Filling knowledge gaps with roadkill data

Study evaluates quality of volunteer-collected animal vehicle collision data and its potential contribution to ecological research

GBIF-mediated data resources used : 26,853,168 species occurrences
Lontra canadensis
North American river otter (Lontra canadensis) found dead by state highway 113 in Rio Vista, CA, USA. Photo by dwright via iNaturalist (CC BY-NC 4.0)

Knowledge about the distributions of species is increasingly being improved thanks to data collected by volunteers through citizen science projects. Dominated by non-experts, some projects rely on experts for validation while others verify identifications through community efforts.

This study seeks to evaluate the quality of a volunteer-based animal-vehicle collision programme—the California Roadkill Observation System (CROS)—and its contribution to knowledge about species ranges. Assessing more than 35,000 observations of more than 400 species, the authors found that species identification accuracy was high for both professional (99 per cent) and non-professional users (95 per cent).

The ranges derived from the CROS dataset overlapped to a large extent with GBIF-mediated occurrences. However, for 139 species the roadkill data expanded ranges by a mean of 3,735 km². For some ten per cent of species, this would increase their environmental niche—0.13 degrees C for mammals, suggesting a clear significance of the dataset, that if published to GBIF could fill gaps in knowledge.

Original article

Tiedeman K, Hijmans RJ, Mandel A, Waetjen DP and Shilling F (2019) The quality and contribution of volunteer collected animal vehicle collision data in ecological research. Ecological Indicators. Elsevier BV 106: 105431. Available at: