Inferring abundance from detection frequency with metabarcoding

Here we assessed the potential for replicate samples to allow for the inference of abundance data from DNA metabarcoding of macroinvertebrates from benthic kick net samples in New Brunswick, Canada. This was done by comparing field samples where fish were excluded by closed cages at a site, to control samples with open cages. A total of 31 samples were collected at each of 5 sampling dates. Multi-species models trained with replicate detections of 184 genera and 318 species were able to identify seasonal and experimental (predator exclusion) shifts in abundance, using samples with morphologically identified specimens to validate.