How does airborne eDNA compare to water eDNA and soil eDNA?

1. The eDNA substrate landscape

Environmental DNA is not a single method — it is a family of approaches that share a common analytical core (DNA extraction and sequencing) but differ fundamentally in how genetic material accumulates, persists, and is sampled in different environmental matrices.

The three primary substrates are:

A fourth approach — sometimes termed forensic environmental sampling — collects deposited biological material from surfaces rather than from a fluid medium. This includes surface swabbing (e.g. vegetation surfaces, rocks, fences), faecal DNA collection, and targeted substrate sampling at points of known animal contact (urine marks, resting sites, den entrances). Forensic sampling is not a substrate in the same sense as air, water, or soil, but it is increasingly used alongside them for targeted single-species detection and attribution. It shares the analytical workflow of eDNA but differs in that the sampling is targeted rather than general.

Each substrate reflects a different ecological signal, operates at a different spatial and temporal scale, and is subject to different physical and chemical controls on DNA persistence and transport.

3. Direct comparisons: what the studies show

3.1 Air vs. water for aquatic organisms

Ip et al. (2025a) made the first systematic comparison of air and water sampling for the same aquatic target species (spawning Coho salmon in a Washington State stream). They found that DNA from the salmon was consistently detectable from passive air samples collected above the water surface, demonstrating that aquatic organisms can be detected from air through water-to-air transfer of eDNA via surface aerosols. Despite an approximately 25,000-fold dilution relative to the water signal, passive air collectors captured quantitative airborne eDNA signals that closely paralleled salmon counts over a six-week spawning period.

A follow-up study by the same group (Ip et al. 2025b; preprint) extended this approach to community-level vertebrate metabarcoding at two urban–wildland interface sites, collecting 27 paired air-and-water samples. They found substantial overlap in vertebrate species detected across both media: abundant aquatic species tracked temporally across air and water within 24 hours, while rare or low-abundance taxa were the first to drop out of cross-medium transfer. This finding has practical implications: it suggests that community-level signals do cross the air-water boundary, but detection sensitivity is strongly modulated by species abundance, meaning that airborne detection of aquatic organisms cannot replace in-water sampling for rare species.

Taken together, these results suggest that air monitoring near rivers and lakes may provide information about aquatic communities as a byproduct, and that the strict separation between "aquatic" and "terrestrial" eDNA sampling may be less absolute than previously assumed. They also raise a less discussed possibility — that airborne eDNA originating from one water body could be deposited in an adjacent one via atmospheric transport and wet or dry deposition. Until recently this was not recognised as a meaningful route of eDNA dispersal between water bodies (transfer via birds carrying eDNA on their bodies was acknowledged, but airborne particle transport was not). This has potential implications for the interpretation of aquatic eDNA signals in water bodies close to each other and for biosecurity protocols around invasive aquatic species.

3.2 Air vs. soil for microbial and fungal communities

The comparison between airborne and soil eDNA is most developed for microorganisms and fungi — the dominant signals in both matrices. Fungal communities are exceptionally well-represented in air via spore release and in soil via mycelial networks and spore banks; the two substrates capture overlapping but distinct fungal communities at different life-cycle stages (Abrego et al. 2024; Ovaskainen et al. 2024). Bacterial communities in air are strongly shaped by nearby soil surfaces, with terrestrial soil the dominant source of outdoor airborne bacteria (Bowers et al. 2011).

For plant communities specifically, the picture is more nuanced. Airborne eDNA integrates over a landscape-scale catchment and is strongest during reproductive periods; soil eDNA captures local, point-specific plant communities with high fidelity and longer temporal depth. Johnson et al. (2021) found that airborne eDNA metabarcoding detected more plant diversity than traditional field surveys, though direct systematic comparisons with soil eDNA for the same sites are limited. Detection of plant DNA from soil eDNA approaches is rarely the primary method of choice — it is incidental to microorganism-focused studies — and caution is warranted before treating airborne vs soil plant comparisons as established.

Strandberg et al. (2025) compared capture probe, metabarcoding, and shotgun sequencing approaches for vegetation characterisation and found that different methods captured different components of diversity, underlining the value of multi-approach design.

3.3 Air vs forensic approaches for terrestrial vertebrates

Newton et al. (2026) conducted the most rigorous multi-substrate comparison for terrestrial vertebrates, comparing airborne eDNA against spider webs, soil, leaf swabs, and other passive substrates. Airborne eDNA and spider webs together outperformed other substrates for total species richness detected, while soil was less effective for highly mobile species such as birds and bats. Berard et al. (2025) confirmed that no single substrate dominated across all taxonomic groups and environments, reinforcing the complementarity principle.

5. Spatial scale: the fundamental contrast

The most important practical difference between the three substrates is their spatial scale of operation.

A single water eDNA sample from a river represents species present upstream within a specific hydrological catchment — typically a few kilometres of channel, if we neglect contamination sources. A soil core or swab represents a patch of a few square metres to, at most, a few hundred square metres. An air sample represents a landscape footprint of several square kilometres to several tens of kilometres, depending on atmospheric conditions and sampler volume.

This scaling difference has direct implications for monitoring programme design:

The corollary is that airborne eDNA is most powerful as a landscape-level biodiversity indicator rather than a site-level species inventory tool. It is the method of choice when the question is "what is the overall state of biodiversity across this region?" rather than "does this specific pond contain great crested newts?"

7. Caveats and open questions

Direct substrate comparisons are rare. Most studies use a single substrate. The few direct comparisons that exist (Newton et al. 2026; Berard et al. 2025) are at single sites and cannot be generalised to all habitat types or taxa. Systematic multi-substrate comparison studies across diverse ecosystems are an urgent research priority.

Cost-effectiveness benchmarking is immature. Tournayre et al. (2025) showed airborne eDNA has low marginal cost when using existing infrastructure, but no study has yet conducted a rigorous cost-per-species-detected comparison across substrates for the same sites, questions and taxa.

The cross-medium transfer question is advancing but not resolved. Ip et al. (2025a, 2025b) showed water-to-air eDNA transfer for both single species (salmon qPCR) and community-level vertebrate assemblages (metabarcoding). The community-level data show that abundant species track reliably across media, while rare taxa drop out — implying that cross-medium transfer is real but detection-limited for low-abundance organisms. How this generalises across water body types (lentic vs lotic, marine vs freshwater) and environmental conditions remains to be established. An analogous question applies to air-to-soil transfer: DNA deposited from the atmosphere onto soil surfaces could in principle be recovered from soil samples, particularly after rain events that wash particulate matter into the upper soil layer. This pathway has received almost no direct study but would imply that soil eDNA near intensive land use areas (e.g. arable fields, roads, high-traffic natural sites) may contain atmospheric DNA contributions that are not of strictly local origin. This is a research gap with direct relevance to soil eDNA monitoring programme design.