Landmark Studies

Thompson and colleagues examined sediments at six sites around Plymouth, UK, and on 17 more British beaches, plus North Atlantic plankton samples archived since the 1960s. Infrared spectroscopy identified nylon, polyester and other polymers; the archive showed a significant increase. In tanks, amphipods, lugworms and barnacles ingested the particles. A 2024 review says the term was first applied to microscopic plastic debris here⁶. The core finding, microscopic plastic widespread at sea, is now consistently found worldwide. Limitation: UK shores and two plankton routes, counts not mass, smallest particles unidentifiable.

Browne and colleagues found microplastic (here, below 1 mm) on shorelines at 18 sites on six continents, with more near dense populations. Polyester and acrylic fibers occurred in proportions resembling sewage effluent, and a single garment shed more than 1,900 fibers per wash. Later laboratory work found over 700,000 fibers from a 6 kg acrylic load²⁶; per-wash numbers vary with fabric and conditions. Limitation: counts only, and the sewage link was inferred from polymer proportions, not traced.

Eriksen and colleagues combined 680 net tows and 891 visual transects from 24 expeditions (2007–2013) with a dispersal model. They estimated at least 5.25 trillion floating particles weighing 268,940 metric tons, with far fewer small particles than fragmentation predicts. In 2023, pooling 11,777 stations, the same lead author's team estimated about 82–358 trillion particles weighing 1.1–4.9 million metric tons (Mt)²⁷. We grade the numbers C, each being one modeled estimate; widespread floating plastic is not in doubt. Limitation: surface only, nets miss particles below 0.33 mm.

Jambeck and colleagues linked waste, population and management data for 192 coastal countries. They estimated 275 Mt of plastic waste generated there in 2010, with 4.8–12.7 Mt entering the ocean. Later models changed the scope: 19–23 Mt into all aquatic ecosystems in 2016²⁸, and 0.8–2.7 Mt per year from more than 1,000 rivers²⁹. Limitation: an estimate resting on assumed shares of mismanaged waste reaching the sea, with no measured flux.

In a viewpoint, Nizzetto, Futter and Langaas combined sludge volumes, the share spread on farmland, and microplastic contents of sludge. They estimated that 63,000–430,000 metric tons of microplastics reach European farmland each year, and 44,000–300,000 metric tons North American farmland. Each range exceeds the 93,000–236,000 metric tons then estimated afloat in the oceans. It put soil on the research agenda. Limitation: a theoretical estimate, not checked against measured soil stocks.

Geyer, Jambeck and Law compiled production, use and end-of-life data for resins, fibers and additives. By 2015, they estimated, 8,300 Mt of virgin plastic had been made and 6,300 Mt had become waste: about 9% recycled, 12% incinerated and 79% landfilled or left in the environment. The OECD's 2022 model again put recycling at 9%, with 460 Mt produced in 2019³⁰. Limitation: global mass-balance estimates from industry statistics, not tracked flows.

Allen and colleagues collected five months of deposition in winter 2017–2018 at the Bernadouze station, 1,425 m up in the French Pyrenees. They counted an average of 365 particles per square meter per day, mostly fragments, from source areas up to 95 km away. The atmosphere became a transport route in its own right; a 2026 compilation holds 2,782 measurements²⁴. The Author Correction only added acknowledgements and fixed a reference. Limitation: one site, one season, counts without mass.

Brahney and colleagues sampled deposition at 11 sites in national parks and wilderness areas of the western United States. It averaged 132 particles per square meter per day, scaled to more than 1,000 metric tons per year on the region's protected lands. Wet deposition tracked nearby cities; dry deposition came from farther away. Limitation: the tonnage is an extrapolation from counts, and a global emission estimate later built on western US data³¹ fed models that overestimate measured air concentrations²⁴.

Li and colleagues exposed wheat and lettuce to fluorescent polystyrene (PS) beads of 0.2 µm and 2 µm, and to polymethylmethacrylate particles, in hydroponic culture, sand and sandy soil. Particles entered roots through cracks where lateral roots emerge, reached the vascular core and moved up to stems and leaves. Limitation: pristine spherical beads in the laboratory; uptake of weathered particles in working fields is not established.

Schwabl and colleagues collected one stool sample each from eight healthy volunteers in Europe and Asia. Infrared microspectroscopy found microplastics in all eight: a median of 20 particles per 10 g, 50–500 µm across, nine polymer types, mostly polypropylene (PP) and polyethylene terephthalate (PET). It was the first published detection in human stool. Limitation: eight people, one sample each, only particles of 50 µm and larger.

Ragusa and colleagues found 12 pigmented particles of 5–10 µm in four of six placentas by Raman microspectroscopy, on fetal and maternal sides. Three were polypropylene; for nine, only the pigments could be identified. A 2024 study using pyrolysis gas chromatography–mass spectrometry (py-GC/MS) reported plastic in all 62 placentas tested, mean 126.8 µg/g, mostly polyethylene³²; in blood, that method gives false polyethylene signals³. Limitation: six samples, and most identifications rest on pigments, not polymers.

