Predicting the effectiveness of wildlife fencing along roads using an individual-based model: How do fence-following distances influence the fence-end effect?

Wildlife-vehicle collisions on roads pose a major threat to biodiversity and a danger to human motorists. Wildlife fencing prevents animals' access to roads and reduces road mortality significantly. However, mitigation is often constrained by cost, and fences that are too short can be rendered ineffective because of the fence-end effect where collision locations are shifted towards the fence ends. We created an individual-based model to study processes related to the fence-end effect and predict the effectiveness of fences at preventing road crossings based on fence length, home-range size, and movement distances along the fence. The model was created using the JavaScript programming language, runs in a web browser, and includes a visualization that can help identify emerging patterns. The model generates a mathematical function that relates fence effectiveness to fence length. We parameterized the model for wood turtles (Glyptemys insculpta) and ran simulations for the equivalent of 1 year of movement. We compared 8 fence-following distances and 10 fence lengths up to the home-range diameter. The model recreated patterns characteristic of the fence-end effect, including the presence of high-risk collision zones located at the fence ends. Fence effectiveness was calculated by comparing the number of road encounters prevented by the fence to the number of road encounters without a fence present, and a mathematical function was created to predict effectiveness of fences longer than the home-range diameter. Fences shorter than the home-range diameter ranged from 0 to 69 % effective. Longer fences exhibited significantly higher effectiveness but never reached 100 % due to the fence-end effect. Fence effectiveness dropped proportionately to the animals' fence-following distance. The predicted effectiveness can be used in road mitigation planning. Empirical data are needed to quantify fence-following behaviors of a range of species as they can significantly influence a fence's effectiveness.