When AI encounters natural history: Morphological OTUs reshape our understanding of Earth's life

MorphOTU enables accurate, species-level biodiversity estimation directly from specimen images, matching expert identifications across diverse biological datasets.

• Why it matters: Understanding Earth's biodiversity is limited by the difficulty of reliably identifying species in the field, especially for unnamed or poorly characterized organisms.
• What they did: The authors developed morphOTU, a framework combining self-supervised learning, metric supervision, and hierarchical clustering to organize specimen images in phenotypic space, tested on five datasets with over 4,700 insects.
• The result: MorphOTU accurately recovers species-level diversity estimates, even with sparse data or limited species labels, providing a practical tool for biodiversity assessment without formal taxonomy.