A Global Inventory of Small Plastic Debris

Microplastic debris floating at the ocean surface can harm marine life. Understanding the severity of this harm requires knowledge of plastic abundance and distributions. Dozens of expeditions measuring microplastics have been carried out since the 1970s, but they have primarily focused on the North Atlantic and North Pacific accumulation zones, with much sparser coverage elsewhere. Here, we use the largest dataset of microplastic measurements assembled to date to assess the confidence we can have in global estimates of microplastic abundance and mass. We use a rigorous statistical framework to standardize a global dataset of plastic marine debris measured using surface-trawling plankton nets and coupled this with three different ocean circulation models to spatially interpolate the observations. Our estimates show that the accumulated number of microplastic particles in 2014 ranges from 15 to 51 trillion particles, weighing between 93 and 236 thousand metric tons, which is only approximately 1% of global plastic waste estimated to enter the ocean in the year 2010. These estimates are larger than previous global estimates, but vary widely because the scarcity of data in most of the world ocean, differences in model formulations, and fundamental knowledge gaps in the sources, transformations and fates of microplastics in the ocean.

Plastic debris has been documented in all marine environments, from coastlines to the open ocean (Barnes et al 2009), from the sea surface to the sea floor (Schlining et al 2013), in deep-sea sediments (Woodall et al 2014) and even in Arctic sea ice (Obbard et al 2014). The best-measured reservoir of plastic marine debris on a global scale is that of buoyant plastics floating at the sea surface. Yet observational data, even in the extensively surveyed Western North Atlantic Ocean (Law et al 2010) and Eastern North Pacific Ocean (e.g. Goldstein et al 2012, Law et al 2014), have not yet determined the full extent of large accumulations of debris associated with the converging surface currents in ocean subtropical gyres. In the Southern hemisphere gyres there are scarcely enough data to confirm the presence of floating plastic debris (Eriksen et al 2013, 2014, Cózar et al 2014), and the vast majority of the sea surface outside the gyres remains unsurveyed, introducing potentially large errors in global estimates of the amount of floating plastic.

Little is known about the transformations of plastics in seawater, including the time scales of degradation and its ultimate sinks. Weakened by UV radiation, chemical degradation, wave mechanics and grazing by marine life, plastics fragment into smaller and smaller pieces; plastic particles smaller than 5 mm in size are commonly referred to as microplastics. It has been suggested that plastic never fully degrades, yet expected increases in plastic concentration in response to increased production and use have not been consistently observed (e.g. Thompson et al 2004, Law et al 2010), and global budgeting exercises find less material on the ocean surface than expected (Cózar et al 2014, Eriksen et al 2014). To properly evaluate the risk of plastic contamination to marine organisms, understanding the amount, form and distribution of plastic in the marine environment, and how these evolve in time, is necessary. In this study, we focus on assessing the amount and distribution of 'small' (nominally <200 mm) plastic debris on the ocean surface, as these are by far the most sampled data set and also have demonstrated biological impact (Rochman et al 2015), although larger items can also impact biota.

At the sea surface, microplastic marine debris is typically measured by surface-towing plankton nets with mesh ranging from 0.1 to 0.5 mm, which capture particles limited to the size of the net aperture. Net tow sampling efforts typically capture plastic particles smaller than 10 mm in size (Morét-Ferguson et al 2010), while less numerous larger items are observed by visual surveys with ships or aircraft. The vast majority of observations since the 1970s have been made using plankton nets, with broadly similar sampling methodologies but variable reporting units (particle count per area or volume, or mass per area or volume). In contrast, visual surveys of macroplastic debris are conducted using a wide range of survey protocols ranging from (non-quantitative) opportunistic sightings to rigorous distance sampling methods (e.g. Williams et al 2011) for which it is difficult to satisfy all underlying methodological assumptions (Buckland et al 2001). In addition to the difficulty in reconciling different visual survey techniques (although useful reference standardized approaches based on distance sampling have been proposed, e.g. Ryan 2013), large debris is less numerically abundant than microplastics and its drift behavior and accumulation patterns are likely quite different because of its size, buoyancy and windage. Even though large debris accounts for a substantial mass of ocean plastics, for the reasons described above, we consider only data from plankton net trawls, which primarily collect microplastics in this analysis.

