Plastic debris is a contaminant of emerging concern that is often discussed in society, science, the media and policy1,2. It is visible to the naked eye and easily linked to our daily lives, which explains part of the public concern2,3. One size fraction of plastic debris is called microplastics, arbitrarily defined as particles smaller than 5 mm (refs1,4,5). Plastic particles smaller than 0.1 μm or 1 μm are often referred to as nanoplastics, although some recent reports put the lower limit for microplastics at 1 nm (refs6,7).

The vast majority of microplastics come from the breakdown of larger plastic waste. The diversity of sources is reflected in the heterogeneity of microplastic properties (shape, size, density and polymer type)810, transport characteristics1,11,12,13, and in vivo and in vitro biological effects — and therefore also in its risks14. The presence of contaminants in microplastic adds to this diversity15. Together with a high probability of being ingested and absorbed by a large range of species, this diversity in multiple dimensions has contributed to the concern that microplastics may constitute a risk to humans and the environment1,16,17,18,19,20,21,22.

Assessing such risks is challenging. Microplastics have been detected in air, soil, fresh water, drinking water, the oceans, aquatic and terrestrial biota, food products, and human placenta and stools1,19,23,24. Microplastic is both diverse and ill-defined. Even more uncertainties exist for assessing the risks of microplastics owing to the diversity of exposure and impact assessment methods used by scientists, and their inability to be compared with one another25,26,27. These challenges have been addressed in a growing body of literature1,2,4,8,9,10,11,15,16,17,18,25,27,28,29,30,31.

However, the way scientists currently look at microplastic is missing a critical nuance. Microplastic is generally viewed as an environmental contaminant with unique properties, one of which is its unparalleled complexity9,10. However, when each of its properties is considered individually, microplastic particles do not differ much from their natural counterparts. Nevertheless, researchers studying microplastics have focused mainly on the persistence profiles, fate processes and effect mechanisms of microplastic particles. By considering the complexity of microplastics in addition to that of natural particles, a more balanced picture of the risks posed by microplastics can emerge.

No risk assessment framework adequately examines the multidimensionality of microplastics while also taking into account the numerous types of natural particles present in the background, which together can encompass an enormous number of combinations of sizes, shapes, densities and chemical characteristics1,2,17,25. Natural particles should also be considered because they are much more ubiquitous than microplastics. Most organisms are therefore equipped to deal with natural particles32. Natural and microplastic particles that share characteristics can cause similar adverse effects, and differences in properties can help to explain differences in effects32.

In this Review, we survey the current tenets in microplastic research and emphasize progress in efficiently describing the diversity of microplastic particles. We compare the characteristics of microplastics to those of natural particles, and discuss the main mechanisms of their effects on organisms and the relevant dose metrics for these mechanisms. We describe how exposure and effect thresholds can be tuned and quantified using probability density functions (PDFs) that capture the diversity of microplastic particles in order to characterize risks consistently. Finally, we propose a framework to assess these risks through multiple simultaneous effect mechanisms, while accounting for the multidimensionality of the particles.

The multidimensionality of microplastic

Scientists have coined the term ‘environmentally relevant microplastic’ to refer to the plastic particles found in nature, including in the human diet. These differ substantially from the virgin plastic particles ‘from the shelf’ that are used in many laboratory experiments28,33. The difference between laboratory and nature needs to be addressed to understand the risks of microplastic for humans and the environment9,28. Scientists often portray environmental microplastic as a diverse and complex material, given its origins from a wide variety of materials and products1,2,4,8,9,29, and define it simply as ‘all plastic particles smaller than 5 mm’. Because this definition is vague, microplastic constitutes a heterogeneous mixture of polymers, sizes and shapes, and is associated with all kinds of chemicals1,4,8,10,17,29,34,35. For instance, polybrominated diphenyl ethers (PBDEs), phthalates, nonylphenols (NPs), bisphenol A (BPA) and antioxidants, all common additives in plastic products, slowly desorb into the environment when plastic items age and fragment36. At the same time, microplastic particles are passive samplers37,38, which means that any contaminant that has higher fugacity in the environment than in the plastic, will adsorb to microplastic until chemical equilibrium is reached39,40. Leached additives also re-adsorb to other microplastic particles in the environment, whenever their aqueous concentration is higher than the chemical equilibrium concentration41. Therefore, the often postulated distinction between desorbing additive chemicals and adsorbing environmental chemicals as separate categories36 does not truly exist. As for the particles themselves, once released into or formed in the environment, they undergo biofouling, weathering and ageing, and interact with chemicals, organisms and natural particles under a wide range of environmental conditions that are highly variable in time and space30,42.


The sources of microplastic vary widely, as plastics have a broad range of uses and applications. Primary microplastic refers to the micrometre-sized particles deliberately manufactured for specific applications or products, such as pellets for industrial production (for example, for drink bottles) and microbeads (such as those used in personal care products), that enter the environment43. Secondary microplastic refers to particles formed from the fragmentation and breakdown of larger plastic debris1,4,8. In the environment, these two classes ultimately become virtually indistinguishable as a result of the small proportion of primary microplastic44 and effects of weathering and ageing. Nevertheless, the diversity and complexity of sources continues to be reflected in the diversity of the material once it reaches the microplastic scale33.

