods in the tailor-made beach litter
software, Litter Analyst.
The method was also developed to fulfil the requirements of the
MSFD. Information on trends in abundance of individual litter types
enables the effectiveness of measures targeted to reduce litter pollution
to be evaluated. In this study, a novel standard data analysis and as-
sessment method for OSPAR beach litter data and the software tool,
Litter Analyst, used to apply this method are presented. Litter Analyst
can be used to analyse the OSPAR beach litter data and any other data
relying on the same or a comparable monitoring protocol, which can
produce a suitable data input format. Beach litter data have been
analysed statistically on a number of occasions, partly for marine re-
gions as large as the OSPAR region (Ribic et al., 2010, 2012). However,
to the knowledge of the authors, Litter Analyst is the first statistical
assessment software dedicated to the analysis of beach litter data.
M. Schulz et al.
Fig. 2. Map of the study site, the Dutch and German North Sea coast. Filled circles mark the positions of seven beaches considered for assessment of beach litter data. The size of circles
corresponds to median total abundances of beach litter in the study period from 2009 to 2014.
2. Methods
2.1. Study area
The data analysed in this study was collected on Dutch and German
beaches situated in the south-eastern North Sea, which is part of the
OSPAR region. The North Sea is a shelf sea in north-western Europe,
which is connected to the Atlantic via the English Channel and the
Norwegian Sea. In the southern North Sea, major currents flow counter-
clockwise from southwest to northeast along the Dutch and German
coasts. Predominant westerly winds promote the transport of floating
and submerged litter from the English Channel along the Dutch coast to
the German Bight. In this part of the OSPAR region, beach litter mainly
originates from sea-based sources, such as fishing and shipping, while
tourism and river inflow contribute less to total abundances of beach
litter (Schulz et al., 2015). The main fraction of beach litter consists of
plastic polymers, which are mainly packaging, remains of fishing gear
and unidentified plastic fragments (Schulz et al., 2013).
The results of beach litter surveys from seven survey sites of 100 m
length on the Dutch and German North Sea coast (Fig. 2) for the period
January 1st 2009 to December 31st 2014 were selected from the OSPAR
beach litter database. The beaches were selected, because they are
composed of similar sandy substrates, have similar morphometry and
are exposed to the same water currents. They also all have a long and
consistent time series of beach litter data. The aggregation of their
monitoring data was thus justifiable. The beaches were situated on the
islands of Sylt (OSPAR beach DE1), Minsener Oog (OSPAR beach DE3),
Juist (OSPAR beach DE5) and Terschelling (OSPAR beach NL4), as well
as on the mainland coast near Bergen (OSPAR beach NL1), Noordwijk
168
Marine Pollution Bulletin 122 (2017) 166–175
(OSPAR beach NL2) and Veere (OSPAR beach NL3). While all OSPAR
beach litter data from 2002 to 2014 have been used for the optimization
of trend analyses, the German and Dutch data have been used to de-
monstrate the functioning of Litter Analyst.
2.2. OSPAR beach litter monitoring
Since 2002, regular surveys of litter items on beaches have been
carried out within the framework of the OSPAR convention on standard
100 m sections of beach on the coasts of the North-East Atlantic in-
cluding the North Sea (OSPAR, 2010). A detailed description of the
OSPAR monitoring procedures and their justification as well as results
of descriptive and analytical statistics of OSPAR beach litter time series
are given in Schulz et al. (2013).
The OSPAR beach litter database contains the results of all the
OSPAR beach litter surveys carried out since 2001. It can be accessed by
members of the OSPAR ICG ML at http://www.mcsuk.org/ospar/user
or upon request from the OSPAR secretariat. Prior to analyses, data
from beaches with an insufficient number of surveys for trend analysis
in the assessment period (< 15 surveys within 6 years) were removed
from the input data.
2.3. Statistical analysis methods
2.3.1. Evaluation of trend tests
The results of all OSPAR beach litter surveys carried out in the
period 2002 to 2014 have been used for the comparison of trend ana-
lyses. The following trend tests were evaluated using the OSPAR beach
litter dataset for the southern North Sea for the period 2002–2013: the
M. Schulz et al.
Mann-Kendall test (MK), the seasonal MK test (MKs), the autocorrela-
tion MK test (MKa), the seasonal and autocorrelation MK test (MKsa),
the linear regression test (LR), the seasonal LR test (LRs), the auto-
correlation LR test (LRa), and the seasonal and autocorrelation LR test
(LRsa). The most suitable test for the analysis of trends in the abun-
dance of beach litter recorded during OSPAR surveys was determined.
This was done for the abundance of the individual litter types, which in
total on average constitute 80% of all recorded litter items (top-80%) as
well as for the total abundance of all litter items. The potential problem
of ties (zero or equal values in a time series) was investigated for all
OSPAR beach litter time series.
2.3.2. Comparison of the results of trend analysis using different levels of
spatial aggregation
For the comparison of different spatial aggregation levels, data from
Dutch and German beach litter surveys from the period 2009 to 2014
were used. Three aggregation levels, individual beach, all national
beaches and all beaches in the southern North Sea region, were com-
pared.
At first, trends in the total abundance of all litter types and in the
abundance of individual litter types in the top-80% were analysed using
the Mann-Kendall and Theil-Sen tests. Beach litter data were then ag-
gregated at the national level for each quarter (i.e. four aggregated data
points per year), and again trends were calculated. Finally, all beach
survey sites in the southern North Sea, with at least 15 surveys per
period of six years, were aggregated. Data were only aggregated if
survey results were available for at least 75% of the survey sites in a
given time period. Trend slopes and significances were calculated and
compared for all three spatial aggregation levels, for the total abun-
dance of all items and for the individual abundances of litter types,
which make up the top-80%.
2.3.3. Relationship between slope and level of significance
For the analysis of the relationship between slope and significance,
data from Dutch and German beaches from 2009 to 2014 were used.
The relationship between the slope, relative slope and the trend sig-
