Abstract
Macroinvertebrates are widely used in freshwater biomonitoring, but current methods are labor-intensive and require taxonomic expertise. DNA metabarcoding offers a potential solution, though results vary per applied sampling protocol. Here, we compare several protocols using macroinvertebrate samples from a Dutch peatland. Live-sorted specimens were (i) identified morphologically followed by (ii) non-destructive (soft-lysis) and (iii) destructive DNA extraction protocols (aggressive-lysis). Additionally, (iv) unsorted samples (including substrate and plant material) were homogenized, and (v) water-derived eDNA collected. Only 25% of macroinvertebrate taxa were detected across all methods. Aggressive-lysis of live-sorted macroinvertebrates provided a community composition most comparable to the morphology-based identification (mean ± sd = 70 ± 6%), followed by soft-lysis (58 ± 7%). Some beetle taxa were missed using soft-lysis, likely due to insufficient DNA release. Unsorted-debris and eDNA samples showed lower similarity to morphological identifications, with only 31 ± 9% and 20 ± 9% community overlap, respectively. These results suggest aggressive-lysis best replicates traditional assessments, while soft-lysis offers a non-destructive alternative with some taxonomic bias. Unsorted samples and eDNA approaches capture additional diversity but show lower overlap with traditional methods. Our results underscore the importance of protocol selection in DNA-based biomonitoring of freshwater macroinvertebrates.