Polymer composition and density

Microplastic polymer composition and thus particle density are a function of the polymers used in products, how much those polymers emit into the environment45, and the alteration of polymers during ageing and weathering in the environment42. Environmental altering involves processes such as photo-oxidation, embrittlement, crack propagation, abrasion, erosion, biodegradation, biofouling and aggregation. These processes affect such polymer properties as crystallinity, density and tensile strength42,46,47,48.

The most abundant polymers in microplastic are polyethylene (PE), polyethylene terephthalate (PET), polyamide (PA), polypropylene (PP), polystyrene (PS), polyvinyl alcohol (PVA) and polyvinyl chloride (PVC)17,31. However, in the environment, each of these discrete polymers has a certain range of densities, owing to differences in manufacture, crystallinity, additives, age and level of weathering and biofouling. This range ensures that the density distribution is essentially continuous for large numbers of particles9. Overall, the densities of microplastic polymers found in nature range between roughly 0.8 and 2 g cm–3, with a density of around 1 g cm–3 being most abundant9. An environmentally relevant microplastic particle on average has a weight of 12.5 μg, a volume of 0.011 mm3 and a density of 1.14 g cm–3 (ref.49).


Among microplastic characteristics, size spans the widest range. Microplastic particles include nanometre to millimetre sizes, a range of more than six orders of magnitude. The size distribution of microplastic particles has been shown often to follow a power law with a negative exponent, whose magnitude is determined by processes through which the particles are formed (by fragmentation) or removed (by erosion, size-selective transport or settling) from environmental media, including air9,19,33,50,51,52 (Fig. 1a). This implies that number concentrations increase dramatically with smaller size, which may have serious implications for the abundance of as-yet-undetectable nanometre-sized plastic particles53. Microplastic components in the human diet follow this size trend as well19.

Fig. 1: Relationships between processes and characteristics of environmental microplastic particles.
figure 1

a | Relation between size distribution and particle-formation processes. The formation of small particles through fragmentation and erosion of larger particles, in combination with the removal of large particles by erosion, size-dependent transport and settling, produce a higher abundance of smaller particles. b | Contour plot for width and length of environmentally realistic microplastic, showing how uptake of bioavailable microplastic (ingested) represents a fraction of the total exposure to environmentally realistic microplastic (exposure). Data are for the uptake of microplastic from sediment by Gammaruspulex9,13. c | Contour plots of longevity and size for microplastics compared with several categories of natural particles. Contours are plotted to guide the eye based on triangular distributions using reported median, minimum and maximum values (natural particles, as detailed in Table 1) or a power-law distribution (microplastics), based on refs9,33.


Fragments, fibres and films are the most frequently reported microplastic shape categories9,10,29,54. The occurrence of these shapes follows in part from product or material categories such as fibres, beads and films9,10,29. Primary dimensions such as length, width and height are similar for spherical particles, but can also differ greatly from each other, with very small heights for films, or very small heights and widths for fibres. Once in the environment, the categories remain recognizable as such, although fibre length or film area may reduce over time. Shapes that are irregular in all three dimensions, such as fragments, also remain in the same category upon further fragmentation, although change is random in each of the three dimensions, and fragments are far more difficult to trace to the original material.


Plastic is designed to be durable, and therefore the persistence and longevity of environmental plastic are very high. As the timescales of degradation of some polymers are much longer than what can be simulated in laboratory or field experiments, lifetime estimates must rely on extrapolations with considerable uncertainties. Extrapolated persistence data have been used to estimate half-lives ranging from 58 years for bottles to 1,200 years for pipes55. Under laboratory conditions, a particle of 1 mm in diameter would require about 320 years to reach a nanoscale diameter of 100 nm, based on particle shrinking from photo-oxidation and biodegradation at the polymer surface53. In the environment, however, degradation can be assumed to proceed much more slowly owing to the limited availability of oxygen, light and bacteria. Modelling of fragmentation and calibration of experimental weathering data found that fragments at the ocean surface and beaches have a 176-year lifespan52. It was calculated in 2015 that about 50% of the plastic in the oceans had been present for more than 13 years, whereas 80% and 90% of the plastic were older than 4 and 2 years, respectively41. Given these half-lives and timescales of 100 to 1,000 years, a major fraction of present-day environmental plastic is thus still in the early degradation stage. On the other hand, the fact that microplastics are widespread in the environment despite the production of plastics having started only 70 years ago implies that the formation of microplastics is quite rapid, with a timescale of at most decades.

Microplastic characteristics follow a continuous distribution

There are two approaches to define the characteristics of microplastic particles and their diversity. As in the discussion above, characteristics have traditionally been described using discrete categories, such as polymer type, shape category and size class10,17,29. This categorization is useful for tracking sources and origins of particles, and may be useful from a regulatory perspective29. However, it also makes microplastics look complex in an artificial way, owing to the high number of descriptors required. For instance, the category of polymer type would easily require more than 20 descriptors or parameters, shape would need more than 10, and size would also need more than 10, the latter depending on the accuracy required.

These predetermined categories are simplifications, because microplastic particles actually represent a continuum9. Although two microplastic particles will never be the same, the distributions of characteristics of large populations of particles will converge and become similar9,19,33,49,52. A PDF is a mathematical function that describes the actual distribution of a certain microplastic characteristic. A PDF is created by fitting the function to empirical data measured for a large number of microplastic particles in an environmental compartment of interest33. Size, shape and density can be captured in three PDFs with a total of twelve parameters describing the full diversity of environmental microplastics9. PDFs can be used to probabilistically quantify microplastic in the context of transport and fate modelling56; to rescale number concentrations obtained for limited size ranges to the full 1 to 5,000 μm microplastic size range; to convert number into mass concentrations; to assess exposure, effect and risk; and to quantify and visualize the bioavailability of microplastic19,33,49 (Fig. 1b). PDFs have been calibrated based on a meta-analysis of >60,000 microplastic particles, whose characteristics were measured with Fourier transform infrared (FTIR) imaging33. In this way, polymer-specific and environmental-compartment-specific PDFs can be obtained, which lend themselves to site-specific risk assessments.