nificance (p-value) was investigated for the five most abundant items
(plastic nets and ropes, plastic and polystyrene pieces, plastic caps and
lids, rubber balloons, plastic crisp/sweet packets) and three additional
items, which displayed slopes (small plastic bags, plastic [shopping]
bags & cotton bud sticks). These items were analysed for each beach,
and for aggregations of beaches at the national level. The p-values were
plotted against the slopes and the relative slopes. The exponential re-
lationships, for both the increasing and decreasing slopes, were fitted
into the data using Excel.
2.4. Statistical software Litter Analyst
Litter Analyst is a Windows programme designed for the statistical
analyses of OSPAR beach litter data. A full description of the software,
which was developed as a standalone Matlab routine, is given in its user
manual (see van der Meulen and Baggelaar, 2016).
Litter Analyst provides a selection of options for analysis. Different
time periods and different levels of aggregation of data can be chosen as
well as the number of litter types included in the top-80% list. This list
can also be supplemented with additional litter types considered to be
particularly harmful to the marine environment.
As a prerequisite for the aggregation of data from a number of
survey sites, each site has to have results from at least 75% of all pos-
sible surveys within a given time period. Litter Analyst also allows for
the setting of stricter levels of coverage i.e. 80% and 100%. If required,
the beaches can be weighted in order to adjust for sample bias due to an
uneven distribution of survey sites.
Litter Analyst establishes a list of the most abundant and harmful
litter types (top-80%) according to the following criteria: a) the
minimum percentage of total abundance, the list must represent, b) the
169
Marine Pollution Bulletin 122 (2017) 166–175
minimum number of litter types in the list, and c) the minimum number
of harmful litter types in the list. A list of litter types can be selected
manually from the litter types on the OSPAR survey protocol, if re-
quired. Harmful litter types were predefined by expert knowledge and
include litter types, which can easily be ingested or in which animals
can be entangled. Clustering (aggregation) of litter types is optional but
is necessary, when trend analyses include data from 2010 or preceding
years. This is because some OSPAR litter types were redefined or split
into two or more size classes in that year. Clustering allows for the
comparison of data sets before and after 2010.
Litter Analyst can read OSPAR beach litter data as csv-files or Excel
spreadsheet files. Litter Analyst calculates abundances for individual
litter types and for data aggregated at the levels of sources, material
categories and total abundance. Subsequently, Litter Analyst calculates
descriptive statistics, data density tables, trends and trend indices, and
generates output tables of statistical results for litter types, material
categories and sources. Graphs of trend plots and boxplots can also be
generated.
3. Results
3.1. Evaluation of trend tests
By exploring the statistical characteristics of a large number of
beach litter time series, it was found that no trend test is optimal for all
beach litter time series. However, the Mann-Kendall test showed the
best results for assessing trends in abundance of individual litter types
in 67% of the cases investigated and for trends in the total abundance of
all litter items in 48% of the cases investigated; linear regression is the
second best option with respective values of 16% and 30%. For trends
in total abundance, the performance of linear regression is improved,
probably due to an increased normalization of the distribution of total
abundance data compared to that of individual litter types. Therefore,
the two-sided Mann-Kendall test was selected as the standard method
for measuring trend significance, because this was found to be the best
for the majority of time series of abundances of individual litter types
and total abundance of all litter items. The Theil-Sen is the logical slope
estimation method, which is often used in combination with the Mann-
Kendall trend significance test.
Under the litter types, which make up the top-80%, 80% have <
20% zero-slopes, which is manageable when using the Mann-Kendall
test. For the remaining 20% of the litter types, the number of zero-
slopes is much larger, which probably impedes the correct functioning
of the Mann-Kendall test. These results show that it is important to use
the abundances of litter types, which make up the top 80%, in order to
obtain correct results for trend significance.
The slopes of the trends of the abundances of the five most common
litter types and of total abundance of all litter items for the seven
German and Dutch beaches are presented in Table 1. These results show
that, even for this set of survey sites, which in many respects are si-
milar, there is a large variation in the magnitude and direction of the
trend slopes for individual litter types and for total abundance. The
same large spatial variation is also visible at a national level, i.e. when
considering the Dutch and German beaches separately. These results
show that average aggregated values for a selection of beaches, which
are subject to similar physical conditions and litter sources can be used
as basis for assessments e.g. under the MSFD. This selection can be at a
national level, as in the Netherlands, but can also be at a sub-national
level i.e. in countries with an extensive coastline, such as Spain or the
United Kingdom. This emphasises the importance of having the spatial
aggregation function in Litter Analyst.
With increasing magnitude of relative slope, the trend significance
improved, as indicated by a lower p-value. This is very useful in-
formation for national MSFD assessment purposes, because it will be
important to be able to detect decreasing trends for individual litter
types, for which national reduction measures have been implemented.
M. Schulz et al.
Table 1
Demonstration of the large variation of magnitude and direction of slope of total abundance and the five most abundant litter types between the seven German and Dutch beaches in the
period 2009 to 2014. In each of the columns, the slopes are presented in order of magnitude and direction of slope. To improve readability, the beach names are not presented.
Total abundance Nets & ropes (OSPAR ID Plastic & polystyrene pieces (OSPAR ID
300) 301)
− 73.3 −13.7 −13.7
− 19.6 −4.7 0.2
2.3 −1.6 0.4
9.5 1.6 0.7
14.3 1.7 1.7
31.7 16.3 5.6
50.7 19.3 12.9
For the five most common litter types, results aggregated at a na-
tional level showed a higher r2 value. This is expected, because through
