Metabarcoding of storage ethanol vs. conventional morphometric identification in relation to the use of stream macroinvertebrates as ecological indicators in forest management

Abstract

Aquatic macroinvertebrate communities are often used to assess the ecological integrity of streams. However, conventional methods involving morphometric identification of macroinvertebrates are usually costly and time-consuming. Here we compare stream macroinvertebrate community metrics based on conventional morphometrics vs. non-destructive DNA metabarcoding from storage ethanol to assess forest management impacts on headwater streams across a gradient of intensively managed forest catchments in eastern Canada. The two approaches demonstrated substantial congruence in the detection of taxa (81% and 69% at the family and genus level, respectively) and in the characterization of community composition and richness. However, DNA metabarcoding from preservative ethanol identified significantly fewer genera (3.3 on average, 15.9%) and families (2.0, 11.5%) than conventional morphometrics. Taxa missed by metabarcoding of storage ethanol were typically those low in proportional mass or poorly represented in the CO1 reference database. This led to some differences in the explanatory variables identified as being related to macroinvertebrate metrics, which could have implications on conclusions and management actions that might result therefrom. For example, the negative relationships between richness and reach-scale variables associated with forest management intensity were weaker when richness was based on metabarcoding as compared to conventional morphometrics. Discriminatory power was greater when data at the genus level were used. The congruence between functional feeding group results derived from morphometric (based on relative abundance) vs. metabarcoding (based on relative frequency and read abundance) identifications was group specific (r = 0.16–0.63), but low overall. We conclude that DNA metabarcoding of storage ethanol provides a promising approach for characterizing stream macroinvertebrate communities, but that its full deployment in biomonitoring projects requires developing more complete reference libraries and enhancing the sensitivity for detecting taxa with low sample biomass.