Environmental DNA, usually abbreviated eDNA, is DNA recovered from environmental material such as water, soil, sediment, snow, or air, rather than obtained directly from a selected organism. It includes DNA inside living cells and microscopic organisms, DNA contained in shed tissue or organelles, and extracellular fragments released by secretion, excretion, decomposition, or cell damage. By identifying these sequences, researchers can investigate organisms that are difficult to observe, capture, or distinguish by appearance.
Environmental DNA analysis has become an important approach to monitoring biodiversity, detecting rare or introduced species, studying microbial communities, and reconstructing past environments. Its principal advantage is that a sample can contain biological evidence from many organisms without requiring their collection. Its principal interpretive limitation is that a DNA sequence is not equivalent to a direct observation: genetic material may persist after an organism has died, move away from its source, or enter a sample through contamination.
The term encompasses both contemporary ecological surveys and investigations of ancient environmental deposits. Methods range from highly specific tests for one species to sequencing-based inventories of entire biological communities. Reliable interpretation therefore depends on linking molecular measurements to the processes that produce, transport, preserve, and ultimately destroy DNA in the environment.
Origins and conceptual scope
Environmental DNA research developed from several overlapping traditions. Environmental microbiology had long investigated organisms that could not readily be cultured, while molecular ecology introduced genetic methods for identifying organisms and their relationships. In 1987, Andrew Ogram and colleagues published a method for extracting and purifying microbial DNA from sediments. Such work helped establish direct environmental extraction as an alternative to first growing organisms in the laboratory.
The development of the polymerase chain reaction made it possible to amplify small quantities of particular DNA sequences. Subsequent improvements in DNA sequencing allowed researchers to examine increasingly complex mixtures. Microbial surveys often targeted genes encoding ribosomal RNA, particularly the small-subunit ribosomal genes used to distinguish major bacterial, archaeal, and eukaryotic lineages. These approaches revealed extensive diversity overlooked by cultivation and conventional microscopy.
A related development was the recovery of ancient DNA from sediments and other deposits without identifiable body fossils. In 2003, Eske Willerslev and colleagues reported ancient plant and animal DNA from sedimentary material, helping establish that environmental deposits could preserve records of former communities. In contemporary ecology, a widely cited 2008 study by Gentile Francesco Ficetola and colleagues demonstrated the detection of American bullfrogs in pond water. It showed that aquatic vertebrates could be surveyed through DNA released into their surroundings.
The meaning of “environmental DNA” varies somewhat between disciplines. In broad usage, it includes all DNA extracted directly from an environmental sample, including the genomes of intact bacteria, archaea, and microscopic eukaryotes. In some conservation literature, it refers more narrowly to genetic traces shed by larger organisms. The distinction matters because a bacterium collected on a filter may itself be the organism under study, whereas fish DNA on the same filter may be only a fragment of tissue or mucus.
Environmental DNA describes the source of genetic material, not one analytical technique. DNA metabarcoding identifies multiple taxa by sequencing selected marker regions from mixed samples; metagenomics examines environmental genetic material more broadly, often without targeting a single marker. Conversely, metabarcoding of a deliberately collected mixture of whole insects is not necessarily eDNA analysis in the narrower sense. Clear descriptions of sample material and laboratory procedures are therefore more informative than the label alone.
Biological sources, transport, and persistence
Organisms release DNA through several pathways. Animals shed skin cells, mucus, feces, urine, reproductive material, and fragments of tissue. Plants contribute pollen, roots, leaves, and decomposing material. Microorganisms release DNA through cell death, viral infection, and other processes. Consequently, an environmental sample can contain nuclear DNA, mitochondrial DNA, chloroplast DNA, and microbial genomes in different physical states. DNA enclosed within a cell or a mitochondrion may behave differently from an unprotected extracellular fragment.
The rate of release is not constant. Body size, abundance, developmental stage, feeding, activity, stress, and reproduction can all affect shedding. A spawning event may produce a sharp rise in target DNA without a corresponding increase in the number of adult animals. Conversely, dormant or sparsely distributed organisms may contribute little detectable material. These differences explain why equal numbers of two species need not produce comparable environmental DNA concentrations.
