Cumulative spatial assessment of observed wildlife-vehicle collisions based on indicator species thresholds

Abstract

Global road network expansion has exacerbated wildlife-vehicle collisions. However, traditional risk assessments predominantly assume a linear relationship between landscape features and observed collision quantities, overlooking the non-linear threshold characteristics of ecological responses. This study aimed to quantify the non-linear response thresholds of terrestrial vertebrates to diverse landscape conditions and to characterize national spatial patterns of threshold-based cumulative scores. We further examined how positive- and negative-response taxa were associated with different landscape-cover intervals. Using extensive citizen science survey data across mainland China, we employed Threshold Indicator Taxa Analysis at an optimal spatial scale of 500 m. We then integrated community-level abrupt change points into a threshold-based cumulative scoring system. Positive-response taxa experienced abrupt increases in observed roadkill quantities within specific landscape-cover intervals. Negative-response taxa exhibited declines in reported records under relatively low levels of impervious surface coverage or specific landscape-cover conditions. High-scoring road segments were concentrated in the agricultural and urban landscapes of southeastern China, reflecting the geographic co-occurrence of multiple landscape thresholds associated with increases in reported roadkill quantities. Low-scoring road segments contained fewer reported collisions, but this did not necessarily indicate ecological safety. These contrasting patterns could arise from multiple ecological and observation-related processes, including differences in wildlife occurrence, movement behavior, road-crossing probability, reporting effort, and detectability. The identified landscape thresholds provide a hypothesis-generating framework for exploring spatial variation in reported WVC patterns. To transition from broad-scale screening to targeted mitigation prioritization, future research should validate these patterns by incorporating localized reporting effort, road exposure, traffic intensity, and biogeographic turnover.