Georeferencing herbarium specimens of the Naturalis botanical collection using automation and crowdsourcing

Digitalized botanical collection data often lacks location data in the form of geographical coordinates, limiting a wider use of the data. Specimens of Dutch plants, mosses and fungi in the Naturalis botanical collection (herbarium codes AMD, L, U, and WAG) were initially digitized with only 3% of the records having geographical coordinates. In a two year project, coordinates were added to the specimen records by means of (semi)automated techniques and crowdsourcing. Coordinates were inferred from location data present on the specimen labels such as toponyms (including those with typographical errors and abbreviations), grid square codes and other local coordinate systems. Finally, 75% of about half a million specimens of vascular plants could be georeferenced with a precision of 5 km or better.