Microplastic and natural particles

Natural particles are ubiquitous in our living environment32. Their numbers, compositions and characteristics are highly diverse and exceed those of microplastic particles (Table 1; Fig. 1c). Differences between natural particles relate to their chemistry, reactivity, physical density, geometry, size, persistence and abundance57. If particles of natural versus plastic origin nevertheless have similar characteristics, then comparable transport and fate processes — and thus toxicological effects — are also to be expected. Here we look at microplastic characteristics from the perspective of several major categories of natural particles, and compare their key features.

Table 1 Indicative medians and ranges of characteristics of microplastic and natural particles taken from various studies

Categories of natural particles

Inert minerals

The mineral particles occurring in sediments, soils, suspended solids or desert dust are sand, silt, silica, and clay, in order of decreasing size. They originate from prolonged chemical weathering of silicate-bearing rocks or locally from hydrothermal activity58. These minerals have a size range of about 60 nm to 2 mm (refs59,60,61,62), a density of 1.1 to 2.8 g cm–3 (refs60,63) and an environmental longevity of 104 to 109 days64,65,66. The concentrations in soils, sediment or suspended solids can range up to 99% by weight67,68. Minerals have a low affinity for hydrophobic organic contaminants69.

Organic matter

Non-living organic-matter particles in sediments, soils and aquatic systems include decomposing algae, detritus and natural fibres. Detritus is composed of organic compounds originating from the remains of organisms such as animals and plants and their waste products. Detritus particles have a size range of roughly 200 nm to 2.5 mm (refs60,70), a density of 0.8 to 1.2 g cm–3 (refs60,71) and an environmental longevity of about 20 days up to 10,000 days72. Organic-matter contents of soils, sediment or suspended solids can range from very low (<1%) to several tens of per cent67,73,74. Studies suggest that 80–90% of oceanic fibres are actually of natural origin75,76,77. This includes purely natural fibres as well as man-made (textile) cellulosic fibres, which are difficult to differentiate78. Data that include both marine and freshwater samples suggest that 75% of the fibres are of natural origin33 and show that natural and microplastic fibres have a similar length distribution (Fig. 2). The affinity of hydrophobic organic contaminants for organic matter is high, with partition coefficients of up to 107 litres kg–1 (ref.69).

Fig. 2: Comparison of lengths for natural and microplastic fibres.
figure 2

Minimum, median and maximum lengths for natural fibres (55, 120 and 6,090 μm) versus microplastic fibres (55, 120 and 7,470 μm), obtained from six water types. A minimum diameter of 15 μm was assumed, implying a detection limit for length of 15 \(\times \) 3 = 45 μm, based on a minimum aspect ratio of 3. Data from ref.33.

Black carbon

Environmental condensed carbon or black carbon is the collective term for a range of carbonaceous substances, encompassing partially charred plant residues (char, charcoal) to highly graphitized soot, produced from incomplete combustion of biomass and fossil fuel in the absence of oxygen79,80,81,82,83. Black carbon is of mixed natural (natural wildfires, vulcanism) and anthropogenic (industry, traffic, domestic fires) origin. The particles have a size range of about 1 nm to 200 μm (refs84,85,86), a density of 0.13 to 2.1 g cm–3 (refs87,88) and an environmental longevity of 105.4 to 107.6 days89,90. Soots, comprising the smallest size fraction of black carbon, can contain aggregated fractions of single-walled carbon nanotubes, multi-walled carbon nanotubes and multi-concentric fullerenes, rendering engineered organic nanomaterials part of the black carbon space91,92. Black carbon particles are abundant in the air and make up on average 3% and 9% of organic carbon in soils and freshwater sediments, respectively80,86,88, with extremes of up to 40% at specific locations80. They can also be a significant component (up to 20%) of sediment organic carbon in the remote ocean93. The affinity of hydrophobic organic contaminants for black carbon is high, with partition coefficients two to three orders of magnitude higher than those for organic matter69,80,81,85.

Microplastics as a phase in particulate matter, sediments, suspended solids or soils

The above natural particles exist in nature in the form of mixtures and aggregates. They are composite particles that we often refer to with functional names such as aerosols, dust, sediments, soils, particulate matter, suspended solids or settling solids. The physical characteristics of these natural particles are somewhere in between those of their individual components, with the range of the characteristics determined by the physical characteristics and relative proportion of the components. Like other particles of anthropogenic origin, microplastics can be considered as a new artificial phase of composite natural particles94,95. This raises the semantic question of whether one should speak of ‘the sediment plastic fraction’, or of ‘the extent to which the actual sediment is diluted by plastic’.

