DNA barcoding of aquatic insects reveals unforeseen diversity and recurrent population divergence patterns through broad-scale sampling in northern Canada

Automated identification of aquatic insects: A case study using deep learning and computer vision techniques

Deep learning techniques have recently found application in biodiversity research. Mayflies (Ephemeroptera), stoneflies (Plecoptera) and caddisflies (Trichoptera), often abbreviated as EPT, are frequently used for freshwater biomonitoring due to their large numbers and sensitivity to environmental changes. However, the morphological identification of EPT species is a challenging but fundamental task. Morphological identification of these freshwater insects is therefore not only extremely time-consuming and costly, but also often leads to misjudgments or generates datasets with low taxonomic resolution. Here, we investigated the application of deep learning to increase the efficiency and taxonomic resolution of biomonitoring programs. Our database contains 90 EPT taxa (genus or species level), with the number of images per category ranging from 21 to 300 (16,650 in total). Upon completion of training, a CNN (Convolutional Neural Network) model was created, capable of automatically classifying these taxa into their appropriate taxonomic categories with an accuracy of 98.7 %. Our model achieved a perfect classification rate of 100 % for 68 of the taxa in our dataset. We achieved noteworthy classification accuracy with morphologically closely related taxa within the training data (e.g., species of the genus Baetis, Hydropsyche, Perla). Gradient-weighted Class Activation Mapping (Grad-CAM) visualized the morphological features responsible for the classification of the treated species in the CNN models. Within Ephemeroptera, the head was the most important feature, while the thorax and abdomen were equally important for the classification of Plecoptera taxa. For the order Trichoptera, the head and thorax were almost equally important. Our database is recognized as the most extensive aquatic insect database, notably distinguished by its wealth of included categories (taxa). Our approach can help solve long-standing challenges in biodiversity research and address pressing issues in monitoring programs by saving time in sample identification.

DNA barcoding unveils a high diversity of caddisflies (Trichoptera) in the Mount Halimun Salak National Park (West Java; Indonesia)

Background

Trichoptera are one of the most diverse groups of freshwater insects worldwide and one of the main bioindicators for freshwater quality. However, in many areas, caddisflies remain understudied due to lack of taxonomic expertise. Meanwhile, globally increasing anthropogenic stress on freshwater streams also threatens Trichoptera diversity.

Methods

To assess the Trichoptera diversity of the area within and around the Mount Halimun Salak National Park (MHSNP or Taman Nasional Gunung Halimun Salak) in West Java (Indonesia), we conducted a molecular-morphological study on Trichoptera diversity using larvae from a benthic survey and adults from hand-netting. In addition to morphological identification, we applied four different molecular taxon delimitation approaches (Generalized Mixed Yule Coalescent, Bayesian Poisson Tree Processes, Automatic Barcode Gap Discovery and Assemble Species by Automatic Partitioning) based on DNA barcoding of Cytochrome-C-Oxidase I (COI).

Results

The molecular delimitation detected 72 to 81 Operational Taxonomic Units (OTU). Only five OTUs could be identified to species level by comparing sequences against the BOLD database using BLAST, and four more to the genus level. Adults and larvae could be successfully associated in 18 cases across six families. The high diversity of Trichoptera in this area highlights their potential as bioindicators for water quality assessment.

Conclusions

This study provides an example of how molecular approaches can benefit the exploration of hidden diversity in unexplored areas and can be a valuable tool to link life stages. However, our study also highlights the need to improve DNA barcode reference libraries of Trichoptera for the Oriental region.

Application of propylene glycol in DNA-based studies of invertebrates

High-throughput sequencing (HTS) studies on invertebrates commonly use ethanol as the main sample fixative (upon collection) and preservative (for storage and curation). However, alternative agents exists, which should not be automatically neglected when studies are newly designed. This review provides an overview of the application of propylene glycol (PG) in DNA-based studies of invertebrates, thus to stimulate an evidence-based discussion.

The use of PG in DNA-based studies of invertebrates is still limited (n = 79), but a steady increase has been visible since 2011. Most studies used PG as a fixative for passive trapping (73%) and performed Sanger sequencing (66%; e.g. DNA barcoding). More recently, HTS setups joined the field (11%). Terrestrial Coleoptera (30%) and Diptera (20%) were the most studied groups. Very often, information on the grade of PG used (75%) or storage conditions (duration, temperature) were lacking. This rendered direct comparisons of study results difficult, and highlight the need for further systematic studies on these subjects.

