Abstract
It is often necessary to estimate the number of wind turbine collision fatalities to assess impacts to birds following construction of wind farms. Detection of bird carcasses at wind turbines in the field is affected by carcass persistence and searcher detection rate. Integrated detection trials, which integrate carcass persistence and searcher detection trials into the periodic fatality search, have been proposed as an effective method for estimating these parameters. The purpose of our study was to test whether and how environmental factors affect integrated detection trial outcomes at multiple wind farms. We conducted this study at 10 wind farms in various environments of Japan. Binary data on trial outcomes in open versus forested areas served as our response variable in a generalized additive mixed model informed by days into trial, carcass body mass, season, whether snow covered the ground, and precipitation. For both ground cover types, days into trial and body mass were included in all the top models, suggesting that these factors most influenced bird carcass detection probability in integrated trials. The best model in open areas included days into trial, body mass, snow, and precipitation, and the best model in forested areas included days into trial, body mass, snow, precipitation, and season. Values of area under the curve indicated high accuracy of the best model for both ground cover types. The survey design needs to be appropriate to the size of the target species and to the environment in which the impacts will occur, such as the site's seasonality, its ground cover, and whether snow will cover the ground. Frequency of post-construction fatality monitoring should also be set cautiously, especially at wind farms located on small-bird migration routes, at wind farms in open areas, in areas with snow-covered ground in winter, or in forested areas during spring and summer because detection probabilities decline fastest under such conditions.