Abstract
The Fowler Ridge Wind Farm (FRWF) is collectively owned by BP Wind Energy, Sempra Generation, and Dominion Energy, and is being developed in five separate phases for a total of 1,000 megawatts (MW). The project currently consists of 355 wind turbines in three phases in Benton County, Indiana. A post-construction casualty study of birds and bats was conducted within Phases I and III during 2009 by Western EcoSystems Technology, Inc (WEST). One carcass of an Indiana bat (Myotis sodalis), a federally endangered species, was found by wind plant personnel during the fall of 2009. As a result of this discovery, the US Fish and Wildlife Service (USFWS) Field Office in Bloomington, Indiana, has recommended that the FRWF develop a Habitat Conservation Plan (HCP). Consequently, the owners of the FRWF requested that WEST conduct further research of potential bat casualty and use rates at the FRWF during 2010 for use in completing a HCP and obtaining an Incidental Take Permit (ITP) from Region 3 of the USFWS. The results presented in this report were collected under a Scientific Research and Recovery Permit for the Indiana bat (TE15075A).
Studies at the FRWF were comprised of two components: a bat casualty study and an acoustical bat survey. The goals of the casualty study were: 1) to determine if Indiana bat fatalities were widespread and if fatalities occurred at levels that may reduce the viability of Indiana bat populations; and 2) determine if facility operation management can be utilized as an effective tool for reducing bat casualty rates. The goals of the bat use study were to: 1) measure the overall rates of use of the FRWF by Indiana bats and other bat species; and to 2) provide information that may help to better define periods when potential mitigation and minimization measures for Indiana bats may be most effective.
Casualty searches were conducted during the spring (April 13 – May 15, 2010) and fall (August 1 – October 15, 2010) migration periods. Casualty searches were completed on two types of plots: 1) 80 x 80-meter (m; 262 x 262-foot [ft]) square plots centered on 36 turbines and cleared of vegetation; and 2) roads and turbine gravel pads within 40 m (131 ft) of 100 additional turbines. Carcass searches were conducted daily at the 36 square plots, and weekly at the road and pad plots. During the spring, 128 turbines were sampled during the study. Surveys were unable to be completed on eight turbines originally selected for road and pad searches as part of the study plan due to construction activities. Surveys were completed at all 136 turbines in the fall.
Turbines were selected for carcass searches using a systematic sampling design with a random start. During the fall migration period the effectiveness of raising turbine cut-in speeds was tested at a sample of the daily searched turbines. Bat casualty rates were measured at two different cut-in speed adjustments or “treatments” and two sets of “control” or “reference” turbines with no turbine cut-in speed adjustment. Nine turbines were randomly selected from the sample of 36 daily searched turbines for use as a “control” sample, and had no treatments for the duration of the study. Treatments for cut-in speed adjustment and a second set of “control” turbines were rotated on a weekly basis between the remaining 27 daily search turbines, with nine turbines assigned to each group. The treatments included turbines with cut-in speeds raised to a lower level (5 m/s), and turbines with cut-in speeds raised to higher level (6.5 m/s). Turbines were randomly assigned to control and treatment groups among the 27 turbines, and treatments were distributed to ensure each turbine received 3 - 4 weeks of each treatment or control status.
During the spring migration period (April 13, 2010 – May 15, 2010) 80 x 80-m square plots were surveyed 33 times for a total of 1,169 searches. A total of 423 surveys were completed at roads and pads during spring 2010. Thirty-six bat casualties (18 silver-haired bats [Lasionycteris noctivagans], 15 eastern red bats [Lasiurus cinereus], two hoary bats [Lasiurus borealis], and one big brown bat [Eptesicus fuscus]) were found during the spring. Of the 36 bat casualties found, 32 were found during scheduled searches and four were found incidentally. All fresh, non-myotis bat casualties were collected and utilized in searcher efficiency and carcass removal trials. Four raptor casualties (three red-tailed hawks [Buteo jamaicensis] and one rough-legged hawk [Buteo lagopus]), eight small bird casualties (two European starlings [Sturnus vulgaris], one horned lark [Eremophila alpestris], one Tennessee warbler [Vermivora peregrina], one common yellowthroat [Geothlypis trichas], one unidentified dove and two unidentified birds) were also found during spring migration searches.