Natural particles as a proxy for microplastics

Microplastic particles can be transported in air and water, are subject to vertical mixing owing to their small size and can settle in a manner similar to that of these natural solids1,4,8,11,18,19,96,97. Because the continuum of characteristics for natural particles overlaps with that of microplastic9,33 (Fig. 1c), natural particles may serve as a proxy for microplastics, and models simulating the transport of such natural particles can form the basis of transport models for microplastic11. In natural waters, for instance, several processes drive the transport of natural and microplastic particles11, including turbulent transport, aggregation, biofouling, settling, resuspension and burial. Microplastics rapidly acquire a biofilm and become captured in low-density aggregates or flocs. This implies that the transport of the plastic-inclusive floc or aggregate becomes virtually indistinguishable from that of a fully natural floc or aggregate. The extent to which the aggregation affects the fate and transport of microplastic or microplastic-embedded composite particles is an open question94.

Many transport modelling studies assume spherical particles or aggregates11. A sphere has the smallest possible surface area for a given particle volume or size, and thus constitutes only a very small fraction of the actual shape distribution of microplastic and natural particles. The PDFs discussed above provide opportunities for more realism in transport models with respect to shape.

Microplastics have a distinctive combination of characteristics

Here we compare the key characteristics of environmental microplastics with other categories of particles11 (Fig. 1c). It appears that some particle categories can have a similar size to microplastics, but then have a higher density (minerals, sand, silt, clay, metal-based nanoparticles and colloids). Other particle categories can have similar density, but then are far less persistent (organic-matter flocs, detritus, algae, detritus or organic colloids) than microplastics. Yet other particle categories do not exist in a nanometre-to-centimetre size range, or in a fibre-to-sphere range of shapes, with all other properties being similar to those of microplastics.

In summary, it is the combination of low density (often close to that of water)4,8,9,10,29, high persistence52,55, wide size range4,8,9,10,29,51,53 and variable shape4,8,9,10,29,33,54 that makes microplastic particles unique. However, apart from their manufactured polymer composition, none of the individual features of microplastic is unique by itself. Microplastic sizes overlap with those of minerals, organic matter and black carbon particles, and densities are similar to those of organic matter and black carbon. Microplastic longevity in the environment is high, of an order of magnitude similar to that of inert persistent minerals such as clays and black carbon (Fig. 1c). Natural fibres, spherical soot black carbon particles and film-like flake graphite exist, so these shape categories are not unique to microplastic. The sorption affinity of hydrophobic organic contaminants for microplastic is similar to that of organic matter but orders of magnitude lower than that of black carbon80,85,98,99. It is often said that chemical leaching from microplastic particles distinguishes them from natural particles. However, black carbon and soot particles are highly contaminated and therefore also leach chemicals99,100,101. The same applies to aquatic sediments contaminated with legacy compounds that leach chemicals into a cleaner overlying water column102. In other words, leaching is not an essential difference, and given the abundance of contaminated sediment and black carbon compared with microplastic masses in the environment, the fluxes of organic chemicals leaching from natural particles will often overwhelm those of additives leaching from plastics in many habitats15,103.

Microplastics, nanoparticles and nanoplastics

In addition to the natural microparticles described above, smaller nanoparticles are also present in the environment. Research in the field of engineered nanoparticles has provided a picture of the main similarities and differences between microparticles and nanoparticles104. Here, we briefly reflect on the main differences between micrometre-sized and nanometre-sized environmental particles.

Important categories of environmental nanoparticles are organic and inorganic natural colloids, soot, the submicrometre size fractions of polydisperse particles such as black carbon80 and desert dust aerosols61,62, engineered nanoparticles32 and nanoplastics53. At present, the relative contribution (abundance) of nanoplastic particles to this complex mixture is unknown.

Nanoplastics can be expected to differ from microplastics with respect to their transport and fate, aggregation with natural colloids, surface area and reactivity104,105, chemical affinity, and adsorption and desorption rates for sorbed chemicals98,106. Hydrophobic organic contaminant sorption affinities for nanoplastic particles are much higher than those for microplastics and approach those for soot and black carbon85,98. However, this information is largely based on laboratory studies using synthesized spherical nanoplastics. Apart from polymer identity, the actual characteristics of nanoplastic particles in the environment remain unknown107. Common engineered nanoparticles include metal nanomaterials such as Ag or Au, stable oxides such as TiO2 or CeO2, iron oxides and carbon nanotubes105,108. Nanoplastics are distinguished from these engineered nanoparticles by their primary origin and particle heterogeneity, which is likely to be higher for nanoplastics. Metal-based engineered nanomaterials have a higher density than nanoplastics, whereas densities and environmental persistence of carbon-based nanoparticles (fullerenes, carbon nanotubes) are likely to be in the same range. Some metal-based nanomaterials are subject to dissolution processes. For nanoplastic, it is not clear to what extent dissolution, particle shrinking, further fragmentation and/or degradation processes contribute to removal53, which constitutes a major research gap.

Interactions of microplastic with biota

Owing to the multidimensionality of microplastic, the mechanisms for uptake, bioaccumulation and adverse effects, as well as the dose metrics to quantify these effects, are diverse. Systematic research is needed to decipher these mechanisms28.

