The Gold Standard of Ecology | DNA Degradation, Hydrodynamic Transport and Source-Area Attribution

1. Define the ecological source question before sampling

Back to contents

Decide whether the study asks for local presence, upstream catchment occupancy, recent passage, relative source strength or the location of an unknown release. These targets require different designs. A shoreline biodiversity survey can tolerate broad integration; a management search for an invasive population needs tighter spatial attribution. State the time window and water body represented by one sample.

2. Treat eDNA as a mixture of physical states

Back to contents

Environmental DNA can occur in cells, organelles, tissue fragments, particles, aggregates, free molecules or material attached to sediment. These states settle, resuspend, filter and degrade at different rates. Filter pore size and extraction method select part of the mixture. A concentration is therefore a method-dependent signal, not a universal count of molecules moving identically through water.

3. Measure hydrodynamics at the scale of the claim

Back to contents

Record discharge, velocity, depth, cross-section, stratification, wind, tide, tributaries and barriers as relevant. Straight-line distance is a poor transport metric when currents curve, water masses mix unevenly or retention zones store particles. Sampling only the centreline can miss bank storage or a vertical plume. Pair eDNA with tracers or a calibrated flow model when source attribution is central.

4. Model degradation without pretending one half-life is universal

Back to contents

Decay depends on temperature, microbial activity, salinity, ultraviolet exposure, particle protection and the target fragment. Laboratory rates can bound mechanisms but may not transfer directly to field conditions. Use time-series experiments in relevant water, report detection limits and distinguish loss from the water column through degradation, settling, adsorption and dilution. Several processes can produce the same declining concentration.

5. Design spatial and temporal replication around transport time

Back to contents

Place stations upstream, at plausible sources and along travel-time paths rather than equal map distances alone. Repeat across flow conditions and collect concurrent field blanks and controls. A single downstream detection cannot reveal whether the source was brief and near, sustained and far, or resuspended from storage. Travel-time sampling and repeated pulses are more informative than one snapshot.

6. Worked example: an upstream mussel signal after rain

Back to contents

A threatened mussel is detected at a bridge but not at two nearby bank sites. The first interpretation places a population within hundreds of metres. Rainfall records show a high-flow pulse, and a tributary eight kilometres upstream contains known habitat. A conservative tracer and discharge model place its water at the bridge during sampling, while sediment cores near the bridge hold old DNA. The team reports catchment-scale evidence, adds travel-time stations and postpones a local population claim.

7. Fit source, transport and observation together

Back to contents

A useful model links source shedding, advection, dispersion, settling or resuspension, degradation, sampling volume, extraction recovery and assay detection. Parameters may be weakly identifiable from eDNA alone, so use priors or independent measurements transparently. Estimate a distribution of plausible source areas rather than selecting the map cell with the highest fitted concentration. Validate with held-out stations or known releases.

8. Changed-context transfer check: a stratified lake

Back to contents

Move from a river with mostly downstream transport to a lake influenced by wind, density layers and internal circulation. The transferable elements are source timing, movement, loss and observation. The changed system needs three-dimensional currents, vertical sampling and sensitivity to particle depth. A river decay-distance rule cannot identify a lake shoreline source, and surface detections may miss a plume moving below the thermocline.

9. Use negative and positive evidence cautiously

Back to contents

A non-detection can reflect low source strength, dilution, degradation, poor transport to the station, inhibition or limited sampling. A detection can integrate distant sources. Occupancy models or repeated-sample detection models handle laboratory and field uncertainty, but source attribution also needs a movement model. Distinguish confidence that DNA was present in the sample from confidence about where organisms were.

10. Know the limits and common misconceptions

Back to contents

eDNA does not always indicate a living organism at the sampling point, and concentration is not automatically abundance. Faster flow can increase transport distance while also changing dilution, suspension and sampling. A fitted decay constant is not the same as removal from the entire system. Hydrodynamic models add assumptions and calibration needs; they do not turn every sequence into an exact origin coordinate.

11. Audit the field-to-source inference chain

Back to contents

Publish station coordinates and depths, times, discharge and weather, sample volumes, filters, controls, extraction and assay details, detection thresholds, particle or tracer assumptions, hydrodynamic inputs, decay priors, source scenarios and code. Map posterior or ensemble uncertainty and show sensitivity to transport, degradation and resuspension choices. Preserve raw detections so future models can reanalyse the evidence.

12. The gold standard reports a source area, not false precision

Back to contents

Environmental DNA can connect molecular evidence with landscape-scale ecology when transport and decay are part of the question. The gold standard designs stations around water movement, distinguishes physical DNA states, estimates detection loss, tests alternative sources and reports the area and time window supported by evidence. A precise pin is not progress if the water could have carried the signal from somewhere else.