Christa Deeleman-Reinhold (1930-2025): An AI-generated memorial dataset of specimens cited across a career in arachnology

Specimen citation records were extracted from the oeuvre of Christa Deeleman-Reinhold (1930-2025), automated using large language models (LLMs). The legacy of an exemplary career in arachnology featured 55 publications from 1971 to 2025, 49 of which contain a total of 677 taxonomic treatments that cite specimens. Updated to account for current nomenclature, these treatments contributed to the taxonomy of 644 species (including two subspecies and four with less than species level identification) across 29 spider families; 408 species were described by CDR as sole or co-authority. CDR was listed as the sole or co-collector for 125 specimen records. A total of 2719 material citation records accounting for more than 6000 individual specimens were digitized and structured according to Darwin Core standards to facilitate data mobilization. Records were georeferenced and plotted by year of collection on an animated map and shared with the online biodiversity informatics datasphere. Both ChatGPT and Claude.ai LLMs struggled with large monographs, but Claude.ai performed better overall. In some cases, ChatGPT hallucinated non-existent species, something Claude.ai never did.