In silico and empirical evaluation of twelve metabarcoding primer sets for insectivorous diet analyses.

Orianne Tournayre,Marine Trillat,Nathalie Charbonnel,Maxime Leuchtmann,Maxime Galan,Sylvain Piry,Ondine Filippi-Codaccioni,Dominique Pontier

doi:10.1002/ece3.6362

Abstract

During the most recent decade, environmental DNA metabarcoding approaches have been both developed and improved to minimize the biological and technical biases in these protocols. However, challenges remain, notably those relating to primer design. In the current study, we comprehensively assessed the performance of ten COI and two 16S primer pairs for eDNA metabarcoding, including novel and previously published primers. We used a combined approach of in silico, in vivo‐mock community (33 arthropod taxa from 16 orders), and guano‐based analyses to identify primer sets that would maximize arthropod detection and taxonomic identification, successfully identify the predator (bat) species, and minimize the time and financial costs of the experiment. We focused on two insectivorous bat species that live together in mixed colonies: the greater horseshoe bat (Rhinolophus ferrumequinum) and Geoffroy's bat (Myotis emarginatus). We found that primer degeneracy is the main factor that influences arthropod detection in silico and mock community analyses, while amplicon length is critical for the detection of arthropods from degraded DNA samples. Our guano‐based results highlight the importance of detecting and identifying both predator and prey, as guano samples can be contaminated by other insectivorous species. Moreover, we demonstrate that amplifying bat DNA does not reduce the primers' capacity to detect arthropods. We therefore recommend the simultaneous identification of predator and prey. Finally, our results suggest that up to one‐third of prey occurrences may be unreliable and are probably not of primary interest in diet studies, which may decrease the relevance of combining several primer sets instead of using a single efficient one. In conclusion, this study provides a pragmatic framework for eDNA primer selection with respect to scientific and methodological constraints.

Highlights

The genetic analysis of environmental samples such as soil, water, or feces, known as environmental DNA metabarcoding, is a rapid and cost-effective tool for the study of species that are difficult to detect or monitor (Bohmann et al, 2014). This approach allows the simultaneous identification of multiple taxa in environmental samples, bypassing the need to isolate organisms prior to identification (Clare, 2014; Taberlet, Coissac, Hajibabaei, & Rieseberg, 2012). eDNA metabarcoding is of particular interest in dietary analysis of rare or elusive species, and this approach has been applied to a large spectrum of organisms (Clare, Fraser, Braid, Fenton, & Hebert, 2009; Corse et al, 2017; Kartzinel & Pringle, 2015; Rytkönen et al, 2019; Shehzad et al, 2012)
We have shown that some of the guano sampled in the roosts of mixed colonies (R. ferrumequinum/M. emarginatus) was contaminated with excreta belonging to other bat species, including M. myotis, for which a few individuals were known to be present in the studied colony
Our results revealed that there is no effect of the percentage of bat reads on the percentage of arthropod taxa detected in mock communities, nor on the number of arthropod occurrences in guano samples

Summary

Introduction

The genetic analysis of environmental samples such as soil, water, or feces, known as environmental DNA (eDNA) metabarcoding, is a rapid and cost-effective tool for the study of species that are difficult to detect or monitor (Bohmann et al, 2014). This approach allows the simultaneous identification of multiple taxa in environmental samples, bypassing the need to isolate organisms prior to identification (Clare, 2014; Taberlet, Coissac, Hajibabaei, & Rieseberg, 2012). Many methodological improvements have been made to limit some of these biases and to introduce best-practice guidelines for metabarcoding protocols, such as the systematic inclusion of technical replicates and negative controls (Alberdi, Aizpurua, Gilbert, & Bohmann, 2018; Corse et al, 2017; Elbrecht & Steinke, 2019; Galan et al, 2016; Mata et al, 2018)

Objectives

Findings

Discussion

Conclusion