Compromised taxonomy in the Eastern Himalaya: a critical appraisal of three recently described Impatiens (Balsaminaceae) species, with evidence of fabricated references, unverified AI-generated content, and failures of peer review and editorial oversight

The genus Impatiens L. (Balsaminaceae) has been the subject of discovery of new species from the Eastern Himalaya and Northeast India over the past decade (30 new taxa between 2016 and 2025). But in some cases, rapid publication of narrowly circumscribed taxa, combined with the increasing availability of generative artificial-intelligence (AI) writing tools, has created conditions in which descriptive and bibliographic errors can pass into the permanent taxonomic record. Here we provide a detailed, evidence based appraisal of three Impatiens species described in 2025–2026 by an overlapping author group: I. chakhesangiana M.Borah & S.Singh (Phytotaxa 742), I. pfutserensis S.Singh, B.Singh & M.Bhuyan (Taiwania 71), and I. shamaii S.Singh (Phytotaxa 728). Across the three papers we document a consistent set of problems: biologically impossible or internally contradictory measurements (e.g., a capsule simultaneously described as ‘linear’ and ‘fusiform’); diagnostic characters that contradict the accompanying figures and illustrations; failure to compare the new taxa with their most morphologically similar/allied and geographically proximate relatives; misleading distribution maps that share an identical, mis-scaled frame; inappropriate or self-serving citation; and reference lists containing non-existent papers, impossible volume numbers, and duplicated or internally inconsistent DOIs. The reference level errors in particular are characteristic of AI-assembled bibliographies generated without database verification and possible lack of review. One of the three taxa, I. shamaii, has already been formally synonymised under I. leptocarpa Hook.f. within weeks of publication. We argue that these are not isolated lapses but a recurrent pattern indicating inadequate manuscript preparation and a clear failure of editorial and peer review at the journals concerned. We recommend corrigenda or, where warranted, retraction, and propose concrete safeguards for the description of new taxa from biodiversity rich hotspot regions.