aggregation the variation within the data is partially averaged out. For
increasing slopes, aggregation does not improve the p-value at a re-
lative slope of 0.1. However, for decreasing slopes the p-value increases
by a factor of two, resulting in a p-value of 0.11 (89% confidence) for a
relative trend of − 0.1 (10% decrease per year). This result was ob-
tained by using the data from three to four beaches per national region,
corresponding to a spatial coverage of approximately one beach per
125 km. We propose that this could be a useful minimum requirement
for the number of beaches necessary to obtain the power to detect an
annual 10% decrease in abundance for individual litter types, with a
confidence of 89%.
3.2. Exemplary evaluation of results from the statistical analyses of Dutch
and German beach litter time series
In the period 2009–2014, the average and median total abundances
of all litter items varied considerably between the survey sites on the
southeastern coast of the North Sea. Median abundance ranged from 84
litter items on the beach of Juist to 421 on the beach of Veere. Dutch
beaches tended to have higher median total abundances than German
beaches, indicating a decreasing spatial gradient in the level of litter
pollution from west to east. In addition, beaches which faced to the
west had higher median abundances than other beaches (Fig. 2).
The composition of litter types was similar on all seven beaches.
Similarly, material composition differed little between beaches, and it
was dominated by plastic/polystyrene. Nets and ropes [300] were the
most abundant litter types recorded on all beaches, thus showing a
significant input from sea-based sources, such as fisheries. Caps and lids
[15], plastic and polystyrene pieces [301], as well as tangled nets [33]
were among the five most abundant litter types on all seven beaches.
Tables 2 and 3 give an overview of the statistical results for in-
dividual litter types and total abundance, obtained from Litter Analyst.
Trends in abundances of litter types were not consistent among the
survey sites analysed. For example, on the survey site on Veere beach
nine significant trends were identified, while on the beach of Ter-
schelling no significant trends could be detected. On the survey site on
Veere beach, only decreasing trends were detected. On the other six
beaches, 47 trends were increasing, and only 14 were decreasing. The
trend characteristics of data from the survey site on Veere beach are not
typical for the selected set of beaches.
The same litter types showed both decreasing and increasing trends,
depending on the beach. For example, nets and ropes [300] had in-
creasing trends on four beaches and decreasing trends on the other
three. Fig. 3 illustrates that national and sub-regional aggregation of
beach litter data partly resulted in a decrease in the number of sig-
nificant trends, which could be identified, and in the magnitude of
those trends. However, for the survey sites in the Netherlands no such
decrease in numbers of significant trends could be observed (Tables 2
and 3). In the data for individual litter items aggregated for all selected
sites on the Dutch and German North Sea coasts, which are part of the
170
Marine Pollution Bulletin 122 (2017) 166–175
Caps and lids (OSPAR Rubber balloons (OSPAR Crisp/sweet packets
ID 15) ID 49) (OSPAR ID 19)
−7 − 2.7 −4
−1.3 0.5 − 1.4
0.2 0.7 1.1
0.9 1.6 1.7
1.8 0 0
2.7 0 0
0 0 0
OSPAR Greater North Sea region, seven increasing trends and six de-
creasing trends were apparent, however, only one of these, a decreasing
trend in plastic bags [2], was significant.
Aggregations of material categories for survey sites showed seven
significant trends on the beach of Veere, while other beaches revealed
only few or non-significant trends (Fig. 4). Data aggregated according
to material categories revealed four significant trends in the Nether-
lands, none in Germany and one for the total of all survey sites. Total
abundance showed only one significant trend, which was on the survey
site on Veere beach (Table 3).
In the period 2009 to 2014, no general pattern in trends in the
amount of litter recorded on beaches in the south-eastern North Sea
could be recognised.
4. Discussion
The relationship between relative slope and p-value can be used to
demonstrate the power of the data from beach litter monitoring to
detect significant trends in the abundance of litter items on individual
beaches as well as in the aggregated data from a spatially coherent set
of beaches. For example, both the Dutch and German coasts can be
considered as spatially coherent beach sub-units, for which beach litter
data can be aggregated at a national level. Aggregations of beaches
according to well-defined sub-units are suitable for national and sub-
regional MSFD reporting. The German and Dutch results presented in
this study indicate that at least three or four survey sites are necessary
within a sub-unit, in order to obtain sufficient power to detect an an-
nual 10% decrease in the abundance of top-80% litter types, with 90%
confidence (p ~ 0.1). In the Dutch and German cases, one survey site
per 125 km of coast is present. These aspects can be used as a guideline
for the selection of the number of survey sites, which are required for
this indicator. However, further analysis of beach litter data from
countries where the coast is not as homogeneous as it is on the Dutch
and German North Sea coasts, is recommended to confirm this proposed
guideline.
The observed spatial gradient in median total abundances can
probably be attributed to the differences in the density of ship traffic in
waters adjacent to the beach litter survey sites in Germany and the
Netherlands. Thiel et al., 2011 identified the Street of Dover as a major
source of floating macrolitter in the south-eastern North Sea. This could
be the reason for the high values on the Dutch coast. Even higher total
abundances have been documented for litter on the Belgian coast
(Schulz et al., 2013), which is located close to the Street of Dover. In
addition, the Belgian coast is exposed to predominantly westerly winds
and currents, which would transport floating litter toward the coast. In
this study, beaches exposed to westerly winds and currents also re-
vealed higher median abundances.
The observed dominance of plastics in material composition is in
good agreement with previous studies in the North Sea and worldwide
(Barnes et al., 2009). Similar assemblages of litter types hint at similar
sources and transport pathways of beach litter in the south-eastern
North Sea, as was discussed by Schulz et al. (2015). Some of the most
M. Schulz et al. Marine Pollution Bulletin 122 (2017) 166–175

