Estimated spatial distributions for Australia's threatened species

This data set contains spatial layers of the estimated extant and historical distributions for species listed as Critically Endangered, Endangered, or Vulnerable as of 28 May 2025 and that occurred within the contiguous Australian exclusive economic zone. This includes 1,785 species with terrestrial distributions, 122 species with aquatic distributions, and 95 species with marine distributions, with 1,962 species in total (noting that some species can occur in multiple realms). For each species, their extant and historical distributions were estimated using a simple logical algorithm based on Species of National Environmental Significance Distribution grids (DCCEEW 2025a), occurrence records (ALA 2025), and spatial environmental layers, described in full in Giljohann et al. (2025) and in the Lineage text associated with this data publication.\nLineage: The species considered are from the 1,962 species listed nationally as Critically Endangered, Endangered or Vulnerable (DCCEEW, 2025) as of 28 May 2025. Species whose distributions and occurrences did not intersect with the Australian Exclusive Economic zone (AMSIS 2023) were excluded. For each species, distribution layers were downloaded from the Species of National Environmental Significance (SNES) database (DCCEEW, 2025), providing polygons of “species or habitat may occur” and “species or habitat likely to occur”.\n\nOccurrence data for each species were downloaded on 28 May 2025 from the Atlas of Living Australia (ALA, 2025). Occurrence records were filtered to exclude records within 100 m of biodiversity institutions or where the record type was unknown. Records classified as spatial outliers (modified z-score > 3.5 std. dev. from the mean) were assessed and cross-checked against published information, and retained or excluded on a case-by-case basis. To ensure the historical estimates included species’ current habitat, an additional occurrence record was created within each SNES “likely to occur” polygon that did not contain an occurrence record. For riverine aquatic species a regular grid of points was sampled from the “likely to occur” polygon and appended to the occurrence data to preserve the shape of rivers. For marine species, a regular grid of points was sampled from the SNES “may occur” polygon and used as occurrence records.\n\nSpecies were assessed as belonging to the realms: aquatic, marine, or terrestrial, and in some cases a combination of realms (e.g. aquatic/marine). The Australian coastline (OSM 2024) was used to differentiate the marine realm from the aquatic and terrestrial realms. Each realm was processed separately, using only data within that realm.\n\nFor terrestrial species, their historical distribution was derived from broad habitat preferences identified by intersecting occurrence records with fine resolution (100 m) layers of the pre-clearing (pre-1750) extent of major vegetation subgroups (DCCEEW, 2020b), bioclimatic subregions (DCCEEW, 2020a) and elevation (Hutchinson et al., 2008). The historical distribution was calculated by selecting all areas of preferred habitat, falling within observed altitudinal limits, from the spatial extent of pre-clearing vegetation. Historical distribution estimates were then constrained using a rule-based approach. For each species with four or more occurrences, a concave hull was created around occurrence records (20 km buffer; 50 km maximum segment length); when the hull area was less than 100,000 km2 the distribution was clipped by the hull, otherwise the entire distribution was used. For species with fewer than four occurrences, the historical distribution was clipped using a 20 km buffer around occurrence records. \n\nFor aquatic species, their historical distribution was derived in a similar way to terrestrial species, with some additional refinement. A map of Australian catchments (BOM 2024) was intersected with the occurrence point data and the SNES polygons, creating a maximum catchment map. The historical distribution was masked by the maximum catchment map.\n\nFor marine species, habitat preference was taken as the intersection of distance from shore and water depth from observed occurrences (ALA 2025) excluding the assumed additional occurrences generated from the regular grid sample. This area was intersected with a convex hull polygon (using a 20 km buffer and 50 km maximum segment length) of all occurrence data (observed and assumed) to create the historical distribution. \n\nFor all realms, the currently occupied portion of the historical distribution (‘extant’) was created by clipping historical distribution by the SNES "likely" and “may" occur distribution polygons.\n\nMethods for Release 1 (2022) are documented in Giljohann et al. (2025), while Release 2 and 3 applied the refinements above.\n\nData products are described in the 'Data_description' file.