Figure 4. Evidence of amplification bias in quantification of fish taxa in a complex mixture. Proportion of sequencing reads for each reference taxon added to the fish mixture identified to either species or genus. Reads could not be attributed unambiguously for family-level matches. Each point represents a single taxon, which were added to the fishmeal mixture in one of six discrete proportions (% tissue in fishmeal; colors). Lines connect the same taxon across three experimental feeds composed of different proportions fishmeal and filler (10% fishmeal = 90% filler). Taxa that were not detected at all three % fishmeal levels are not displayed (e.g., some taxa at 0.01% tissue).