With the recent addition of relatively large and more geographically widespread datasets, and oceanographic numerical models that predict debris accumulation at the sea surface from surface current patterns, the first global estimates of the reservoir of floating plastic debris have recently been reported (Cózar et al 2014, Eriksen et al 2014). Using plankton net data (1127 trawls) spatially averaged in accumulation and non-accumulation zones defined from a statistical oceanographic model (Maximenko et al 2012), Cózar et al (2014) estimated between 7000 and 35 000 tons of floating plastic (0.20–100 mm in size) in the Atlantic, Pacific and Indian Oceans combined. Using a nearly independent plankton net dataset (680 trawls), Eriksen et al (2014) computed a global estimate of floating plastic (0.33–200 mm in size; 66 140 metric tons) using a different oceanographic model (Lebreton et al 2012) whose output was scaled by the globally measured plastic concentration. Given the methodological differences between these studies, it is encouraging that the resulting estimates are so close.

Here we estimate the global standing stock of small floating plastic debris with the most comprehensive dataset, ocean models and ocean plastic input estimates available. We compiled all available plastic data collected with surface-trawling plankton nets (more than 11 000 observations, including those in Cózar et al 2014 and Eriksen et al 2014), resolved sampling biases and other variations using a statistical model, and then used the standardized dataset to scale the outputs of three ocean circulation models. By comparing the three scaled model solutions, we assessed where debris patterns are well predicted and identified regions where discrepancies between solutions must be resolved through improved process description in models, additional oceanographic data collection and/or increased understanding of sources, composition, and lifecycle of plastic debris.

Plankton nets can capture any debris larger than the net mesh and smaller than the net mouth, but net dimensions vary between studies and maximum particle size is often not reported. Since most particles collected in plankton nets are millimeters in size or smaller, from here forward we use the term 'microplastics' not in its strict definition (as particles <5 mm in size), but instead to conveniently refer to all plastic debris collected in surface-trawling plankton nets.

There are two relevant measures for net-collected plastic debris: particle count and mass. Both have their merits. Samples are easier to count than to weigh, especially while underway at sea, and the number of particles may be more relevant for an exposure assessment. On the other hand, as a conservative variable, mass can more easily be related to source estimates, and will eventually be needed to close the mass balance of ocean plastics. Because of these considerations, we report both measures.

Plastic data collected using surface-trawling plankton nets were identified by literature search and data were assembled either directly from the publication or by contacting the corresponding author (table S1). Additional unpublished data were provided by contributing authors. In total 27 floating debris studies were identified, with 11 854 surface trawls carried out between 1971 and 2013, spanning all major ocean basins except the Arctic. Given the long time span over which samples were collected, we addressed sampling year as a potential bias when we standardized the data (see section 2.2). Net mesh ranged from 0.15 to 3.0 mm in size, although more than 90% of observations were collected using a manta net or neuston net with 0.333 or 0.335 mm mesh. Most studies did not report the maximum size of plastic debris collected. All data reported in units of #m?3 were converted to #m?2 by multiplying by the submerged height of the net, and then cast into units of #km?2. Nearly all studies reported plastic abundance in count units, and two-thirds reported data in mass units. However, the three largest datasets (comprising 82% of total observations) only reported counts. Conversions to mass for datasets in which only count was reported were made using factors derived from empirical data collected in similar geographic regions, during similar time periods and/or using similar sampling methods (table S1).

Microplastic abundance at the sea surface has been shown to vary with wind speed due to vertical mixing (Kukulka et al 2012, Reisser et al 2015), yet most studies did not report wind data. To evaluate the relationship between wind speed and plastic abundance as a source of variability in the data set, we used daily-averaged wind speed from the ECMWF ERA-Interim global atmospheric reanalysis (Dee et al 2011) interpolated to each surface trawl date and location. ERA-Interim output is available beginning 1 January 1979; thus, 222 surface trawls collected prior to 1979 were omitted from our analysis.

A major objective of this analysis is to inform the abundance and distribution of plastic debris in the ocean in order to ultimately assess marine animals' exposure to and impact from interaction with debris. Ours is the third study to estimate the amount and distribution of small floating plastic particles in the global ocean, using the largest dataset to date and three different ocean circulation models. While the previous two studies found coarse agreement in the global mass of plastics collected using surface-trawling plankton nets (7–35 thousand tons by Cózar et al 2014; 66 thousand metric tons by Eriksen et al 2014), our model solutions not only exceed these but also vary substantially from 93 to 236 thousand metric tons.

Despite the wide discrepancy in these standing stock estimates, all analyses find the highest concentrations of net-collected plastics in the subtropical gyres, with the largest mass reservoir in the North Pacific Ocean, presumably because of its vast area and also the large inputs of plastic waste from coastlines of Asia and the United States (Jambeck et al 2015).

