Double-blind validation of alternative wild bee identification techniques: DNA metabarcoding and in vivo determination in the field

Abstract

Over the past few decades, several investigations around the globe have reported alarming declines in the abundance and diversity of bee species. The success of effective conservation strategies targeting these important pollinators relies heavily on accurate biodiversity assessments. The shortage of taxonomic experts and the escalation of the ongoing biodiversity crisis call for the development of alternative identification tools to implement efficient monitoring programs. The validation of such techniques is crucial to ensure that they provide results comparable to those of traditional morphotaxonomy. Here we performed two double-blind experiments to evaluate the accuracy of a pair of new techniques used for wild bee identification: DNA metabarcoding and in vivo identification in the field. The methods were tested on sets of wild bees from Germany and their results compared against evaluations done by panels of bee experts using traditional morphotaxonomy. On average the congruency of species identification between metabarcoding and morphotaxonomy was 88.98% across samples (N = 10), while in vivo identification and morphotaxonomy were 91.81% congruent (N = 7) for bees considered feasible for in vivo identification in the field. Traditional morphotaxonomy showed similar congruencies when compared to itself: 93.65% in the metabarcoding study and 92.96% in the in vivo study. Overall, these results support both new methods as viable alternatives to traditional microscopy-based assessment, with neither method being error-free. Metabarcoding provides a suitable option to analyze large numbers of specimens in the absence of highly trained taxonomic experts, while in vivo identification is recommended for repeated long-term monitoring, and when working in areas where the sampling of individuals could threaten local populations of endangered wild bee species. Further research is still needed to explore the potential of both techniques for conservation management and wildlife monitoring, as well as to overcome their current limitations as taxonomic tools.

Comparing modern identification methods for wild bees: Metabarcoding and image-based morphological taxonomic assignment

“…Identification through DNA metabarcoding relies on comparison to sequences in currently accessible databases [ 12 , 13 , 23 , 52 ], which is lacking for some taxa. Most sequences can currently be attributed to genus level; however, assignments will improve as additional sequences are added to public databases ( S2 Table ).…”

“…The hypotheses addressed in this study included: (1) there would be general agreement among the two identification methods, (2) the morphological identification would provide definitive information on the presence of specific taxa, and (3) the molecular data would resolve some cryptic species that were indiscernible from the images using the photographic approach. This study assessed non-traditional identification methods that make efficient use of collected bee specimens and discusses the most applicable identification method for future field studies of wild bees [ 13 ].…”

DNA barcoding insufficiently identifies European wild bees (Hymenoptera, Anthophila) due to undefined species diversity, genus‐specific barcoding gaps and database errors

“…This is an expected mistake type in large invertebrate inventories and was also detected in previous bee-inventories. For example, Herrera-Mesías et al (2022) found an 11% mismatch between morphological and metabarcoding determinations on average, with differences among experts, where 3% of these differences were attributed to false positives due to environmental contaminations and 1.3% to false negatives due to insufficient sequencing depth. A similar identification mismatch of 10%-20% was found in other regional bee inventories (e.g., Gibbs, 2018;Herrera-Mesías et al, 2022;Magnacca & Brown, 2012;Schmidt et al, 2015).…”

“…For example, Herrera-Mesías et al (2022) found an 11% mismatch between morphological and metabarcoding determinations on average, with differences among experts, where 3% of these differences were attributed to false positives due to environmental contaminations and 1.3% to false negatives due to insufficient sequencing depth. A similar identification mismatch of 10%-20% was found in other regional bee inventories (e.g., Gibbs, 2018;Herrera-Mesías et al, 2022;Magnacca & Brown, 2012;Schmidt et al, 2015). This mismatch is probably higher in countries with the largest gaps in DNA data availability, that is, southern and eastern Europe (Leclercq et al, 2023).…”

DNA metabarcoding and morphological identification reveal similar richness, taxonomic composition and body size patterns among flying insect communities

“…In addition, our understanding of how well metabarcoding represents the composition of flying insect communities compared with conventional morphological methods is still incomplete. Several studies have shown that metabarcoding can perform similarly to morphological approaches (Elbrecht, Vamos, et al, 2017; Herrera‐Mesías et al, 2022; Liu et al, 2020), but the majority of studies come from aquatic ecosystems (Buchner, Macher, et al, 2021; Elbrecht & Leese, 2015; Groendahl et al, 2017; Mora et al, 2019; Steyaert et al, 2020; Zimmermann et al, 2015) or terrestrial soil fauna (Basset et al, 2022; Oliverio et al, 2018). Moreover, most studies only investigate a small subset of defined target species or higher taxonomic levels (e.g., family level or genus level), necessitating examinations of how well metabarcoding performs across the entire community at species level.…”