An audit of some processing effects in aggregated occurrence records

Area of Science:

• Biodiversity Informatics
• Taxonomy
• Data Curation

Background:

• Occurrence records are crucial for biodiversity research.
• Data aggregators process records from multiple institutions for wider accessibility.
• Ensuring data integrity after aggregation is vital for scientific use.

Purpose of the Study:

• To audit changes in Darwin Core fields for Australian Museum (AM), Museums Victoria (MV), and New Zealand Arthropod Collection (NZAC) records after processing by the Atlas of Living Australia (ALA) and Global Biodiversity Information Facility (GBIF).
• To assess the impact of data aggregation on taxonomic names and specimen type status.
• To identify data loss and inconsistencies introduced by aggregators.

Main Methods:

• Auditing approximately 800,000 occurrence records from AM, MV, and NZAC.
• Comparing Darwin Core fields before and after processing by ALA and GBIF.
• Analyzing changes in formal taxon names and type status.
• Quantifying data loss in specific fields.

Main Results:

• Formal taxon names were altered in 13-21% of AM and MV records, with low agreement between ALA and GBIF.
• Specimen type status remained unchanged despite name modifications, causing taxonomic confusion.
• Significant data loss (up to 100%) occurred in certain fields, likely due to programming errors.
• Discrepancies were observed in name changes between aggregators, with one changing names in two to three times more records than both combined.

Conclusions:

• Data aggregation processes can introduce substantial changes to taxonomic names and cause data loss, impacting the scientific utility of biodiversity records.
• The lack of agreement between aggregators on name processing and the persistence of type status with changed names create ambiguity.
• It is recommended that aggregators provide both original and processed data, and end-users meticulously check for alterations and losses.