To a considerable extent, our mass estimates may be larger than previously published estimates because of the data standardization used. Adjusting each observation forward in time to a common sampling year of 2014 and to no-wind sampling conditions increased the observed plastic concentrations in nearly all samples (figure S2). Previous studies have taken vertical wind-mixing of buoyant plastic debris into account by employing a simple one-dimensional model (Kukulka et al 2012) whose dynamics capture only a fraction of deep mixing observed (Brunner et al 2015). Certainly the variation in data collection (e.g., net mesh size); sample analysis (e.g., visual versus microscope identification); count-to-mass conversions (which are strongly dependent on particle size); and model design (e.g., source functions and removal processes) also contribute to the discrepancies.

The variation in model solutions in our study emphasizes that most of the ocean surface is undersampled for microplastics. Uncertainties in the Southern Hemisphere basins illustrate the lack of data even in high concentration subtropical gyres. The least sampled regions are areas of low plastic concentrations, where models predict between 30% and 70% of particles may reside (figure 6). Perhaps the starkest illustration is in the Mediterranean Sea, where models predict between 21% and 54% of global microplastic particles, equivalent to between 5% and 10% of global mass (because of small average particle size), are located. Our dataset has only 105 surface trawls concentrated in a very small region of the Western basin, whereas models predict the highest concentrations in the Eastern basin. One might expect to find very large plastic concentrations given the predicted large inputs of land-based plastic waste (Jambeck et al 2015) and the very long residence time of surface waters due to lack of exchange with the North Atlantic. Indeed, recent field data not included in this study confirmed very high mean surface concentrations in the Southern Adriatic Sea from 29 surface trawls (1.05 million particles km?2; 442 g km?2; (Suaria et al 2015)), yet more data, especially in the Eastern basin, is strongly needed.

Any global estimate of total accumulated floating microplastic debris is only on order of 1% or less of the amount of plastic waste available to enter the ocean annually from land-based sources. While these source estimates from Jambeck et al (2015) have relatively large uncertainties themselves (for example because they omit the tonnage of plastic locally burned, buried and recovered by self-employed wastepickers), it is hard to see their source and our floating stock estimates converge. While some of the 'missing' mass would be in plastic items larger than 200 mm (e.g. Eriksen et al 2014), and hence not included in our study, this is unlikely to account for the two orders of magnitude difference.

Importantly, however, there is no reason that standing stock estimates should equal an annual input estimate, especially since the input is of all plastic materials, not just those that float. Seafloor deposits of dense plastics, coastal deposits, and debris larger than typically captured in plankton nets are undoubtedly important reservoirs of plastic debris. In addition, standing stock reflects inputs and removal over time. The input rate is a function of not only the amount of plastic entering the ocean, but also of the rate at which these presumably large items fragment into the microplastics that surface trawls mostly collect. Removal processes are hypothesized (Law et al 2010), but their rates are essentially unknown. Multi-decadal time series of industrial resin pellets in the North Atlantic subtropical gyre and in North Sea seabirds indicate that removal can be quite rapid (van Franeker and Law 2015). Microplastics might fragment to as-yet undetectable sizes, sink due to buoyancy loss (Ye and Andrady 1991), be deposited on shorelines (McDermid and McMullen 2004), or be ingested and subsequently reduced in size (e.g., due to digestive grinding) and/or transported to land or the seafloor upon egestion. Biota represent the only other reservoir for which microplastic mass estimates exist. Myctophid fishes in the North Pacific gyre were estimated to hold 12–24 thousand metric tons of microplastic (Davison and Asch 2011), and the growing knowledge on ingestion of plastics by fishes (Kühn et al 2015) could imply a reservoir comparable in size to the sea surface.

The order-of-magnitude discrepancies in these global-scale budgeting exercises reveal a fundamental gap in understanding akin to the 'missing' anthropogenic carbon dioxide in the carbon budgeting exercise of the early 2000s (e.g. Stephens et al 2007). Until these discrepancies are resolved at even a coarse scale, we cannot quantify the full suite of impacts of plastic debris on the marine ecosystem.

Part of the article of:

Erik van Sebille1, Chris Wilcox, Laurent Lebreton4, Nikolai Maximenko, Britta Denise Hardesty, Jan A van Franeker, Marcus Eriksen, David Siegel, Francois Galgani and Kara Lavender Law1

Published 8 December 2015 ? ? 2015 IOP Publishing Ltd