Table 2. The automated filters used in this study.

Test	Type	Basis	Rationale
Biodiversity institutions	Error	Gazetteer-based	Records may have coordinates at the location of biodiversity institutions, e.g., because they were erroneously entered with the physical location of the specimen or because they represent individuals from captivity or horticulture, which were not clearly labeled as such
Equal lat/lon	Error	Gazetteer-based	Coordinates with equal latitude and longitude are usually indicative of data entry errors
Sea	Error	Gazetteer-based	Coordinates from terrestrial organisms in the sea are usually indicative of data entry errors, e.g., swapped latitude and longitude
Zeros	Error	Gazetteer-based	Coordinates with plain zeros are often indicative of data entry errors
Capitals	Unfit	Gazetteer-based	Records may be assigned to the coordinates of country capitals based on a vague locality description
Duplicates	Unfit	Gazetteer-based	Duplicated records may add unnecessary computational burden, in particular for large scale biodiversity analyses and distribution modelling for many species
Political centroids	Unfit	Gazetteer-based	Records may be assigned to the coordinates of the centroids of political entities based on a vague locality description
Urban areas	Unfit	Gazetteer-based	Records from urban areas are not necessarily errors, but often represent imprecise records automatically geo-referenced from vague locality descriptions or old records from different land-use types
Basis of record	Unfit	Meta-data	Records might be unsuitable or unreliable for certain analyses dependent on their source, e.g., ‘fossil’ or ‘unknown’
Collection year	Unfit	Meta-data	Coordinates from old records are more likely to be imprecise or erroneous coordinates since they are derived from geo-referencing based on the locality description. This is more problematic for older records, since names or borders of places may change
Coordinate precision	Unfit	Meta-data	Records may be unsuitable for a study if their precision is lower than the study analysis scale
Identification level	Unfit	Meta-data	Records may be unsuitable if they are not identified to species level.
Individual count	Unfit	Meta-data	Records may be unsuitable if the number of recorded individuals is 0 or if the count is too high. This may be related to data-entry or data-basing problems (e.g., defaulting to 0 for numerical values), indicate records from DNA barcoding and in some cases indicate records of absence.