Approaches to integrating genetic data into ecological networks

As molecular tools for assessing trophic interactions become common, research is increasingly focused on the construction of interaction networks. Here we demonstrate three key methods for incorporating DNA data into network ecology and discuss analytical considerations using a model consisting of plants, insects, bats and their parasites from the Costa Rican dry forest. The simplest method involves the use of Sanger sequencing to acquire long sequences to validate or refine field identifications, for example of bats and their parasites, where one specimen yields one sequence and one identification. This method can be fully quantified and resolved and these data resemble traditional ecological networks. For more complex taxonomic identifications, we target multiple DNA loci e.g. from a seed or fruit pulp sample in faeces. These networks are also well resolved but gene targets vary in resolution and quantification is difficult. Finally for mixed templates such as faecal contents of insectivorous bats we use DNA metabarcoding targeting two sequence lengths (157bp, 407bp) of one gene region and a MOTU, BLAST and BIN association approach to resolve nodes. This network type is complex to generate and analyse and we discuss the implications of this type of resolution on network analysis. Using these data we construct the first molecular-based network of networks containing 3304 interactions between 762 nodes of 8 trophic functions and involving parasitic, mutualistic, and predatory interactions. We provide a comparison of the relative strengths and weaknesses of these data types in network ecology.