Data from the BioNet Data repository, which holds records from a number of custodians. The data are only indicative and cannot be considered a comprehensive inventory, and may contain errors and omissions.

Species classified as sensitive have been removed as well as fields that contain names or personally identifying information. Location accuracy varies by record.

Copyright the State of NSW through the Department of Planning, Industry and Environment. The records included in this release are CC-BY4.0.

The data query was constructed using the following criteria to represent animals considered significant to biodiversity or human safety:

• BioNet Records in NSW only
• Only records with the statuses of ‘Valid and accepted without modification’ or 5 ‘Accepted as valid from quarantine’
• Observations of roadkill defined as ‘Road Injured', 'Road kill' or where occurrenceRemarks contains the phrase 'Encounter broad: Collision; Encounter narrow: Collision -Motor Vehicle'

Species records were included using the following criteria:

• Marsupials filtered by family (Vombatidae or Macropodidae only).
• Ungulates filtered by order (Artiodactyla).
• Koala (Phascolarctos cinereus) explicitly listed since it's the only member of family Phascolarctidae.
• Emu (Dromaius novaehollandiae) explicitly listed.
• Dogs and Dingoes included by filtering for family (Canidae).
• Additional threatened species with conservation concern (not 'Not Listed' at federal or NSW level) are included.

The BioNet database contains records with different levels of locational accuracy. This has necessitated additional filtering to support the use of the data for decision-making. The filtering process involved removing:

a) Records that were accurate to the suburb only level

b) Records where the co-ordinates given were inconsistent with the location information in the record field

c) Records that were otherwise identical but had different dates

d) All records without a CC-BY 4.0 licence. Any threatened species under NSW or federal law (species defined as ‘sensitive’ are excluded for this public release as they cannot be distributed without written consent of DCCEW).

This filtering for b) and c) was achieved by removing records whose co-ordinates were within 1m of each other. The filtering process removed 73% (67% from b) and c), 6% from d) of all Bionet wildlife vehicle strike records identified by the data query leaving 18,367 state-wide.