Table 2
Examples of the results of trend analyses of abundance for individual litter types calculated by Litter Analyst based on a) non-aggregated data from the selected seven beaches, b)
nationally aggregated data, and c) sub-regionally aggregated data for the period from January 1, 2009 to December 31, 2014. n gives the numbers of surveys. p-value gives the level of
significance.

Beach name/country/sub- Top-X litter Litter type definition Median Average % of total Trend (slope) p-value
region types [counts [counts counts [counts year− 1]
Litter type survey− 1] survey− 1]
code

Juist (Germany, n = 23) 300 Nets and ropes [300] 27.0 49.0 40.6% 1.7 0.566
301 Plastic polystyrene 4.0 12.7 10.6% 0.7 0.222
pieces < 50 cm [301]
15 Plastic: Caps [15] 2.0 7.4 6.1% 0.9 0.018
49 Rubber: Balloons [49] 2.0 5.3 4.4% 0.7 0.244
6 Plastic: Food [6] 2.0 4.4 3.6% 0.0 0.854
19 Plastic: Crisp [19] 2.0 3.9 3.2% 0.0 0.975
40 Plastic: Industrial [40] 3.0 3.6 3.0% 0.3 0.310
3 Plastic: Small bags [3] 3.0 3.5 2.9% 0.3 0.377
4 Plastic: Drinks [4] 1.0 2.4 2.0% −0.3 0.193
74 Wood: Other small [74] 0.0 2.1 1.8% 0.0 1.000
39 Plastic: Strapping [39] 1.0 2.0 1.7% 0.3 0.033
23 Plastic: Fertiliser [23] 0.0 2.0 1.7% 0.0 0.598
75 Wood: Other large [75] 1.0 1.7 1.4% 0.0 0.824
302 All cartons/tetra-paks [302] 1.0 1.6 1.3% 0.0 0.924
5 Plastic: Cleaner [5] 0.0 1.5 1.3% 0.0 0.717
Minsener Oog (island, Germany, 300 Nets and ropes [300] 31.0 46.9 39.6% −1.6 0.731
n = 24) 93 Glass: Other [93] 3.5 10.5 8.9% 1.0 0.200
33 Plastic: Tangled [33] 2.0 9.3 7.8% 1.0 0.006
301 Plastic polystyrene 3.5 5.8 4.9% 0.4 0.425
pieces < 50 cm [301]
48 Plastic: Other [48] 1.5 4.4 3.7% 0.7 0.046
90 Metal: Other large [90] 0.0 4.0 3.4% 0.0 0.506
91 Glass: Bottles [91] 1.0 4.0 3.3% 0.6 0.002
40 Plastic: Industrial [40] 1.0 2.6 2.2% 0.7 0.018
74 Wood: Other small [74] 0.0 2.4 2.0% 0.0 0.319
39 Plastic: Strapping [39] 2.0 2.3 2.0% 0.0 0.342
3 Plastic: Small bags [3] 1.5 2.2 1.8% 0.0 0.759
24 Plastic: Mesh bags [24] 0.5 1.5 1.2% 0.0 0.480
89 Metal: Other small [89] 0.5 1.4 1.2% 0.2 0.045
6 Plastic: Food [6] 0.0 1.2 1.0% 0.0 0.102
96 Pottery: Other [96] 0.0 1.2 1.0% 0.0 0.003
Sylt (island, Germany, n = 24) 300 Nets and ropes [300] 40.0 46.0 25.1% 1.6 0.713
301 Plastic polystyrene 13.5 23.0 12.6% 5.6 0.054
pieces < 50 cm [301]
15 Plastic: Caps [15] 9.0 12.3 6.7% −1.3 0.184
48 Plastic: Other [48] 4.0 12.0 6.6% −0.8 0.253
4 Plastic: Drinks [4] 5.5 8.0 4.4% 2.0 0.044
19 Plastic: Crisp [19] 1.5 5.1 2.8% 0.0 0.681
3 Plastic: Small bags [3] 2.5 5.0 2.7% 0.0 0.384
45 Plastic: Foam sponge [45] 3.0 4.9 2.7% −0.2 0.650
39 Plastic: Strapping [39] 4.0 4.6 2.5% 0.0 0.865
49 Rubber: Balloons [49] 3.0 4.6 2.5% 0.0 0.627
40 Plastic: Industrial [40] 4.5 4.5 2.4% 0.0 0.909
89 Metal: Other small [89] 0.0 4.4 2.4% 0.0 0.718
6 Plastic: Food [6] 1.5 4.3 2.4% 0.9 0.010
74 Wood: Other small [74] 2.0 4.1 2.3% 0.0 0.797
93 Glass: Other [93] 2.0 2.8 1.6% 0.0 0.932
91 Glass: Bottles [91] 1.0 2.8 1.5% 0.0 0.745
Bergen (The Netherlands, 300 Nets and ropes [300] 68.0 121.0 38.4% 19.3 0.056
n = 24) 301 Plastic polystyrene 35.5 51.3 16.3% 1.7 0.637
pieces < 50 cm [301]
49 Rubber: Balloons [49] 10.5 14.0 4.4% 1.6 0.066
15 Plastic: Caps [15] 11.0 12.7 4.0% 1.8 0.117
3 Plastic: Small bags [3] 9.0 12.5 4.0% 0.0 0.940
19 Plastic: Crisp [19] 8.0 12.4 3.9% 1.1 0.261
33 Plastic: Tangled [33] 7.5 9.7 3.1% 0.0 1.000
2 Plastic: Bags [2] 6.5 8.9 2.8% 0.1 0.765
45 Plastic: Foam sponge [45] 6.0 7.3 2.3% 1.8 0.008
40 Plastic: Industrial [40] 5.5 6.3 2.0% 0.5 0.241
98 San: Buds [98] 3.0 5.4 1.7% 0.2 0.437
4 Plastic: Drinks [4] 4.0 4.8 1.5% 0.0 0.599
22 Plastic: Cutlery [22] 3.0 4.8 1.5% 0.4 0.305
48 Plastic: Other [48] 1.0 4.8 1.5% 0.0 0.288
6 Plastic: Food [6] 3.5 3.5 1.1% −0.4 0.219
Noordwijk (The Netherlands, 300 Nets and ropes [300] 103.0 163.1 36.1% 16.3 0.126
n = 23) 301 Plastic polystyrene 68.0 94.3 20.9% 12.9 0.050
pieces < 50 cm [301]
15 Plastic: Caps [15] 19.0 24.7 5.5% 2.7 0.214
(continued on next page)

171
M. Schulz et al. Marine Pollution Bulletin 122 (2017) 166–175

Table 2 (continued)

Beach name/country/sub- Top-X litter Litter type definition Median Average % of total Trend (slope) p-value
region types [counts [counts counts [counts year− 1]
Litter type survey− 1] survey− 1]
code

