The Gold Standard of Ecology | Cryptic Species Complexes, DNA Barcoding Thresholds and Richness Sensitivity

1. Separate identification, discovery and formal delimitation

Back to contents

Identification assigns a specimen to an existing reference. Discovery flags unexpected genetic structure. Species delimitation evaluates whether lineages warrant separate species hypotheses, ideally using independent evidence. These are related but not interchangeable tasks. A barcode that fails to match a library may reflect an undescribed lineage, a missing reference, contamination, sequencing error or geographic variation within one species.

2. Understand what a threshold does

Back to contents

A distance threshold groups sequences that differ by less than a chosen amount or separates candidate clusters across a barcode gap. It does not observe reproductive isolation or ecological independence directly. A universal percentage is risky because mutation rates and within-species diversity differ among taxa. Report the algorithm, marker, distance model and threshold, and explain whether the rule was calibrated for the focal group.

3. Build a reference set with vouchers and geography

Back to contents

Sparse reference libraries can make common variation look like a new species, while misidentified records can anchor the wrong name. Retain voucher specimens, collection coordinates, photographs, sequence quality and determinations by taxonomic experts where possible. Sample across the focal species’ range and include close relatives. Local ecological samples alone are rarely enough to estimate the full within-species distribution.

4. Inspect the barcode gap rather than assuming it

Back to contents

Plot within- and between-cluster distances, nearest-neighbour relationships and the effect of uneven sample size. A gap may be local to a group even when no global threshold separates every species. Deep mitochondrial splits can occur within one species, and recent species can share barcodes. Test several thresholds and algorithms. Where conclusions change abruptly, the instability is a result to report.

5. Propagate alternative taxonomies into richness

Back to contents

Create at least three community matrices when appropriate: conservative named species, barcode clusters under the primary rule and plausible alternative delimitations. Recalculate alpha richness, beta diversity, occupancy and trend estimates. Splitting a widespread name can raise local richness, increase apparent endemism and alter turnover; lumping can do the opposite. The ecological claim should show which findings survive those choices.

6. Worked example: one stream insect becomes four clusters

Back to contents

Surveys record one named mayfly at twelve streams. COI sequences form four clusters under a 2% threshold, each concentrated in a different catchment. Treating clusters as species raises regional richness from 18 to 21 and makes between-catchment turnover much larger. At 3%, two clusters merge and the turnover result weakens. Nuclear loci support two lineages, while larval morphology remains ambiguous. The defensible conclusion is taxonomically sensitive catchment structure, not four confirmed new species.

7. Distinguish biological structure from technical artefact

Back to contents

Nuclear mitochondrial pseudogenes, heteroplasmy, contamination, sequencing errors and alignment problems can create spurious divergence. Introgression and incomplete lineage sorting can hide real species boundaries. Check chromatograms or read quality, stop codons where relevant, laboratory controls and replicate sequences. Compare mitochondrial results with nuclear, morphological, ecological or behavioural evidence rather than using more decimal places as a substitute for independent support.

8. Changed-context transfer check: metabarcoding an eDNA sample

Back to contents

A water sample contains many short sequence reads rather than individually vouched specimens. Operational or amplicon sequence units are filtered and matched to a reference library. The transferable rule is to vary clustering or assignment thresholds and propagate identity uncertainty. The changed context adds read error, copy-number differences and incomplete reference coverage, so a read cluster should not automatically be counted as one species.

9. Know the limits and common misconceptions

Back to contents

Cryptic does not mean invisible to every method, and barcode difference does not equal species difference. More clusters do not automatically mean better biodiversity estimates. A clean barcode gap can arise from incomplete sampling. Conversely, the absence of a gap does not prove one species. Formal nomenclature, ecological units and conservation units can have different evidential standards, so state which decision the analysis supports.

10. Audit specimens, sequences and ecological consequences

Back to contents

Publish specimen identifiers, vouchers, permits, coordinates at an ethically appropriate precision, primers, quality filters, alignments, reference versions, genetic distances, delimitation settings and cluster membership. Link every ecological observation to its taxonomic unit. Provide sensitivity tables showing community metrics across plausible lumping and splitting schemes, and record which management conclusions change.

11. Frequently asked questions

Back to contents

Is there a universal barcode threshold? No. Can COI alone discover useful structure? Yes, as a hypothesis-generating marker. Does every cluster require a species name? No. Why do richness estimates change? The counting unit changes. Should uncertain specimens be discarded? Not automatically; retain them with explicit confidence and test their influence. What strengthens delimitation? Geographic sampling, vouchers and independent molecular, morphological, ecological or behavioural evidence.

12. The gold standard carries taxonomic uncertainty into ecology

Back to contents

A credible barcode study treats clusters as model-dependent hypotheses, grounds sequences in vouchers and samples the geography needed to interpret divergence. It recalculates ecological results under plausible taxonomies and distinguishes robust patterns from threshold artefacts. The gold standard does not hide uncertainty before counting species; it shows exactly how alternative identities change the ecological story.