When Current Taxids Make Historical Taxa Look New

Abstract

A taxon can keep its biological identity while its numeric identifier changes over time. Reconstructing a past taxonomy state by resolving each sequence to its current-day identifier ignores this, and can make an already-existing taxon look newly created. Among 56 viral sequences flagged as newly created between two NCBI taxonomy snapshots one year apart, only 17 (30.4%) were genuinely absent from the earlier snapshot; the remaining 39 (69.6%) already existed there under a predecessor identifier later merged into today's. The same mechanism recovers 218–269 identifiers per snapshot library-wide, for ten of thirteen dated snapshots studied (2014–2022). An illustrative classification experiment shows the practical consequence: retagging the 39 artifacts with their period-valid identifier restores correct classification. Because this correction relies on identifier-merge history alone, it establishes apparent novelty caused by identifier evolution, not biological novelty. We give five recommendations to avoid this artifact.