49 Rubber: Balloons [49] 18.0 19.4 4.3% 0.5 0.633

19 Plastic: Crisp [19] 14.0 17.8 3.9% 1.7 0.095
45 Plastic: Foam sponge [45] 8.0 17.2 3.8% 4.4 0.002
3 Plastic: Small bags [3] 11.0 13.5 3.0% −1.4 0.158
33 Plastic: Tangled [33] 7.0 10.8 2.4% 0.2 0.750
98 San: Buds [98] 4.0 7.7 1.7% 0.6 0.276
48 Plastic: Other [48] 6.0 7.5 1.7% 1.1 0.024
4 Plastic: Drinks [4] 6.0 7.0 1.5% 0.0 1.000
2 Plastic: Bags [2] 6.0 6.8 1.5% −0.2 0.710
40 Plastic: Industrial [40] 5.0 6.1 1.3% 0.3 0.296
22 Plastic: Cutlery [22] 5.0 5.9 1.3% 0.4 0.472
6 Plastic: Food [6] 3.0 4.8 1.1% −0.3 0.439
Terschelling (The Netherlands, 300 Nets and ropes [300] 132.5 150.7 37.2% −4.7 0.812
n = 24) 301 Plastic polystyrene 50.5 61.1 15.1% 0.2 0.916
pieces < 50 cm [301]
15 Plastic: Caps [15] 16.5 23.4 5.8% 0.2 0.895
33 Plastic: Tangled [33] 11.5 17.9 4.4% −1.0 0.427
49 Rubber: Balloons [49] 14.5 16.8 4.1% 0.7 0.578
19 Plastic: Crisp [19] 9.0 15.0 3.7% −1.4 0.459
45 Plastic: Foam sponge [45] 7.0 13.1 3.2% 0.7 0.366
40 Plastic: Industrial [40] 9.0 11.7 2.9% 1.3 0.160
3 Plastic: Small bags [3] 10.0 11.0 2.7% −1.5 0.100
2 Plastic: Bags [2] 4.0 11.0 2.7% −1.0 0.310
4 Plastic: Drinks [4] 6.0 6.7 1.7% 0.5 0.352
39 Plastic: Strapping [39] 3.0 5.1 1.3% 0.4 0.363
6 Plastic: Food [6] 5.0 4.9 1.2% 0.0 0.611
48 Plastic: Other [48] 1.0 4.6 1.1% 0.0 0.477
98 San: Buds [98] 2.0 3.8 0.9% 0.0 0.537
Veere (The Netherlands, 300 Nets and ropes [300] 144.0 132.0 34.0% −15.1 0.078
n = 24) 301 Plastic polystyrene 71.5 63.9 16.5% −13.7 0.010
pieces < 50 cm [301]
15 Plastic: Caps [15] 23.0 27.2 7.0% −7.0 0.000
33 Plastic: Tangled [33] 11.5 16.5 4.3% −1.5 0.172
48 Plastic: Other [48] 11.0 15.0 3.9% −1.7 0.253
19 Plastic: Crisp [19] 10.5 13.7 3.5% −4.0 0.000
45 Plastic: Foam sponge [45] 7.0 13.7 3.5% −0.6 0.727
98 San: Buds [98] 12.5 12.5 3.2% −2.4 0.046
49 Rubber: Balloons [49] 11.5 12.0 3.1% −2.7 0.006
3 Plastic: Small bags [3] 4.0 6.3 1.6% −0.8 0.128
40 Plastic: Industrial [40] 3.5 5.9 1.5% −1.5 0.002
4 Plastic: Drinks [4] 4.5 5.4 1.4% −1.2 0.006
74 Wood: Other small [74] 2.5 5.0 1.3% −1.1 0.156
22 Plastic: Cutlery [22] 4.0 4.7 1.2% −1.2 0.035
2 Plastic: Bags [2] 2.5 4.1 1.1% −1.1 0.011
The Netherlands (n = 22) 300 Nets and ropes [300] 141.3 144.6 36.6% 5.3 0.463
301 Plastic polystyrene 63.3 68.5 17.3% 1.7 0.481
pieces < 50 cm [301]
15 Plastic: Caps [15] 20.8 22.1 5.6% −0.4 0.611
49 Rubber: Balloons [49] 14.0 15.5 3.9% −0.2 0.866
19 Plastic: Crisp [19] 14.6 14.9 3.8% −0.8 0.463
33 Plastic: Tangled [33] 12.3 13.8 3.5% −1.3 0.176
45 Plastic: Foam sponge [45] 12.8 12.7 3.2% 1.8 0.113
3 Plastic: Small bags [3] 9.0 10.9 2.8% −1.3 0.039
48 Plastic: Other [48] 6.5 8.0 2.0% 0.0 0.932
2 Plastic: Bags [2] 6.9 7.9 2.0% −1.4 0.004
98 San: Buds [98] 6.6 7.6 1.9% −0.5 0.381
40 Plastic: Industrial [40] 6.9 7.4 1.9% 0.4 0.352
4 Plastic: Drinks [4] 5.3 6.0 1.5% −0.4 0.183
22 Plastic: Cutlery [22] 5.0 4.7 1.2% 0.0 0.821
6 Plastic: Food [6] 3.9 4.3 1.1% −0.7 0.007
Germany (n = 20) 300 Nets and ropes [300] 37.5 48.4 34.6% −1.1 0.795
301 Plastic polystyrene 9.7 13.0 9.3% 2.9 0.074
pieces < 50 cm [301]
15 Plastic: Caps [15] 3.7 6.5 4.7% 0.3 0.495
48 Plastic: Other [48] 3.2 5.5 3.9% −0.1 0.922
93 Glass: Other [93] 2.7 5.1 3.7% 0.4 0.435
4 Plastic: Drinks [4] 3.0 3.7 2.6% 0.4 0.345
40 Plastic: Industrial [40] 3.3 3.6 2.6% 0.6 0.074
33 Plastic: Tangled [33] 1.3 3.6 2.6% 0.3 0.084
3 Plastic: Small bags [3] 3.0 3.6 2.5% −0.1 0.820
6 Plastic: Food [6] 2.3 3.4 2.4% 0.5 0.026
49 Rubber: Balloons [49] 2.5 3.2 2.3% 0.3 0.435
(continued on next page)