Once released, DNA is redistributed by water movement, wind, settling, and interactions with particles. In a river, a detected sequence may have originated upstream rather than at the collection point. Transport distance depends on river discharge, channel structure, retention in sediments, and degradation. In marine environments, currents, stratification, and mixing influence the relationship between a sampled water mass and the organisms that contributed DNA. Terrestrial runoff can also introduce land-derived DNA into lakes and coastal waters.
DNA disappears through several interacting mechanisms. Enzymes produced by organisms can break it down; chemical reactions damage bases and cleave its backbone; sunlight can cause molecular damage. Temperature, pH, microbial activity, and exposure conditions influence these processes. Adsorption onto clay minerals or organic particles can protect DNA against some forms of degradation, although strong binding can also make laboratory extraction difficult. Sediment disturbance may subsequently return previously deposited DNA to the water column.
A simplified model represents loss as first-order decay:
where is DNA concentration after time , is the initial concentration, and is a decay constant. The corresponding half-life is . This approximation is useful in experiments but does not establish a universal environmental clock. Real samples contain multiple DNA fractions, receive continuing inputs, and experience changing transport and preservation conditions. Detectable DNA may persist for hours or days in some exposed waters, substantially longer in other settings, and for geological timescales in exceptionally protected sediments.
Sampling design and laboratory recovery
An eDNA investigation begins with a biological question and a defined sampling unit. A survey intended to detect one rare amphibian requires a different design from an inventory of marine plankton or a reconstruction of vegetation from lake sediment. Relevant choices include where and when to sample, the number of independent sites, the volume or mass collected, and whether samples are pooled. Spatial replication addresses patchy distributions, while repeated visits help distinguish persistent patterns from temporary changes.
Water samples are commonly filtered to concentrate cells, particles, and associated DNA. Filters with pore sizes around 0.2–0.8 micrometres are frequently used, although both finer and coarser filters have applications. Smaller pores can retain more material but clog more rapidly in turbid water. Sample volumes range from small fractions of a litre to many litres, depending on the environment and equipment. Increasing volume can improve detection, but it can also concentrate substances that interfere with amplification.
Soil and sediment require particular attention to heterogeneity. A small subsample may contain very different DNA from material collected only centimetres away. Root fragments, mineral surfaces, organic layers, and sediment grain size affect recovery. Air sampling typically draws air through filters or captures suspended particles using other collection devices. Snow, ice, surface swabs, and water retained in plant structures provide additional sampling media, each with its own spatial and temporal meaning.
Preservation limits changes between collection and analysis. Common approaches include freezing, rapid filtration followed by preservation of the filter, and chemical buffers that stabilize DNA. A preservation method must be compatible with the extraction procedure and the intended assay. Metadata such as sampling time, water temperature, depth, recent rainfall, and storage duration can be essential for explaining differences among samples. Archived extracts and filters may permit later analysis, but storage does not eliminate all degradation or contamination risks.
Extraction generally combines disruption of cells or particles, removal of interfering compounds, and purification using silica membranes or magnetic beads. Humic substances, some salts, and other sample constituents can inhibit amplification. Dilution may reduce inhibition but also lowers target concentration; additional cleanup can remove inhibitors while losing DNA. Field blanks, filtration blanks, extraction blanks, and laboratory negative controls help locate contamination introduced at different stages. Separating pre-amplification work from amplified products is particularly important because the latter can contain enormous numbers of target copies. A carefully designed sampling campaign can therefore fail if laboratory handling introduces a small but systematic contaminating signal.
Targeted detection and molecular quantification
Targeted assays ask whether DNA from a specified organism or group is present. They usually employ short primers that bind to a diagnostic sequence and, in many cases, a fluorescent probe that provides an additional layer of specificity. Mitochondrial markers are frequently used for animals because mitochondrial genomes occur in multiple copies per cell and often contain useful differences between species. Nuclear and chloroplast markers are also used when they better match the biological question.
Quantitative PCR, or qPCR, monitors fluorescence during amplification. Under appropriate conditions, samples with more initial target molecules reach a fluorescence threshold earlier. A standard curve based on known quantities links threshold cycles to starting concentration. Results may be expressed as target copies per reaction, per extraction, or per volume of original water. These are different measurement levels: conversion between them requires records of filtered volume, extraction volume, dilution, and the amount of extract added to each reaction.
