Occurrence data collection for wild species

We collected occurrence data from GBIF using higher taxon searches for Artiodactyla and Perissodactyla. The results of these searches are shown in the following table:

Taxon	Date	Occurrences	Data sets	DOI
Artiodactyla	2018-10-19	1,221,109	1,019	https://doi.org/10.15468/dl.qqwyhp
Perissodactyla	2018-10-20	226,805	289	https://doi.org/10.15468/dl.jxwvia

We then pre-processed the DarwinCore records resulting from these searches by the following steps:

1. extracted selected columns, normalized boolean field for geospatial issues, looked up the long taxon name for the record, and wrote the result to tab separated data
2. assessed whether records had been annotated with non-accepted taxonomic names, and wrote these to a synonyms table
3. loaded canonical and synonymous names into a sqlite schema to form an extended taxonomic backbone, whose topology is based on Mammal Species of the World, 3rd edition
4. loaded the occurrences themselves, anchored on the taxonomic backbone. At this step we removed records with incomplete lat/long/event_date fields, records with geospatial issues, and records whose basis was UNKNOWN

After these steps, the number of remaining records was 834,182. We then exported these per species to CSV files, which we stored here. For any given species, the following steps were taken to produce its output file:

1. collect all records for the species, including all subspecies
2. keep an initial random sample of 1000 records
3. do not filter on basisOfRecord (in plants one would normally only keep PHYSICAL_SPECIMEN, here we keep all types)
4. keep records whose eventDate is onwards from 1900-01-01
5. keep records where latitude and longitude have a precision of at least two decimal places
6. keep records whose coordinates are distinct
7. keep records whose coordinates fall within polygons for the species ranges (in a shape file) if available
8. keep records whose mean pairwise distance to all others does not differ from the species mean by more than 1 standard deviation
9. only keep species with more than 10 records (as per Raes & Aguirre-Gutierrez, 2018)

For more information on the data collection for domesticated species, and for selected (data-poor) wild species, refer to the ../domesticated folder.