Historical occurrences of lynx (Lynx spp.) in Iberia

Long-term biodiversity data are essential for understanding historical range dynamics, extinction processes, and the impacts of global changes on species distributions. However, the relevant species records pre-dating systematic surveys are often dispersed across heterogeneous sources and rarely consolidated into standardized datasets suitable for ecological analysis. Here, we present a curated dataset of historical occurrence records of the Eurasian lynx (Lynx lynx) and the Iberian lynx (Lynx pardinus) in the Iberian Peninsula, which integrates information from written historical sources, fossil evidence, and museum specimens. The dataset comprises 637 unique lynx occurrence records, with approximately 60% attributed to L. pardinus and 38% to L. lynx. The remaining records could not be assigned to species-level with sufficient confidence and were retained without species attribution. Our occurrence records span six centuries, with dated records from the 15th century onward. However, the temporal representation is uneven, reflecting historical documentation practices rather than actual species abundance. The 18th, 19th, and 20th centuries are the most represented periods, with the 19th century alone accounting for approximately one-third of all records. We chose to end our sampling in 1980, when the first systematic scientific studies on the Iberian lynx began to be published. Occurrences were compiled from three complementary evidence streams: written historical accounts, fossil remains, and museum specimens. The two fossil specimens were exclusively obtained from public paleontological collections, whereas museum specimens were sourced from both public and private collections. All physical specimens were identified to species using DNA sequencing, providing high-confidence taxonomic assignments. Written records were extracted from geographical dictionaries, legal and administrative documents, hunting records, travel narratives, and early zoological works. Species differentiation within written sources followed a hierarchical and conservative protocol. When available, explicit morphological or behavioral descriptions were used for species assignment. When such information was absent or insufficient, species attribution followed framework which evaluates the diagnostic reliability of vernacular names while accounting for regional and temporal linguistic variation. Records lacking sufficient evidence for confident attribution were left unassigned. To further minimize misclassification, spatial outliers were individually examined, allowing the identification of inconsistencies caused by transcription errors or implausible species assignments. All records were standardized using contemporary geographic references and georeferenced with associated spatial uncertainty estimates. Each occurrence is accompanied by metadata describing source type, temporal precision, and confidence in taxonomic attribution. The dataset is provided in three complementary versions: (1) without species assignment, enabling independent reassessment; (2) with species assignments based on the described protocols; and (3) a gridded dataset (i.e. 10 × 10 km) summarizing occurrences by species and century, intended for broad-scale spatial and temporal analyses. This openly available dataset adheres to FAIR data principles and is designed to support research in historical ecology, macroecology, and conservation research. By combining DNA-validated physical evidence with systematically curated historical records and explicitly documenting uncertainty, this dataset expands ecological baselines for both lynxes, while allowing users to evaluate and address uncertainty according to their analytical needs. The data are available for reuse under CC BY 4.0 licensing. Please cite the data release and this data paper when reusing the dataset.