Digital PCR partitions a reaction into many small compartments, some containing target molecules and others containing none. After amplification, the fraction of positive partitions supports an estimate of the original number of molecules, usually through a Poisson model. Droplet digital PCR creates these partitions as droplets. Digital approaches can be useful for low concentrations and some inhibitor-rich samples, but they still require validated specificity, appropriate controls, and careful interpretation near the detection limit.
Assay validation has several distinct components. Sequence comparisons evaluate whether primers and probes might bind to non-target organisms. Tests against tissue DNA assess amplification of the intended species and closely related species. Environmental samples then test performance in realistic mixtures. An assay developed for one region may fail elsewhere if target populations possess different alleles or if an untested local species carries a similar marker. No single percentage of sequence similarity guarantees correct discrimination.
The limit of detection is the concentration at which an assay achieves a specified probability of detecting the target; the limit of quantification concerns reliable numerical estimation. Their definitions and estimation procedures should be reported explicitly. At very low concentrations, random allocation of molecules among reactions means that replicate tests can disagree even when all procedures are functioning correctly. A weak positive is therefore not automatically contamination, just as a negative reaction is not proof of absence. Quantitative reporting also requires attention to calibration, extraction recovery, inhibition, and measurement uncertainty, rather than treating the instrument’s reported number as an exact ecological measurement.
Community profiling and sequence interpretation
Community-level analysis replaces a single-species question with the identification of multiple organisms in a sample. The most widely used approach is metabarcoding: primers amplify a marker shared across a broad group, and high-throughput sequencing reads the resulting mixture. This extends the principle of DNA barcoding, which identifies individual specimens using standardized sequence regions. In an environmental mixture, however, sequences must first be separated computationally into distinguishable biological signals.
Marker choice determines much of what can be detected. Portions of the mitochondrial 12S gene are commonly used for vertebrates, while mitochondrial cytochrome c oxidase subunit I, often abbreviated COI, is widely used for animals, especially invertebrates. The 16S ribosomal RNA gene is extensively used for bacteria and archaea; 18S markers target broad eukaryotic diversity; internal transcribed spacer regions are important for fungi. Chloroplast regions such as trnL can recover plant information, including from short, degraded fragments. No marker provides equal resolution across all organisms.
Laboratory processing usually attaches sample-specific index sequences to amplified products, allowing many samples to be pooled in one sequencing run. Computational analysis then removes low-quality reads, separates samples, identifies artificial chimeras, and accounts for sequencing errors. Some workflows cluster reads into operational taxonomic units using a similarity threshold. Others infer amplicon sequence variants that distinguish exact sequences after error correction. Neither category automatically corresponds to a biological species: one species can possess multiple marker variants, and different species can share the same marker.
Bioinformatics connects these sequences to taxonomic names through reference databases. Reliable assignment depends on correctly identified reference specimens, adequate geographic coverage, and a marker that actually distinguishes the organisms concerned. A sequence may be confidently assigned to a family but not to a species. Forcing a species-level name onto an ambiguous match creates false precision. Taxonomy also changes, requiring versioned databases and records of the classification system used.
Read counts are shaped by extraction efficiency, primer mismatches, amplification competition, genome copy number, and sequencing depth. Thus, a species producing half the reads need not account for half the biomass. Mock communities with known composition help reveal these biases. Shotgun metagenomic sequencing avoids the targeted amplification of one barcode but often requires more sequencing to detect scarce organisms and can be dominated by abundant microbial or host DNA. Target-enrichment methods occupy an intermediate position by capturing selected sequences before sequencing. Each approach trades breadth, sensitivity, resolution, and cost differently; combining markers can expand coverage but does not remove all shared sampling biases.
From detection to ecological inference
Environmental DNA measurements support several levels of inference that should remain distinct. A sequence match concerns molecular identity. Detection in a sample concerns analytical presence. Inferring that an organism occupies a particular site adds assumptions about transport and contamination. Estimating population size adds further assumptions about shedding, persistence, and sampling efficiency. Evidence adequate for one level may be insufficient for the next.
