eDNA metabarcoding vs traditional bioassessment for macroinvertebrates

Answer

The three verified studies do not give one direction. In Swiss rivers, kick-net sampling captured higher EPT species richness per site than water eDNA metabarcoding, and the richness of the two methods correlated only weakly (Pearson r = 0.27) a07a2e0a. In a Japanese river basin, eDNA metabarcoding detected 90 species and 51 families against 29 species and 35 families by morphological identification 36701dd7. In French streams, DNA from sorted bulk samples found 127 taxa against 140 by morphology, with the same ecological quality class at 86% of site sampling events 5a454efa. These studies differ in taxon scope, DNA source and taxonomic rank, so they are not three measurements of one quantity.

Summary

Species richness: the sign of the molecular-versus-morphology difference changes between studies. Kick-net exceeded eDNA for EPT only a07a2e0a; both molecular methods exceeded morphology in all-macroinvertebrate species counts in Japan 36701dd7; morphology exceeded bulk-specimen DNA in France 5a454efa. They point in different directions and are not matched on taxon group, DNA source or morphological rank, so the corpus does not say which method gives more richness in general.

Biodiversity index scores: only the French study reports an index-level result, with morphology and bulk-DNA metabarcoding assigning the same ecological quality class at 86% of 18 site sampling events, and combining the markers improving the match of biotic index values 5a454efa. That DNA came from sorted specimens, so it speaks to DNA versus morphological identification, not to water eDNA versus kick-net capture 5a454efa. No verified fact gives an index score for water eDNA.

Evidence

• Keck et al. 2022 - eDNA metabarcoding with two primer sets (fwhF2/EPTDr2n and mlCOIintF/HCO2198) against kick-net sampling for EPT macroinvertebrates at 24 Swiss river sites. Kick-net gave higher per-site EPT richness; eDNA-versus-kick-net richness Pearson r = 0.27; the model of predicted richness correlated at 0.44 with kick-net and 0.30 with eDNA; community composition differed substantially; the EPT-specific primers detected more targeted EPT species than the general primers. a07a2e0a
• Doloiras-Larano et al. 2023 (bioRxiv) - morphology, eDNA and bulk-sample metabarcoding for benthic macroinvertebrates in the Shigenobu River Basin, Japan. Morphology: 8,052 individuals, 35 families, 29 species. eDNA: 51 families, 84 genera, 90 species. Bulk-sample metabarcoding: 37 families, 55 genera, 107 species, the finest species-level resolution. A preprint. 36701dd7
• Meyer et al. 2021 (Molecular Ecology) - 18 sites on 15 French wadeable streams; morphology 140 taxa, DNA metabarcoding of sorted bulk material (COI, 18S and 16S combined) 127 taxa with 99 shared; same ecological quality class at 86% of site sampling events. Abstract only; the field gear is not stated as a kick-net. 5a454efa

Methodology

All three verified facts are used. I compared the studies on the axes that decide whether richness figures are the same quantity: taxon group (EPT only versus all macroinvertebrates), DNA source (water eDNA versus bulk specimens), and the rank to which the morphological arm was identified (species-level EPT, versus 35 families and 29 species). Counts of species, families, genera and taxa were not pooled or converted, and the Swiss correlation, the Japanese counts and the French concordance percentage were not set on one scale.

Limitations

• Only one fact compares water eDNA directly with kick-net sampling, and it covers EPT only a07a2e0a.
• The Japanese study is a preprint and does not state the morphological sampling gear in the fact; its morphology arm is largely family-level 36701dd7.
• The French DNA arm is bulk specimens and the fact rests on an abstract 5a454efa.
• No fact gives a biodiversity index score from water eDNA, so the index half of the question is only partly answered.