Assessing diet variability is of main importance to better understand the biology of bats and design conservation strategies. Although the advent of metabarcoding has facilitated such analyses, this approach does not come without challenges. Biases may occur throughout the whole experiment, from fieldwork to biostatistics, resulting in the detection of false negatives, false positives or low taxonomic resolution. We detail a rigorous metabarcoding approach based on a short COI minibarcode and two-step PCR protocol enabling the "all at once" taxonomic identification of bats and their arthropod prey for several hundreds of samples. Our study includes faecal pellets collected in France from 357 bats representing 16 species, as well as insect mock communities that mimic bat meals of known composition, negative and positive controls. All samples were analysed using three replicates. We compare the efficiency of DNA extraction methods, and we evaluate the effectiveness of our protocol using identification success, taxonomic resolution, sensitivity and amplification biases. Our parallel identification strategy of predators and prey reduces the risk of mis-assigning prey to wrong predators and decreases the number of molecular steps. Controls and replicates enable to filter the data and limit the risk of false positives, hence guaranteeing high confidence results for both prey occurrence and bat species identification. We validate 551 COI variants from arthropod including 18 orders, 117 family, 282 genus and 290 species. Our method therefore provides a rapid, resolutive and cost-effective screening tool for addressing evolutionary ecological issues or developing "chirosurveillance" and conservation strategies.

Aphids constitute a diverse group of plant-feeding insects and are among the most important crop pests in temperate regions. Their morphological identification is time-consuming and requires specific knowledge, training and skills that may take years to acquire. We assessed the advantages and limits of DNA barcoding with the standard COI barcode fragment for the identification of European aphids. We constructed a large reference dataset of barcodes from 1020 specimens belonging to 274 species and 87 genera sampled throughout Europe and set up a database-driven website allowing species identification from query sequences.ResultsIn this unbiased sampling of the taxonomic diversity of European aphids, intraspecific divergence ranged from 0.0% to 3.9%, with a mean value of 0.29%, whereas mean congeneric divergence was 6.4%, ranging from 0.0% to 15%. Neighbor-joining analysis generated a tree in which most species clustered in distinct genetic units. Most of the species with undifferentiated or overlapping barcodes belonged to the genus Aphis or, to a lesser extent, the genera Brachycaudus, Dysaphis and Macrosiphum. The taxa involved were always morphologically similar or closely related and belonged to species groups known to present taxonomic difficulties.ConclusionsThese data confirm that COI barcoding is a useful identification tool for aphids. Barcode identification is straightforward and reliable for 80% of species, including some difficult to distinguish on the basis of morphological characters alone. Unsurprisingly, barcodes often failed to distinguish between species from groups for which classical taxonomy has also reached its limits, leading to endless revisions and discussions about species and subspecies definitions. In such cases, the development of an effective procedure for the accurate identification of aphid specimens continues to pose a difficult challenge.