False negatives arise when organisms are present but their DNA is not collected, recovered, or amplified. False positives can result from contamination, sequence misidentification, or a laboratory artefact. They can also occur at the ecological interpretation stage: a correctly identified sequence may originate from an organism outside the sampled area. Fish DNA carried in wastewater, prey DNA transported by a predator, or material introduced with bait can produce genuine molecular detections without indicating a local living population.
Occupancy modeling explicitly separates the probability that a site is occupied from the probability of detecting a target when it is present. In a simple formulation, if each survey has detection probability and surveys are conditionally independent, the probability of at least one detection in surveys is
Environmental DNA studies often need hierarchical models because sampling, extraction, and amplification introduce separate detection processes. Repeating PCR on one extract improves knowledge of laboratory detection but does not substitute for collecting independent environmental samples. Some models additionally estimate false-positive probabilities.
Estimating abundance is more demanding. Under controlled conditions, DNA concentration can correlate with biomass or individual abundance, but the relationship may change across temperatures, life stages, species, or habitats. A simple well-mixed model illustrates the problem:
where is the rate of target DNA input, is water volume, and is the loss constant. At equilibrium, . Even this simplified expression shows that the same concentration can result from different combinations of shedding and loss. Flowing systems additionally require transport and exchange terms.
Comparisons with conventional surveys help establish what a molecular signal means. Netting, visual observations, acoustic surveys, and camera records have their own detection biases, so disagreement does not automatically invalidate either method. Integrated studies can reveal complementary coverage: eDNA may detect elusive species, while direct observations supply age, size, behavior, or evidence of reproduction. Interpretation is strongest when the spatial extent, time window, and biological meaning of each method are explicitly matched rather than assumed to be identical.
Biodiversity monitoring and environmental management
One major application is the detection of rare, elusive, or introduced organisms. Water sampling can reveal aquatic species that occur at low densities or occupy habitats difficult to survey physically. Tests developed for Asian carps in North America and great crested newts in the United Kingdom illustrate the use of species-specific assays in management. In 2014, a large English study demonstrated the utility of pond-water eDNA for great crested newt detection, contributing to subsequent use of the method in ecological assessments.
For invasive species, early detection can be valuable because small populations are more difficult to locate by conventional methods. Nevertheless, an isolated positive does not by itself establish a reproducing population or its location. Investigations may combine repeated sampling, independent assays, hydrological information, and conventional searches. After control or eradication, residual DNA creates the converse difficulty: continued detection may reflect surviving organisms, recently dead material, or DNA retained in sediments.
Metabarcoding can characterize assemblages of fish, amphibians, invertebrates, plants, fungi, and microorganisms. Repeated surveys can track community change across seasons or gradients of pollution and habitat alteration. In rivers, samples can integrate signals from parts of a drainage basin, potentially providing broad coverage but complicating the assignment of detections to individual reaches. In lakes and the ocean, depth, stratification, currents, and seasonal migrations influence which communities are represented.
Environmental assessment also uses molecular information to investigate ecosystem condition. Community composition can respond to nutrient enrichment, oxygen depletion, salinity, and other pressures. Sequence-based indicators can supplement established biological indices, but transferring an index from morphology to DNA requires validation because the two approaches do not necessarily detect the same organisms with equal efficiency. Reference conditions, sampling season, marker choice, and regional species coverage all influence comparisons among sites.
Airborne and terrestrial applications have expanded the range of possible surveys. Two independent studies published in 2022 recovered vertebrate DNA from air collected in zoos, demonstrating that airborne material could reveal nearby animals, including species not directly handled by researchers. Soil, surface water, and material washed from vegetation can likewise contain traces of terrestrial organisms. These applications remain strongly dependent on collection context: airborne particles can travel, food or animal feed can contribute sequences, and enclosed spaces differ from open landscapes. Across environments, eDNA is most informative when used as a defined evidence stream within a broader monitoring program rather than as a universal replacement for organism-based observation.
Sedimentary archives and past ecosystems
Environmental deposits can preserve DNA long after the organisms that produced it have disappeared. Lake sediments, cave deposits, frozen ground, and marine sediments accumulate genetic material from local organisms and from surrounding landscapes. Sedimentary ancient DNA is often abbreviated sedaDNA. Unlike an identifiable bone or seed, a DNA-bearing sediment sample may integrate material from numerous organisms and pathways, making its interpretation dependent on sediment formation as well as molecular analysis.
