Back to the interactive map

Reckless Roosters data and methodology

Reckless Roosters is an Etashi geospatial demo that combines biodiversity observations with road proximity. The “jaywalking” theme is a joke; the spatial analysis is real.

GBIF source scale

On 2026-11-11, the GBIF index returned 276,092,387 animal observations for 2025. Filters: Animalia, human observation, coordinates present, no flagged geospatial issue. These are occurrence records, not a count of unique animals, and the records in this source count were not all downloaded or checked against roads.

Ten most represented country codes in that query:

  1. US: 120,461,960 observations
  2. CA: 19,127,809 observations
  3. GB: 11,782,858 observations
  4. AU: 11,001,598 observations
  5. NO: 10,536,298 observations
  6. NL: 10,475,766 observations
  7. IN: 9,398,975 observations
  8. SE: 7,102,680 observations
  9. ES: 6,559,904 observations
  10. DE: 5,299,540 observations

Explore the GBIF filters · View the GBIF API query (live totals may differ from the captured count).

Worldwide bubble map

The worldwide view uses country and territory counts from the GBIF index query for 2025, captured 2026-11-11. The bubbles represent 276,090,733 records in 249 mapped countries and territories. Each bubble is drawn at a Natural Earth label point. The worldwide view measures observation volume. Distance-to-road analysis covers the study window in southwestern South Africa.

Offline road proximity pilot

We staged 2,420,260 observations of passerines, shorebirds and their relatives, waterfowl, and hawks and their relatives from southwestern South Africa, dated 2010–2025. The archived inputs came from GBIF and the Western Cape OpenStreetMap extract dated 2026-11-11. Road data © OpenStreetMap contributors.

We screened for duplicate or missing GBIF IDs, selected CC0 and CC BY records, and excluded flagged, generalized or withheld records. We indexed 114,445 mapped road lines in a regional metre projection, then checked 2,229,803 reported locations for a road within 100 m. The selected road classes run from residential streets to motorways. Ways marked private or prohibited were excluded.

The map publishes counts in 0.25° grid cells. Cells with fewer than 25 eligible records in a view are hidden from that view. Overall totals include those cells. Switch to “Uncertainty ≤20 m” to explore the smaller subset with a reported uncertainty between 0 and 20 m, exclusive of 0. This subset may have a different mix of sources, species and locations.

Distances refer to reported points and mapped road centre lines. They measure proximity in the two snapshots and may reflect road access by observers, coordinate uncertainty or road changes since the observation date.

Original road proximity demo

The original map and rankings used 1,000 records in a bundled export dated 2026-06-21. The historical workflow used OpenStreetMap Overpass to find nearby highway geometry. The map draws a subset for speed. The rankings group named taxa and omit missing names. The legacy export contains non-animal taxa and incomplete road-distance results, so its counts describe only this demo.

Distance from a reported coordinate does not establish that an animal crossed a road. Coordinate uncertainty, where people choose to observe, and uneven reporting all matter.

Built by Etashi.