When compared to absolute ethanol, PG can be more widely and cheaply acquired (e.g. as an antifreeze, 13% of studies). It also enables longer trapping intervals, being especially relevant at remote or hard-to-reach places. Shipping of PG-conserved samples is regarded as risk-free and is authorised, pinpointing its potential for larger trapping programs or citizen science projects. Its property to retain flexibility of morphological characters as well as to lead to a reduced shrinkage effect was especially appraised by integrative study designs. Finally, the so far limited application of PG in the context of HTS showed promising results for short read amplicon sequencing and reduced representation methods. Knowledge of the influence of PG fixation and storage for long(er) read HTS setups is currently unavailable.

Given our review results and taking difficulties of direct methodological comparisons into account, future DNA-based studies of invertebrates should on a case-by-case basis critically scrutinise if the application of PG in their anticipated study design can be of benefit.

A DNA Barcoding Survey of an Arctic Arthropod Community: Implications for Future Monitoring

Accurate and cost-effective methods for tracking changes in arthropod communities are needed to develop integrative environmental monitoring programs in the Arctic. To date, even baseline data on their species composition at established ecological monitoring sites are severely lacking. We present the results of a pilot assessment of non-marine arthropod diversity in a middle arctic tundra area near Ikaluktutiak (Cambridge Bay), Victoria Island, Nunavut, undertaken in 2018 using DNA barcodes. A total of 1264 Barcode Index Number (BIN) clusters, used as a proxy for species, were recorded. The efficacy of widely used sampling methods was assessed. Yellow pan traps captured 62% of the entire BIN diversity at the study sites. When complemented with soil and leaf litter sifting, the coverage rose up to 74.6%. Combining community-based data collection with high-throughput DNA barcoding has the potential to overcome many of the logistic, financial, and taxonomic obstacles for large-scale monitoring of the Arctic arthropod fauna.

Trichoptera of Canada

Trichoptera, or caddisflies, are common members of freshwater ecosystems as larvae and are important indicators of aquatic system health. As such, the species are relatively well studied, with keys available for larvae and adults of many of the taxa occurring in Canada. The number of species recorded from Canada since 1979 (Wiggins 1979) has increased from 546 to 636, an increase of 16.4%. Of those species newly recorded, 17 represent newly described taxa since 1979. Taking into consideration the species likely to be subsequently found in Canada based on records in adjacent parts of the United States and results from DNA barcoding, an estimated 129-181 species remain to be documented in Canada.

Ephemeroptera of Canada

Thus far, 335 currently valid species in 82 genera and 21 families of mayflies (Ephemeroptera) have been documented from Canada, remarkably representing a little more than half of the combined species richness of Canada, Mexico and the USA. The current known species richness for Canada represents an increase of 11.3% as compared to that reported in 1979. Species richness is greatest in the families Heptageniidae (83), Baetidae (76) and Ephemerellidae (45). A total of 328 DNA Barcode Index Numbers (BINs) are available for Canadian mayfly species. The greatest net gains anticipated for future species tallies are for Baetidae (25), Heptageniidae (10) and Leptophlebiidae (10). A total of 66 more species overall is anticipated for Canada, with greatest gains potentially coming from lentic habitats across Canada and from far eastern and far western areas in general. However, even metropolitan areas should not be overlooked for the potential of discovery.

DNA barcode-based survey of Trichoptera in the Crooked River reveals three new species records for British Columbia

Anthropogenic pressures on aquatic systems have placed a renewed focus on biodiversity of aquatic macroinvertebrates. By combining classical taxonomy and DNA barcoding we identified 39 species of caddisflies from the Crooked River, a unique and sensitive system in the southernmost arctic watershed in British Columbia. Our records include three species never before recorded in British Columbia: Lepidostoma togatum (Lepidostomatidae), Ceraclea annulicornis (Leptoceridae), and possibly Cheumatopsyche harwoodi (Hydropsychidae). Three other specimens may represent new occurrence records and a number of other records seem to be substantial observed geographic range expansions within British Columbia.