Abstract
The United States Fish and Wildlife Service (USFWS) recommends using a Bayesian modeling framework to predict the annual golden eagle (Aquila chrysaetos) fatality rate at a wind energy facility, and the modeling approach defines prior distributions for collision rate and exposure rate from data at existing wind projects. Collision rate is defined as the number of collisions per exposure. Exposure rate is a function of minutes of eagle activity and survey effort; we used site‐specific data to update the prior distribution, resulting in the posterior distribution. An expansion factor adjusts the fatality prediction by accounting for daylight hours and the hazardous area within a wind project footprint. The product of the collision rate, posterior exposure rate, and expansion factor is the predicted annual fatality rate. We reviewed the input data for the prior distribution for collision rate, and provided an updated prior distribution for collision rate using more contemporary information. As suggested by the current USFWS guidance, we updated the prior distribution for collision rate from the USFWS baseline model with data from a site with modern specifications to obtain an updated prior distribution. We also created alternative prior distributions by estimating parameters for the distributions from data at 26 modern facilities only. Using more recent data and a larger data set, we determined the predictions using the alternative prior distributions for collision rate are approximately half the estimates using the original distribution.