A novel method to estimate actual infrastructure-induced mortality by integrating sampling biases

1. Human infrastructures are among the most impactful threads to wildlife. While estimates exist on the number of animals killed by these structures over a given period, such estimates typically do no account for several detection biases. Consequently, true mortality rates may be severely underestimated, as well as their impact on populations and species. 2. We present a hierarchical Bayesian latent-state modelling framework that sequentially accounts for three main biases probabilities in estimating mortality abundance: the probability that a hit animal dies on the surveyed area (carcass location probability), the probability that the carcass remains on the surveyed area until the survey is conducted (carcass persistence probability), and the probability that the carcass is observed during the survey process (carcass observation probability). We employ a comprehensive simulation study where we test the effects of variability in species characteristics, sampling design, latent-state parameters, and prior information on the ability of our model to estimate mortality abundance on roads as total number of roadkills. We then demonstrate the applicability of our framework on a case study to estimate the total number of roadkills per km in different Mediterranean ecosystems while evaluating the cross-efficiency of different sampling methods. 3. Our framework is able to accurately recover the total number of roadkills from simulated census data for most simulation scenarios. We detected the highest disagreement between modelling outcomes and simulated data when variability in simulated carcass persistence probability, as well as related prior information in the Bayesian model, were high. In the case study, our results showed notably high roadkill numbers (e.g., for passerines, we estimate a total of 48.92 roadkills per km rate based on 8.04 observed rate during the road survey), along with substantial variation across different vertebrate groups. Furthermore, our case study confirms that walking and cycling surveys are more effective than driving surveys in detecting carcasses. 4. Our modelling framework offers an efficient approach to estimate mortality rates for a wide range of taxa. To optimize its application, extensive fieldwork for bias estimation and integration in analysis is needed. The accuracy of our framework may help managers to assess the impact of infrastructure-related mortality and prioritize conservation efforts to mitigate it.