eDNA: measuring diversity within a fish species

A fish inventory asks which species have left detectable traces in the water. Population genetics asks how individuals within a species differ. Environmental DNA can contribute to the second question, but detecting a variant, estimating its frequency and showing that this frequency has changed require different evidence.

Reviews by Adams and colleagues in 2019 and Andres and colleagues in 2023 describe the opportunities and analytical difficulties. A 2021 study by Jensen and colleagues illustrates the recovery of mitochondrial and nuclear information from marine fish eDNA. These papers help interpret results while leaving the monitoring method to be validated for its intended use.

Species identification and genetic variation need different resolution

A marker suited to distinguishing species may not contain enough variation to distinguish populations of the same species. Several similar sequences can represent genuine variation, but amplification, sequencing and assignment errors can also generate apparent differences. Marker selection therefore needs to follow the question being asked.

A haplotype is a combination of variants carried together within a sequence or genetic region. Multiple mitochondrial haplotypes may reveal variation among maternal lineages. They are not complete descriptions of separate individual genomes, and one shared haplotype does not reveal how many fish carry it.

An allele is a version of a locus, the genetic position under consideration. Counting detected alleles is different from estimating their frequencies in a population. That distinction is particularly important when a water sample contains unequal contributions from an unknown number of individuals.

The sample is a mixture with unequal contributions

Fish do not all release the same amount of DNA at the same time. Transport, dilution and degradation then influence what reaches the filter. The proportion of sequencing reads assigned to a variant cannot automatically be interpreted as the proportion of fish carrying it.

The 2019 and 2023 reviews discuss these difficulties, including errors that create false variants or mask genuine rare ones. Controls, replication and explicit filtering rules help assess reliability. Stricter filtering can also remove true low-abundance variants, so the shortest final list is not necessarily the most biologically accurate.

Failure to detect a rare sequence may originate in field sampling or laboratory analysis. It does not establish that the variant has disappeared from the population. Comparing campaigns requires information about detection capability as well as the lists of sequences recovered.

A marine target-capture study shows both possibilities and limits

Jensen and colleagues analysed two one-litre water samples collected in Qatar within an aggregation of more than fifty whale sharks, Rhincodon typus. They used target capture, in which probes enrich selected sequences before analysis. This high local concentration of animals differs from a sparsely distributed species in a river.

The researchers recovered data spanning the mitochondrial genome and variation consistent with previously known haplotypes. They also retrieved nuclear sequences, but whale shark coverage was insufficient for detailed population genetic inference. The paper explicitly discusses the possibility that retained nuclear variants included analytical errors.

This is a demonstration of technical capacity under particular conditions. It does not show that every filtered water sample yields a complete genome or that read counts provide a fish census. Recovery of non-target sequences also underlines the need to check probe specificity and sequence assignment.

The example concerns a fish species; it should not be extended automatically to all aquatic animals or every sampling environment. Its practical lesson is to ask which type of genetic information was recovered, at what coverage and with what remaining uncertainty.

Why nuclear markers can add information

Mitochondrial DNA is often targeted because it occurs in multiple copies and useful reference sequences are available. In the fish considered here it primarily represents maternal inheritance. Limited mitochondrial variation does not, on its own, describe the full diversity of the nuclear genome.

Nuclear markers can broaden the information, but recovery from environmental mixtures creates further challenges involving quantity, coverage and assignment. Andres and colleagues also discuss genetic mixture models for estimating contributions. Results obtained through simulations or controlled systems should not be described as a counting method already validated across all natural populations.

Marker choice, reference data and statistical methods need to be considered together. A dataset sufficient to detect a species may be unsuitable for examining population structure, relatedness or individual diversity. These purposes cannot be substituted after sequencing simply because genetic data are available.

Showing genetic erosion requires a justified comparison

Genetic erosion implies a decline, so it requires a temporal reference or another defensible comparison framework. A single measurement cannot demonstrate that process. Comparable campaigns need to account for location, season, temperature, hydrology, preservation and analytical methods.

Field replicates and laboratory replicates have different purposes. Field replication helps describe environmental variability, while laboratory replication captures part of the analytical variability. Repeating an assay is not a substitute for independent water collections. A method change between years requires an assessment of comparability before an apparent decline is interpreted biologically.

The final report should name the measurement: variant detection, observed richness, an estimated frequency or a diversity indicator. It should retain uncertainty and identify missing information. A signal may justify further investigation without already proving a loss of genetic diversity or identifying its cause.

Conclusion and Vetofish support

eDNA can reveal variation within fish species, provided the level of evidence remains explicit. Detecting a variant, measuring population diversity and demonstrating its decline are different objectives requiring different validation.

Vetofish’s advice and support service can help environmental teams define their question and discuss an appropriate plan with the analytical laboratory. Choosing a method that can inform the intended management decision should come before interpreting a list of sequences as evidence of genetic erosion.