Maximizing Identification Precision of Hymenoptera and Brachycera (Diptera) With a Non-Destructive DNA Metabarcoding Approach

In recent years, DNA metabarcoding has been used for a more efficient assessment of bulk samples. However, there remains a paucity of studies examining potential disparities in species identification methodologies. Here, we explore the outcomes of diverse clustering and filtering techniques on data from a non-destructive metabarcoding approach, compared to species-level morphological identification of Brachycera (Diptera) and Hymenoptera of two bulk samples collected with Malaise traps. The study evaluated four distinct approaches, namely clustering to Amplicon Sequence Variants (ASVs) or ASVs clustered to Operational Taxonomic Units (OTUs) coupled with subsequent filtering using the LULU algorithm at 84% and 96% minimum match. In total, 114 species of Brachycera (35 families) and 85 species of Hymenoptera (27 families) were identified morphologically. Depending on the selected approach, DNA metabarcoding results strongly varied in terms of detected molecular units blasted to brachyceran and hymenopteran species. For Brachycera, ASVs clustered into OTUs followed by LULU using a 96% minimum match (OTU96) inferred the number of molecular units closest to the number of morphologically identified species. Using Syrphidae as an exemplary family, we found an overlap ranging from 9% to 81% between the morphological identification and the different clustering and filtering approaches, OTU96 being also here the closest one. For Hymenoptera, while OTU96 also yielded the highest number of molecular units, it was still considerably low compared to the number of morphologically identified species. Our results show that metabarcoding methodology needs to be significantly improved to be applied to Hymenoptera. Conversely, for Brachycera, we acknowledge the promise of employing a non-destructive metabarcoding approach, incorporating ASV clustering into OTUs and filtering with LULU, to derive dependable species lists. Such lists hold significant potential for applications in biomonitoring, conservation efforts, and other related fields.