Tropical Species Backbone: Taxonomy, Bilingual Vernacular Names, and Evidence-Grounded Functional Traits
收藏资源简介：
Tropical Species Backbone — a relational Darwin Core Archive for 409,880 tropical and subtropical species (plants, ornamental aquatics, exotic pets). v1.0.0, published 2026-07-09. One Darwin Core Archive uniting three layers on a shared taxonomic key: a taxonomic backbone (aligned to POWO, IPNI, Catalogue of Life), a bilingual (Chinese + English) vernacular-name lexicon, and an evidence-grounded functional-trait matrix. Contents (Darwin Core Archive) Taxon core — 409,880 species with the full Linnaean hierarchy (kingdom→genus); ~24% carry a resolvable GBIF backbone cross-reference (taxonConceptID). VernacularName extension — 1,204,618 names (1,060,013 Chinese + 144,605 English), each source-tagged. ExtendedMeasurementOrFact extension — 2,682,576 taxon-level functional-trait measurements across 13 trait keys, each tied to an evidence quote and an ontology URI. Provenance & intended use (please read) This dataset stratifies by provenance. The taxonomy and English vernacular names are sourced from / aligned to external authorities. Chinese vernacular names are editorially curated (some pipeline-generated for taxa lacking an established Chinese name). The trait measurements are produced by an evidence-grounded, registry-bound large-language-model extraction pipeline (methods: arXiv:2606.00994): each measurement is high/medium-confidence and tied to a verbatim evidence quote, but is not expert-verified ground truth — best used as broad-coverage priors, weak supervision, or hypotheses to validate. A held-out audit estimated an overall trait value-error rate on the order of ~1–2%. Two compliance-adjacent traits (CITES appendix, IUCN category) are text-scoped indicators derived from descriptions and must not be used as legal, trade, or conservation determinations; consult Species+ (speciesplus.net) and the IUCN Red List for authoritative status. Format & standards Darwin Core Archive (Taxon core + VernacularName + ExtendedMeasurementOrFact + EML metadata). Traits follow the EOL TraitBank model and the Ecological Trait-data Standard (ETS). The three layers are also downloadable as standalone TSV files at tropicals.cn/datasets. The bundled DATASHEET.md gives per-trait coverage, the audited error rate, and full limitations. License Text data licensed under CC-BY 4.0. Cited third-party authorities (POWO, IPNI, Catalogue of Life, GBIF, iNaturalist, EOL, WoRMS, Reptile Database) retain their own terms; attribution travels in the source fields. Related Companion to the Tropicals.cn Tropical Species Encyclopedia (DOI 10.5281/zenodo.20377811). Documented by the preprints arXiv:2606.03156 (data paper) and arXiv:2606.00994 (methods). Contact data@tropicals.cn — institutional use · errata reports