During the fall 3,871 plot surveys were conducted during 110 visits to the study area at the 36 daily search turbines, with an additional 1,618 surveys being conducted during 17 visits to the facility at the 100 turbines selected for road and pad surveys. A total of 556 bats were found during regularly scheduled searches, with an additional 25 bats found incidentally at turbine search plots. The remaining 228 bat carcasses were found outside of the study period during clearing searches or not on scheduled search turbines and were not included in the casualty estimates. The most commonly found bat species during scheduled searches was eastern red bat (353 fatalities, 63.49% of fatalities), followed by silver haired bat (98 fatalities, 17.62%), hoary bat (84 fatalities, 15.11%), and big brown bat (16 fatalities, 2.88%). Two tri-colored bats (Perimyotis subflavus), two little brown bats (Myotis lucifigus) and one Indiana bat (Myotis sodalis) were also found during scheduled searches.
During these studies, some casualties may have been missed on search plots because either searchers missed them, or scavengers removed them prior to the search. Experimental trials were conducted to estimate these adjustment factors, including an estimate of the probability a carcass is available to be found during a search (i.e. not removed by scavengers) and is detected by observers. A total of 222 fresh bat carcasses were used in calculating an empirical estimate of the probability of availability and detection. These carcasses were allowed to remain where placed for a 28 day period and the date when searchers found the carcasses was noted. Of the 222 carcasses placed 77 were placed on turbines where only the road and pad was searched and 145 were placed on turbines where cleared plots were searched. The empirical estimate for the probability of available and detected on road/pad searches was 0.51 for weekly searches and 0.58 for cleared plot turbines (i.e. daily searches).
Two casualty estimators were used. The Schoenfeld method utilizes a formula to estimate the probability of a carcass being available and detected by search, and is typically used for studies with infrequent search intervals. The Schoenfeld estimate was calculated to ensure a casualty estimate could be compared to the 2009 bat casualty estimate. The 2nd estimator used, the empirical estimate, was based on daily checks of bat carcasses, and represents an actual measure of carcass availability and detectability.
The estimated number of bat fatalities per turbine (based on Shoenfeld’s pi estimates) was 0.91 for daily search plots and 0.43 for weekly search plots during the spring season. These estimates increased during the fall season, with 17.69 bat fatalities per turbine per season at daily search turbines and 15.73 bats per turbine at weekly searches. Adjusted casualty estimates based on empirical estimates for the probability of available and detected were 1.25 bats per turbine for daily search turbines and 0.55 bats per turbine for weekly searches in the spring and 24.17 bats per turbine for daily search turbines and 20.96 per turbine for weekly searches in the fall, with an overall adjusted casualty estimate of 22.20 bats per turbine for the study period.
Bat casualty rates were lower at turbines where cut-in speeds were adjusted. Bat casualty rates and corresponding 90% bootstrap confidence intervals of 14.0 (11.6 - 16.5), 7.0 (7.0 - 9.1), and 3.0 (1.8 - 4.2) bats/turbine/season were recorded at control, 5.0 m/s, and 6.5 m/s treatment conditions respectively. Non-overlapping confidence intervals for observed casualty rates under each cut-in speed condition indicate a significant difference between treatments. An approximate 50% reduction in overall bat mortality was realized by raising the cut-in speed by 1.5 m/s (from 3.5 m/s to 5.0 m/s). An approximate 78% reduction in overall bat mortality was realized by raising the cut-in speed by 3.0 m/s (from 3.5 m/s to 6.5 m/s).
Relationships between bat casualty rates and weather factors were evaluated utilizing Poisson regression during the fall migration period of 2010. Bat casualty rates were highest on nights with higher bat activity, lower mean wind speeds, higher mean temperature and increasing variance in temperature, and increasing barometric pressure. In other words, higher bat fatalities were associated with warmer, calmer evenings prior to or after the passage of weather fronts and increasing bat activity. Bat casualty rates were also related to turbine type, with higher bat casualty rates observed at turbines with greater rotor swept areas.
Bat activity was monitored at the base of four daily search turbines on 36 nights during the spring period of April 9 through May 14. Bat activity was monitored at ground locations previously monitored during 2007 on 95 nights during the fall period of July 15 through October 18, 2010. Turbine locations were monitored at the base from August 1 through October 18, 2010; and nacelle units from August 17 through October 18, 2010. Anabat units recorded 185 bat passes on 137 detector-nights during the spring, and 5,721 bat passes on 685 detector-nights during the fall. The average bat activity for ground stations was 1.34 ± 0.29 bat passes per detector-night in the spring, and 11.46 ± 1.29 at ground stations and 3.10 ± 0.42 at raised stations during the fall.
