Evaluating and Adapting Surveys Designed to Predict Wind-Wildlife Interactions: A Simulation Approach Focused on Golden Eagles

Wind energy development is rapidly expanding in North America, and this expansion causes potential conflict with conservation-reliant wildlife. Within the USA, estimates of golden eagle (Aquila chrysaetos) use of areas... [ view full abstract ]

Wind energy development is rapidly expanding in North America, and this expansion causes potential conflict with conservation-reliant wildlife. Within the USA, estimates of golden eagle (Aquila chrysaetos) use of areas proposed for wind development are often generated with modified point counts. However, it is not always clear to what degree the data collected at on-site point counts relates to actual eagle use of an area. We investigated this relationship with GPS tracking data from individual eagles to understand the relationship between potentially observed and actual use of simulated wind project footprints by individual eagles.
Our analyses used telemetry data collected every 15 minutes from seven territorial eagles and six fledglings tracked in the Mojave Desert of California. We overlaid telemetry data on simulated project footprints and generated survey data for a variety of potential point count sampling strategies. We then compared eagle use within point count areas to eagle use within the simulated project footprints and evaluated the strength of that relationship for a variety of (a) point count sampling types (n = 3 designs; random, systematic, stratified), (b) sampling intensities (n=2 intensities; 30% or 60% area coverage), and (c) project footprint sizes (n=4 footprint sizes; 20, 40, 90 and 180 square km) by month in 2012 and 2013. We also evaluated how survey results would vary if surveys were conducted weekly, bi-weekly, monthly or every 4 months (variation in sampling frequency).
We found that in 2012, 3-way interactions among month of year, sampling intensity and size of the project footprint influenced the effectiveness of surveys in describing use of project footprints. Similarly, in 2013, 2-way interactions between month of year and intensity of sampling and between month of year and size of project footprints influenced the effectiveness of the surveys. We also found that when sampling frequency was reduced, the effectiveness of the surveys was drastically reduced. Although our work focuses on individual eagles (not eagle populations), our analysis shows the utility of simulations to improve surveys at wind energy facilities by considering the context-specific way point counts are laid out on the landscape.