Estimating avian road mortality when only a single observer is available

Gulf Islands National Seashore in Florida, USA, hosts breeding populations of migratory beach-nesting birds, along with other avian species. It consists of 2 barrier islands with a public paved road that runs through the center. For the past decade, bird carcasses were observed opportunistically along the roadway, and this mortality sparked interest among wildlife managers about population-level effects on species of conservation concern. Accurately estimating wildlife mortality from carcass counts requires correcting negative bias due to imperfect detection and scavenging between surveys. We had only 1 observer available to perform road surveys, leading to a carcass detection rate of 1.0 after discovery, and a persistence rate that was only estimable after discovery. We therefore employed a robust design version of a time-of-detection model to estimate detection and persistence rates. We used a post hoc derivation of the superpopulation size for carcasses, which is an estimate of the total number of carcasses that were ever present on the road for any period of time. Each primary survey consisted of 2 passes on the road within 4 hours. Probability of detection for 2 passes was near 100%. We counted 77 avian carcasses between 22 May and 4 July 2013 and estimated that this represented 106.8 ± 21.3 (SE) birds killed on the road during that time period. Opportunistic surveying by park service staff detected between 0% and 100% of carcasses we estimated to be present, depending on the date. Standardized surveys are needed going forward to estimate annual variation in carcass numbers and to determine effectiveness of mortality mitigation. Moreover, further research can use these estimates to understand the contribution of road mortality to population trends.