Abstract
Understanding interactions between herbivores and parasitoids is essential for successful biodiversity protection and monitoring and for biological pest control. Morphological identifications employ insect rearing and are complicated by insects' high diversity and crypsis. DNA barcoding has been successfully used in studies of host-parasitoid interactions as it can substantially increase the recovered real host-parasitoid diversity distorted by overlooked species complexes, or by species with slight morphological differences. However, this approach does not allow the simultaneous detection and identification of host(s) and parasitoid(s). Recently, high-throughput sequencing has shown high potential for surveying ecological communities and trophic interactions. Using mock samples comprising insect larvae and their parasitoids, we tested the potential of DNA metabarcoding for identifying individuals involved in host-parasitoid interactions to different taxonomic levels, and compared it to standard DNA barcoding and morphological approaches. For DNA metabarcoding, we targeted the standard barcoding marker cytochrome oxidase subunit I using highly degenerate primers, 2*300 bp sequencing on a MiSeq platform, and RTAX classification using paired-end reads. Additionally, using a large host-parasitoid dataset from a Central European floodplain forest, we assess the completeness and usability of a local reference library by confronting the number of Barcoding Index Numbers obtained by standard barcoding with the number of morphotypes. Overall, metabarcoding recovery was high, identifying 92.8% of the taxa present in mock samples, and identification success within individual taxonomic levels did not significantly differ among metabarcoding, standard barcoding, and morphology. Based on the current local reference library, 39.4% parasitoid and 90.7% host taxa were identified to the species level. DNA barcoding estimated higher parasitoid diversity than morphotyping, especially in groups with high level of crypsis. This study suggests the potential of metabarcoding for effectively recovering host-parasitoid diversity, together with more accurate identifications obtained from building reliable and comprehensive reference libraries, especially for parasitoids.