Monitoring carcass persistence in windfarms: Recommendations for estimating mortality

Environmental impact assessments (EIAs) of windfarms require counting bat and bird carcasses around turbines to assess the number of individuals killed by collision or barotrauma. In order to accurately estimate mortality from carcass counts, most studies correct mortality by the probability of carcass removal between fatality searches, data that is obtained through persistence trials. However, there is currently no consensus on persistence trial design and estimation methodology. This study aimed to assess the sensitivity of several estimators employed during EIAs to collect persistence data. To this end, the persistence time of 266 carcasses was surveyed continuously using camera traps. This continuous observation data was resampled to generate virtual datasets collected using different survey intervals. Several estimators were then used to estimate mortality from a subsample of carcasses for each carcass survey interval. Mortality estimates obtained by the different estimators were very similar when persistence was observed continuously. However, differences between estimators emerged when persistence survey intervals widened, and even more so when fatality searches were conducted less frequently. The GenEst estimator seemed to produce less biased and more accurate mortality estimates compared to other estimators frequently used for EIAs. Based on these findings, we recommend limiting search intervals during persistence surveys as much as possible. If persistence time is low, we also recommend increasing fatality search frequency. For estimators using mean persistence time, the latter should be calculated using the median time between the last observation of the carcass and its first absence.