Biting midges of the genus Culicoides are tiny, often just 1–3 mm long, but their impact on animal health and agricultural economies can be vast. They vector more than 50 viruses of veterinary concern, including bluetongue and African horse sickness. New Zealand remains free of Culicoides and the diseases they transmit, and it runs a national surveillance programme to keep it that way. This article shares a study that shows advancements in the utilisation of DNA-based tools, specifically DNA metabarcoding of bulk insect samples and environmental DNA (eDNA) recovered from trap fluids to make surveillance faster, more scalable and no less reliable.
Early detection is everything in biosecurity. If Culicoides arrived and established silently, the first sign might be an animal disease outbreak that is both costly and difficult to contain. Yet scanning 15,000–20,000 insects a year by eye is neither efficient nor easily expandable. Molecular approaches promise to read the genetic “fingerprints” of all organisms in a mixed sample in one go, potentially flagging rare or unexpected species that taxonomists might miss.
A brief primer on metabarcoding and eDNA
Metabarcoding is a high-throughput method that uses universal or semi‑universal genetic primers to amplify barcode regions (here, COI) from all organisms present in a mixed sample, which are then sequenced en masse. Bioinformatic clustering of the reads into operational taxonomic units (OTUs) allows rapid estimation of which taxa are present. Environmental DNA (eDNA) refers to the genetic material shed by organisms into their surroundings—water, soil, air, or, in the case of this study, the ethanol inside a light trap—so that it can be captured without handling the organism itself. Both approaches can dramatically reduce the time to screen complex samples, but both come with biases linked to primer choice, DNA degradation, biomass differences and amplification stochasticity.
What the team actually did
The researchers analysed 38 trap samples collected weekly over nine to ten weeks in 2020 from four cattle farms in different North Island districts: Morrinsville, Okaihau, Warkworth and Whakatāne. Each site deployed green LED light traps baited with carbon dioxide and octenol to attract midges and other flying insects. The traps were filled with ethanol to preserve the catch. Every sample was first processed morphologically: insects were counted, and any members of the midge family Ceratopogonidae were recorded. As Culicoides does not occur in New Zealand, native Ceratopogonidae served as a proxy “target group” to test how well the DNA methods could recover what morphology saw and spot what morphology might have missed.
Two parallel molecular routes were then compared. In one, the entire insect catch was homogenised (destroying the specimens) and DNA extracted from the tissue mixture. In the other, DNA was filtered from the ethanol preservative—the eDNA that insects shed into the fluid—leaving the intact specimens available for later taxonomic work if needed. Both DNA types were amplified using two commonly used mitochondrial cytochrome oxidase I (COI) primer pairs (LCO1490/HCO2198 and mlCOIintF/jgHCO2198) and sequenced on an Illumina MiSeq platform. An in‑house Ceratopogonidae COI reference library, supplemented by public databases, supported species assignment.
What the microscope found
Across the 38 trap samples, a total of 45,745 insects were counted. No exotic Culicoides were detected, but native Ceratopogonidae were present in 22 traps (58%), albeit at low abundance—just 114 individuals overall, representing a mere 0.25% of total catch. Most of those were from Whakatāne, where Ceratopogonidae appeared in every sample. These figures served as the benchmark for evaluating molecular performance.
Bulk metabarcoding outperformed eDNA for target detection
When it came to reproducing the morphological verdict on whether Ceratopogonidae were present or absent, metabarcoding of homogenised bulk samples was consistently more accurate than metabarcoding of eDNA from the trap fluid. Using the classic LCO1490/HCO2198 primers, bulk metabarcoding achieved an overall detection accuracy of 81.94%, compared with 68.42% for the ethanol-derived eDNA. With the mlCOIintF/jgHCO2198 primers, bulk metabarcoding again reached roughly the same accuracy (81.58%), while eDNA accuracy dropped to 55.26%. In short, across both primer sets, the bulk approach surpassed the 80% mark, whereas eDNA lagged behind.
False negatives—cases where Ceratopogonidae were seen under the microscope but not detected by sequencing—were the main reason eDNA underperformed. The team also observed a few ‘false positives’, where metabarcoding detected Ceratopogonidae that morphology did not; these could reflect genuinely missed specimens, tiny fragments invisible to the taxonomist, or technical artefacts such as low‑level index “cross‑talk” among samples during sequencing. Importantly, metabarcoding (both eDNA and bulk) did sometimes rescue detections that morphology missed, reminding us that the microscope is not infallible either.
Why did eDNA struggle?
The biology and physics are not in eDNA’s favour here. Ceratopogonidae are extremely small. Small-bodied insects contribute little biomass and therefore little DNA to the preservative ethanol, especially compared with the large moths and flies that dominate trap catches. eDNA is also not evenly distributed through the fluid and can degrade quickly. Even when present, amplification biases during PCR can down‑weight scarce templates further, particularly with broad “universal” primers. Together, these factors make it hard for eDNA metabarcoding to reliably pick up low‑abundance, tiny-bodied targets amid a noisy community background—at least with generalist primers and the workflows used here.
Primer choice shaped the community picture more than sample type
Although eDNA was weaker for yes/no detection of Ceratopogonidae, both molecular approaches painted broadly similar pictures of the overall insect communities in each trap. Non-metric multidimensional scaling showed that ethanol and bulk samples from the same trap usually clustered together, meaning they recovered comparable community structure. What did make a bigger difference was the primer pair. LCO1490/HCO2198 skewed strongly towards Lepidoptera, whereas mlCOIintF/jgHCO2198 yielded a more even spread across orders such as Lepidoptera, Diptera and Trichoptera. This reinforces a central truth of metabarcoding: your primers define your window on biodiversity.
So, what value does eDNA bring?
Despite its lower sensitivity for the specific, tiny-bodied target group tested here, eDNA retains several important advantages. It is non‑destructive, allowing taxonomists to re‑examine the original specimens. It is operationally simpler—filter the fluid, extract DNA, and you can process far more samples more quickly than microscopy allows. It produces community‑level data that align well with bulk metabarcoding, so it can still track spatial and temporal shifts in insect assemblages at scale. Crucially, eDNA’s performance could be markedly improved if the aim is targeted detection rather than broad community profiling—for example, by switching from metabarcoding to quantitative PCR (qPCR) or digital PCR assays that lock onto Culicoides‑specific markers with much higher sensitivity. In other words, use eDNA metabarcoding to understand the community and screen widely, then deploy species‑specific assays to decisively confirm or rule out incursions.
Practical implications for biosecurity programmes
For routine, rapid, early‑warning surveillance of unwanted midges, bulk-sample metabarcoding currently delivers the best balance of accuracy and throughput when using universal COI markers. Where preserving specimens is essential, an eDNA-first workflow could facilitate triage of samples: rapidly screen trap fluids to flag suspect traps, then prioritise those specimens for either bulk metabarcoding, species-specific qPCR, or morphological confirmation. Programmes should also invest in well-curated, regionally comprehensive reference libraries (the in-house Ceratopogonidae database used in this study was invaluable) and in validating primer sets against the taxa of greatest concern.
Conclusion
This study offers a measured, practice‑ready message. If your immediate need is to know, with high confidence, whether a delicate, low‑abundance target like Ceratopogonidae is present in a trap, homogenised bulk metabarcoding with carefully chosen primers currently does the job better than ethanol‑derived eDNA. But if your goal is to scale surveillance, keep specimens intact, and track whole insect communities efficiently, eDNA metabarcoding of trap fluids is already useful, and with species‑specific assays layered on top, it could become a powerful, non‑destructive frontline tool for early detection.
Leave a Reply