Estimating actual versus detected road mortality rates for a northern viper

Abstract

Mitigation of adverse effects of roads on wildlife benefits from a fundamental understanding of the number of animals killed by traffic. Road surveys are a key part in quantifying this mortality; however, without accounting for scavenging and observer detection rates, this method only reveals a minimum number of roadkilled animals. We quantified western rattlesnake (Crotalus oreganus) road mortality on a two-lane road in the dry interior of British Columbia, Canada. In 2015, 2016 we repeatedly surveyed 11.7 km of this road by walking, driving, and cycling. We determined the rate of carcass removal by scavengers and observer detection probability during walking surveys using planted snake carcasses. Fifty-two percent of carcasses were removed from the road by scavengers in two days, while 11% remained for >14 days. The mean observer detection probability was 0.76 for a team of two observers conducting surveys by walking. The mean road mortality rate, which accounts for scavenging and observer detection during walking surveys, was 0.06 rattlesnake deaths/km/day or 124 rattlesnake deaths/year based on mean rattlesnake active season duration. The estimated number of rattlesnake deaths was 2.7times the number of dead rattlesnakes detected through all surveys and incidental observations combined. Overall, walking surveys detected only 8% of the estimated rattlesnake deaths. Incidental observations of carcasses made while driving at other times in the study area were numerous, however the detection rate per kilometer was very low compared to more rigorous survey methods. The results highlight the magnitude of snake road deaths that can occur on a two-lane road with just 350 vehicles per day bisecting a wildlife reserve, the importance of standardized surveys and accurate assessments of sources of error in road mortality research, and the value of a model that can be used to calculate detection rates of roadkilled wildlife by trained observers. Keywords: Bias experiments, Mortality estimates, Observer detection, Snake, Road mortality, Scavenging