An air filter does more than collect soot, pollen and dust. Mixed into those particles are microscopic traces shed by nearly every living thing nearby: fragments of skin, hair, feathers, leaves, spores, saliva, faeces, microbes and viruses. Scientists have shown that DNA recovered from this material can reveal a biological portrait of a place without anyone seeing or capturing the organisms that produced it.
The idea has advanced in stages. Experiments first demonstrated that animal DNA could be vacuumed from the air. Researchers then discovered that routine pollution-monitoring stations were already collecting environmental DNA as an accidental by-product of measuring airborne particles. More recent work has used powerful, untargeted sequencing to detect wildlife, pathogens, antimicrobial-resistance genes and even detailed human genetic variation from airborne samples.
The filter is not a magical species detector. It simply traps particles as air passes through. DNA must still be extracted, sequenced and compared with reference databases. Yet the finding means that a vast monitoring system built for one purpose may already contain an unexpected record of the life surrounding it.
DNA leaves the body and enters the environment
Environmental DNA, usually shortened to eDNA, is genetic material collected from a place rather than directly from an organism. Fish leave DNA in rivers, mammals shed it into soil and plants release it in pollen. Scientists have long used water and sediment samples to identify species that may be difficult to catch or observe.
Air seemed harder. Biological material is much more dilute in the atmosphere than in water, and wind can carry it away from its source. Even so, all organisms continuously release particles. A fragment does not need to be a complete cell, and researchers do not need an entire genome to make many identifications. A short sequence can sometimes be matched to a species through a genetic reference library.
Two independent teams demonstrated the potential in zoo experiments published in 2022. As the Natural History Museum’s account explains, researchers in Britain and Denmark pumped air through filters and detected species inside enclosures, in open habitats and even hundreds of metres from the sampling point. The British team identified DNA from 25 animal species, including 17 known zoo animals, as well as nearby wildlife and animals used as food.
Those controlled settings proved that vertebrate DNA could move through air and remain detectable. But zoos are unusually dense collections of large animals. The next question was whether the approach could work through existing equipment in ordinary landscapes.
Pollution stations were collecting biodiversity by accident
Air-quality stations routinely pull known volumes of air through filters so laboratories can measure particulate pollution, metals and other contaminants. In 2023, researchers tested whether the same filters also preserved biological information.
The resulting study in Current Biology analysed environmental DNA trapped alongside particulate matter at routine monitoring stations in the United Kingdom. The team identified more than 180 vertebrate, arthropod, plant and fungal taxa representative of nearby biodiversity.
This mattered because air-quality networks are standardised, geographically widespread and already operating. Instead of building a continental system of wildlife traps from scratch, ecologists might adapt infrastructure that governments maintain for public health. In some regions, filters have been stored for years or decades, potentially preserving a biological time series from the past.
The discovery does not mean every archived filter is immediately usable. Storage temperature, filter material, sampling duration, contamination controls and laboratory protocols all affect the DNA that survives. Still, the researchers found usable eDNA on an eight-month-old filter kept at room temperature, while frozen storage could preserve material much longer.
New sequencing reads far more than a species barcode
Many early eDNA surveys used metabarcoding. Researchers select a short genetic region, amplify it with a polymerase chain reaction and compare the resulting sequences with a database. This is efficient, but the chosen primers favour some groups and miss others. It also reveals only a small part of the genetic information on the filter.
Shotgun sequencing takes a broader approach by reading the mixture of DNA without first targeting one narrow barcode. A 2025 study in Nature Ecology & Evolution applied long-read and short-read shotgun sequencing to airborne eDNA collected in natural, urban and indoor settings. The researchers showed that a single sample could contain information spanning microbes, plants, wildlife, pests, pathogens and people.
In outdoor air from a Florida forest, the team detected elusive or difficult-to-survey animals including a bobcat, bats, moths, squirrels, birds, spiders, snakes and an alligator. Bobcat mitochondrial sequences were detailed enough to place their source close to a north-eastern Florida population. This goes beyond saying that a species was present. It suggests that airborne DNA can sometimes carry information about population relationships and genetic variation.
The method was also fast. According to the US Geological Survey’s publication record, one investigator could move from airborne sample collection to cloud-based analysis in two days using compact sequencing equipment.
Viruses can be detected, but detection is not infection
The same broad sequencing found genetic signatures from viruses, bacteria, fungi and other organisms with pathogenic potential. A long-read Dublin air sample contained sequences assigned to 63 viruses. Researchers recovered extensive genome coverage of an avian-associated virus from pumped air samples and a windowpane swab. They also found antimicrobial-resistance genes in air samples.
These results point towards environmental surveillance for plant, animal and human pathogens. Air sampling might help identify where a virus or disease vector is circulating, track agricultural threats or reveal changes before traditional surveys do.
There is an important limit. Finding a viral sequence does not by itself show that infectious virus particles are present, that a person is ill or that transmission is occurring. DNA can persist after an organism has died, and sequence assignments depend on the quality of reference databases and analytical thresholds. Airborne eDNA is a signal requiring interpretation, not a diagnosis.
The air also records human presence
Humans shed DNA as readily as other animals. The 2025 researchers recovered human genetic variation from indoor air, including insertions, deletions and more than 217,000 single-nucleotide polymorphisms after targeted enrichment. Earlier work had already warned that deep sequencing of wildlife samples can unintentionally collect what researchers call human genetic bycatch.
A 2023 analysis of human eDNA showed that genetic material from people can be recovered from air, water and sand. The authors argued that the technology could support medical, forensic and environmental applications, but also raised questions about consent, ownership, privacy and surveillance.
Other sequences may indicate human-associated activity. Dublin air contained DNA from food plants and from cannabis, poppy and fungi associated with psychoactive mushrooms, as described by the University of Florida. Such detections must be interpreted cautiously. A plant’s DNA can arrive through food, horticulture, legal products, transported dust or other sources. It cannot automatically prove that a particular person used a drug or performed a specific act.
The privacy issue becomes more serious as sequencing grows more sensitive. A filter designed to monitor air pollution could unintentionally collect genetic information from workers, residents or passers-by. Researchers therefore need rules governing when human sequences are analysed, how they are stored, who can access them and whether they should be deleted from biodiversity datasets.
A powerful tool with unresolved blind spots
Airborne DNA cannot yet provide a perfect census. Wind determines where particles travel, different organisms shed DNA at different rates and rain or sunlight may remove or degrade it. A positive result does not always prove that the source was standing beside the sampler, while a negative result does not prove absence.
Reference libraries are another limitation. A sequence can be identified only if suitable comparison data exist, and well-studied species are overrepresented. Closely related organisms may be difficult to distinguish. Contamination from laboratory staff, equipment or previous samples can also create false signals, making controls essential.
For these reasons, air eDNA is most useful alongside camera traps, acoustic sensors, field observations and established disease-surveillance methods. Its advantage is breadth. One filter can sample organisms that are nocturnal, microscopic, underground, high in the canopy, dangerous to approach or simply too elusive to see.
The atmosphere is not empty space between visible creatures. It is a moving archive of biological particles. Ordinary filters have been trapping that archive for years, and modern sequencing has finally made much of it readable. The result could transform biodiversity monitoring and pathogen surveillance, while forcing society to decide how much information should be allowed to drift unnoticed through the air.