172
M. Schulz et al. Marine Pollution Bulletin 122 (2017) 166–175

Table 2 (continued)

Beach name/country/sub- Top-X litter Litter type definition Median Average % of total Trend (slope) p-value
region types [counts [counts counts [counts year− 1]
Litter type survey− 1] survey− 1]
code

19 Plastic: Crisp [19] 2.3 3.1 2.2% 0.1 0.579

74 Wood: Other small [74] 1.8 2.9 2.1% 0.4 0.059
39 Plastic: Strapping [39] 2.0 2.8 2.0% 0.0 0.844
91 Glass: Bottles [91] 1.3 2.5 1.8% 0.4 0.083
89 Metal: Other small [89] 0.4 2.3 1.6% 0.1 0.233
South-eastern North Sea 300 Nets and ropes [300] 104.7 101.6 36.0% 6.0 0.154
(n = 23) 301 Plastic polystyrene 42.6 43.9 15.6% 2.2 0.187
pieces < 50 cm [301]
15 Plastic: Caps [15] 14.3 15.2 5.4% −0.5 0.509
49 Rubber: Balloons [49] 8.8 10.0 3.5% 0.3 0.491
19 Plastic: Crisp [19] 9.4 9.6 3.4% −0.1 0.792
33 Plastic: Tangled [33] 9.3 9.6 3.4% 0.2 0.812
45 Plastic: Foam sponge [45] 7.6 8.1 2.9% 0.5 0.315
3 Plastic: Small bags [3] 7.3 7.7 2.7% −0.7 0.077
48 Plastic: Other [48] 5.1 7.0 2.5% 0.0 0.979
40 Plastic: Industrial [40] 5.6 5.7 2.0% 0.5 0.090
2 Plastic: Bags [2] 4.6 5.1 1.8% −0.8 0.019
4 Plastic: Drinks [4] 4.9 4.9 1.7% 0.2 0.492
98 San: Buds [98] 3.7 4.4 1.5% −0.1 0.751
6 Plastic: Food [6] 3.7 3.8 1.3% 0.0 0.915
39 Plastic: Strapping [39] 3.2 3.3 1.2% 0.0 0.874

abundant litter types can definitely be assigned to fisheries, a sea-based
source of litter. However, the sources of other common litter types re-
corded on southern North Sea survey sites e.g. plastic packaging have
not yet been identified with certainty. More advanced methods (see
Veiga et al., 2016) should be applied to the OSPAR beach litter data set
in order to gain more insight into their sources.
In this study, there was a pronounced lack of spatial patterns in
observed trends of abundance in litter on survey sites, as already de-
scribed by Schulz et al. (2013, 2015). These authors discuss the role of
wind drift and near-shore currents as driving factors of small-scale
spatial and temporal variation in beach litter data. Neumann et al.
(2014) also explained temporal and spatial heterogeneities in floating
macrolitter by the influence of wind fields, which also vary in time and
space.
This leads to the question of spatial representativeness of beaches,
and how trend results can be aggregated spatially. From a statistical
point of view, beach surveys should be performed at a number of new,
randomly selected, sites for each survey period, in order to ensure that
the data collected at those sites can be considered to be representative
for the region being assessed. However, this is neither convenient nor
workable for beach litter surveys for logistical reasons and because the
availability of suitable sites is limited. Therefore, aggregations of the
results from a selection of survey sites are only considered to be re-
presentative for that selection of sites and not for a regional or national
population of beaches. This implies that care should be taken in using
these results to assess the effects of measures implemented to reduce
litter pollution and to set environmental reduction targets.
Aggregation of data led to a considerable loss of significance and
magnitude of trends, because through aggregation, spatial variation
was clearly increased. At a single beach level, spatial variation is
eliminated, which means that temporal variation and possible trends
can be observed more precisely. This phenomenon was also observed by
van Franeker (2013). Therefore, it remains questionable, whether trend
analyses should be made on aggregated data on a sub-regional scale.
With the relatively small spatial distribution of survey sites in the
Netherlands, aggregation of the data from the four beaches still showed
many significant trends for the abundance of individual litter types and
total abundance. Small spatial-scale aggregation can, therefore, in some
cases, provide results, which are useful for MSFD assessments. The re-
sults of small spatial-scale aggregations of sites e.g. at a national level
could then be combined to produce an assessment for larger sub-regions
or regions.
The application of the Wilcoxon rank sum test for significant dif-
ferences in median abundances of six-year periods is justified for sev-
eral reasons: Firstly, subdivision into six-year periods with a minimum
of 4.5 years guarantees a minimum number of replicates and thus a
minimum level of statistical power for analyses. Secondly, six-year
periods appear to be sufficiently long to identify many trends in the
abundance of individual litter types on individual survey sites and in
aggregations of survey sites into sensible sub-units (see also Hougee and

Table 3
Summary of statistics of data on total abundance of beach litter calculated by Litter Analyst for a) non-aggregated data from the selected seven beaches, b) nationally aggregated data, and
c) sub-regionally aggregated data, for the period from January 1, 2009 to December 31, 2014.

