Plant–orthoptera network reconstruction
The reconstruction of plant–insect trophic networks from fecal samples relies on a DNA metabarcoding procedure that uses a two-step DNA amplification PCR-based approach in which samples are individually tagged by dual-indexing. A full protocol of the wet-lab procedure, from DNA extraction to sequencing, is provided in Supporting information, Appendix S2, section Methods 1. In short, after DNA extraction from the insect feces, the ITS2 nuclear plant marker (360bp) was amplified in the amplicon PCR. We selected this marker based on its ease of amplification, high taxonomic resolution, good coverage of the reference database and successful application to degraded DNA samples (Li et al. 2011; García-Robledo et al. 2013). In parallel to DNA metabarcoding library preparation and sequencing, we compiled an ITS2 reference database by recovering sequences from Genbank (Clark et al. 2016). The database was filtered using in-silico PCRs that allow only one mismatch between the primers and the priming sites. We expanded the reference database with custom sequences generated for 54% of the plant species (see Appendix S2, section Methods 2). After processing of the raw sequencing data, OTU calling and taxonomic assignment against the DNA reference database, the OTU table was streamlined to reconstruct individual networks using R (R Core Team 2019; see complete descriptions of the bioinformatic and OTU table cleaning procedures in Appendix S2, section Methods 3). We discarded OTUs that were non-monophyletic and/or identified above the family level, summed the OTUs belonging to the same taxon, used vegetation surveys to filter out the OTUs with low read numbers, and redistributed the count of OTUs assigned above the species level. In a final step, we computed the relative read abundance (RRA) for each sample as an estimate of the interaction intensity (Deagleet al. 2019, Roslin et al. 2019).