Predicting wildlife road-crossing probability from roadkill data using occupancy-detection models.

Wildlife-vehicle collisions (WVC) represent a major threat for wildlife and understanding how WVC spatial patterns relate to surrounding land cover can provide valuable information for deciding where to implement mitigation measures. However, these relations may be heavily biased as many casualties are undetected in roadkill surveys, e.g. due to scavenger activity, which may ultimately jeopardize conservation actions. We suggest using occupancy models to overcome imperfect detection issues, assuming that ‘occupancy’ represents the preference for crossing the road in a given site, i.e. is a proxy for the roadkill risk; and that the ‘detectability’ is the joint probability of an animal being hit in the crossing site and its carcass being detected afterwards. Our main objective was to assess the roadkill risk along roads while accounting for imperfect detection issues and relate it to land cover information. We conducted roadkill surveys over 114 km in nine different roads, biweekly, for five years (total of 484 surveys), and developed a Bayesian hierarchical occupancy model to assess the roadkill risk for the six most road-killed taxa for each road section and season (WET and DRY). Overall, we estimated a higher roadkill risk in road sections surrounded by agriculture, open habitats; and a higher detectability within the 4-lane road sections. Our modeling framework has a great potential to overcome the limitations related to imperfect detection when assessing the roadkill risk and may become an important tool to predict which road sections have a higher mortality risk.

Data and Resources

Cite as

Santos R.A.L. Mota-Ferreira M. Aguiar L.M.S. y Ascensão F. Predicting wildlife road-crossing probability from roadkill data using occupancy-detection models. Elsevier, 2018. https://doi.org/10.1016/j.scitotenv.2018.06.107

Clipboard Icon
Retrieved: 21 Jan 2025 14:23:53

Metadata

Basic information
Resource type Text
Date of creation 2024-12-02
Date of last revision 2025-01-21
Show changelog
Metadata identifier cf9d12f8-cd5b-55a0-b45d-3f70a01f0508
Metadata language Spanish
Themes (NTI-RISP)
High-value dataset category
ISO 19115 topic category
Keyword URIs
Bibliographic information
Name of the dataset creator Santos, R.A.L., Mota-Ferreira, M., Aguiar, L.M.S. y Ascensão, F.
Name of the dataset editor Elsevier
Other identifier DOI: 10.1016/j.scitotenv.2018.06.107
Identifier of the dataset creator
Email of the dataset creator
Website of the dataset creator
Provenance
Lineage statement
Metadata Standard
Version notes
Version