Beach name/country/sub-region Number of surveys Median Average Standard deviation Coefficient of variance[%] Trend (slope) p-value
[counts survey− 1] [counts survey− 1] [counts survey− 1] [counts survey− 1]

Juist (island) 23 84.0 120.7 131.2 108.6% 2.3 0.832

Minsener Oog (island) 24 91.0 118.4 88.3 74.6% 14.3 0.205
Sylt (island) 24 175.5 182.8 78.0 42.6% 9.5 0.430
Bergen 24 249.0 315.2 223.0 70.8% 31.7 0.244
Noordwijk 23 368.0 451.7 334.0 74.0% 50.7 0.102
Terschelling 24 405.0 405.5 286.4 70.6% − 19.6 0.509
Veere 24 421.0 388.2 226.0 58.2% − 73.3 0.004
The Netherlands 22 369.3 395.4 104.7 26.5% − 1.7 0.866
Germany 20 140.5 139.8 60.8 43.5% 7.8 0.673
South-eastern North Sea 23 293.0 282.0 68.3 24.2% 6.1 0.561

173
M. Schulz et al. Marine Pollution Bulletin 122 (2017) 166–175

Fig. 3. Output of Litter Analyst: Examples of trend plots of the litter type Caps and lids [15] on the beach of Veere (left upper, slope = −7.0 year− 1, p < 0.001, n = 24), b) based on
aggregated data from the Netherlands (right upper, slope = −0.4 year− 1, p = 0.611, n = 22), c) based on aggregated data from Germany (left lower, slope = 0.3 year− 1, p = 0.495,
n = 20), and d) based on aggregated data from the south-eastern North Sea (right lower, slope = − 0.5 year− 1, p = 0.509, n = 23).

Fig. 4. Output of Litter Analyst: Examples of trend plots of the material category “plastic/polystyrene” a) on the beach of Veere (left upper, slope = − 64.0 year− 1, p = 0.009, n = 24),
b) based on aggregated data from the Netherlands (right upper, slope = − 1.3 year− 1, p = 0,977, n = 22), c) based on aggregated data from Germany (left lower, slope = 0.4 year− 1,
p = 0.871, n = 20), and d) based on aggregated data from the south-eastern North Sea (right lower, slope = 4.9 year− 1, p = 0.526, n = 23).

Boonstra, 2016). For example, in the UK and on the Iberian coast the
use of a number of sub-units is probably necessary. As Litter Analyst can
analyse the six-year reporting periods set down by the MSFD, it can be
applied in all areas in the European Union, where the OSPAR mon-
itoring protocol is used.
In general, non-parametrical tests, such as the Mann-Kendall test,
are robust against non-normal distributions and outliers and are,
therefore, applicable to most beach litter data (Schulz et al., 2013,
2015). Ribic et al. (2010, 2012) used generalized additive models, in
order to identify and quantify trends and sources of marine litter on the
U. S. Atlantic and Pacific coasts. Schulz et al. (2015) used multivariate
techniques, such as cluster analyses, to group beaches according to si-
milar trends in and similar degrees of pollution with litter. Both studies
succeeded in relating beach litter to its sources and identifying regions

