Abstract
The diets of pelagic marine larvae are difficult to analyze due to their small size and even smaller prey. Furthermore, different methods may lead to alternative interpretations of trophic interactions. Conventionally, diet studies have relied primarily on visual identification of prey through dissection and microscopy. While microscopy has clear benefits, it can yield an incomplete assessment of diet since smaller and soft-bodied prey items are often difficult to identify. Here, we combined conventional microscopy and two contemporary environmental DNA (eDNA) methods: DNA metabarcode sequencing (metabarcoding) and real-time Polymerase Chain Reaction (rtPCR), comparing their advantages/disadvantages in a diet analysis of planktonic American lobster (Homarus americanus) postlarvae. This is the first application of these molecular techniques on the postlarval lobster diet. We also describe the testing and development of a novel blocking primer designed to inhibit the amplification of lobster DNA, enhancing prey amplification. This approach allowed finer-scale identification of a greater variety of prey than microscopy. The targeted rtPCR approach identified a specific prey taxon with high fidelity - but involves a priori decisions regarding the choice of target. Here, an rtPCR assay was developed to target Calanus finmarchicus, an abundant copepod species in the Gulf of Maine, suspected to be an important prey item of larval lobsters. Microscopy revealed broad prey categories and the importance of arthropod prey in the postlarval diet. Metabarcoding confirmed the importance of arthropod prey, while filling in the gaps with additional prey species. Finally, rtPCR was able to detect a significant level of predation on Calanus finmarchicus that neither of the other two approaches identified. The combination of methods provided a richer understanding of diet than any single method alone and future diet studies of a wide range of consumers would benefit from the application of a mixture of microscopy and molecular-based methodologies.