Uptake and bioaccessibility

Many effect mechanisms require microplastic to be ingested, and microplastic ingestion has been demonstrated in the laboratory and field for a wide range of species109,110,111,112,113,114,115,116,117,118. The ingestion of microplastics by aquatic biota depends on their bioaccessibility, which is mostly determined by particle size, the species-specific characteristics of the exposed organisms, such as mouth opening or resistance to translocation, and the environmental conditions19,49,119. The formation of biofilms can affect bioaccessibility by increasing the size and density of microplastics and modifying their shape. For example, biofilms were found to be able to promote particle uptake under some conditions120,121, but to decrease it when the biofilm increased particle size and promoted aggregation122. After ingestion, microplastics can be transported along the digestive tract until excretion, or they can accumulate in the gut123, the digestive gland124, the gills123,124 or the liver123 of some organisms. Cell internalization of microplastics in the digestive system125 and their translocation from the gut cavity to the bloodstream126 have been demonstrated in the marine mussel Mytilus edulis and the transport of microplastics along the circulatory system has been shown in the marine clam Scrobicularia plana124. Digestive processes can change microplastic characteristics: the Antarctic krill Euphausia superba127 and the amphipod Gammarus47 were found to turn microplastics into nanoplastics through digestive fragmentation. Besides being ingested, microplastics can adsorb to the surface of microalgae128, aquatic plants129, cnidarians130 and crustaceans131.

Adverse effects

The ingestion or adsorption of microplastics can cause adverse effects on aquatic organisms16. At the population level, the presence of microplastics can reduce the number of species or their biomass132,133,134. At the individual level, microplastics can affect survival134,135,136,137, reproduction135,137,138,139,140,141, growth13,122,128,134,138,141,142,143,144, feeding130,131,139,145, emergence146, embryonic development147, mobility148,149 and photosynthetic efficiency128. At the sub-organismal level, microplastics can cause increased oxygen consumption145,150, inflammation125,151, reduced lysosomal stability in the digestive gland125, reduced antioxidant capacity124, DNA damage16,124, neurotoxicity124, oxidative damage124, gut dysbiosis152,153, alteration of the genetic expression136, ionic exchange150 and enzymatic activity149. The mechanisms leading to these effects are often unknown, but many studies have speculated that microplastic-induced physical damage or reduced feeding may contribute13,121,122,128,130,134,135,136,137,138,139,140,141,142,144,148. Microplastics can cause physical damage by adsorbing and aggregating on the surface of microalgae128, restrict movement by accumulating in tentacles130, or promote satiation and reduce food assimilation by blocking the food passage121,133,134,135,137,139,141,143,144,147. Other studies have attributed the effects to specific plastic properties, such as the surface groups28,31,35,36,135,138,142,143, or the leaching of toxic chemicals from the microplastics136,140,145,147. The severity of the effects vary depending on the properties of the microplastics, their concentrations and the exposure time.

The weight of evidence of these reported effects and effect mechanisms has been analysed30,154, including a quantitative analysis that also took the quality of studies into account28. The four most relevant effect mechanisms, in order of decreasing weight of evidence, were food dilution (inhibited food assimilation or decreased nutritional value); internal physical damage; external physical damage; and, with lower certainty, oxidative stress28. Consequently, these mechanisms should take priority when assessing the risks of microplastic particles.

Differences between microplastic and natural particle effects

Because microplastic particles are present among numerous inert and degradable natural particles, microplastic and natural particle effects are likely to occur together. Although there are calls to address the complexity of microplastic as a mixture of plastic-based particles, conceptually, the broader array of particle types present should also be considered. After all, for protecting the habitat quality of both aquatic and terrestrial organisms, it is relevant to address risks from mixtures that also contain other man-made and low-calorific-value natural particles. Often these particles can have similar effect mechanisms and toxicological profiles, given that they have similar combinations of particle, particle-associated chemical, surface area or size-related toxicities155,156,157,158,159,160,161.

In general, studies comparing microplastics with natural particles (for instance, red clay, kaolin or natural sediment) under controlled settings show that microplastic-induced adverse effects occur at lower concentrations than for natural particles13,135,159,162,163,164, although some of these studies lack confirmation of quality assurance and control (QA/QC)28. That microplastic effects occur at lower concentrations does not necessarily imply that they cause effects when mixed with natural particles, because natural particles can be more abundant. Because most of the published comparisons between microplastic and natural particle effects concern clays or natural sediments, a clear research gap exists as to how other abundant and hazardous microparticles compare, such as micrometre-sized soot and black carbon particles165. Thus, we argue that the implications of microplastic particles in the biosphere should not remain fully separated from those of other, natural particles.


Although many uncertainties remain, years of particle toxicity research have provided a picture of the relationship between particle characteristics and toxicity, and of the main similarities and differences between microparticles and nanoparticles32,104,166,167,168,169,170,171,172,173 (Table 2). However, for nanoplastic particles, our knowledge of the relationship between particle characteristics and toxicity is based on laboratory studies with fabricated particles, mainly submicrometre-sized polystyrene spheres. For environmentally realistic conditions, this relationship is difficult to address, because as mentioned, apart from polymer identity107, toxicologically relevant characteristics of environmental nanoplastic particles (such as shape, submicrometre-size range, area, volume, surface chemistry, biopersistence and zeta potential32,104,166) are unknown.

Table 2 Key differences between microparticles and nanoparticles in the context of toxicity

Risks of microplastic particles

Current status of risk assessments

Although assessments of exposure to microplastics for humans exist19,174,175, there has not been an assessment of the risk posed by such exposure, nor of the risk to terrestrial ecosystems. However, risk assessments have been performed for microplastic particles in the aquatic environment, based on a comparison of exposure concentrations and threshold effect concentrations14,16,17,49,176,177,178,179,180,181. Exposure concentrations were taken from either measured literature data or modelled data176,177. Because of the different analytical methods used, these exposure data are often difficult to compare and have limited quality25,49,54. Similar limitations with respect to comparability and quality exist for effect assessment methods28,49, which in combination seriously limits the reliability of the resulting risk characterizations.