Calls were placed in three groups, based on call frequency. During the spring study season, passes by low-frequency bats (LF; 46.5% of all bat passes) outnumbered passes by mid-frequency bats (MF; 28.1%) and high-frequency bats (HF; 25.4%), and this pattern was consistent among ground stations. Temporal patterns of use were similar among all three species groups during the study period, with the highest periods of use occurring during the first two weeks in August. During the fall study season, passes by LF bats (55.2% of all bat passes) outnumbered passes by HF (22.4%), and MF (22.3%), and this pattern was consistent among ground stations. Among raised stations, LF bats comprised 71.6% of passes. LF and MF comprised a greater proportion, and HF comprised a lower proportion, of calls recorded on the nacelle of turbines compared to ground stations. Temporal patterns of use were similar among all three species groups during the study period.
Using a call library, Myotis bat calls were analyzed to determine if call characteristics resembled Indiana bat calls. Three methods were utilized to identify potential Indiana bat calls, including a Discriminant Function Analysis, the “Britzke” filter, and a qualitative analysis by WEST biologists. Thirty calls were identified by the Discriminant Function Analysis as resembling calls made by Indiana bats in 2010, two calls passed the Britzke filter, and WEST biologists identified all 30 calls as being made by Myotis, but were not able to conclusively determine if calls were made by Indiana or little brown bat. Most potential Myotis calls were identified as occurring on the evening of August 9-10, 2010 at a ground-based station located away from turbines. Based on the presence of multiple potential calls, Myotis were present on the evening of August 9-10 at Station 2 (located away from turbine locations), and some potential exists that these calls were made by Indiana bats. The recording of individual potential Myotis sodalis calls identified by the Discriminant Function Analysis on other nights and stations does not provide strong evidence of Myotis sodalis activity due to the potential for incorrect identifications.
Bat activity was also actively sampled during both the spring and fall study seasons by driving east-west transects across the study area twice per week and recording calls at water sources and other potential bat habitat features. During the spring study period (five detector nights), active bat monitoring efforts recorded five HF bat passes, one MF bat pass, and nine LF bat passes. During the fall study season (21 detector-nights), active bat monitoring efforts recorded 32 HF bat passes, 59 MF bat pass, and 43 LF bat passes. None of the passes were identified as Indiana bat.
Bat activity was positively correlated with observed casualty rates, although other factors, such as weather and turbine type also played a role in observed bat casualty rates. Correlation between bat activity (number of bat passes per detector night) recorded with Anabat units located on the nacelles of selected turbines and observed casualty rates of fresh bat carcass finds was stronger than the correlation of casualty rates to Anabat detectors located at the base of turbines (Pearson’s correlation coefficients; 0.45 and 0.07 respectively).
The primary goals of the 2010 casualty study were to: 1) determine if Indiana bat fatalities were widespread, and if fatalities occurred at levels that may reduce the viability of Indiana bat populations; and 2) determine if facility operation management can be utilized as an effective tool for reducing bat casualty rates. Despite the increased survey effort over studies conducted in 2009, and despite reasonable probabilities of detecting bat carcasses during the study, very few Myotis carcasses were found, with only one Indiana bat carcass found during the study. Very few Myotis calls were recorded during the course of the study, and multiple calls potentially resembling Indiana bat calls were recorded at a single station on one evening during the fall migration period. It is clear that Indiana bat casualties were not widespread and did not occur at a high rate during the spring and fall migration period of Indiana bats. The data collected during this study also provide a reasonable estimate of overall bat casualties within the FRWF. More detailed estimates of Indiana bat casualties for future operations of the FRWF will be calculated within a Habitat Conservation Plan, currently being prepared for the FRWF.
The data collected at the FRWF show bat casualty rates were affected by bat activity, weather, turbine types, and habitat. These observed relationships provide a baseline for developing additional hypotheses for implementing and testing measures to reduce bat casualty rates. The results also show that Indiana bat fatalities occur at the FRWF, albeit at an extremely low rate. While the low rate of Indiana bat fatalities likely indicate impacts of operation of the FRWF are also low, the low casualty rates also make it difficult to recommendations specific measures for Indiana bats. Thus, recommendations provided in this report are aimed at providing a basis for designing a strategy for reducing overall bat casualty rates under the assumption that Indiana bat casualty rates will also be reduced.
Raising turbine cut-in speeds showed a reduction in bat casualty rates at FRWF during the fall of 2010, and similar reductions in bat casualty rates have also been documented by researchers at other wind-energy facilities. However; correlations observed between weather variables and bat fatalities suggest that focusing or adjusting cut-in speeds during periods when bats may be at most risk, or at locations where bats may be more likely to be found as fatalities, may provide similar reductions in bat fatalities.