An 11-year spatiotemporal roadkill dataset derived from citizen reports in Tsukuba, Japan

Roadkill data are vital for assessing biodiversity and human–wildlife conflicts. However, datasets are often constrained by linear sampling along major highways, a lack of missing-data mechanism quantification, and geographical bias toward Western regions. This Data Descriptor presents a high-resolution dataset of 14,249 roadkill records collected over 11 years (2005–2007 and 2016–2025) in Tsukuba, Japan. Unlike traditional surveys, this dataset was generated through an “unintended citizen science,” in which citizens report carcasses to the municipal administration, triggering collection by contracted professionals. This mechanism ensured comprehensive “planar” coverage of the entire 284 km² administrative area. This standardized dataset features three methodological strengths: (1) a dual-resolution georeferencing system that assigns “point” and “polygon” precisions to detailed and ambiguous records, respectively; (2) a missing-data proxy via the retention of invalid reports; and (3) a three-tier taxonomic verification process involving citizens, contractors, and experts. The dataset offers a baseline for analyzing the impacts of urbanization, monitoring the invasion of non-native species, and evaluating the reliability of other administrative biodiversity monitoring systems.