How eDNA compares to conventional survey methods

The question

eDNA methods began appearing in the scientific literature in the early 2000s. Supporters have made bold claims about their potential: more species detected, faster, with less effort, and without the need for highly trained field taxonomists. Critics have pushed back, pointing to studies where eDNA performed no better than conventional approaches, or produced results inconsistent with the existing literature.

Both sides were right. The problem was that studies comparing eDNA to conventional methods had used at least 16 different metrics to evaluate performance, and no one had systematically looked at all of them together. A 2025 review published in Ecography by Nicholas Iacaruso and colleagues at the University of Illinois set out to do exactly that.¹

What they found

Across the full dataset of 398 studies, eDNA was more likely to produce higher estimates of sensitivity, meaning it found more species and detected target species more reliably, at lower costs. eDNA analysis required less time and money to achieve comparable sensitivity to conventional methods. Those findings applied broadly across environments, taxa, substrates, and the type of conventional method used.

The sensitivity advantage -- and its limits

For species richness and detection probability, eDNA outperformed conventional methods in the majority of studies. This was consistent across environments (rivers, lakes, marine, terrestrial), substrates (water, soil, air), and taxonomic groups -- with only two notable exceptions. Birds and reptiles were groups where eDNA more often returned lower species richness and detection probability than conventional methods.

The authors note a likely biological explanation. Reptiles, which often have hard keratinized outer layers, may shed less DNA into the environment than other vertebrates. Birds are highly mobile and may not spend enough time in any sampled location to leave detectable traces. These are not fatal limitations, but they suggest that eDNA sampling protocols may need to be specifically adapted for these groups before they become reliable survey tools for them.

There is also an important caveat to the sensitivity numbers. Species richness can be inflated by eDNA detecting DNA that has drifted in from elsewhere. Some examples include an animal upstream, a species that was present days earlier, or even laboratory contamination. The review is careful to note that detecting eDNA and inferring presence are not the same thing, and conservation professionals need to understand that distinction before acting on eDNA survey data.

Efficiency: faster and cheaper, once set up

Nearly all studies that compared eDNA to conventional methods on cost or sampling time found eDNA required less effort to achieve equivalent sensitivity. The savings in hands-on time were significant across multiple studies, and per-sample costs were consistently lower than comparable conventional methods.

The review flags an important nuance though: the studies it examined only compared per-sample costs, not the substantial upfront investment in sequencing equipment, laboratory space, and bioinformatics infrastructure. For organisations that already have laboratory capacity, eDNA is genuinely cheaper to run. For those starting from scratch, the setup costs are considerable. Once those upfront costs are absorbed, the review concludes, eDNA effort costs do not scale linearly. Larger surveys become proportionally cheaper, which makes eDNA particularly attractive for landscape-scale monitoring programs.

The community composition problem

The most significant finding in the review is not about what eDNA does well, but about what it does differently. Across nearly all environments, substrates, taxa, and conventional methods tested, eDNA and conventional approaches consistently returned different community compositions. They were not measuring the same things and arriving at the same answer -- they were measuring different slices of the same community.

How each metric performed

Where eDNA has not been tested

The review reveals a pronounced geographic bias. The vast majority of comparative studies were conducted in North America, Europe, and parts of Asia. The only countries in the Global South with more than three comparative studies in the dataset were South Africa and Brazil, each with six. This is a serious gap because the most species-rich ecosystems on Earth are in the tropics, where eDNA reference databases are also thinnest and where DNA may degrade faster in warmer, more acidic conditions.

The review is explicit that eDNA methods are not being tested where they could make the most difference. Countries in the Global South often lack the funding and expertise to run conventional biodiversity monitoring programs, which is precisely the situation where eDNA's lower per-sample cost and reduced need for field taxonomists could have the greatest practical impact. Until more comparative studies are done in tropical and subtropical systems, conservation professionals working in those regions cannot know how much to trust eDNA results there.

Temporal monitoring: a critical gap

One of the review's clearest conclusions is that eDNA has barely been tested as a tool for tracking biodiversity change over time. Only four of the 398 studies explicitly evaluated the temporal distribution of taxa as a metric. Most studies collected just one temporal replicate per sampling location. This allows the characterization of what is present at a moment in time, but not enough to track whether populations are growing, declining, or shifting seasonally.

For rewilding specifically, this is an important limitation. The question that matters most for a rewilding project is not just "what species are here now?" but "is biodiversity recovering?" Answering that question requires repeated sampling over years, with consistent methodology that allows genuine comparison over time. The review calls for more investment in long-term eDNA monitoring networks that can provide this kind of temporal data.

What does this mean for rewilding and conservation?

The review's practical conclusions for conservation professionals are clear. eDNA is a genuinely useful addition to the biodiversity monitoring toolkit, but it is not a replacement for conventional methods. Its advantages are strongest for aquatic environments, for rare and elusive species, and for large-scale surveys where conventional effort would be prohibitively expensive. Its limitations are most pronounced for birds and reptiles, in terrestrial environments where less sampling infrastructure exists, and for any question that requires abundance estimates rather than presence-absence data.

For rewilding projects, the recommendation is to use eDNA alongside conventional methods at the start of a project, where both can establish complementary baselines, and to invest in temporal replication from the beginning rather than treating monitoring as a one-time exercise. The community composition findings suggest that rewilding projects relying solely on eDNA may miss important parts of the ecosystem they are trying to restore and vice versa.