Transport Distance of Invertebrate Environmental DNA in a Natural River

You are a conservation biologist. You take a water sample from a river, run it through a filter, extract the DNA, and detect genetic material from an endangered freshwater mussel. You mark that spot on the map. That is where the mussel lives. That is the habitat to protect. Except — what if the mussel lives ten kilometers upstream? What if its DNA simply traveled to you, carried by the current, while the animal itself never moved at all? That gap between where the DNA is and where the animal is — that is the problem Kristy Deiner and Florian Altermatt set out to measure. What they found should change how every river environmental DNA survey is designed and interpreted. Environmental DNA, or eDNA, is exactly what it sounds like: DNA collected not from the organism itself but from the environment it moves through — the water, the soil, the air. Organisms shed DNA constantly, through mucus, molting, gametes, and decay. You filter the water, amplify what you catch using polymerase chain reaction, or PCR, which makes millions of copies of a target DNA sequence, and then confirm the identity with sequencing. No trapping, no sighting, no disturbance. It is a powerful technique. However, most early aquatic eDNA work was done in lakes and ponds, where water does not move. In a lake, if you detect a species' DNA, that species is probably nearby. Rivers break that assumption entirely. Water flows, and DNA flows with it. Deiner and Altermatt designed their study around a beautifully clean natural experiment. Lake Greifensee in Switzerland sits at the head of the Glatt River. The lake is 8.5 square kilometers, up to 33 meters deep, and it hosts two invertebrate species that have never been recorded in the Glatt itself, despite standardized surveys over the past 20 years: Daphnia longispina, a planktonic crustacean about 2 millimeters long, and Unio tumidus, a sessile freshwater mussel around 10 to 15 centimeters. Both live in the lake. Neither lives in the river. So any detection of their DNA downstream is not a question of local presence — it is pure transport signal. The river is a conveyor belt, and the researchers wanted to know how far the belt runs before the signal disappears. They sampled water at eleven sites along the Glatt, spaced at regular intervals up to 12.3 kilometers from the lake. Sampling happened twice — in July and in October 2012. At each site, they collected 900 milliliters of water, filtered it through fine glass fiber filters, extracted the DNA, and ran three PCR replicates per sample, confirming every positive hit with Sanger sequencing — a method that reads out the actual DNA sequence, not just a fluorescent signal, so you can be certain you have the right species. They also ran negative controls for filtration, extraction, and PCR to make sure nothing was contaminating the results. The design is careful. The logic is airtight. The Glatt itself moves at an average velocity that the authors calculated at roughly 1.2 kilometers per hour. That means water — and any eDNA it carries — takes at minimum about 16 hours to travel from the lake to the farthest sampling site. Accounting for slower flow near the banks and the river bed, effective travel time could stretch anywhere from 5 to 40 hours. That is a lot of time for DNA to degrade. And that degradation is exactly what the results reveal, in two very different ways for two very different animals. Daphnia longispina eDNA was detected at every single river site, across all 12.3 kilometers, in both July and October. At nearly every site, all three PCR replicates came back positive. This is a remarkably clean result — a tiny crustacean, shedding DNA at high rates due to its continuous overlapping generations and regular molting, sending a detectable signal the full length of the study corridor. Unio tumidus told a different story. Detection declined with distance. By 9.1 kilometers, the mussel's signal had dropped out entirely. Unlike Daphnia, Unio showed variation between seasons. It was not detected at the 1.6 kilometer site in July but was detected there in October, and no site produced all three PCR replicates positive. Deiner and Altermatt put these patterns through a generalized linear model — a statistical test that asks whether the observed differences in detection rate are real rather than random. The model found an extremely strong effect of species identity, with an F-statistic of 71.25 and a p-value below 0.001. Distance had a significant negative effect on detection. There was also a significant interaction between species and sampling time, meaning the two species responded differently to the seasons. The raw finding is intuitive. The statistics confirm it is not a coincidence. The biological reasons for the difference make sense once you think about the two animals. Daphnia is small, abundant, and constantly shedding — it molts, it reproduces, it dies in large numbers. The concentration of its DNA entering the river from the lake is presumably high, giving the signal more staying power as it degrades downstream. Unio is large, sessile, and relatively sparse. It releases planktonic gametes and larvae only at particular times of year, and its overall DNA shedding rate into the water is likely far lower. Deiner and Altermatt interpret the contrasting footprints as arising from the combined effects of species-specific shedding rates, life history timing, and time-dependent decay during transport. A lower initial concentration degrades below the detection threshold sooner. That is why the mussel's signal fades before the crustacean's does. The sequencing step added something beyond simple presence or absence detection. When Deiner and Altermatt aligned the Daphnia eDNA sequences from river samples, they found that the haplotype — the specific genetic variant — shifted between July and October, differing by four base pairs. This shows that eDNA in river water is not just a yes or no signal. It carries actual genetic information about the source population, potentially enough to track population dynamics or genetic diversity from a water sample alone. Model extrapolations extended the picture beyond the 12.3 kilometers actually sampled. The generalized linear model predicts that Unio's detection probability would fall below 5 percent at roughly 15 kilometers in autumn and 25 kilometers in summer. For Daphnia, that same threshold sits at about 50 kilometers in summer. These are projections, not measurements — but they give a sense of the scale at which river eDNA operates. That scale is the crux of what this study means for conservation and monitoring. Detecting a species' DNA 10 or even 50 kilometers downstream of its actual location is not just a methodological curiosity. It fundamentally changes what a positive eDNA result in a river means. Deiner and Altermatt frame this as a feature as much as a problem: a single water sample from a downstream point could integrate the biological signal from an entire upstream catchment, giving you a biodiversity snapshot of a large network from one location. That is genuinely powerful for large-scale surveys of hard-to-detect invertebrates. However, it is a problem if you need to know where exactly a species lives. If you are trying to identify critical habitat for a protected mussel, detecting its DNA nine kilometers downstream of its actual population is not useful for siting a protected area. The paper recommends designing river eDNA surveys with sampling nodes spaced according to the logic of the river network, using intervals on the order of 5 to 10 kilometers to keep detections interpretable. Hydrology — flow rate, channel geometry, discharge — needs to be factored into any interpretation. This study is the first demonstration that invertebrate eDNA can persist and remain detectable over these distances in a natural river system. Invertebrates are ecologically critical and notoriously difficult to survey using traditional methods. Expanding eDNA monitoring to this group, across the taxonomic range from tiny planktonic crustaceans to large bivalves, opens a significant new capability. But only if the surveys are designed with an honest understanding of how far the signal travels before it disappears — and that the answer is not one number, but one for each species, each season, each river. This lecture was created by ennepō. Go to https://ennepo.ai to Discover, Create and Follow the latest research in your field. Read when you can. Listen when you want to.