Two 2021 intake estimates differ by a factor of about 24,000 to 1.2 million (our calculation). Senathirajah and colleagues combined particle counts from 59 publications, for particles up to 1 mm, with average particle masses: 0.1–5 g per person per week. The upper end became WWF's 2019 "credit card a week"³³. Mohamed Nor and colleagues modeled eight foods and inhalation for particles of 1–5,000 µm: median adult intake 583 ng per day, about 4.1 µg per week (our calculation)². Pletz argues the 5 g figure is orders of magnitude too high because sizes and masses were combined wrongly³⁴. Limitation: both are estimates with different size windows.

Leslie and colleagues applied double-shot py-GC/MS, for particles of 700 nm and larger, to whole blood from 22 healthy donors in the Netherlands. Plastic was quantifiable in 17, mostly PET, styrene polymers and polyethylene (PVC was not targeted), with a mean summed concentration of 1.6 µg/mL, the first mass of plastic reported in blood. Rauert and colleagues later found that blood lipids create false polyethylene signals and matrix suppression makes PET unreliable³. A 2025 follow-up by overlapping authors, with a changed workflow, detected plastic in all 102 samples but quantified it in only 20%, averaging 386 ng/mL³⁵. Limitation: 22 donors, nothing below 700 nm.

Marfella and colleagues analyzed carotid plaque from 304 surgical patients by py-GC/MS, with supporting electron microscopy and carbon-isotope analysis, and followed 257 for a mean of 34 months. Polyethylene was detected in 150 of the 257 (58%) and PVC in 31. Those with detectable plastic had a higher risk of heart attack, stroke or death (hazard ratio 4.53; 95% CI 2.00–10.27). Hołda and Batko's letter notes no anti-contamination steps beyond glass tubes and no operating-room blanks; the authors' reply was inaccessible to us³⁶. The paper says contamination "cannot be firmly ruled out". Limitation: one cohort, residual confounding, an association.

Nihart and colleagues analyzed liver, kidney and frontal cortex from New Mexico decedents who died in 2016 or 2024 (20–28 per year group), plus 12 brains from people with dementia. By py-GC/MS, median brain concentration was 4,917 µg/g in 2024 against 3,345 µg/g in 2016, mostly polyethylene, and highest with dementia. These are different people at each date, not accumulation within people, and the authors assume no causal role in dementia. A Matters Arising argues that limited contamination controls, missing validation and incomplete digestion of lipid-rich brain may have inflated the values⁴. Limitation: one pyrolysis method.

Ter Halle and colleagues isolated colloids from North Atlantic subtropical gyre seawater, confirmed nanoscale particles by dynamic light scattering and fingerprinted them by py-GC/MS. The anthropogenic signal matched PVC, PET, PS and polyethylene: the first nanoplastic reported in a natural sample. Later basin-scale measurements support it²³. Limitation: polymers were identified but concentrations not reported, and polyethylene signals changed with weathering.

Materić and colleagues adapted thermal desorption–proton transfer reaction–mass spectrometry (TD-PTR-MS) to plastics, detecting below 1 ng of polystyrene in 1 mL samples, about 100 times more sensitive than earlier methods by their comparison. In Austrian Alpine snow they found several microplastic polymers, but only PET in the nanometer range. The method later enabled basin-scale ocean measurements²³. Limitation: semi-quantitative, one site, and fingerprints that dissolved organic matter can complicate.

Qian and colleagues built a stimulated Raman scattering microscope that matches single particles, down to about 100–200 nm, against seven common polymers. Bottled water from three brands bought at one retailer held about 240,000 plastic particles per liter (±130,000), about 90% nanoplastics. The count is disputed: Materić notes that the Milli-Q procedural blanks held about as many particles as the samples⁴⁰; the authors reply that this water was itself contaminated and so was not a valid blank⁴¹. Limitation: three brands, seven polymers, unidentified particles not counted.

Ten Hietbrink and colleagues sampled 12 stations on a November 2020 cruise of RV Pelagia, from the subtropical gyre to the European shelf, surface to seafloor. TD-PTR-MS found 1.5–32 mg/m³ of PET, PS and PVC nanoplastics. Extrapolated from the nine open-ocean stations to the temperate to subtropical North Atlantic mixed layer, this gives an estimate of 27 Mt, rivaling estimates for larger floating plastic. Limitation: one cruise extrapolated to a basin; polyethylene and PP were not detected.

Evangelou, Bucci and Stohl compared 2,782 airborne microplastic measurements from 76 studies at 283 locations (2014–2024) with size-matched model runs. Models driven by three earlier emission estimates overestimated measured median concentrations by two to four orders of magnitude. Measured medians were 0.08 particles/m³ over land and 0.003 over the ocean. After rescaling, land emits about 23 times more particles than the ocean (our calculation), but the ocean dominates by mass. Limitation: particle numbers from methods and size ranges that differ.

Li and colleagues analyzed 156 samples of tumor and surrounding brain from 113 surgical patients in China, and 35 healthy brain samples from five post-mortem donors. Infrared imaging gave particle counts, py-GC/MS gave mass, including a fraction below 1 µm, and a second infrared method disagreed on 2 of 18 checked particles. A hot-ethanol wash removed a lipid peak before pyrolysis; operating-room air, saline and surgical materials were tested. Plastic was in 99.4% of patient and all healthy samples, at higher levels beside tumors than in healthy brain; the healthy-brain py-GC/MS median was 50.3 µg/g, a value from the publisher's source-data file. A correction fixed figure rounding⁴⁵. Limitation: five healthy donors, surgical versus post-mortem tissue, correlations only.