When publishers supply GBIF (Global Biodiversity Information Facility) with a dwc:scientificName, this name is sometimes not found in the GBIF taxonomic backbone. The backbone is needed to organize occurrences on GBIF. In these cases, the occurrence records get a data quality flag called taxon match higher rank. This means that GBIF was only able to match the name to a higher rank. Matching is a process whereby a name supplied by the publisher is compared to a name in the already existing in the GBIF backbone taxonomy. <br> At GBIF, we would always like to match the name supplied by the publisher to the lowest rank possible, so that when a user comes to GBIF looking for a certain name, they will have access to the largest amount of occurrence data possible. <br> The main goals of this project were:<br> <br> <br> <br> Identify the types of issues that prevent matching occurrences to the backbone that come in with an identification at species level (or below) to backbone names at that same rank.<br> <br> <br> Identify the responsible actors (GBIF processing, occurrence record curators, missing checklist) who are best placed to help improve the name.<br> <br> <br> <br> Identify the types of issues that prevent matching occurrences to the backbone that come in with an identification at species level (or below) to backbone names at that same rank.<br> Identify the responsible actors (GBIF processing, occurrence record curators, missing checklist) who are best placed to help improve the name.<br> In Fig. 1, I divide unique names from occurrences supplied to GBIF from publishers that have received the taxon match higher rank flag. Here we see that GBIF is probably missing many names from Coleoptera (Beetles) and Lepidoptera (Butterflies/Moths). <br> Publishers to GBIF sometimes do not provide enough information in the dwc:scientificName for GBIF to choose between names in the backbone Fig. 2. If a publisher only supplied GBIF with 'Glocianus punctiger' we would not be able to determine between the two choices, and it would get moved to the higher rank (genus Glocianus). <br> Publishers also supply GBIF with a variety of what I call unmatchable names, which are names that are impossible to match to the GBIF backbone. Sometimes these names are acceptable names, but still missing from the backbone, like missing hybrids or OTUs (Operational Taxonomic Units). Other names are simply bad names that we can’t expect to fix. Some examples below:<br> Table 1<br> It is often hard to tell if a missing name is a real data gap. To check, I randomly sampled five possibly missing names from each group from Fig. 1 to check if I could manually locate a source outside GBIF with the name.<br> Around 50% (44 of 86) of the possibly missing names appear to be genuinely missing from the GBIF backbone. We can therefore conservatively assume that there are thousands of missing names in the GBIF backbone. Keep in mind, however, that many missing names are missing synonyms—that is, they are not unique taxon concepts. Taking half of 50% (25%), we can make a conservative minimum missing names Table 2.<br> As a data publisher, there are a few things that can be done to improve name matching to the GBIF backbone.<br> <br> <br> <br> Run your dataset through the data validator<br> <br> <br> Match your names to the GBIF backbone before publishing using species lookup or rgbif<br> <br> <br> Add authorship if appropriate<br> <br> <br> Fill known higher-taxonomy<br> <br> <br> Try to avoid working name placeholders for the dwc:scientificName<br> <br> <br> Do not put identification qualifiers in the dwc:scientificName field but rather use the dwc:identificationQualifier field.<br> <br> <br> <br> Run your dataset through the data validator<br> Match your names to the GBIF backbone before publishing using species lookup or rgbif<br> Add authorship if appropriate<br> Fill known higher-taxonomy<br> Try to avoid working name placeholders for the dwc:scientificName<br> Do not put identification qualifiers in the dwc:scientificName field but rather use the dwc:identificationQualifier field.