The reported risk assessments used species sensitivity distributions (SSDs), which aim to determine the affected fraction of a series of species at a given concentration182. SSDs increase the relevance of the assessment for the community level, and are used to obtain the hazard concentration for 5% of the species (HC5) (Fig. 3). Ideally, SSDs use the effect threshold values for one well-defined type of stressor and end point (one type of harm) for more than ten different species while environmental variables are kept constant. However, such effect threshold data are not yet available for microplastic particles. Even though ‘real’ microplastic represents a continuum of particle types, laboratory tests have either used monodisperse particles that are all ingested28, or particles with wider size distributions, where larger particles may have disrupted the intake or effects of the smaller particles or had more limited bioavailability. Microplastic particles can also trigger responses through different modes of action (different types of harm)28, suggesting that the HC5 values obtained from SSDs remain ambiguous with respect to identifying the relevant particles and their associated effects.

Fig. 3: Interactions of microplastic with biota.
figure 3

a | Plastic continuum. Erosion and fragmentation of plastic objects leads to an increasing number of ever smaller microplastic particles that are bioavailable to an increasing group of species. b | Species continuum. As soon as microplastic particles become bioavailable, they may trigger interactions with a variety of species. For instance, larger particles may be bioavailable through ingestion, but not to the smallest organisms. Smaller particles can also be bioavailable to small organisms, through ingestion but also through translocation. c | Species sensitivity distribution (SSD). The interactions between the bioavailable part of the plastic continuum, and the species continuum, may lead to species-specific effects. Threshold effect concentrations for these species-specific effects can be assessed for a range of species and combined in an SSD for environmentally relevant microplastic. Hazardous dose metric concentration for 5% of the species (HC5) provides input for risk assessment. d | Hazardous concentration in particles per litre for 5% of the species (HC5) present in an aquatic community with error bars relating to the 95% confidence intervals or 25–75 interquartile ranges14,180, according to Everaert et al. (2018)176, Besseling et al. (2019)16, Skåre et al. (2019)179, Adam et al. (2019)14, Zhang et al. (2020)178, Everaert et al. (2020)177, Koelmans et al. (2020)49, Adam et al. (2021)180 and Jung et al. (2021)181. Later studies take more data points into account. Median HC5 for these nine studies is 75.6 particles per litre.

Strategies to improve risk assessment

Harmonizing the setups between studies, strictly adhering to quality criteria and using environmentally realistic microplastic mixtures in tests may solve these problems of limited comparability and quality. This, however, also implies that much of the work has yet to be done. As an alternative to redoing all of these tests from scratch, microplastic PDFs can be applied to rescale and align currently available exposure and effect threshold concentration data, including those used in SSDs9,19,33,49. The ambiguity of multiple causation — that is, effects originating from a range of mechanisms — can be addressed by first identifying the effect mechanism, and then aligning and quantifying the relevant exposure and effect metrics. For instance, upon identifying ‘food dilution’ as a relevant microplastic effect mechanism, the corresponding dose metric (ingested volume), as well as the exposure and effect threshold data with respect to that dose metric, can be aligned and quantified49. This mechanism-specific quantification of effects is a well established concept in toxicology and should be applied to microplastic particles, including to distinguish between the particles’ physical effects and particle-associated chemical effects2,183. Besides application in risk characterizations based on SSDs, this concept can be used in prospective risk assessments based on food-web modelling184.

Risks from chemicals

Plastic-associated chemicals

In addition to the physical effects and risks of microplastic particles, microplastics can simultaneously cause chemical effects if chemical exposure is high enough. Chemicals such as additives and sorbed environmental pollutants are abundant on microplastic particles in the environment1,17,30,34,36,39,185, and are often referred to as ‘plastic chemicals’185 or ‘plastic-associated’ chemicals15. These terms are imprecise because phase distribution ensures that the same chemicals are also present in other environmental media or particles, such as soil, sediment, water, organic particles, algae, black carbon or detritus15,17,35,37,38,41,85,100,102,103. Depending on exposure concentrations, any of these chemicals from any such media can pose a risk to biota. For instance, microplastic can be a significant source of hazardous chemicals through water as an exposure medium, leading to mortality under environmentally relevant conditions. Acute mortality in coho salmon could be ascribed to a tyre-rubber-derived chemical that was present in the water at toxic concentrations186.

The microplastic vector effect

It has long been hypothesized that desorption of chemicals from microplastic after particle ingestion may increase exposure to these chemicals, and subsequently lead to chemical risks to biota (the ‘microplastic vector effect’). However, the microplastic vector effect is now thought unlikely to play a major part in most habitats, and no risk has been demonstrated to date15,17,38. There are two main reasons for this conclusion. First, most empirical studies considered chemical exposure only through ingestion of microplastics, neglecting parallel uptake pathways. This overestimates the relative importance of microplastics as a vector for transport and uptake, but underestimates the total exposure to chemicals15,35. When all uptake processes are taken into account, uptake through microplastic ingestion is seen to be minor15,41,187. This also means that testing the chemical toxicity of microplastic particles, ignoring simultaneous exposure to the same chemicals through other media, is less useful when trying to mimic environmental realism. Second, most studies also used an unrealistic chemical concentration gradient between plastic and biological receptors. For example, desorption of plastic-associated chemicals was studied in a clean animal test system, or effects were tested in vitro using extracts obtained at unrealistically high plastic-to-water ratios and small extract volumes188,189. Because additives are subject to significant dilution in the natural environment, which is not mirrored in such tests, any observed toxicity does not directly indicate whether or not a risk occurs188. It is true that at ‘hotspot locations’, ingestion of high concentrations of microplastic particles may occur. However, given the typical chemical dilution in most environments, uptake from other media, the low digestibility of plastic, the abovementioned ‘food dilution effect’ and the lack of a chemical concentration gradient under environmentally realistic conditions, it is plausible that an organism would starve before it would exceed a chemical toxicity threshold.