174
M. Schulz et al.
of similar trends of litter types. In this study, the principal aims differed
from those of Ribic et al. (2010, 2012) and Schulz et al. (2015). Here,
more simple and robust non-parametrical methods were chosen, in
order to establish a standard method and analysis software for the
purposes of OSPAR and the MSFD.
Only a small number of litter types were clustered in order to ac-
count for changes in the OSPAR beach litter protocol in 2010, and only
litter types of the same materials and sources were clustered together.
Clustering is not necessary if only data after 2010 is used. These tech-
nical aspects and other characteristics of the tool make Litter Analyst a
robust, efficient and user-friendly software, which satisfies the demands
of OSPAR and the MSFD with regard to reporting and assessment.
In the MSFD, quantitative threshold values for beach litter have not
yet been defined. According to the results of this study, setting
threshold values for the abundance of individual litter types appears to
be most appropriate. Setting threshold values for overall trends in the
total abundance of beach litter will be ineffective, because the results of
analysis will, in most cases, be uncertain due to lack of significance.
Schulz et al. (2013) made a proposal for the definition of Good En-
vironmental Status (GES) for beach litter in the OSPAR region, based on
the degree of pollution rather than on trends. In addition to the demand
for the establishment of threshold values for litter on the coastline
(COMMISSION DECISION (EU) 2017/848) the European Commission
has proposed an aspirational target of reducing marine litter by 30% by
2020 for the ten most common types of litter found on beaches, as well
as for fishing gear found at sea (COM (2014) 398 final/2). In order to
achieve this target, measures are expected to be most effective, when
they are directed against pollution with single litter types or groups of
similar litter types, such as nets and ropes [300], which are attributable
to discarded or lost fishing gear. A statistical software for trend analyses
of beach litter at the level of litter types, such as Litter Analyst, would
be the software of choice, when assessing the fulfilment of a reduction
target of an individual litter type or groups of similar litter types. In
addition, different spatial aggregation options, which are included in
Litter Analyst, support surveillance of the success of measures, which
can be implemented at different spatial levels e.g. OSPAR or national.
As shown above, the non-parametrical trend analyses implemented in
Litter Analyst work well on the beach level and on national aggrega-
tions of beaches. However, this has still to be confirmed for other
countries with longer and less homogeneous coastlines than the coast of
the south-eastern North Sea. Moreover, as shown above, Litter Analyst,
with its ensemble of statistical methods, is well fitted to the demands of
the OSPAR Convention and the MSFD with regard to reporting.
5. Conclusions
A transparent, effective and efficient beach litter assessment method
for the analyses of state and trends of total abundance and abundance of
individual litter types was developed. The OSPAR beach litter dataset
was used for the development process and for illustrative case studies.
Three spatial aggregation levels for trend analysis were compared, and
it was found, that aggregation at a national scale can provide a number
of significant trends in abundance of individual litter types. Mainly non-
parametrical methods, such as the Mann-Kendall- test and the Wilcoxon
rank sum test were selected for an optimal ensemble of state and trend
analysis methods. These statistical methods were implemented in a
tailor-made beach litter software, Litter Analyst, and were successfully
applied to data from German and Dutch beach litter surveys. The
software Litter Analyst is an efficient and user-friendly software, which
can identify and quantify trends of abundance in individual litter types,
total abundances of all items and in aggregations of litter types such as
material categories. Potentially, Litter Analyst is also applicable to
other marine regions (e.g. HELCOM), if the same or a comparable
175
Marine Pollution Bulletin 122 (2017) 166–175
monitoring protocol for recording beach litter abundance is applied.
However, the software would require adaptation if the list of litter types
used for recording litter is different to the OSPAR list. The statistical
analyses used are agreed standard methods for assessments of marine
litter in the OSPAR region, and have been proposed to be used for as-
sessments in other European regions as well as for the MSFD. Litter
Analyst is designed to support the monitoring of success of measures
against pollution implemented under the MSFD.
Acknowledgements
We kindly acknowledge all persons actively involved in the OSPAR
monitoring of European beaches, especially the OSPAR ICG ML and all
persons performing OSPAR beach surveys.
References
Alkalay, R., Pasternak, G., Zask, A., 2007. Clean-coast index—a new approach for beach
cleanliness assessment. Ocean Coast. Manag. 50, 352–362.
Barnes, D.K.A., Galgani, F., Thompson, R.C., Barlaz, M., 2009. Accumulation and frag-
mentation of plastic debris in global environments. Philos. Trans. R. Soc. B 364,
1985–1998.
Bravo, M., de los Ángeles Gallardo, M., Luna-Jorquera, G., Núñez, P., Vásquez, N., Thiel,
M., 2009. Anthropogenic debris on beaches in the SE Pacific (Chile): results from a
national survey supported by volunteers. Mar. Pollut. Bull. 58, 1718–1726.
Browne, M.A., Galloway, T.S., Thompson, R.C., 2010. Spatial patterns of plastic debris
along estuarine shorelines. Environ. Sci. Technol. 44, 3404–3409.
Cheshire, A., Adler, E., Barbière, J., Cohen, Y., Evans, S., Jarayabhand, S., Jeftic, L., Jung,
R.-T., Kinsey, S., Kusui, E.T., Lavine, I., Manyara, P., Oosterbaan, L., Pereira, M.A.,
Sheavly, S., Tkalin, A., Varadarajan, S., Wenneker, B., Westphalen, G., 2009. UNEP/
IOC Guidelines on Survey and Monitoring of Marine Litter. Regional Seas Reports and
Studies No. 186. IOC Technical Series No. 83. (96 pp.).
Derraik, J.G.B., 2002. The pollution of the marine environment by plastic debris: a re-
view. Mar. Pollut. Bull. 44, 842–852.
EU, 2008. Marine Strategy Framework Directive 2008/56/EC of the European Parliament
and of the Council of 17 June 2008 establishing a framework for community action in
the field of marine environmental policy. http://eur-lex.europa.eu/LexUriServ/
LexUriServ.do?uri=OJ:L:2008:164:0019:0040:EN:PDF (accessed 28.2.2013).
EU, 2010. Commission Decision of 1 September 2010 on criteria and methodological
standards on good environmental status of marine waters (2010/477/EU). http://
eur-lex.europa.eu/LexUriServ/LexUriServ.do?uri=
OJ:L:2010:232:0014:0024:EN:PDF (accessed 28.2.2013).
van Franeker, J.A., 2013. Survey of Methods and Data Analyses in the Netherlands OSPAR
Beach Litter Monitoring Program. (Imares unpublished report, June 2013).
van Franeker, J.A., Blaize, C., Danielsen, J., Fairclough, K., Gollan, J., Guse, N., Hansen,
P.-L., Heubeck, M., Jensen, J.-K., Le Guillou, G., Olsen, B., Olsen, K.-O., Pedersen, J.,
Stienen, E.W.M., Turner, D.M., 2011. Monitoring plastic ingestion by the northern
fulmar Fulmarus glacialis in the North Sea. Environ. Pollut. 159, 2609–2615.
Hougee, M., Boonstra, M., 2016. OSPAR Beach Litter Monitoring in the Netherlands
2010–2015. (Annual Report). http://publicaties.minienm.nl/documenten/ospar-
beach-litter-monitoring-in-the-netherlands-2010-2015-annual-report.
van der Meulen, E.C.J., Baggelaar, P.K., 2016. Litter Analyst 3.0 User Manual. AMO/
Icastat, February 2016. (35 pp.).
Neumann, D., Callies, U., Matthies, M., 2014. Marine litter ensemble transport simula-
tions in the southern North Sea. Mar. Pollut. Bull. 86 (1–2), 219–228.
Opfer, S., Arthur, C., Lippiatt, S., 2012. NOAA Marine Debris Shoreline Survey Field
Guide. (14 pp.).
OSPAR, 2010. Guideline for Monitoring Marine Litter on the Beaches in the OSPAR
Maritime Area. OSPAR Commission (ISBN 90-3631-973).
Ribic, C.A., Sheavly, S.B., Rugg, D.J., Erdmann, E.S., 2010. Trends and drivers of marine
debris on the Atlantic coast of the United States 1997–2007. Mar. Pollut. Bull. 60,
1231–1242.
Ribic, C.A., Sheavly, S.B., Rugg, D.J., Erdmann, E.S., 2012. Trends in marine debris along
the U.S. Pacific Coast and Hawai'i 1998–2007. Mar. Pollut. Bull. 64, 994–1004.
Schulz, M., Neumann, D., Fleet, D.M., Matthies, M., 2013. A multi-criteria evaluation
system for marine litter pollution based on statistical analyses of OSPAR beach litter
monitoring time series. Mar. Environ. Res. http://dx.doi.org/10.1016/j.marenvres.
2013.08.013.
Schulz, M., Clemens, T., Förster, H., Harder, T., Fleet, D.M., Gaus, S., Grave, C., Flegel, I.,
Schrey, E., Hartwig, E., 2015. Statistical analyses of the results of 25 years of beach
litter surveys on the south-eastern North Sea coast. Mar. Environ. Res. 109, 21–27.
Thiel, M., Hinojosa, I.A., Joschko, T., Gutow, L., 2011. Spatio-temporal distribution of
floating objects in the German Bight (North Sea). J. Sea Res. 65, 368–379.
Veiga, J.M., Fleet, D., Kinsey, S., Nilsson, P., Vlachogianni, T., Werner, S., Galgani, F.,
Thompson, R.C., Dagevos, J., Gago, J., Sobral, P., Cronin, R., 2016. Identifying
Sources of Marine Litter. MSFD GES TG Marine Litter Thematic Report; JRC
Technical Report; EUR 28309. http://dx.doi.org/10.2788/018068.