Preservation is favored by conditions that slow molecular damage, including low temperatures and protection on mineral surfaces. Permafrost is especially important, but DNA can also survive in other protected deposits. Over time, molecules become shorter and accumulate characteristic chemical changes, including cytosine deamination that can appear as excess terminal C-to-T substitutions in sequence reads. Such damage patterns can support authenticity, although their absence or presence must be interpreted in relation to preservation conditions and laboratory treatment.
Chronology is central. The age assigned to a DNA sequence usually comes from the surrounding deposit or associated material rather than direct dating of the individual DNA molecule. Radiocarbon dating, stratigraphic relationships, and other geochronological methods establish the sequence of deposition. Sediment mixing, erosion, redeposition, and movement through pore spaces can disconnect DNA from its apparent layer. Consequently, the date of a sediment sample is not automatically the exact age of every genetic fragment within it.
Ancient environmental DNA has reconstructed vegetation, animal distributions, and changes associated with glaciation and human activity. A 2022 study of the Kap København Formation in northern Greenland reported DNA from deposits approximately two million years old, identifying components of an ecosystem markedly different from modern northern Greenland. In 2017, Viviane Slon and colleagues reported Neanderthal and Denisovan mitochondrial DNA from cave sediments, demonstrating that human evolutionary evidence could be recovered from layers without recognizable hominin skeletal remains.
These records complement pollen, fossils, chemical proxies, and archaeological evidence rather than reproducing them exactly. Pollen can travel far and some plants produce much more of it than others; skeletal remains favor organisms and tissues that preserve well; DNA recovery depends on molecular survival and reference coverage. Agreement among independent proxies strengthens reconstruction, while discrepancies can reveal different source areas or preservation processes. Ancient DNA laboratories therefore use stringent contamination controls, specialized facilities, and methods adapted to short molecules. Claims about local survival, disappearance, or community composition depend jointly on molecular authenticity, sediment history, chronology, and ecological plausibility.
Standardization, emerging methods, and ethical questions
Routine use requires a chain of quality assurance extending from field design to reporting. A reproducible study records sample locations, dates, volumes, filter characteristics, preservation, extraction procedures, assay sequences, controls, and data-processing parameters. Reporting only a species list conceals major sources of variation. Reference materials, interlaboratory comparisons, and validated operating procedures help determine whether differences between surveys reflect ecological change or differences in measurement.
Standardization does not imply that one protocol suits every environment. Turbid rivers, oligotrophic lakes, forest soils, and airborne dust present different sampling and inhibition problems. The central requirement is fitness for the intended purpose: a method designed to detect one species at low concentration need not provide an unbiased community profile. Long-term monitoring is especially sensitive to changes in primers, sequencing platforms, databases, and software. Bridge studies and reanalysis of archived data can help distinguish methodological discontinuities from biological trends.
New technologies are moving parts of the workflow closer to the sampling site. Automated samplers can collect repeated samples without a field team at every visit, while autonomous underwater instruments can filter water and preserve material. Portable amplification and sequencing systems can shorten the interval between collection and a provisional result. These approaches do not remove the need for contamination controls or reference databases; rather, they shift some quality-control requirements into less controlled field conditions.
Environmental RNA offers a related but distinct evidence stream. Because RNA participates in gene expression and is often relatively labile, it can provide information not contained in DNA surveys alone. However, RNA detection is not a universal test of organism viability or current activity: stability varies with molecule type and environmental conditions, and protected RNA can persist. Other developing approaches seek population-level variation, longer genomic regions, or improved abundance estimation. Such applications demand greater sequencing depth and more careful separation of individuals, populations, and closely related species.
Environmental sampling also raises questions of data privacy, ownership, and access. Human DNA can be recovered incidentally from environmental samples; a 2023 study demonstrated that such material could contain substantially more human genomic information than a simple species detection. Public deposition of sequence data can therefore require consideration of incidental human information. Fine-scale locations of threatened species may also be sensitive. Work involving Indigenous lands, culturally significant places, or genetic resources intersects with consent, community authority, and access-and-benefit-sharing arrangements. These concerns affect decisions about sample collection, analysis, metadata disclosure, and long-term storage, even when the original research objective is non-human biodiversity.