Dealing with chemical and particle effects

The above does not mean that chemical effects of ingested particles will always be minor compared with physical particle effects. After all, the chemical effects and thus potential risks are highly context-dependent, influenced by the composition of the chemical mixture, chemical concentrations, the directions of chemical transfer, presence and co-exposure through water, through natural organic particles or through prey organisms15. The chemical and physical toxicity components of microplastic hazards are best assessed separately2,15, following the principle of separate dose metrics and effect mechanisms, to be combined only at the last stage of the assessment33.

A new microplastic risk assessment framework

Several scientists have introduced risk assessment frameworks for microplastic particles2,16,20,183,190. These provided general strategies and recognized the challenges that come with the complexity of microplastic. However, the frameworks did not yet include the theory and tools required to meet these challenges. They simplified the actual diversity of microplastic by using categories and bins for particle properties and/or chemical content, thereby neglecting the considerable variation that exists within categories2,16,183,190. Second, they did not implement a strategy to account for the limited quality of data or the fact that the data cannot be compared2,16,183,190. Third, one proposed framework did not include an effect threshold component, neglected dose dependency of particle and chemical effects, and inaccurately distinguished between additive and environmental chemicals190. Because scientific progress in the past 3 years has tackled many of these challenges, we can now define a generic microplastic risk assessment framework that applies to either the environment or to human health. Three new elements determine the framework: use of PDFs to describe toxicologically relevant particle characteristics, so that no simplification with categories is necessary; use of QA/QC screening methods to evaluate whether exposure and effect data are fit for purpose; and use of a calculation framework to assess exposure to plastic-associated chemicals through all relevant pathways. Here, we provide the outline for a new approach to consistently account for the multidimensionality of microplastic.

The risk assessment needs three levels of information (Fig. 4). First, steps are taken on a process level, which includes defining the problem, designing exposure and effect assessments, risk characterization and quality assurance and control for the screening of input data2,16. Second, based on a particular problem definition, mechanisms of adverse effects on biota should be identified. Because microplastic is a diverse mixture of contaminants, it causes effects simultaneously through different effect mechanisms9,33. Each of these mechanisms is linked to an ecologically or toxicologically relevant end point, and is quantified by the ecologically or toxicologically relevant dose metric (ERM or TRM) for that mechanism2,33,49. Third, because dose metrics are relevant to only part of the continuum, they need to be extracted from the exposure data covering the full continuum33,49 (Fig. 5).

Fig. 4: Risk assessment scheme addressing the multidimensionality of microplastics.
figure 4

Step 1: problem definition (blue box). A problem definition is given on the basis of a protective objective (for example, protecting a population). This guides the design of a hazard and exposure assessment, taking into account harmonization of methods and quality assurance and control (QA/QC) to select input data that are fit for purpose. Step 2: dose metric for mechanism of effect (yellow box). Based on selected mechanisms of effect, ecologically or toxicologically relevant dose metrics (ERM or TRM, respectively) are defined, such as mass, surface area, volume and aspect ratio. Subsequently, exposure concentrations and effect threshold concentrations are assessed for these metrics, leading to an exposure profile and a stressor-response profile. Step 3: aligning dose to microplastic continuum (red box). For each of the ERMs or TRMs, the profiles from step 2 apply only to the bioavailable fraction of the microplastic continuum. Thus, they must be expressed in terms of the full (from 1 to 5,000 μm), environmentally realistic microplastic continuum. This conversion is done using probability density functions (PDFs). Finally, for each ERM or TRM, actual exposure and effects thresholds are compared in a risk characterization33,49. In this way, a risk is calculated for each of the individual effect mechanisms.

Fig. 5: The concept of simultaneously acting effect mechanisms.
figure 5

A heterogeneous mixture of plastic particles can initiate effects through different mechanisms acting simultaneously. For each mechanism, an ecologically or toxicologically relevant dose metric is defined. These are, for example, particle surface area, specific surface area, aspect ratio (length to width ratio) or particle volume (not shown). For each of the ecologically or toxicologically relevant dose metrics, a probability density function (PDF) is defined, based on the best available data on the characteristics of environmentally relevant microplastic particles9,33,49. Finally, using these PDFs, exposure data are translated into the ecologically or toxicologically relevant dose metrics for a risk characterization for each of the effect mechanisms.

Step 1: problem definition

The problem to be defined — often the protection of a population or an individual flagship species — comes from the regulatory domain. From there, exposure and effect assessments are carried out, which rely on QA/QC and harmonization of analytical techniques and effect test methods28,54. Exposure and effect assessments have benefited from advances, particularly in automated FTIR imaging techniques33,191, and the cost and effort required for these measurements can be expected to drop as soon as they become mainstream. A crucial requirement for microplastic risk characterization is that both exposure and effect assessments relate to the same ecologically or toxicologically relevant dose metric2. The gold standard would be to test with a consistent environmentally relevant diverse microplastic mixture, but data correction and alignment methods can bridge the differences in the current data33,49. For retrospective risk assessments, quantitative quality assurance and criteria evaluation tools have now been developed28,54. These tools allow the risk assessor to select data that are fit for the purpose of the assessment, which is crucial when the assessment is intended to have regulatory implications.

