“Dark taxonomy”: a new protocol for overcoming the taxonomic impediments for dark taxa and broadening the taxon base for biodiversity assessment

Abstract

We are entering the 6th mass extinction event on the planet with scarcely any data for "dark taxa" that comprise most animal species. These taxa have been neglected, because conventional taxonomic methods are not well-equipped to process tens of thousands of specimens belonging to thousands of species. We here test a new protocol for tackling this data deficiency problem by simultaneously addressing (1) taxonomic impediments, (2) lack of biodiversity baseline data, (3) and low impact of revisionary research. We propose to overcome the taxonomic impediments by carrying out revisions at small geographic scales to keep the number of specimens and species manageable. To lower the risk of taxonomic error, species delimitation and description are based on multiple data sources ("integrative taxonomy"). Secondly, we propose to create baseline data for dark taxa by initially revising them only based on fresh specimens obtained with the same sampling methods that are also used for biodiversity monitoring. This ensures that the revision constitutes a biodiversity baseline and the species most relevant for biomonitoring are revised first. Thirdly, we propose to improve the impact of taxonomic revisions by publishing two papers addressing different readerships (general and specialists). We illustrate our proposals by carrying out a taxonomic revision of the fungus gnats (Diptera: Mycetophilidae) of Singapore based on specimens obtained with Malaise traps placed at 107 sites across different habitats. We show that a first batch of specimens (N=1,456) contains 120 species, of which 115 are new to science and described in a separate taxonomic monograph. Species delimitation started with obtaining NGS barcodes (313-bp) that were used for estimating the number of MOTUs. This revealed 116-129 MOTUs depending on whether PTP, ABGD (P=0.001-0.060), or objective clustering (2-5%) was used. MOTU boundaries were then revised with "LIT" (Large-scale Integrative Taxonomy) which optimizes the integration of morphological and molecular data by using the molecular data. LIT revealed a match ratio of 91% for 3% MOTUs, but an even higher match ratio of 97% was found for MOTUs obtained with 5% Objective Clustering and ABGD (P=0.060). Using these parameters, only two MOTUs had to be fused for perfect congruence between molecular and morphological data for the 120 species. To test the completeness of our monograph, we then analyzed a second batch of 1,567 specimens. All clustering algorithms and settings revealed only 20 additional MOTUs; i.e., the first batch of specimens already covered 85% of the diversity, although 22% and 10% of the species are singletons or doubletons, respectively. Remarkably, >97% of all specimens in the second batch belonged to species described based on the first batch. The study thus demonstrates that revisions of dark taxa at moderate geographic scales with a moderate number of specimens from standardized traps can move dark taxa from being unavailable for biomonitoring to being well suited. In addition, the study increases the number of described species of Mycetophilidae in the entire Oriental Region by >25% indicating the size of taxonomic impediments for dark taxa in Southeast Asia.

Towards holistic insect monitoring: species discovery, description, identification and traits for all insects

“…Barcodes were obtained for all specimens, but the targeted morphological checks only required revisiting just over 5% of specimens. A second successful validation of LIT was recently undertaken by Meier et al [37] for 1456 specimens of Mycetophilidae (Diptera) which were shown to contain 120 species based on a LIT analysis. Compared to the first test based on phorid flies, a larger proportion of specimens had to be checked for MOTU validation because studying only one specimen for each of the 120 species already meant inspecting greater than 8% of all specimens in the sample (n = 1456).…”

“…Compared to the first test based on phorid flies, a larger proportion of specimens had to be checked for MOTU validation because studying only one specimen for each of the 120 species already meant inspecting greater than 8% of all specimens in the sample ( n = 1456). Still, LIT was again found to be effective at resolving conflict between MOTUs and morphology by studying the morphology of only a small subset of all specimens [ 37 , 38 ].…”

“…When Meier et al ., analysed a second set of samples, more than 90% of the specimens belonged to the species described based on the first set samples. Indeed, the barcodes suggested the presence of fewer than 20 additional new species [ 37 ]. This implies that a few iterations of taxonomic revision of a taxon at any one site will be sufficient for allowing most species to be assigned to a species.…”