Bird genetic databases need improved curation and error reporting to NCBI

Abstract

The ongoing biodiversity crisis is causing rapid species losses faster than taxonomists' capacity to describe new species. Integrative taxonomic approaches need robust taxonomic baseline data to correctly describe and conserve global species diversity, in which genetic data are one of the pillars. However, despite their broad use throughout the biological sciences, the quality of genetic data within public repositories is still not guaranteed. Here, we curated GenBank Cytochrome‐b records of Aves, a well‐known taxon undergoing continuous taxonomic changes, and provide a curated database to aid taxonomic and conservation efforts. After curation of 54 114 records, the database represents 50 280 sequences from 6867 species (63% of current bird biodiversity), with a per‐genus median of two species (50% interquartile ranges of 1–4) and three sequences (1–9). Overall, 4469 bird species have ≥ 2 sequences representing 91.7% and 97.5% of currently known families and orders, respectively. We flagged 1336 erroneous records (2.5% of Aves Cytochrome‐b records) in need of taxonomic curation (71%) or removal, 45% of which lack any voucher information preventing a proper taxonomic assignation. Compared with Amphibia, a group for which a similar analysis was recently published, Aves records have a five‐fold higher prevalence of errors caused by contamination, sequencing errors or diverging mutation patterns. We provide a species‐level taxonomic update for 839 GenBank records, including changes at the genus (76), family (21) and order (16) levels. Scientific conclusions from the manuscripts that have ever used those data might be compromised. A further concern is the current availability of GenBank records previously identified as erroneous in published manuscripts, demonstrating the need for improved communication between NCBI and the community. We call for caution when utilizing GenBank records without curation of retrieved data, despite new improvements.