Step 2: dose metric for the mechanism of effect

As mentioned above, food dilution, internal damage and oxidative stress are plausible microplastic particle effects28,154. Relevant dose metrics then need to be selected for these mechanisms33. For instance, volume is a relevant dose metric for a food dilution mechanism28,192, and aspect ratio is relevant in cases where internal damage correlates to fibre toxicity32,170,171,193. Surface area or specific surface area are appropriate for endpoints such as oxidative stress32,167,168,169. When potential risks arise from chemicals or pathogens bound to either the surface of plastic or absorbed by its bulk, concentrations in terms of surface area and mass, respectively, are suitable metrics15. Environmentally realistic mixtures of particles are likely to cause effects through multiple mechanisms simultaneously, implying that there are multiple relevant dose metrics33. We note that some of these metrics have boundaries to account for bioavailability. For instance, if particles are too large to be ingested, food dilution would apply only to the ingestible bioavailable fraction of the microplastic continuum28,49. Cellular-level mechanisms like oxidative stress require translocation, so only much smaller particles would be bioavailable33,49 (Table 2). Bioavailability is a well-established concept in risk assessment for traditional chemicals, and is relevant for microplastic as well.

Step 3: aligning dose to microplastic continuum

Once the relevant effect mechanism and dose metrics are known, they need to be aligned to exposure data2,16,17. Well-established procedures are available for risk assessment of chemicals, which have been extended to chemicals absorbed to microplastic particles2,15,37,41,103. To align particle effects to exposure data, exposure needs to be expressed in parameters such as microplastic volume, aspect ratio, (specific) surface area or mass33,49. Often mass provides a fair proxy for volume, given the proportionality between the two, as the densities of environmental microplastic particle mixtures are close to 1 and less variable than other microplastic characteristics such as size and shape13,49. The various relevant exposure and dose metrics can be quantified using the above-mentioned microplastic PDFs, obtained prospectively9 or retrospectively19,33 using existing data. However, given the diversity of microplastic, the need to take multiple effect mechanisms and thus dose metrics into account is the rule rather than an exception, and requires a thorough characterization of the microplastic continuum. State-of-the-art focal-plane-array FTIR imaging provides the necessary detail to obtain such PDFs33,194, although otherwise existing PDFs for environmental microplastic can be used in the assessment33. For instance, detailed PDFs for volume, area, mass, aspect ratio and specific surface area have been provided from a meta-analysis of characteristics of individual particles sampled from different environmental compartments33. This represents a promising approach to linking microplastic doses and characteristics to effect mechanisms and thresholds. A risk characterization is then performed for each of the effect mechanisms separately (Fig. 4), after which the risk assessor can use a precautionary approach by basing the conclusion about the risk on the metric for which the highest risk has been calculated.


Particles are ubiquitous in our natural environment, and microplastic particles are a relatively recent subcategory. Owing to their anthropogenic origin, continued plastic production and fragmentation of plastic items, both the mass and number concentrations of microplastic are expected to increase. Size distributions are expected to gradually shift towards a greater proportion of smaller particles, increasing bioavailability and potential risk to a wider range of species, including humans. Risks do not appear to be widespread at this point, but most scientists agree that it is not a question of if, but rather when, the environmental and human health risks of microplastic particles become apparent. Improving risk assessment methods for microplastics will enable us to determine the timeline for these risks better.

Viewing environmental microplastic as a continuum of characteristics rather than as a categorical phenomenon is a relatively new concept. Although newer frameworks address the diversity of microplastics with respect to analytical detection, exposure assessment, impact and risk assessment, they have not been applied to perform actual risk assessments in a regulatory context. It is critical that these postulated frameworks are tested, validated and simply tried out across a wide range of environmental and human health-related risk assessment problems.

These new frameworks provide opportunities not only to assess the risks of microplastic particles, but also to unify and align research approaches that have hitherto been fragmented and disparate. The conventional wisdom regarding the risk assessment for any stressor is that a framework is needed first. Only then can the acquisition of empirical data and the development of technical and theoretical instruments optimally match the intended purpose. Loss of time and efficiency can thus be avoided. In this context, we present the framework described in this Review and hope that it can lead to further alignment and efficiency in this complex area of research.

Microplastic particles form a continuum within a broader continuum of natural and other anthropogenic particles. Although some studies address the differences in effects between microplastic and clay particles, environmental microplastics are not typically viewed in the context of the multiple other particle types present. We suggest that the approach proposed here to distinguish effect metrics associated with potentially co-occurring specific effect mechanisms is equally applicable to other particle types. An ecological risk assessment based on a ‘food dilution’ effect mechanism from the ingestion of low-calorific microplastic particles could, in theory, also work for low-calorie clay, silt or sand microparticles. Oxidative stress after translocation of small particles, owing to inhalation of microplastics and other particles such as PM2.5, could ideally be assessed together. Soot and other black carbon particles are highly aromatic, with substantial amounts of polycyclic aromatic hydrocarbons, making it logical to assess the implications of plastic-associated chemicals in the context of exposure to other such chemically contaminated particles. Thus, the conceptual approaches presented in this Review to address the multidimensionality of microplastics may perhaps be extended to address the multidimensionality of all categories of hazardous particles.