Answers are grounded in this paper and cite their sources.
Explores species discovery using unsupervised machine learning in parasitoids of oak gall wasps, suggesting improved methodologies.
Species discovery (inferring species limits de novo, without a priori hypotheses) from genetic data has become more common as molecular tools have expanded and has been a helpful initial step in tackling the taxonomic impediment for small insects. Often species discovery involves a single locus (eg mitochondrial cytochrome oxidase I [mtCOI]), but the accessibility of techniques for large sub-genomic sequencing projects (1000 s of loci) makes it possible to approach molecular species discovery with more robust datasets. Here, we test unsupervised machine learning (UML) methods for species discovery on a set of ultra-conserved element loci for a large collection of parasitic wasps reared from North American oak galls, all initially thought to be in genus Eurytoma Illiger. UML methods produced species hypotheses that largely aligned with those that emerge from a commonly used mtCOI-based species partitioning method, and that also tended to match existing species descriptions. Results revealed a new genus-level association with oak galls (Phylloxeroxenus Ashmead) hidden among the Eurytoma, 2 distinct lineages of Eurytoma, including a new lineage of Eurytoma more closely related to the South American genus Kavayva Zhang, Gates, and Silvestre, evidence for one or more cryptic Eurytoma species, and a mix of generalist and specialist host ranges. We make recommendations for how best to employ UML methods to similar datasets.
No takes yet. Share an insight, caveat, or question.
Weinrich et al. (2026) studied this question.
Synapse has enriched 10 related papers. Consider them for comparative context: