Experimental and Simulated Approaches to Roadkill Persistence: Implications for Road Mortality Assessment

The time during which roadkill remains available for detection following a collision is a central issue in road ecology. Not accounting for roadkill persistence leads to underestimating the actual impact of vehicular traffic on populations. Moreover, disregarding the inconsistent rates of disappearance across species, seasons and roads prevents carcass counts from being comparable both within and across studies. This can lead to a misidentification of roadkill hotspots and species for which road mortality is the highest. However, finding the appropriate experimental method to estimate roadkill persistence is difficult: both placing carcasses on roads to conduct removal trials and periodic monitoring of roadkill already present may fail to accurately describe the underlying distribution in persistence times. In this study, in order to assess the potential for systematic bias in roadkill persistence studies depending on the methodology, we simulate roadkill appearance and removal from the road by fitting a survival distribution based on experimental data. We then explore the discrepancies between methods by analysing empirical persistence data derived from both carcass trials and periodic monitoring. We find that, both in theory and in practice, monitoring tends to produce inflated estimates of roadkill persistence compared to carcass removal trials. Accurately describing the distribution of roadkill persistence times may therefore present a greater challenge than previously anticipated.