Abstract
We initiated a 5-year study in mid-summer 2005 to determine patterns of bat activity and evaluate the use of acoustic monitoring to predict fatality of bats at a proposed wind energy facility in south-central Pennsylvania. The primary objectives of this study are to 1) determine activity of different bat species and species groups using the Casselman Wind Project in southcentral Pennsylvania prior to and after construction; 2) determine if indices of pre-construction bat activity can predict post-construction bat fatalities at turbine locations at the Casselman Wind Project (we also will combine data from this study with that from several facilities to determine if indices of pre-construction bat activity can predict post-construction bat fatalities at the wind facility-scale); 3) determine searcher efficiency and scavenger removal rates to adjust number of carcasses found and estimate bat fatality rates at the Casselman Wind Project; and 4) evaluate patterns of post-construction bat fatality in relation to wind speed, temperature, rotor speed, and other factors and assess the predictability of fatality based on these factors. This report focuses exclusively on the first year of post-construction fatality searches and estimates and patterns of bat fatality at the Casselman Wind Project in 2008.
We searched 10 of 23 turbines on 204 out of the total possible 211 days between 19 April and 15 November 2008 at the Casselman Wind Project. A total of 2,040 turbines searches were conducted at the 10 turbines during the 2008 field season.
We found 16 bird carcasses of 8 species (6 carcasses were of unknown species of birds) during searches between 19 April and 15 November 2008. Of 21 bird carcasses used for searcher efficiency trials, 5 of 7 placed in the easy class were found by searchers, while 4 of 7 and 2 of 7 were found in moderate and difficult classes, respectively. Average carcass persistence time was estimated to be 12.7 days (95% CI: 6.65, 24.37) for all birds. We estimated bat and bird fatality with three different estimators (two previous reported estimators [naïve and modified estimators] and a new estimator proposed and under review in the scientific literature [MH]; simulation studies suggest that the MH estimator is less biased than other estimators). The estimated number of bird fatalities per turbine was 4.69 (95% CI: 1.25, 14.31), 0.37 (95% CI: 0.07, 1.20), and 2.27 (95% CI: 0.88, 3.92) for the MH, naïve, and modified estimators, respectively.
We found 148 carcasses of 6 species of bats (1 carcasses was not identifiable to species) in the search plots from 19 April through 15 November 2009. Hoary bats, silver-haired bats, and eastern red bats were killed most frequently, representing 75% of estimated fatalities, and eastern pipistrelles and little brown bats represented 11 and 10% of estimated fatalities, respectively. Fatalities were found at all 10 turbines searched during the study. The total number of bat carcasses (all species and decay conditions combined) found at all turbines during searches on each day of the study increased in late summer and early fall (beginning around the middle of July), and no bat carcasses were found after 24 October 2008. One hundred-twenty-four (84%) of all bat carcasses were found between 15 July and 15 October 2009.
Data from 70 searcher efficiency trial carcasses were fit to a logistic regression model, and there was strong evidence of a difference in searcher efficiency among the visibility classes ( 2 2 = 14.32, p = 0.0002). All carcasses in the „easy‟ class that persisted long enough to be observed were found by searchers, while 71% of carcasses in the „moderate‟ class that persisted long enough to be observed were found. Only 13% of carcasses that persisted more than 2 weeks in the „difficult‟ class were found. Data from 114 scavenger removal trial carcasses were fit to an interval-censored parametric failure time model and there was no difference among visibility classes ( 2 2 = 1.78, p = 0.41). Average persistence time was estimated to be 31.9 days (95% CI: 17.4, 57.7 days).
The estimated number of bat fatalities per turbine was 32.30 (95% CI: 20.76, 51.43), 1.11 (95% CI: 0.50, 2.26), and 18.91 (95% CI: 15.27, 22.88) for the MH, naïve, and modified estimators, respectively. The average bat fatality estimate per turbine using the MH estimator was 1.7 times greater than that of the modified estimator and 29.1 times greater than estimates using the naïve estimator. Mean fatalities per turbine from the forested ridge portion of the project was 32.34 (n = 7, 95% CI: 22.39, 43.66), while mean fatalities per turbine from the strip mined ridge was 32.53 (n = 3, 95% CI: 24.53, 42.48), indicating no difference in fatalities between these two ridges and habitat conditions on the project site.
We will initiate a second year of post-construction fatality searches beginning 1 April and continuing through 15 November 2009 at the Casselman facility. A final report on the 3-years of pre-construction acoustic data is in preparation and will be distributed in early summer 2009. Also, we plan to correlate bat activity with fatality data for each turbine where these data were collected to determine if activity can predict turbine-scale fatality. That report will be prepared and distributed later in 2009.