Discussion
We found that a portfolio of four markers targeting three different barcoding genes identified as many of our set of 103 marine and freshwater taxa as possible to the species level and that no additional taxa could be identified by applying an additional 15 markers to our DNA reference pool. These same four markers identified the full set of 103 taxa to the family level. COI markers recovered the greatest number of reference taxa and at the highest taxonomic resolution, but markers for 12S and 16S identified unique taxa missed by COI. These patterns offer strong evidence of the benefits of a portfolio approach, but also suggest that primer portfolios may need to be optimized for each target taxonomic group or geographic region rather than considered broadly transferable.
While we found an overall tendency towards lower read counts for taxa added in lower input amounts, the patterns were highly inconsistent. Accordingly, our read counts could not reliably be interpreted as even a semi-quantitative proxy for relative abundance of a species in our experimental tissue mixture, highlighting a key limitation for PCR-based assays. Detection (presence/absence) of reference taxa from this complex mixture became inconsistent with <1% of input tissue, which points to a threshold of ~1% representation for robust PCR-based detection in a heterogeneous sample. Interestingly, the likelihoods of detecting species across the range of relative abundance in the mixture was independent of whether fish tissues were analyzed alone or after being diluted with either a plant-based or plant/animal filler, probably thanks to the relative affinity of the markers tested for fishes.