In silico and empirical evaluation of twelve COI & 16S metabarcoding primer sets for insectivorous diet analyses

Abstract

This last decade, environmental DNA metabarcoding approaches have been developed and improved to minimize biological and technical biases; some challenges, however, remain, as the design of primers. Here we have performed a comprehensive assessment of ten COI and two 16S primer sets. We have combined in silico, in vivo-mock community of 33 arthropod taxa from 16 orders and guano analyses to identify primer sets that should maximize arthropod detection and taxonomic identification, whilst identifying bat species and minimizing labour time and cost. We have focused on two insectivorous bat species living in mixed-colonies, the greater horseshoe bat (Rhinolophus ferrumequinum) and Geoffroys bat (Myotis emarginatus). We have found that the level of primer degeneracy is the main factor influencing arthropod detection for in silico and mock community analyses, while the amplicon length is critical for the detection of arthropods from degraded DNA samples. Our results confirm the importance of performing predator detection and taxonomic identification, simultaneously with arthropod sequencing, as faeces samples can be contaminated by different insectivorous species. Moreover, amplifying bat DNA does not affect the primers capacity to detect arthropods. We therefore recommend the systematic simultaneous identification of predator and prey. Finally, we evidenced that one third of the prey occurrences are unreliable and probably not of primary interest in diet studies, which might decrease the relevance of combining several primer sets instead of using one efficient primer set. In conclusion, this study provides general criteria enabling the selection of primers whilst considering different scientific and methodological constraints.